diff options
Diffstat (limited to 'drivers')
115 files changed, 6262 insertions, 5260 deletions
diff --git a/drivers/acpi/sleep.c b/drivers/acpi/sleep.c index 01574a06653..42159a28f43 100644 --- a/drivers/acpi/sleep.c +++ b/drivers/acpi/sleep.c @@ -397,6 +397,14 @@ static struct dmi_system_id __initdata acpisleep_dmi_table[] = { }, }, { + .callback = init_set_sci_en_on_resume, + .ident = "Hewlett-Packard HP G7000 Notebook PC", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"), + DMI_MATCH(DMI_PRODUCT_NAME, "HP G7000 Notebook PC"), + }, + }, + { .callback = init_old_suspend_ordering, .ident = "Panasonic CF51-2L", .matches = { diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c index 336eb1ed73c..958c1fa4190 100644 --- a/drivers/ata/ahci.c +++ b/drivers/ata/ahci.c @@ -515,10 +515,14 @@ static const struct pci_device_id ahci_pci_tbl[] = { { PCI_VDEVICE(INTEL, 0x3a05), board_ahci }, /* ICH10 */ { PCI_VDEVICE(INTEL, 0x3a22), board_ahci }, /* ICH10 */ { PCI_VDEVICE(INTEL, 0x3a25), board_ahci }, /* ICH10 */ + { PCI_VDEVICE(INTEL, 0x3b22), board_ahci }, /* PCH AHCI */ + { PCI_VDEVICE(INTEL, 0x3b23), board_ahci }, /* PCH AHCI */ { PCI_VDEVICE(INTEL, 0x3b24), board_ahci }, /* PCH RAID */ { PCI_VDEVICE(INTEL, 0x3b25), board_ahci }, /* PCH RAID */ + { PCI_VDEVICE(INTEL, 0x3b29), board_ahci }, /* PCH AHCI */ { PCI_VDEVICE(INTEL, 0x3b2b), board_ahci }, /* PCH RAID */ { PCI_VDEVICE(INTEL, 0x3b2c), board_ahci }, /* PCH RAID */ + { PCI_VDEVICE(INTEL, 0x3b2f), board_ahci }, /* PCH AHCI */ /* JMicron 360/1/3/5/6, match class to avoid IDE function */ { PCI_VENDOR_ID_JMICRON, PCI_ANY_ID, PCI_ANY_ID, PCI_ANY_ID, diff --git a/drivers/ata/ata_piix.c b/drivers/ata/ata_piix.c index d0a14cf2bd7..56b8a3ff128 100644 --- a/drivers/ata/ata_piix.c +++ b/drivers/ata/ata_piix.c @@ -596,9 +596,12 @@ static const struct ich_laptop ich_laptop[] = { { 0x27DF, 0x0005, 0x0280 }, /* ICH7 on Acer 5602WLMi */ { 0x27DF, 0x1025, 0x0102 }, /* ICH7 on Acer 5602aWLMi */ { 0x27DF, 0x1025, 0x0110 }, /* ICH7 on Acer 3682WLMi */ + { 0x27DF, 0x1028, 0x02b0 }, /* ICH7 on unknown Dell */ { 0x27DF, 0x1043, 0x1267 }, /* ICH7 on Asus W5F */ { 0x27DF, 0x103C, 0x30A1 }, /* ICH7 on HP Compaq nc2400 */ + { 0x27DF, 0x103C, 0x361a }, /* ICH7 on unkown HP */ { 0x27DF, 0x1071, 0xD221 }, /* ICH7 on Hercules EC-900 */ + { 0x27DF, 0x152D, 0x0778 }, /* ICH7 on unknown Intel */ { 0x24CA, 0x1025, 0x0061 }, /* ICH4 on ACER Aspire 2023WLMi */ { 0x24CA, 0x1025, 0x003d }, /* ICH4 on ACER TM290 */ { 0x266F, 0x1025, 0x0066 }, /* ICH6 on ACER Aspire 1694WLMi */ diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c index 2c6aedaef71..8ac98ff16d7 100644 --- a/drivers/ata/libata-core.c +++ b/drivers/ata/libata-core.c @@ -1515,6 +1515,7 @@ static int ata_hpa_resize(struct ata_device *dev) return rc; } + dev->n_native_sectors = native_sectors; /* nothing to do? */ if (native_sectors <= sectors || !ata_ignore_hpa) { @@ -4099,6 +4100,7 @@ int ata_dev_revalidate(struct ata_device *dev, unsigned int new_class, unsigned int readid_flags) { u64 n_sectors = dev->n_sectors; + u64 n_native_sectors = dev->n_native_sectors; int rc; if (!ata_dev_enabled(dev)) @@ -4128,16 +4130,30 @@ int ata_dev_revalidate(struct ata_device *dev, unsigned int new_class, /* verify n_sectors hasn't changed */ if (dev->class == ATA_DEV_ATA && n_sectors && dev->n_sectors != n_sectors) { - ata_dev_printk(dev, KERN_INFO, "n_sectors mismatch " + ata_dev_printk(dev, KERN_WARNING, "n_sectors mismatch " "%llu != %llu\n", (unsigned long long)n_sectors, (unsigned long long)dev->n_sectors); - - /* restore original n_sectors */ - dev->n_sectors = n_sectors; - - rc = -ENODEV; - goto fail; + /* + * Something could have caused HPA to be unlocked + * involuntarily. If n_native_sectors hasn't changed + * and the new size matches it, keep the device. + */ + if (dev->n_native_sectors == n_native_sectors && + dev->n_sectors > n_sectors && + dev->n_sectors == n_native_sectors) { + ata_dev_printk(dev, KERN_WARNING, + "new n_sectors matches native, probably " + "late HPA unlock, continuing\n"); + /* keep using the old n_sectors */ + dev->n_sectors = n_sectors; + } else { + /* restore original n_[native]_sectors and fail */ + dev->n_native_sectors = n_native_sectors; + dev->n_sectors = n_sectors; + rc = -ENODEV; + goto fail; + } } return 0; diff --git a/drivers/ata/libata-eh.c b/drivers/ata/libata-eh.c index 1a07c061f64..79711b64054 100644 --- a/drivers/ata/libata-eh.c +++ b/drivers/ata/libata-eh.c @@ -2327,7 +2327,7 @@ int ata_eh_reset(struct ata_link *link, int classify, struct ata_port *ap = link->ap; struct ata_link *slave = ap->slave_link; struct ata_eh_context *ehc = &link->eh_context; - struct ata_eh_context *sehc = &slave->eh_context; + struct ata_eh_context *sehc = slave ? &slave->eh_context : NULL; unsigned int *classes = ehc->classes; unsigned int lflags = link->flags; int verbose = !(ehc->i.flags & ATA_EHI_QUIET); diff --git a/drivers/ata/pata_at91.c b/drivers/ata/pata_at91.c index 8561a9f195c..5702affcb32 100644 --- a/drivers/ata/pata_at91.c +++ b/drivers/ata/pata_at91.c @@ -26,9 +26,7 @@ #include <linux/platform_device.h> #include <linux/ata_platform.h> -#include <mach/at91sam9260_matrix.h> #include <mach/at91sam9_smc.h> -#include <mach/at91sam9260.h> #include <mach/board.h> #include <mach/gpio.h> @@ -44,65 +42,62 @@ struct at91_ide_info { unsigned long mode; unsigned int cs; + struct clk *mck; + void __iomem *ide_addr; void __iomem *alt_addr; }; -const struct ata_timing initial_timing = +static const struct ata_timing initial_timing = {XFER_PIO_0, 70, 290, 240, 600, 165, 150, 600, 0}; -static unsigned int calc_mck_cycles(unsigned int ns, unsigned int mck_hz) +static unsigned long calc_mck_cycles(unsigned long ns, unsigned long mck_hz) { unsigned long mul; - /* - * cycles = x [nsec] * f [Hz] / 10^9 [ns in sec] = - * x * (f / 1_000_000_000) = - * x * ((f * 65536) / 1_000_000_000) / 65536 = - * x * (((f / 10_000) * 65536) / 100_000) / 65536 = - */ + /* + * cycles = x [nsec] * f [Hz] / 10^9 [ns in sec] = + * x * (f / 1_000_000_000) = + * x * ((f * 65536) / 1_000_000_000) / 65536 = + * x * (((f / 10_000) * 65536) / 100_000) / 65536 = + */ - mul = (mck_hz / 10000) << 16; - mul /= 100000; + mul = (mck_hz / 10000) << 16; + mul /= 100000; - return (ns * mul + 65536) >> 16; /* rounding */ + return (ns * mul + 65536) >> 16; /* rounding */ } static void set_smc_mode(struct at91_ide_info *info) { - at91_sys_write(AT91_SMC_MODE(info->cs), info->mode); - return; + at91_sys_write(AT91_SMC_MODE(info->cs), info->mode); + return; } static void set_smc_timing(struct device *dev, struct at91_ide_info *info, const struct ata_timing *ata) { - int read_cycle, write_cycle, active, recover; - int nrd_setup, nrd_pulse, nrd_recover; - int nwe_setup, nwe_pulse; + unsigned long read_cycle, write_cycle, active, recover; + unsigned long nrd_setup, nrd_pulse, nrd_recover; + unsigned long nwe_setup, nwe_pulse; - int ncs_write_setup, ncs_write_pulse; - int ncs_read_setup, ncs_read_pulse; + unsigned long ncs_write_setup, ncs_write_pulse; + unsigned long ncs_read_setup, ncs_read_pulse; - unsigned int mck_hz; - struct clk *mck; + unsigned long mck_hz; read_cycle = ata->cyc8b; nrd_setup = ata->setup; nrd_pulse = ata->act8b; nrd_recover = ata->rec8b; - mck = clk_get(NULL, "mck"); - BUG_ON(IS_ERR(mck)); - mck_hz = clk_get_rate(mck); + mck_hz = clk_get_rate(info->mck); read_cycle = calc_mck_cycles(read_cycle, mck_hz); nrd_setup = calc_mck_cycles(nrd_setup, mck_hz); nrd_pulse = calc_mck_cycles(nrd_pulse, mck_hz); nrd_recover = calc_mck_cycles(nrd_recover, mck_hz); - clk_put(mck); - active = nrd_setup + nrd_pulse; recover = read_cycle - active; @@ -121,13 +116,13 @@ static void set_smc_timing(struct device *dev, ncs_write_setup = ncs_read_setup; ncs_write_pulse = ncs_read_pulse; - dev_dbg(dev, "ATA timings: nrd_setup = %d nrd_pulse = %d nrd_cycle = %d\n", + dev_dbg(dev, "ATA timings: nrd_setup = %lu nrd_pulse = %lu nrd_cycle = %lu\n", nrd_setup, nrd_pulse, read_cycle); - dev_dbg(dev, "ATA timings: nwe_setup = %d nwe_pulse = %d nwe_cycle = %d\n", + dev_dbg(dev, "ATA timings: nwe_setup = %lu nwe_pulse = %lu nwe_cycle = %lu\n", nwe_setup, nwe_pulse, write_cycle); - dev_dbg(dev, "ATA timings: ncs_read_setup = %d ncs_read_pulse = %d\n", + dev_dbg(dev, "ATA timings: ncs_read_setup = %lu ncs_read_pulse = %lu\n", ncs_read_setup, ncs_read_pulse); - dev_dbg(dev, "ATA timings: ncs_write_setup = %d ncs_write_pulse = %d\n", + dev_dbg(dev, "ATA timings: ncs_write_setup = %lu ncs_write_pulse = %lu\n", ncs_write_setup, ncs_write_pulse); at91_sys_write(AT91_SMC_SETUP(info->cs), @@ -217,6 +212,7 @@ static int __devinit pata_at91_probe(struct platform_device *pdev) struct resource *mem_res; struct ata_host *host; struct ata_port *ap; + int irq_flags = 0; int irq = 0; int ret; @@ -261,6 +257,13 @@ static int __devinit pata_at91_probe(struct platform_device *pdev) return -ENOMEM; } + info->mck = clk_get(NULL, "mck"); + + if (IS_ERR(info->mck)) { + dev_err(dev, "failed to get access to mck clock\n"); + return -ENODEV; + } + info->cs = board->chipselect; info->mode = AT91_SMC_READMODE | AT91_SMC_WRITEMODE | AT91_SMC_EXNWMODE_READY | AT91_SMC_BAT_SELECT | @@ -304,6 +307,7 @@ err_alt_ioremap: devm_iounmap(dev, info->ide_addr); err_ide_ioremap: + clk_put(info->mck); kfree(info); return ret; @@ -326,6 +330,7 @@ static int __devexit pata_at91_remove(struct platform_device *pdev) devm_iounmap(dev, info->ide_addr); devm_iounmap(dev, info->alt_addr); + clk_put(info->mck); kfree(info); return 0; diff --git a/drivers/ata/pata_octeon_cf.c b/drivers/ata/pata_octeon_cf.c index 8d9343accf3..abdd19fe990 100644 --- a/drivers/ata/pata_octeon_cf.c +++ b/drivers/ata/pata_octeon_cf.c @@ -653,7 +653,8 @@ static irqreturn_t octeon_cf_interrupt(int irq, void *dev_instance) ap = host->ports[i]; ocd = ap->dev->platform_data; - if (!ap || (ap->flags & ATA_FLAG_DISABLED)) + + if (ap->flags & ATA_FLAG_DISABLED) continue; ocd = ap->dev->platform_data; diff --git a/drivers/ata/pata_pcmcia.c b/drivers/ata/pata_pcmcia.c index f4d009ed50a..dc99e26f8e5 100644 --- a/drivers/ata/pata_pcmcia.c +++ b/drivers/ata/pata_pcmcia.c @@ -411,6 +411,7 @@ static struct pcmcia_device_id pcmcia_devices[] = { PCMCIA_DEVICE_PROD_ID123("PCMCIA", "IDE CARD", "F1", 0x281f1c5d, 0x1907960c, 0xf7fde8b9), PCMCIA_DEVICE_PROD_ID12("ARGOSY", "CD-ROM", 0x78f308dc, 0x66536591), PCMCIA_DEVICE_PROD_ID12("ARGOSY", "PnPIDE", 0x78f308dc, 0x0c694728), + PCMCIA_DEVICE_PROD_ID12("CNF ", "CD-ROM", 0x46d7db81, 0x66536591), PCMCIA_DEVICE_PROD_ID12("CNF CD-M", "CD-ROM", 0x7d93b852, 0x66536591), PCMCIA_DEVICE_PROD_ID12("Creative Technology Ltd.", "PCMCIA CD-ROM Interface Card", 0xff8c8a45, 0xfe8020c4), PCMCIA_DEVICE_PROD_ID12("Digital Equipment Corporation.", "Digital Mobile Media CD-ROM", 0x17692a66, 0xef1dcbde), diff --git a/drivers/ata/sata_mv.c b/drivers/ata/sata_mv.c index 23714aefb82..c19417e0220 100644 --- a/drivers/ata/sata_mv.c +++ b/drivers/ata/sata_mv.c @@ -2514,7 +2514,7 @@ static void mv_unexpected_intr(struct ata_port *ap, int edma_was_enabled) char *when = "idle"; ata_ehi_clear_desc(ehi); - if (!ap || (ap->flags & ATA_FLAG_DISABLED)) { + if (ap->flags & ATA_FLAG_DISABLED) { when = "disabled"; } else if (edma_was_enabled) { when = "EDMA enabled"; diff --git a/drivers/ata/sata_sil.c b/drivers/ata/sata_sil.c index 030ec079b18..35bd5cc7f28 100644 --- a/drivers/ata/sata_sil.c +++ b/drivers/ata/sata_sil.c @@ -532,7 +532,7 @@ static irqreturn_t sil_interrupt(int irq, void *dev_instance) struct ata_port *ap = host->ports[i]; u32 bmdma2 = readl(mmio_base + sil_port[ap->port_no].bmdma2); - if (unlikely(!ap || ap->flags & ATA_FLAG_DISABLED)) + if (unlikely(ap->flags & ATA_FLAG_DISABLED)) continue; /* turn off SATA_IRQ if not supported */ diff --git a/drivers/base/firmware_class.c b/drivers/base/firmware_class.c index f285f441fab..7376367bcb8 100644 --- a/drivers/base/firmware_class.c +++ b/drivers/base/firmware_class.c @@ -180,7 +180,6 @@ static ssize_t firmware_loading_store(struct device *dev, goto err; } /* Pages will be freed by vfree() */ - fw_priv->pages = NULL; fw_priv->page_array_size = 0; fw_priv->nr_pages = 0; complete(&fw_priv->completion); diff --git a/drivers/base/sys.c b/drivers/base/sys.c index 79a9ae5238a..0d903909af7 100644 --- a/drivers/base/sys.c +++ b/drivers/base/sys.c @@ -275,9 +275,9 @@ int sysdev_register(struct sys_device *sysdev) drv->add(sysdev); } mutex_unlock(&sysdev_drivers_lock); + kobject_uevent(&sysdev->kobj, KOBJ_ADD); } - kobject_uevent(&sysdev->kobj, KOBJ_ADD); return error; } diff --git a/drivers/char/n_tty.c b/drivers/char/n_tty.c index ff47907ff1b..973be2f4419 100644 --- a/drivers/char/n_tty.c +++ b/drivers/char/n_tty.c @@ -1583,6 +1583,7 @@ static int n_tty_open(struct tty_struct *tty) static inline int input_available_p(struct tty_struct *tty, int amt) { + tty_flush_to_ldisc(tty); if (tty->icanon) { if (tty->canon_data) return 1; diff --git a/drivers/char/pty.c b/drivers/char/pty.c index 3850a68f265..6e6942c45f5 100644 --- a/drivers/char/pty.c +++ b/drivers/char/pty.c @@ -52,7 +52,6 @@ static void pty_close(struct tty_struct *tty, struct file *filp) return; tty->link->packet = 0; set_bit(TTY_OTHER_CLOSED, &tty->link->flags); - tty_flip_buffer_push(tty->link); wake_up_interruptible(&tty->link->read_wait); wake_up_interruptible(&tty->link->write_wait); if (tty->driver->subtype == PTY_TYPE_MASTER) { @@ -208,7 +207,6 @@ static int pty_open(struct tty_struct *tty, struct file *filp) clear_bit(TTY_OTHER_CLOSED, &tty->link->flags); set_bit(TTY_THROTTLED, &tty->flags); retval = 0; - tty->low_latency = 1; out: return retval; } diff --git a/drivers/char/sysrq.c b/drivers/char/sysrq.c index 0db35857e4d..5d7a02f63e1 100644 --- a/drivers/char/sysrq.c +++ b/drivers/char/sysrq.c @@ -35,7 +35,6 @@ #include <linux/spinlock.h> #include <linux/vt_kern.h> #include <linux/workqueue.h> -#include <linux/kexec.h> #include <linux/hrtimer.h> #include <linux/oom.h> @@ -124,9 +123,12 @@ static struct sysrq_key_op sysrq_unraw_op = { static void sysrq_handle_crash(int key, struct tty_struct *tty) { char *killer = NULL; + + panic_on_oops = 1; /* force panic */ + wmb(); *killer = 1; } -static struct sysrq_key_op sysrq_crashdump_op = { +static struct sysrq_key_op sysrq_crash_op = { .handler = sysrq_handle_crash, .help_msg = "Crash", .action_msg = "Trigger a crash", @@ -401,7 +403,7 @@ static struct sysrq_key_op *sysrq_key_table[36] = { */ NULL, /* a */ &sysrq_reboot_op, /* b */ - &sysrq_crashdump_op, /* c & ibm_emac driver debug */ + &sysrq_crash_op, /* c & ibm_emac driver debug */ &sysrq_showlocks_op, /* d */ &sysrq_term_op, /* e */ &sysrq_moom_op, /* f */ diff --git a/drivers/char/tty_buffer.c b/drivers/char/tty_buffer.c index 810ee25d66a..3108991c5c8 100644 --- a/drivers/char/tty_buffer.c +++ b/drivers/char/tty_buffer.c @@ -462,6 +462,19 @@ static void flush_to_ldisc(struct work_struct *work) } /** + * tty_flush_to_ldisc + * @tty: tty to push + * + * Push the terminal flip buffers to the line discipline. + * + * Must not be called from IRQ context. + */ +void tty_flush_to_ldisc(struct tty_struct *tty) +{ + flush_to_ldisc(&tty->buf.work.work); +} + +/** * tty_flip_buffer_push - terminal * @tty: tty to push * diff --git a/drivers/char/vr41xx_giu.c b/drivers/char/vr41xx_giu.c deleted file mode 100644 index e69de29bb2d..00000000000 --- a/drivers/char/vr41xx_giu.c +++ /dev/null diff --git a/drivers/edac/x38_edac.c b/drivers/edac/x38_edac.c index 2406c2ce284..d4ec6059317 100644 --- a/drivers/edac/x38_edac.c +++ b/drivers/edac/x38_edac.c @@ -30,7 +30,7 @@ /* Intel X38 register addresses - device 0 function 0 - DRAM Controller */ #define X38_MCHBAR_LOW 0x48 /* MCH Memory Mapped Register BAR */ -#define X38_MCHBAR_HIGH 0x4b +#define X38_MCHBAR_HIGH 0x4c #define X38_MCHBAR_MASK 0xfffffc000ULL /* bits 35:14 */ #define X38_MMR_WINDOW_SIZE 16384 diff --git a/drivers/gpu/drm/radeon/Makefile b/drivers/gpu/drm/radeon/Makefile index 5fae1e074b4..013d3805994 100644 --- a/drivers/gpu/drm/radeon/Makefile +++ b/drivers/gpu/drm/radeon/Makefile @@ -13,7 +13,8 @@ radeon-$(CONFIG_DRM_RADEON_KMS) += radeon_device.o radeon_kms.o \ radeon_encoders.o radeon_display.o radeon_cursor.o radeon_i2c.o \ radeon_clocks.o radeon_fb.o radeon_gem.o radeon_ring.o radeon_irq_kms.o \ radeon_cs.o radeon_bios.o radeon_benchmark.o r100.o r300.o r420.o \ - rs400.o rs600.o rs690.o rv515.o r520.o r600.o rs780.o rv770.o + rs400.o rs600.o rs690.o rv515.o r520.o r600.o rs780.o rv770.o \ + radeon_test.o radeon-$(CONFIG_COMPAT) += radeon_ioc32.o diff --git a/drivers/gpu/drm/radeon/atombios_crtc.c b/drivers/gpu/drm/radeon/atombios_crtc.c index c0080cc9bf8..74d034f77c6 100644 --- a/drivers/gpu/drm/radeon/atombios_crtc.c +++ b/drivers/gpu/drm/radeon/atombios_crtc.c @@ -31,6 +31,132 @@ #include "atom.h" #include "atom-bits.h" +static void atombios_overscan_setup(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + SET_CRTC_OVERSCAN_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_OverScan); + int a1, a2; + + memset(&args, 0, sizeof(args)); + + args.usOverscanRight = 0; + args.usOverscanLeft = 0; + args.usOverscanBottom = 0; + args.usOverscanTop = 0; + args.ucCRTC = radeon_crtc->crtc_id; + + switch (radeon_crtc->rmx_type) { + case RMX_CENTER: + args.usOverscanTop = (adjusted_mode->crtc_vdisplay - mode->crtc_vdisplay) / 2; + args.usOverscanBottom = (adjusted_mode->crtc_vdisplay - mode->crtc_vdisplay) / 2; + args.usOverscanLeft = (adjusted_mode->crtc_hdisplay - mode->crtc_hdisplay) / 2; + args.usOverscanRight = (adjusted_mode->crtc_hdisplay - mode->crtc_hdisplay) / 2; + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + break; + case RMX_ASPECT: + a1 = mode->crtc_vdisplay * adjusted_mode->crtc_hdisplay; + a2 = adjusted_mode->crtc_vdisplay * mode->crtc_hdisplay; + + if (a1 > a2) { + args.usOverscanLeft = (adjusted_mode->crtc_hdisplay - (a2 / mode->crtc_vdisplay)) / 2; + args.usOverscanRight = (adjusted_mode->crtc_hdisplay - (a2 / mode->crtc_vdisplay)) / 2; + } else if (a2 > a1) { + args.usOverscanLeft = (adjusted_mode->crtc_vdisplay - (a1 / mode->crtc_hdisplay)) / 2; + args.usOverscanRight = (adjusted_mode->crtc_vdisplay - (a1 / mode->crtc_hdisplay)) / 2; + } + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + break; + case RMX_FULL: + default: + args.usOverscanRight = 0; + args.usOverscanLeft = 0; + args.usOverscanBottom = 0; + args.usOverscanTop = 0; + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + break; + } +} + +static void atombios_scaler_setup(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + ENABLE_SCALER_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, EnableScaler); + /* fixme - fill in enc_priv for atom dac */ + enum radeon_tv_std tv_std = TV_STD_NTSC; + + if (!ASIC_IS_AVIVO(rdev) && radeon_crtc->crtc_id) + return; + + memset(&args, 0, sizeof(args)); + + args.ucScaler = radeon_crtc->crtc_id; + + if (radeon_crtc->devices & (ATOM_DEVICE_TV_SUPPORT)) { + switch (tv_std) { + case TV_STD_NTSC: + default: + args.ucTVStandard = ATOM_TV_NTSC; + break; + case TV_STD_PAL: + args.ucTVStandard = ATOM_TV_PAL; + break; + case TV_STD_PAL_M: + args.ucTVStandard = ATOM_TV_PALM; + break; + case TV_STD_PAL_60: + args.ucTVStandard = ATOM_TV_PAL60; + break; + case TV_STD_NTSC_J: + args.ucTVStandard = ATOM_TV_NTSCJ; + break; + case TV_STD_SCART_PAL: + args.ucTVStandard = ATOM_TV_PAL; /* ??? */ + break; + case TV_STD_SECAM: + args.ucTVStandard = ATOM_TV_SECAM; + break; + case TV_STD_PAL_CN: + args.ucTVStandard = ATOM_TV_PALCN; + break; + } + args.ucEnable = SCALER_ENABLE_MULTITAP_MODE; + } else if (radeon_crtc->devices & (ATOM_DEVICE_CV_SUPPORT)) { + args.ucTVStandard = ATOM_TV_CV; + args.ucEnable = SCALER_ENABLE_MULTITAP_MODE; + } else { + switch (radeon_crtc->rmx_type) { + case RMX_FULL: + args.ucEnable = ATOM_SCALER_EXPANSION; + break; + case RMX_CENTER: + args.ucEnable = ATOM_SCALER_CENTER; + break; + case RMX_ASPECT: + args.ucEnable = ATOM_SCALER_EXPANSION; + break; + default: + if (ASIC_IS_AVIVO(rdev)) + args.ucEnable = ATOM_SCALER_DISABLE; + else + args.ucEnable = ATOM_SCALER_CENTER; + break; + } + } + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + if (radeon_crtc->devices & (ATOM_DEVICE_CV_SUPPORT | ATOM_DEVICE_TV_SUPPORT) + && rdev->family >= CHIP_RV515 && rdev->family <= CHIP_RV570) { + atom_rv515_force_tv_scaler(rdev); + } +} + static void atombios_lock_crtc(struct drm_crtc *crtc, int lock) { struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); @@ -203,6 +329,12 @@ void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode) if (ASIC_IS_AVIVO(rdev)) { uint32_t ss_cntl; + if ((rdev->family == CHIP_RS600) || + (rdev->family == CHIP_RS690) || + (rdev->family == CHIP_RS740)) + pll_flags |= (RADEON_PLL_USE_FRAC_FB_DIV | + RADEON_PLL_PREFER_CLOSEST_LOWER); + if (ASIC_IS_DCE32(rdev) && mode->clock > 200000) /* range limits??? */ pll_flags |= RADEON_PLL_PREFER_HIGH_FB_DIV; else @@ -321,7 +453,7 @@ int atombios_crtc_set_base(struct drm_crtc *crtc, int x, int y, struct drm_gem_object *obj; struct drm_radeon_gem_object *obj_priv; uint64_t fb_location; - uint32_t fb_format, fb_pitch_pixels; + uint32_t fb_format, fb_pitch_pixels, tiling_flags; if (!crtc->fb) return -EINVAL; @@ -358,7 +490,14 @@ int atombios_crtc_set_base(struct drm_crtc *crtc, int x, int y, return -EINVAL; } - /* TODO tiling */ + radeon_object_get_tiling_flags(obj->driver_private, + &tiling_flags, NULL); + if (tiling_flags & RADEON_TILING_MACRO) + fb_format |= AVIVO_D1GRPH_MACRO_ADDRESS_MODE; + + if (tiling_flags & RADEON_TILING_MICRO) + fb_format |= AVIVO_D1GRPH_TILED; + if (radeon_crtc->crtc_id == 0) WREG32(AVIVO_D1VGA_CONTROL, 0); else @@ -509,6 +648,9 @@ int atombios_crtc_mode_set(struct drm_crtc *crtc, radeon_crtc_set_base(crtc, x, y, old_fb); radeon_legacy_atom_set_surface(crtc); } + atombios_overscan_setup(crtc, mode, adjusted_mode); + atombios_scaler_setup(crtc); + radeon_bandwidth_update(rdev); return 0; } @@ -516,6 +658,8 @@ static bool atombios_crtc_mode_fixup(struct drm_crtc *crtc, struct drm_display_mode *mode, struct drm_display_mode *adjusted_mode) { + if (!radeon_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode)) + return false; return true; } @@ -548,148 +692,3 @@ void radeon_atombios_init_crtc(struct drm_device *dev, AVIVO_D2CRTC_H_TOTAL - AVIVO_D1CRTC_H_TOTAL; drm_crtc_helper_add(&radeon_crtc->base, &atombios_helper_funcs); } - -void radeon_init_disp_bw_avivo(struct drm_device *dev, - struct drm_display_mode *mode1, - uint32_t pixel_bytes1, - struct drm_display_mode *mode2, - uint32_t pixel_bytes2) -{ - struct radeon_device *rdev = dev->dev_private; - fixed20_12 min_mem_eff; - fixed20_12 peak_disp_bw, mem_bw, pix_clk, pix_clk2, temp_ff; - fixed20_12 sclk_ff, mclk_ff; - uint32_t dc_lb_memory_split, temp; - - min_mem_eff.full = rfixed_const_8(0); - if (rdev->disp_priority == 2) { - uint32_t mc_init_misc_lat_timer = 0; - if (rdev->family == CHIP_RV515) - mc_init_misc_lat_timer = - RREG32_MC(RV515_MC_INIT_MISC_LAT_TIMER); - else if (rdev->family == CHIP_RS690) - mc_init_misc_lat_timer = - RREG32_MC(RS690_MC_INIT_MISC_LAT_TIMER); - - mc_init_misc_lat_timer &= - ~(R300_MC_DISP1R_INIT_LAT_MASK << - R300_MC_DISP1R_INIT_LAT_SHIFT); - mc_init_misc_lat_timer &= - ~(R300_MC_DISP0R_INIT_LAT_MASK << - R300_MC_DISP0R_INIT_LAT_SHIFT); - - if (mode2) - mc_init_misc_lat_timer |= - (1 << R300_MC_DISP1R_INIT_LAT_SHIFT); - if (mode1) - mc_init_misc_lat_timer |= - (1 << R300_MC_DISP0R_INIT_LAT_SHIFT); - - if (rdev->family == CHIP_RV515) - WREG32_MC(RV515_MC_INIT_MISC_LAT_TIMER, - mc_init_misc_lat_timer); - else if (rdev->family == CHIP_RS690) - WREG32_MC(RS690_MC_INIT_MISC_LAT_TIMER, - mc_init_misc_lat_timer); - } - - /* - * determine is there is enough bw for current mode - */ - temp_ff.full = rfixed_const(100); - mclk_ff.full = rfixed_const(rdev->clock.default_mclk); - mclk_ff.full = rfixed_div(mclk_ff, temp_ff); - sclk_ff.full = rfixed_const(rdev->clock.default_sclk); - sclk_ff.full = rfixed_div(sclk_ff, temp_ff); - - temp = (rdev->mc.vram_width / 8) * (rdev->mc.vram_is_ddr ? 2 : 1); - temp_ff.full = rfixed_const(temp); - mem_bw.full = rfixed_mul(mclk_ff, temp_ff); - mem_bw.full = rfixed_mul(mem_bw, min_mem_eff); - - pix_clk.full = 0; - pix_clk2.full = 0; - peak_disp_bw.full = 0; - if (mode1) { - temp_ff.full = rfixed_const(1000); - pix_clk.full = rfixed_const(mode1->clock); /* convert to fixed point */ - pix_clk.full = rfixed_div(pix_clk, temp_ff); - temp_ff.full = rfixed_const(pixel_bytes1); - peak_disp_bw.full += rfixed_mul(pix_clk, temp_ff); - } - if (mode2) { - temp_ff.full = rfixed_const(1000); - pix_clk2.full = rfixed_const(mode2->clock); /* convert to fixed point */ - pix_clk2.full = rfixed_div(pix_clk2, temp_ff); - temp_ff.full = rfixed_const(pixel_bytes2); - peak_disp_bw.full += rfixed_mul(pix_clk2, temp_ff); - } - - if (peak_disp_bw.full >= mem_bw.full) { - DRM_ERROR - ("You may not have enough display bandwidth for current mode\n" - "If you have flickering problem, try to lower resolution, refresh rate, or color depth\n"); - printk("peak disp bw %d, mem_bw %d\n", - rfixed_trunc(peak_disp_bw), rfixed_trunc(mem_bw)); - } - - /* - * Line Buffer Setup - * There is a single line buffer shared by both display controllers. - * DC_LB_MEMORY_SPLIT controls how that line buffer is shared between the display - * controllers. The paritioning can either be done manually or via one of four - * preset allocations specified in bits 1:0: - * 0 - line buffer is divided in half and shared between each display controller - * 1 - D1 gets 3/4 of the line buffer, D2 gets 1/4 - * 2 - D1 gets the whole buffer - * 3 - D1 gets 1/4 of the line buffer, D2 gets 3/4 - * Setting bit 2 of DC_LB_MEMORY_SPLIT controls switches to manual allocation mode. - * In manual allocation mode, D1 always starts at 0, D1 end/2 is specified in bits - * 14:4; D2 allocation follows D1. - */ - - /* is auto or manual better ? */ - dc_lb_memory_split = - RREG32(AVIVO_DC_LB_MEMORY_SPLIT) & ~AVIVO_DC_LB_MEMORY_SPLIT_MASK; - dc_lb_memory_split &= ~AVIVO_DC_LB_MEMORY_SPLIT_SHIFT_MODE; -#if 1 - /* auto */ - if (mode1 && mode2) { - if (mode1->hdisplay > mode2->hdisplay) { - if (mode1->hdisplay > 2560) - dc_lb_memory_split |= - AVIVO_DC_LB_MEMORY_SPLIT_D1_3Q_D2_1Q; - else - dc_lb_memory_split |= - AVIVO_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; - } else if (mode2->hdisplay > mode1->hdisplay) { - if (mode2->hdisplay > 2560) - dc_lb_memory_split |= - AVIVO_DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q; - else - dc_lb_memory_split |= - AVIVO_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; - } else - dc_lb_memory_split |= - AVIVO_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; - } else if (mode1) { - dc_lb_memory_split |= AVIVO_DC_LB_MEMORY_SPLIT_D1_ONLY; - } else if (mode2) { - dc_lb_memory_split |= AVIVO_DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q; - } -#else - /* manual */ - dc_lb_memory_split |= AVIVO_DC_LB_MEMORY_SPLIT_SHIFT_MODE; - dc_lb_memory_split &= - ~(AVIVO_DC_LB_DISP1_END_ADR_MASK << - AVIVO_DC_LB_DISP1_END_ADR_SHIFT); - if (mode1) { - dc_lb_memory_split |= - ((((mode1->hdisplay / 2) + 64) & AVIVO_DC_LB_DISP1_END_ADR_MASK) - << AVIVO_DC_LB_DISP1_END_ADR_SHIFT); - } else if (mode2) { - dc_lb_memory_split |= (0 << AVIVO_DC_LB_DISP1_END_ADR_SHIFT); - } -#endif - WREG32(AVIVO_DC_LB_MEMORY_SPLIT, dc_lb_memory_split); -} diff --git a/drivers/gpu/drm/radeon/r100.c b/drivers/gpu/drm/radeon/r100.c index c550932a108..05a44896dff 100644 --- a/drivers/gpu/drm/radeon/r100.c +++ b/drivers/gpu/drm/radeon/r100.c @@ -110,7 +110,7 @@ int r100_pci_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr) if (i < 0 || i > rdev->gart.num_gpu_pages) { return -EINVAL; } - rdev->gart.table.ram.ptr[i] = cpu_to_le32((uint32_t)addr); + rdev->gart.table.ram.ptr[i] = cpu_to_le32(lower_32_bits(addr)); return 0; } @@ -173,8 +173,12 @@ void r100_mc_setup(struct radeon_device *rdev) DRM_ERROR("Failed to register debugfs file for R100 MC !\n"); } /* Write VRAM size in case we are limiting it */ - WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.vram_size); - tmp = rdev->mc.vram_location + rdev->mc.vram_size - 1; + WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.real_vram_size); + /* Novell bug 204882 for RN50/M6/M7 with 8/16/32MB VRAM, + * if the aperture is 64MB but we have 32MB VRAM + * we report only 32MB VRAM but we have to set MC_FB_LOCATION + * to 64MB, otherwise the gpu accidentially dies */ + tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; tmp = REG_SET(RADEON_MC_FB_TOP, tmp >> 16); tmp |= REG_SET(RADEON_MC_FB_START, rdev->mc.vram_location >> 16); WREG32(RADEON_MC_FB_LOCATION, tmp); @@ -215,7 +219,6 @@ int r100_mc_init(struct radeon_device *rdev) r100_pci_gart_disable(rdev); /* Setup GPU memory space */ - rdev->mc.vram_location = 0xFFFFFFFFUL; rdev->mc.gtt_location = 0xFFFFFFFFUL; if (rdev->flags & RADEON_IS_AGP) { r = radeon_agp_init(rdev); @@ -753,6 +756,102 @@ int r100_cs_packet_parse(struct radeon_cs_parser *p, } /** + * r100_cs_packet_next_vline() - parse userspace VLINE packet + * @parser: parser structure holding parsing context. + * + * Userspace sends a special sequence for VLINE waits. + * PACKET0 - VLINE_START_END + value + * PACKET0 - WAIT_UNTIL +_value + * RELOC (P3) - crtc_id in reloc. + * + * This function parses this and relocates the VLINE START END + * and WAIT UNTIL packets to the correct crtc. + * It also detects a switched off crtc and nulls out the + * wait in that case. + */ +int r100_cs_packet_parse_vline(struct radeon_cs_parser *p) +{ + struct radeon_cs_chunk *ib_chunk; + struct drm_mode_object *obj; + struct drm_crtc *crtc; + struct radeon_crtc *radeon_crtc; + struct radeon_cs_packet p3reloc, waitreloc; + int crtc_id; + int r; + uint32_t header, h_idx, reg; + + ib_chunk = &p->chunks[p->chunk_ib_idx]; + + /* parse the wait until */ + r = r100_cs_packet_parse(p, &waitreloc, p->idx); + if (r) + return r; + + /* check its a wait until and only 1 count */ + if (waitreloc.reg != RADEON_WAIT_UNTIL || + waitreloc.count != 0) { + DRM_ERROR("vline wait had illegal wait until segment\n"); + r = -EINVAL; + return r; + } + + if (ib_chunk->kdata[waitreloc.idx + 1] != RADEON_WAIT_CRTC_VLINE) { + DRM_ERROR("vline wait had illegal wait until\n"); + r = -EINVAL; + return r; + } + + /* jump over the NOP */ + r = r100_cs_packet_parse(p, &p3reloc, p->idx); + if (r) + return r; + + h_idx = p->idx - 2; + p->idx += waitreloc.count; + p->idx += p3reloc.count; + + header = ib_chunk->kdata[h_idx]; + crtc_id = ib_chunk->kdata[h_idx + 5]; + reg = ib_chunk->kdata[h_idx] >> 2; + mutex_lock(&p->rdev->ddev->mode_config.mutex); + obj = drm_mode_object_find(p->rdev->ddev, crtc_id, DRM_MODE_OBJECT_CRTC); + if (!obj) { + DRM_ERROR("cannot find crtc %d\n", crtc_id); + r = -EINVAL; + goto out; + } + crtc = obj_to_crtc(obj); + radeon_crtc = to_radeon_crtc(crtc); + crtc_id = radeon_crtc->crtc_id; + + if (!crtc->enabled) { + /* if the CRTC isn't enabled - we need to nop out the wait until */ + ib_chunk->kdata[h_idx + 2] = PACKET2(0); + ib_chunk->kdata[h_idx + 3] = PACKET2(0); + } else if (crtc_id == 1) { + switch (reg) { + case AVIVO_D1MODE_VLINE_START_END: + header &= R300_CP_PACKET0_REG_MASK; + header |= AVIVO_D2MODE_VLINE_START_END >> 2; + break; + case RADEON_CRTC_GUI_TRIG_VLINE: + header &= R300_CP_PACKET0_REG_MASK; + header |= RADEON_CRTC2_GUI_TRIG_VLINE >> 2; + break; + default: + DRM_ERROR("unknown crtc reloc\n"); + r = -EINVAL; + goto out; + } + ib_chunk->kdata[h_idx] = header; + ib_chunk->kdata[h_idx + 3] |= RADEON_ENG_DISPLAY_SELECT_CRTC1; + } +out: + mutex_unlock(&p->rdev->ddev->mode_config.mutex); + return r; +} + +/** * r100_cs_packet_next_reloc() - parse next packet which should be reloc packet3 * @parser: parser structure holding parsing context. * @data: pointer to relocation data @@ -814,6 +913,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p, unsigned idx; bool onereg; int r; + u32 tile_flags = 0; ib = p->ib->ptr; ib_chunk = &p->chunks[p->chunk_ib_idx]; @@ -825,6 +925,15 @@ static int r100_packet0_check(struct radeon_cs_parser *p, } for (i = 0; i <= pkt->count; i++, idx++, reg += 4) { switch (reg) { + case RADEON_CRTC_GUI_TRIG_VLINE: + r = r100_cs_packet_parse_vline(p); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + break; /* FIXME: only allow PACKET3 blit? easier to check for out of * range access */ case RADEON_DST_PITCH_OFFSET: @@ -838,7 +947,20 @@ static int r100_packet0_check(struct radeon_cs_parser *p, } tmp = ib_chunk->kdata[idx] & 0x003fffff; tmp += (((u32)reloc->lobj.gpu_offset) >> 10); - ib[idx] = (ib_chunk->kdata[idx] & 0xffc00000) | tmp; + + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + tile_flags |= RADEON_DST_TILE_MACRO; + if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) { + if (reg == RADEON_SRC_PITCH_OFFSET) { + DRM_ERROR("Cannot src blit from microtiled surface\n"); + r100_cs_dump_packet(p, pkt); + return -EINVAL; + } + tile_flags |= RADEON_DST_TILE_MICRO; + } + + tmp |= tile_flags; + ib[idx] = (ib_chunk->kdata[idx] & 0x3fc00000) | tmp; break; case RADEON_RB3D_DEPTHOFFSET: case RADEON_RB3D_COLOROFFSET: @@ -869,6 +991,11 @@ static int r100_packet0_check(struct radeon_cs_parser *p, case R300_TX_OFFSET_0+52: case R300_TX_OFFSET_0+56: case R300_TX_OFFSET_0+60: + /* rn50 has no 3D engine so fail on any 3d setup */ + if (ASIC_IS_RN50(p->rdev)) { + DRM_ERROR("attempt to use RN50 3D engine failed\n"); + return -EINVAL; + } r = r100_cs_packet_next_reloc(p, &reloc); if (r) { DRM_ERROR("No reloc for ib[%d]=0x%04X\n", @@ -878,6 +1005,25 @@ static int r100_packet0_check(struct radeon_cs_parser *p, } ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); break; + case R300_RB3D_COLORPITCH0: + case RADEON_RB3D_COLORPITCH: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + tile_flags |= RADEON_COLOR_TILE_ENABLE; + if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + tile_flags |= RADEON_COLOR_MICROTILE_ENABLE; + + tmp = ib_chunk->kdata[idx] & ~(0x7 << 16); + tmp |= tile_flags; + ib[idx] = tmp; + break; default: /* FIXME: we don't want to allow anyothers packet */ break; @@ -1256,29 +1402,100 @@ static void r100_vram_get_type(struct radeon_device *rdev) } } -void r100_vram_info(struct radeon_device *rdev) +static u32 r100_get_accessible_vram(struct radeon_device *rdev) { - r100_vram_get_type(rdev); + u32 aper_size; + u8 byte; + + aper_size = RREG32(RADEON_CONFIG_APER_SIZE); + + /* Set HDP_APER_CNTL only on cards that are known not to be broken, + * that is has the 2nd generation multifunction PCI interface + */ + if (rdev->family == CHIP_RV280 || + rdev->family >= CHIP_RV350) { + WREG32_P(RADEON_HOST_PATH_CNTL, RADEON_HDP_APER_CNTL, + ~RADEON_HDP_APER_CNTL); + DRM_INFO("Generation 2 PCI interface, using max accessible memory\n"); + return aper_size * 2; + } + + /* Older cards have all sorts of funny issues to deal with. First + * check if it's a multifunction card by reading the PCI config + * header type... Limit those to one aperture size + */ + pci_read_config_byte(rdev->pdev, 0xe, &byte); + if (byte & 0x80) { + DRM_INFO("Generation 1 PCI interface in multifunction mode\n"); + DRM_INFO("Limiting VRAM to one aperture\n"); + return aper_size; + } + + /* Single function older card. We read HDP_APER_CNTL to see how the BIOS + * have set it up. We don't write this as it's broken on some ASICs but + * we expect the BIOS to have done the right thing (might be too optimistic...) + */ + if (RREG32(RADEON_HOST_PATH_CNTL) & RADEON_HDP_APER_CNTL) + return aper_size * 2; + return aper_size; +} + +void r100_vram_init_sizes(struct radeon_device *rdev) +{ + u64 config_aper_size; + u32 accessible; + + config_aper_size = RREG32(RADEON_CONFIG_APER_SIZE); if (rdev->flags & RADEON_IS_IGP) { uint32_t tom; /* read NB_TOM to get the amount of ram stolen for the GPU */ tom = RREG32(RADEON_NB_TOM); - rdev->mc.vram_size = (((tom >> 16) - (tom & 0xffff) + 1) << 16); - WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.vram_size); + rdev->mc.real_vram_size = (((tom >> 16) - (tom & 0xffff) + 1) << 16); + /* for IGPs we need to keep VRAM where it was put by the BIOS */ + rdev->mc.vram_location = (tom & 0xffff) << 16; + WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.real_vram_size); + rdev->mc.mc_vram_size = rdev->mc.real_vram_size; } else { - rdev->mc.vram_size = RREG32(RADEON_CONFIG_MEMSIZE); + rdev->mc.real_vram_size = RREG32(RADEON_CONFIG_MEMSIZE); /* Some production boards of m6 will report 0 * if it's 8 MB */ - if (rdev->mc.vram_size == 0) { - rdev->mc.vram_size = 8192 * 1024; - WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.vram_size); + if (rdev->mc.real_vram_size == 0) { + rdev->mc.real_vram_size = 8192 * 1024; + WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.real_vram_size); } + /* let driver place VRAM */ + rdev->mc.vram_location = 0xFFFFFFFFUL; + /* Fix for RN50, M6, M7 with 8/16/32(??) MBs of VRAM - + * Novell bug 204882 + along with lots of ubuntu ones */ + if (config_aper_size > rdev->mc.real_vram_size) + rdev->mc.mc_vram_size = config_aper_size; + else + rdev->mc.mc_vram_size = rdev->mc.real_vram_size; } + /* work out accessible VRAM */ + accessible = r100_get_accessible_vram(rdev); + rdev->mc.aper_base = drm_get_resource_start(rdev->ddev, 0); rdev->mc.aper_size = drm_get_resource_len(rdev->ddev, 0); + + if (accessible > rdev->mc.aper_size) + accessible = rdev->mc.aper_size; + + if (rdev->mc.mc_vram_size > rdev->mc.aper_size) + rdev->mc.mc_vram_size = rdev->mc.aper_size; + + if (rdev->mc.real_vram_size > rdev->mc.aper_size) + rdev->mc.real_vram_size = rdev->mc.aper_size; +} + +void r100_vram_info(struct radeon_device *rdev) +{ + r100_vram_get_type(rdev); + + r100_vram_init_sizes(rdev); } @@ -1533,3 +1750,530 @@ int r100_debugfs_mc_info_init(struct radeon_device *rdev) return 0; #endif } + +int r100_set_surface_reg(struct radeon_device *rdev, int reg, + uint32_t tiling_flags, uint32_t pitch, + uint32_t offset, uint32_t obj_size) +{ + int surf_index = reg * 16; + int flags = 0; + + /* r100/r200 divide by 16 */ + if (rdev->family < CHIP_R300) + flags = pitch / 16; + else + flags = pitch / 8; + + if (rdev->family <= CHIP_RS200) { + if ((tiling_flags & (RADEON_TILING_MACRO|RADEON_TILING_MICRO)) + == (RADEON_TILING_MACRO|RADEON_TILING_MICRO)) + flags |= RADEON_SURF_TILE_COLOR_BOTH; + if (tiling_flags & RADEON_TILING_MACRO) + flags |= RADEON_SURF_TILE_COLOR_MACRO; + } else if (rdev->family <= CHIP_RV280) { + if (tiling_flags & (RADEON_TILING_MACRO)) + flags |= R200_SURF_TILE_COLOR_MACRO; + if (tiling_flags & RADEON_TILING_MICRO) + flags |= R200_SURF_TILE_COLOR_MICRO; + } else { + if (tiling_flags & RADEON_TILING_MACRO) + flags |= R300_SURF_TILE_MACRO; + if (tiling_flags & RADEON_TILING_MICRO) + flags |= R300_SURF_TILE_MICRO; + } + + DRM_DEBUG("writing surface %d %d %x %x\n", reg, flags, offset, offset+obj_size-1); + WREG32(RADEON_SURFACE0_INFO + surf_index, flags); + WREG32(RADEON_SURFACE0_LOWER_BOUND + surf_index, offset); + WREG32(RADEON_SURFACE0_UPPER_BOUND + surf_index, offset + obj_size - 1); + return 0; +} + +void r100_clear_surface_reg(struct radeon_device *rdev, int reg) +{ + int surf_index = reg * 16; + WREG32(RADEON_SURFACE0_INFO + surf_index, 0); +} + +void r100_bandwidth_update(struct radeon_device *rdev) +{ + fixed20_12 trcd_ff, trp_ff, tras_ff, trbs_ff, tcas_ff; + fixed20_12 sclk_ff, mclk_ff, sclk_eff_ff, sclk_delay_ff; + fixed20_12 peak_disp_bw, mem_bw, pix_clk, pix_clk2, temp_ff, crit_point_ff; + uint32_t temp, data, mem_trcd, mem_trp, mem_tras; + fixed20_12 memtcas_ff[8] = { + fixed_init(1), + fixed_init(2), + fixed_init(3), + fixed_init(0), + fixed_init_half(1), + fixed_init_half(2), + fixed_init(0), + }; + fixed20_12 memtcas_rs480_ff[8] = { + fixed_init(0), + fixed_init(1), + fixed_init(2), + fixed_init(3), + fixed_init(0), + fixed_init_half(1), + fixed_init_half(2), + fixed_init_half(3), + }; + fixed20_12 memtcas2_ff[8] = { + fixed_init(0), + fixed_init(1), + fixed_init(2), + fixed_init(3), + fixed_init(4), + fixed_init(5), + fixed_init(6), + fixed_init(7), + }; + fixed20_12 memtrbs[8] = { + fixed_init(1), + fixed_init_half(1), + fixed_init(2), + fixed_init_half(2), + fixed_init(3), + fixed_init_half(3), + fixed_init(4), + fixed_init_half(4) + }; + fixed20_12 memtrbs_r4xx[8] = { + fixed_init(4), + fixed_init(5), + fixed_init(6), + fixed_init(7), + fixed_init(8), + fixed_init(9), + fixed_init(10), + fixed_init(11) + }; + fixed20_12 min_mem_eff; + fixed20_12 mc_latency_sclk, mc_latency_mclk, k1; + fixed20_12 cur_latency_mclk, cur_latency_sclk; + fixed20_12 disp_latency, disp_latency_overhead, disp_drain_rate, + disp_drain_rate2, read_return_rate; + fixed20_12 time_disp1_drop_priority; + int c; + int cur_size = 16; /* in octawords */ + int critical_point = 0, critical_point2; +/* uint32_t read_return_rate, time_disp1_drop_priority; */ + int stop_req, max_stop_req; + struct drm_display_mode *mode1 = NULL; + struct drm_display_mode *mode2 = NULL; + uint32_t pixel_bytes1 = 0; + uint32_t pixel_bytes2 = 0; + + if (rdev->mode_info.crtcs[0]->base.enabled) { + mode1 = &rdev->mode_info.crtcs[0]->base.mode; + pixel_bytes1 = rdev->mode_info.crtcs[0]->base.fb->bits_per_pixel / 8; + } + if (rdev->mode_info.crtcs[1]->base.enabled) { + mode2 = &rdev->mode_info.crtcs[1]->base.mode; + pixel_bytes2 = rdev->mode_info.crtcs[1]->base.fb->bits_per_pixel / 8; + } + + min_mem_eff.full = rfixed_const_8(0); + /* get modes */ + if ((rdev->disp_priority == 2) && ASIC_IS_R300(rdev)) { + uint32_t mc_init_misc_lat_timer = RREG32(R300_MC_INIT_MISC_LAT_TIMER); + mc_init_misc_lat_timer &= ~(R300_MC_DISP1R_INIT_LAT_MASK << R300_MC_DISP1R_INIT_LAT_SHIFT); + mc_init_misc_lat_timer &= ~(R300_MC_DISP0R_INIT_LAT_MASK << R300_MC_DISP0R_INIT_LAT_SHIFT); + /* check crtc enables */ + if (mode2) + mc_init_misc_lat_timer |= (1 << R300_MC_DISP1R_INIT_LAT_SHIFT); + if (mode1) + mc_init_misc_lat_timer |= (1 << R300_MC_DISP0R_INIT_LAT_SHIFT); + WREG32(R300_MC_INIT_MISC_LAT_TIMER, mc_init_misc_lat_timer); + } + + /* + * determine is there is enough bw for current mode + */ + mclk_ff.full = rfixed_const(rdev->clock.default_mclk); + temp_ff.full = rfixed_const(100); + mclk_ff.full = rfixed_div(mclk_ff, temp_ff); + sclk_ff.full = rfixed_const(rdev->clock.default_sclk); + sclk_ff.full = rfixed_div(sclk_ff, temp_ff); + + temp = (rdev->mc.vram_width / 8) * (rdev->mc.vram_is_ddr ? 2 : 1); + temp_ff.full = rfixed_const(temp); + mem_bw.full = rfixed_mul(mclk_ff, temp_ff); + + pix_clk.full = 0; + pix_clk2.full = 0; + peak_disp_bw.full = 0; + if (mode1) { + temp_ff.full = rfixed_const(1000); + pix_clk.full = rfixed_const(mode1->clock); /* convert to fixed point */ + pix_clk.full = rfixed_div(pix_clk, temp_ff); + temp_ff.full = rfixed_const(pixel_bytes1); + peak_disp_bw.full += rfixed_mul(pix_clk, temp_ff); + } + if (mode2) { + temp_ff.full = rfixed_const(1000); + pix_clk2.full = rfixed_const(mode2->clock); /* convert to fixed point */ + pix_clk2.full = rfixed_div(pix_clk2, temp_ff); + temp_ff.full = rfixed_const(pixel_bytes2); + peak_disp_bw.full += rfixed_mul(pix_clk2, temp_ff); + } + + mem_bw.full = rfixed_mul(mem_bw, min_mem_eff); + if (peak_disp_bw.full >= mem_bw.full) { + DRM_ERROR("You may not have enough display bandwidth for current mode\n" + "If you have flickering problem, try to lower resolution, refresh rate, or color depth\n"); + } + + /* Get values from the EXT_MEM_CNTL register...converting its contents. */ + temp = RREG32(RADEON_MEM_TIMING_CNTL); + if ((rdev->family == CHIP_RV100) || (rdev->flags & RADEON_IS_IGP)) { /* RV100, M6, IGPs */ + mem_trcd = ((temp >> 2) & 0x3) + 1; + mem_trp = ((temp & 0x3)) + 1; + mem_tras = ((temp & 0x70) >> 4) + 1; + } else if (rdev->family == CHIP_R300 || + rdev->family == CHIP_R350) { /* r300, r350 */ + mem_trcd = (temp & 0x7) + 1; + mem_trp = ((temp >> 8) & 0x7) + 1; + mem_tras = ((temp >> 11) & 0xf) + 4; + } else if (rdev->family == CHIP_RV350 || + rdev->family <= CHIP_RV380) { + /* rv3x0 */ + mem_trcd = (temp & 0x7) + 3; + mem_trp = ((temp >> 8) & 0x7) + 3; + mem_tras = ((temp >> 11) & 0xf) + 6; + } else if (rdev->family == CHIP_R420 || + rdev->family == CHIP_R423 || + rdev->family == CHIP_RV410) { + /* r4xx */ + mem_trcd = (temp & 0xf) + 3; + if (mem_trcd > 15) + mem_trcd = 15; + mem_trp = ((temp >> 8) & 0xf) + 3; + if (mem_trp > 15) + mem_trp = 15; + mem_tras = ((temp >> 12) & 0x1f) + 6; + if (mem_tras > 31) + mem_tras = 31; + } else { /* RV200, R200 */ + mem_trcd = (temp & 0x7) + 1; + mem_trp = ((temp >> 8) & 0x7) + 1; + mem_tras = ((temp >> 12) & 0xf) + 4; + } + /* convert to FF */ + trcd_ff.full = rfixed_const(mem_trcd); + trp_ff.full = rfixed_const(mem_trp); + tras_ff.full = rfixed_const(mem_tras); + + /* Get values from the MEM_SDRAM_MODE_REG register...converting its */ + temp = RREG32(RADEON_MEM_SDRAM_MODE_REG); + data = (temp & (7 << 20)) >> 20; + if ((rdev->family == CHIP_RV100) || rdev->flags & RADEON_IS_IGP) { + if (rdev->family == CHIP_RS480) /* don't think rs400 */ + tcas_ff = memtcas_rs480_ff[data]; + else + tcas_ff = memtcas_ff[data]; + } else + tcas_ff = memtcas2_ff[data]; + + if (rdev->family == CHIP_RS400 || + rdev->family == CHIP_RS480) { + /* extra cas latency stored in bits 23-25 0-4 clocks */ + data = (temp >> 23) & 0x7; + if (data < 5) + tcas_ff.full += rfixed_const(data); + } + + if (ASIC_IS_R300(rdev) && !(rdev->flags & RADEON_IS_IGP)) { + /* on the R300, Tcas is included in Trbs. + */ + temp = RREG32(RADEON_MEM_CNTL); + data = (R300_MEM_NUM_CHANNELS_MASK & temp); + if (data == 1) { + if (R300_MEM_USE_CD_CH_ONLY & temp) { + temp = RREG32(R300_MC_IND_INDEX); + temp &= ~R300_MC_IND_ADDR_MASK; + temp |= R300_MC_READ_CNTL_CD_mcind; + WREG32(R300_MC_IND_INDEX, temp); + temp = RREG32(R300_MC_IND_DATA); + data = (R300_MEM_RBS_POSITION_C_MASK & temp); + } else { + temp = RREG32(R300_MC_READ_CNTL_AB); + data = (R300_MEM_RBS_POSITION_A_MASK & temp); + } + } else { + temp = RREG32(R300_MC_READ_CNTL_AB); + data = (R300_MEM_RBS_POSITION_A_MASK & temp); + } + if (rdev->family == CHIP_RV410 || + rdev->family == CHIP_R420 || + rdev->family == CHIP_R423) + trbs_ff = memtrbs_r4xx[data]; + else + trbs_ff = memtrbs[data]; + tcas_ff.full += trbs_ff.full; + } + + sclk_eff_ff.full = sclk_ff.full; + + if (rdev->flags & RADEON_IS_AGP) { + fixed20_12 agpmode_ff; + agpmode_ff.full = rfixed_const(radeon_agpmode); + temp_ff.full = rfixed_const_666(16); + sclk_eff_ff.full -= rfixed_mul(agpmode_ff, temp_ff); + } + /* TODO PCIE lanes may affect this - agpmode == 16?? */ + + if (ASIC_IS_R300(rdev)) { + sclk_delay_ff.full = rfixed_const(250); + } else { + if ((rdev->family == CHIP_RV100) || + rdev->flags & RADEON_IS_IGP) { + if (rdev->mc.vram_is_ddr) + sclk_delay_ff.full = rfixed_const(41); + else + sclk_delay_ff.full = rfixed_const(33); + } else { + if (rdev->mc.vram_width == 128) + sclk_delay_ff.full = rfixed_const(57); + else + sclk_delay_ff.full = rfixed_const(41); + } + } + + mc_latency_sclk.full = rfixed_div(sclk_delay_ff, sclk_eff_ff); + + if (rdev->mc.vram_is_ddr) { + if (rdev->mc.vram_width == 32) { + k1.full = rfixed_const(40); + c = 3; + } else { + k1.full = rfixed_const(20); + c = 1; + } + } else { + k1.full = rfixed_const(40); + c = 3; + } + + temp_ff.full = rfixed_const(2); + mc_latency_mclk.full = rfixed_mul(trcd_ff, temp_ff); + temp_ff.full = rfixed_const(c); + mc_latency_mclk.full += rfixed_mul(tcas_ff, temp_ff); + temp_ff.full = rfixed_const(4); + mc_latency_mclk.full += rfixed_mul(tras_ff, temp_ff); + mc_latency_mclk.full += rfixed_mul(trp_ff, temp_ff); + mc_latency_mclk.full += k1.full; + + mc_latency_mclk.full = rfixed_div(mc_latency_mclk, mclk_ff); + mc_latency_mclk.full += rfixed_div(temp_ff, sclk_eff_ff); + + /* + HW cursor time assuming worst case of full size colour cursor. + */ + temp_ff.full = rfixed_const((2 * (cur_size - (rdev->mc.vram_is_ddr + 1)))); + temp_ff.full += trcd_ff.full; + if (temp_ff.full < tras_ff.full) + temp_ff.full = tras_ff.full; + cur_latency_mclk.full = rfixed_div(temp_ff, mclk_ff); + + temp_ff.full = rfixed_const(cur_size); + cur_latency_sclk.full = rfixed_div(temp_ff, sclk_eff_ff); + /* + Find the total latency for the display data. + */ + disp_latency_overhead.full = rfixed_const(80); + disp_latency_overhead.full = rfixed_div(disp_latency_overhead, sclk_ff); + mc_latency_mclk.full += disp_latency_overhead.full + cur_latency_mclk.full; + mc_latency_sclk.full += disp_latency_overhead.full + cur_latency_sclk.full; + + if (mc_latency_mclk.full > mc_latency_sclk.full) + disp_latency.full = mc_latency_mclk.full; + else + disp_latency.full = mc_latency_sclk.full; + + /* setup Max GRPH_STOP_REQ default value */ + if (ASIC_IS_RV100(rdev)) + max_stop_req = 0x5c; + else + max_stop_req = 0x7c; + + if (mode1) { + /* CRTC1 + Set GRPH_BUFFER_CNTL register using h/w defined optimal values. + GRPH_STOP_REQ <= MIN[ 0x7C, (CRTC_H_DISP + 1) * (bit depth) / 0x10 ] + */ + stop_req = mode1->hdisplay * pixel_bytes1 / 16; + + if (stop_req > max_stop_req) + stop_req = max_stop_req; + + /* + Find the drain rate of the display buffer. + */ + temp_ff.full = rfixed_const((16/pixel_bytes1)); + disp_drain_rate.full = rfixed_div(pix_clk, temp_ff); + + /* + Find the critical point of the display buffer. + */ + crit_point_ff.full = rfixed_mul(disp_drain_rate, disp_latency); + crit_point_ff.full += rfixed_const_half(0); + + critical_point = rfixed_trunc(crit_point_ff); + + if (rdev->disp_priority == 2) { + critical_point = 0; + } + + /* + The critical point should never be above max_stop_req-4. Setting + GRPH_CRITICAL_CNTL = 0 will thus force high priority all the time. + */ + if (max_stop_req - critical_point < 4) + critical_point = 0; + + if (critical_point == 0 && mode2 && rdev->family == CHIP_R300) { + /* some R300 cards have problem with this set to 0, when CRTC2 is enabled.*/ + critical_point = 0x10; + } + + temp = RREG32(RADEON_GRPH_BUFFER_CNTL); + temp &= ~(RADEON_GRPH_STOP_REQ_MASK); + temp |= (stop_req << RADEON_GRPH_STOP_REQ_SHIFT); + temp &= ~(RADEON_GRPH_START_REQ_MASK); + if ((rdev->family == CHIP_R350) && + (stop_req > 0x15)) { + stop_req -= 0x10; + } + temp |= (stop_req << RADEON_GRPH_START_REQ_SHIFT); + temp |= RADEON_GRPH_BUFFER_SIZE; + temp &= ~(RADEON_GRPH_CRITICAL_CNTL | + RADEON_GRPH_CRITICAL_AT_SOF | + RADEON_GRPH_STOP_CNTL); + /* + Write the result into the register. + */ + WREG32(RADEON_GRPH_BUFFER_CNTL, ((temp & ~RADEON_GRPH_CRITICAL_POINT_MASK) | + (critical_point << RADEON_GRPH_CRITICAL_POINT_SHIFT))); + +#if 0 + if ((rdev->family == CHIP_RS400) || + (rdev->family == CHIP_RS480)) { + /* attempt to program RS400 disp regs correctly ??? */ + temp = RREG32(RS400_DISP1_REG_CNTL); + temp &= ~(RS400_DISP1_START_REQ_LEVEL_MASK | + RS400_DISP1_STOP_REQ_LEVEL_MASK); + WREG32(RS400_DISP1_REQ_CNTL1, (temp | + (critical_point << RS400_DISP1_START_REQ_LEVEL_SHIFT) | + (critical_point << RS400_DISP1_STOP_REQ_LEVEL_SHIFT))); + temp = RREG32(RS400_DMIF_MEM_CNTL1); + temp &= ~(RS400_DISP1_CRITICAL_POINT_START_MASK | + RS400_DISP1_CRITICAL_POINT_STOP_MASK); + WREG32(RS400_DMIF_MEM_CNTL1, (temp | + (critical_point << RS400_DISP1_CRITICAL_POINT_START_SHIFT) | + (critical_point << RS400_DISP1_CRITICAL_POINT_STOP_SHIFT))); + } +#endif + + DRM_DEBUG("GRPH_BUFFER_CNTL from to %x\n", + /* (unsigned int)info->SavedReg->grph_buffer_cntl, */ + (unsigned int)RREG32(RADEON_GRPH_BUFFER_CNTL)); + } + + if (mode2) { + u32 grph2_cntl; + stop_req = mode2->hdisplay * pixel_bytes2 / 16; + + if (stop_req > max_stop_req) + stop_req = max_stop_req; + + /* + Find the drain rate of the display buffer. + */ + temp_ff.full = rfixed_const((16/pixel_bytes2)); + disp_drain_rate2.full = rfixed_div(pix_clk2, temp_ff); + + grph2_cntl = RREG32(RADEON_GRPH2_BUFFER_CNTL); + grph2_cntl &= ~(RADEON_GRPH_STOP_REQ_MASK); + grph2_cntl |= (stop_req << RADEON_GRPH_STOP_REQ_SHIFT); + grph2_cntl &= ~(RADEON_GRPH_START_REQ_MASK); + if ((rdev->family == CHIP_R350) && + (stop_req > 0x15)) { + stop_req -= 0x10; + } + grph2_cntl |= (stop_req << RADEON_GRPH_START_REQ_SHIFT); + grph2_cntl |= RADEON_GRPH_BUFFER_SIZE; + grph2_cntl &= ~(RADEON_GRPH_CRITICAL_CNTL | + RADEON_GRPH_CRITICAL_AT_SOF | + RADEON_GRPH_STOP_CNTL); + + if ((rdev->family == CHIP_RS100) || + (rdev->family == CHIP_RS200)) + critical_point2 = 0; + else { + temp = (rdev->mc.vram_width * rdev->mc.vram_is_ddr + 1)/128; + temp_ff.full = rfixed_const(temp); + temp_ff.full = rfixed_mul(mclk_ff, temp_ff); + if (sclk_ff.full < temp_ff.full) + temp_ff.full = sclk_ff.full; + + read_return_rate.full = temp_ff.full; + + if (mode1) { + temp_ff.full = read_return_rate.full - disp_drain_rate.full; + time_disp1_drop_priority.full = rfixed_div(crit_point_ff, temp_ff); + } else { + time_disp1_drop_priority.full = 0; + } + crit_point_ff.full = disp_latency.full + time_disp1_drop_priority.full + disp_latency.full; + crit_point_ff.full = rfixed_mul(crit_point_ff, disp_drain_rate2); + crit_point_ff.full += rfixed_const_half(0); + + critical_point2 = rfixed_trunc(crit_point_ff); + + if (rdev->disp_priority == 2) { + critical_point2 = 0; + } + + if (max_stop_req - critical_point2 < 4) + critical_point2 = 0; + + } + + if (critical_point2 == 0 && rdev->family == CHIP_R300) { + /* some R300 cards have problem with this set to 0 */ + critical_point2 = 0x10; + } + + WREG32(RADEON_GRPH2_BUFFER_CNTL, ((grph2_cntl & ~RADEON_GRPH_CRITICAL_POINT_MASK) | + (critical_point2 << RADEON_GRPH_CRITICAL_POINT_SHIFT))); + + if ((rdev->family == CHIP_RS400) || + (rdev->family == CHIP_RS480)) { +#if 0 + /* attempt to program RS400 disp2 regs correctly ??? */ + temp = RREG32(RS400_DISP2_REQ_CNTL1); + temp &= ~(RS400_DISP2_START_REQ_LEVEL_MASK | + RS400_DISP2_STOP_REQ_LEVEL_MASK); + WREG32(RS400_DISP2_REQ_CNTL1, (temp | + (critical_point2 << RS400_DISP1_START_REQ_LEVEL_SHIFT) | + (critical_point2 << RS400_DISP1_STOP_REQ_LEVEL_SHIFT))); + temp = RREG32(RS400_DISP2_REQ_CNTL2); + temp &= ~(RS400_DISP2_CRITICAL_POINT_START_MASK | + RS400_DISP2_CRITICAL_POINT_STOP_MASK); + WREG32(RS400_DISP2_REQ_CNTL2, (temp | + (critical_point2 << RS400_DISP2_CRITICAL_POINT_START_SHIFT) | + (critical_point2 << RS400_DISP2_CRITICAL_POINT_STOP_SHIFT))); +#endif + WREG32(RS400_DISP2_REQ_CNTL1, 0x105DC1CC); + WREG32(RS400_DISP2_REQ_CNTL2, 0x2749D000); + WREG32(RS400_DMIF_MEM_CNTL1, 0x29CA71DC); + WREG32(RS400_DISP1_REQ_CNTL1, 0x28FBC3AC); + } + + DRM_DEBUG("GRPH2_BUFFER_CNTL from to %x\n", + (unsigned int)RREG32(RADEON_GRPH2_BUFFER_CNTL)); + } +} diff --git a/drivers/gpu/drm/radeon/r300.c b/drivers/gpu/drm/radeon/r300.c index e2ed5bc0817..9c8d41534a5 100644 --- a/drivers/gpu/drm/radeon/r300.c +++ b/drivers/gpu/drm/radeon/r300.c @@ -30,6 +30,8 @@ #include "drm.h" #include "radeon_reg.h" #include "radeon.h" +#include "radeon_drm.h" +#include "radeon_share.h" /* r300,r350,rv350,rv370,rv380 depends on : */ void r100_hdp_reset(struct radeon_device *rdev); @@ -44,6 +46,7 @@ int r100_gui_wait_for_idle(struct radeon_device *rdev); int r100_cs_packet_parse(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx); +int r100_cs_packet_parse_vline(struct radeon_cs_parser *p); int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, struct radeon_cs_reloc **cs_reloc); int r100_cs_parse_packet0(struct radeon_cs_parser *p, @@ -150,8 +153,13 @@ int rv370_pcie_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr) if (i < 0 || i > rdev->gart.num_gpu_pages) { return -EINVAL; } - addr = (((u32)addr) >> 8) | ((upper_32_bits(addr) & 0xff) << 4) | 0xC; - writel(cpu_to_le32(addr), ((void __iomem *)ptr) + (i * 4)); + addr = (lower_32_bits(addr) >> 8) | + ((upper_32_bits(addr) & 0xff) << 24) | + 0xc; + /* on x86 we want this to be CPU endian, on powerpc + * on powerpc without HW swappers, it'll get swapped on way + * into VRAM - so no need for cpu_to_le32 on VRAM tables */ + writel(addr, ((void __iomem *)ptr) + (i * 4)); return 0; } @@ -579,10 +587,8 @@ void r300_vram_info(struct radeon_device *rdev) } else { rdev->mc.vram_width = 64; } - rdev->mc.vram_size = RREG32(RADEON_CONFIG_MEMSIZE); - rdev->mc.aper_base = drm_get_resource_start(rdev->ddev, 0); - rdev->mc.aper_size = drm_get_resource_len(rdev->ddev, 0); + r100_vram_init_sizes(rdev); } @@ -970,7 +976,7 @@ static inline void r300_cs_track_clear(struct r300_cs_track *track) static const unsigned r300_reg_safe_bm[159] = { 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, - 0xFFFFFFBF, 0xFFFFFFFF, 0xFFFFFFBF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, @@ -1019,7 +1025,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, struct radeon_cs_reloc *reloc; struct r300_cs_track *track; volatile uint32_t *ib; - uint32_t tmp; + uint32_t tmp, tile_flags = 0; unsigned i; int r; @@ -1027,6 +1033,16 @@ static int r300_packet0_check(struct radeon_cs_parser *p, ib_chunk = &p->chunks[p->chunk_ib_idx]; track = (struct r300_cs_track*)p->track; switch(reg) { + case AVIVO_D1MODE_VLINE_START_END: + case RADEON_CRTC_GUI_TRIG_VLINE: + r = r100_cs_packet_parse_vline(p); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + break; case RADEON_DST_PITCH_OFFSET: case RADEON_SRC_PITCH_OFFSET: r = r100_cs_packet_next_reloc(p, &reloc); @@ -1038,7 +1054,19 @@ static int r300_packet0_check(struct radeon_cs_parser *p, } tmp = ib_chunk->kdata[idx] & 0x003fffff; tmp += (((u32)reloc->lobj.gpu_offset) >> 10); - ib[idx] = (ib_chunk->kdata[idx] & 0xffc00000) | tmp; + + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + tile_flags |= RADEON_DST_TILE_MACRO; + if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) { + if (reg == RADEON_SRC_PITCH_OFFSET) { + DRM_ERROR("Cannot src blit from microtiled surface\n"); + r100_cs_dump_packet(p, pkt); + return -EINVAL; + } + tile_flags |= RADEON_DST_TILE_MICRO; + } + tmp |= tile_flags; + ib[idx] = (ib_chunk->kdata[idx] & 0x3fc00000) | tmp; break; case R300_RB3D_COLOROFFSET0: case R300_RB3D_COLOROFFSET1: @@ -1127,6 +1155,23 @@ static int r300_packet0_check(struct radeon_cs_parser *p, /* RB3D_COLORPITCH1 */ /* RB3D_COLORPITCH2 */ /* RB3D_COLORPITCH3 */ + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + tile_flags |= R300_COLOR_TILE_ENABLE; + if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + tile_flags |= R300_COLOR_MICROTILE_ENABLE; + + tmp = ib_chunk->kdata[idx] & ~(0x7 << 16); + tmp |= tile_flags; + ib[idx] = tmp; + i = (reg - 0x4E38) >> 2; track->cb[i].pitch = ib_chunk->kdata[idx] & 0x3FFE; switch (((ib_chunk->kdata[idx] >> 21) & 0xF)) { @@ -1182,6 +1227,23 @@ static int r300_packet0_check(struct radeon_cs_parser *p, break; case 0x4F24: /* ZB_DEPTHPITCH */ + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + tile_flags |= R300_DEPTHMACROTILE_ENABLE; + if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + tile_flags |= R300_DEPTHMICROTILE_TILED;; + + tmp = ib_chunk->kdata[idx] & ~(0x7 << 16); + tmp |= tile_flags; + ib[idx] = tmp; + track->zb.pitch = ib_chunk->kdata[idx] & 0x3FFC; break; case 0x4104: diff --git a/drivers/gpu/drm/radeon/r300_reg.h b/drivers/gpu/drm/radeon/r300_reg.h index 70f48609515..4b7afef35a6 100644 --- a/drivers/gpu/drm/radeon/r300_reg.h +++ b/drivers/gpu/drm/radeon/r300_reg.h @@ -27,7 +27,9 @@ #ifndef _R300_REG_H_ #define _R300_REG_H_ - +#define R300_SURF_TILE_MACRO (1<<16) +#define R300_SURF_TILE_MICRO (2<<16) +#define R300_SURF_TILE_BOTH (3<<16) #define R300_MC_INIT_MISC_LAT_TIMER 0x180 diff --git a/drivers/gpu/drm/radeon/r500_reg.h b/drivers/gpu/drm/radeon/r500_reg.h index 9070a1c2ce2..036691b38cb 100644 --- a/drivers/gpu/drm/radeon/r500_reg.h +++ b/drivers/gpu/drm/radeon/r500_reg.h @@ -445,6 +445,7 @@ #define AVIVO_D1MODE_DATA_FORMAT 0x6528 # define AVIVO_D1MODE_INTERLEAVE_EN (1 << 0) #define AVIVO_D1MODE_DESKTOP_HEIGHT 0x652C +#define AVIVO_D1MODE_VLINE_START_END 0x6538 #define AVIVO_D1MODE_VIEWPORT_START 0x6580 #define AVIVO_D1MODE_VIEWPORT_SIZE 0x6584 #define AVIVO_D1MODE_EXT_OVERSCAN_LEFT_RIGHT 0x6588 @@ -496,6 +497,7 @@ #define AVIVO_D2CUR_SIZE 0x6c10 #define AVIVO_D2CUR_POSITION 0x6c14 +#define AVIVO_D2MODE_VLINE_START_END 0x6d38 #define AVIVO_D2MODE_VIEWPORT_START 0x6d80 #define AVIVO_D2MODE_VIEWPORT_SIZE 0x6d84 #define AVIVO_D2MODE_EXT_OVERSCAN_LEFT_RIGHT 0x6d88 diff --git a/drivers/gpu/drm/radeon/r520.c b/drivers/gpu/drm/radeon/r520.c index 570a244bd88..09fb0b6ec7d 100644 --- a/drivers/gpu/drm/radeon/r520.c +++ b/drivers/gpu/drm/radeon/r520.c @@ -28,6 +28,7 @@ #include "drmP.h" #include "radeon_reg.h" #include "radeon.h" +#include "radeon_share.h" /* r520,rv530,rv560,rv570,r580 depends on : */ void r100_hdp_reset(struct radeon_device *rdev); @@ -94,8 +95,8 @@ int r520_mc_init(struct radeon_device *rdev) "programming pipes. Bad things might happen.\n"); } /* Write VRAM size in case we are limiting it */ - WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.vram_size); - tmp = rdev->mc.vram_location + rdev->mc.vram_size - 1; + WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.real_vram_size); + tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; tmp = REG_SET(R520_MC_FB_TOP, tmp >> 16); tmp |= REG_SET(R520_MC_FB_START, rdev->mc.vram_location >> 16); WREG32_MC(R520_MC_FB_LOCATION, tmp); @@ -226,9 +227,20 @@ static void r520_vram_get_type(struct radeon_device *rdev) void r520_vram_info(struct radeon_device *rdev) { + fixed20_12 a; + r520_vram_get_type(rdev); - rdev->mc.vram_size = RREG32(RADEON_CONFIG_MEMSIZE); - rdev->mc.aper_base = drm_get_resource_start(rdev->ddev, 0); - rdev->mc.aper_size = drm_get_resource_len(rdev->ddev, 0); + r100_vram_init_sizes(rdev); + /* FIXME: we should enforce default clock in case GPU is not in + * default setup + */ + a.full = rfixed_const(100); + rdev->pm.sclk.full = rfixed_const(rdev->clock.default_sclk); + rdev->pm.sclk.full = rfixed_div(rdev->pm.sclk, a); +} + +void r520_bandwidth_update(struct radeon_device *rdev) +{ + rv515_bandwidth_avivo_update(rdev); } diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c index c45559fc97f..538cd907df6 100644 --- a/drivers/gpu/drm/radeon/r600.c +++ b/drivers/gpu/drm/radeon/r600.c @@ -67,7 +67,7 @@ int r600_mc_init(struct radeon_device *rdev) "programming pipes. Bad things might happen.\n"); } - tmp = rdev->mc.vram_location + rdev->mc.vram_size - 1; + tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; tmp = REG_SET(R600_MC_FB_TOP, tmp >> 24); tmp |= REG_SET(R600_MC_FB_BASE, rdev->mc.vram_location >> 24); WREG32(R600_MC_VM_FB_LOCATION, tmp); @@ -140,7 +140,8 @@ void r600_vram_get_type(struct radeon_device *rdev) void r600_vram_info(struct radeon_device *rdev) { r600_vram_get_type(rdev); - rdev->mc.vram_size = RREG32(R600_CONFIG_MEMSIZE); + rdev->mc.real_vram_size = RREG32(R600_CONFIG_MEMSIZE); + rdev->mc.mc_vram_size = rdev->mc.real_vram_size; /* Could aper size report 0 ? */ rdev->mc.aper_base = drm_get_resource_start(rdev->ddev, 0); diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index d61f2fc61df..b1d945b8ed6 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -64,6 +64,7 @@ extern int radeon_agpmode; extern int radeon_vram_limit; extern int radeon_gart_size; extern int radeon_benchmarking; +extern int radeon_testing; extern int radeon_connector_table; /* @@ -113,6 +114,7 @@ enum radeon_family { CHIP_RV770, CHIP_RV730, CHIP_RV710, + CHIP_RS880, CHIP_LAST, }; @@ -201,6 +203,14 @@ int radeon_fence_wait_last(struct radeon_device *rdev); struct radeon_fence *radeon_fence_ref(struct radeon_fence *fence); void radeon_fence_unref(struct radeon_fence **fence); +/* + * Tiling registers + */ +struct radeon_surface_reg { + struct radeon_object *robj; +}; + +#define RADEON_GEM_MAX_SURFACES 8 /* * Radeon buffer. @@ -213,6 +223,7 @@ struct radeon_object_list { uint64_t gpu_offset; unsigned rdomain; unsigned wdomain; + uint32_t tiling_flags; }; int radeon_object_init(struct radeon_device *rdev); @@ -242,8 +253,15 @@ void radeon_object_list_clean(struct list_head *head); int radeon_object_fbdev_mmap(struct radeon_object *robj, struct vm_area_struct *vma); unsigned long radeon_object_size(struct radeon_object *robj); - - +void radeon_object_clear_surface_reg(struct radeon_object *robj); +int radeon_object_check_tiling(struct radeon_object *robj, bool has_moved, + bool force_drop); +void radeon_object_set_tiling_flags(struct radeon_object *robj, + uint32_t tiling_flags, uint32_t pitch); +void radeon_object_get_tiling_flags(struct radeon_object *robj, uint32_t *tiling_flags, uint32_t *pitch); +void radeon_bo_move_notify(struct ttm_buffer_object *bo, + struct ttm_mem_reg *mem); +void radeon_bo_fault_reserve_notify(struct ttm_buffer_object *bo); /* * GEM objects. */ @@ -315,8 +333,11 @@ struct radeon_mc { unsigned gtt_location; unsigned gtt_size; unsigned vram_location; - unsigned vram_size; + /* for some chips with <= 32MB we need to lie + * about vram size near mc fb location */ + unsigned mc_vram_size; unsigned vram_width; + unsigned real_vram_size; int vram_mtrr; bool vram_is_ddr; }; @@ -474,6 +495,39 @@ struct radeon_wb { uint64_t gpu_addr; }; +/** + * struct radeon_pm - power management datas + * @max_bandwidth: maximum bandwidth the gpu has (MByte/s) + * @igp_sideport_mclk: sideport memory clock Mhz (rs690,rs740,rs780,rs880) + * @igp_system_mclk: system clock Mhz (rs690,rs740,rs780,rs880) + * @igp_ht_link_clk: ht link clock Mhz (rs690,rs740,rs780,rs880) + * @igp_ht_link_width: ht link width in bits (rs690,rs740,rs780,rs880) + * @k8_bandwidth: k8 bandwidth the gpu has (MByte/s) (IGP) + * @sideport_bandwidth: sideport bandwidth the gpu has (MByte/s) (IGP) + * @ht_bandwidth: ht bandwidth the gpu has (MByte/s) (IGP) + * @core_bandwidth: core GPU bandwidth the gpu has (MByte/s) (IGP) + * @sclk: GPU clock Mhz (core bandwith depends of this clock) + * @needed_bandwidth: current bandwidth needs + * + * It keeps track of various data needed to take powermanagement decision. + * Bandwith need is used to determine minimun clock of the GPU and memory. + * Equation between gpu/memory clock and available bandwidth is hw dependent + * (type of memory, bus size, efficiency, ...) + */ +struct radeon_pm { + fixed20_12 max_bandwidth; + fixed20_12 igp_sideport_mclk; + fixed20_12 igp_system_mclk; + fixed20_12 igp_ht_link_clk; + fixed20_12 igp_ht_link_width; + fixed20_12 k8_bandwidth; + fixed20_12 sideport_bandwidth; + fixed20_12 ht_bandwidth; + fixed20_12 core_bandwidth; + fixed20_12 sclk; + fixed20_12 needed_bandwidth; +}; + /* * Benchmarking @@ -482,6 +536,12 @@ void radeon_benchmark(struct radeon_device *rdev); /* + * Testing + */ +void radeon_test_moves(struct radeon_device *rdev); + + +/* * Debugfs */ int radeon_debugfs_add_files(struct radeon_device *rdev, @@ -535,6 +595,11 @@ struct radeon_asic { void (*set_memory_clock)(struct radeon_device *rdev, uint32_t mem_clock); void (*set_pcie_lanes)(struct radeon_device *rdev, int lanes); void (*set_clock_gating)(struct radeon_device *rdev, int enable); + int (*set_surface_reg)(struct radeon_device *rdev, int reg, + uint32_t tiling_flags, uint32_t pitch, + uint32_t offset, uint32_t obj_size); + int (*clear_surface_reg)(struct radeon_device *rdev, int reg); + void (*bandwidth_update)(struct radeon_device *rdev); }; union radeon_asic_config { @@ -566,6 +631,10 @@ int radeon_gem_busy_ioctl(struct drm_device *dev, void *data, int radeon_gem_wait_idle_ioctl(struct drm_device *dev, void *data, struct drm_file *filp); int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp); +int radeon_gem_set_tiling_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp); +int radeon_gem_get_tiling_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp); /* @@ -594,8 +663,8 @@ struct radeon_device { struct radeon_object *fbdev_robj; struct radeon_framebuffer *fbdev_rfb; /* Register mmio */ - unsigned long rmmio_base; - unsigned long rmmio_size; + resource_size_t rmmio_base; + resource_size_t rmmio_size; void *rmmio; radeon_rreg_t mm_rreg; radeon_wreg_t mm_wreg; @@ -619,11 +688,14 @@ struct radeon_device { struct radeon_irq irq; struct radeon_asic *asic; struct radeon_gem gem; + struct radeon_pm pm; struct mutex cs_mutex; struct radeon_wb wb; bool gpu_lockup; bool shutdown; bool suspend; + bool need_dma32; + struct radeon_surface_reg surface_regs[RADEON_GEM_MAX_SURFACES]; }; int radeon_device_init(struct radeon_device *rdev, @@ -670,6 +742,8 @@ void r100_pll_errata_after_index(struct radeon_device *rdev); /* * ASICs helpers. */ +#define ASIC_IS_RN50(rdev) ((rdev->pdev->device == 0x515e) || \ + (rdev->pdev->device == 0x5969)) #define ASIC_IS_RV100(rdev) ((rdev->family == CHIP_RV100) || \ (rdev->family == CHIP_RV200) || \ (rdev->family == CHIP_RS100) || \ @@ -796,5 +870,8 @@ static inline void radeon_ring_write(struct radeon_device *rdev, uint32_t v) #define radeon_set_memory_clock(rdev, e) (rdev)->asic->set_engine_clock((rdev), (e)) #define radeon_set_pcie_lanes(rdev, l) (rdev)->asic->set_pcie_lanes((rdev), (l)) #define radeon_set_clock_gating(rdev, e) (rdev)->asic->set_clock_gating((rdev), (e)) +#define radeon_set_surface_reg(rdev, r, f, p, o, s) ((rdev)->asic->set_surface_reg((rdev), (r), (f), (p), (o), (s))) +#define radeon_clear_surface_reg(rdev, r) ((rdev)->asic->clear_surface_reg((rdev), (r))) +#define radeon_bandwidth_update(rdev) (rdev)->asic->bandwidth_update((rdev)) #endif diff --git a/drivers/gpu/drm/radeon/radeon_asic.h b/drivers/gpu/drm/radeon/radeon_asic.h index e2e567395df..9a75876e0c3 100644 --- a/drivers/gpu/drm/radeon/radeon_asic.h +++ b/drivers/gpu/drm/radeon/radeon_asic.h @@ -71,6 +71,11 @@ int r100_copy_blit(struct radeon_device *rdev, uint64_t dst_offset, unsigned num_pages, struct radeon_fence *fence); +int r100_set_surface_reg(struct radeon_device *rdev, int reg, + uint32_t tiling_flags, uint32_t pitch, + uint32_t offset, uint32_t obj_size); +int r100_clear_surface_reg(struct radeon_device *rdev, int reg); +void r100_bandwidth_update(struct radeon_device *rdev); static struct radeon_asic r100_asic = { .init = &r100_init, @@ -100,6 +105,9 @@ static struct radeon_asic r100_asic = { .set_memory_clock = NULL, .set_pcie_lanes = NULL, .set_clock_gating = &radeon_legacy_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &r100_bandwidth_update, }; @@ -128,6 +136,7 @@ int r300_copy_dma(struct radeon_device *rdev, uint64_t dst_offset, unsigned num_pages, struct radeon_fence *fence); + static struct radeon_asic r300_asic = { .init = &r300_init, .errata = &r300_errata, @@ -156,6 +165,9 @@ static struct radeon_asic r300_asic = { .set_memory_clock = NULL, .set_pcie_lanes = &rv370_set_pcie_lanes, .set_clock_gating = &radeon_legacy_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &r100_bandwidth_update, }; /* @@ -193,6 +205,9 @@ static struct radeon_asic r420_asic = { .set_memory_clock = &radeon_atom_set_memory_clock, .set_pcie_lanes = &rv370_set_pcie_lanes, .set_clock_gating = &radeon_atom_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &r100_bandwidth_update, }; @@ -237,6 +252,9 @@ static struct radeon_asic rs400_asic = { .set_memory_clock = NULL, .set_pcie_lanes = NULL, .set_clock_gating = &radeon_legacy_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &r100_bandwidth_update, }; @@ -254,6 +272,7 @@ void rs600_gart_tlb_flush(struct radeon_device *rdev); int rs600_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr); uint32_t rs600_mc_rreg(struct radeon_device *rdev, uint32_t reg); void rs600_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); +void rs600_bandwidth_update(struct radeon_device *rdev); static struct radeon_asic rs600_asic = { .init = &r300_init, .errata = &rs600_errata, @@ -282,6 +301,7 @@ static struct radeon_asic rs600_asic = { .set_memory_clock = &radeon_atom_set_memory_clock, .set_pcie_lanes = NULL, .set_clock_gating = &radeon_atom_set_clock_gating, + .bandwidth_update = &rs600_bandwidth_update, }; @@ -294,6 +314,7 @@ int rs690_mc_init(struct radeon_device *rdev); void rs690_mc_fini(struct radeon_device *rdev); uint32_t rs690_mc_rreg(struct radeon_device *rdev, uint32_t reg); void rs690_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); +void rs690_bandwidth_update(struct radeon_device *rdev); static struct radeon_asic rs690_asic = { .init = &r300_init, .errata = &rs690_errata, @@ -322,6 +343,9 @@ static struct radeon_asic rs690_asic = { .set_memory_clock = &radeon_atom_set_memory_clock, .set_pcie_lanes = NULL, .set_clock_gating = &radeon_atom_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &rs690_bandwidth_update, }; @@ -339,6 +363,7 @@ void rv515_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); void rv515_ring_start(struct radeon_device *rdev); uint32_t rv515_pcie_rreg(struct radeon_device *rdev, uint32_t reg); void rv515_pcie_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); +void rv515_bandwidth_update(struct radeon_device *rdev); static struct radeon_asic rv515_asic = { .init = &rv515_init, .errata = &rv515_errata, @@ -367,6 +392,9 @@ static struct radeon_asic rv515_asic = { .set_memory_clock = &radeon_atom_set_memory_clock, .set_pcie_lanes = &rv370_set_pcie_lanes, .set_clock_gating = &radeon_atom_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &rv515_bandwidth_update, }; @@ -377,6 +405,7 @@ void r520_errata(struct radeon_device *rdev); void r520_vram_info(struct radeon_device *rdev); int r520_mc_init(struct radeon_device *rdev); void r520_mc_fini(struct radeon_device *rdev); +void r520_bandwidth_update(struct radeon_device *rdev); static struct radeon_asic r520_asic = { .init = &rv515_init, .errata = &r520_errata, @@ -405,6 +434,9 @@ static struct radeon_asic r520_asic = { .set_memory_clock = &radeon_atom_set_memory_clock, .set_pcie_lanes = &rv370_set_pcie_lanes, .set_clock_gating = &radeon_atom_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &r520_bandwidth_update, }; /* diff --git a/drivers/gpu/drm/radeon/radeon_atombios.c b/drivers/gpu/drm/radeon/radeon_atombios.c index 1f5a1a49098..fcfe5c02d74 100644 --- a/drivers/gpu/drm/radeon/radeon_atombios.c +++ b/drivers/gpu/drm/radeon/radeon_atombios.c @@ -103,7 +103,8 @@ static inline struct radeon_i2c_bus_rec radeon_lookup_gpio(struct drm_device static bool radeon_atom_apply_quirks(struct drm_device *dev, uint32_t supported_device, int *connector_type, - struct radeon_i2c_bus_rec *i2c_bus) + struct radeon_i2c_bus_rec *i2c_bus, + uint8_t *line_mux) { /* Asus M2A-VM HDMI board lists the DVI port as HDMI */ @@ -127,8 +128,10 @@ static bool radeon_atom_apply_quirks(struct drm_device *dev, if ((dev->pdev->device == 0x5653) && (dev->pdev->subsystem_vendor == 0x1462) && (dev->pdev->subsystem_device == 0x0291)) { - if (*connector_type == DRM_MODE_CONNECTOR_LVDS) + if (*connector_type == DRM_MODE_CONNECTOR_LVDS) { i2c_bus->valid = false; + *line_mux = 53; + } } /* Funky macbooks */ @@ -526,7 +529,7 @@ bool radeon_get_atom_connector_info_from_supported_devices_table(struct if (!radeon_atom_apply_quirks (dev, (1 << i), &bios_connectors[i].connector_type, - &bios_connectors[i].ddc_bus)) + &bios_connectors[i].ddc_bus, &bios_connectors[i].line_mux)) continue; bios_connectors[i].valid = true; diff --git a/drivers/gpu/drm/radeon/radeon_benchmark.c b/drivers/gpu/drm/radeon/radeon_benchmark.c index c44403a2ca7..2e938f7496f 100644 --- a/drivers/gpu/drm/radeon/radeon_benchmark.c +++ b/drivers/gpu/drm/radeon/radeon_benchmark.c @@ -63,7 +63,7 @@ void radeon_benchmark_move(struct radeon_device *rdev, unsigned bsize, if (r) { goto out_cleanup; } - r = radeon_copy_dma(rdev, saddr, daddr, size >> 14, fence); + r = radeon_copy_dma(rdev, saddr, daddr, size / 4096, fence); if (r) { goto out_cleanup; } @@ -88,7 +88,7 @@ void radeon_benchmark_move(struct radeon_device *rdev, unsigned bsize, if (r) { goto out_cleanup; } - r = radeon_copy_blit(rdev, saddr, daddr, size >> 14, fence); + r = radeon_copy_blit(rdev, saddr, daddr, size / 4096, fence); if (r) { goto out_cleanup; } diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c index b843f9bdfb1..a169067efc4 100644 --- a/drivers/gpu/drm/radeon/radeon_cs.c +++ b/drivers/gpu/drm/radeon/radeon_cs.c @@ -127,17 +127,23 @@ int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data) sizeof(struct drm_radeon_cs_chunk))) { return -EFAULT; } + p->chunks[i].length_dw = user_chunk.length_dw; + p->chunks[i].kdata = NULL; p->chunks[i].chunk_id = user_chunk.chunk_id; + if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_RELOCS) { p->chunk_relocs_idx = i; } if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_IB) { p->chunk_ib_idx = i; + /* zero length IB isn't useful */ + if (p->chunks[i].length_dw == 0) + return -EINVAL; } + p->chunks[i].length_dw = user_chunk.length_dw; cdata = (uint32_t *)(unsigned long)user_chunk.chunk_data; - p->chunks[i].kdata = NULL; size = p->chunks[i].length_dw * sizeof(uint32_t); p->chunks[i].kdata = kzalloc(size, GFP_KERNEL); if (p->chunks[i].kdata == NULL) { diff --git a/drivers/gpu/drm/radeon/radeon_cursor.c b/drivers/gpu/drm/radeon/radeon_cursor.c index 5232441f119..b13c79e38bc 100644 --- a/drivers/gpu/drm/radeon/radeon_cursor.c +++ b/drivers/gpu/drm/radeon/radeon_cursor.c @@ -111,9 +111,11 @@ static void radeon_set_cursor(struct drm_crtc *crtc, struct drm_gem_object *obj, if (ASIC_IS_AVIVO(rdev)) WREG32(AVIVO_D1CUR_SURFACE_ADDRESS + radeon_crtc->crtc_offset, gpu_addr); - else + else { + radeon_crtc->legacy_cursor_offset = gpu_addr - radeon_crtc->legacy_display_base_addr; /* offset is from DISP(2)_BASE_ADDRESS */ - WREG32(RADEON_CUR_OFFSET + radeon_crtc->crtc_offset, gpu_addr); + WREG32(RADEON_CUR_OFFSET + radeon_crtc->crtc_offset, radeon_crtc->legacy_cursor_offset); + } } int radeon_crtc_cursor_set(struct drm_crtc *crtc, @@ -245,6 +247,9 @@ int radeon_crtc_cursor_move(struct drm_crtc *crtc, (RADEON_CUR_LOCK | ((xorigin ? 0 : x) << 16) | (yorigin ? 0 : y))); + /* offset is from DISP(2)_BASE_ADDRESS */ + WREG32(RADEON_CUR_OFFSET + radeon_crtc->crtc_offset, (radeon_crtc->legacy_cursor_offset + + (yorigin * 256))); } radeon_lock_cursor(crtc, false); diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c index f97563db4e5..a162ade74b7 100644 --- a/drivers/gpu/drm/radeon/radeon_device.c +++ b/drivers/gpu/drm/radeon/radeon_device.c @@ -48,6 +48,8 @@ static void radeon_surface_init(struct radeon_device *rdev) i * (RADEON_SURFACE1_INFO - RADEON_SURFACE0_INFO), 0); } + /* enable surfaces */ + WREG32(RADEON_SURFACE_CNTL, 0); } } @@ -119,7 +121,7 @@ int radeon_mc_setup(struct radeon_device *rdev) if (rdev->mc.vram_location != 0xFFFFFFFFUL) { /* vram location was already setup try to put gtt after * if it fits */ - tmp = rdev->mc.vram_location + rdev->mc.vram_size; + tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size; tmp = (tmp + rdev->mc.gtt_size - 1) & ~(rdev->mc.gtt_size - 1); if ((0xFFFFFFFFUL - tmp) >= rdev->mc.gtt_size) { rdev->mc.gtt_location = tmp; @@ -134,13 +136,13 @@ int radeon_mc_setup(struct radeon_device *rdev) } else if (rdev->mc.gtt_location != 0xFFFFFFFFUL) { /* gtt location was already setup try to put vram before * if it fits */ - if (rdev->mc.vram_size < rdev->mc.gtt_location) { + if (rdev->mc.mc_vram_size < rdev->mc.gtt_location) { rdev->mc.vram_location = 0; } else { tmp = rdev->mc.gtt_location + rdev->mc.gtt_size; - tmp += (rdev->mc.vram_size - 1); - tmp &= ~(rdev->mc.vram_size - 1); - if ((0xFFFFFFFFUL - tmp) >= rdev->mc.vram_size) { + tmp += (rdev->mc.mc_vram_size - 1); + tmp &= ~(rdev->mc.mc_vram_size - 1); + if ((0xFFFFFFFFUL - tmp) >= rdev->mc.mc_vram_size) { rdev->mc.vram_location = tmp; } else { printk(KERN_ERR "[drm] vram too big to fit " @@ -150,12 +152,14 @@ int radeon_mc_setup(struct radeon_device *rdev) } } else { rdev->mc.vram_location = 0; - rdev->mc.gtt_location = rdev->mc.vram_size; + rdev->mc.gtt_location = rdev->mc.mc_vram_size; } - DRM_INFO("radeon: VRAM %uM\n", rdev->mc.vram_size >> 20); + DRM_INFO("radeon: VRAM %uM\n", rdev->mc.real_vram_size >> 20); DRM_INFO("radeon: VRAM from 0x%08X to 0x%08X\n", rdev->mc.vram_location, - rdev->mc.vram_location + rdev->mc.vram_size - 1); + rdev->mc.vram_location + rdev->mc.mc_vram_size - 1); + if (rdev->mc.real_vram_size != rdev->mc.mc_vram_size) + DRM_INFO("radeon: VRAM less than aperture workaround enabled\n"); DRM_INFO("radeon: GTT %uM\n", rdev->mc.gtt_size >> 20); DRM_INFO("radeon: GTT from 0x%08X to 0x%08X\n", rdev->mc.gtt_location, @@ -450,6 +454,7 @@ int radeon_device_init(struct radeon_device *rdev, uint32_t flags) { int r, ret; + int dma_bits; DRM_INFO("radeon: Initializing kernel modesetting.\n"); rdev->shutdown = false; @@ -492,8 +497,20 @@ int radeon_device_init(struct radeon_device *rdev, return r; } - /* Report DMA addressing limitation */ - r = pci_set_dma_mask(rdev->pdev, DMA_BIT_MASK(32)); + /* set DMA mask + need_dma32 flags. + * PCIE - can handle 40-bits. + * IGP - can handle 40-bits (in theory) + * AGP - generally dma32 is safest + * PCI - only dma32 + */ + rdev->need_dma32 = false; + if (rdev->flags & RADEON_IS_AGP) + rdev->need_dma32 = true; + if (rdev->flags & RADEON_IS_PCI) + rdev->need_dma32 = true; + + dma_bits = rdev->need_dma32 ? 32 : 40; + r = pci_set_dma_mask(rdev->pdev, DMA_BIT_MASK(dma_bits)); if (r) { printk(KERN_WARNING "radeon: No suitable DMA available.\n"); } @@ -546,27 +563,22 @@ int radeon_device_init(struct radeon_device *rdev, radeon_combios_asic_init(rdev->ddev); } } + /* Initialize clocks */ + r = radeon_clocks_init(rdev); + if (r) { + return r; + } /* Get vram informations */ radeon_vram_info(rdev); - /* Device is severly broken if aper size > vram size. - * for RN50/M6/M7 - Novell bug 204882 ? - */ - if (rdev->mc.vram_size < rdev->mc.aper_size) { - rdev->mc.aper_size = rdev->mc.vram_size; - } + /* Add an MTRR for the VRAM */ rdev->mc.vram_mtrr = mtrr_add(rdev->mc.aper_base, rdev->mc.aper_size, MTRR_TYPE_WRCOMB, 1); DRM_INFO("Detected VRAM RAM=%uM, BAR=%uM\n", - rdev->mc.vram_size >> 20, + rdev->mc.real_vram_size >> 20, (unsigned)rdev->mc.aper_size >> 20); DRM_INFO("RAM width %dbits %cDR\n", rdev->mc.vram_width, rdev->mc.vram_is_ddr ? 'D' : 'S'); - /* Initialize clocks */ - r = radeon_clocks_init(rdev); - if (r) { - return r; - } /* Initialize memory controller (also test AGP) */ r = radeon_mc_init(rdev); if (r) { @@ -626,6 +638,9 @@ int radeon_device_init(struct radeon_device *rdev, if (!ret) { DRM_INFO("radeon: kernel modesetting successfully initialized.\n"); } + if (radeon_testing) { + radeon_test_moves(rdev); + } if (radeon_benchmarking) { radeon_benchmark(rdev); } diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c index 3efcf1a526b..a8fa1bb84cf 100644 --- a/drivers/gpu/drm/radeon/radeon_display.c +++ b/drivers/gpu/drm/radeon/radeon_display.c @@ -187,6 +187,7 @@ static void radeon_crtc_init(struct drm_device *dev, int index) drm_mode_crtc_set_gamma_size(&radeon_crtc->base, 256); radeon_crtc->crtc_id = index; + rdev->mode_info.crtcs[index] = radeon_crtc; radeon_crtc->mode_set.crtc = &radeon_crtc->base; radeon_crtc->mode_set.connectors = (struct drm_connector **)(radeon_crtc + 1); @@ -491,7 +492,11 @@ void radeon_compute_pll(struct radeon_pll *pll, tmp += (uint64_t)pll->reference_freq * 1000 * frac_feedback_div; current_freq = radeon_div(tmp, ref_div * post_div); - error = abs(current_freq - freq); + if (flags & RADEON_PLL_PREFER_CLOSEST_LOWER) { + error = freq - current_freq; + error = error < 0 ? 0xffffffff : error; + } else + error = abs(current_freq - freq); vco_diff = abs(vco - best_vco); if ((best_vco == 0 && error < best_error) || @@ -657,36 +662,51 @@ void radeon_modeset_fini(struct radeon_device *rdev) } } -void radeon_init_disp_bandwidth(struct drm_device *dev) +bool radeon_crtc_scaling_mode_fixup(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) { - struct radeon_device *rdev = dev->dev_private; - struct drm_display_mode *modes[2]; - int pixel_bytes[2]; - struct drm_crtc *crtc; - - pixel_bytes[0] = pixel_bytes[1] = 0; - modes[0] = modes[1] = NULL; - - list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { - struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct drm_encoder *encoder; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct radeon_encoder *radeon_encoder; + bool first = true; - if (crtc->enabled && crtc->fb) { - modes[radeon_crtc->crtc_id] = &crtc->mode; - pixel_bytes[radeon_crtc->crtc_id] = crtc->fb->bits_per_pixel / 8; + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + radeon_encoder = to_radeon_encoder(encoder); + if (encoder->crtc != crtc) + continue; + if (first) { + radeon_crtc->rmx_type = radeon_encoder->rmx_type; + radeon_crtc->devices = radeon_encoder->devices; + memcpy(&radeon_crtc->native_mode, + &radeon_encoder->native_mode, + sizeof(struct radeon_native_mode)); + first = false; + } else { + if (radeon_crtc->rmx_type != radeon_encoder->rmx_type) { + /* WARNING: Right now this can't happen but + * in the future we need to check that scaling + * are consistent accross different encoder + * (ie all encoder can work with the same + * scaling). + */ + DRM_ERROR("Scaling not consistent accross encoder.\n"); + return false; + } } } - - if (ASIC_IS_AVIVO(rdev)) { - radeon_init_disp_bw_avivo(dev, - modes[0], - pixel_bytes[0], - modes[1], - pixel_bytes[1]); + if (radeon_crtc->rmx_type != RMX_OFF) { + fixed20_12 a, b; + a.full = rfixed_const(crtc->mode.vdisplay); + b.full = rfixed_const(radeon_crtc->native_mode.panel_xres); + radeon_crtc->vsc.full = rfixed_div(a, b); + a.full = rfixed_const(crtc->mode.hdisplay); + b.full = rfixed_const(radeon_crtc->native_mode.panel_yres); + radeon_crtc->hsc.full = rfixed_div(a, b); } else { - radeon_init_disp_bw_legacy(dev, - modes[0], - pixel_bytes[0], - modes[1], - pixel_bytes[1]); + radeon_crtc->vsc.full = rfixed_const(1); + radeon_crtc->hsc.full = rfixed_const(1); } + return true; } diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c index 84ba69f4878..3cfcee17dc5 100644 --- a/drivers/gpu/drm/radeon/radeon_drv.c +++ b/drivers/gpu/drm/radeon/radeon_drv.c @@ -89,6 +89,7 @@ int radeon_agpmode = 0; int radeon_vram_limit = 0; int radeon_gart_size = 512; /* default gart size */ int radeon_benchmarking = 0; +int radeon_testing = 0; int radeon_connector_table = 0; #endif @@ -117,6 +118,9 @@ module_param_named(gartsize, radeon_gart_size, int, 0600); MODULE_PARM_DESC(benchmark, "Run benchmark"); module_param_named(benchmark, radeon_benchmarking, int, 0444); +MODULE_PARM_DESC(test, "Run tests"); +module_param_named(test, radeon_testing, int, 0444); + MODULE_PARM_DESC(connector_table, "Force connector table"); module_param_named(connector_table, radeon_connector_table, int, 0444); #endif diff --git a/drivers/gpu/drm/radeon/radeon_encoders.c b/drivers/gpu/drm/radeon/radeon_encoders.c index c8ef0d14ffa..0a92706eac1 100644 --- a/drivers/gpu/drm/radeon/radeon_encoders.c +++ b/drivers/gpu/drm/radeon/radeon_encoders.c @@ -154,7 +154,6 @@ void radeon_rmx_mode_fixup(struct drm_encoder *encoder, if (mode->hdisplay < native_mode->panel_xres || mode->vdisplay < native_mode->panel_yres) { - radeon_encoder->flags |= RADEON_USE_RMX; if (ASIC_IS_AVIVO(rdev)) { adjusted_mode->hdisplay = native_mode->panel_xres; adjusted_mode->vdisplay = native_mode->panel_yres; @@ -197,15 +196,13 @@ void radeon_rmx_mode_fixup(struct drm_encoder *encoder, } } + static bool radeon_atom_mode_fixup(struct drm_encoder *encoder, struct drm_display_mode *mode, struct drm_display_mode *adjusted_mode) { - struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); - radeon_encoder->flags &= ~RADEON_USE_RMX; - drm_mode_set_crtcinfo(adjusted_mode, 0); if (radeon_encoder->rmx_type != RMX_OFF) @@ -808,234 +805,6 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action) } -static void atom_rv515_force_tv_scaler(struct radeon_device *rdev) -{ - - WREG32(0x659C, 0x0); - WREG32(0x6594, 0x705); - WREG32(0x65A4, 0x10001); - WREG32(0x65D8, 0x0); - WREG32(0x65B0, 0x0); - WREG32(0x65C0, 0x0); - WREG32(0x65D4, 0x0); - WREG32(0x6578, 0x0); - WREG32(0x657C, 0x841880A8); - WREG32(0x6578, 0x1); - WREG32(0x657C, 0x84208680); - WREG32(0x6578, 0x2); - WREG32(0x657C, 0xBFF880B0); - WREG32(0x6578, 0x100); - WREG32(0x657C, 0x83D88088); - WREG32(0x6578, 0x101); - WREG32(0x657C, 0x84608680); - WREG32(0x6578, 0x102); - WREG32(0x657C, 0xBFF080D0); - WREG32(0x6578, 0x200); - WREG32(0x657C, 0x83988068); - WREG32(0x6578, 0x201); - WREG32(0x657C, 0x84A08680); - WREG32(0x6578, 0x202); - WREG32(0x657C, 0xBFF080F8); - WREG32(0x6578, 0x300); - WREG32(0x657C, 0x83588058); - WREG32(0x6578, 0x301); - WREG32(0x657C, 0x84E08660); - WREG32(0x6578, 0x302); - WREG32(0x657C, 0xBFF88120); - WREG32(0x6578, 0x400); - WREG32(0x657C, 0x83188040); - WREG32(0x6578, 0x401); - WREG32(0x657C, 0x85008660); - WREG32(0x6578, 0x402); - WREG32(0x657C, 0xBFF88150); - WREG32(0x6578, 0x500); - WREG32(0x657C, 0x82D88030); - WREG32(0x6578, 0x501); - WREG32(0x657C, 0x85408640); - WREG32(0x6578, 0x502); - WREG32(0x657C, 0xBFF88180); - WREG32(0x6578, 0x600); - WREG32(0x657C, 0x82A08018); - WREG32(0x6578, 0x601); - WREG32(0x657C, 0x85808620); - WREG32(0x6578, 0x602); - WREG32(0x657C, 0xBFF081B8); - WREG32(0x6578, 0x700); - WREG32(0x657C, 0x82608010); - WREG32(0x6578, 0x701); - WREG32(0x657C, 0x85A08600); - WREG32(0x6578, 0x702); - WREG32(0x657C, 0x800081F0); - WREG32(0x6578, 0x800); - WREG32(0x657C, 0x8228BFF8); - WREG32(0x6578, 0x801); - WREG32(0x657C, 0x85E085E0); - WREG32(0x6578, 0x802); - WREG32(0x657C, 0xBFF88228); - WREG32(0x6578, 0x10000); - WREG32(0x657C, 0x82A8BF00); - WREG32(0x6578, 0x10001); - WREG32(0x657C, 0x82A08CC0); - WREG32(0x6578, 0x10002); - WREG32(0x657C, 0x8008BEF8); - WREG32(0x6578, 0x10100); - WREG32(0x657C, 0x81F0BF28); - WREG32(0x6578, 0x10101); - WREG32(0x657C, 0x83608CA0); - WREG32(0x6578, 0x10102); - WREG32(0x657C, 0x8018BED0); - WREG32(0x6578, 0x10200); - WREG32(0x657C, 0x8148BF38); - WREG32(0x6578, 0x10201); - WREG32(0x657C, 0x84408C80); - WREG32(0x6578, 0x10202); - WREG32(0x657C, 0x8008BEB8); - WREG32(0x6578, 0x10300); - WREG32(0x657C, 0x80B0BF78); - WREG32(0x6578, 0x10301); - WREG32(0x657C, 0x85008C20); - WREG32(0x6578, 0x10302); - WREG32(0x657C, 0x8020BEA0); - WREG32(0x6578, 0x10400); - WREG32(0x657C, 0x8028BF90); - WREG32(0x6578, 0x10401); - WREG32(0x657C, 0x85E08BC0); - WREG32(0x6578, 0x10402); - WREG32(0x657C, 0x8018BE90); - WREG32(0x6578, 0x10500); - WREG32(0x657C, 0xBFB8BFB0); - WREG32(0x6578, 0x10501); - WREG32(0x657C, 0x86C08B40); - WREG32(0x6578, 0x10502); - WREG32(0x657C, 0x8010BE90); - WREG32(0x6578, 0x10600); - WREG32(0x657C, 0xBF58BFC8); - WREG32(0x6578, 0x10601); - WREG32(0x657C, 0x87A08AA0); - WREG32(0x6578, 0x10602); - WREG32(0x657C, 0x8010BE98); - WREG32(0x6578, 0x10700); - WREG32(0x657C, 0xBF10BFF0); - WREG32(0x6578, 0x10701); - WREG32(0x657C, 0x886089E0); - WREG32(0x6578, 0x10702); - WREG32(0x657C, 0x8018BEB0); - WREG32(0x6578, 0x10800); - WREG32(0x657C, 0xBED8BFE8); - WREG32(0x6578, 0x10801); - WREG32(0x657C, 0x89408940); - WREG32(0x6578, 0x10802); - WREG32(0x657C, 0xBFE8BED8); - WREG32(0x6578, 0x20000); - WREG32(0x657C, 0x80008000); - WREG32(0x6578, 0x20001); - WREG32(0x657C, 0x90008000); - WREG32(0x6578, 0x20002); - WREG32(0x657C, 0x80008000); - WREG32(0x6578, 0x20003); - WREG32(0x657C, 0x80008000); - WREG32(0x6578, 0x20100); - WREG32(0x657C, 0x80108000); - WREG32(0x6578, 0x20101); - WREG32(0x657C, 0x8FE0BF70); - WREG32(0x6578, 0x20102); - WREG32(0x657C, 0xBFE880C0); - WREG32(0x6578, 0x20103); - WREG32(0x657C, 0x80008000); - WREG32(0x6578, 0x20200); - WREG32(0x657C, 0x8018BFF8); - WREG32(0x6578, 0x20201); - WREG32(0x657C, 0x8F80BF08); - WREG32(0x6578, 0x20202); - WREG32(0x657C, 0xBFD081A0); - WREG32(0x6578, 0x20203); - WREG32(0x657C, 0xBFF88000); - WREG32(0x6578, 0x20300); - WREG32(0x657C, 0x80188000); - WREG32(0x6578, 0x20301); - WREG32(0x657C, 0x8EE0BEC0); - WREG32(0x6578, 0x20302); - WREG32(0x657C, 0xBFB082A0); - WREG32(0x6578, 0x20303); - WREG32(0x657C, 0x80008000); - WREG32(0x6578, 0x20400); - WREG32(0x657C, 0x80188000); - WREG32(0x6578, 0x20401); - WREG32(0x657C, 0x8E00BEA0); - WREG32(0x6578, 0x20402); - WREG32(0x657C, 0xBF8883C0); - WREG32(0x6578, 0x20403); - WREG32(0x657C, 0x80008000); - WREG32(0x6578, 0x20500); - WREG32(0x657C, 0x80188000); - WREG32(0x6578, 0x20501); - WREG32(0x657C, 0x8D00BE90); - WREG32(0x6578, 0x20502); - WREG32(0x657C, 0xBF588500); - WREG32(0x6578, 0x20503); - WREG32(0x657C, 0x80008008); - WREG32(0x6578, 0x20600); - WREG32(0x657C, 0x80188000); - WREG32(0x6578, 0x20601); - WREG32(0x657C, 0x8BC0BE98); - WREG32(0x6578, 0x20602); - WREG32(0x657C, 0xBF308660); - WREG32(0x6578, 0x20603); - WREG32(0x657C, 0x80008008); - WREG32(0x6578, 0x20700); - WREG32(0x657C, 0x80108000); - WREG32(0x6578, 0x20701); - WREG32(0x657C, 0x8A80BEB0); - WREG32(0x6578, 0x20702); - WREG32(0x657C, 0xBF0087C0); - WREG32(0x6578, 0x20703); - WREG32(0x657C, 0x80008008); - WREG32(0x6578, 0x20800); - WREG32(0x657C, 0x80108000); - WREG32(0x6578, 0x20801); - WREG32(0x657C, 0x8920BED0); - WREG32(0x6578, 0x20802); - WREG32(0x657C, 0xBED08920); - WREG32(0x6578, 0x20803); - WREG32(0x657C, 0x80008010); - WREG32(0x6578, 0x30000); - WREG32(0x657C, 0x90008000); - WREG32(0x6578, 0x30001); - WREG32(0x657C, 0x80008000); - WREG32(0x6578, 0x30100); - WREG32(0x657C, 0x8FE0BF90); - WREG32(0x6578, 0x30101); - WREG32(0x657C, 0xBFF880A0); - WREG32(0x6578, 0x30200); - WREG32(0x657C, 0x8F60BF40); - WREG32(0x6578, 0x30201); - WREG32(0x657C, 0xBFE88180); - WREG32(0x6578, 0x30300); - WREG32(0x657C, 0x8EC0BF00); - WREG32(0x6578, 0x30301); - WREG32(0x657C, 0xBFC88280); - WREG32(0x6578, 0x30400); - WREG32(0x657C, 0x8DE0BEE0); - WREG32(0x6578, 0x30401); - WREG32(0x657C, 0xBFA083A0); - WREG32(0x6578, 0x30500); - WREG32(0x657C, 0x8CE0BED0); - WREG32(0x6578, 0x30501); - WREG32(0x657C, 0xBF7884E0); - WREG32(0x6578, 0x30600); - WREG32(0x657C, 0x8BA0BED8); - WREG32(0x6578, 0x30601); - WREG32(0x657C, 0xBF508640); - WREG32(0x6578, 0x30700); - WREG32(0x657C, 0x8A60BEE8); - WREG32(0x6578, 0x30701); - WREG32(0x657C, 0xBF2087A0); - WREG32(0x6578, 0x30800); - WREG32(0x657C, 0x8900BF00); - WREG32(0x6578, 0x30801); - WREG32(0x657C, 0xBF008900); -} - static void atombios_yuv_setup(struct drm_encoder *encoder, bool enable) { @@ -1074,129 +843,6 @@ atombios_yuv_setup(struct drm_encoder *encoder, bool enable) } static void -atombios_overscan_setup(struct drm_encoder *encoder, - struct drm_display_mode *mode, - struct drm_display_mode *adjusted_mode) -{ - struct drm_device *dev = encoder->dev; - struct radeon_device *rdev = dev->dev_private; - struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); - struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); - SET_CRTC_OVERSCAN_PS_ALLOCATION args; - int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_OverScan); - - memset(&args, 0, sizeof(args)); - - args.usOverscanRight = 0; - args.usOverscanLeft = 0; - args.usOverscanBottom = 0; - args.usOverscanTop = 0; - args.ucCRTC = radeon_crtc->crtc_id; - - if (radeon_encoder->flags & RADEON_USE_RMX) { - if (radeon_encoder->rmx_type == RMX_FULL) { - args.usOverscanRight = 0; - args.usOverscanLeft = 0; - args.usOverscanBottom = 0; - args.usOverscanTop = 0; - } else if (radeon_encoder->rmx_type == RMX_CENTER) { - args.usOverscanTop = (adjusted_mode->crtc_vdisplay - mode->crtc_vdisplay) / 2; - args.usOverscanBottom = (adjusted_mode->crtc_vdisplay - mode->crtc_vdisplay) / 2; - args.usOverscanLeft = (adjusted_mode->crtc_hdisplay - mode->crtc_hdisplay) / 2; - args.usOverscanRight = (adjusted_mode->crtc_hdisplay - mode->crtc_hdisplay) / 2; - } else if (radeon_encoder->rmx_type == RMX_ASPECT) { - int a1 = mode->crtc_vdisplay * adjusted_mode->crtc_hdisplay; - int a2 = adjusted_mode->crtc_vdisplay * mode->crtc_hdisplay; - - if (a1 > a2) { - args.usOverscanLeft = (adjusted_mode->crtc_hdisplay - (a2 / mode->crtc_vdisplay)) / 2; - args.usOverscanRight = (adjusted_mode->crtc_hdisplay - (a2 / mode->crtc_vdisplay)) / 2; - } else if (a2 > a1) { - args.usOverscanLeft = (adjusted_mode->crtc_vdisplay - (a1 / mode->crtc_hdisplay)) / 2; - args.usOverscanRight = (adjusted_mode->crtc_vdisplay - (a1 / mode->crtc_hdisplay)) / 2; - } - } - } - - atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); - -} - -static void -atombios_scaler_setup(struct drm_encoder *encoder) -{ - struct drm_device *dev = encoder->dev; - struct radeon_device *rdev = dev->dev_private; - struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); - struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); - ENABLE_SCALER_PS_ALLOCATION args; - int index = GetIndexIntoMasterTable(COMMAND, EnableScaler); - /* fixme - fill in enc_priv for atom dac */ - enum radeon_tv_std tv_std = TV_STD_NTSC; - - if (!ASIC_IS_AVIVO(rdev) && radeon_crtc->crtc_id) - return; - - memset(&args, 0, sizeof(args)); - - args.ucScaler = radeon_crtc->crtc_id; - - if (radeon_encoder->devices & (ATOM_DEVICE_TV_SUPPORT)) { - switch (tv_std) { - case TV_STD_NTSC: - default: - args.ucTVStandard = ATOM_TV_NTSC; - break; - case TV_STD_PAL: - args.ucTVStandard = ATOM_TV_PAL; - break; - case TV_STD_PAL_M: - args.ucTVStandard = ATOM_TV_PALM; - break; - case TV_STD_PAL_60: - args.ucTVStandard = ATOM_TV_PAL60; - break; - case TV_STD_NTSC_J: - args.ucTVStandard = ATOM_TV_NTSCJ; - break; - case TV_STD_SCART_PAL: - args.ucTVStandard = ATOM_TV_PAL; /* ??? */ - break; - case TV_STD_SECAM: - args.ucTVStandard = ATOM_TV_SECAM; - break; - case TV_STD_PAL_CN: - args.ucTVStandard = ATOM_TV_PALCN; - break; - } - args.ucEnable = SCALER_ENABLE_MULTITAP_MODE; - } else if (radeon_encoder->devices & (ATOM_DEVICE_CV_SUPPORT)) { - args.ucTVStandard = ATOM_TV_CV; - args.ucEnable = SCALER_ENABLE_MULTITAP_MODE; - } else if (radeon_encoder->flags & RADEON_USE_RMX) { - if (radeon_encoder->rmx_type == RMX_FULL) - args.ucEnable = ATOM_SCALER_EXPANSION; - else if (radeon_encoder->rmx_type == RMX_CENTER) - args.ucEnable = ATOM_SCALER_CENTER; - else if (radeon_encoder->rmx_type == RMX_ASPECT) - args.ucEnable = ATOM_SCALER_EXPANSION; - } else { - if (ASIC_IS_AVIVO(rdev)) - args.ucEnable = ATOM_SCALER_DISABLE; - else - args.ucEnable = ATOM_SCALER_CENTER; - } - - atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); - - if (radeon_encoder->devices & (ATOM_DEVICE_CV_SUPPORT | ATOM_DEVICE_TV_SUPPORT) - && rdev->family >= CHIP_RV515 && rdev->family <= CHIP_RV570) { - atom_rv515_force_tv_scaler(rdev); - } - -} - -static void radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode) { struct drm_device *dev = encoder->dev; @@ -1448,8 +1094,6 @@ radeon_atom_encoder_mode_set(struct drm_encoder *encoder, radeon_encoder->pixel_clock = adjusted_mode->clock; radeon_atombios_encoder_crtc_scratch_regs(encoder, radeon_crtc->crtc_id); - atombios_overscan_setup(encoder, mode, adjusted_mode); - atombios_scaler_setup(encoder); atombios_set_encoder_crtc_source(encoder); if (ASIC_IS_AVIVO(rdev)) { @@ -1667,6 +1311,7 @@ radeon_add_atom_encoder(struct drm_device *dev, uint32_t encoder_id, uint32_t su radeon_encoder->encoder_id = encoder_id; radeon_encoder->devices = supported_device; + radeon_encoder->rmx_type = RMX_OFF; switch (radeon_encoder->encoder_id) { case ENCODER_OBJECT_ID_INTERNAL_LVDS: diff --git a/drivers/gpu/drm/radeon/radeon_fb.c b/drivers/gpu/drm/radeon/radeon_fb.c index 9e8f191eb64..3206c0ad7b6 100644 --- a/drivers/gpu/drm/radeon/radeon_fb.c +++ b/drivers/gpu/drm/radeon/radeon_fb.c @@ -101,9 +101,10 @@ static int radeonfb_setcolreg(unsigned regno, break; case 24: case 32: - fb->pseudo_palette[regno] = ((red & 0xff00) << 8) | - (green & 0xff00) | - ((blue & 0xff00) >> 8); + fb->pseudo_palette[regno] = + (((red >> 8) & 0xff) << info->var.red.offset) | + (((green >> 8) & 0xff) << info->var.green.offset) | + (((blue >> 8) & 0xff) << info->var.blue.offset); break; } } @@ -154,6 +155,7 @@ static int radeonfb_check_var(struct fb_var_screeninfo *var, var->transp.length = 0; var->transp.offset = 0; break; +#ifdef __LITTLE_ENDIAN case 15: var->red.offset = 10; var->green.offset = 5; @@ -194,6 +196,28 @@ static int radeonfb_check_var(struct fb_var_screeninfo *var, var->transp.length = 8; var->transp.offset = 24; break; +#else + case 24: + var->red.offset = 8; + var->green.offset = 16; + var->blue.offset = 24; + var->red.length = 8; + var->green.length = 8; + var->blue.length = 8; + var->transp.length = 0; + var->transp.offset = 0; + break; + case 32: + var->red.offset = 8; + var->green.offset = 16; + var->blue.offset = 24; + var->red.length = 8; + var->green.length = 8; + var->blue.length = 8; + var->transp.length = 8; + var->transp.offset = 0; + break; +#endif default: return -EINVAL; } @@ -447,10 +471,10 @@ static struct notifier_block paniced = { .notifier_call = radeonfb_panic, }; -static int radeon_align_pitch(struct radeon_device *rdev, int width, int bpp) +static int radeon_align_pitch(struct radeon_device *rdev, int width, int bpp, bool tiled) { int aligned = width; - int align_large = (ASIC_IS_AVIVO(rdev)); + int align_large = (ASIC_IS_AVIVO(rdev)) || tiled; int pitch_mask = 0; switch (bpp / 8) { @@ -488,12 +512,13 @@ int radeonfb_create(struct radeon_device *rdev, u64 fb_gpuaddr; void *fbptr = NULL; unsigned long tmp; + bool fb_tiled = false; /* useful for testing */ mode_cmd.width = surface_width; mode_cmd.height = surface_height; mode_cmd.bpp = 32; /* need to align pitch with crtc limits */ - mode_cmd.pitch = radeon_align_pitch(rdev, mode_cmd.width, mode_cmd.bpp) * ((mode_cmd.bpp + 1) / 8); + mode_cmd.pitch = radeon_align_pitch(rdev, mode_cmd.width, mode_cmd.bpp, fb_tiled) * ((mode_cmd.bpp + 1) / 8); mode_cmd.depth = 24; size = mode_cmd.pitch * mode_cmd.height; @@ -511,6 +536,8 @@ int radeonfb_create(struct radeon_device *rdev, } robj = gobj->driver_private; + if (fb_tiled) + radeon_object_set_tiling_flags(robj, RADEON_TILING_MACRO|RADEON_TILING_SURFACE, mode_cmd.pitch); mutex_lock(&rdev->ddev->struct_mutex); fb = radeon_framebuffer_create(rdev->ddev, &mode_cmd, gobj); if (fb == NULL) { @@ -539,6 +566,9 @@ int radeonfb_create(struct radeon_device *rdev, } rfbdev = info->par; + if (fb_tiled) + radeon_object_check_tiling(robj, 0, 0); + ret = radeon_object_kmap(robj, &fbptr); if (ret) { goto out_unref; @@ -572,6 +602,11 @@ int radeonfb_create(struct radeon_device *rdev, info->var.width = -1; info->var.xres = fb_width; info->var.yres = fb_height; + + /* setup aperture base/size for vesafb takeover */ + info->aperture_base = rdev->ddev->mode_config.fb_base; + info->aperture_size = rdev->mc.real_vram_size; + info->fix.mmio_start = 0; info->fix.mmio_len = 0; info->pixmap.size = 64*1024; @@ -600,6 +635,7 @@ int radeonfb_create(struct radeon_device *rdev, info->var.transp.offset = 0; info->var.transp.length = 0; break; +#ifdef __LITTLE_ENDIAN case 15: info->var.red.offset = 10; info->var.green.offset = 5; @@ -639,7 +675,29 @@ int radeonfb_create(struct radeon_device *rdev, info->var.transp.offset = 24; info->var.transp.length = 8; break; +#else + case 24: + info->var.red.offset = 8; + info->var.green.offset = 16; + info->var.blue.offset = 24; + info->var.red.length = 8; + info->var.green.length = 8; + info->var.blue.length = 8; + info->var.transp.offset = 0; + info->var.transp.length = 0; + break; + case 32: + info->var.red.offset = 8; + info->var.green.offset = 16; + info->var.blue.offset = 24; + info->var.red.length = 8; + info->var.green.length = 8; + info->var.blue.length = 8; + info->var.transp.offset = 0; + info->var.transp.length = 8; + break; default: +#endif break; } diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c index 96afbf5ae2a..b4e48dd2e85 100644 --- a/drivers/gpu/drm/radeon/radeon_fence.c +++ b/drivers/gpu/drm/radeon/radeon_fence.c @@ -195,7 +195,7 @@ retry: r = wait_event_interruptible_timeout(rdev->fence_drv.queue, radeon_fence_signaled(fence), timeout); if (unlikely(r == -ERESTARTSYS)) { - return -ERESTART; + return -EBUSY; } } else { r = wait_event_timeout(rdev->fence_drv.queue, diff --git a/drivers/gpu/drm/radeon/radeon_gart.c b/drivers/gpu/drm/radeon/radeon_gart.c index d343a15316e..2977539880f 100644 --- a/drivers/gpu/drm/radeon/radeon_gart.c +++ b/drivers/gpu/drm/radeon/radeon_gart.c @@ -177,7 +177,7 @@ int radeon_gart_bind(struct radeon_device *rdev, unsigned offset, return -ENOMEM; } rdev->gart.pages[p] = pagelist[i]; - page_base = (uint32_t)rdev->gart.pages_addr[p]; + page_base = rdev->gart.pages_addr[p]; for (j = 0; j < (PAGE_SIZE / 4096); j++, t++) { radeon_gart_set_page(rdev, t, page_base); page_base += 4096; diff --git a/drivers/gpu/drm/radeon/radeon_gem.c b/drivers/gpu/drm/radeon/radeon_gem.c index eb516034235..cded5180c75 100644 --- a/drivers/gpu/drm/radeon/radeon_gem.c +++ b/drivers/gpu/drm/radeon/radeon_gem.c @@ -157,9 +157,9 @@ int radeon_gem_info_ioctl(struct drm_device *dev, void *data, struct radeon_device *rdev = dev->dev_private; struct drm_radeon_gem_info *args = data; - args->vram_size = rdev->mc.vram_size; + args->vram_size = rdev->mc.real_vram_size; /* FIXME: report somethings that makes sense */ - args->vram_visible = rdev->mc.vram_size - (4 * 1024 * 1024); + args->vram_visible = rdev->mc.real_vram_size - (4 * 1024 * 1024); args->gart_size = rdev->mc.gtt_size; return 0; } @@ -285,3 +285,44 @@ int radeon_gem_wait_idle_ioctl(struct drm_device *dev, void *data, mutex_unlock(&dev->struct_mutex); return r; } + +int radeon_gem_set_tiling_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + struct drm_radeon_gem_set_tiling *args = data; + struct drm_gem_object *gobj; + struct radeon_object *robj; + int r = 0; + + DRM_DEBUG("%d \n", args->handle); + gobj = drm_gem_object_lookup(dev, filp, args->handle); + if (gobj == NULL) + return -EINVAL; + robj = gobj->driver_private; + radeon_object_set_tiling_flags(robj, args->tiling_flags, args->pitch); + mutex_lock(&dev->struct_mutex); + drm_gem_object_unreference(gobj); + mutex_unlock(&dev->struct_mutex); + return r; +} + +int radeon_gem_get_tiling_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + struct drm_radeon_gem_get_tiling *args = data; + struct drm_gem_object *gobj; + struct radeon_object *robj; + int r = 0; + + DRM_DEBUG("\n"); + gobj = drm_gem_object_lookup(dev, filp, args->handle); + if (gobj == NULL) + return -EINVAL; + robj = gobj->driver_private; + radeon_object_get_tiling_flags(robj, &args->tiling_flags, + &args->pitch); + mutex_lock(&dev->struct_mutex); + drm_gem_object_unreference(gobj); + mutex_unlock(&dev->struct_mutex); + return r; +} diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c index 4612a7c146d..937a2f1cdb4 100644 --- a/drivers/gpu/drm/radeon/radeon_kms.c +++ b/drivers/gpu/drm/radeon/radeon_kms.c @@ -291,5 +291,7 @@ struct drm_ioctl_desc radeon_ioctls_kms[] = { DRM_IOCTL_DEF(DRM_RADEON_GEM_WAIT_IDLE, radeon_gem_wait_idle_ioctl, DRM_AUTH), DRM_IOCTL_DEF(DRM_RADEON_CS, radeon_cs_ioctl, DRM_AUTH), DRM_IOCTL_DEF(DRM_RADEON_INFO, radeon_info_ioctl, DRM_AUTH), + DRM_IOCTL_DEF(DRM_RADEON_GEM_SET_TILING, radeon_gem_set_tiling_ioctl, DRM_AUTH), + DRM_IOCTL_DEF(DRM_RADEON_GEM_GET_TILING, radeon_gem_get_tiling_ioctl, DRM_AUTH), }; int radeon_max_kms_ioctl = DRM_ARRAY_SIZE(radeon_ioctls_kms); diff --git a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c index 8086ecf7f03..7d06dc98a42 100644 --- a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c +++ b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c @@ -29,6 +29,171 @@ #include "radeon_fixed.h" #include "radeon.h" +static void radeon_legacy_rmx_mode_set(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + int xres = mode->hdisplay; + int yres = mode->vdisplay; + bool hscale = true, vscale = true; + int hsync_wid; + int vsync_wid; + int hsync_start; + int blank_width; + u32 scale, inc, crtc_more_cntl; + u32 fp_horz_stretch, fp_vert_stretch, fp_horz_vert_active; + u32 fp_h_sync_strt_wid, fp_crtc_h_total_disp; + u32 fp_v_sync_strt_wid, fp_crtc_v_total_disp; + struct radeon_native_mode *native_mode = &radeon_crtc->native_mode; + + fp_vert_stretch = RREG32(RADEON_FP_VERT_STRETCH) & + (RADEON_VERT_STRETCH_RESERVED | + RADEON_VERT_AUTO_RATIO_INC); + fp_horz_stretch = RREG32(RADEON_FP_HORZ_STRETCH) & + (RADEON_HORZ_FP_LOOP_STRETCH | + RADEON_HORZ_AUTO_RATIO_INC); + + crtc_more_cntl = 0; + if ((rdev->family == CHIP_RS100) || + (rdev->family == CHIP_RS200)) { + /* This is to workaround the asic bug for RMX, some versions + of BIOS dosen't have this register initialized correctly. */ + crtc_more_cntl |= RADEON_CRTC_H_CUTOFF_ACTIVE_EN; + } + + + fp_crtc_h_total_disp = ((((mode->crtc_htotal / 8) - 1) & 0x3ff) + | ((((mode->crtc_hdisplay / 8) - 1) & 0x1ff) << 16)); + + hsync_wid = (mode->crtc_hsync_end - mode->crtc_hsync_start) / 8; + if (!hsync_wid) + hsync_wid = 1; + hsync_start = mode->crtc_hsync_start - 8; + + fp_h_sync_strt_wid = ((hsync_start & 0x1fff) + | ((hsync_wid & 0x3f) << 16) + | ((mode->flags & DRM_MODE_FLAG_NHSYNC) + ? RADEON_CRTC_H_SYNC_POL + : 0)); + + fp_crtc_v_total_disp = (((mode->crtc_vtotal - 1) & 0xffff) + | ((mode->crtc_vdisplay - 1) << 16)); + + vsync_wid = mode->crtc_vsync_end - mode->crtc_vsync_start; + if (!vsync_wid) + vsync_wid = 1; + + fp_v_sync_strt_wid = (((mode->crtc_vsync_start - 1) & 0xfff) + | ((vsync_wid & 0x1f) << 16) + | ((mode->flags & DRM_MODE_FLAG_NVSYNC) + ? RADEON_CRTC_V_SYNC_POL + : 0)); + + fp_horz_vert_active = 0; + + if (native_mode->panel_xres == 0 || + native_mode->panel_yres == 0) { + hscale = false; + vscale = false; + } else { + if (xres > native_mode->panel_xres) + xres = native_mode->panel_xres; + if (yres > native_mode->panel_yres) + yres = native_mode->panel_yres; + + if (xres == native_mode->panel_xres) + hscale = false; + if (yres == native_mode->panel_yres) + vscale = false; + } + + switch (radeon_crtc->rmx_type) { + case RMX_FULL: + case RMX_ASPECT: + if (!hscale) + fp_horz_stretch |= ((xres/8-1) << 16); + else { + inc = (fp_horz_stretch & RADEON_HORZ_AUTO_RATIO_INC) ? 1 : 0; + scale = ((xres + inc) * RADEON_HORZ_STRETCH_RATIO_MAX) + / native_mode->panel_xres + 1; + fp_horz_stretch |= (((scale) & RADEON_HORZ_STRETCH_RATIO_MASK) | + RADEON_HORZ_STRETCH_BLEND | + RADEON_HORZ_STRETCH_ENABLE | + ((native_mode->panel_xres/8-1) << 16)); + } + + if (!vscale) + fp_vert_stretch |= ((yres-1) << 12); + else { + inc = (fp_vert_stretch & RADEON_VERT_AUTO_RATIO_INC) ? 1 : 0; + scale = ((yres + inc) * RADEON_VERT_STRETCH_RATIO_MAX) + / native_mode->panel_yres + 1; + fp_vert_stretch |= (((scale) & RADEON_VERT_STRETCH_RATIO_MASK) | + RADEON_VERT_STRETCH_ENABLE | + RADEON_VERT_STRETCH_BLEND | + ((native_mode->panel_yres-1) << 12)); + } + break; + case RMX_CENTER: + fp_horz_stretch |= ((xres/8-1) << 16); + fp_vert_stretch |= ((yres-1) << 12); + + crtc_more_cntl |= (RADEON_CRTC_AUTO_HORZ_CENTER_EN | + RADEON_CRTC_AUTO_VERT_CENTER_EN); + + blank_width = (mode->crtc_hblank_end - mode->crtc_hblank_start) / 8; + if (blank_width > 110) + blank_width = 110; + + fp_crtc_h_total_disp = (((blank_width) & 0x3ff) + | ((((mode->crtc_hdisplay / 8) - 1) & 0x1ff) << 16)); + + hsync_wid = (mode->crtc_hsync_end - mode->crtc_hsync_start) / 8; + if (!hsync_wid) + hsync_wid = 1; + + fp_h_sync_strt_wid = ((((mode->crtc_hsync_start - mode->crtc_hblank_start) / 8) & 0x1fff) + | ((hsync_wid & 0x3f) << 16) + | ((mode->flags & DRM_MODE_FLAG_NHSYNC) + ? RADEON_CRTC_H_SYNC_POL + : 0)); + + fp_crtc_v_total_disp = (((mode->crtc_vblank_end - mode->crtc_vblank_start) & 0xffff) + | ((mode->crtc_vdisplay - 1) << 16)); + + vsync_wid = mode->crtc_vsync_end - mode->crtc_vsync_start; + if (!vsync_wid) + vsync_wid = 1; + + fp_v_sync_strt_wid = ((((mode->crtc_vsync_start - mode->crtc_vblank_start) & 0xfff) + | ((vsync_wid & 0x1f) << 16) + | ((mode->flags & DRM_MODE_FLAG_NVSYNC) + ? RADEON_CRTC_V_SYNC_POL + : 0))); + + fp_horz_vert_active = (((native_mode->panel_yres) & 0xfff) | + (((native_mode->panel_xres / 8) & 0x1ff) << 16)); + break; + case RMX_OFF: + default: + fp_horz_stretch |= ((xres/8-1) << 16); + fp_vert_stretch |= ((yres-1) << 12); + break; + } + + WREG32(RADEON_FP_HORZ_STRETCH, fp_horz_stretch); + WREG32(RADEON_FP_VERT_STRETCH, fp_vert_stretch); + WREG32(RADEON_CRTC_MORE_CNTL, crtc_more_cntl); + WREG32(RADEON_FP_HORZ_VERT_ACTIVE, fp_horz_vert_active); + WREG32(RADEON_FP_H_SYNC_STRT_WID, fp_h_sync_strt_wid); + WREG32(RADEON_FP_V_SYNC_STRT_WID, fp_v_sync_strt_wid); + WREG32(RADEON_FP_CRTC_H_TOTAL_DISP, fp_crtc_h_total_disp); + WREG32(RADEON_FP_CRTC_V_TOTAL_DISP, fp_crtc_v_total_disp); +} + void radeon_restore_common_regs(struct drm_device *dev) { /* don't need this yet */ @@ -235,6 +400,7 @@ int radeon_crtc_set_base(struct drm_crtc *crtc, int x, int y, uint64_t base; uint32_t crtc_offset, crtc_offset_cntl, crtc_tile_x0_y0 = 0; uint32_t crtc_pitch, pitch_pixels; + uint32_t tiling_flags; DRM_DEBUG("\n"); @@ -244,7 +410,12 @@ int radeon_crtc_set_base(struct drm_crtc *crtc, int x, int y, if (radeon_gem_object_pin(obj, RADEON_GEM_DOMAIN_VRAM, &base)) { return -EINVAL; } - crtc_offset = (u32)base; + /* if scanout was in GTT this really wouldn't work */ + /* crtc offset is from display base addr not FB location */ + radeon_crtc->legacy_display_base_addr = rdev->mc.vram_location; + + base -= radeon_crtc->legacy_display_base_addr; + crtc_offset_cntl = 0; pitch_pixels = crtc->fb->pitch / (crtc->fb->bits_per_pixel / 8); @@ -253,8 +424,12 @@ int radeon_crtc_set_base(struct drm_crtc *crtc, int x, int y, (crtc->fb->bits_per_pixel * 8)); crtc_pitch |= crtc_pitch << 16; - /* TODO tiling */ - if (0) { + radeon_object_get_tiling_flags(obj->driver_private, + &tiling_flags, NULL); + if (tiling_flags & RADEON_TILING_MICRO) + DRM_ERROR("trying to scanout microtiled buffer\n"); + + if (tiling_flags & RADEON_TILING_MACRO) { if (ASIC_IS_R300(rdev)) crtc_offset_cntl |= (R300_CRTC_X_Y_MODE_EN | R300_CRTC_MICRO_TILE_BUFFER_DIS | @@ -270,15 +445,13 @@ int radeon_crtc_set_base(struct drm_crtc *crtc, int x, int y, crtc_offset_cntl &= ~RADEON_CRTC_TILE_EN; } - - /* TODO more tiling */ - if (0) { + if (tiling_flags & RADEON_TILING_MACRO) { if (ASIC_IS_R300(rdev)) { crtc_tile_x0_y0 = x | (y << 16); base &= ~0x7ff; } else { int byteshift = crtc->fb->bits_per_pixel >> 4; - int tile_addr = (((y >> 3) * crtc->fb->width + x) >> (8 - byteshift)) << 11; + int tile_addr = (((y >> 3) * pitch_pixels + x) >> (8 - byteshift)) << 11; base += tile_addr + ((x << byteshift) % 256) + ((y % 8) << 8); crtc_offset_cntl |= (y % 16); } @@ -303,11 +476,9 @@ int radeon_crtc_set_base(struct drm_crtc *crtc, int x, int y, base &= ~7; - /* update sarea TODO */ - crtc_offset = (u32)base; - WREG32(RADEON_DISPLAY_BASE_ADDR + radeon_crtc->crtc_offset, rdev->mc.vram_location); + WREG32(RADEON_DISPLAY_BASE_ADDR + radeon_crtc->crtc_offset, radeon_crtc->legacy_display_base_addr); if (ASIC_IS_R300(rdev)) { if (radeon_crtc->crtc_id) @@ -751,6 +922,8 @@ static bool radeon_crtc_mode_fixup(struct drm_crtc *crtc, struct drm_display_mode *mode, struct drm_display_mode *adjusted_mode) { + if (!radeon_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode)) + return false; return true; } @@ -759,16 +932,25 @@ static int radeon_crtc_mode_set(struct drm_crtc *crtc, struct drm_display_mode *adjusted_mode, int x, int y, struct drm_framebuffer *old_fb) { - - DRM_DEBUG("\n"); + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; /* TODO TV */ - radeon_crtc_set_base(crtc, x, y, old_fb); radeon_set_crtc_timing(crtc, adjusted_mode); radeon_set_pll(crtc, adjusted_mode); - radeon_init_disp_bandwidth(crtc->dev); - + radeon_bandwidth_update(rdev); + if (radeon_crtc->crtc_id == 0) { + radeon_legacy_rmx_mode_set(crtc, mode, adjusted_mode); + } else { + if (radeon_crtc->rmx_type != RMX_OFF) { + /* FIXME: only first crtc has rmx what should we + * do ? + */ + DRM_ERROR("Mode need scaling but only first crtc can do that.\n"); + } + } return 0; } @@ -799,478 +981,3 @@ void radeon_legacy_init_crtc(struct drm_device *dev, radeon_crtc->crtc_offset = RADEON_CRTC2_H_TOTAL_DISP - RADEON_CRTC_H_TOTAL_DISP; drm_crtc_helper_add(&radeon_crtc->base, &legacy_helper_funcs); } - -void radeon_init_disp_bw_legacy(struct drm_device *dev, - struct drm_display_mode *mode1, - uint32_t pixel_bytes1, - struct drm_display_mode *mode2, - uint32_t pixel_bytes2) -{ - struct radeon_device *rdev = dev->dev_private; - fixed20_12 trcd_ff, trp_ff, tras_ff, trbs_ff, tcas_ff; - fixed20_12 sclk_ff, mclk_ff, sclk_eff_ff, sclk_delay_ff; - fixed20_12 peak_disp_bw, mem_bw, pix_clk, pix_clk2, temp_ff, crit_point_ff; - uint32_t temp, data, mem_trcd, mem_trp, mem_tras; - fixed20_12 memtcas_ff[8] = { - fixed_init(1), - fixed_init(2), - fixed_init(3), - fixed_init(0), - fixed_init_half(1), - fixed_init_half(2), - fixed_init(0), - }; - fixed20_12 memtcas_rs480_ff[8] = { - fixed_init(0), - fixed_init(1), - fixed_init(2), - fixed_init(3), - fixed_init(0), - fixed_init_half(1), - fixed_init_half(2), - fixed_init_half(3), - }; - fixed20_12 memtcas2_ff[8] = { - fixed_init(0), - fixed_init(1), - fixed_init(2), - fixed_init(3), - fixed_init(4), - fixed_init(5), - fixed_init(6), - fixed_init(7), - }; - fixed20_12 memtrbs[8] = { - fixed_init(1), - fixed_init_half(1), - fixed_init(2), - fixed_init_half(2), - fixed_init(3), - fixed_init_half(3), - fixed_init(4), - fixed_init_half(4) - }; - fixed20_12 memtrbs_r4xx[8] = { - fixed_init(4), - fixed_init(5), - fixed_init(6), - fixed_init(7), - fixed_init(8), - fixed_init(9), - fixed_init(10), - fixed_init(11) - }; - fixed20_12 min_mem_eff; - fixed20_12 mc_latency_sclk, mc_latency_mclk, k1; - fixed20_12 cur_latency_mclk, cur_latency_sclk; - fixed20_12 disp_latency, disp_latency_overhead, disp_drain_rate, - disp_drain_rate2, read_return_rate; - fixed20_12 time_disp1_drop_priority; - int c; - int cur_size = 16; /* in octawords */ - int critical_point = 0, critical_point2; -/* uint32_t read_return_rate, time_disp1_drop_priority; */ - int stop_req, max_stop_req; - - min_mem_eff.full = rfixed_const_8(0); - /* get modes */ - if ((rdev->disp_priority == 2) && ASIC_IS_R300(rdev)) { - uint32_t mc_init_misc_lat_timer = RREG32(R300_MC_INIT_MISC_LAT_TIMER); - mc_init_misc_lat_timer &= ~(R300_MC_DISP1R_INIT_LAT_MASK << R300_MC_DISP1R_INIT_LAT_SHIFT); - mc_init_misc_lat_timer &= ~(R300_MC_DISP0R_INIT_LAT_MASK << R300_MC_DISP0R_INIT_LAT_SHIFT); - /* check crtc enables */ - if (mode2) - mc_init_misc_lat_timer |= (1 << R300_MC_DISP1R_INIT_LAT_SHIFT); - if (mode1) - mc_init_misc_lat_timer |= (1 << R300_MC_DISP0R_INIT_LAT_SHIFT); - WREG32(R300_MC_INIT_MISC_LAT_TIMER, mc_init_misc_lat_timer); - } - - /* - * determine is there is enough bw for current mode - */ - mclk_ff.full = rfixed_const(rdev->clock.default_mclk); - temp_ff.full = rfixed_const(100); - mclk_ff.full = rfixed_div(mclk_ff, temp_ff); - sclk_ff.full = rfixed_const(rdev->clock.default_sclk); - sclk_ff.full = rfixed_div(sclk_ff, temp_ff); - - temp = (rdev->mc.vram_width / 8) * (rdev->mc.vram_is_ddr ? 2 : 1); - temp_ff.full = rfixed_const(temp); - mem_bw.full = rfixed_mul(mclk_ff, temp_ff); - - pix_clk.full = 0; - pix_clk2.full = 0; - peak_disp_bw.full = 0; - if (mode1) { - temp_ff.full = rfixed_const(1000); - pix_clk.full = rfixed_const(mode1->clock); /* convert to fixed point */ - pix_clk.full = rfixed_div(pix_clk, temp_ff); - temp_ff.full = rfixed_const(pixel_bytes1); - peak_disp_bw.full += rfixed_mul(pix_clk, temp_ff); - } - if (mode2) { - temp_ff.full = rfixed_const(1000); - pix_clk2.full = rfixed_const(mode2->clock); /* convert to fixed point */ - pix_clk2.full = rfixed_div(pix_clk2, temp_ff); - temp_ff.full = rfixed_const(pixel_bytes2); - peak_disp_bw.full += rfixed_mul(pix_clk2, temp_ff); - } - - mem_bw.full = rfixed_mul(mem_bw, min_mem_eff); - if (peak_disp_bw.full >= mem_bw.full) { - DRM_ERROR("You may not have enough display bandwidth for current mode\n" - "If you have flickering problem, try to lower resolution, refresh rate, or color depth\n"); - } - - /* Get values from the EXT_MEM_CNTL register...converting its contents. */ - temp = RREG32(RADEON_MEM_TIMING_CNTL); - if ((rdev->family == CHIP_RV100) || (rdev->flags & RADEON_IS_IGP)) { /* RV100, M6, IGPs */ - mem_trcd = ((temp >> 2) & 0x3) + 1; - mem_trp = ((temp & 0x3)) + 1; - mem_tras = ((temp & 0x70) >> 4) + 1; - } else if (rdev->family == CHIP_R300 || - rdev->family == CHIP_R350) { /* r300, r350 */ - mem_trcd = (temp & 0x7) + 1; - mem_trp = ((temp >> 8) & 0x7) + 1; - mem_tras = ((temp >> 11) & 0xf) + 4; - } else if (rdev->family == CHIP_RV350 || - rdev->family <= CHIP_RV380) { - /* rv3x0 */ - mem_trcd = (temp & 0x7) + 3; - mem_trp = ((temp >> 8) & 0x7) + 3; - mem_tras = ((temp >> 11) & 0xf) + 6; - } else if (rdev->family == CHIP_R420 || - rdev->family == CHIP_R423 || - rdev->family == CHIP_RV410) { - /* r4xx */ - mem_trcd = (temp & 0xf) + 3; - if (mem_trcd > 15) - mem_trcd = 15; - mem_trp = ((temp >> 8) & 0xf) + 3; - if (mem_trp > 15) - mem_trp = 15; - mem_tras = ((temp >> 12) & 0x1f) + 6; - if (mem_tras > 31) - mem_tras = 31; - } else { /* RV200, R200 */ - mem_trcd = (temp & 0x7) + 1; - mem_trp = ((temp >> 8) & 0x7) + 1; - mem_tras = ((temp >> 12) & 0xf) + 4; - } - /* convert to FF */ - trcd_ff.full = rfixed_const(mem_trcd); - trp_ff.full = rfixed_const(mem_trp); - tras_ff.full = rfixed_const(mem_tras); - - /* Get values from the MEM_SDRAM_MODE_REG register...converting its */ - temp = RREG32(RADEON_MEM_SDRAM_MODE_REG); - data = (temp & (7 << 20)) >> 20; - if ((rdev->family == CHIP_RV100) || rdev->flags & RADEON_IS_IGP) { - if (rdev->family == CHIP_RS480) /* don't think rs400 */ - tcas_ff = memtcas_rs480_ff[data]; - else - tcas_ff = memtcas_ff[data]; - } else - tcas_ff = memtcas2_ff[data]; - - if (rdev->family == CHIP_RS400 || - rdev->family == CHIP_RS480) { - /* extra cas latency stored in bits 23-25 0-4 clocks */ - data = (temp >> 23) & 0x7; - if (data < 5) - tcas_ff.full += rfixed_const(data); - } - - if (ASIC_IS_R300(rdev) && !(rdev->flags & RADEON_IS_IGP)) { - /* on the R300, Tcas is included in Trbs. - */ - temp = RREG32(RADEON_MEM_CNTL); - data = (R300_MEM_NUM_CHANNELS_MASK & temp); - if (data == 1) { - if (R300_MEM_USE_CD_CH_ONLY & temp) { - temp = RREG32(R300_MC_IND_INDEX); - temp &= ~R300_MC_IND_ADDR_MASK; - temp |= R300_MC_READ_CNTL_CD_mcind; - WREG32(R300_MC_IND_INDEX, temp); - temp = RREG32(R300_MC_IND_DATA); - data = (R300_MEM_RBS_POSITION_C_MASK & temp); - } else { - temp = RREG32(R300_MC_READ_CNTL_AB); - data = (R300_MEM_RBS_POSITION_A_MASK & temp); - } - } else { - temp = RREG32(R300_MC_READ_CNTL_AB); - data = (R300_MEM_RBS_POSITION_A_MASK & temp); - } - if (rdev->family == CHIP_RV410 || - rdev->family == CHIP_R420 || - rdev->family == CHIP_R423) - trbs_ff = memtrbs_r4xx[data]; - else - trbs_ff = memtrbs[data]; - tcas_ff.full += trbs_ff.full; - } - - sclk_eff_ff.full = sclk_ff.full; - - if (rdev->flags & RADEON_IS_AGP) { - fixed20_12 agpmode_ff; - agpmode_ff.full = rfixed_const(radeon_agpmode); - temp_ff.full = rfixed_const_666(16); - sclk_eff_ff.full -= rfixed_mul(agpmode_ff, temp_ff); - } - /* TODO PCIE lanes may affect this - agpmode == 16?? */ - - if (ASIC_IS_R300(rdev)) { - sclk_delay_ff.full = rfixed_const(250); - } else { - if ((rdev->family == CHIP_RV100) || - rdev->flags & RADEON_IS_IGP) { - if (rdev->mc.vram_is_ddr) - sclk_delay_ff.full = rfixed_const(41); - else - sclk_delay_ff.full = rfixed_const(33); - } else { - if (rdev->mc.vram_width == 128) - sclk_delay_ff.full = rfixed_const(57); - else - sclk_delay_ff.full = rfixed_const(41); - } - } - - mc_latency_sclk.full = rfixed_div(sclk_delay_ff, sclk_eff_ff); - - if (rdev->mc.vram_is_ddr) { - if (rdev->mc.vram_width == 32) { - k1.full = rfixed_const(40); - c = 3; - } else { - k1.full = rfixed_const(20); - c = 1; - } - } else { - k1.full = rfixed_const(40); - c = 3; - } - - temp_ff.full = rfixed_const(2); - mc_latency_mclk.full = rfixed_mul(trcd_ff, temp_ff); - temp_ff.full = rfixed_const(c); - mc_latency_mclk.full += rfixed_mul(tcas_ff, temp_ff); - temp_ff.full = rfixed_const(4); - mc_latency_mclk.full += rfixed_mul(tras_ff, temp_ff); - mc_latency_mclk.full += rfixed_mul(trp_ff, temp_ff); - mc_latency_mclk.full += k1.full; - - mc_latency_mclk.full = rfixed_div(mc_latency_mclk, mclk_ff); - mc_latency_mclk.full += rfixed_div(temp_ff, sclk_eff_ff); - - /* - HW cursor time assuming worst case of full size colour cursor. - */ - temp_ff.full = rfixed_const((2 * (cur_size - (rdev->mc.vram_is_ddr + 1)))); - temp_ff.full += trcd_ff.full; - if (temp_ff.full < tras_ff.full) - temp_ff.full = tras_ff.full; - cur_latency_mclk.full = rfixed_div(temp_ff, mclk_ff); - - temp_ff.full = rfixed_const(cur_size); - cur_latency_sclk.full = rfixed_div(temp_ff, sclk_eff_ff); - /* - Find the total latency for the display data. - */ - disp_latency_overhead.full = rfixed_const(80); - disp_latency_overhead.full = rfixed_div(disp_latency_overhead, sclk_ff); - mc_latency_mclk.full += disp_latency_overhead.full + cur_latency_mclk.full; - mc_latency_sclk.full += disp_latency_overhead.full + cur_latency_sclk.full; - - if (mc_latency_mclk.full > mc_latency_sclk.full) - disp_latency.full = mc_latency_mclk.full; - else - disp_latency.full = mc_latency_sclk.full; - - /* setup Max GRPH_STOP_REQ default value */ - if (ASIC_IS_RV100(rdev)) - max_stop_req = 0x5c; - else - max_stop_req = 0x7c; - - if (mode1) { - /* CRTC1 - Set GRPH_BUFFER_CNTL register using h/w defined optimal values. - GRPH_STOP_REQ <= MIN[ 0x7C, (CRTC_H_DISP + 1) * (bit depth) / 0x10 ] - */ - stop_req = mode1->hdisplay * pixel_bytes1 / 16; - - if (stop_req > max_stop_req) - stop_req = max_stop_req; - - /* - Find the drain rate of the display buffer. - */ - temp_ff.full = rfixed_const((16/pixel_bytes1)); - disp_drain_rate.full = rfixed_div(pix_clk, temp_ff); - - /* - Find the critical point of the display buffer. - */ - crit_point_ff.full = rfixed_mul(disp_drain_rate, disp_latency); - crit_point_ff.full += rfixed_const_half(0); - - critical_point = rfixed_trunc(crit_point_ff); - - if (rdev->disp_priority == 2) { - critical_point = 0; - } - - /* - The critical point should never be above max_stop_req-4. Setting - GRPH_CRITICAL_CNTL = 0 will thus force high priority all the time. - */ - if (max_stop_req - critical_point < 4) - critical_point = 0; - - if (critical_point == 0 && mode2 && rdev->family == CHIP_R300) { - /* some R300 cards have problem with this set to 0, when CRTC2 is enabled.*/ - critical_point = 0x10; - } - - temp = RREG32(RADEON_GRPH_BUFFER_CNTL); - temp &= ~(RADEON_GRPH_STOP_REQ_MASK); - temp |= (stop_req << RADEON_GRPH_STOP_REQ_SHIFT); - temp &= ~(RADEON_GRPH_START_REQ_MASK); - if ((rdev->family == CHIP_R350) && - (stop_req > 0x15)) { - stop_req -= 0x10; - } - temp |= (stop_req << RADEON_GRPH_START_REQ_SHIFT); - temp |= RADEON_GRPH_BUFFER_SIZE; - temp &= ~(RADEON_GRPH_CRITICAL_CNTL | - RADEON_GRPH_CRITICAL_AT_SOF | - RADEON_GRPH_STOP_CNTL); - /* - Write the result into the register. - */ - WREG32(RADEON_GRPH_BUFFER_CNTL, ((temp & ~RADEON_GRPH_CRITICAL_POINT_MASK) | - (critical_point << RADEON_GRPH_CRITICAL_POINT_SHIFT))); - -#if 0 - if ((rdev->family == CHIP_RS400) || - (rdev->family == CHIP_RS480)) { - /* attempt to program RS400 disp regs correctly ??? */ - temp = RREG32(RS400_DISP1_REG_CNTL); - temp &= ~(RS400_DISP1_START_REQ_LEVEL_MASK | - RS400_DISP1_STOP_REQ_LEVEL_MASK); - WREG32(RS400_DISP1_REQ_CNTL1, (temp | - (critical_point << RS400_DISP1_START_REQ_LEVEL_SHIFT) | - (critical_point << RS400_DISP1_STOP_REQ_LEVEL_SHIFT))); - temp = RREG32(RS400_DMIF_MEM_CNTL1); - temp &= ~(RS400_DISP1_CRITICAL_POINT_START_MASK | - RS400_DISP1_CRITICAL_POINT_STOP_MASK); - WREG32(RS400_DMIF_MEM_CNTL1, (temp | - (critical_point << RS400_DISP1_CRITICAL_POINT_START_SHIFT) | - (critical_point << RS400_DISP1_CRITICAL_POINT_STOP_SHIFT))); - } -#endif - - DRM_DEBUG("GRPH_BUFFER_CNTL from to %x\n", - /* (unsigned int)info->SavedReg->grph_buffer_cntl, */ - (unsigned int)RREG32(RADEON_GRPH_BUFFER_CNTL)); - } - - if (mode2) { - u32 grph2_cntl; - stop_req = mode2->hdisplay * pixel_bytes2 / 16; - - if (stop_req > max_stop_req) - stop_req = max_stop_req; - - /* - Find the drain rate of the display buffer. - */ - temp_ff.full = rfixed_const((16/pixel_bytes2)); - disp_drain_rate2.full = rfixed_div(pix_clk2, temp_ff); - - grph2_cntl = RREG32(RADEON_GRPH2_BUFFER_CNTL); - grph2_cntl &= ~(RADEON_GRPH_STOP_REQ_MASK); - grph2_cntl |= (stop_req << RADEON_GRPH_STOP_REQ_SHIFT); - grph2_cntl &= ~(RADEON_GRPH_START_REQ_MASK); - if ((rdev->family == CHIP_R350) && - (stop_req > 0x15)) { - stop_req -= 0x10; - } - grph2_cntl |= (stop_req << RADEON_GRPH_START_REQ_SHIFT); - grph2_cntl |= RADEON_GRPH_BUFFER_SIZE; - grph2_cntl &= ~(RADEON_GRPH_CRITICAL_CNTL | - RADEON_GRPH_CRITICAL_AT_SOF | - RADEON_GRPH_STOP_CNTL); - - if ((rdev->family == CHIP_RS100) || - (rdev->family == CHIP_RS200)) - critical_point2 = 0; - else { - temp = (rdev->mc.vram_width * rdev->mc.vram_is_ddr + 1)/128; - temp_ff.full = rfixed_const(temp); - temp_ff.full = rfixed_mul(mclk_ff, temp_ff); - if (sclk_ff.full < temp_ff.full) - temp_ff.full = sclk_ff.full; - - read_return_rate.full = temp_ff.full; - - if (mode1) { - temp_ff.full = read_return_rate.full - disp_drain_rate.full; - time_disp1_drop_priority.full = rfixed_div(crit_point_ff, temp_ff); - } else { - time_disp1_drop_priority.full = 0; - } - crit_point_ff.full = disp_latency.full + time_disp1_drop_priority.full + disp_latency.full; - crit_point_ff.full = rfixed_mul(crit_point_ff, disp_drain_rate2); - crit_point_ff.full += rfixed_const_half(0); - - critical_point2 = rfixed_trunc(crit_point_ff); - - if (rdev->disp_priority == 2) { - critical_point2 = 0; - } - - if (max_stop_req - critical_point2 < 4) - critical_point2 = 0; - - } - - if (critical_point2 == 0 && rdev->family == CHIP_R300) { - /* some R300 cards have problem with this set to 0 */ - critical_point2 = 0x10; - } - - WREG32(RADEON_GRPH2_BUFFER_CNTL, ((grph2_cntl & ~RADEON_GRPH_CRITICAL_POINT_MASK) | - (critical_point2 << RADEON_GRPH_CRITICAL_POINT_SHIFT))); - - if ((rdev->family == CHIP_RS400) || - (rdev->family == CHIP_RS480)) { -#if 0 - /* attempt to program RS400 disp2 regs correctly ??? */ - temp = RREG32(RS400_DISP2_REQ_CNTL1); - temp &= ~(RS400_DISP2_START_REQ_LEVEL_MASK | - RS400_DISP2_STOP_REQ_LEVEL_MASK); - WREG32(RS400_DISP2_REQ_CNTL1, (temp | - (critical_point2 << RS400_DISP1_START_REQ_LEVEL_SHIFT) | - (critical_point2 << RS400_DISP1_STOP_REQ_LEVEL_SHIFT))); - temp = RREG32(RS400_DISP2_REQ_CNTL2); - temp &= ~(RS400_DISP2_CRITICAL_POINT_START_MASK | - RS400_DISP2_CRITICAL_POINT_STOP_MASK); - WREG32(RS400_DISP2_REQ_CNTL2, (temp | - (critical_point2 << RS400_DISP2_CRITICAL_POINT_START_SHIFT) | - (critical_point2 << RS400_DISP2_CRITICAL_POINT_STOP_SHIFT))); -#endif - WREG32(RS400_DISP2_REQ_CNTL1, 0x105DC1CC); - WREG32(RS400_DISP2_REQ_CNTL2, 0x2749D000); - WREG32(RS400_DMIF_MEM_CNTL1, 0x29CA71DC); - WREG32(RS400_DISP1_REQ_CNTL1, 0x28FBC3AC); - } - - DRM_DEBUG("GRPH2_BUFFER_CNTL from to %x\n", - (unsigned int)RREG32(RADEON_GRPH2_BUFFER_CNTL)); - } -} diff --git a/drivers/gpu/drm/radeon/radeon_legacy_encoders.c b/drivers/gpu/drm/radeon/radeon_legacy_encoders.c index 2c2f42de1d4..34d0f58eb94 100644 --- a/drivers/gpu/drm/radeon/radeon_legacy_encoders.c +++ b/drivers/gpu/drm/radeon/radeon_legacy_encoders.c @@ -30,170 +30,6 @@ #include "atom.h" -static void radeon_legacy_rmx_mode_set(struct drm_encoder *encoder, - struct drm_display_mode *mode, - struct drm_display_mode *adjusted_mode) -{ - struct drm_device *dev = encoder->dev; - struct radeon_device *rdev = dev->dev_private; - struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); - int xres = mode->hdisplay; - int yres = mode->vdisplay; - bool hscale = true, vscale = true; - int hsync_wid; - int vsync_wid; - int hsync_start; - uint32_t scale, inc; - uint32_t fp_horz_stretch, fp_vert_stretch, crtc_more_cntl, fp_horz_vert_active; - uint32_t fp_h_sync_strt_wid, fp_v_sync_strt_wid, fp_crtc_h_total_disp, fp_crtc_v_total_disp; - struct radeon_native_mode *native_mode = &radeon_encoder->native_mode; - - DRM_DEBUG("\n"); - - fp_vert_stretch = RREG32(RADEON_FP_VERT_STRETCH) & - (RADEON_VERT_STRETCH_RESERVED | - RADEON_VERT_AUTO_RATIO_INC); - fp_horz_stretch = RREG32(RADEON_FP_HORZ_STRETCH) & - (RADEON_HORZ_FP_LOOP_STRETCH | - RADEON_HORZ_AUTO_RATIO_INC); - - crtc_more_cntl = 0; - if ((rdev->family == CHIP_RS100) || - (rdev->family == CHIP_RS200)) { - /* This is to workaround the asic bug for RMX, some versions - of BIOS dosen't have this register initialized correctly. */ - crtc_more_cntl |= RADEON_CRTC_H_CUTOFF_ACTIVE_EN; - } - - - fp_crtc_h_total_disp = ((((mode->crtc_htotal / 8) - 1) & 0x3ff) - | ((((mode->crtc_hdisplay / 8) - 1) & 0x1ff) << 16)); - - hsync_wid = (mode->crtc_hsync_end - mode->crtc_hsync_start) / 8; - if (!hsync_wid) - hsync_wid = 1; - hsync_start = mode->crtc_hsync_start - 8; - - fp_h_sync_strt_wid = ((hsync_start & 0x1fff) - | ((hsync_wid & 0x3f) << 16) - | ((mode->flags & DRM_MODE_FLAG_NHSYNC) - ? RADEON_CRTC_H_SYNC_POL - : 0)); - - fp_crtc_v_total_disp = (((mode->crtc_vtotal - 1) & 0xffff) - | ((mode->crtc_vdisplay - 1) << 16)); - - vsync_wid = mode->crtc_vsync_end - mode->crtc_vsync_start; - if (!vsync_wid) - vsync_wid = 1; - - fp_v_sync_strt_wid = (((mode->crtc_vsync_start - 1) & 0xfff) - | ((vsync_wid & 0x1f) << 16) - | ((mode->flags & DRM_MODE_FLAG_NVSYNC) - ? RADEON_CRTC_V_SYNC_POL - : 0)); - - fp_horz_vert_active = 0; - - if (native_mode->panel_xres == 0 || - native_mode->panel_yres == 0) { - hscale = false; - vscale = false; - } else { - if (xres > native_mode->panel_xres) - xres = native_mode->panel_xres; - if (yres > native_mode->panel_yres) - yres = native_mode->panel_yres; - - if (xres == native_mode->panel_xres) - hscale = false; - if (yres == native_mode->panel_yres) - vscale = false; - } - - if (radeon_encoder->flags & RADEON_USE_RMX) { - if (radeon_encoder->rmx_type != RMX_CENTER) { - if (!hscale) - fp_horz_stretch |= ((xres/8-1) << 16); - else { - inc = (fp_horz_stretch & RADEON_HORZ_AUTO_RATIO_INC) ? 1 : 0; - scale = ((xres + inc) * RADEON_HORZ_STRETCH_RATIO_MAX) - / native_mode->panel_xres + 1; - fp_horz_stretch |= (((scale) & RADEON_HORZ_STRETCH_RATIO_MASK) | - RADEON_HORZ_STRETCH_BLEND | - RADEON_HORZ_STRETCH_ENABLE | - ((native_mode->panel_xres/8-1) << 16)); - } - - if (!vscale) - fp_vert_stretch |= ((yres-1) << 12); - else { - inc = (fp_vert_stretch & RADEON_VERT_AUTO_RATIO_INC) ? 1 : 0; - scale = ((yres + inc) * RADEON_VERT_STRETCH_RATIO_MAX) - / native_mode->panel_yres + 1; - fp_vert_stretch |= (((scale) & RADEON_VERT_STRETCH_RATIO_MASK) | - RADEON_VERT_STRETCH_ENABLE | - RADEON_VERT_STRETCH_BLEND | - ((native_mode->panel_yres-1) << 12)); - } - } else if (radeon_encoder->rmx_type == RMX_CENTER) { - int blank_width; - - fp_horz_stretch |= ((xres/8-1) << 16); - fp_vert_stretch |= ((yres-1) << 12); - - crtc_more_cntl |= (RADEON_CRTC_AUTO_HORZ_CENTER_EN | - RADEON_CRTC_AUTO_VERT_CENTER_EN); - - blank_width = (mode->crtc_hblank_end - mode->crtc_hblank_start) / 8; - if (blank_width > 110) - blank_width = 110; - - fp_crtc_h_total_disp = (((blank_width) & 0x3ff) - | ((((mode->crtc_hdisplay / 8) - 1) & 0x1ff) << 16)); - - hsync_wid = (mode->crtc_hsync_end - mode->crtc_hsync_start) / 8; - if (!hsync_wid) - hsync_wid = 1; - - fp_h_sync_strt_wid = ((((mode->crtc_hsync_start - mode->crtc_hblank_start) / 8) & 0x1fff) - | ((hsync_wid & 0x3f) << 16) - | ((mode->flags & DRM_MODE_FLAG_NHSYNC) - ? RADEON_CRTC_H_SYNC_POL - : 0)); - - fp_crtc_v_total_disp = (((mode->crtc_vblank_end - mode->crtc_vblank_start) & 0xffff) - | ((mode->crtc_vdisplay - 1) << 16)); - - vsync_wid = mode->crtc_vsync_end - mode->crtc_vsync_start; - if (!vsync_wid) - vsync_wid = 1; - - fp_v_sync_strt_wid = ((((mode->crtc_vsync_start - mode->crtc_vblank_start) & 0xfff) - | ((vsync_wid & 0x1f) << 16) - | ((mode->flags & DRM_MODE_FLAG_NVSYNC) - ? RADEON_CRTC_V_SYNC_POL - : 0))); - - fp_horz_vert_active = (((native_mode->panel_yres) & 0xfff) | - (((native_mode->panel_xres / 8) & 0x1ff) << 16)); - } - } else { - fp_horz_stretch |= ((xres/8-1) << 16); - fp_vert_stretch |= ((yres-1) << 12); - } - - WREG32(RADEON_FP_HORZ_STRETCH, fp_horz_stretch); - WREG32(RADEON_FP_VERT_STRETCH, fp_vert_stretch); - WREG32(RADEON_CRTC_MORE_CNTL, crtc_more_cntl); - WREG32(RADEON_FP_HORZ_VERT_ACTIVE, fp_horz_vert_active); - WREG32(RADEON_FP_H_SYNC_STRT_WID, fp_h_sync_strt_wid); - WREG32(RADEON_FP_V_SYNC_STRT_WID, fp_v_sync_strt_wid); - WREG32(RADEON_FP_CRTC_H_TOTAL_DISP, fp_crtc_h_total_disp); - WREG32(RADEON_FP_CRTC_V_TOTAL_DISP, fp_crtc_v_total_disp); - -} - static void radeon_legacy_lvds_dpms(struct drm_encoder *encoder, int mode) { struct drm_device *dev = encoder->dev; @@ -287,9 +123,6 @@ static void radeon_legacy_lvds_mode_set(struct drm_encoder *encoder, DRM_DEBUG("\n"); - if (radeon_crtc->crtc_id == 0) - radeon_legacy_rmx_mode_set(encoder, mode, adjusted_mode); - lvds_pll_cntl = RREG32(RADEON_LVDS_PLL_CNTL); lvds_pll_cntl &= ~RADEON_LVDS_PLL_EN; @@ -318,7 +151,7 @@ static void radeon_legacy_lvds_mode_set(struct drm_encoder *encoder, if (radeon_crtc->crtc_id == 0) { if (ASIC_IS_R300(rdev)) { - if (radeon_encoder->flags & RADEON_USE_RMX) + if (radeon_encoder->rmx_type != RMX_OFF) lvds_pll_cntl |= R300_LVDS_SRC_SEL_RMX; } else lvds_gen_cntl &= ~RADEON_LVDS_SEL_CRTC2; @@ -350,8 +183,6 @@ static bool radeon_legacy_lvds_mode_fixup(struct drm_encoder *encoder, drm_mode_set_crtcinfo(adjusted_mode, 0); - radeon_encoder->flags &= ~RADEON_USE_RMX; - if (radeon_encoder->rmx_type != RMX_OFF) radeon_rmx_mode_fixup(encoder, mode, adjusted_mode); @@ -455,9 +286,6 @@ static void radeon_legacy_primary_dac_mode_set(struct drm_encoder *encoder, DRM_DEBUG("\n"); - if (radeon_crtc->crtc_id == 0) - radeon_legacy_rmx_mode_set(encoder, mode, adjusted_mode); - if (radeon_crtc->crtc_id == 0) { if (rdev->family == CHIP_R200 || ASIC_IS_R300(rdev)) { disp_output_cntl = RREG32(RADEON_DISP_OUTPUT_CNTL) & @@ -653,9 +481,6 @@ static void radeon_legacy_tmds_int_mode_set(struct drm_encoder *encoder, DRM_DEBUG("\n"); - if (radeon_crtc->crtc_id == 0) - radeon_legacy_rmx_mode_set(encoder, mode, adjusted_mode); - tmp = tmds_pll_cntl = RREG32(RADEON_TMDS_PLL_CNTL); tmp &= 0xfffff; if (rdev->family == CHIP_RV280) { @@ -711,7 +536,7 @@ static void radeon_legacy_tmds_int_mode_set(struct drm_encoder *encoder, if (radeon_crtc->crtc_id == 0) { if (ASIC_IS_R300(rdev) || rdev->family == CHIP_R200) { fp_gen_cntl &= ~R200_FP_SOURCE_SEL_MASK; - if (radeon_encoder->flags & RADEON_USE_RMX) + if (radeon_encoder->rmx_type != RMX_OFF) fp_gen_cntl |= R200_FP_SOURCE_SEL_RMX; else fp_gen_cntl |= R200_FP_SOURCE_SEL_CRTC1; @@ -820,9 +645,6 @@ static void radeon_legacy_tmds_ext_mode_set(struct drm_encoder *encoder, DRM_DEBUG("\n"); - if (radeon_crtc->crtc_id == 0) - radeon_legacy_rmx_mode_set(encoder, mode, adjusted_mode); - if (rdev->is_atom_bios) { radeon_encoder->pixel_clock = adjusted_mode->clock; atombios_external_tmds_setup(encoder, ATOM_ENABLE); @@ -856,7 +678,7 @@ static void radeon_legacy_tmds_ext_mode_set(struct drm_encoder *encoder, if (radeon_crtc->crtc_id == 0) { if ((rdev->family == CHIP_R200) || ASIC_IS_R300(rdev)) { fp2_gen_cntl &= ~R200_FP2_SOURCE_SEL_MASK; - if (radeon_encoder->flags & RADEON_USE_RMX) + if (radeon_encoder->rmx_type != RMX_OFF) fp2_gen_cntl |= R200_FP2_SOURCE_SEL_RMX; else fp2_gen_cntl |= R200_FP2_SOURCE_SEL_CRTC1; @@ -1014,9 +836,6 @@ static void radeon_legacy_tv_dac_mode_set(struct drm_encoder *encoder, DRM_DEBUG("\n"); - if (radeon_crtc->crtc_id == 0) - radeon_legacy_rmx_mode_set(encoder, mode, adjusted_mode); - if (rdev->family != CHIP_R200) { tv_dac_cntl = RREG32(RADEON_TV_DAC_CNTL); if (rdev->family == CHIP_R420 || @@ -1243,6 +1062,7 @@ radeon_add_legacy_encoder(struct drm_device *dev, uint32_t encoder_id, uint32_t radeon_encoder->encoder_id = encoder_id; radeon_encoder->devices = supported_device; + radeon_encoder->rmx_type = RMX_OFF; switch (radeon_encoder->encoder_id) { case ENCODER_OBJECT_ID_INTERNAL_LVDS: diff --git a/drivers/gpu/drm/radeon/radeon_mode.h b/drivers/gpu/drm/radeon/radeon_mode.h index 9173b687462..3b09a1f2d8f 100644 --- a/drivers/gpu/drm/radeon/radeon_mode.h +++ b/drivers/gpu/drm/radeon/radeon_mode.h @@ -36,6 +36,9 @@ #include <linux/i2c.h> #include <linux/i2c-id.h> #include <linux/i2c-algo-bit.h> +#include "radeon_fixed.h" + +struct radeon_device; #define to_radeon_crtc(x) container_of(x, struct radeon_crtc, base) #define to_radeon_connector(x) container_of(x, struct radeon_connector, base) @@ -124,6 +127,7 @@ struct radeon_tmds_pll { #define RADEON_PLL_PREFER_LOW_POST_DIV (1 << 8) #define RADEON_PLL_PREFER_HIGH_POST_DIV (1 << 9) #define RADEON_PLL_USE_FRAC_FB_DIV (1 << 10) +#define RADEON_PLL_PREFER_CLOSEST_LOWER (1 << 11) struct radeon_pll { uint16_t reference_freq; @@ -170,6 +174,18 @@ struct radeon_mode_info { struct atom_context *atom_context; enum radeon_connector_table connector_table; bool mode_config_initialized; + struct radeon_crtc *crtcs[2]; +}; + +struct radeon_native_mode { + /* preferred mode */ + uint32_t panel_xres, panel_yres; + uint32_t hoverplus, hsync_width; + uint32_t hblank; + uint32_t voverplus, vsync_width; + uint32_t vblank; + uint32_t dotclock; + uint32_t flags; }; struct radeon_crtc { @@ -185,19 +201,13 @@ struct radeon_crtc { uint64_t cursor_addr; int cursor_width; int cursor_height; -}; - -#define RADEON_USE_RMX 1 - -struct radeon_native_mode { - /* preferred mode */ - uint32_t panel_xres, panel_yres; - uint32_t hoverplus, hsync_width; - uint32_t hblank; - uint32_t voverplus, vsync_width; - uint32_t vblank; - uint32_t dotclock; - uint32_t flags; + uint32_t legacy_display_base_addr; + uint32_t legacy_cursor_offset; + enum radeon_rmx_type rmx_type; + uint32_t devices; + fixed20_12 vsc; + fixed20_12 hsc; + struct radeon_native_mode native_mode; }; struct radeon_encoder_primary_dac { @@ -383,16 +393,9 @@ void radeon_enc_destroy(struct drm_encoder *encoder); void radeon_copy_fb(struct drm_device *dev, struct drm_gem_object *dst_obj); void radeon_combios_asic_init(struct drm_device *dev); extern int radeon_static_clocks_init(struct drm_device *dev); -void radeon_init_disp_bw_legacy(struct drm_device *dev, - struct drm_display_mode *mode1, - uint32_t pixel_bytes1, - struct drm_display_mode *mode2, - uint32_t pixel_bytes2); -void radeon_init_disp_bw_avivo(struct drm_device *dev, - struct drm_display_mode *mode1, - uint32_t pixel_bytes1, - struct drm_display_mode *mode2, - uint32_t pixel_bytes2); -void radeon_init_disp_bandwidth(struct drm_device *dev); +bool radeon_crtc_scaling_mode_fixup(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode); +void atom_rv515_force_tv_scaler(struct radeon_device *rdev); #endif diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c index bac0d06c52a..dd9ac2fed6d 100644 --- a/drivers/gpu/drm/radeon/radeon_object.c +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -44,6 +44,9 @@ struct radeon_object { uint64_t gpu_addr; void *kptr; bool is_iomem; + uint32_t tiling_flags; + uint32_t pitch; + int surface_reg; }; int radeon_ttm_init(struct radeon_device *rdev); @@ -70,6 +73,7 @@ static void radeon_ttm_object_object_destroy(struct ttm_buffer_object *tobj) robj = container_of(tobj, struct radeon_object, tobj); list_del_init(&robj->list); + radeon_object_clear_surface_reg(robj); kfree(robj); } @@ -99,16 +103,16 @@ static inline uint32_t radeon_object_flags_from_domain(uint32_t domain) { uint32_t flags = 0; if (domain & RADEON_GEM_DOMAIN_VRAM) { - flags |= TTM_PL_FLAG_VRAM; + flags |= TTM_PL_FLAG_VRAM | TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED; } if (domain & RADEON_GEM_DOMAIN_GTT) { - flags |= TTM_PL_FLAG_TT; + flags |= TTM_PL_FLAG_TT | TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED; } if (domain & RADEON_GEM_DOMAIN_CPU) { - flags |= TTM_PL_FLAG_SYSTEM; + flags |= TTM_PL_FLAG_SYSTEM | TTM_PL_MASK_CACHING; } if (!flags) { - flags |= TTM_PL_FLAG_SYSTEM; + flags |= TTM_PL_FLAG_SYSTEM | TTM_PL_MASK_CACHING; } return flags; } @@ -141,6 +145,7 @@ int radeon_object_create(struct radeon_device *rdev, } robj->rdev = rdev; robj->gobj = gobj; + robj->surface_reg = -1; INIT_LIST_HEAD(&robj->list); flags = radeon_object_flags_from_domain(domain); @@ -304,7 +309,7 @@ int radeon_object_wait(struct radeon_object *robj) } spin_lock(&robj->tobj.lock); if (robj->tobj.sync_obj) { - r = ttm_bo_wait(&robj->tobj, true, false, false); + r = ttm_bo_wait(&robj->tobj, true, true, false); } spin_unlock(&robj->tobj.lock); radeon_object_unreserve(robj); @@ -403,7 +408,6 @@ int radeon_object_list_validate(struct list_head *head, void *fence) struct radeon_object *robj; struct radeon_fence *old_fence = NULL; struct list_head *i; - uint32_t flags; int r; r = radeon_object_list_reserve(head); @@ -414,27 +418,25 @@ int radeon_object_list_validate(struct list_head *head, void *fence) list_for_each(i, head) { lobj = list_entry(i, struct radeon_object_list, list); robj = lobj->robj; - if (lobj->wdomain) { - flags = radeon_object_flags_from_domain(lobj->wdomain); - flags |= TTM_PL_FLAG_TT; - } else { - flags = radeon_object_flags_from_domain(lobj->rdomain); - flags |= TTM_PL_FLAG_TT; - flags |= TTM_PL_FLAG_VRAM; - } if (!robj->pin_count) { - robj->tobj.proposed_placement = flags | TTM_PL_MASK_CACHING; + if (lobj->wdomain) { + robj->tobj.proposed_placement = + radeon_object_flags_from_domain(lobj->wdomain); + } else { + robj->tobj.proposed_placement = + radeon_object_flags_from_domain(lobj->rdomain); + } r = ttm_buffer_object_validate(&robj->tobj, robj->tobj.proposed_placement, true, false); if (unlikely(r)) { - radeon_object_list_unreserve(head); DRM_ERROR("radeon: failed to validate.\n"); return r; } radeon_object_gpu_addr(robj); } lobj->gpu_offset = robj->gpu_addr; + lobj->tiling_flags = robj->tiling_flags; if (fence) { old_fence = (struct radeon_fence *)robj->tobj.sync_obj; robj->tobj.sync_obj = radeon_fence_ref(fence); @@ -479,3 +481,127 @@ unsigned long radeon_object_size(struct radeon_object *robj) { return robj->tobj.num_pages << PAGE_SHIFT; } + +int radeon_object_get_surface_reg(struct radeon_object *robj) +{ + struct radeon_device *rdev = robj->rdev; + struct radeon_surface_reg *reg; + struct radeon_object *old_object; + int steal; + int i; + + if (!robj->tiling_flags) + return 0; + + if (robj->surface_reg >= 0) { + reg = &rdev->surface_regs[robj->surface_reg]; + i = robj->surface_reg; + goto out; + } + + steal = -1; + for (i = 0; i < RADEON_GEM_MAX_SURFACES; i++) { + + reg = &rdev->surface_regs[i]; + if (!reg->robj) + break; + + old_object = reg->robj; + if (old_object->pin_count == 0) + steal = i; + } + + /* if we are all out */ + if (i == RADEON_GEM_MAX_SURFACES) { + if (steal == -1) + return -ENOMEM; + /* find someone with a surface reg and nuke their BO */ + reg = &rdev->surface_regs[steal]; + old_object = reg->robj; + /* blow away the mapping */ + DRM_DEBUG("stealing surface reg %d from %p\n", steal, old_object); + ttm_bo_unmap_virtual(&old_object->tobj); + old_object->surface_reg = -1; + i = steal; + } + + robj->surface_reg = i; + reg->robj = robj; + +out: + radeon_set_surface_reg(rdev, i, robj->tiling_flags, robj->pitch, + robj->tobj.mem.mm_node->start << PAGE_SHIFT, + robj->tobj.num_pages << PAGE_SHIFT); + return 0; +} + +void radeon_object_clear_surface_reg(struct radeon_object *robj) +{ + struct radeon_device *rdev = robj->rdev; + struct radeon_surface_reg *reg; + + if (robj->surface_reg == -1) + return; + + reg = &rdev->surface_regs[robj->surface_reg]; + radeon_clear_surface_reg(rdev, robj->surface_reg); + + reg->robj = NULL; + robj->surface_reg = -1; +} + +void radeon_object_set_tiling_flags(struct radeon_object *robj, + uint32_t tiling_flags, uint32_t pitch) +{ + robj->tiling_flags = tiling_flags; + robj->pitch = pitch; +} + +void radeon_object_get_tiling_flags(struct radeon_object *robj, + uint32_t *tiling_flags, + uint32_t *pitch) +{ + if (tiling_flags) + *tiling_flags = robj->tiling_flags; + if (pitch) + *pitch = robj->pitch; +} + +int radeon_object_check_tiling(struct radeon_object *robj, bool has_moved, + bool force_drop) +{ + if (!(robj->tiling_flags & RADEON_TILING_SURFACE)) + return 0; + + if (force_drop) { + radeon_object_clear_surface_reg(robj); + return 0; + } + + if (robj->tobj.mem.mem_type != TTM_PL_VRAM) { + if (!has_moved) + return 0; + + if (robj->surface_reg >= 0) + radeon_object_clear_surface_reg(robj); + return 0; + } + + if ((robj->surface_reg >= 0) && !has_moved) + return 0; + + return radeon_object_get_surface_reg(robj); +} + +void radeon_bo_move_notify(struct ttm_buffer_object *bo, + struct ttm_mem_reg *mem) +{ + struct radeon_object *robj = container_of(bo, struct radeon_object, tobj); + radeon_object_check_tiling(robj, 0, 1); +} + +void radeon_bo_fault_reserve_notify(struct ttm_buffer_object *bo) +{ + struct radeon_object *robj = container_of(bo, struct radeon_object, tobj); + radeon_object_check_tiling(robj, 0, 0); +} diff --git a/drivers/gpu/drm/radeon/radeon_ring.c b/drivers/gpu/drm/radeon/radeon_ring.c index a853261d188..60d159308b8 100644 --- a/drivers/gpu/drm/radeon/radeon_ring.c +++ b/drivers/gpu/drm/radeon/radeon_ring.c @@ -126,32 +126,19 @@ static void radeon_ib_align(struct radeon_device *rdev, struct radeon_ib *ib) } } -static void radeon_ib_cpu_flush(struct radeon_device *rdev, - struct radeon_ib *ib) -{ - unsigned long tmp; - unsigned i; - - /* To force CPU cache flush ugly but seems reliable */ - for (i = 0; i < ib->length_dw; i += (rdev->cp.align_mask + 1)) { - tmp = readl(&ib->ptr[i]); - } -} - int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib) { int r = 0; mutex_lock(&rdev->ib_pool.mutex); radeon_ib_align(rdev, ib); - radeon_ib_cpu_flush(rdev, ib); if (!ib->length_dw || !rdev->cp.ready) { /* TODO: Nothings in the ib we should report. */ mutex_unlock(&rdev->ib_pool.mutex); DRM_ERROR("radeon: couldn't schedule IB(%lu).\n", ib->idx); return -EINVAL; } - /* 64 dwords should be enought for fence too */ + /* 64 dwords should be enough for fence too */ r = radeon_ring_lock(rdev, 64); if (r) { DRM_ERROR("radeon: scheduling IB failled (%d).\n", r); diff --git a/drivers/gpu/drm/radeon/radeon_share.h b/drivers/gpu/drm/radeon/radeon_share.h new file mode 100644 index 00000000000..63a773578f1 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_share.h @@ -0,0 +1,39 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RADEON_SHARE_H__ +#define __RADEON_SHARE_H__ + +void r100_vram_init_sizes(struct radeon_device *rdev); + +void rs690_line_buffer_adjust(struct radeon_device *rdev, + struct drm_display_mode *mode1, + struct drm_display_mode *mode2); + +void rv515_bandwidth_avivo_update(struct radeon_device *rdev); + +#endif diff --git a/drivers/gpu/drm/radeon/radeon_test.c b/drivers/gpu/drm/radeon/radeon_test.c new file mode 100644 index 00000000000..03c33cf4e14 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_test.c @@ -0,0 +1,209 @@ +/* + * Copyright 2009 VMware, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Michel Dänzer + */ +#include <drm/drmP.h> +#include <drm/radeon_drm.h> +#include "radeon_reg.h" +#include "radeon.h" + + +/* Test BO GTT->VRAM and VRAM->GTT GPU copies across the whole GTT aperture */ +void radeon_test_moves(struct radeon_device *rdev) +{ + struct radeon_object *vram_obj = NULL; + struct radeon_object **gtt_obj = NULL; + struct radeon_fence *fence = NULL; + uint64_t gtt_addr, vram_addr; + unsigned i, n, size; + int r; + + size = 1024 * 1024; + + /* Number of tests = + * (Total GTT - IB pool - writeback page - ring buffer) / test size + */ + n = (rdev->mc.gtt_size - RADEON_IB_POOL_SIZE*64*1024 - 4096 - + rdev->cp.ring_size) / size; + + gtt_obj = kzalloc(n * sizeof(*gtt_obj), GFP_KERNEL); + if (!gtt_obj) { + DRM_ERROR("Failed to allocate %d pointers\n", n); + r = 1; + goto out_cleanup; + } + + r = radeon_object_create(rdev, NULL, size, true, RADEON_GEM_DOMAIN_VRAM, + false, &vram_obj); + if (r) { + DRM_ERROR("Failed to create VRAM object\n"); + goto out_cleanup; + } + + r = radeon_object_pin(vram_obj, RADEON_GEM_DOMAIN_VRAM, &vram_addr); + if (r) { + DRM_ERROR("Failed to pin VRAM object\n"); + goto out_cleanup; + } + + for (i = 0; i < n; i++) { + void *gtt_map, *vram_map; + void **gtt_start, **gtt_end; + void **vram_start, **vram_end; + + r = radeon_object_create(rdev, NULL, size, true, + RADEON_GEM_DOMAIN_GTT, false, gtt_obj + i); + if (r) { + DRM_ERROR("Failed to create GTT object %d\n", i); + goto out_cleanup; + } + + r = radeon_object_pin(gtt_obj[i], RADEON_GEM_DOMAIN_GTT, >t_addr); + if (r) { + DRM_ERROR("Failed to pin GTT object %d\n", i); + goto out_cleanup; + } + + r = radeon_object_kmap(gtt_obj[i], >t_map); + if (r) { + DRM_ERROR("Failed to map GTT object %d\n", i); + goto out_cleanup; + } + + for (gtt_start = gtt_map, gtt_end = gtt_map + size; + gtt_start < gtt_end; + gtt_start++) + *gtt_start = gtt_start; + + radeon_object_kunmap(gtt_obj[i]); + + r = radeon_fence_create(rdev, &fence); + if (r) { + DRM_ERROR("Failed to create GTT->VRAM fence %d\n", i); + goto out_cleanup; + } + + r = radeon_copy(rdev, gtt_addr, vram_addr, size / 4096, fence); + if (r) { + DRM_ERROR("Failed GTT->VRAM copy %d\n", i); + goto out_cleanup; + } + + r = radeon_fence_wait(fence, false); + if (r) { + DRM_ERROR("Failed to wait for GTT->VRAM fence %d\n", i); + goto out_cleanup; + } + + radeon_fence_unref(&fence); + + r = radeon_object_kmap(vram_obj, &vram_map); + if (r) { + DRM_ERROR("Failed to map VRAM object after copy %d\n", i); + goto out_cleanup; + } + + for (gtt_start = gtt_map, gtt_end = gtt_map + size, + vram_start = vram_map, vram_end = vram_map + size; + vram_start < vram_end; + gtt_start++, vram_start++) { + if (*vram_start != gtt_start) { + DRM_ERROR("Incorrect GTT->VRAM copy %d: Got 0x%p, " + "expected 0x%p (GTT map 0x%p-0x%p)\n", + i, *vram_start, gtt_start, gtt_map, + gtt_end); + radeon_object_kunmap(vram_obj); + goto out_cleanup; + } + *vram_start = vram_start; + } + + radeon_object_kunmap(vram_obj); + + r = radeon_fence_create(rdev, &fence); + if (r) { + DRM_ERROR("Failed to create VRAM->GTT fence %d\n", i); + goto out_cleanup; + } + + r = radeon_copy(rdev, vram_addr, gtt_addr, size / 4096, fence); + if (r) { + DRM_ERROR("Failed VRAM->GTT copy %d\n", i); + goto out_cleanup; + } + + r = radeon_fence_wait(fence, false); + if (r) { + DRM_ERROR("Failed to wait for VRAM->GTT fence %d\n", i); + goto out_cleanup; + } + + radeon_fence_unref(&fence); + + r = radeon_object_kmap(gtt_obj[i], >t_map); + if (r) { + DRM_ERROR("Failed to map GTT object after copy %d\n", i); + goto out_cleanup; + } + + for (gtt_start = gtt_map, gtt_end = gtt_map + size, + vram_start = vram_map, vram_end = vram_map + size; + gtt_start < gtt_end; + gtt_start++, vram_start++) { + if (*gtt_start != vram_start) { + DRM_ERROR("Incorrect VRAM->GTT copy %d: Got 0x%p, " + "expected 0x%p (VRAM map 0x%p-0x%p)\n", + i, *gtt_start, vram_start, vram_map, + vram_end); + radeon_object_kunmap(gtt_obj[i]); + goto out_cleanup; + } + } + + radeon_object_kunmap(gtt_obj[i]); + + DRM_INFO("Tested GTT->VRAM and VRAM->GTT copy for GTT offset 0x%llx\n", + gtt_addr - rdev->mc.gtt_location); + } + +out_cleanup: + if (vram_obj) { + radeon_object_unpin(vram_obj); + radeon_object_unref(&vram_obj); + } + if (gtt_obj) { + for (i = 0; i < n; i++) { + if (gtt_obj[i]) { + radeon_object_unpin(gtt_obj[i]); + radeon_object_unref(>t_obj[i]); + } + } + kfree(gtt_obj); + } + if (fence) { + radeon_fence_unref(&fence); + } + if (r) { + printk(KERN_WARNING "Error while testing BO move.\n"); + } +} + diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c index 1227a97f516..15c3531377e 100644 --- a/drivers/gpu/drm/radeon/radeon_ttm.c +++ b/drivers/gpu/drm/radeon/radeon_ttm.c @@ -355,23 +355,26 @@ static int radeon_bo_move(struct ttm_buffer_object *bo, if (!rdev->cp.ready) { /* use memcpy */ DRM_ERROR("CP is not ready use memcpy.\n"); - return ttm_bo_move_memcpy(bo, evict, no_wait, new_mem); + goto memcpy; } if (old_mem->mem_type == TTM_PL_VRAM && new_mem->mem_type == TTM_PL_SYSTEM) { - return radeon_move_vram_ram(bo, evict, interruptible, + r = radeon_move_vram_ram(bo, evict, interruptible, no_wait, new_mem); } else if (old_mem->mem_type == TTM_PL_SYSTEM && new_mem->mem_type == TTM_PL_VRAM) { - return radeon_move_ram_vram(bo, evict, interruptible, + r = radeon_move_ram_vram(bo, evict, interruptible, no_wait, new_mem); } else { r = radeon_move_blit(bo, evict, no_wait, new_mem, old_mem); - if (unlikely(r)) { - return r; - } } + + if (r) { +memcpy: + r = ttm_bo_move_memcpy(bo, evict, no_wait, new_mem); + } + return r; } @@ -429,6 +432,8 @@ static struct ttm_bo_driver radeon_bo_driver = { .sync_obj_flush = &radeon_sync_obj_flush, .sync_obj_unref = &radeon_sync_obj_unref, .sync_obj_ref = &radeon_sync_obj_ref, + .move_notify = &radeon_bo_move_notify, + .fault_reserve_notify = &radeon_bo_fault_reserve_notify, }; int radeon_ttm_init(struct radeon_device *rdev) @@ -442,13 +447,14 @@ int radeon_ttm_init(struct radeon_device *rdev) /* No others user of address space so set it to 0 */ r = ttm_bo_device_init(&rdev->mman.bdev, rdev->mman.mem_global_ref.object, - &radeon_bo_driver, DRM_FILE_PAGE_OFFSET); + &radeon_bo_driver, DRM_FILE_PAGE_OFFSET, + rdev->need_dma32); if (r) { DRM_ERROR("failed initializing buffer object driver(%d).\n", r); return r; } r = ttm_bo_init_mm(&rdev->mman.bdev, TTM_PL_VRAM, 0, - ((rdev->mc.aper_size) >> PAGE_SHIFT)); + ((rdev->mc.real_vram_size) >> PAGE_SHIFT)); if (r) { DRM_ERROR("Failed initializing VRAM heap.\n"); return r; @@ -465,7 +471,7 @@ int radeon_ttm_init(struct radeon_device *rdev) return r; } DRM_INFO("radeon: %uM of VRAM memory ready\n", - rdev->mc.vram_size / (1024 * 1024)); + rdev->mc.real_vram_size / (1024 * 1024)); r = ttm_bo_init_mm(&rdev->mman.bdev, TTM_PL_TT, 0, ((rdev->mc.gtt_size) >> PAGE_SHIFT)); if (r) { diff --git a/drivers/gpu/drm/radeon/rs400.c b/drivers/gpu/drm/radeon/rs400.c index cc074b5a8f7..b29affd9c5d 100644 --- a/drivers/gpu/drm/radeon/rs400.c +++ b/drivers/gpu/drm/radeon/rs400.c @@ -29,6 +29,7 @@ #include <drm/drmP.h> #include "radeon_reg.h" #include "radeon.h" +#include "radeon_share.h" /* rs400,rs480 depends on : */ void r100_hdp_reset(struct radeon_device *rdev); @@ -164,7 +165,9 @@ int rs400_gart_enable(struct radeon_device *rdev) WREG32(RADEON_BUS_CNTL, tmp); } /* Table should be in 32bits address space so ignore bits above. */ - tmp = rdev->gart.table_addr & 0xfffff000; + tmp = (u32)rdev->gart.table_addr & 0xfffff000; + tmp |= (upper_32_bits(rdev->gart.table_addr) & 0xff) << 4; + WREG32_MC(RS480_GART_BASE, tmp); /* TODO: more tweaking here */ WREG32_MC(RS480_GART_FEATURE_ID, @@ -201,10 +204,17 @@ void rs400_gart_disable(struct radeon_device *rdev) int rs400_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr) { + uint32_t entry; + if (i < 0 || i > rdev->gart.num_gpu_pages) { return -EINVAL; } - rdev->gart.table.ram.ptr[i] = cpu_to_le32(((uint32_t)addr) | 0xC); + + entry = (lower_32_bits(addr) & PAGE_MASK) | + ((upper_32_bits(addr) & 0xff) << 4) | + 0xc; + entry = cpu_to_le32(entry); + rdev->gart.table.ram.ptr[i] = entry; return 0; } @@ -223,10 +233,9 @@ int rs400_mc_init(struct radeon_device *rdev) rs400_gpu_init(rdev); rs400_gart_disable(rdev); - rdev->mc.gtt_location = rdev->mc.vram_size; + rdev->mc.gtt_location = rdev->mc.mc_vram_size; rdev->mc.gtt_location += (rdev->mc.gtt_size - 1); rdev->mc.gtt_location &= ~(rdev->mc.gtt_size - 1); - rdev->mc.vram_location = 0xFFFFFFFFUL; r = radeon_mc_setup(rdev); if (r) { return r; @@ -238,7 +247,7 @@ int rs400_mc_init(struct radeon_device *rdev) "programming pipes. Bad things might happen.\n"); } - tmp = rdev->mc.vram_location + rdev->mc.vram_size - 1; + tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; tmp = REG_SET(RADEON_MC_FB_TOP, tmp >> 16); tmp |= REG_SET(RADEON_MC_FB_START, rdev->mc.vram_location >> 16); WREG32(RADEON_MC_FB_LOCATION, tmp); @@ -284,21 +293,12 @@ void rs400_gpu_init(struct radeon_device *rdev) */ void rs400_vram_info(struct radeon_device *rdev) { - uint32_t tom; - rs400_gart_adjust_size(rdev); /* DDR for all card after R300 & IGP */ rdev->mc.vram_is_ddr = true; rdev->mc.vram_width = 128; - /* read NB_TOM to get the amount of ram stolen for the GPU */ - tom = RREG32(RADEON_NB_TOM); - rdev->mc.vram_size = (((tom >> 16) - (tom & 0xffff) + 1) << 16); - WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.vram_size); - - /* Could aper size report 0 ? */ - rdev->mc.aper_base = drm_get_resource_start(rdev->ddev, 0); - rdev->mc.aper_size = drm_get_resource_len(rdev->ddev, 0); + r100_vram_init_sizes(rdev); } diff --git a/drivers/gpu/drm/radeon/rs600.c b/drivers/gpu/drm/radeon/rs600.c index ab0c967553e..bbea6dee4a9 100644 --- a/drivers/gpu/drm/radeon/rs600.c +++ b/drivers/gpu/drm/radeon/rs600.c @@ -223,7 +223,7 @@ int rs600_mc_init(struct radeon_device *rdev) printk(KERN_WARNING "Failed to wait MC idle while " "programming pipes. Bad things might happen.\n"); } - tmp = rdev->mc.vram_location + rdev->mc.vram_size - 1; + tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; tmp = REG_SET(RS600_MC_FB_TOP, tmp >> 16); tmp |= REG_SET(RS600_MC_FB_START, rdev->mc.vram_location >> 16); WREG32_MC(RS600_MC_FB_LOCATION, tmp); @@ -301,6 +301,11 @@ void rs600_vram_info(struct radeon_device *rdev) rdev->mc.vram_width = 128; } +void rs600_bandwidth_update(struct radeon_device *rdev) +{ + /* FIXME: implement, should this be like rs690 ? */ +} + /* * Indirect registers accessor diff --git a/drivers/gpu/drm/radeon/rs690.c b/drivers/gpu/drm/radeon/rs690.c index 79ba85042b5..839595b0072 100644 --- a/drivers/gpu/drm/radeon/rs690.c +++ b/drivers/gpu/drm/radeon/rs690.c @@ -28,6 +28,9 @@ #include "drmP.h" #include "radeon_reg.h" #include "radeon.h" +#include "rs690r.h" +#include "atom.h" +#include "atom-bits.h" /* rs690,rs740 depends on : */ void r100_hdp_reset(struct radeon_device *rdev); @@ -64,7 +67,7 @@ int rs690_mc_init(struct radeon_device *rdev) rs400_gart_disable(rdev); /* Setup GPU memory space */ - rdev->mc.gtt_location = rdev->mc.vram_size; + rdev->mc.gtt_location = rdev->mc.mc_vram_size; rdev->mc.gtt_location += (rdev->mc.gtt_size - 1); rdev->mc.gtt_location &= ~(rdev->mc.gtt_size - 1); rdev->mc.vram_location = 0xFFFFFFFFUL; @@ -79,7 +82,7 @@ int rs690_mc_init(struct radeon_device *rdev) printk(KERN_WARNING "Failed to wait MC idle while " "programming pipes. Bad things might happen.\n"); } - tmp = rdev->mc.vram_location + rdev->mc.vram_size - 1; + tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; tmp = REG_SET(RS690_MC_FB_TOP, tmp >> 16); tmp |= REG_SET(RS690_MC_FB_START, rdev->mc.vram_location >> 16); WREG32_MC(RS690_MCCFG_FB_LOCATION, tmp); @@ -138,9 +141,82 @@ void rs690_gpu_init(struct radeon_device *rdev) /* * VRAM info. */ +void rs690_pm_info(struct radeon_device *rdev) +{ + int index = GetIndexIntoMasterTable(DATA, IntegratedSystemInfo); + struct _ATOM_INTEGRATED_SYSTEM_INFO *info; + struct _ATOM_INTEGRATED_SYSTEM_INFO_V2 *info_v2; + void *ptr; + uint16_t data_offset; + uint8_t frev, crev; + fixed20_12 tmp; + + atom_parse_data_header(rdev->mode_info.atom_context, index, NULL, + &frev, &crev, &data_offset); + ptr = rdev->mode_info.atom_context->bios + data_offset; + info = (struct _ATOM_INTEGRATED_SYSTEM_INFO *)ptr; + info_v2 = (struct _ATOM_INTEGRATED_SYSTEM_INFO_V2 *)ptr; + /* Get various system informations from bios */ + switch (crev) { + case 1: + tmp.full = rfixed_const(100); + rdev->pm.igp_sideport_mclk.full = rfixed_const(info->ulBootUpMemoryClock); + rdev->pm.igp_sideport_mclk.full = rfixed_div(rdev->pm.igp_sideport_mclk, tmp); + rdev->pm.igp_system_mclk.full = rfixed_const(le16_to_cpu(info->usK8MemoryClock)); + rdev->pm.igp_ht_link_clk.full = rfixed_const(le16_to_cpu(info->usFSBClock)); + rdev->pm.igp_ht_link_width.full = rfixed_const(info->ucHTLinkWidth); + break; + case 2: + tmp.full = rfixed_const(100); + rdev->pm.igp_sideport_mclk.full = rfixed_const(info_v2->ulBootUpSidePortClock); + rdev->pm.igp_sideport_mclk.full = rfixed_div(rdev->pm.igp_sideport_mclk, tmp); + rdev->pm.igp_system_mclk.full = rfixed_const(info_v2->ulBootUpUMAClock); + rdev->pm.igp_system_mclk.full = rfixed_div(rdev->pm.igp_system_mclk, tmp); + rdev->pm.igp_ht_link_clk.full = rfixed_const(info_v2->ulHTLinkFreq); + rdev->pm.igp_ht_link_clk.full = rfixed_div(rdev->pm.igp_ht_link_clk, tmp); + rdev->pm.igp_ht_link_width.full = rfixed_const(le16_to_cpu(info_v2->usMinHTLinkWidth)); + break; + default: + tmp.full = rfixed_const(100); + /* We assume the slower possible clock ie worst case */ + /* DDR 333Mhz */ + rdev->pm.igp_sideport_mclk.full = rfixed_const(333); + /* FIXME: system clock ? */ + rdev->pm.igp_system_mclk.full = rfixed_const(100); + rdev->pm.igp_system_mclk.full = rfixed_div(rdev->pm.igp_system_mclk, tmp); + rdev->pm.igp_ht_link_clk.full = rfixed_const(200); + rdev->pm.igp_ht_link_width.full = rfixed_const(8); + DRM_ERROR("No integrated system info for your GPU, using safe default\n"); + break; + } + /* Compute various bandwidth */ + /* k8_bandwidth = (memory_clk / 2) * 2 * 8 * 0.5 = memory_clk * 4 */ + tmp.full = rfixed_const(4); + rdev->pm.k8_bandwidth.full = rfixed_mul(rdev->pm.igp_system_mclk, tmp); + /* ht_bandwidth = ht_clk * 2 * ht_width / 8 * 0.8 + * = ht_clk * ht_width / 5 + */ + tmp.full = rfixed_const(5); + rdev->pm.ht_bandwidth.full = rfixed_mul(rdev->pm.igp_ht_link_clk, + rdev->pm.igp_ht_link_width); + rdev->pm.ht_bandwidth.full = rfixed_div(rdev->pm.ht_bandwidth, tmp); + if (tmp.full < rdev->pm.max_bandwidth.full) { + /* HT link is a limiting factor */ + rdev->pm.max_bandwidth.full = tmp.full; + } + /* sideport_bandwidth = (sideport_clk / 2) * 2 * 2 * 0.7 + * = (sideport_clk * 14) / 10 + */ + tmp.full = rfixed_const(14); + rdev->pm.sideport_bandwidth.full = rfixed_mul(rdev->pm.igp_sideport_mclk, tmp); + tmp.full = rfixed_const(10); + rdev->pm.sideport_bandwidth.full = rfixed_div(rdev->pm.sideport_bandwidth, tmp); +} + void rs690_vram_info(struct radeon_device *rdev) { uint32_t tmp; + fixed20_12 a; rs400_gart_adjust_size(rdev); /* DDR for all card after R300 & IGP */ @@ -152,12 +228,409 @@ void rs690_vram_info(struct radeon_device *rdev) } else { rdev->mc.vram_width = 64; } - rdev->mc.vram_size = RREG32(RADEON_CONFIG_MEMSIZE); + rdev->mc.real_vram_size = RREG32(RADEON_CONFIG_MEMSIZE); + rdev->mc.mc_vram_size = rdev->mc.real_vram_size; rdev->mc.aper_base = drm_get_resource_start(rdev->ddev, 0); rdev->mc.aper_size = drm_get_resource_len(rdev->ddev, 0); + rs690_pm_info(rdev); + /* FIXME: we should enforce default clock in case GPU is not in + * default setup + */ + a.full = rfixed_const(100); + rdev->pm.sclk.full = rfixed_const(rdev->clock.default_sclk); + rdev->pm.sclk.full = rfixed_div(rdev->pm.sclk, a); + a.full = rfixed_const(16); + /* core_bandwidth = sclk(Mhz) * 16 */ + rdev->pm.core_bandwidth.full = rfixed_div(rdev->pm.sclk, a); +} + +void rs690_line_buffer_adjust(struct radeon_device *rdev, + struct drm_display_mode *mode1, + struct drm_display_mode *mode2) +{ + u32 tmp; + + /* + * Line Buffer Setup + * There is a single line buffer shared by both display controllers. + * DC_LB_MEMORY_SPLIT controls how that line buffer is shared between + * the display controllers. The paritioning can either be done + * manually or via one of four preset allocations specified in bits 1:0: + * 0 - line buffer is divided in half and shared between crtc + * 1 - D1 gets 3/4 of the line buffer, D2 gets 1/4 + * 2 - D1 gets the whole buffer + * 3 - D1 gets 1/4 of the line buffer, D2 gets 3/4 + * Setting bit 2 of DC_LB_MEMORY_SPLIT controls switches to manual + * allocation mode. In manual allocation mode, D1 always starts at 0, + * D1 end/2 is specified in bits 14:4; D2 allocation follows D1. + */ + tmp = RREG32(DC_LB_MEMORY_SPLIT) & ~DC_LB_MEMORY_SPLIT_MASK; + tmp &= ~DC_LB_MEMORY_SPLIT_SHIFT_MODE; + /* auto */ + if (mode1 && mode2) { + if (mode1->hdisplay > mode2->hdisplay) { + if (mode1->hdisplay > 2560) + tmp |= DC_LB_MEMORY_SPLIT_D1_3Q_D2_1Q; + else + tmp |= DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; + } else if (mode2->hdisplay > mode1->hdisplay) { + if (mode2->hdisplay > 2560) + tmp |= DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q; + else + tmp |= DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; + } else + tmp |= AVIVO_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; + } else if (mode1) { + tmp |= DC_LB_MEMORY_SPLIT_D1_ONLY; + } else if (mode2) { + tmp |= DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q; + } + WREG32(DC_LB_MEMORY_SPLIT, tmp); } +struct rs690_watermark { + u32 lb_request_fifo_depth; + fixed20_12 num_line_pair; + fixed20_12 estimated_width; + fixed20_12 worst_case_latency; + fixed20_12 consumption_rate; + fixed20_12 active_time; + fixed20_12 dbpp; + fixed20_12 priority_mark_max; + fixed20_12 priority_mark; + fixed20_12 sclk; +}; + +void rs690_crtc_bandwidth_compute(struct radeon_device *rdev, + struct radeon_crtc *crtc, + struct rs690_watermark *wm) +{ + struct drm_display_mode *mode = &crtc->base.mode; + fixed20_12 a, b, c; + fixed20_12 pclk, request_fifo_depth, tolerable_latency, estimated_width; + fixed20_12 consumption_time, line_time, chunk_time, read_delay_latency; + /* FIXME: detect IGP with sideport memory, i don't think there is any + * such product available + */ + bool sideport = false; + + if (!crtc->base.enabled) { + /* FIXME: wouldn't it better to set priority mark to maximum */ + wm->lb_request_fifo_depth = 4; + return; + } + + if (crtc->vsc.full > rfixed_const(2)) + wm->num_line_pair.full = rfixed_const(2); + else + wm->num_line_pair.full = rfixed_const(1); + + b.full = rfixed_const(mode->crtc_hdisplay); + c.full = rfixed_const(256); + a.full = rfixed_mul(wm->num_line_pair, b); + request_fifo_depth.full = rfixed_div(a, c); + if (a.full < rfixed_const(4)) { + wm->lb_request_fifo_depth = 4; + } else { + wm->lb_request_fifo_depth = rfixed_trunc(request_fifo_depth); + } + + /* Determine consumption rate + * pclk = pixel clock period(ns) = 1000 / (mode.clock / 1000) + * vtaps = number of vertical taps, + * vsc = vertical scaling ratio, defined as source/destination + * hsc = horizontal scaling ration, defined as source/destination + */ + a.full = rfixed_const(mode->clock); + b.full = rfixed_const(1000); + a.full = rfixed_div(a, b); + pclk.full = rfixed_div(b, a); + if (crtc->rmx_type != RMX_OFF) { + b.full = rfixed_const(2); + if (crtc->vsc.full > b.full) + b.full = crtc->vsc.full; + b.full = rfixed_mul(b, crtc->hsc); + c.full = rfixed_const(2); + b.full = rfixed_div(b, c); + consumption_time.full = rfixed_div(pclk, b); + } else { + consumption_time.full = pclk.full; + } + a.full = rfixed_const(1); + wm->consumption_rate.full = rfixed_div(a, consumption_time); + + + /* Determine line time + * LineTime = total time for one line of displayhtotal + * LineTime = total number of horizontal pixels + * pclk = pixel clock period(ns) + */ + a.full = rfixed_const(crtc->base.mode.crtc_htotal); + line_time.full = rfixed_mul(a, pclk); + + /* Determine active time + * ActiveTime = time of active region of display within one line, + * hactive = total number of horizontal active pixels + * htotal = total number of horizontal pixels + */ + a.full = rfixed_const(crtc->base.mode.crtc_htotal); + b.full = rfixed_const(crtc->base.mode.crtc_hdisplay); + wm->active_time.full = rfixed_mul(line_time, b); + wm->active_time.full = rfixed_div(wm->active_time, a); + + /* Maximun bandwidth is the minimun bandwidth of all component */ + rdev->pm.max_bandwidth = rdev->pm.core_bandwidth; + if (sideport) { + if (rdev->pm.max_bandwidth.full > rdev->pm.sideport_bandwidth.full && + rdev->pm.sideport_bandwidth.full) + rdev->pm.max_bandwidth = rdev->pm.sideport_bandwidth; + read_delay_latency.full = rfixed_const(370 * 800 * 1000); + read_delay_latency.full = rfixed_div(read_delay_latency, + rdev->pm.igp_sideport_mclk); + } else { + if (rdev->pm.max_bandwidth.full > rdev->pm.k8_bandwidth.full && + rdev->pm.k8_bandwidth.full) + rdev->pm.max_bandwidth = rdev->pm.k8_bandwidth; + if (rdev->pm.max_bandwidth.full > rdev->pm.ht_bandwidth.full && + rdev->pm.ht_bandwidth.full) + rdev->pm.max_bandwidth = rdev->pm.ht_bandwidth; + read_delay_latency.full = rfixed_const(5000); + } + + /* sclk = system clocks(ns) = 1000 / max_bandwidth / 16 */ + a.full = rfixed_const(16); + rdev->pm.sclk.full = rfixed_mul(rdev->pm.max_bandwidth, a); + a.full = rfixed_const(1000); + rdev->pm.sclk.full = rfixed_div(a, rdev->pm.sclk); + /* Determine chunk time + * ChunkTime = the time it takes the DCP to send one chunk of data + * to the LB which consists of pipeline delay and inter chunk gap + * sclk = system clock(ns) + */ + a.full = rfixed_const(256 * 13); + chunk_time.full = rfixed_mul(rdev->pm.sclk, a); + a.full = rfixed_const(10); + chunk_time.full = rfixed_div(chunk_time, a); + + /* Determine the worst case latency + * NumLinePair = Number of line pairs to request(1=2 lines, 2=4 lines) + * WorstCaseLatency = worst case time from urgent to when the MC starts + * to return data + * READ_DELAY_IDLE_MAX = constant of 1us + * ChunkTime = time it takes the DCP to send one chunk of data to the LB + * which consists of pipeline delay and inter chunk gap + */ + if (rfixed_trunc(wm->num_line_pair) > 1) { + a.full = rfixed_const(3); + wm->worst_case_latency.full = rfixed_mul(a, chunk_time); + wm->worst_case_latency.full += read_delay_latency.full; + } else { + a.full = rfixed_const(2); + wm->worst_case_latency.full = rfixed_mul(a, chunk_time); + wm->worst_case_latency.full += read_delay_latency.full; + } + + /* Determine the tolerable latency + * TolerableLatency = Any given request has only 1 line time + * for the data to be returned + * LBRequestFifoDepth = Number of chunk requests the LB can + * put into the request FIFO for a display + * LineTime = total time for one line of display + * ChunkTime = the time it takes the DCP to send one chunk + * of data to the LB which consists of + * pipeline delay and inter chunk gap + */ + if ((2+wm->lb_request_fifo_depth) >= rfixed_trunc(request_fifo_depth)) { + tolerable_latency.full = line_time.full; + } else { + tolerable_latency.full = rfixed_const(wm->lb_request_fifo_depth - 2); + tolerable_latency.full = request_fifo_depth.full - tolerable_latency.full; + tolerable_latency.full = rfixed_mul(tolerable_latency, chunk_time); + tolerable_latency.full = line_time.full - tolerable_latency.full; + } + /* We assume worst case 32bits (4 bytes) */ + wm->dbpp.full = rfixed_const(4 * 8); + + /* Determine the maximum priority mark + * width = viewport width in pixels + */ + a.full = rfixed_const(16); + wm->priority_mark_max.full = rfixed_const(crtc->base.mode.crtc_hdisplay); + wm->priority_mark_max.full = rfixed_div(wm->priority_mark_max, a); + + /* Determine estimated width */ + estimated_width.full = tolerable_latency.full - wm->worst_case_latency.full; + estimated_width.full = rfixed_div(estimated_width, consumption_time); + if (rfixed_trunc(estimated_width) > crtc->base.mode.crtc_hdisplay) { + wm->priority_mark.full = rfixed_const(10); + } else { + a.full = rfixed_const(16); + wm->priority_mark.full = rfixed_div(estimated_width, a); + wm->priority_mark.full = wm->priority_mark_max.full - wm->priority_mark.full; + } +} + +void rs690_bandwidth_update(struct radeon_device *rdev) +{ + struct drm_display_mode *mode0 = NULL; + struct drm_display_mode *mode1 = NULL; + struct rs690_watermark wm0; + struct rs690_watermark wm1; + u32 tmp; + fixed20_12 priority_mark02, priority_mark12, fill_rate; + fixed20_12 a, b; + + if (rdev->mode_info.crtcs[0]->base.enabled) + mode0 = &rdev->mode_info.crtcs[0]->base.mode; + if (rdev->mode_info.crtcs[1]->base.enabled) + mode1 = &rdev->mode_info.crtcs[1]->base.mode; + /* + * Set display0/1 priority up in the memory controller for + * modes if the user specifies HIGH for displaypriority + * option. + */ + if (rdev->disp_priority == 2) { + tmp = RREG32_MC(MC_INIT_MISC_LAT_TIMER); + tmp &= ~MC_DISP1R_INIT_LAT_MASK; + tmp &= ~MC_DISP0R_INIT_LAT_MASK; + if (mode1) + tmp |= (1 << MC_DISP1R_INIT_LAT_SHIFT); + if (mode0) + tmp |= (1 << MC_DISP0R_INIT_LAT_SHIFT); + WREG32_MC(MC_INIT_MISC_LAT_TIMER, tmp); + } + rs690_line_buffer_adjust(rdev, mode0, mode1); + + if ((rdev->family == CHIP_RS690) || (rdev->family == CHIP_RS740)) + WREG32(DCP_CONTROL, 0); + if ((rdev->family == CHIP_RS780) || (rdev->family == CHIP_RS880)) + WREG32(DCP_CONTROL, 2); + + rs690_crtc_bandwidth_compute(rdev, rdev->mode_info.crtcs[0], &wm0); + rs690_crtc_bandwidth_compute(rdev, rdev->mode_info.crtcs[1], &wm1); + + tmp = (wm0.lb_request_fifo_depth - 1); + tmp |= (wm1.lb_request_fifo_depth - 1) << 16; + WREG32(LB_MAX_REQ_OUTSTANDING, tmp); + + if (mode0 && mode1) { + if (rfixed_trunc(wm0.dbpp) > 64) + a.full = rfixed_mul(wm0.dbpp, wm0.num_line_pair); + else + a.full = wm0.num_line_pair.full; + if (rfixed_trunc(wm1.dbpp) > 64) + b.full = rfixed_mul(wm1.dbpp, wm1.num_line_pair); + else + b.full = wm1.num_line_pair.full; + a.full += b.full; + fill_rate.full = rfixed_div(wm0.sclk, a); + if (wm0.consumption_rate.full > fill_rate.full) { + b.full = wm0.consumption_rate.full - fill_rate.full; + b.full = rfixed_mul(b, wm0.active_time); + a.full = rfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + a.full = a.full + b.full; + b.full = rfixed_const(16 * 1000); + priority_mark02.full = rfixed_div(a, b); + } else { + a.full = rfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + b.full = rfixed_const(16 * 1000); + priority_mark02.full = rfixed_div(a, b); + } + if (wm1.consumption_rate.full > fill_rate.full) { + b.full = wm1.consumption_rate.full - fill_rate.full; + b.full = rfixed_mul(b, wm1.active_time); + a.full = rfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + a.full = a.full + b.full; + b.full = rfixed_const(16 * 1000); + priority_mark12.full = rfixed_div(a, b); + } else { + a.full = rfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + b.full = rfixed_const(16 * 1000); + priority_mark12.full = rfixed_div(a, b); + } + if (wm0.priority_mark.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark.full; + if (rfixed_trunc(priority_mark02) < 0) + priority_mark02.full = 0; + if (wm0.priority_mark_max.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark_max.full; + if (wm1.priority_mark.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark.full; + if (rfixed_trunc(priority_mark12) < 0) + priority_mark12.full = 0; + if (wm1.priority_mark_max.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark_max.full; + WREG32(D1MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark02)); + WREG32(D1MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark02)); + WREG32(D2MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark12)); + WREG32(D2MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark12)); + } else if (mode0) { + if (rfixed_trunc(wm0.dbpp) > 64) + a.full = rfixed_mul(wm0.dbpp, wm0.num_line_pair); + else + a.full = wm0.num_line_pair.full; + fill_rate.full = rfixed_div(wm0.sclk, a); + if (wm0.consumption_rate.full > fill_rate.full) { + b.full = wm0.consumption_rate.full - fill_rate.full; + b.full = rfixed_mul(b, wm0.active_time); + a.full = rfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + a.full = a.full + b.full; + b.full = rfixed_const(16 * 1000); + priority_mark02.full = rfixed_div(a, b); + } else { + a.full = rfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + b.full = rfixed_const(16 * 1000); + priority_mark02.full = rfixed_div(a, b); + } + if (wm0.priority_mark.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark.full; + if (rfixed_trunc(priority_mark02) < 0) + priority_mark02.full = 0; + if (wm0.priority_mark_max.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark_max.full; + WREG32(D1MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark02)); + WREG32(D1MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark02)); + WREG32(D2MODE_PRIORITY_A_CNT, MODE_PRIORITY_OFF); + WREG32(D2MODE_PRIORITY_B_CNT, MODE_PRIORITY_OFF); + } else { + if (rfixed_trunc(wm1.dbpp) > 64) + a.full = rfixed_mul(wm1.dbpp, wm1.num_line_pair); + else + a.full = wm1.num_line_pair.full; + fill_rate.full = rfixed_div(wm1.sclk, a); + if (wm1.consumption_rate.full > fill_rate.full) { + b.full = wm1.consumption_rate.full - fill_rate.full; + b.full = rfixed_mul(b, wm1.active_time); + a.full = rfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + a.full = a.full + b.full; + b.full = rfixed_const(16 * 1000); + priority_mark12.full = rfixed_div(a, b); + } else { + a.full = rfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + b.full = rfixed_const(16 * 1000); + priority_mark12.full = rfixed_div(a, b); + } + if (wm1.priority_mark.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark.full; + if (rfixed_trunc(priority_mark12) < 0) + priority_mark12.full = 0; + if (wm1.priority_mark_max.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark_max.full; + WREG32(D1MODE_PRIORITY_A_CNT, MODE_PRIORITY_OFF); + WREG32(D1MODE_PRIORITY_B_CNT, MODE_PRIORITY_OFF); + WREG32(D2MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark12)); + WREG32(D2MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark12)); + } +} /* * Indirect registers accessor diff --git a/drivers/gpu/drm/radeon/rs690r.h b/drivers/gpu/drm/radeon/rs690r.h new file mode 100644 index 00000000000..c0d9faa2175 --- /dev/null +++ b/drivers/gpu/drm/radeon/rs690r.h @@ -0,0 +1,99 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef RS690R_H +#define RS690R_H + +/* RS690/RS740 registers */ +#define MC_INDEX 0x0078 +# define MC_INDEX_MASK 0x1FF +# define MC_INDEX_WR_EN (1 << 9) +# define MC_INDEX_WR_ACK 0x7F +#define MC_DATA 0x007C +#define HDP_FB_LOCATION 0x0134 +#define DC_LB_MEMORY_SPLIT 0x6520 +#define DC_LB_MEMORY_SPLIT_MASK 0x00000003 +#define DC_LB_MEMORY_SPLIT_SHIFT 0 +#define DC_LB_MEMORY_SPLIT_D1HALF_D2HALF 0 +#define DC_LB_MEMORY_SPLIT_D1_3Q_D2_1Q 1 +#define DC_LB_MEMORY_SPLIT_D1_ONLY 2 +#define DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q 3 +#define DC_LB_MEMORY_SPLIT_SHIFT_MODE (1 << 2) +#define DC_LB_DISP1_END_ADR_SHIFT 4 +#define DC_LB_DISP1_END_ADR_MASK 0x00007FF0 +#define D1MODE_PRIORITY_A_CNT 0x6548 +#define MODE_PRIORITY_MARK_MASK 0x00007FFF +#define MODE_PRIORITY_OFF (1 << 16) +#define MODE_PRIORITY_ALWAYS_ON (1 << 20) +#define MODE_PRIORITY_FORCE_MASK (1 << 24) +#define D1MODE_PRIORITY_B_CNT 0x654C +#define LB_MAX_REQ_OUTSTANDING 0x6D58 +#define LB_D1_MAX_REQ_OUTSTANDING_MASK 0x0000000F +#define LB_D1_MAX_REQ_OUTSTANDING_SHIFT 0 +#define LB_D2_MAX_REQ_OUTSTANDING_MASK 0x000F0000 +#define LB_D2_MAX_REQ_OUTSTANDING_SHIFT 16 +#define DCP_CONTROL 0x6C9C +#define D2MODE_PRIORITY_A_CNT 0x6D48 +#define D2MODE_PRIORITY_B_CNT 0x6D4C + +/* MC indirect registers */ +#define MC_STATUS_IDLE (1 << 0) +#define MC_MISC_CNTL 0x18 +#define DISABLE_GTW (1 << 1) +#define GART_INDEX_REG_EN (1 << 12) +#define BLOCK_GFX_D3_EN (1 << 14) +#define GART_FEATURE_ID 0x2B +#define HANG_EN (1 << 11) +#define TLB_ENABLE (1 << 18) +#define P2P_ENABLE (1 << 19) +#define GTW_LAC_EN (1 << 25) +#define LEVEL2_GART (0 << 30) +#define LEVEL1_GART (1 << 30) +#define PDC_EN (1 << 31) +#define GART_BASE 0x2C +#define GART_CACHE_CNTRL 0x2E +# define GART_CACHE_INVALIDATE (1 << 0) +#define MC_STATUS 0x90 +#define MCCFG_FB_LOCATION 0x100 +#define MC_FB_START_MASK 0x0000FFFF +#define MC_FB_START_SHIFT 0 +#define MC_FB_TOP_MASK 0xFFFF0000 +#define MC_FB_TOP_SHIFT 16 +#define MCCFG_AGP_LOCATION 0x101 +#define MC_AGP_START_MASK 0x0000FFFF +#define MC_AGP_START_SHIFT 0 +#define MC_AGP_TOP_MASK 0xFFFF0000 +#define MC_AGP_TOP_SHIFT 16 +#define MCCFG_AGP_BASE 0x102 +#define MCCFG_AGP_BASE_2 0x103 +#define MC_INIT_MISC_LAT_TIMER 0x104 +#define MC_DISP0R_INIT_LAT_SHIFT 8 +#define MC_DISP0R_INIT_LAT_MASK 0x00000F00 +#define MC_DISP1R_INIT_LAT_SHIFT 12 +#define MC_DISP1R_INIT_LAT_MASK 0x0000F000 + +#endif diff --git a/drivers/gpu/drm/radeon/rv515.c b/drivers/gpu/drm/radeon/rv515.c index ffea37b1b3e..551e608702e 100644 --- a/drivers/gpu/drm/radeon/rv515.c +++ b/drivers/gpu/drm/radeon/rv515.c @@ -27,8 +27,9 @@ */ #include <linux/seq_file.h> #include "drmP.h" -#include "radeon_reg.h" +#include "rv515r.h" #include "radeon.h" +#include "radeon_share.h" /* rv515 depends on : */ void r100_hdp_reset(struct radeon_device *rdev); @@ -99,26 +100,26 @@ int rv515_mc_init(struct radeon_device *rdev) "programming pipes. Bad things might happen.\n"); } /* Write VRAM size in case we are limiting it */ - WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.vram_size); - tmp = REG_SET(RV515_MC_FB_START, rdev->mc.vram_location >> 16); + WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.real_vram_size); + tmp = REG_SET(MC_FB_START, rdev->mc.vram_location >> 16); WREG32(0x134, tmp); - tmp = rdev->mc.vram_location + rdev->mc.vram_size - 1; - tmp = REG_SET(RV515_MC_FB_TOP, tmp >> 16); - tmp |= REG_SET(RV515_MC_FB_START, rdev->mc.vram_location >> 16); - WREG32_MC(RV515_MC_FB_LOCATION, tmp); - WREG32(RS690_HDP_FB_LOCATION, rdev->mc.vram_location >> 16); + tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; + tmp = REG_SET(MC_FB_TOP, tmp >> 16); + tmp |= REG_SET(MC_FB_START, rdev->mc.vram_location >> 16); + WREG32_MC(MC_FB_LOCATION, tmp); + WREG32(HDP_FB_LOCATION, rdev->mc.vram_location >> 16); WREG32(0x310, rdev->mc.vram_location); if (rdev->flags & RADEON_IS_AGP) { tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - 1; - tmp = REG_SET(RV515_MC_AGP_TOP, tmp >> 16); - tmp |= REG_SET(RV515_MC_AGP_START, rdev->mc.gtt_location >> 16); - WREG32_MC(RV515_MC_AGP_LOCATION, tmp); - WREG32_MC(RV515_MC_AGP_BASE, rdev->mc.agp_base); - WREG32_MC(RV515_MC_AGP_BASE_2, 0); + tmp = REG_SET(MC_AGP_TOP, tmp >> 16); + tmp |= REG_SET(MC_AGP_START, rdev->mc.gtt_location >> 16); + WREG32_MC(MC_AGP_LOCATION, tmp); + WREG32_MC(MC_AGP_BASE, rdev->mc.agp_base); + WREG32_MC(MC_AGP_BASE_2, 0); } else { - WREG32_MC(RV515_MC_AGP_LOCATION, 0x0FFFFFFF); - WREG32_MC(RV515_MC_AGP_BASE, 0); - WREG32_MC(RV515_MC_AGP_BASE_2, 0); + WREG32_MC(MC_AGP_LOCATION, 0x0FFFFFFF); + WREG32_MC(MC_AGP_BASE, 0); + WREG32_MC(MC_AGP_BASE_2, 0); } return 0; } @@ -136,95 +137,67 @@ void rv515_mc_fini(struct radeon_device *rdev) */ void rv515_ring_start(struct radeon_device *rdev) { - unsigned gb_tile_config; int r; - /* Sub pixel 1/12 so we can have 4K rendering according to doc */ - gb_tile_config = R300_ENABLE_TILING | R300_TILE_SIZE_16; - switch (rdev->num_gb_pipes) { - case 2: - gb_tile_config |= R300_PIPE_COUNT_R300; - break; - case 3: - gb_tile_config |= R300_PIPE_COUNT_R420_3P; - break; - case 4: - gb_tile_config |= R300_PIPE_COUNT_R420; - break; - case 1: - default: - gb_tile_config |= R300_PIPE_COUNT_RV350; - break; - } - r = radeon_ring_lock(rdev, 64); if (r) { return; } - radeon_ring_write(rdev, PACKET0(RADEON_ISYNC_CNTL, 0)); - radeon_ring_write(rdev, - RADEON_ISYNC_ANY2D_IDLE3D | - RADEON_ISYNC_ANY3D_IDLE2D | - RADEON_ISYNC_WAIT_IDLEGUI | - RADEON_ISYNC_CPSCRATCH_IDLEGUI); - radeon_ring_write(rdev, PACKET0(R300_GB_TILE_CONFIG, 0)); - radeon_ring_write(rdev, gb_tile_config); - radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); + radeon_ring_write(rdev, PACKET0(ISYNC_CNTL, 0)); radeon_ring_write(rdev, - RADEON_WAIT_2D_IDLECLEAN | - RADEON_WAIT_3D_IDLECLEAN); + ISYNC_ANY2D_IDLE3D | + ISYNC_ANY3D_IDLE2D | + ISYNC_WAIT_IDLEGUI | + ISYNC_CPSCRATCH_IDLEGUI); + radeon_ring_write(rdev, PACKET0(WAIT_UNTIL, 0)); + radeon_ring_write(rdev, WAIT_2D_IDLECLEAN | WAIT_3D_IDLECLEAN); radeon_ring_write(rdev, PACKET0(0x170C, 0)); radeon_ring_write(rdev, 1 << 31); - radeon_ring_write(rdev, PACKET0(R300_GB_SELECT, 0)); + radeon_ring_write(rdev, PACKET0(GB_SELECT, 0)); radeon_ring_write(rdev, 0); - radeon_ring_write(rdev, PACKET0(R300_GB_ENABLE, 0)); + radeon_ring_write(rdev, PACKET0(GB_ENABLE, 0)); radeon_ring_write(rdev, 0); radeon_ring_write(rdev, PACKET0(0x42C8, 0)); radeon_ring_write(rdev, (1 << rdev->num_gb_pipes) - 1); - radeon_ring_write(rdev, PACKET0(R500_VAP_INDEX_OFFSET, 0)); + radeon_ring_write(rdev, PACKET0(VAP_INDEX_OFFSET, 0)); radeon_ring_write(rdev, 0); - radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); - radeon_ring_write(rdev, R300_RB3D_DC_FLUSH | R300_RB3D_DC_FREE); - radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0)); - radeon_ring_write(rdev, R300_ZC_FLUSH | R300_ZC_FREE); - radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); - radeon_ring_write(rdev, - RADEON_WAIT_2D_IDLECLEAN | - RADEON_WAIT_3D_IDLECLEAN); - radeon_ring_write(rdev, PACKET0(R300_GB_AA_CONFIG, 0)); + radeon_ring_write(rdev, PACKET0(RB3D_DSTCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, RB3D_DC_FLUSH | RB3D_DC_FREE); + radeon_ring_write(rdev, PACKET0(ZB_ZCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, ZC_FLUSH | ZC_FREE); + radeon_ring_write(rdev, PACKET0(WAIT_UNTIL, 0)); + radeon_ring_write(rdev, WAIT_2D_IDLECLEAN | WAIT_3D_IDLECLEAN); + radeon_ring_write(rdev, PACKET0(GB_AA_CONFIG, 0)); radeon_ring_write(rdev, 0); - radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); - radeon_ring_write(rdev, R300_RB3D_DC_FLUSH | R300_RB3D_DC_FREE); - radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0)); - radeon_ring_write(rdev, R300_ZC_FLUSH | R300_ZC_FREE); - radeon_ring_write(rdev, PACKET0(R300_GB_MSPOS0, 0)); - radeon_ring_write(rdev, - ((6 << R300_MS_X0_SHIFT) | - (6 << R300_MS_Y0_SHIFT) | - (6 << R300_MS_X1_SHIFT) | - (6 << R300_MS_Y1_SHIFT) | - (6 << R300_MS_X2_SHIFT) | - (6 << R300_MS_Y2_SHIFT) | - (6 << R300_MSBD0_Y_SHIFT) | - (6 << R300_MSBD0_X_SHIFT))); - radeon_ring_write(rdev, PACKET0(R300_GB_MSPOS1, 0)); + radeon_ring_write(rdev, PACKET0(RB3D_DSTCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, RB3D_DC_FLUSH | RB3D_DC_FREE); + radeon_ring_write(rdev, PACKET0(ZB_ZCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, ZC_FLUSH | ZC_FREE); + radeon_ring_write(rdev, PACKET0(GB_MSPOS0, 0)); radeon_ring_write(rdev, - ((6 << R300_MS_X3_SHIFT) | - (6 << R300_MS_Y3_SHIFT) | - (6 << R300_MS_X4_SHIFT) | - (6 << R300_MS_Y4_SHIFT) | - (6 << R300_MS_X5_SHIFT) | - (6 << R300_MS_Y5_SHIFT) | - (6 << R300_MSBD1_SHIFT))); - radeon_ring_write(rdev, PACKET0(R300_GA_ENHANCE, 0)); - radeon_ring_write(rdev, R300_GA_DEADLOCK_CNTL | R300_GA_FASTSYNC_CNTL); - radeon_ring_write(rdev, PACKET0(R300_GA_POLY_MODE, 0)); + ((6 << MS_X0_SHIFT) | + (6 << MS_Y0_SHIFT) | + (6 << MS_X1_SHIFT) | + (6 << MS_Y1_SHIFT) | + (6 << MS_X2_SHIFT) | + (6 << MS_Y2_SHIFT) | + (6 << MSBD0_Y_SHIFT) | + (6 << MSBD0_X_SHIFT))); + radeon_ring_write(rdev, PACKET0(GB_MSPOS1, 0)); radeon_ring_write(rdev, - R300_FRONT_PTYPE_TRIANGE | R300_BACK_PTYPE_TRIANGE); - radeon_ring_write(rdev, PACKET0(R300_GA_ROUND_MODE, 0)); - radeon_ring_write(rdev, - R300_GEOMETRY_ROUND_NEAREST | - R300_COLOR_ROUND_NEAREST); + ((6 << MS_X3_SHIFT) | + (6 << MS_Y3_SHIFT) | + (6 << MS_X4_SHIFT) | + (6 << MS_Y4_SHIFT) | + (6 << MS_X5_SHIFT) | + (6 << MS_Y5_SHIFT) | + (6 << MSBD1_SHIFT))); + radeon_ring_write(rdev, PACKET0(GA_ENHANCE, 0)); + radeon_ring_write(rdev, GA_DEADLOCK_CNTL | GA_FASTSYNC_CNTL); + radeon_ring_write(rdev, PACKET0(GA_POLY_MODE, 0)); + radeon_ring_write(rdev, FRONT_PTYPE_TRIANGE | BACK_PTYPE_TRIANGE); + radeon_ring_write(rdev, PACKET0(GA_ROUND_MODE, 0)); + radeon_ring_write(rdev, GEOMETRY_ROUND_NEAREST | COLOR_ROUND_NEAREST); radeon_ring_write(rdev, PACKET0(0x20C8, 0)); radeon_ring_write(rdev, 0); radeon_ring_unlock_commit(rdev); @@ -242,8 +215,8 @@ int rv515_mc_wait_for_idle(struct radeon_device *rdev) for (i = 0; i < rdev->usec_timeout; i++) { /* read MC_STATUS */ - tmp = RREG32_MC(RV515_MC_STATUS); - if (tmp & RV515_MC_STATUS_IDLE) { + tmp = RREG32_MC(MC_STATUS); + if (tmp & MC_STATUS_IDLE) { return 0; } DRM_UDELAY(1); @@ -291,33 +264,33 @@ int rv515_ga_reset(struct radeon_device *rdev) reinit_cp = rdev->cp.ready; rdev->cp.ready = false; for (i = 0; i < rdev->usec_timeout; i++) { - WREG32(RADEON_CP_CSQ_MODE, 0); - WREG32(RADEON_CP_CSQ_CNTL, 0); - WREG32(RADEON_RBBM_SOFT_RESET, 0x32005); - (void)RREG32(RADEON_RBBM_SOFT_RESET); + WREG32(CP_CSQ_MODE, 0); + WREG32(CP_CSQ_CNTL, 0); + WREG32(RBBM_SOFT_RESET, 0x32005); + (void)RREG32(RBBM_SOFT_RESET); udelay(200); - WREG32(RADEON_RBBM_SOFT_RESET, 0); + WREG32(RBBM_SOFT_RESET, 0); /* Wait to prevent race in RBBM_STATUS */ mdelay(1); - tmp = RREG32(RADEON_RBBM_STATUS); + tmp = RREG32(RBBM_STATUS); if (tmp & ((1 << 20) | (1 << 26))) { DRM_ERROR("VAP & CP still busy (RBBM_STATUS=0x%08X)\n", tmp); /* GA still busy soft reset it */ WREG32(0x429C, 0x200); - WREG32(R300_VAP_PVS_STATE_FLUSH_REG, 0); + WREG32(VAP_PVS_STATE_FLUSH_REG, 0); WREG32(0x43E0, 0); WREG32(0x43E4, 0); WREG32(0x24AC, 0); } /* Wait to prevent race in RBBM_STATUS */ mdelay(1); - tmp = RREG32(RADEON_RBBM_STATUS); + tmp = RREG32(RBBM_STATUS); if (!(tmp & ((1 << 20) | (1 << 26)))) { break; } } for (i = 0; i < rdev->usec_timeout; i++) { - tmp = RREG32(RADEON_RBBM_STATUS); + tmp = RREG32(RBBM_STATUS); if (!(tmp & ((1 << 20) | (1 << 26)))) { DRM_INFO("GA reset succeed (RBBM_STATUS=0x%08X)\n", tmp); @@ -331,7 +304,7 @@ int rv515_ga_reset(struct radeon_device *rdev) } DRM_UDELAY(1); } - tmp = RREG32(RADEON_RBBM_STATUS); + tmp = RREG32(RBBM_STATUS); DRM_ERROR("Failed to reset GA ! (RBBM_STATUS=0x%08X)\n", tmp); return -1; } @@ -341,7 +314,7 @@ int rv515_gpu_reset(struct radeon_device *rdev) uint32_t status; /* reset order likely matter */ - status = RREG32(RADEON_RBBM_STATUS); + status = RREG32(RBBM_STATUS); /* reset HDP */ r100_hdp_reset(rdev); /* reset rb2d */ @@ -353,12 +326,12 @@ int rv515_gpu_reset(struct radeon_device *rdev) rv515_ga_reset(rdev); } /* reset CP */ - status = RREG32(RADEON_RBBM_STATUS); + status = RREG32(RBBM_STATUS); if (status & (1 << 16)) { r100_cp_reset(rdev); } /* Check if GPU is idle */ - status = RREG32(RADEON_RBBM_STATUS); + status = RREG32(RBBM_STATUS); if (status & (1 << 31)) { DRM_ERROR("Failed to reset GPU (RBBM_STATUS=0x%08X)\n", status); return -1; @@ -377,8 +350,7 @@ static void rv515_vram_get_type(struct radeon_device *rdev) rdev->mc.vram_width = 128; rdev->mc.vram_is_ddr = true; - tmp = RREG32_MC(RV515_MC_CNTL); - tmp &= RV515_MEM_NUM_CHANNELS_MASK; + tmp = RREG32_MC(RV515_MC_CNTL) & MEM_NUM_CHANNELS_MASK; switch (tmp) { case 0: rdev->mc.vram_width = 64; @@ -394,11 +366,16 @@ static void rv515_vram_get_type(struct radeon_device *rdev) void rv515_vram_info(struct radeon_device *rdev) { + fixed20_12 a; + rv515_vram_get_type(rdev); - rdev->mc.vram_size = RREG32(RADEON_CONFIG_MEMSIZE); - rdev->mc.aper_base = drm_get_resource_start(rdev->ddev, 0); - rdev->mc.aper_size = drm_get_resource_len(rdev->ddev, 0); + /* FIXME: we should enforce default clock in case GPU is not in + * default setup + */ + a.full = rfixed_const(100); + rdev->pm.sclk.full = rfixed_const(rdev->clock.default_sclk); + rdev->pm.sclk.full = rfixed_div(rdev->pm.sclk, a); } @@ -409,35 +386,35 @@ uint32_t rv515_mc_rreg(struct radeon_device *rdev, uint32_t reg) { uint32_t r; - WREG32(R520_MC_IND_INDEX, 0x7f0000 | (reg & 0xffff)); - r = RREG32(R520_MC_IND_DATA); - WREG32(R520_MC_IND_INDEX, 0); + WREG32(MC_IND_INDEX, 0x7f0000 | (reg & 0xffff)); + r = RREG32(MC_IND_DATA); + WREG32(MC_IND_INDEX, 0); return r; } void rv515_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) { - WREG32(R520_MC_IND_INDEX, 0xff0000 | ((reg) & 0xffff)); - WREG32(R520_MC_IND_DATA, (v)); - WREG32(R520_MC_IND_INDEX, 0); + WREG32(MC_IND_INDEX, 0xff0000 | ((reg) & 0xffff)); + WREG32(MC_IND_DATA, (v)); + WREG32(MC_IND_INDEX, 0); } uint32_t rv515_pcie_rreg(struct radeon_device *rdev, uint32_t reg) { uint32_t r; - WREG32(RADEON_PCIE_INDEX, ((reg) & 0x7ff)); - (void)RREG32(RADEON_PCIE_INDEX); - r = RREG32(RADEON_PCIE_DATA); + WREG32(PCIE_INDEX, ((reg) & 0x7ff)); + (void)RREG32(PCIE_INDEX); + r = RREG32(PCIE_DATA); return r; } void rv515_pcie_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) { - WREG32(RADEON_PCIE_INDEX, ((reg) & 0x7ff)); - (void)RREG32(RADEON_PCIE_INDEX); - WREG32(RADEON_PCIE_DATA, (v)); - (void)RREG32(RADEON_PCIE_DATA); + WREG32(PCIE_INDEX, ((reg) & 0x7ff)); + (void)RREG32(PCIE_INDEX); + WREG32(PCIE_DATA, (v)); + (void)RREG32(PCIE_DATA); } @@ -452,13 +429,13 @@ static int rv515_debugfs_pipes_info(struct seq_file *m, void *data) struct radeon_device *rdev = dev->dev_private; uint32_t tmp; - tmp = RREG32(R400_GB_PIPE_SELECT); + tmp = RREG32(GB_PIPE_SELECT); seq_printf(m, "GB_PIPE_SELECT 0x%08x\n", tmp); - tmp = RREG32(R500_SU_REG_DEST); + tmp = RREG32(SU_REG_DEST); seq_printf(m, "SU_REG_DEST 0x%08x\n", tmp); - tmp = RREG32(R300_GB_TILE_CONFIG); + tmp = RREG32(GB_TILE_CONFIG); seq_printf(m, "GB_TILE_CONFIG 0x%08x\n", tmp); - tmp = RREG32(R300_DST_PIPE_CONFIG); + tmp = RREG32(DST_PIPE_CONFIG); seq_printf(m, "DST_PIPE_CONFIG 0x%08x\n", tmp); return 0; } @@ -509,9 +486,9 @@ int rv515_debugfs_ga_info_init(struct radeon_device *rdev) /* * Asic initialization */ -static const unsigned r500_reg_safe_bm[159] = { +static const unsigned r500_reg_safe_bm[219] = { + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, - 0xFFFFFFBF, 0xFFFFFFFF, 0xFFFFFFBF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, @@ -549,14 +526,575 @@ static const unsigned r500_reg_safe_bm[159] = { 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFF80FFFF, 0x00000000, 0x00000000, 0x00000000, 0x00000000, - 0x0003FC01, 0x3FFFFCF8, 0xFE800B19, + 0x0003FC01, 0x3FFFFCF8, 0xFE800B19, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, }; - - int rv515_init(struct radeon_device *rdev) { rdev->config.r300.reg_safe_bm = r500_reg_safe_bm; rdev->config.r300.reg_safe_bm_size = ARRAY_SIZE(r500_reg_safe_bm); return 0; } + +void atom_rv515_force_tv_scaler(struct radeon_device *rdev) +{ + + WREG32(0x659C, 0x0); + WREG32(0x6594, 0x705); + WREG32(0x65A4, 0x10001); + WREG32(0x65D8, 0x0); + WREG32(0x65B0, 0x0); + WREG32(0x65C0, 0x0); + WREG32(0x65D4, 0x0); + WREG32(0x6578, 0x0); + WREG32(0x657C, 0x841880A8); + WREG32(0x6578, 0x1); + WREG32(0x657C, 0x84208680); + WREG32(0x6578, 0x2); + WREG32(0x657C, 0xBFF880B0); + WREG32(0x6578, 0x100); + WREG32(0x657C, 0x83D88088); + WREG32(0x6578, 0x101); + WREG32(0x657C, 0x84608680); + WREG32(0x6578, 0x102); + WREG32(0x657C, 0xBFF080D0); + WREG32(0x6578, 0x200); + WREG32(0x657C, 0x83988068); + WREG32(0x6578, 0x201); + WREG32(0x657C, 0x84A08680); + WREG32(0x6578, 0x202); + WREG32(0x657C, 0xBFF080F8); + WREG32(0x6578, 0x300); + WREG32(0x657C, 0x83588058); + WREG32(0x6578, 0x301); + WREG32(0x657C, 0x84E08660); + WREG32(0x6578, 0x302); + WREG32(0x657C, 0xBFF88120); + WREG32(0x6578, 0x400); + WREG32(0x657C, 0x83188040); + WREG32(0x6578, 0x401); + WREG32(0x657C, 0x85008660); + WREG32(0x6578, 0x402); + WREG32(0x657C, 0xBFF88150); + WREG32(0x6578, 0x500); + WREG32(0x657C, 0x82D88030); + WREG32(0x6578, 0x501); + WREG32(0x657C, 0x85408640); + WREG32(0x6578, 0x502); + WREG32(0x657C, 0xBFF88180); + WREG32(0x6578, 0x600); + WREG32(0x657C, 0x82A08018); + WREG32(0x6578, 0x601); + WREG32(0x657C, 0x85808620); + WREG32(0x6578, 0x602); + WREG32(0x657C, 0xBFF081B8); + WREG32(0x6578, 0x700); + WREG32(0x657C, 0x82608010); + WREG32(0x6578, 0x701); + WREG32(0x657C, 0x85A08600); + WREG32(0x6578, 0x702); + WREG32(0x657C, 0x800081F0); + WREG32(0x6578, 0x800); + WREG32(0x657C, 0x8228BFF8); + WREG32(0x6578, 0x801); + WREG32(0x657C, 0x85E085E0); + WREG32(0x6578, 0x802); + WREG32(0x657C, 0xBFF88228); + WREG32(0x6578, 0x10000); + WREG32(0x657C, 0x82A8BF00); + WREG32(0x6578, 0x10001); + WREG32(0x657C, 0x82A08CC0); + WREG32(0x6578, 0x10002); + WREG32(0x657C, 0x8008BEF8); + WREG32(0x6578, 0x10100); + WREG32(0x657C, 0x81F0BF28); + WREG32(0x6578, 0x10101); + WREG32(0x657C, 0x83608CA0); + WREG32(0x6578, 0x10102); + WREG32(0x657C, 0x8018BED0); + WREG32(0x6578, 0x10200); + WREG32(0x657C, 0x8148BF38); + WREG32(0x6578, 0x10201); + WREG32(0x657C, 0x84408C80); + WREG32(0x6578, 0x10202); + WREG32(0x657C, 0x8008BEB8); + WREG32(0x6578, 0x10300); + WREG32(0x657C, 0x80B0BF78); + WREG32(0x6578, 0x10301); + WREG32(0x657C, 0x85008C20); + WREG32(0x6578, 0x10302); + WREG32(0x657C, 0x8020BEA0); + WREG32(0x6578, 0x10400); + WREG32(0x657C, 0x8028BF90); + WREG32(0x6578, 0x10401); + WREG32(0x657C, 0x85E08BC0); + WREG32(0x6578, 0x10402); + WREG32(0x657C, 0x8018BE90); + WREG32(0x6578, 0x10500); + WREG32(0x657C, 0xBFB8BFB0); + WREG32(0x6578, 0x10501); + WREG32(0x657C, 0x86C08B40); + WREG32(0x6578, 0x10502); + WREG32(0x657C, 0x8010BE90); + WREG32(0x6578, 0x10600); + WREG32(0x657C, 0xBF58BFC8); + WREG32(0x6578, 0x10601); + WREG32(0x657C, 0x87A08AA0); + WREG32(0x6578, 0x10602); + WREG32(0x657C, 0x8010BE98); + WREG32(0x6578, 0x10700); + WREG32(0x657C, 0xBF10BFF0); + WREG32(0x6578, 0x10701); + WREG32(0x657C, 0x886089E0); + WREG32(0x6578, 0x10702); + WREG32(0x657C, 0x8018BEB0); + WREG32(0x6578, 0x10800); + WREG32(0x657C, 0xBED8BFE8); + WREG32(0x6578, 0x10801); + WREG32(0x657C, 0x89408940); + WREG32(0x6578, 0x10802); + WREG32(0x657C, 0xBFE8BED8); + WREG32(0x6578, 0x20000); + WREG32(0x657C, 0x80008000); + WREG32(0x6578, 0x20001); + WREG32(0x657C, 0x90008000); + WREG32(0x6578, 0x20002); + WREG32(0x657C, 0x80008000); + WREG32(0x6578, 0x20003); + WREG32(0x657C, 0x80008000); + WREG32(0x6578, 0x20100); + WREG32(0x657C, 0x80108000); + WREG32(0x6578, 0x20101); + WREG32(0x657C, 0x8FE0BF70); + WREG32(0x6578, 0x20102); + WREG32(0x657C, 0xBFE880C0); + WREG32(0x6578, 0x20103); + WREG32(0x657C, 0x80008000); + WREG32(0x6578, 0x20200); + WREG32(0x657C, 0x8018BFF8); + WREG32(0x6578, 0x20201); + WREG32(0x657C, 0x8F80BF08); + WREG32(0x6578, 0x20202); + WREG32(0x657C, 0xBFD081A0); + WREG32(0x6578, 0x20203); + WREG32(0x657C, 0xBFF88000); + WREG32(0x6578, 0x20300); + WREG32(0x657C, 0x80188000); + WREG32(0x6578, 0x20301); + WREG32(0x657C, 0x8EE0BEC0); + WREG32(0x6578, 0x20302); + WREG32(0x657C, 0xBFB082A0); + WREG32(0x6578, 0x20303); + WREG32(0x657C, 0x80008000); + WREG32(0x6578, 0x20400); + WREG32(0x657C, 0x80188000); + WREG32(0x6578, 0x20401); + WREG32(0x657C, 0x8E00BEA0); + WREG32(0x6578, 0x20402); + WREG32(0x657C, 0xBF8883C0); + WREG32(0x6578, 0x20403); + WREG32(0x657C, 0x80008000); + WREG32(0x6578, 0x20500); + WREG32(0x657C, 0x80188000); + WREG32(0x6578, 0x20501); + WREG32(0x657C, 0x8D00BE90); + WREG32(0x6578, 0x20502); + WREG32(0x657C, 0xBF588500); + WREG32(0x6578, 0x20503); + WREG32(0x657C, 0x80008008); + WREG32(0x6578, 0x20600); + WREG32(0x657C, 0x80188000); + WREG32(0x6578, 0x20601); + WREG32(0x657C, 0x8BC0BE98); + WREG32(0x6578, 0x20602); + WREG32(0x657C, 0xBF308660); + WREG32(0x6578, 0x20603); + WREG32(0x657C, 0x80008008); + WREG32(0x6578, 0x20700); + WREG32(0x657C, 0x80108000); + WREG32(0x6578, 0x20701); + WREG32(0x657C, 0x8A80BEB0); + WREG32(0x6578, 0x20702); + WREG32(0x657C, 0xBF0087C0); + WREG32(0x6578, 0x20703); + WREG32(0x657C, 0x80008008); + WREG32(0x6578, 0x20800); + WREG32(0x657C, 0x80108000); + WREG32(0x6578, 0x20801); + WREG32(0x657C, 0x8920BED0); + WREG32(0x6578, 0x20802); + WREG32(0x657C, 0xBED08920); + WREG32(0x6578, 0x20803); + WREG32(0x657C, 0x80008010); + WREG32(0x6578, 0x30000); + WREG32(0x657C, 0x90008000); + WREG32(0x6578, 0x30001); + WREG32(0x657C, 0x80008000); + WREG32(0x6578, 0x30100); + WREG32(0x657C, 0x8FE0BF90); + WREG32(0x6578, 0x30101); + WREG32(0x657C, 0xBFF880A0); + WREG32(0x6578, 0x30200); + WREG32(0x657C, 0x8F60BF40); + WREG32(0x6578, 0x30201); + WREG32(0x657C, 0xBFE88180); + WREG32(0x6578, 0x30300); + WREG32(0x657C, 0x8EC0BF00); + WREG32(0x6578, 0x30301); + WREG32(0x657C, 0xBFC88280); + WREG32(0x6578, 0x30400); + WREG32(0x657C, 0x8DE0BEE0); + WREG32(0x6578, 0x30401); + WREG32(0x657C, 0xBFA083A0); + WREG32(0x6578, 0x30500); + WREG32(0x657C, 0x8CE0BED0); + WREG32(0x6578, 0x30501); + WREG32(0x657C, 0xBF7884E0); + WREG32(0x6578, 0x30600); + WREG32(0x657C, 0x8BA0BED8); + WREG32(0x6578, 0x30601); + WREG32(0x657C, 0xBF508640); + WREG32(0x6578, 0x30700); + WREG32(0x657C, 0x8A60BEE8); + WREG32(0x6578, 0x30701); + WREG32(0x657C, 0xBF2087A0); + WREG32(0x6578, 0x30800); + WREG32(0x657C, 0x8900BF00); + WREG32(0x6578, 0x30801); + WREG32(0x657C, 0xBF008900); +} + +struct rv515_watermark { + u32 lb_request_fifo_depth; + fixed20_12 num_line_pair; + fixed20_12 estimated_width; + fixed20_12 worst_case_latency; + fixed20_12 consumption_rate; + fixed20_12 active_time; + fixed20_12 dbpp; + fixed20_12 priority_mark_max; + fixed20_12 priority_mark; + fixed20_12 sclk; +}; + +void rv515_crtc_bandwidth_compute(struct radeon_device *rdev, + struct radeon_crtc *crtc, + struct rv515_watermark *wm) +{ + struct drm_display_mode *mode = &crtc->base.mode; + fixed20_12 a, b, c; + fixed20_12 pclk, request_fifo_depth, tolerable_latency, estimated_width; + fixed20_12 consumption_time, line_time, chunk_time, read_delay_latency; + + if (!crtc->base.enabled) { + /* FIXME: wouldn't it better to set priority mark to maximum */ + wm->lb_request_fifo_depth = 4; + return; + } + + if (crtc->vsc.full > rfixed_const(2)) + wm->num_line_pair.full = rfixed_const(2); + else + wm->num_line_pair.full = rfixed_const(1); + + b.full = rfixed_const(mode->crtc_hdisplay); + c.full = rfixed_const(256); + a.full = rfixed_mul(wm->num_line_pair, b); + request_fifo_depth.full = rfixed_div(a, c); + if (a.full < rfixed_const(4)) { + wm->lb_request_fifo_depth = 4; + } else { + wm->lb_request_fifo_depth = rfixed_trunc(request_fifo_depth); + } + + /* Determine consumption rate + * pclk = pixel clock period(ns) = 1000 / (mode.clock / 1000) + * vtaps = number of vertical taps, + * vsc = vertical scaling ratio, defined as source/destination + * hsc = horizontal scaling ration, defined as source/destination + */ + a.full = rfixed_const(mode->clock); + b.full = rfixed_const(1000); + a.full = rfixed_div(a, b); + pclk.full = rfixed_div(b, a); + if (crtc->rmx_type != RMX_OFF) { + b.full = rfixed_const(2); + if (crtc->vsc.full > b.full) + b.full = crtc->vsc.full; + b.full = rfixed_mul(b, crtc->hsc); + c.full = rfixed_const(2); + b.full = rfixed_div(b, c); + consumption_time.full = rfixed_div(pclk, b); + } else { + consumption_time.full = pclk.full; + } + a.full = rfixed_const(1); + wm->consumption_rate.full = rfixed_div(a, consumption_time); + + + /* Determine line time + * LineTime = total time for one line of displayhtotal + * LineTime = total number of horizontal pixels + * pclk = pixel clock period(ns) + */ + a.full = rfixed_const(crtc->base.mode.crtc_htotal); + line_time.full = rfixed_mul(a, pclk); + + /* Determine active time + * ActiveTime = time of active region of display within one line, + * hactive = total number of horizontal active pixels + * htotal = total number of horizontal pixels + */ + a.full = rfixed_const(crtc->base.mode.crtc_htotal); + b.full = rfixed_const(crtc->base.mode.crtc_hdisplay); + wm->active_time.full = rfixed_mul(line_time, b); + wm->active_time.full = rfixed_div(wm->active_time, a); + + /* Determine chunk time + * ChunkTime = the time it takes the DCP to send one chunk of data + * to the LB which consists of pipeline delay and inter chunk gap + * sclk = system clock(Mhz) + */ + a.full = rfixed_const(600 * 1000); + chunk_time.full = rfixed_div(a, rdev->pm.sclk); + read_delay_latency.full = rfixed_const(1000); + + /* Determine the worst case latency + * NumLinePair = Number of line pairs to request(1=2 lines, 2=4 lines) + * WorstCaseLatency = worst case time from urgent to when the MC starts + * to return data + * READ_DELAY_IDLE_MAX = constant of 1us + * ChunkTime = time it takes the DCP to send one chunk of data to the LB + * which consists of pipeline delay and inter chunk gap + */ + if (rfixed_trunc(wm->num_line_pair) > 1) { + a.full = rfixed_const(3); + wm->worst_case_latency.full = rfixed_mul(a, chunk_time); + wm->worst_case_latency.full += read_delay_latency.full; + } else { + wm->worst_case_latency.full = chunk_time.full + read_delay_latency.full; + } + + /* Determine the tolerable latency + * TolerableLatency = Any given request has only 1 line time + * for the data to be returned + * LBRequestFifoDepth = Number of chunk requests the LB can + * put into the request FIFO for a display + * LineTime = total time for one line of display + * ChunkTime = the time it takes the DCP to send one chunk + * of data to the LB which consists of + * pipeline delay and inter chunk gap + */ + if ((2+wm->lb_request_fifo_depth) >= rfixed_trunc(request_fifo_depth)) { + tolerable_latency.full = line_time.full; + } else { + tolerable_latency.full = rfixed_const(wm->lb_request_fifo_depth - 2); + tolerable_latency.full = request_fifo_depth.full - tolerable_latency.full; + tolerable_latency.full = rfixed_mul(tolerable_latency, chunk_time); + tolerable_latency.full = line_time.full - tolerable_latency.full; + } + /* We assume worst case 32bits (4 bytes) */ + wm->dbpp.full = rfixed_const(2 * 16); + + /* Determine the maximum priority mark + * width = viewport width in pixels + */ + a.full = rfixed_const(16); + wm->priority_mark_max.full = rfixed_const(crtc->base.mode.crtc_hdisplay); + wm->priority_mark_max.full = rfixed_div(wm->priority_mark_max, a); + + /* Determine estimated width */ + estimated_width.full = tolerable_latency.full - wm->worst_case_latency.full; + estimated_width.full = rfixed_div(estimated_width, consumption_time); + if (rfixed_trunc(estimated_width) > crtc->base.mode.crtc_hdisplay) { + wm->priority_mark.full = rfixed_const(10); + } else { + a.full = rfixed_const(16); + wm->priority_mark.full = rfixed_div(estimated_width, a); + wm->priority_mark.full = wm->priority_mark_max.full - wm->priority_mark.full; + } +} + +void rv515_bandwidth_avivo_update(struct radeon_device *rdev) +{ + struct drm_display_mode *mode0 = NULL; + struct drm_display_mode *mode1 = NULL; + struct rv515_watermark wm0; + struct rv515_watermark wm1; + u32 tmp; + fixed20_12 priority_mark02, priority_mark12, fill_rate; + fixed20_12 a, b; + + if (rdev->mode_info.crtcs[0]->base.enabled) + mode0 = &rdev->mode_info.crtcs[0]->base.mode; + if (rdev->mode_info.crtcs[1]->base.enabled) + mode1 = &rdev->mode_info.crtcs[1]->base.mode; + rs690_line_buffer_adjust(rdev, mode0, mode1); + + rv515_crtc_bandwidth_compute(rdev, rdev->mode_info.crtcs[0], &wm0); + rv515_crtc_bandwidth_compute(rdev, rdev->mode_info.crtcs[1], &wm1); + + tmp = wm0.lb_request_fifo_depth; + tmp |= wm1.lb_request_fifo_depth << 16; + WREG32(LB_MAX_REQ_OUTSTANDING, tmp); + + if (mode0 && mode1) { + if (rfixed_trunc(wm0.dbpp) > 64) + a.full = rfixed_div(wm0.dbpp, wm0.num_line_pair); + else + a.full = wm0.num_line_pair.full; + if (rfixed_trunc(wm1.dbpp) > 64) + b.full = rfixed_div(wm1.dbpp, wm1.num_line_pair); + else + b.full = wm1.num_line_pair.full; + a.full += b.full; + fill_rate.full = rfixed_div(wm0.sclk, a); + if (wm0.consumption_rate.full > fill_rate.full) { + b.full = wm0.consumption_rate.full - fill_rate.full; + b.full = rfixed_mul(b, wm0.active_time); + a.full = rfixed_const(16); + b.full = rfixed_div(b, a); + a.full = rfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + priority_mark02.full = a.full + b.full; + } else { + a.full = rfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + b.full = rfixed_const(16 * 1000); + priority_mark02.full = rfixed_div(a, b); + } + if (wm1.consumption_rate.full > fill_rate.full) { + b.full = wm1.consumption_rate.full - fill_rate.full; + b.full = rfixed_mul(b, wm1.active_time); + a.full = rfixed_const(16); + b.full = rfixed_div(b, a); + a.full = rfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + priority_mark12.full = a.full + b.full; + } else { + a.full = rfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + b.full = rfixed_const(16 * 1000); + priority_mark12.full = rfixed_div(a, b); + } + if (wm0.priority_mark.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark.full; + if (rfixed_trunc(priority_mark02) < 0) + priority_mark02.full = 0; + if (wm0.priority_mark_max.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark_max.full; + if (wm1.priority_mark.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark.full; + if (rfixed_trunc(priority_mark12) < 0) + priority_mark12.full = 0; + if (wm1.priority_mark_max.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark_max.full; + WREG32(D1MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark02)); + WREG32(D1MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark02)); + WREG32(D2MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark12)); + WREG32(D2MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark12)); + } else if (mode0) { + if (rfixed_trunc(wm0.dbpp) > 64) + a.full = rfixed_div(wm0.dbpp, wm0.num_line_pair); + else + a.full = wm0.num_line_pair.full; + fill_rate.full = rfixed_div(wm0.sclk, a); + if (wm0.consumption_rate.full > fill_rate.full) { + b.full = wm0.consumption_rate.full - fill_rate.full; + b.full = rfixed_mul(b, wm0.active_time); + a.full = rfixed_const(16); + b.full = rfixed_div(b, a); + a.full = rfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + priority_mark02.full = a.full + b.full; + } else { + a.full = rfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + b.full = rfixed_const(16); + priority_mark02.full = rfixed_div(a, b); + } + if (wm0.priority_mark.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark.full; + if (rfixed_trunc(priority_mark02) < 0) + priority_mark02.full = 0; + if (wm0.priority_mark_max.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark_max.full; + WREG32(D1MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark02)); + WREG32(D1MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark02)); + WREG32(D2MODE_PRIORITY_A_CNT, MODE_PRIORITY_OFF); + WREG32(D2MODE_PRIORITY_B_CNT, MODE_PRIORITY_OFF); + } else { + if (rfixed_trunc(wm1.dbpp) > 64) + a.full = rfixed_div(wm1.dbpp, wm1.num_line_pair); + else + a.full = wm1.num_line_pair.full; + fill_rate.full = rfixed_div(wm1.sclk, a); + if (wm1.consumption_rate.full > fill_rate.full) { + b.full = wm1.consumption_rate.full - fill_rate.full; + b.full = rfixed_mul(b, wm1.active_time); + a.full = rfixed_const(16); + b.full = rfixed_div(b, a); + a.full = rfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + priority_mark12.full = a.full + b.full; + } else { + a.full = rfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + b.full = rfixed_const(16 * 1000); + priority_mark12.full = rfixed_div(a, b); + } + if (wm1.priority_mark.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark.full; + if (rfixed_trunc(priority_mark12) < 0) + priority_mark12.full = 0; + if (wm1.priority_mark_max.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark_max.full; + WREG32(D1MODE_PRIORITY_A_CNT, MODE_PRIORITY_OFF); + WREG32(D1MODE_PRIORITY_B_CNT, MODE_PRIORITY_OFF); + WREG32(D2MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark12)); + WREG32(D2MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark12)); + } +} + +void rv515_bandwidth_update(struct radeon_device *rdev) +{ + uint32_t tmp; + struct drm_display_mode *mode0 = NULL; + struct drm_display_mode *mode1 = NULL; + + if (rdev->mode_info.crtcs[0]->base.enabled) + mode0 = &rdev->mode_info.crtcs[0]->base.mode; + if (rdev->mode_info.crtcs[1]->base.enabled) + mode1 = &rdev->mode_info.crtcs[1]->base.mode; + /* + * Set display0/1 priority up in the memory controller for + * modes if the user specifies HIGH for displaypriority + * option. + */ + if (rdev->disp_priority == 2) { + tmp = RREG32_MC(MC_MISC_LAT_TIMER); + tmp &= ~MC_DISP1R_INIT_LAT_MASK; + tmp &= ~MC_DISP0R_INIT_LAT_MASK; + if (mode1) + tmp |= (1 << MC_DISP1R_INIT_LAT_SHIFT); + if (mode0) + tmp |= (1 << MC_DISP0R_INIT_LAT_SHIFT); + WREG32_MC(MC_MISC_LAT_TIMER, tmp); + } + rv515_bandwidth_avivo_update(rdev); +} diff --git a/drivers/gpu/drm/radeon/rv515r.h b/drivers/gpu/drm/radeon/rv515r.h new file mode 100644 index 00000000000..f3cf8403990 --- /dev/null +++ b/drivers/gpu/drm/radeon/rv515r.h @@ -0,0 +1,170 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef RV515R_H +#define RV515R_H + +/* RV515 registers */ +#define PCIE_INDEX 0x0030 +#define PCIE_DATA 0x0034 +#define MC_IND_INDEX 0x0070 +#define MC_IND_WR_EN (1 << 24) +#define MC_IND_DATA 0x0074 +#define RBBM_SOFT_RESET 0x00F0 +#define CONFIG_MEMSIZE 0x00F8 +#define HDP_FB_LOCATION 0x0134 +#define CP_CSQ_CNTL 0x0740 +#define CP_CSQ_MODE 0x0744 +#define CP_CSQ_ADDR 0x07F0 +#define CP_CSQ_DATA 0x07F4 +#define CP_CSQ_STAT 0x07F8 +#define CP_CSQ2_STAT 0x07FC +#define RBBM_STATUS 0x0E40 +#define DST_PIPE_CONFIG 0x170C +#define WAIT_UNTIL 0x1720 +#define WAIT_2D_IDLE (1 << 14) +#define WAIT_3D_IDLE (1 << 15) +#define WAIT_2D_IDLECLEAN (1 << 16) +#define WAIT_3D_IDLECLEAN (1 << 17) +#define ISYNC_CNTL 0x1724 +#define ISYNC_ANY2D_IDLE3D (1 << 0) +#define ISYNC_ANY3D_IDLE2D (1 << 1) +#define ISYNC_TRIG2D_IDLE3D (1 << 2) +#define ISYNC_TRIG3D_IDLE2D (1 << 3) +#define ISYNC_WAIT_IDLEGUI (1 << 4) +#define ISYNC_CPSCRATCH_IDLEGUI (1 << 5) +#define VAP_INDEX_OFFSET 0x208C +#define VAP_PVS_STATE_FLUSH_REG 0x2284 +#define GB_ENABLE 0x4008 +#define GB_MSPOS0 0x4010 +#define MS_X0_SHIFT 0 +#define MS_Y0_SHIFT 4 +#define MS_X1_SHIFT 8 +#define MS_Y1_SHIFT 12 +#define MS_X2_SHIFT 16 +#define MS_Y2_SHIFT 20 +#define MSBD0_Y_SHIFT 24 +#define MSBD0_X_SHIFT 28 +#define GB_MSPOS1 0x4014 +#define MS_X3_SHIFT 0 +#define MS_Y3_SHIFT 4 +#define MS_X4_SHIFT 8 +#define MS_Y4_SHIFT 12 +#define MS_X5_SHIFT 16 +#define MS_Y5_SHIFT 20 +#define MSBD1_SHIFT 24 +#define GB_TILE_CONFIG 0x4018 +#define ENABLE_TILING (1 << 0) +#define PIPE_COUNT_MASK 0x0000000E +#define PIPE_COUNT_SHIFT 1 +#define TILE_SIZE_8 (0 << 4) +#define TILE_SIZE_16 (1 << 4) +#define TILE_SIZE_32 (2 << 4) +#define SUBPIXEL_1_12 (0 << 16) +#define SUBPIXEL_1_16 (1 << 16) +#define GB_SELECT 0x401C +#define GB_AA_CONFIG 0x4020 +#define GB_PIPE_SELECT 0x402C +#define GA_ENHANCE 0x4274 +#define GA_DEADLOCK_CNTL (1 << 0) +#define GA_FASTSYNC_CNTL (1 << 1) +#define GA_POLY_MODE 0x4288 +#define FRONT_PTYPE_POINT (0 << 4) +#define FRONT_PTYPE_LINE (1 << 4) +#define FRONT_PTYPE_TRIANGE (2 << 4) +#define BACK_PTYPE_POINT (0 << 7) +#define BACK_PTYPE_LINE (1 << 7) +#define BACK_PTYPE_TRIANGE (2 << 7) +#define GA_ROUND_MODE 0x428C +#define GEOMETRY_ROUND_TRUNC (0 << 0) +#define GEOMETRY_ROUND_NEAREST (1 << 0) +#define COLOR_ROUND_TRUNC (0 << 2) +#define COLOR_ROUND_NEAREST (1 << 2) +#define SU_REG_DEST 0x42C8 +#define RB3D_DSTCACHE_CTLSTAT 0x4E4C +#define RB3D_DC_FLUSH (2 << 0) +#define RB3D_DC_FREE (2 << 2) +#define RB3D_DC_FINISH (1 << 4) +#define ZB_ZCACHE_CTLSTAT 0x4F18 +#define ZC_FLUSH (1 << 0) +#define ZC_FREE (1 << 1) +#define DC_LB_MEMORY_SPLIT 0x6520 +#define DC_LB_MEMORY_SPLIT_MASK 0x00000003 +#define DC_LB_MEMORY_SPLIT_SHIFT 0 +#define DC_LB_MEMORY_SPLIT_D1HALF_D2HALF 0 +#define DC_LB_MEMORY_SPLIT_D1_3Q_D2_1Q 1 +#define DC_LB_MEMORY_SPLIT_D1_ONLY 2 +#define DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q 3 +#define DC_LB_MEMORY_SPLIT_SHIFT_MODE (1 << 2) +#define DC_LB_DISP1_END_ADR_SHIFT 4 +#define DC_LB_DISP1_END_ADR_MASK 0x00007FF0 +#define D1MODE_PRIORITY_A_CNT 0x6548 +#define MODE_PRIORITY_MARK_MASK 0x00007FFF +#define MODE_PRIORITY_OFF (1 << 16) +#define MODE_PRIORITY_ALWAYS_ON (1 << 20) +#define MODE_PRIORITY_FORCE_MASK (1 << 24) +#define D1MODE_PRIORITY_B_CNT 0x654C +#define LB_MAX_REQ_OUTSTANDING 0x6D58 +#define LB_D1_MAX_REQ_OUTSTANDING_MASK 0x0000000F +#define LB_D1_MAX_REQ_OUTSTANDING_SHIFT 0 +#define LB_D2_MAX_REQ_OUTSTANDING_MASK 0x000F0000 +#define LB_D2_MAX_REQ_OUTSTANDING_SHIFT 16 +#define D2MODE_PRIORITY_A_CNT 0x6D48 +#define D2MODE_PRIORITY_B_CNT 0x6D4C + +/* ix[MC] registers */ +#define MC_FB_LOCATION 0x01 +#define MC_FB_START_MASK 0x0000FFFF +#define MC_FB_START_SHIFT 0 +#define MC_FB_TOP_MASK 0xFFFF0000 +#define MC_FB_TOP_SHIFT 16 +#define MC_AGP_LOCATION 0x02 +#define MC_AGP_START_MASK 0x0000FFFF +#define MC_AGP_START_SHIFT 0 +#define MC_AGP_TOP_MASK 0xFFFF0000 +#define MC_AGP_TOP_SHIFT 16 +#define MC_AGP_BASE 0x03 +#define MC_AGP_BASE_2 0x04 +#define MC_CNTL 0x5 +#define MEM_NUM_CHANNELS_MASK 0x00000003 +#define MC_STATUS 0x08 +#define MC_STATUS_IDLE (1 << 4) +#define MC_MISC_LAT_TIMER 0x09 +#define MC_CPR_INIT_LAT_MASK 0x0000000F +#define MC_VF_INIT_LAT_MASK 0x000000F0 +#define MC_DISP0R_INIT_LAT_MASK 0x00000F00 +#define MC_DISP0R_INIT_LAT_SHIFT 8 +#define MC_DISP1R_INIT_LAT_MASK 0x0000F000 +#define MC_DISP1R_INIT_LAT_SHIFT 12 +#define MC_FIXED_INIT_LAT_MASK 0x000F0000 +#define MC_E2R_INIT_LAT_MASK 0x00F00000 +#define SAME_PAGE_PRIO_MASK 0x0F000000 +#define MC_GLOBW_INIT_LAT_MASK 0xF0000000 + + +#endif + diff --git a/drivers/gpu/drm/radeon/rv770.c b/drivers/gpu/drm/radeon/rv770.c index da50cc51ede..21d8ffd5730 100644 --- a/drivers/gpu/drm/radeon/rv770.c +++ b/drivers/gpu/drm/radeon/rv770.c @@ -67,7 +67,7 @@ int rv770_mc_init(struct radeon_device *rdev) "programming pipes. Bad things might happen.\n"); } - tmp = rdev->mc.vram_location + rdev->mc.vram_size - 1; + tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; tmp = REG_SET(R700_MC_FB_TOP, tmp >> 24); tmp |= REG_SET(R700_MC_FB_BASE, rdev->mc.vram_location >> 24); WREG32(R700_MC_VM_FB_LOCATION, tmp); diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index c1c407f7cca..6538d423698 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -43,7 +43,6 @@ #define TTM_BO_HASH_ORDER 13 static int ttm_bo_setup_vm(struct ttm_buffer_object *bo); -static void ttm_bo_unmap_virtual(struct ttm_buffer_object *bo); static int ttm_bo_swapout(struct ttm_mem_shrink *shrink); static inline uint32_t ttm_bo_type_flags(unsigned type) @@ -224,6 +223,9 @@ static int ttm_bo_add_ttm(struct ttm_buffer_object *bo, bool zero_alloc) TTM_ASSERT_LOCKED(&bo->mutex); bo->ttm = NULL; + if (bdev->need_dma32) + page_flags |= TTM_PAGE_FLAG_DMA32; + switch (bo->type) { case ttm_bo_type_device: if (zero_alloc) @@ -304,6 +306,9 @@ static int ttm_bo_handle_move_mem(struct ttm_buffer_object *bo, } + if (bdev->driver->move_notify) + bdev->driver->move_notify(bo, mem); + if (!(old_man->flags & TTM_MEMTYPE_FLAG_FIXED) && !(new_man->flags & TTM_MEMTYPE_FLAG_FIXED)) ret = ttm_bo_move_ttm(bo, evict, no_wait, mem); @@ -655,31 +660,52 @@ retry_pre_get: return 0; } +static uint32_t ttm_bo_select_caching(struct ttm_mem_type_manager *man, + uint32_t cur_placement, + uint32_t proposed_placement) +{ + uint32_t caching = proposed_placement & TTM_PL_MASK_CACHING; + uint32_t result = proposed_placement & ~TTM_PL_MASK_CACHING; + + /** + * Keep current caching if possible. + */ + + if ((cur_placement & caching) != 0) + result |= (cur_placement & caching); + else if ((man->default_caching & caching) != 0) + result |= man->default_caching; + else if ((TTM_PL_FLAG_CACHED & caching) != 0) + result |= TTM_PL_FLAG_CACHED; + else if ((TTM_PL_FLAG_WC & caching) != 0) + result |= TTM_PL_FLAG_WC; + else if ((TTM_PL_FLAG_UNCACHED & caching) != 0) + result |= TTM_PL_FLAG_UNCACHED; + + return result; +} + + static bool ttm_bo_mt_compatible(struct ttm_mem_type_manager *man, bool disallow_fixed, uint32_t mem_type, - uint32_t mask, uint32_t *res_mask) + uint32_t proposed_placement, + uint32_t *masked_placement) { uint32_t cur_flags = ttm_bo_type_flags(mem_type); if ((man->flags & TTM_MEMTYPE_FLAG_FIXED) && disallow_fixed) return false; - if ((cur_flags & mask & TTM_PL_MASK_MEM) == 0) + if ((cur_flags & proposed_placement & TTM_PL_MASK_MEM) == 0) return false; - if ((mask & man->available_caching) == 0) + if ((proposed_placement & man->available_caching) == 0) return false; - if (mask & man->default_caching) - cur_flags |= man->default_caching; - else if (mask & TTM_PL_FLAG_CACHED) - cur_flags |= TTM_PL_FLAG_CACHED; - else if (mask & TTM_PL_FLAG_WC) - cur_flags |= TTM_PL_FLAG_WC; - else - cur_flags |= TTM_PL_FLAG_UNCACHED; - *res_mask = cur_flags; + cur_flags |= (proposed_placement & man->available_caching); + + *masked_placement = cur_flags; return true; } @@ -723,6 +749,9 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, if (!type_ok) continue; + cur_flags = ttm_bo_select_caching(man, bo->mem.placement, + cur_flags); + if (mem_type == TTM_PL_SYSTEM) break; @@ -779,6 +808,9 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, proposed_placement, &cur_flags)) continue; + cur_flags = ttm_bo_select_caching(man, bo->mem.placement, + cur_flags); + ret = ttm_bo_mem_force_space(bdev, mem, mem_type, interruptible, no_wait); @@ -1305,7 +1337,8 @@ EXPORT_SYMBOL(ttm_bo_device_release); int ttm_bo_device_init(struct ttm_bo_device *bdev, struct ttm_mem_global *mem_glob, - struct ttm_bo_driver *driver, uint64_t file_page_offset) + struct ttm_bo_driver *driver, uint64_t file_page_offset, + bool need_dma32) { int ret = -EINVAL; @@ -1342,6 +1375,7 @@ int ttm_bo_device_init(struct ttm_bo_device *bdev, INIT_LIST_HEAD(&bdev->ddestroy); INIT_LIST_HEAD(&bdev->swap_lru); bdev->dev_mapping = NULL; + bdev->need_dma32 = need_dma32; ttm_mem_init_shrink(&bdev->shrink, ttm_bo_swapout); ret = ttm_mem_register_shrink(mem_glob, &bdev->shrink); if (unlikely(ret != 0)) { @@ -1419,6 +1453,7 @@ void ttm_bo_unmap_virtual(struct ttm_buffer_object *bo) unmap_mapping_range(bdev->dev_mapping, offset, holelen, 1); } +EXPORT_SYMBOL(ttm_bo_unmap_virtual); static void ttm_bo_vm_insert_rb(struct ttm_buffer_object *bo) { diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c index bdec583901e..ce2e6f38ea0 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_util.c +++ b/drivers/gpu/drm/ttm/ttm_bo_util.c @@ -136,7 +136,8 @@ static int ttm_copy_io_page(void *dst, void *src, unsigned long page) } static int ttm_copy_io_ttm_page(struct ttm_tt *ttm, void *src, - unsigned long page) + unsigned long page, + pgprot_t prot) { struct page *d = ttm_tt_get_page(ttm, page); void *dst; @@ -145,17 +146,35 @@ static int ttm_copy_io_ttm_page(struct ttm_tt *ttm, void *src, return -ENOMEM; src = (void *)((unsigned long)src + (page << PAGE_SHIFT)); - dst = kmap(d); + +#ifdef CONFIG_X86 + dst = kmap_atomic_prot(d, KM_USER0, prot); +#else + if (prot != PAGE_KERNEL) + dst = vmap(&d, 1, 0, prot); + else + dst = kmap(d); +#endif if (!dst) return -ENOMEM; memcpy_fromio(dst, src, PAGE_SIZE); - kunmap(d); + +#ifdef CONFIG_X86 + kunmap_atomic(dst, KM_USER0); +#else + if (prot != PAGE_KERNEL) + vunmap(dst); + else + kunmap(d); +#endif + return 0; } static int ttm_copy_ttm_io_page(struct ttm_tt *ttm, void *dst, - unsigned long page) + unsigned long page, + pgprot_t prot) { struct page *s = ttm_tt_get_page(ttm, page); void *src; @@ -164,12 +183,28 @@ static int ttm_copy_ttm_io_page(struct ttm_tt *ttm, void *dst, return -ENOMEM; dst = (void *)((unsigned long)dst + (page << PAGE_SHIFT)); - src = kmap(s); +#ifdef CONFIG_X86 + src = kmap_atomic_prot(s, KM_USER0, prot); +#else + if (prot != PAGE_KERNEL) + src = vmap(&s, 1, 0, prot); + else + src = kmap(s); +#endif if (!src) return -ENOMEM; memcpy_toio(dst, src, PAGE_SIZE); - kunmap(s); + +#ifdef CONFIG_X86 + kunmap_atomic(src, KM_USER0); +#else + if (prot != PAGE_KERNEL) + vunmap(src); + else + kunmap(s); +#endif + return 0; } @@ -214,11 +249,17 @@ int ttm_bo_move_memcpy(struct ttm_buffer_object *bo, for (i = 0; i < new_mem->num_pages; ++i) { page = i * dir + add; - if (old_iomap == NULL) - ret = ttm_copy_ttm_io_page(ttm, new_iomap, page); - else if (new_iomap == NULL) - ret = ttm_copy_io_ttm_page(ttm, old_iomap, page); - else + if (old_iomap == NULL) { + pgprot_t prot = ttm_io_prot(old_mem->placement, + PAGE_KERNEL); + ret = ttm_copy_ttm_io_page(ttm, new_iomap, page, + prot); + } else if (new_iomap == NULL) { + pgprot_t prot = ttm_io_prot(new_mem->placement, + PAGE_KERNEL); + ret = ttm_copy_io_ttm_page(ttm, old_iomap, page, + prot); + } else ret = ttm_copy_io_page(new_iomap, old_iomap, page); if (ret) goto out1; @@ -509,8 +550,8 @@ int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo, if (evict) { ret = ttm_bo_wait(bo, false, false, false); spin_unlock(&bo->lock); - driver->sync_obj_unref(&bo->sync_obj); - + if (tmp_obj) + driver->sync_obj_unref(&tmp_obj); if (ret) return ret; @@ -532,6 +573,8 @@ int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo, set_bit(TTM_BO_PRIV_FLAG_MOVING, &bo->priv_flags); spin_unlock(&bo->lock); + if (tmp_obj) + driver->sync_obj_unref(&tmp_obj); ret = ttm_buffer_object_transfer(bo, &ghost_obj); if (ret) diff --git a/drivers/gpu/drm/ttm/ttm_bo_vm.c b/drivers/gpu/drm/ttm/ttm_bo_vm.c index fe949a12fe4..33de7637c0c 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_vm.c +++ b/drivers/gpu/drm/ttm/ttm_bo_vm.c @@ -101,6 +101,9 @@ static int ttm_bo_vm_fault(struct vm_area_struct *vma, struct vm_fault *vmf) return VM_FAULT_NOPAGE; } + if (bdev->driver->fault_reserve_notify) + bdev->driver->fault_reserve_notify(bo); + /* * Wait for buffer data in transit, due to a pipelined * move. diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index 75dc8bd2459..b8b6c4a5f98 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -86,10 +86,16 @@ void ttm_tt_cache_flush(struct page *pages[], unsigned long num_pages) unsigned long i; for (i = 0; i < num_pages; ++i) { - if (pages[i]) { - unsigned long start = (unsigned long)page_address(pages[i]); - flush_dcache_range(start, start + PAGE_SIZE); - } + struct page *page = pages[i]; + void *page_virtual; + + if (unlikely(page == NULL)) + continue; + + page_virtual = kmap_atomic(page, KM_USER0); + flush_dcache_range((unsigned long) page_virtual, + (unsigned long) page_virtual + PAGE_SIZE); + kunmap_atomic(page_virtual, KM_USER0); } #else if (on_each_cpu(ttm_tt_ipi_handler, NULL, 1) != 0) @@ -131,10 +137,17 @@ static void ttm_tt_free_page_directory(struct ttm_tt *ttm) static struct page *ttm_tt_alloc_page(unsigned page_flags) { + gfp_t gfp_flags = GFP_USER; + if (page_flags & TTM_PAGE_FLAG_ZERO_ALLOC) - return alloc_page(GFP_HIGHUSER | __GFP_ZERO); + gfp_flags |= __GFP_ZERO; + + if (page_flags & TTM_PAGE_FLAG_DMA32) + gfp_flags |= __GFP_DMA32; + else + gfp_flags |= __GFP_HIGHMEM; - return alloc_page(GFP_HIGHUSER); + return alloc_page(gfp_flags); } static void ttm_tt_free_user_pages(struct ttm_tt *ttm) diff --git a/drivers/hwmon/asus_atk0110.c b/drivers/hwmon/asus_atk0110.c index bff0103610c..fe4fa29c921 100644 --- a/drivers/hwmon/asus_atk0110.c +++ b/drivers/hwmon/asus_atk0110.c @@ -593,7 +593,11 @@ static int atk_add_sensor(struct atk_data *data, union acpi_object *obj) sensor->data = data; sensor->id = flags->integer.value; sensor->limit1 = limit1->integer.value; - sensor->limit2 = limit2->integer.value; + if (data->old_interface) + sensor->limit2 = limit2->integer.value; + else + /* The upper limit is expressed as delta from lower limit */ + sensor->limit2 = sensor->limit1 + limit2->integer.value; snprintf(sensor->input_attr_name, ATTR_NAME_SIZE, "%s%d_input", base_name, start + *num); diff --git a/drivers/hwmon/smsc47m1.c b/drivers/hwmon/smsc47m1.c index a92dbb97ee9..ba75bfcf14c 100644 --- a/drivers/hwmon/smsc47m1.c +++ b/drivers/hwmon/smsc47m1.c @@ -86,6 +86,7 @@ superio_exit(void) #define SUPERIO_REG_ACT 0x30 #define SUPERIO_REG_BASE 0x60 #define SUPERIO_REG_DEVID 0x20 +#define SUPERIO_REG_DEVREV 0x21 /* Logical device registers */ @@ -429,6 +430,9 @@ static int __init smsc47m1_find(unsigned short *addr, * The LPC47M292 (device id 0x6B) is somewhat compatible, but it * supports a 3rd fan, and the pin configuration registers are * unfortunately different. + * The LPC47M233 has the same device id (0x6B) but is not compatible. + * We check the high bit of the device revision register to + * differentiate them. */ switch (val) { case 0x51: @@ -448,6 +452,13 @@ static int __init smsc47m1_find(unsigned short *addr, sio_data->type = smsc47m1; break; case 0x6B: + if (superio_inb(SUPERIO_REG_DEVREV) & 0x80) { + pr_debug(DRVNAME ": " + "Found SMSC LPC47M233, unsupported\n"); + superio_exit(); + return -ENODEV; + } + pr_info(DRVNAME ": Found SMSC LPC47M292\n"); sio_data->type = smsc47m2; break; diff --git a/drivers/i2c/busses/i2c-omap.c b/drivers/i2c/busses/i2c-omap.c index fdd83277c8a..d258b02aef4 100644 --- a/drivers/i2c/busses/i2c-omap.c +++ b/drivers/i2c/busses/i2c-omap.c @@ -672,9 +672,10 @@ omap_i2c_isr(int this_irq, void *dev_id) break; } + err = 0; +complete: omap_i2c_write_reg(dev, OMAP_I2C_STAT_REG, stat); - err = 0; if (stat & OMAP_I2C_STAT_NACK) { err |= OMAP_I2C_STAT_NACK; omap_i2c_write_reg(dev, OMAP_I2C_CON_REG, @@ -685,16 +686,19 @@ omap_i2c_isr(int this_irq, void *dev_id) err |= OMAP_I2C_STAT_AL; } if (stat & (OMAP_I2C_STAT_ARDY | OMAP_I2C_STAT_NACK | - OMAP_I2C_STAT_AL)) + OMAP_I2C_STAT_AL)) { omap_i2c_complete_cmd(dev, err); + return IRQ_HANDLED; + } if (stat & (OMAP_I2C_STAT_RRDY | OMAP_I2C_STAT_RDR)) { u8 num_bytes = 1; if (dev->fifo_size) { if (stat & OMAP_I2C_STAT_RRDY) num_bytes = dev->fifo_size; - else - num_bytes = omap_i2c_read_reg(dev, - OMAP_I2C_BUFSTAT_REG); + else /* read RXSTAT on RDR interrupt */ + num_bytes = (omap_i2c_read_reg(dev, + OMAP_I2C_BUFSTAT_REG) + >> 8) & 0x3F; } while (num_bytes) { num_bytes--; @@ -731,9 +735,10 @@ omap_i2c_isr(int this_irq, void *dev_id) if (dev->fifo_size) { if (stat & OMAP_I2C_STAT_XRDY) num_bytes = dev->fifo_size; - else + else /* read TXSTAT on XDR interrupt */ num_bytes = omap_i2c_read_reg(dev, - OMAP_I2C_BUFSTAT_REG); + OMAP_I2C_BUFSTAT_REG) + & 0x3F; } while (num_bytes) { num_bytes--; @@ -760,6 +765,27 @@ omap_i2c_isr(int this_irq, void *dev_id) "data to send\n"); break; } + + /* + * OMAP3430 Errata 1.153: When an XRDY/XDR + * is hit, wait for XUDF before writing data + * to DATA_REG. Otherwise some data bytes can + * be lost while transferring them from the + * memory to the I2C interface. + */ + + if (cpu_is_omap34xx()) { + while (!(stat & OMAP_I2C_STAT_XUDF)) { + if (stat & (OMAP_I2C_STAT_NACK | OMAP_I2C_STAT_AL)) { + omap_i2c_ack_stat(dev, stat & (OMAP_I2C_STAT_XRDY | OMAP_I2C_STAT_XDR)); + err |= OMAP_I2C_STAT_XUDF; + goto complete; + } + cpu_relax(); + stat = omap_i2c_read_reg(dev, OMAP_I2C_STAT_REG); + } + } + omap_i2c_write_reg(dev, OMAP_I2C_DATA_REG, w); } omap_i2c_ack_stat(dev, @@ -879,7 +905,7 @@ omap_i2c_probe(struct platform_device *pdev) i2c_set_adapdata(adap, dev); adap->owner = THIS_MODULE; adap->class = I2C_CLASS_HWMON; - strncpy(adap->name, "OMAP I2C adapter", sizeof(adap->name)); + strlcpy(adap->name, "OMAP I2C adapter", sizeof(adap->name)); adap->algo = &omap_i2c_algo; adap->dev.parent = &pdev->dev; diff --git a/drivers/i2c/busses/i2c-s3c2410.c b/drivers/i2c/busses/i2c-s3c2410.c index 8f42a4536cd..20bb0ceb027 100644 --- a/drivers/i2c/busses/i2c-s3c2410.c +++ b/drivers/i2c/busses/i2c-s3c2410.c @@ -763,11 +763,6 @@ static int s3c24xx_i2c_init(struct s3c24xx_i2c *i2c) dev_info(i2c->dev, "bus frequency set to %d KHz\n", freq); dev_dbg(i2c->dev, "S3C2410_IICCON=0x%02lx\n", iicon); - /* check for s3c2440 i2c controller */ - - if (s3c24xx_i2c_is2440(i2c)) - writel(0x0, i2c->regs + S3C2440_IICLC); - return 0; } diff --git a/drivers/i2c/busses/i2c-sh_mobile.c b/drivers/i2c/busses/i2c-sh_mobile.c index 4f3d99cd169..820487d0d5c 100644 --- a/drivers/i2c/busses/i2c-sh_mobile.c +++ b/drivers/i2c/busses/i2c-sh_mobile.c @@ -637,7 +637,7 @@ static void __exit sh_mobile_i2c_adap_exit(void) platform_driver_unregister(&sh_mobile_i2c_driver); } -module_init(sh_mobile_i2c_adap_init); +subsys_initcall(sh_mobile_i2c_adap_init); module_exit(sh_mobile_i2c_adap_exit); MODULE_DESCRIPTION("SuperH Mobile I2C Bus Controller driver"); diff --git a/drivers/i2c/chips/tsl2550.c b/drivers/i2c/chips/tsl2550.c index 1a9cc135219..b96f3025e58 100644 --- a/drivers/i2c/chips/tsl2550.c +++ b/drivers/i2c/chips/tsl2550.c @@ -27,7 +27,7 @@ #include <linux/delay.h> #define TSL2550_DRV_NAME "tsl2550" -#define DRIVER_VERSION "1.1.1" +#define DRIVER_VERSION "1.1.2" /* * Defines @@ -189,13 +189,16 @@ static int tsl2550_calculate_lux(u8 ch0, u8 ch1) u8 r = 128; /* Avoid division by 0 and count 1 cannot be greater than count 0 */ - if (c0 && (c1 <= c0)) - r = c1 * 128 / c0; + if (c1 <= c0) + if (c0) { + r = c1 * 128 / c0; + + /* Calculate LUX */ + lux = ((c0 - c1) * ratio_lut[r]) / 256; + } else + lux = 0; else - return -1; - - /* Calculate LUX */ - lux = ((c0 - c1) * ratio_lut[r]) / 256; + return -EAGAIN; /* LUX range check */ return lux > TSL2550_MAX_LUX ? TSL2550_MAX_LUX : lux; diff --git a/drivers/isdn/mISDN/l1oip_core.c b/drivers/isdn/mISDN/l1oip_core.c index 990e6a7e667..c3b661a666c 100644 --- a/drivers/isdn/mISDN/l1oip_core.c +++ b/drivers/isdn/mISDN/l1oip_core.c @@ -731,10 +731,10 @@ l1oip_socket_thread(void *data) while (!signal_pending(current)) { struct kvec iov = { .iov_base = recvbuf, - .iov_len = sizeof(recvbuf), + .iov_len = recvbuf_size, }; recvlen = kernel_recvmsg(socket, &msg, &iov, 1, - sizeof(recvbuf), 0); + recvbuf_size, 0); if (recvlen > 0) { l1oip_socket_parse(hc, &sin_rx, recvbuf, recvlen); } else { diff --git a/drivers/lguest/core.c b/drivers/lguest/core.c index a6974e9b8eb..1e2cb846b3c 100644 --- a/drivers/lguest/core.c +++ b/drivers/lguest/core.c @@ -1,6 +1,8 @@ -/*P:400 This contains run_guest() which actually calls into the Host<->Guest +/*P:400 + * This contains run_guest() which actually calls into the Host<->Guest * Switcher and analyzes the return, such as determining if the Guest wants the - * Host to do something. This file also contains useful helper routines. :*/ + * Host to do something. This file also contains useful helper routines. +:*/ #include <linux/module.h> #include <linux/stringify.h> #include <linux/stddef.h> @@ -24,7 +26,8 @@ static struct page **switcher_page; /* This One Big lock protects all inter-guest data structures. */ DEFINE_MUTEX(lguest_lock); -/*H:010 We need to set up the Switcher at a high virtual address. Remember the +/*H:010 + * We need to set up the Switcher at a high virtual address. Remember the * Switcher is a few hundred bytes of assembler code which actually changes the * CPU to run the Guest, and then changes back to the Host when a trap or * interrupt happens. @@ -33,7 +36,8 @@ DEFINE_MUTEX(lguest_lock); * Host since it will be running as the switchover occurs. * * Trying to map memory at a particular address is an unusual thing to do, so - * it's not a simple one-liner. */ + * it's not a simple one-liner. + */ static __init int map_switcher(void) { int i, err; @@ -47,8 +51,10 @@ static __init int map_switcher(void) * easy. */ - /* We allocate an array of struct page pointers. map_vm_area() wants - * this, rather than just an array of pages. */ + /* + * We allocate an array of struct page pointers. map_vm_area() wants + * this, rather than just an array of pages. + */ switcher_page = kmalloc(sizeof(switcher_page[0])*TOTAL_SWITCHER_PAGES, GFP_KERNEL); if (!switcher_page) { @@ -56,8 +62,10 @@ static __init int map_switcher(void) goto out; } - /* Now we actually allocate the pages. The Guest will see these pages, - * so we make sure they're zeroed. */ + /* + * Now we actually allocate the pages. The Guest will see these pages, + * so we make sure they're zeroed. + */ for (i = 0; i < TOTAL_SWITCHER_PAGES; i++) { unsigned long addr = get_zeroed_page(GFP_KERNEL); if (!addr) { @@ -67,19 +75,23 @@ static __init int map_switcher(void) switcher_page[i] = virt_to_page(addr); } - /* First we check that the Switcher won't overlap the fixmap area at + /* + * First we check that the Switcher won't overlap the fixmap area at * the top of memory. It's currently nowhere near, but it could have - * very strange effects if it ever happened. */ + * very strange effects if it ever happened. + */ if (SWITCHER_ADDR + (TOTAL_SWITCHER_PAGES+1)*PAGE_SIZE > FIXADDR_START){ err = -ENOMEM; printk("lguest: mapping switcher would thwack fixmap\n"); goto free_pages; } - /* Now we reserve the "virtual memory area" we want: 0xFFC00000 + /* + * Now we reserve the "virtual memory area" we want: 0xFFC00000 * (SWITCHER_ADDR). We might not get it in theory, but in practice * it's worked so far. The end address needs +1 because __get_vm_area - * allocates an extra guard page, so we need space for that. */ + * allocates an extra guard page, so we need space for that. + */ switcher_vma = __get_vm_area(TOTAL_SWITCHER_PAGES * PAGE_SIZE, VM_ALLOC, SWITCHER_ADDR, SWITCHER_ADDR + (TOTAL_SWITCHER_PAGES+1) * PAGE_SIZE); @@ -89,11 +101,13 @@ static __init int map_switcher(void) goto free_pages; } - /* This code actually sets up the pages we've allocated to appear at + /* + * This code actually sets up the pages we've allocated to appear at * SWITCHER_ADDR. map_vm_area() takes the vma we allocated above, the * kind of pages we're mapping (kernel pages), and a pointer to our * array of struct pages. It increments that pointer, but we don't - * care. */ + * care. + */ pagep = switcher_page; err = map_vm_area(switcher_vma, PAGE_KERNEL_EXEC, &pagep); if (err) { @@ -101,8 +115,10 @@ static __init int map_switcher(void) goto free_vma; } - /* Now the Switcher is mapped at the right address, we can't fail! - * Copy in the compiled-in Switcher code (from <arch>_switcher.S). */ + /* + * Now the Switcher is mapped at the right address, we can't fail! + * Copy in the compiled-in Switcher code (from <arch>_switcher.S). + */ memcpy(switcher_vma->addr, start_switcher_text, end_switcher_text - start_switcher_text); @@ -124,8 +140,7 @@ out: } /*:*/ -/* Cleaning up the mapping when the module is unloaded is almost... - * too easy. */ +/* Cleaning up the mapping when the module is unloaded is almost... too easy. */ static void unmap_switcher(void) { unsigned int i; @@ -151,16 +166,19 @@ static void unmap_switcher(void) * But we can't trust the Guest: it might be trying to access the Launcher * code. We have to check that the range is below the pfn_limit the Launcher * gave us. We have to make sure that addr + len doesn't give us a false - * positive by overflowing, too. */ + * positive by overflowing, too. + */ bool lguest_address_ok(const struct lguest *lg, unsigned long addr, unsigned long len) { return (addr+len) / PAGE_SIZE < lg->pfn_limit && (addr+len >= addr); } -/* This routine copies memory from the Guest. Here we can see how useful the +/* + * This routine copies memory from the Guest. Here we can see how useful the * kill_lguest() routine we met in the Launcher can be: we return a random - * value (all zeroes) instead of needing to return an error. */ + * value (all zeroes) instead of needing to return an error. + */ void __lgread(struct lg_cpu *cpu, void *b, unsigned long addr, unsigned bytes) { if (!lguest_address_ok(cpu->lg, addr, bytes) @@ -181,9 +199,11 @@ void __lgwrite(struct lg_cpu *cpu, unsigned long addr, const void *b, } /*:*/ -/*H:030 Let's jump straight to the the main loop which runs the Guest. +/*H:030 + * Let's jump straight to the the main loop which runs the Guest. * Remember, this is called by the Launcher reading /dev/lguest, and we keep - * going around and around until something interesting happens. */ + * going around and around until something interesting happens. + */ int run_guest(struct lg_cpu *cpu, unsigned long __user *user) { /* We stop running once the Guest is dead. */ @@ -195,10 +215,17 @@ int run_guest(struct lg_cpu *cpu, unsigned long __user *user) if (cpu->hcall) do_hypercalls(cpu); - /* It's possible the Guest did a NOTIFY hypercall to the - * Launcher, in which case we return from the read() now. */ + /* + * It's possible the Guest did a NOTIFY hypercall to the + * Launcher. + */ if (cpu->pending_notify) { + /* + * Does it just needs to write to a registered + * eventfd (ie. the appropriate virtqueue thread)? + */ if (!send_notify_to_eventfd(cpu)) { + /* OK, we tell the main Laucher. */ if (put_user(cpu->pending_notify, user)) return -EFAULT; return sizeof(cpu->pending_notify); @@ -209,29 +236,39 @@ int run_guest(struct lg_cpu *cpu, unsigned long __user *user) if (signal_pending(current)) return -ERESTARTSYS; - /* Check if there are any interrupts which can be delivered now: + /* + * Check if there are any interrupts which can be delivered now: * if so, this sets up the hander to be executed when we next - * run the Guest. */ + * run the Guest. + */ irq = interrupt_pending(cpu, &more); if (irq < LGUEST_IRQS) try_deliver_interrupt(cpu, irq, more); - /* All long-lived kernel loops need to check with this horrible + /* + * All long-lived kernel loops need to check with this horrible * thing called the freezer. If the Host is trying to suspend, - * it stops us. */ + * it stops us. + */ try_to_freeze(); - /* Just make absolutely sure the Guest is still alive. One of - * those hypercalls could have been fatal, for example. */ + /* + * Just make absolutely sure the Guest is still alive. One of + * those hypercalls could have been fatal, for example. + */ if (cpu->lg->dead) break; - /* If the Guest asked to be stopped, we sleep. The Guest's - * clock timer will wake us. */ + /* + * If the Guest asked to be stopped, we sleep. The Guest's + * clock timer will wake us. + */ if (cpu->halted) { set_current_state(TASK_INTERRUPTIBLE); - /* Just before we sleep, make sure no interrupt snuck in - * which we should be doing. */ + /* + * Just before we sleep, make sure no interrupt snuck in + * which we should be doing. + */ if (interrupt_pending(cpu, &more) < LGUEST_IRQS) set_current_state(TASK_RUNNING); else @@ -239,8 +276,10 @@ int run_guest(struct lg_cpu *cpu, unsigned long __user *user) continue; } - /* OK, now we're ready to jump into the Guest. First we put up - * the "Do Not Disturb" sign: */ + /* + * OK, now we're ready to jump into the Guest. First we put up + * the "Do Not Disturb" sign: + */ local_irq_disable(); /* Actually run the Guest until something happens. */ @@ -327,8 +366,10 @@ static void __exit fini(void) } /*:*/ -/* The Host side of lguest can be a module. This is a nice way for people to - * play with it. */ +/* + * The Host side of lguest can be a module. This is a nice way for people to + * play with it. + */ module_init(init); module_exit(fini); MODULE_LICENSE("GPL"); diff --git a/drivers/lguest/hypercalls.c b/drivers/lguest/hypercalls.c index c29ffa19cb7..83511eb0923 100644 --- a/drivers/lguest/hypercalls.c +++ b/drivers/lguest/hypercalls.c @@ -1,8 +1,10 @@ -/*P:500 Just as userspace programs request kernel operations through a system +/*P:500 + * Just as userspace programs request kernel operations through a system * call, the Guest requests Host operations through a "hypercall". You might * notice this nomenclature doesn't really follow any logic, but the name has * been around for long enough that we're stuck with it. As you'd expect, this - * code is basically a one big switch statement. :*/ + * code is basically a one big switch statement. +:*/ /* Copyright (C) 2006 Rusty Russell IBM Corporation @@ -28,30 +30,41 @@ #include <asm/pgtable.h> #include "lg.h" -/*H:120 This is the core hypercall routine: where the Guest gets what it wants. - * Or gets killed. Or, in the case of LHCALL_SHUTDOWN, both. */ +/*H:120 + * This is the core hypercall routine: where the Guest gets what it wants. + * Or gets killed. Or, in the case of LHCALL_SHUTDOWN, both. + */ static void do_hcall(struct lg_cpu *cpu, struct hcall_args *args) { switch (args->arg0) { case LHCALL_FLUSH_ASYNC: - /* This call does nothing, except by breaking out of the Guest - * it makes us process all the asynchronous hypercalls. */ + /* + * This call does nothing, except by breaking out of the Guest + * it makes us process all the asynchronous hypercalls. + */ break; case LHCALL_SEND_INTERRUPTS: - /* This call does nothing too, but by breaking out of the Guest - * it makes us process any pending interrupts. */ + /* + * This call does nothing too, but by breaking out of the Guest + * it makes us process any pending interrupts. + */ break; case LHCALL_LGUEST_INIT: - /* You can't get here unless you're already initialized. Don't - * do that. */ + /* + * You can't get here unless you're already initialized. Don't + * do that. + */ kill_guest(cpu, "already have lguest_data"); break; case LHCALL_SHUTDOWN: { - /* Shutdown is such a trivial hypercall that we do it in four - * lines right here. */ char msg[128]; - /* If the lgread fails, it will call kill_guest() itself; the - * kill_guest() with the message will be ignored. */ + /* + * Shutdown is such a trivial hypercall that we do it in five + * lines right here. + * + * If the lgread fails, it will call kill_guest() itself; the + * kill_guest() with the message will be ignored. + */ __lgread(cpu, msg, args->arg1, sizeof(msg)); msg[sizeof(msg)-1] = '\0'; kill_guest(cpu, "CRASH: %s", msg); @@ -60,16 +73,17 @@ static void do_hcall(struct lg_cpu *cpu, struct hcall_args *args) break; } case LHCALL_FLUSH_TLB: - /* FLUSH_TLB comes in two flavors, depending on the - * argument: */ + /* FLUSH_TLB comes in two flavors, depending on the argument: */ if (args->arg1) guest_pagetable_clear_all(cpu); else guest_pagetable_flush_user(cpu); break; - /* All these calls simply pass the arguments through to the right - * routines. */ + /* + * All these calls simply pass the arguments through to the right + * routines. + */ case LHCALL_NEW_PGTABLE: guest_new_pagetable(cpu, args->arg1); break; @@ -112,15 +126,16 @@ static void do_hcall(struct lg_cpu *cpu, struct hcall_args *args) kill_guest(cpu, "Bad hypercall %li\n", args->arg0); } } -/*:*/ -/*H:124 Asynchronous hypercalls are easy: we just look in the array in the +/*H:124 + * Asynchronous hypercalls are easy: we just look in the array in the * Guest's "struct lguest_data" to see if any new ones are marked "ready". * * We are careful to do these in order: obviously we respect the order the * Guest put them in the ring, but we also promise the Guest that they will * happen before any normal hypercall (which is why we check this before - * checking for a normal hcall). */ + * checking for a normal hcall). + */ static void do_async_hcalls(struct lg_cpu *cpu) { unsigned int i; @@ -133,22 +148,28 @@ static void do_async_hcalls(struct lg_cpu *cpu) /* We process "struct lguest_data"s hcalls[] ring once. */ for (i = 0; i < ARRAY_SIZE(st); i++) { struct hcall_args args; - /* We remember where we were up to from last time. This makes + /* + * We remember where we were up to from last time. This makes * sure that the hypercalls are done in the order the Guest - * places them in the ring. */ + * places them in the ring. + */ unsigned int n = cpu->next_hcall; /* 0xFF means there's no call here (yet). */ if (st[n] == 0xFF) break; - /* OK, we have hypercall. Increment the "next_hcall" cursor, - * and wrap back to 0 if we reach the end. */ + /* + * OK, we have hypercall. Increment the "next_hcall" cursor, + * and wrap back to 0 if we reach the end. + */ if (++cpu->next_hcall == LHCALL_RING_SIZE) cpu->next_hcall = 0; - /* Copy the hypercall arguments into a local copy of - * the hcall_args struct. */ + /* + * Copy the hypercall arguments into a local copy of the + * hcall_args struct. + */ if (copy_from_user(&args, &cpu->lg->lguest_data->hcalls[n], sizeof(struct hcall_args))) { kill_guest(cpu, "Fetching async hypercalls"); @@ -164,19 +185,25 @@ static void do_async_hcalls(struct lg_cpu *cpu) break; } - /* Stop doing hypercalls if they want to notify the Launcher: - * it needs to service this first. */ + /* + * Stop doing hypercalls if they want to notify the Launcher: + * it needs to service this first. + */ if (cpu->pending_notify) break; } } -/* Last of all, we look at what happens first of all. The very first time the - * Guest makes a hypercall, we end up here to set things up: */ +/* + * Last of all, we look at what happens first of all. The very first time the + * Guest makes a hypercall, we end up here to set things up: + */ static void initialize(struct lg_cpu *cpu) { - /* You can't do anything until you're initialized. The Guest knows the - * rules, so we're unforgiving here. */ + /* + * You can't do anything until you're initialized. The Guest knows the + * rules, so we're unforgiving here. + */ if (cpu->hcall->arg0 != LHCALL_LGUEST_INIT) { kill_guest(cpu, "hypercall %li before INIT", cpu->hcall->arg0); return; @@ -185,32 +212,44 @@ static void initialize(struct lg_cpu *cpu) if (lguest_arch_init_hypercalls(cpu)) kill_guest(cpu, "bad guest page %p", cpu->lg->lguest_data); - /* The Guest tells us where we're not to deliver interrupts by putting - * the range of addresses into "struct lguest_data". */ + /* + * The Guest tells us where we're not to deliver interrupts by putting + * the range of addresses into "struct lguest_data". + */ if (get_user(cpu->lg->noirq_start, &cpu->lg->lguest_data->noirq_start) || get_user(cpu->lg->noirq_end, &cpu->lg->lguest_data->noirq_end)) kill_guest(cpu, "bad guest page %p", cpu->lg->lguest_data); - /* We write the current time into the Guest's data page once so it can - * set its clock. */ + /* + * We write the current time into the Guest's data page once so it can + * set its clock. + */ write_timestamp(cpu); /* page_tables.c will also do some setup. */ page_table_guest_data_init(cpu); - /* This is the one case where the above accesses might have been the + /* + * This is the one case where the above accesses might have been the * first write to a Guest page. This may have caused a copy-on-write * fault, but the old page might be (read-only) in the Guest - * pagetable. */ + * pagetable. + */ guest_pagetable_clear_all(cpu); } /*:*/ -/*M:013 If a Guest reads from a page (so creates a mapping) that it has never +/*M:013 + * If a Guest reads from a page (so creates a mapping) that it has never * written to, and then the Launcher writes to it (ie. the output of a virtual * device), the Guest will still see the old page. In practice, this never * happens: why would the Guest read a page which it has never written to? But - * a similar scenario might one day bite us, so it's worth mentioning. :*/ + * a similar scenario might one day bite us, so it's worth mentioning. + * + * Note that if we used a shared anonymous mapping in the Launcher instead of + * mapping /dev/zero private, we wouldn't worry about cop-on-write. And we + * need that to switch the Launcher to processes (away from threads) anyway. +:*/ /*H:100 * Hypercalls @@ -229,17 +268,22 @@ void do_hypercalls(struct lg_cpu *cpu) return; } - /* The Guest has initialized. + /* + * The Guest has initialized. * - * Look in the hypercall ring for the async hypercalls: */ + * Look in the hypercall ring for the async hypercalls: + */ do_async_hcalls(cpu); - /* If we stopped reading the hypercall ring because the Guest did a + /* + * If we stopped reading the hypercall ring because the Guest did a * NOTIFY to the Launcher, we want to return now. Otherwise we do - * the hypercall. */ + * the hypercall. + */ if (!cpu->pending_notify) { do_hcall(cpu, cpu->hcall); - /* Tricky point: we reset the hcall pointer to mark the + /* + * Tricky point: we reset the hcall pointer to mark the * hypercall as "done". We use the hcall pointer rather than * the trap number to indicate a hypercall is pending. * Normally it doesn't matter: the Guest will run again and @@ -248,13 +292,16 @@ void do_hypercalls(struct lg_cpu *cpu) * However, if we are signalled or the Guest sends I/O to the * Launcher, the run_guest() loop will exit without running the * Guest. When it comes back it would try to re-run the - * hypercall. Finding that bug sucked. */ + * hypercall. Finding that bug sucked. + */ cpu->hcall = NULL; } } -/* This routine supplies the Guest with time: it's used for wallclock time at - * initial boot and as a rough time source if the TSC isn't available. */ +/* + * This routine supplies the Guest with time: it's used for wallclock time at + * initial boot and as a rough time source if the TSC isn't available. + */ void write_timestamp(struct lg_cpu *cpu) { struct timespec now; diff --git a/drivers/lguest/interrupts_and_traps.c b/drivers/lguest/interrupts_and_traps.c index 0e9067b0d50..18648180db0 100644 --- a/drivers/lguest/interrupts_and_traps.c +++ b/drivers/lguest/interrupts_and_traps.c @@ -1,4 +1,5 @@ -/*P:800 Interrupts (traps) are complicated enough to earn their own file. +/*P:800 + * Interrupts (traps) are complicated enough to earn their own file. * There are three classes of interrupts: * * 1) Real hardware interrupts which occur while we're running the Guest, @@ -10,7 +11,8 @@ * just like real hardware would deliver them. Traps from the Guest can be set * up to go directly back into the Guest, but sometimes the Host wants to see * them first, so we also have a way of "reflecting" them into the Guest as if - * they had been delivered to it directly. :*/ + * they had been delivered to it directly. +:*/ #include <linux/uaccess.h> #include <linux/interrupt.h> #include <linux/module.h> @@ -26,8 +28,10 @@ static unsigned long idt_address(u32 lo, u32 hi) return (lo & 0x0000FFFF) | (hi & 0xFFFF0000); } -/* The "type" of the interrupt handler is a 4 bit field: we only support a - * couple of types. */ +/* + * The "type" of the interrupt handler is a 4 bit field: we only support a + * couple of types. + */ static int idt_type(u32 lo, u32 hi) { return (hi >> 8) & 0xF; @@ -39,8 +43,10 @@ static bool idt_present(u32 lo, u32 hi) return (hi & 0x8000); } -/* We need a helper to "push" a value onto the Guest's stack, since that's a - * big part of what delivering an interrupt does. */ +/* + * We need a helper to "push" a value onto the Guest's stack, since that's a + * big part of what delivering an interrupt does. + */ static void push_guest_stack(struct lg_cpu *cpu, unsigned long *gstack, u32 val) { /* Stack grows upwards: move stack then write value. */ @@ -48,7 +54,8 @@ static void push_guest_stack(struct lg_cpu *cpu, unsigned long *gstack, u32 val) lgwrite(cpu, *gstack, u32, val); } -/*H:210 The set_guest_interrupt() routine actually delivers the interrupt or +/*H:210 + * The set_guest_interrupt() routine actually delivers the interrupt or * trap. The mechanics of delivering traps and interrupts to the Guest are the * same, except some traps have an "error code" which gets pushed onto the * stack as well: the caller tells us if this is one. @@ -59,7 +66,8 @@ static void push_guest_stack(struct lg_cpu *cpu, unsigned long *gstack, u32 val) * * We set up the stack just like the CPU does for a real interrupt, so it's * identical for the Guest (and the standard "iret" instruction will undo - * it). */ + * it). + */ static void set_guest_interrupt(struct lg_cpu *cpu, u32 lo, u32 hi, bool has_err) { @@ -67,20 +75,26 @@ static void set_guest_interrupt(struct lg_cpu *cpu, u32 lo, u32 hi, u32 eflags, ss, irq_enable; unsigned long virtstack; - /* There are two cases for interrupts: one where the Guest is already + /* + * There are two cases for interrupts: one where the Guest is already * in the kernel, and a more complex one where the Guest is in - * userspace. We check the privilege level to find out. */ + * userspace. We check the privilege level to find out. + */ if ((cpu->regs->ss&0x3) != GUEST_PL) { - /* The Guest told us their kernel stack with the SET_STACK - * hypercall: both the virtual address and the segment */ + /* + * The Guest told us their kernel stack with the SET_STACK + * hypercall: both the virtual address and the segment. + */ virtstack = cpu->esp1; ss = cpu->ss1; origstack = gstack = guest_pa(cpu, virtstack); - /* We push the old stack segment and pointer onto the new + /* + * We push the old stack segment and pointer onto the new * stack: when the Guest does an "iret" back from the interrupt * handler the CPU will notice they're dropping privilege - * levels and expect these here. */ + * levels and expect these here. + */ push_guest_stack(cpu, &gstack, cpu->regs->ss); push_guest_stack(cpu, &gstack, cpu->regs->esp); } else { @@ -91,18 +105,22 @@ static void set_guest_interrupt(struct lg_cpu *cpu, u32 lo, u32 hi, origstack = gstack = guest_pa(cpu, virtstack); } - /* Remember that we never let the Guest actually disable interrupts, so + /* + * Remember that we never let the Guest actually disable interrupts, so * the "Interrupt Flag" bit is always set. We copy that bit from the * Guest's "irq_enabled" field into the eflags word: we saw the Guest - * copy it back in "lguest_iret". */ + * copy it back in "lguest_iret". + */ eflags = cpu->regs->eflags; if (get_user(irq_enable, &cpu->lg->lguest_data->irq_enabled) == 0 && !(irq_enable & X86_EFLAGS_IF)) eflags &= ~X86_EFLAGS_IF; - /* An interrupt is expected to push three things on the stack: the old + /* + * An interrupt is expected to push three things on the stack: the old * "eflags" word, the old code segment, and the old instruction - * pointer. */ + * pointer. + */ push_guest_stack(cpu, &gstack, eflags); push_guest_stack(cpu, &gstack, cpu->regs->cs); push_guest_stack(cpu, &gstack, cpu->regs->eip); @@ -111,15 +129,19 @@ static void set_guest_interrupt(struct lg_cpu *cpu, u32 lo, u32 hi, if (has_err) push_guest_stack(cpu, &gstack, cpu->regs->errcode); - /* Now we've pushed all the old state, we change the stack, the code - * segment and the address to execute. */ + /* + * Now we've pushed all the old state, we change the stack, the code + * segment and the address to execute. + */ cpu->regs->ss = ss; cpu->regs->esp = virtstack + (gstack - origstack); cpu->regs->cs = (__KERNEL_CS|GUEST_PL); cpu->regs->eip = idt_address(lo, hi); - /* There are two kinds of interrupt handlers: 0xE is an "interrupt - * gate" which expects interrupts to be disabled on entry. */ + /* + * There are two kinds of interrupt handlers: 0xE is an "interrupt + * gate" which expects interrupts to be disabled on entry. + */ if (idt_type(lo, hi) == 0xE) if (put_user(0, &cpu->lg->lguest_data->irq_enabled)) kill_guest(cpu, "Disabling interrupts"); @@ -130,7 +152,8 @@ static void set_guest_interrupt(struct lg_cpu *cpu, u32 lo, u32 hi, * * interrupt_pending() returns the first pending interrupt which isn't blocked * by the Guest. It is called before every entry to the Guest, and just before - * we go to sleep when the Guest has halted itself. */ + * we go to sleep when the Guest has halted itself. + */ unsigned int interrupt_pending(struct lg_cpu *cpu, bool *more) { unsigned int irq; @@ -140,8 +163,10 @@ unsigned int interrupt_pending(struct lg_cpu *cpu, bool *more) if (!cpu->lg->lguest_data) return LGUEST_IRQS; - /* Take our "irqs_pending" array and remove any interrupts the Guest - * wants blocked: the result ends up in "blk". */ + /* + * Take our "irqs_pending" array and remove any interrupts the Guest + * wants blocked: the result ends up in "blk". + */ if (copy_from_user(&blk, cpu->lg->lguest_data->blocked_interrupts, sizeof(blk))) return LGUEST_IRQS; @@ -154,16 +179,20 @@ unsigned int interrupt_pending(struct lg_cpu *cpu, bool *more) return irq; } -/* This actually diverts the Guest to running an interrupt handler, once an - * interrupt has been identified by interrupt_pending(). */ +/* + * This actually diverts the Guest to running an interrupt handler, once an + * interrupt has been identified by interrupt_pending(). + */ void try_deliver_interrupt(struct lg_cpu *cpu, unsigned int irq, bool more) { struct desc_struct *idt; BUG_ON(irq >= LGUEST_IRQS); - /* They may be in the middle of an iret, where they asked us never to - * deliver interrupts. */ + /* + * They may be in the middle of an iret, where they asked us never to + * deliver interrupts. + */ if (cpu->regs->eip >= cpu->lg->noirq_start && (cpu->regs->eip < cpu->lg->noirq_end)) return; @@ -187,29 +216,37 @@ void try_deliver_interrupt(struct lg_cpu *cpu, unsigned int irq, bool more) } } - /* Look at the IDT entry the Guest gave us for this interrupt. The + /* + * Look at the IDT entry the Guest gave us for this interrupt. The * first 32 (FIRST_EXTERNAL_VECTOR) entries are for traps, so we skip - * over them. */ + * over them. + */ idt = &cpu->arch.idt[FIRST_EXTERNAL_VECTOR+irq]; /* If they don't have a handler (yet?), we just ignore it */ if (idt_present(idt->a, idt->b)) { /* OK, mark it no longer pending and deliver it. */ clear_bit(irq, cpu->irqs_pending); - /* set_guest_interrupt() takes the interrupt descriptor and a + /* + * set_guest_interrupt() takes the interrupt descriptor and a * flag to say whether this interrupt pushes an error code onto - * the stack as well: virtual interrupts never do. */ + * the stack as well: virtual interrupts never do. + */ set_guest_interrupt(cpu, idt->a, idt->b, false); } - /* Every time we deliver an interrupt, we update the timestamp in the + /* + * Every time we deliver an interrupt, we update the timestamp in the * Guest's lguest_data struct. It would be better for the Guest if we * did this more often, but it can actually be quite slow: doing it * here is a compromise which means at least it gets updated every - * timer interrupt. */ + * timer interrupt. + */ write_timestamp(cpu); - /* If there are no other interrupts we want to deliver, clear - * the pending flag. */ + /* + * If there are no other interrupts we want to deliver, clear + * the pending flag. + */ if (!more) put_user(0, &cpu->lg->lguest_data->irq_pending); } @@ -217,24 +254,29 @@ void try_deliver_interrupt(struct lg_cpu *cpu, unsigned int irq, bool more) /* And this is the routine when we want to set an interrupt for the Guest. */ void set_interrupt(struct lg_cpu *cpu, unsigned int irq) { - /* Next time the Guest runs, the core code will see if it can deliver - * this interrupt. */ + /* + * Next time the Guest runs, the core code will see if it can deliver + * this interrupt. + */ set_bit(irq, cpu->irqs_pending); - /* Make sure it sees it; it might be asleep (eg. halted), or - * running the Guest right now, in which case kick_process() - * will knock it out. */ + /* + * Make sure it sees it; it might be asleep (eg. halted), or running + * the Guest right now, in which case kick_process() will knock it out. + */ if (!wake_up_process(cpu->tsk)) kick_process(cpu->tsk); } /*:*/ -/* Linux uses trap 128 for system calls. Plan9 uses 64, and Ron Minnich sent +/* + * Linux uses trap 128 for system calls. Plan9 uses 64, and Ron Minnich sent * me a patch, so we support that too. It'd be a big step for lguest if half * the Plan 9 user base were to start using it. * * Actually now I think of it, it's possible that Ron *is* half the Plan 9 - * userbase. Oh well. */ + * userbase. Oh well. + */ static bool could_be_syscall(unsigned int num) { /* Normal Linux SYSCALL_VECTOR or reserved vector? */ @@ -274,9 +316,11 @@ void free_interrupts(void) clear_bit(syscall_vector, used_vectors); } -/*H:220 Now we've got the routines to deliver interrupts, delivering traps like +/*H:220 + * Now we've got the routines to deliver interrupts, delivering traps like * page fault is easy. The only trick is that Intel decided that some traps - * should have error codes: */ + * should have error codes: + */ static bool has_err(unsigned int trap) { return (trap == 8 || (trap >= 10 && trap <= 14) || trap == 17); @@ -285,13 +329,17 @@ static bool has_err(unsigned int trap) /* deliver_trap() returns true if it could deliver the trap. */ bool deliver_trap(struct lg_cpu *cpu, unsigned int num) { - /* Trap numbers are always 8 bit, but we set an impossible trap number - * for traps inside the Switcher, so check that here. */ + /* + * Trap numbers are always 8 bit, but we set an impossible trap number + * for traps inside the Switcher, so check that here. + */ if (num >= ARRAY_SIZE(cpu->arch.idt)) return false; - /* Early on the Guest hasn't set the IDT entries (or maybe it put a - * bogus one in): if we fail here, the Guest will be killed. */ + /* + * Early on the Guest hasn't set the IDT entries (or maybe it put a + * bogus one in): if we fail here, the Guest will be killed. + */ if (!idt_present(cpu->arch.idt[num].a, cpu->arch.idt[num].b)) return false; set_guest_interrupt(cpu, cpu->arch.idt[num].a, @@ -299,7 +347,8 @@ bool deliver_trap(struct lg_cpu *cpu, unsigned int num) return true; } -/*H:250 Here's the hard part: returning to the Host every time a trap happens +/*H:250 + * Here's the hard part: returning to the Host every time a trap happens * and then calling deliver_trap() and re-entering the Guest is slow. * Particularly because Guest userspace system calls are traps (usually trap * 128). @@ -311,69 +360,87 @@ bool deliver_trap(struct lg_cpu *cpu, unsigned int num) * the other hypervisors would beat it up at lunchtime. * * This routine indicates if a particular trap number could be delivered - * directly. */ + * directly. + */ static bool direct_trap(unsigned int num) { - /* Hardware interrupts don't go to the Guest at all (except system - * call). */ + /* + * Hardware interrupts don't go to the Guest at all (except system + * call). + */ if (num >= FIRST_EXTERNAL_VECTOR && !could_be_syscall(num)) return false; - /* The Host needs to see page faults (for shadow paging and to save the + /* + * The Host needs to see page faults (for shadow paging and to save the * fault address), general protection faults (in/out emulation) and * device not available (TS handling), invalid opcode fault (kvm hcall), - * and of course, the hypercall trap. */ + * and of course, the hypercall trap. + */ return num != 14 && num != 13 && num != 7 && num != 6 && num != LGUEST_TRAP_ENTRY; } /*:*/ -/*M:005 The Guest has the ability to turn its interrupt gates into trap gates, +/*M:005 + * The Guest has the ability to turn its interrupt gates into trap gates, * if it is careful. The Host will let trap gates can go directly to the * Guest, but the Guest needs the interrupts atomically disabled for an * interrupt gate. It can do this by pointing the trap gate at instructions - * within noirq_start and noirq_end, where it can safely disable interrupts. */ + * within noirq_start and noirq_end, where it can safely disable interrupts. + */ -/*M:006 The Guests do not use the sysenter (fast system call) instruction, +/*M:006 + * The Guests do not use the sysenter (fast system call) instruction, * because it's hardcoded to enter privilege level 0 and so can't go direct. * It's about twice as fast as the older "int 0x80" system call, so it might * still be worthwhile to handle it in the Switcher and lcall down to the * Guest. The sysenter semantics are hairy tho: search for that keyword in - * entry.S :*/ + * entry.S +:*/ -/*H:260 When we make traps go directly into the Guest, we need to make sure +/*H:260 + * When we make traps go directly into the Guest, we need to make sure * the kernel stack is valid (ie. mapped in the page tables). Otherwise, the * CPU trying to deliver the trap will fault while trying to push the interrupt * words on the stack: this is called a double fault, and it forces us to kill * the Guest. * - * Which is deeply unfair, because (literally!) it wasn't the Guests' fault. */ + * Which is deeply unfair, because (literally!) it wasn't the Guests' fault. + */ void pin_stack_pages(struct lg_cpu *cpu) { unsigned int i; - /* Depending on the CONFIG_4KSTACKS option, the Guest can have one or - * two pages of stack space. */ + /* + * Depending on the CONFIG_4KSTACKS option, the Guest can have one or + * two pages of stack space. + */ for (i = 0; i < cpu->lg->stack_pages; i++) - /* The stack grows *upwards*, so the address we're given is the + /* + * The stack grows *upwards*, so the address we're given is the * start of the page after the kernel stack. Subtract one to * get back onto the first stack page, and keep subtracting to - * get to the rest of the stack pages. */ + * get to the rest of the stack pages. + */ pin_page(cpu, cpu->esp1 - 1 - i * PAGE_SIZE); } -/* Direct traps also mean that we need to know whenever the Guest wants to use +/* + * Direct traps also mean that we need to know whenever the Guest wants to use * a different kernel stack, so we can change the IDT entries to use that * stack. The IDT entries expect a virtual address, so unlike most addresses * the Guest gives us, the "esp" (stack pointer) value here is virtual, not * physical. * * In Linux each process has its own kernel stack, so this happens a lot: we - * change stacks on each context switch. */ + * change stacks on each context switch. + */ void guest_set_stack(struct lg_cpu *cpu, u32 seg, u32 esp, unsigned int pages) { - /* You are not allowed have a stack segment with privilege level 0: bad - * Guest! */ + /* + * You're not allowed a stack segment with privilege level 0: bad Guest! + */ if ((seg & 0x3) != GUEST_PL) kill_guest(cpu, "bad stack segment %i", seg); /* We only expect one or two stack pages. */ @@ -387,11 +454,15 @@ void guest_set_stack(struct lg_cpu *cpu, u32 seg, u32 esp, unsigned int pages) pin_stack_pages(cpu); } -/* All this reference to mapping stacks leads us neatly into the other complex - * part of the Host: page table handling. */ +/* + * All this reference to mapping stacks leads us neatly into the other complex + * part of the Host: page table handling. + */ -/*H:235 This is the routine which actually checks the Guest's IDT entry and - * transfers it into the entry in "struct lguest": */ +/*H:235 + * This is the routine which actually checks the Guest's IDT entry and + * transfers it into the entry in "struct lguest": + */ static void set_trap(struct lg_cpu *cpu, struct desc_struct *trap, unsigned int num, u32 lo, u32 hi) { @@ -407,30 +478,38 @@ static void set_trap(struct lg_cpu *cpu, struct desc_struct *trap, if (type != 0xE && type != 0xF) kill_guest(cpu, "bad IDT type %i", type); - /* We only copy the handler address, present bit, privilege level and + /* + * We only copy the handler address, present bit, privilege level and * type. The privilege level controls where the trap can be triggered * manually with an "int" instruction. This is usually GUEST_PL, - * except for system calls which userspace can use. */ + * except for system calls which userspace can use. + */ trap->a = ((__KERNEL_CS|GUEST_PL)<<16) | (lo&0x0000FFFF); trap->b = (hi&0xFFFFEF00); } -/*H:230 While we're here, dealing with delivering traps and interrupts to the +/*H:230 + * While we're here, dealing with delivering traps and interrupts to the * Guest, we might as well complete the picture: how the Guest tells us where * it wants them to go. This would be simple, except making traps fast * requires some tricks. * * We saw the Guest setting Interrupt Descriptor Table (IDT) entries with the - * LHCALL_LOAD_IDT_ENTRY hypercall before: that comes here. */ + * LHCALL_LOAD_IDT_ENTRY hypercall before: that comes here. + */ void load_guest_idt_entry(struct lg_cpu *cpu, unsigned int num, u32 lo, u32 hi) { - /* Guest never handles: NMI, doublefault, spurious interrupt or - * hypercall. We ignore when it tries to set them. */ + /* + * Guest never handles: NMI, doublefault, spurious interrupt or + * hypercall. We ignore when it tries to set them. + */ if (num == 2 || num == 8 || num == 15 || num == LGUEST_TRAP_ENTRY) return; - /* Mark the IDT as changed: next time the Guest runs we'll know we have - * to copy this again. */ + /* + * Mark the IDT as changed: next time the Guest runs we'll know we have + * to copy this again. + */ cpu->changed |= CHANGED_IDT; /* Check that the Guest doesn't try to step outside the bounds. */ @@ -440,9 +519,11 @@ void load_guest_idt_entry(struct lg_cpu *cpu, unsigned int num, u32 lo, u32 hi) set_trap(cpu, &cpu->arch.idt[num], num, lo, hi); } -/* The default entry for each interrupt points into the Switcher routines which +/* + * The default entry for each interrupt points into the Switcher routines which * simply return to the Host. The run_guest() loop will then call - * deliver_trap() to bounce it back into the Guest. */ + * deliver_trap() to bounce it back into the Guest. + */ static void default_idt_entry(struct desc_struct *idt, int trap, const unsigned long handler, @@ -451,13 +532,17 @@ static void default_idt_entry(struct desc_struct *idt, /* A present interrupt gate. */ u32 flags = 0x8e00; - /* Set the privilege level on the entry for the hypercall: this allows - * the Guest to use the "int" instruction to trigger it. */ + /* + * Set the privilege level on the entry for the hypercall: this allows + * the Guest to use the "int" instruction to trigger it. + */ if (trap == LGUEST_TRAP_ENTRY) flags |= (GUEST_PL << 13); else if (base) - /* Copy priv. level from what Guest asked for. This allows - * debug (int 3) traps from Guest userspace, for example. */ + /* + * Copy privilege level from what Guest asked for. This allows + * debug (int 3) traps from Guest userspace, for example. + */ flags |= (base->b & 0x6000); /* Now pack it into the IDT entry in its weird format. */ @@ -475,16 +560,20 @@ void setup_default_idt_entries(struct lguest_ro_state *state, default_idt_entry(&state->guest_idt[i], i, def[i], NULL); } -/*H:240 We don't use the IDT entries in the "struct lguest" directly, instead +/*H:240 + * We don't use the IDT entries in the "struct lguest" directly, instead * we copy them into the IDT which we've set up for Guests on this CPU, just - * before we run the Guest. This routine does that copy. */ + * before we run the Guest. This routine does that copy. + */ void copy_traps(const struct lg_cpu *cpu, struct desc_struct *idt, const unsigned long *def) { unsigned int i; - /* We can simply copy the direct traps, otherwise we use the default - * ones in the Switcher: they will return to the Host. */ + /* + * We can simply copy the direct traps, otherwise we use the default + * ones in the Switcher: they will return to the Host. + */ for (i = 0; i < ARRAY_SIZE(cpu->arch.idt); i++) { const struct desc_struct *gidt = &cpu->arch.idt[i]; @@ -492,14 +581,16 @@ void copy_traps(const struct lg_cpu *cpu, struct desc_struct *idt, if (!direct_trap(i)) continue; - /* Only trap gates (type 15) can go direct to the Guest. + /* + * Only trap gates (type 15) can go direct to the Guest. * Interrupt gates (type 14) disable interrupts as they are * entered, which we never let the Guest do. Not present * entries (type 0x0) also can't go direct, of course. * * If it can't go direct, we still need to copy the priv. level: * they might want to give userspace access to a software - * interrupt. */ + * interrupt. + */ if (idt_type(gidt->a, gidt->b) == 0xF) idt[i] = *gidt; else @@ -518,7 +609,8 @@ void copy_traps(const struct lg_cpu *cpu, struct desc_struct *idt, * the next timer interrupt (in nanoseconds). We use the high-resolution timer * infrastructure to set a callback at that time. * - * 0 means "turn off the clock". */ + * 0 means "turn off the clock". + */ void guest_set_clockevent(struct lg_cpu *cpu, unsigned long delta) { ktime_t expires; @@ -529,9 +621,11 @@ void guest_set_clockevent(struct lg_cpu *cpu, unsigned long delta) return; } - /* We use wallclock time here, so the Guest might not be running for + /* + * We use wallclock time here, so the Guest might not be running for * all the time between now and the timer interrupt it asked for. This - * is almost always the right thing to do. */ + * is almost always the right thing to do. + */ expires = ktime_add_ns(ktime_get_real(), delta); hrtimer_start(&cpu->hrt, expires, HRTIMER_MODE_ABS); } diff --git a/drivers/lguest/lg.h b/drivers/lguest/lg.h index 01c59192379..bc28745d05a 100644 --- a/drivers/lguest/lg.h +++ b/drivers/lguest/lg.h @@ -16,15 +16,13 @@ void free_pagetables(void); int init_pagetables(struct page **switcher_page, unsigned int pages); -struct pgdir -{ +struct pgdir { unsigned long gpgdir; pgd_t *pgdir; }; /* We have two pages shared with guests, per cpu. */ -struct lguest_pages -{ +struct lguest_pages { /* This is the stack page mapped rw in guest */ char spare[PAGE_SIZE - sizeof(struct lguest_regs)]; struct lguest_regs regs; @@ -54,13 +52,13 @@ struct lg_cpu { unsigned long pending_notify; /* pfn from LHCALL_NOTIFY */ - /* At end of a page shared mapped over lguest_pages in guest. */ + /* At end of a page shared mapped over lguest_pages in guest. */ unsigned long regs_page; struct lguest_regs *regs; struct lguest_pages *last_pages; - int cpu_pgd; /* which pgd this cpu is currently using */ + int cpu_pgd; /* Which pgd this cpu is currently using */ /* If a hypercall was asked for, this points to the arguments. */ struct hcall_args *hcall; @@ -89,15 +87,17 @@ struct lg_eventfd_map { }; /* The private info the thread maintains about the guest. */ -struct lguest -{ +struct lguest { struct lguest_data __user *lguest_data; struct lg_cpu cpus[NR_CPUS]; unsigned int nr_cpus; u32 pfn_limit; - /* This provides the offset to the base of guest-physical - * memory in the Launcher. */ + + /* + * This provides the offset to the base of guest-physical memory in the + * Launcher. + */ void __user *mem_base; unsigned long kernel_address; @@ -122,11 +122,13 @@ bool lguest_address_ok(const struct lguest *lg, void __lgread(struct lg_cpu *, void *, unsigned long, unsigned); void __lgwrite(struct lg_cpu *, unsigned long, const void *, unsigned); -/*H:035 Using memory-copy operations like that is usually inconvient, so we +/*H:035 + * Using memory-copy operations like that is usually inconvient, so we * have the following helper macros which read and write a specific type (often * an unsigned long). * - * This reads into a variable of the given type then returns that. */ + * This reads into a variable of the given type then returns that. + */ #define lgread(cpu, addr, type) \ ({ type _v; __lgread((cpu), &_v, (addr), sizeof(_v)); _v; }) @@ -140,9 +142,11 @@ void __lgwrite(struct lg_cpu *, unsigned long, const void *, unsigned); int run_guest(struct lg_cpu *cpu, unsigned long __user *user); -/* Helper macros to obtain the first 12 or the last 20 bits, this is only the +/* + * Helper macros to obtain the first 12 or the last 20 bits, this is only the * first step in the migration to the kernel types. pte_pfn is already defined - * in the kernel. */ + * in the kernel. + */ #define pgd_flags(x) (pgd_val(x) & ~PAGE_MASK) #define pgd_pfn(x) (pgd_val(x) >> PAGE_SHIFT) #define pmd_flags(x) (pmd_val(x) & ~PAGE_MASK) diff --git a/drivers/lguest/lguest_device.c b/drivers/lguest/lguest_device.c index e082cdac88b..b6200bc39b5 100644 --- a/drivers/lguest/lguest_device.c +++ b/drivers/lguest/lguest_device.c @@ -1,10 +1,12 @@ -/*P:050 Lguest guests use a very simple method to describe devices. It's a +/*P:050 + * Lguest guests use a very simple method to describe devices. It's a * series of device descriptors contained just above the top of normal Guest * memory. * * We use the standard "virtio" device infrastructure, which provides us with a * console, a network and a block driver. Each one expects some configuration - * information and a "virtqueue" or two to send and receive data. :*/ + * information and a "virtqueue" or two to send and receive data. +:*/ #include <linux/init.h> #include <linux/bootmem.h> #include <linux/lguest_launcher.h> @@ -20,8 +22,10 @@ /* The pointer to our (page) of device descriptions. */ static void *lguest_devices; -/* For Guests, device memory can be used as normal memory, so we cast away the - * __iomem to quieten sparse. */ +/* + * For Guests, device memory can be used as normal memory, so we cast away the + * __iomem to quieten sparse. + */ static inline void *lguest_map(unsigned long phys_addr, unsigned long pages) { return (__force void *)ioremap_cache(phys_addr, PAGE_SIZE*pages); @@ -32,8 +36,10 @@ static inline void lguest_unmap(void *addr) iounmap((__force void __iomem *)addr); } -/*D:100 Each lguest device is just a virtio device plus a pointer to its entry - * in the lguest_devices page. */ +/*D:100 + * Each lguest device is just a virtio device plus a pointer to its entry + * in the lguest_devices page. + */ struct lguest_device { struct virtio_device vdev; @@ -41,9 +47,11 @@ struct lguest_device { struct lguest_device_desc *desc; }; -/* Since the virtio infrastructure hands us a pointer to the virtio_device all +/* + * Since the virtio infrastructure hands us a pointer to the virtio_device all * the time, it helps to have a curt macro to get a pointer to the struct - * lguest_device it's enclosed in. */ + * lguest_device it's enclosed in. + */ #define to_lgdev(vd) container_of(vd, struct lguest_device, vdev) /*D:130 @@ -55,7 +63,8 @@ struct lguest_device { * the driver will look at them during setup. * * A convenient routine to return the device's virtqueue config array: - * immediately after the descriptor. */ + * immediately after the descriptor. + */ static struct lguest_vqconfig *lg_vq(const struct lguest_device_desc *desc) { return (void *)(desc + 1); @@ -98,10 +107,12 @@ static u32 lg_get_features(struct virtio_device *vdev) return features; } -/* The virtio core takes the features the Host offers, and copies the - * ones supported by the driver into the vdev->features array. Once - * that's all sorted out, this routine is called so we can tell the - * Host which features we understand and accept. */ +/* + * The virtio core takes the features the Host offers, and copies the ones + * supported by the driver into the vdev->features array. Once that's all + * sorted out, this routine is called so we can tell the Host which features we + * understand and accept. + */ static void lg_finalize_features(struct virtio_device *vdev) { unsigned int i, bits; @@ -112,10 +123,11 @@ static void lg_finalize_features(struct virtio_device *vdev) /* Give virtio_ring a chance to accept features. */ vring_transport_features(vdev); - /* The vdev->feature array is a Linux bitmask: this isn't the - * same as a the simple array of bits used by lguest devices - * for features. So we do this slow, manual conversion which is - * completely general. */ + /* + * The vdev->feature array is a Linux bitmask: this isn't the same as a + * the simple array of bits used by lguest devices for features. So we + * do this slow, manual conversion which is completely general. + */ memset(out_features, 0, desc->feature_len); bits = min_t(unsigned, desc->feature_len, sizeof(vdev->features)) * 8; for (i = 0; i < bits; i++) { @@ -146,15 +158,19 @@ static void lg_set(struct virtio_device *vdev, unsigned int offset, memcpy(lg_config(desc) + offset, buf, len); } -/* The operations to get and set the status word just access the status field - * of the device descriptor. */ +/* + * The operations to get and set the status word just access the status field + * of the device descriptor. + */ static u8 lg_get_status(struct virtio_device *vdev) { return to_lgdev(vdev)->desc->status; } -/* To notify on status updates, we (ab)use the NOTIFY hypercall, with the - * descriptor address of the device. A zero status means "reset". */ +/* + * To notify on status updates, we (ab)use the NOTIFY hypercall, with the + * descriptor address of the device. A zero status means "reset". + */ static void set_status(struct virtio_device *vdev, u8 status) { unsigned long offset = (void *)to_lgdev(vdev)->desc - lguest_devices; @@ -191,8 +207,7 @@ static void lg_reset(struct virtio_device *vdev) */ /*D:140 This is the information we remember about each virtqueue. */ -struct lguest_vq_info -{ +struct lguest_vq_info { /* A copy of the information contained in the device config. */ struct lguest_vqconfig config; @@ -200,13 +215,17 @@ struct lguest_vq_info void *pages; }; -/* When the virtio_ring code wants to prod the Host, it calls us here and we +/* + * When the virtio_ring code wants to prod the Host, it calls us here and we * make a hypercall. We hand the physical address of the virtqueue so the Host - * knows which virtqueue we're talking about. */ + * knows which virtqueue we're talking about. + */ static void lg_notify(struct virtqueue *vq) { - /* We store our virtqueue information in the "priv" pointer of the - * virtqueue structure. */ + /* + * We store our virtqueue information in the "priv" pointer of the + * virtqueue structure. + */ struct lguest_vq_info *lvq = vq->priv; kvm_hypercall1(LHCALL_NOTIFY, lvq->config.pfn << PAGE_SHIFT); @@ -215,7 +234,8 @@ static void lg_notify(struct virtqueue *vq) /* An extern declaration inside a C file is bad form. Don't do it. */ extern void lguest_setup_irq(unsigned int irq); -/* This routine finds the first virtqueue described in the configuration of +/* + * This routine finds the Nth virtqueue described in the configuration of * this device and sets it up. * * This is kind of an ugly duckling. It'd be nicer to have a standard @@ -223,9 +243,7 @@ extern void lguest_setup_irq(unsigned int irq); * everyone wants to do it differently. The KVM coders want the Guest to * allocate its own pages and tell the Host where they are, but for lguest it's * simpler for the Host to simply tell us where the pages are. - * - * So we provide drivers with a "find the Nth virtqueue and set it up" - * function. */ + */ static struct virtqueue *lg_find_vq(struct virtio_device *vdev, unsigned index, void (*callback)(struct virtqueue *vq), @@ -244,9 +262,11 @@ static struct virtqueue *lg_find_vq(struct virtio_device *vdev, if (!lvq) return ERR_PTR(-ENOMEM); - /* Make a copy of the "struct lguest_vqconfig" entry, which sits after + /* + * Make a copy of the "struct lguest_vqconfig" entry, which sits after * the descriptor. We need a copy because the config space might not - * be aligned correctly. */ + * be aligned correctly. + */ memcpy(&lvq->config, lg_vq(ldev->desc)+index, sizeof(lvq->config)); printk("Mapping virtqueue %i addr %lx\n", index, @@ -261,8 +281,10 @@ static struct virtqueue *lg_find_vq(struct virtio_device *vdev, goto free_lvq; } - /* OK, tell virtio_ring.c to set up a virtqueue now we know its size - * and we've got a pointer to its pages. */ + /* + * OK, tell virtio_ring.c to set up a virtqueue now we know its size + * and we've got a pointer to its pages. + */ vq = vring_new_virtqueue(lvq->config.num, LGUEST_VRING_ALIGN, vdev, lvq->pages, lg_notify, callback, name); if (!vq) { @@ -273,18 +295,23 @@ static struct virtqueue *lg_find_vq(struct virtio_device *vdev, /* Make sure the interrupt is allocated. */ lguest_setup_irq(lvq->config.irq); - /* Tell the interrupt for this virtqueue to go to the virtio_ring - * interrupt handler. */ - /* FIXME: We used to have a flag for the Host to tell us we could use + /* + * Tell the interrupt for this virtqueue to go to the virtio_ring + * interrupt handler. + * + * FIXME: We used to have a flag for the Host to tell us we could use * the interrupt as a source of randomness: it'd be nice to have that - * back.. */ + * back. + */ err = request_irq(lvq->config.irq, vring_interrupt, IRQF_SHARED, dev_name(&vdev->dev), vq); if (err) goto destroy_vring; - /* Last of all we hook up our 'struct lguest_vq_info" to the - * virtqueue's priv pointer. */ + /* + * Last of all we hook up our 'struct lguest_vq_info" to the + * virtqueue's priv pointer. + */ vq->priv = lvq; return vq; @@ -358,11 +385,14 @@ static struct virtio_config_ops lguest_config_ops = { .del_vqs = lg_del_vqs, }; -/* The root device for the lguest virtio devices. This makes them appear as - * /sys/devices/lguest/0,1,2 not /sys/devices/0,1,2. */ +/* + * The root device for the lguest virtio devices. This makes them appear as + * /sys/devices/lguest/0,1,2 not /sys/devices/0,1,2. + */ static struct device *lguest_root; -/*D:120 This is the core of the lguest bus: actually adding a new device. +/*D:120 + * This is the core of the lguest bus: actually adding a new device. * It's a separate function because it's neater that way, and because an * earlier version of the code supported hotplug and unplug. They were removed * early on because they were never used. @@ -371,14 +401,14 @@ static struct device *lguest_root; * * It's worth reading this carefully: we start with a pointer to the new device * descriptor in the "lguest_devices" page, and the offset into the device - * descriptor page so we can uniquely identify it if things go badly wrong. */ + * descriptor page so we can uniquely identify it if things go badly wrong. + */ static void add_lguest_device(struct lguest_device_desc *d, unsigned int offset) { struct lguest_device *ldev; - /* Start with zeroed memory; Linux's device layer seems to count on - * it. */ + /* Start with zeroed memory; Linux's device layer counts on it. */ ldev = kzalloc(sizeof(*ldev), GFP_KERNEL); if (!ldev) { printk(KERN_EMERG "Cannot allocate lguest dev %u type %u\n", @@ -388,17 +418,25 @@ static void add_lguest_device(struct lguest_device_desc *d, /* This devices' parent is the lguest/ dir. */ ldev->vdev.dev.parent = lguest_root; - /* We have a unique device index thanks to the dev_index counter. */ + /* + * The device type comes straight from the descriptor. There's also a + * device vendor field in the virtio_device struct, which we leave as + * 0. + */ ldev->vdev.id.device = d->type; - /* We have a simple set of routines for querying the device's - * configuration information and setting its status. */ + /* + * We have a simple set of routines for querying the device's + * configuration information and setting its status. + */ ldev->vdev.config = &lguest_config_ops; /* And we remember the device's descriptor for lguest_config_ops. */ ldev->desc = d; - /* register_virtio_device() sets up the generic fields for the struct + /* + * register_virtio_device() sets up the generic fields for the struct * virtio_device and calls device_register(). This makes the bus - * infrastructure look for a matching driver. */ + * infrastructure look for a matching driver. + */ if (register_virtio_device(&ldev->vdev) != 0) { printk(KERN_ERR "Failed to register lguest dev %u type %u\n", offset, d->type); @@ -406,8 +444,10 @@ static void add_lguest_device(struct lguest_device_desc *d, } } -/*D:110 scan_devices() simply iterates through the device page. The type 0 is - * reserved to mean "end of devices". */ +/*D:110 + * scan_devices() simply iterates through the device page. The type 0 is + * reserved to mean "end of devices". + */ static void scan_devices(void) { unsigned int i; @@ -426,7 +466,8 @@ static void scan_devices(void) } } -/*D:105 Fairly early in boot, lguest_devices_init() is called to set up the +/*D:105 + * Fairly early in boot, lguest_devices_init() is called to set up the * lguest device infrastructure. We check that we are a Guest by checking * pv_info.name: there are other ways of checking, but this seems most * obvious to me. @@ -437,7 +478,8 @@ static void scan_devices(void) * correct sysfs incantation). * * Finally we call scan_devices() which adds all the devices found in the - * lguest_devices page. */ + * lguest_devices page. + */ static int __init lguest_devices_init(void) { if (strcmp(pv_info.name, "lguest") != 0) @@ -456,11 +498,13 @@ static int __init lguest_devices_init(void) /* We do this after core stuff, but before the drivers. */ postcore_initcall(lguest_devices_init); -/*D:150 At this point in the journey we used to now wade through the lguest +/*D:150 + * At this point in the journey we used to now wade through the lguest * devices themselves: net, block and console. Since they're all now virtio * devices rather than lguest-specific, I've decided to ignore them. Mostly, * they're kind of boring. But this does mean you'll never experience the * thrill of reading the forbidden love scene buried deep in the block driver. * * "make Launcher" beckons, where we answer questions like "Where do Guests - * come from?", and "What do you do when someone asks for optimization?". */ + * come from?", and "What do you do when someone asks for optimization?". + */ diff --git a/drivers/lguest/lguest_user.c b/drivers/lguest/lguest_user.c index 9f9a2953b38..b4d3f7ca554 100644 --- a/drivers/lguest/lguest_user.c +++ b/drivers/lguest/lguest_user.c @@ -1,8 +1,9 @@ /*P:200 This contains all the /dev/lguest code, whereby the userspace launcher * controls and communicates with the Guest. For example, the first write will - * tell us the Guest's memory layout, pagetable, entry point and kernel address - * offset. A read will run the Guest until something happens, such as a signal - * or the Guest doing a NOTIFY out to the Launcher. :*/ + * tell us the Guest's memory layout and entry point. A read will run the + * Guest until something happens, such as a signal or the Guest doing a NOTIFY + * out to the Launcher. +:*/ #include <linux/uaccess.h> #include <linux/miscdevice.h> #include <linux/fs.h> @@ -11,14 +12,41 @@ #include <linux/file.h> #include "lg.h" +/*L:056 + * Before we move on, let's jump ahead and look at what the kernel does when + * it needs to look up the eventfds. That will complete our picture of how we + * use RCU. + * + * The notification value is in cpu->pending_notify: we return true if it went + * to an eventfd. + */ bool send_notify_to_eventfd(struct lg_cpu *cpu) { unsigned int i; struct lg_eventfd_map *map; - /* lg->eventfds is RCU-protected */ + /* + * This "rcu_read_lock()" helps track when someone is still looking at + * the (RCU-using) eventfds array. It's not actually a lock at all; + * indeed it's a noop in many configurations. (You didn't expect me to + * explain all the RCU secrets here, did you?) + */ rcu_read_lock(); + /* + * rcu_dereference is the counter-side of rcu_assign_pointer(); it + * makes sure we don't access the memory pointed to by + * cpu->lg->eventfds before cpu->lg->eventfds is set. Sounds crazy, + * but Alpha allows this! Paul McKenney points out that a really + * aggressive compiler could have the same effect: + * http://lists.ozlabs.org/pipermail/lguest/2009-July/001560.html + * + * So play safe, use rcu_dereference to get the rcu-protected pointer: + */ map = rcu_dereference(cpu->lg->eventfds); + /* + * Simple array search: even if they add an eventfd while we do this, + * we'll continue to use the old array and just won't see the new one. + */ for (i = 0; i < map->num; i++) { if (map->map[i].addr == cpu->pending_notify) { eventfd_signal(map->map[i].event, 1); @@ -26,19 +54,50 @@ bool send_notify_to_eventfd(struct lg_cpu *cpu) break; } } + /* We're done with the rcu-protected variable cpu->lg->eventfds. */ rcu_read_unlock(); + + /* If we cleared the notification, it's because we found a match. */ return cpu->pending_notify == 0; } +/*L:055 + * One of the more tricksy tricks in the Linux Kernel is a technique called + * Read Copy Update. Since one point of lguest is to teach lguest journeyers + * about kernel coding, I use it here. (In case you're curious, other purposes + * include learning about virtualization and instilling a deep appreciation for + * simplicity and puppies). + * + * We keep a simple array which maps LHCALL_NOTIFY values to eventfds, but we + * add new eventfds without ever blocking readers from accessing the array. + * The current Launcher only does this during boot, so that never happens. But + * Read Copy Update is cool, and adding a lock risks damaging even more puppies + * than this code does. + * + * We allocate a brand new one-larger array, copy the old one and add our new + * element. Then we make the lg eventfd pointer point to the new array. + * That's the easy part: now we need to free the old one, but we need to make + * sure no slow CPU somewhere is still looking at it. That's what + * synchronize_rcu does for us: waits until every CPU has indicated that it has + * moved on to know it's no longer using the old one. + * + * If that's unclear, see http://en.wikipedia.org/wiki/Read-copy-update. + */ static int add_eventfd(struct lguest *lg, unsigned long addr, int fd) { struct lg_eventfd_map *new, *old = lg->eventfds; + /* + * We don't allow notifications on value 0 anyway (pending_notify of + * 0 means "nothing pending"). + */ if (!addr) return -EINVAL; - /* Replace the old array with the new one, carefully: others can - * be accessing it at the same time */ + /* + * Replace the old array with the new one, carefully: others can + * be accessing it at the same time. + */ new = kmalloc(sizeof(*new) + sizeof(new->map[0]) * (old->num + 1), GFP_KERNEL); if (!new) @@ -52,22 +111,41 @@ static int add_eventfd(struct lguest *lg, unsigned long addr, int fd) new->map[new->num].addr = addr; new->map[new->num].event = eventfd_ctx_fdget(fd); if (IS_ERR(new->map[new->num].event)) { + int err = PTR_ERR(new->map[new->num].event); kfree(new); - return PTR_ERR(new->map[new->num].event); + return err; } new->num++; - /* Now put new one in place. */ + /* + * Now put new one in place: rcu_assign_pointer() is a fancy way of + * doing "lg->eventfds = new", but it uses memory barriers to make + * absolutely sure that the contents of "new" written above is nailed + * down before we actually do the assignment. + * + * We have to think about these kinds of things when we're operating on + * live data without locks. + */ rcu_assign_pointer(lg->eventfds, new); - /* We're not in a big hurry. Wait until noone's looking at old - * version, then delete it. */ + /* + * We're not in a big hurry. Wait until noone's looking at old + * version, then free it. + */ synchronize_rcu(); kfree(old); return 0; } +/*L:052 + * Receiving notifications from the Guest is usually done by attaching a + * particular LHCALL_NOTIFY value to an event filedescriptor. The eventfd will + * become readable when the Guest does an LHCALL_NOTIFY with that value. + * + * This is really convenient for processing each virtqueue in a separate + * thread. + */ static int attach_eventfd(struct lguest *lg, const unsigned long __user *input) { unsigned long addr, fd; @@ -79,15 +157,22 @@ static int attach_eventfd(struct lguest *lg, const unsigned long __user *input) if (get_user(fd, input) != 0) return -EFAULT; + /* + * Just make sure two callers don't add eventfds at once. We really + * only need to lock against callers adding to the same Guest, so using + * the Big Lguest Lock is overkill. But this is setup, not a fast path. + */ mutex_lock(&lguest_lock); err = add_eventfd(lg, addr, fd); mutex_unlock(&lguest_lock); - return 0; + return err; } -/*L:050 Sending an interrupt is done by writing LHREQ_IRQ and an interrupt - * number to /dev/lguest. */ +/*L:050 + * Sending an interrupt is done by writing LHREQ_IRQ and an interrupt + * number to /dev/lguest. + */ static int user_send_irq(struct lg_cpu *cpu, const unsigned long __user *input) { unsigned long irq; @@ -97,12 +182,18 @@ static int user_send_irq(struct lg_cpu *cpu, const unsigned long __user *input) if (irq >= LGUEST_IRQS) return -EINVAL; + /* + * Next time the Guest runs, the core code will see if it can deliver + * this interrupt. + */ set_interrupt(cpu, irq); return 0; } -/*L:040 Once our Guest is initialized, the Launcher makes it run by reading - * from /dev/lguest. */ +/*L:040 + * Once our Guest is initialized, the Launcher makes it run by reading + * from /dev/lguest. + */ static ssize_t read(struct file *file, char __user *user, size_t size,loff_t*o) { struct lguest *lg = file->private_data; @@ -138,8 +229,10 @@ static ssize_t read(struct file *file, char __user *user, size_t size,loff_t*o) return len; } - /* If we returned from read() last time because the Guest sent I/O, - * clear the flag. */ + /* + * If we returned from read() last time because the Guest sent I/O, + * clear the flag. + */ if (cpu->pending_notify) cpu->pending_notify = 0; @@ -147,8 +240,10 @@ static ssize_t read(struct file *file, char __user *user, size_t size,loff_t*o) return run_guest(cpu, (unsigned long __user *)user); } -/*L:025 This actually initializes a CPU. For the moment, a Guest is only - * uniprocessor, so "id" is always 0. */ +/*L:025 + * This actually initializes a CPU. For the moment, a Guest is only + * uniprocessor, so "id" is always 0. + */ static int lg_cpu_start(struct lg_cpu *cpu, unsigned id, unsigned long start_ip) { /* We have a limited number the number of CPUs in the lguest struct. */ @@ -163,8 +258,10 @@ static int lg_cpu_start(struct lg_cpu *cpu, unsigned id, unsigned long start_ip) /* Each CPU has a timer it can set. */ init_clockdev(cpu); - /* We need a complete page for the Guest registers: they are accessible - * to the Guest and we can only grant it access to whole pages. */ + /* + * We need a complete page for the Guest registers: they are accessible + * to the Guest and we can only grant it access to whole pages. + */ cpu->regs_page = get_zeroed_page(GFP_KERNEL); if (!cpu->regs_page) return -ENOMEM; @@ -172,29 +269,38 @@ static int lg_cpu_start(struct lg_cpu *cpu, unsigned id, unsigned long start_ip) /* We actually put the registers at the bottom of the page. */ cpu->regs = (void *)cpu->regs_page + PAGE_SIZE - sizeof(*cpu->regs); - /* Now we initialize the Guest's registers, handing it the start - * address. */ + /* + * Now we initialize the Guest's registers, handing it the start + * address. + */ lguest_arch_setup_regs(cpu, start_ip); - /* We keep a pointer to the Launcher task (ie. current task) for when - * other Guests want to wake this one (eg. console input). */ + /* + * We keep a pointer to the Launcher task (ie. current task) for when + * other Guests want to wake this one (eg. console input). + */ cpu->tsk = current; - /* We need to keep a pointer to the Launcher's memory map, because if + /* + * We need to keep a pointer to the Launcher's memory map, because if * the Launcher dies we need to clean it up. If we don't keep a - * reference, it is destroyed before close() is called. */ + * reference, it is destroyed before close() is called. + */ cpu->mm = get_task_mm(cpu->tsk); - /* We remember which CPU's pages this Guest used last, for optimization - * when the same Guest runs on the same CPU twice. */ + /* + * We remember which CPU's pages this Guest used last, for optimization + * when the same Guest runs on the same CPU twice. + */ cpu->last_pages = NULL; /* No error == success. */ return 0; } -/*L:020 The initialization write supplies 3 pointer sized (32 or 64 bit) - * values (in addition to the LHREQ_INITIALIZE value). These are: +/*L:020 + * The initialization write supplies 3 pointer sized (32 or 64 bit) values (in + * addition to the LHREQ_INITIALIZE value). These are: * * base: The start of the Guest-physical memory inside the Launcher memory. * @@ -206,14 +312,15 @@ static int lg_cpu_start(struct lg_cpu *cpu, unsigned id, unsigned long start_ip) */ static int initialize(struct file *file, const unsigned long __user *input) { - /* "struct lguest" contains everything we (the Host) know about a - * Guest. */ + /* "struct lguest" contains all we (the Host) know about a Guest. */ struct lguest *lg; int err; unsigned long args[3]; - /* We grab the Big Lguest lock, which protects against multiple - * simultaneous initializations. */ + /* + * We grab the Big Lguest lock, which protects against multiple + * simultaneous initializations. + */ mutex_lock(&lguest_lock); /* You can't initialize twice! Close the device and start again... */ if (file->private_data) { @@ -248,8 +355,10 @@ static int initialize(struct file *file, const unsigned long __user *input) if (err) goto free_eventfds; - /* Initialize the Guest's shadow page tables, using the toplevel - * address the Launcher gave us. This allocates memory, so can fail. */ + /* + * Initialize the Guest's shadow page tables, using the toplevel + * address the Launcher gave us. This allocates memory, so can fail. + */ err = init_guest_pagetable(lg); if (err) goto free_regs; @@ -274,20 +383,24 @@ unlock: return err; } -/*L:010 The first operation the Launcher does must be a write. All writes +/*L:010 + * The first operation the Launcher does must be a write. All writes * start with an unsigned long number: for the first write this must be * LHREQ_INITIALIZE to set up the Guest. After that the Launcher can use - * writes of other values to send interrupts. + * writes of other values to send interrupts or set up receipt of notifications. * * Note that we overload the "offset" in the /dev/lguest file to indicate what - * CPU number we're dealing with. Currently this is always 0, since we only + * CPU number we're dealing with. Currently this is always 0 since we only * support uniprocessor Guests, but you can see the beginnings of SMP support - * here. */ + * here. + */ static ssize_t write(struct file *file, const char __user *in, size_t size, loff_t *off) { - /* Once the Guest is initialized, we hold the "struct lguest" in the - * file private data. */ + /* + * Once the Guest is initialized, we hold the "struct lguest" in the + * file private data. + */ struct lguest *lg = file->private_data; const unsigned long __user *input = (const unsigned long __user *)in; unsigned long req; @@ -322,13 +435,15 @@ static ssize_t write(struct file *file, const char __user *in, } } -/*L:060 The final piece of interface code is the close() routine. It reverses +/*L:060 + * The final piece of interface code is the close() routine. It reverses * everything done in initialize(). This is usually called because the * Launcher exited. * * Note that the close routine returns 0 or a negative error number: it can't * really fail, but it can whine. I blame Sun for this wart, and K&R C for - * letting them do it. :*/ + * letting them do it. +:*/ static int close(struct inode *inode, struct file *file) { struct lguest *lg = file->private_data; @@ -338,8 +453,10 @@ static int close(struct inode *inode, struct file *file) if (!lg) return 0; - /* We need the big lock, to protect from inter-guest I/O and other - * Launchers initializing guests. */ + /* + * We need the big lock, to protect from inter-guest I/O and other + * Launchers initializing guests. + */ mutex_lock(&lguest_lock); /* Free up the shadow page tables for the Guest. */ @@ -350,8 +467,10 @@ static int close(struct inode *inode, struct file *file) hrtimer_cancel(&lg->cpus[i].hrt); /* We can free up the register page we allocated. */ free_page(lg->cpus[i].regs_page); - /* Now all the memory cleanups are done, it's safe to release - * the Launcher's memory management structure. */ + /* + * Now all the memory cleanups are done, it's safe to release + * the Launcher's memory management structure. + */ mmput(lg->cpus[i].mm); } @@ -360,8 +479,10 @@ static int close(struct inode *inode, struct file *file) eventfd_ctx_put(lg->eventfds->map[i].event); kfree(lg->eventfds); - /* If lg->dead doesn't contain an error code it will be NULL or a - * kmalloc()ed string, either of which is ok to hand to kfree(). */ + /* + * If lg->dead doesn't contain an error code it will be NULL or a + * kmalloc()ed string, either of which is ok to hand to kfree(). + */ if (!IS_ERR(lg->dead)) kfree(lg->dead); /* Free the memory allocated to the lguest_struct */ @@ -385,7 +506,8 @@ static int close(struct inode *inode, struct file *file) * * We begin our understanding with the Host kernel interface which the Launcher * uses: reading and writing a character device called /dev/lguest. All the - * work happens in the read(), write() and close() routines: */ + * work happens in the read(), write() and close() routines: + */ static struct file_operations lguest_fops = { .owner = THIS_MODULE, .release = close, @@ -393,8 +515,10 @@ static struct file_operations lguest_fops = { .read = read, }; -/* This is a textbook example of a "misc" character device. Populate a "struct - * miscdevice" and register it with misc_register(). */ +/* + * This is a textbook example of a "misc" character device. Populate a "struct + * miscdevice" and register it with misc_register(). + */ static struct miscdevice lguest_dev = { .minor = MISC_DYNAMIC_MINOR, .name = "lguest", diff --git a/drivers/lguest/page_tables.c b/drivers/lguest/page_tables.c index a6fe1abda24..a8d0aee3bc0 100644 --- a/drivers/lguest/page_tables.c +++ b/drivers/lguest/page_tables.c @@ -1,9 +1,11 @@ -/*P:700 The pagetable code, on the other hand, still shows the scars of +/*P:700 + * The pagetable code, on the other hand, still shows the scars of * previous encounters. It's functional, and as neat as it can be in the * circumstances, but be wary, for these things are subtle and break easily. * The Guest provides a virtual to physical mapping, but we can neither trust * it nor use it: we verify and convert it here then point the CPU to the - * converted Guest pages when running the Guest. :*/ + * converted Guest pages when running the Guest. +:*/ /* Copyright (C) Rusty Russell IBM Corporation 2006. * GPL v2 and any later version */ @@ -17,18 +19,20 @@ #include <asm/bootparam.h> #include "lg.h" -/*M:008 We hold reference to pages, which prevents them from being swapped. +/*M:008 + * We hold reference to pages, which prevents them from being swapped. * It'd be nice to have a callback in the "struct mm_struct" when Linux wants * to swap out. If we had this, and a shrinker callback to trim PTE pages, we - * could probably consider launching Guests as non-root. :*/ + * could probably consider launching Guests as non-root. +:*/ /*H:300 * The Page Table Code * - * We use two-level page tables for the Guest. If you're not entirely - * comfortable with virtual addresses, physical addresses and page tables then - * I recommend you review arch/x86/lguest/boot.c's "Page Table Handling" (with - * diagrams!). + * We use two-level page tables for the Guest, or three-level with PAE. If + * you're not entirely comfortable with virtual addresses, physical addresses + * and page tables then I recommend you review arch/x86/lguest/boot.c's "Page + * Table Handling" (with diagrams!). * * The Guest keeps page tables, but we maintain the actual ones here: these are * called "shadow" page tables. Which is a very Guest-centric name: these are @@ -45,16 +49,18 @@ * (v) Flushing (throwing away) page tables, * (vi) Mapping the Switcher when the Guest is about to run, * (vii) Setting up the page tables initially. - :*/ +:*/ - -/* 1024 entries in a page table page maps 1024 pages: 4MB. The Switcher is - * conveniently placed at the top 4MB, so it uses a separate, complete PTE - * page. */ +/* + * The Switcher uses the complete top PTE page. That's 1024 PTE entries (4MB) + * or 512 PTE entries with PAE (2MB). + */ #define SWITCHER_PGD_INDEX (PTRS_PER_PGD - 1) -/* For PAE we need the PMD index as well. We use the last 2MB, so we - * will need the last pmd entry of the last pmd page. */ +/* + * For PAE we need the PMD index as well. We use the last 2MB, so we + * will need the last pmd entry of the last pmd page. + */ #ifdef CONFIG_X86_PAE #define SWITCHER_PMD_INDEX (PTRS_PER_PMD - 1) #define RESERVE_MEM 2U @@ -64,14 +70,18 @@ #define CHECK_GPGD_MASK _PAGE_TABLE #endif -/* We actually need a separate PTE page for each CPU. Remember that after the +/* + * We actually need a separate PTE page for each CPU. Remember that after the * Switcher code itself comes two pages for each CPU, and we don't want this - * CPU's guest to see the pages of any other CPU. */ + * CPU's guest to see the pages of any other CPU. + */ static DEFINE_PER_CPU(pte_t *, switcher_pte_pages); #define switcher_pte_page(cpu) per_cpu(switcher_pte_pages, cpu) -/*H:320 The page table code is curly enough to need helper functions to keep it - * clear and clean. +/*H:320 + * The page table code is curly enough to need helper functions to keep it + * clear and clean. The kernel itself provides many of them; one advantage + * of insisting that the Guest and Host use the same CONFIG_PAE setting. * * There are two functions which return pointers to the shadow (aka "real") * page tables. @@ -79,7 +89,8 @@ static DEFINE_PER_CPU(pte_t *, switcher_pte_pages); * spgd_addr() takes the virtual address and returns a pointer to the top-level * page directory entry (PGD) for that address. Since we keep track of several * page tables, the "i" argument tells us which one we're interested in (it's - * usually the current one). */ + * usually the current one). + */ static pgd_t *spgd_addr(struct lg_cpu *cpu, u32 i, unsigned long vaddr) { unsigned int index = pgd_index(vaddr); @@ -96,9 +107,11 @@ static pgd_t *spgd_addr(struct lg_cpu *cpu, u32 i, unsigned long vaddr) } #ifdef CONFIG_X86_PAE -/* This routine then takes the PGD entry given above, which contains the +/* + * This routine then takes the PGD entry given above, which contains the * address of the PMD page. It then returns a pointer to the PMD entry for the - * given address. */ + * given address. + */ static pmd_t *spmd_addr(struct lg_cpu *cpu, pgd_t spgd, unsigned long vaddr) { unsigned int index = pmd_index(vaddr); @@ -119,9 +132,11 @@ static pmd_t *spmd_addr(struct lg_cpu *cpu, pgd_t spgd, unsigned long vaddr) } #endif -/* This routine then takes the page directory entry returned above, which +/* + * This routine then takes the page directory entry returned above, which * contains the address of the page table entry (PTE) page. It then returns a - * pointer to the PTE entry for the given address. */ + * pointer to the PTE entry for the given address. + */ static pte_t *spte_addr(struct lg_cpu *cpu, pgd_t spgd, unsigned long vaddr) { #ifdef CONFIG_X86_PAE @@ -139,8 +154,10 @@ static pte_t *spte_addr(struct lg_cpu *cpu, pgd_t spgd, unsigned long vaddr) return &page[pte_index(vaddr)]; } -/* These two functions just like the above two, except they access the Guest - * page tables. Hence they return a Guest address. */ +/* + * These functions are just like the above two, except they access the Guest + * page tables. Hence they return a Guest address. + */ static unsigned long gpgd_addr(struct lg_cpu *cpu, unsigned long vaddr) { unsigned int index = vaddr >> (PGDIR_SHIFT); @@ -148,6 +165,7 @@ static unsigned long gpgd_addr(struct lg_cpu *cpu, unsigned long vaddr) } #ifdef CONFIG_X86_PAE +/* Follow the PGD to the PMD. */ static unsigned long gpmd_addr(pgd_t gpgd, unsigned long vaddr) { unsigned long gpage = pgd_pfn(gpgd) << PAGE_SHIFT; @@ -155,6 +173,7 @@ static unsigned long gpmd_addr(pgd_t gpgd, unsigned long vaddr) return gpage + pmd_index(vaddr) * sizeof(pmd_t); } +/* Follow the PMD to the PTE. */ static unsigned long gpte_addr(struct lg_cpu *cpu, pmd_t gpmd, unsigned long vaddr) { @@ -164,6 +183,7 @@ static unsigned long gpte_addr(struct lg_cpu *cpu, return gpage + pte_index(vaddr) * sizeof(pte_t); } #else +/* Follow the PGD to the PTE (no mid-level for !PAE). */ static unsigned long gpte_addr(struct lg_cpu *cpu, pgd_t gpgd, unsigned long vaddr) { @@ -175,17 +195,21 @@ static unsigned long gpte_addr(struct lg_cpu *cpu, #endif /*:*/ -/*M:014 get_pfn is slow: we could probably try to grab batches of pages here as - * an optimization (ie. pre-faulting). :*/ +/*M:014 + * get_pfn is slow: we could probably try to grab batches of pages here as + * an optimization (ie. pre-faulting). +:*/ -/*H:350 This routine takes a page number given by the Guest and converts it to +/*H:350 + * This routine takes a page number given by the Guest and converts it to * an actual, physical page number. It can fail for several reasons: the * virtual address might not be mapped by the Launcher, the write flag is set * and the page is read-only, or the write flag was set and the page was * shared so had to be copied, but we ran out of memory. * * This holds a reference to the page, so release_pte() is careful to put that - * back. */ + * back. + */ static unsigned long get_pfn(unsigned long virtpfn, int write) { struct page *page; @@ -198,33 +222,41 @@ static unsigned long get_pfn(unsigned long virtpfn, int write) return -1UL; } -/*H:340 Converting a Guest page table entry to a shadow (ie. real) page table +/*H:340 + * Converting a Guest page table entry to a shadow (ie. real) page table * entry can be a little tricky. The flags are (almost) the same, but the * Guest PTE contains a virtual page number: the CPU needs the real page - * number. */ + * number. + */ static pte_t gpte_to_spte(struct lg_cpu *cpu, pte_t gpte, int write) { unsigned long pfn, base, flags; - /* The Guest sets the global flag, because it thinks that it is using + /* + * The Guest sets the global flag, because it thinks that it is using * PGE. We only told it to use PGE so it would tell us whether it was * flushing a kernel mapping or a userspace mapping. We don't actually - * use the global bit, so throw it away. */ + * use the global bit, so throw it away. + */ flags = (pte_flags(gpte) & ~_PAGE_GLOBAL); /* The Guest's pages are offset inside the Launcher. */ base = (unsigned long)cpu->lg->mem_base / PAGE_SIZE; - /* We need a temporary "unsigned long" variable to hold the answer from + /* + * We need a temporary "unsigned long" variable to hold the answer from * get_pfn(), because it returns 0xFFFFFFFF on failure, which wouldn't * fit in spte.pfn. get_pfn() finds the real physical number of the - * page, given the virtual number. */ + * page, given the virtual number. + */ pfn = get_pfn(base + pte_pfn(gpte), write); if (pfn == -1UL) { kill_guest(cpu, "failed to get page %lu", pte_pfn(gpte)); - /* When we destroy the Guest, we'll go through the shadow page + /* + * When we destroy the Guest, we'll go through the shadow page * tables and release_pte() them. Make sure we don't think - * this one is valid! */ + * this one is valid! + */ flags = 0; } /* Now we assemble our shadow PTE from the page number and flags. */ @@ -234,8 +266,10 @@ static pte_t gpte_to_spte(struct lg_cpu *cpu, pte_t gpte, int write) /*H:460 And to complete the chain, release_pte() looks like this: */ static void release_pte(pte_t pte) { - /* Remember that get_user_pages_fast() took a reference to the page, in - * get_pfn()? We have to put it back now. */ + /* + * Remember that get_user_pages_fast() took a reference to the page, in + * get_pfn()? We have to put it back now. + */ if (pte_flags(pte) & _PAGE_PRESENT) put_page(pte_page(pte)); } @@ -273,7 +307,8 @@ static void check_gpmd(struct lg_cpu *cpu, pmd_t gpmd) * and return to the Guest without it knowing. * * If we fixed up the fault (ie. we mapped the address), this routine returns - * true. Otherwise, it was a real fault and we need to tell the Guest. */ + * true. Otherwise, it was a real fault and we need to tell the Guest. + */ bool demand_page(struct lg_cpu *cpu, unsigned long vaddr, int errcode) { pgd_t gpgd; @@ -282,6 +317,7 @@ bool demand_page(struct lg_cpu *cpu, unsigned long vaddr, int errcode) pte_t gpte; pte_t *spte; + /* Mid level for PAE. */ #ifdef CONFIG_X86_PAE pmd_t *spmd; pmd_t gpmd; @@ -298,22 +334,26 @@ bool demand_page(struct lg_cpu *cpu, unsigned long vaddr, int errcode) if (!(pgd_flags(*spgd) & _PAGE_PRESENT)) { /* No shadow entry: allocate a new shadow PTE page. */ unsigned long ptepage = get_zeroed_page(GFP_KERNEL); - /* This is not really the Guest's fault, but killing it is - * simple for this corner case. */ + /* + * This is not really the Guest's fault, but killing it is + * simple for this corner case. + */ if (!ptepage) { kill_guest(cpu, "out of memory allocating pte page"); return false; } /* We check that the Guest pgd is OK. */ check_gpgd(cpu, gpgd); - /* And we copy the flags to the shadow PGD entry. The page - * number in the shadow PGD is the page we just allocated. */ + /* + * And we copy the flags to the shadow PGD entry. The page + * number in the shadow PGD is the page we just allocated. + */ set_pgd(spgd, __pgd(__pa(ptepage) | pgd_flags(gpgd))); } #ifdef CONFIG_X86_PAE gpmd = lgread(cpu, gpmd_addr(gpgd, vaddr), pmd_t); - /* middle level not present? We can't map it in. */ + /* Middle level not present? We can't map it in. */ if (!(pmd_flags(gpmd) & _PAGE_PRESENT)) return false; @@ -324,8 +364,10 @@ bool demand_page(struct lg_cpu *cpu, unsigned long vaddr, int errcode) /* No shadow entry: allocate a new shadow PTE page. */ unsigned long ptepage = get_zeroed_page(GFP_KERNEL); - /* This is not really the Guest's fault, but killing it is - * simple for this corner case. */ + /* + * This is not really the Guest's fault, but killing it is + * simple for this corner case. + */ if (!ptepage) { kill_guest(cpu, "out of memory allocating pte page"); return false; @@ -334,27 +376,37 @@ bool demand_page(struct lg_cpu *cpu, unsigned long vaddr, int errcode) /* We check that the Guest pmd is OK. */ check_gpmd(cpu, gpmd); - /* And we copy the flags to the shadow PMD entry. The page - * number in the shadow PMD is the page we just allocated. */ + /* + * And we copy the flags to the shadow PMD entry. The page + * number in the shadow PMD is the page we just allocated. + */ native_set_pmd(spmd, __pmd(__pa(ptepage) | pmd_flags(gpmd))); } - /* OK, now we look at the lower level in the Guest page table: keep its - * address, because we might update it later. */ + /* + * OK, now we look at the lower level in the Guest page table: keep its + * address, because we might update it later. + */ gpte_ptr = gpte_addr(cpu, gpmd, vaddr); #else - /* OK, now we look at the lower level in the Guest page table: keep its - * address, because we might update it later. */ + /* + * OK, now we look at the lower level in the Guest page table: keep its + * address, because we might update it later. + */ gpte_ptr = gpte_addr(cpu, gpgd, vaddr); #endif + + /* Read the actual PTE value. */ gpte = lgread(cpu, gpte_ptr, pte_t); /* If this page isn't in the Guest page tables, we can't page it in. */ if (!(pte_flags(gpte) & _PAGE_PRESENT)) return false; - /* Check they're not trying to write to a page the Guest wants - * read-only (bit 2 of errcode == write). */ + /* + * Check they're not trying to write to a page the Guest wants + * read-only (bit 2 of errcode == write). + */ if ((errcode & 2) && !(pte_flags(gpte) & _PAGE_RW)) return false; @@ -362,8 +414,10 @@ bool demand_page(struct lg_cpu *cpu, unsigned long vaddr, int errcode) if ((errcode & 4) && !(pte_flags(gpte) & _PAGE_USER)) return false; - /* Check that the Guest PTE flags are OK, and the page number is below - * the pfn_limit (ie. not mapping the Launcher binary). */ + /* + * Check that the Guest PTE flags are OK, and the page number is below + * the pfn_limit (ie. not mapping the Launcher binary). + */ check_gpte(cpu, gpte); /* Add the _PAGE_ACCESSED and (for a write) _PAGE_DIRTY flag */ @@ -373,29 +427,40 @@ bool demand_page(struct lg_cpu *cpu, unsigned long vaddr, int errcode) /* Get the pointer to the shadow PTE entry we're going to set. */ spte = spte_addr(cpu, *spgd, vaddr); - /* If there was a valid shadow PTE entry here before, we release it. - * This can happen with a write to a previously read-only entry. */ + + /* + * If there was a valid shadow PTE entry here before, we release it. + * This can happen with a write to a previously read-only entry. + */ release_pte(*spte); - /* If this is a write, we insist that the Guest page is writable (the - * final arg to gpte_to_spte()). */ + /* + * If this is a write, we insist that the Guest page is writable (the + * final arg to gpte_to_spte()). + */ if (pte_dirty(gpte)) *spte = gpte_to_spte(cpu, gpte, 1); else - /* If this is a read, don't set the "writable" bit in the page + /* + * If this is a read, don't set the "writable" bit in the page * table entry, even if the Guest says it's writable. That way * we will come back here when a write does actually occur, so - * we can update the Guest's _PAGE_DIRTY flag. */ + * we can update the Guest's _PAGE_DIRTY flag. + */ native_set_pte(spte, gpte_to_spte(cpu, pte_wrprotect(gpte), 0)); - /* Finally, we write the Guest PTE entry back: we've set the - * _PAGE_ACCESSED and maybe the _PAGE_DIRTY flags. */ + /* + * Finally, we write the Guest PTE entry back: we've set the + * _PAGE_ACCESSED and maybe the _PAGE_DIRTY flags. + */ lgwrite(cpu, gpte_ptr, pte_t, gpte); - /* The fault is fixed, the page table is populated, the mapping + /* + * The fault is fixed, the page table is populated, the mapping * manipulated, the result returned and the code complete. A small * delay and a trace of alliteration are the only indications the Guest - * has that a page fault occurred at all. */ + * has that a page fault occurred at all. + */ return true; } @@ -408,7 +473,8 @@ bool demand_page(struct lg_cpu *cpu, unsigned long vaddr, int errcode) * mapped, so it's overkill. * * This is a quick version which answers the question: is this virtual address - * mapped by the shadow page tables, and is it writable? */ + * mapped by the shadow page tables, and is it writable? + */ static bool page_writable(struct lg_cpu *cpu, unsigned long vaddr) { pgd_t *spgd; @@ -428,21 +494,26 @@ static bool page_writable(struct lg_cpu *cpu, unsigned long vaddr) return false; #endif - /* Check the flags on the pte entry itself: it must be present and - * writable. */ + /* + * Check the flags on the pte entry itself: it must be present and + * writable. + */ flags = pte_flags(*(spte_addr(cpu, *spgd, vaddr))); return (flags & (_PAGE_PRESENT|_PAGE_RW)) == (_PAGE_PRESENT|_PAGE_RW); } -/* So, when pin_stack_pages() asks us to pin a page, we check if it's already +/* + * So, when pin_stack_pages() asks us to pin a page, we check if it's already * in the page tables, and if not, we call demand_page() with error code 2 - * (meaning "write"). */ + * (meaning "write"). + */ void pin_page(struct lg_cpu *cpu, unsigned long vaddr) { if (!page_writable(cpu, vaddr) && !demand_page(cpu, vaddr, 2)) kill_guest(cpu, "bad stack page %#lx", vaddr); } +/*:*/ #ifdef CONFIG_X86_PAE static void release_pmd(pmd_t *spmd) @@ -479,15 +550,21 @@ static void release_pgd(pgd_t *spgd) } #else /* !CONFIG_X86_PAE */ -/*H:450 If we chase down the release_pgd() code, it looks like this: */ +/*H:450 + * If we chase down the release_pgd() code, the non-PAE version looks like + * this. The PAE version is almost identical, but instead of calling + * release_pte it calls release_pmd(), which looks much like this. + */ static void release_pgd(pgd_t *spgd) { /* If the entry's not present, there's nothing to release. */ if (pgd_flags(*spgd) & _PAGE_PRESENT) { unsigned int i; - /* Converting the pfn to find the actual PTE page is easy: turn + /* + * Converting the pfn to find the actual PTE page is easy: turn * the page number into a physical address, then convert to a - * virtual address (easy for kernel pages like this one). */ + * virtual address (easy for kernel pages like this one). + */ pte_t *ptepage = __va(pgd_pfn(*spgd) << PAGE_SHIFT); /* For each entry in the page, we might need to release it. */ for (i = 0; i < PTRS_PER_PTE; i++) @@ -499,9 +576,12 @@ static void release_pgd(pgd_t *spgd) } } #endif -/*H:445 We saw flush_user_mappings() twice: once from the flush_user_mappings() + +/*H:445 + * We saw flush_user_mappings() twice: once from the flush_user_mappings() * hypercall and once in new_pgdir() when we re-used a top-level pgdir page. - * It simply releases every PTE page from 0 up to the Guest's kernel address. */ + * It simply releases every PTE page from 0 up to the Guest's kernel address. + */ static void flush_user_mappings(struct lguest *lg, int idx) { unsigned int i; @@ -510,10 +590,12 @@ static void flush_user_mappings(struct lguest *lg, int idx) release_pgd(lg->pgdirs[idx].pgdir + i); } -/*H:440 (v) Flushing (throwing away) page tables, +/*H:440 + * (v) Flushing (throwing away) page tables, * * The Guest has a hypercall to throw away the page tables: it's used when a - * large number of mappings have been changed. */ + * large number of mappings have been changed. + */ void guest_pagetable_flush_user(struct lg_cpu *cpu) { /* Drop the userspace part of the current page table. */ @@ -551,9 +633,11 @@ unsigned long guest_pa(struct lg_cpu *cpu, unsigned long vaddr) return pte_pfn(gpte) * PAGE_SIZE | (vaddr & ~PAGE_MASK); } -/* We keep several page tables. This is a simple routine to find the page +/* + * We keep several page tables. This is a simple routine to find the page * table (if any) corresponding to this top-level address the Guest has given - * us. */ + * us. + */ static unsigned int find_pgdir(struct lguest *lg, unsigned long pgtable) { unsigned int i; @@ -563,9 +647,11 @@ static unsigned int find_pgdir(struct lguest *lg, unsigned long pgtable) return i; } -/*H:435 And this is us, creating the new page directory. If we really do +/*H:435 + * And this is us, creating the new page directory. If we really do * allocate a new one (and so the kernel parts are not there), we set - * blank_pgdir. */ + * blank_pgdir. + */ static unsigned int new_pgdir(struct lg_cpu *cpu, unsigned long gpgdir, int *blank_pgdir) @@ -575,8 +661,10 @@ static unsigned int new_pgdir(struct lg_cpu *cpu, pmd_t *pmd_table; #endif - /* We pick one entry at random to throw out. Choosing the Least - * Recently Used might be better, but this is easy. */ + /* + * We pick one entry at random to throw out. Choosing the Least + * Recently Used might be better, but this is easy. + */ next = random32() % ARRAY_SIZE(cpu->lg->pgdirs); /* If it's never been allocated at all before, try now. */ if (!cpu->lg->pgdirs[next].pgdir) { @@ -587,8 +675,10 @@ static unsigned int new_pgdir(struct lg_cpu *cpu, next = cpu->cpu_pgd; else { #ifdef CONFIG_X86_PAE - /* In PAE mode, allocate a pmd page and populate the - * last pgd entry. */ + /* + * In PAE mode, allocate a pmd page and populate the + * last pgd entry. + */ pmd_table = (pmd_t *)get_zeroed_page(GFP_KERNEL); if (!pmd_table) { free_page((long)cpu->lg->pgdirs[next].pgdir); @@ -598,8 +688,10 @@ static unsigned int new_pgdir(struct lg_cpu *cpu, set_pgd(cpu->lg->pgdirs[next].pgdir + SWITCHER_PGD_INDEX, __pgd(__pa(pmd_table) | _PAGE_PRESENT)); - /* This is a blank page, so there are no kernel - * mappings: caller must map the stack! */ + /* + * This is a blank page, so there are no kernel + * mappings: caller must map the stack! + */ *blank_pgdir = 1; } #else @@ -615,19 +707,23 @@ static unsigned int new_pgdir(struct lg_cpu *cpu, return next; } -/*H:430 (iv) Switching page tables +/*H:430 + * (iv) Switching page tables * * Now we've seen all the page table setting and manipulation, let's see * what happens when the Guest changes page tables (ie. changes the top-level - * pgdir). This occurs on almost every context switch. */ + * pgdir). This occurs on almost every context switch. + */ void guest_new_pagetable(struct lg_cpu *cpu, unsigned long pgtable) { int newpgdir, repin = 0; /* Look to see if we have this one already. */ newpgdir = find_pgdir(cpu->lg, pgtable); - /* If not, we allocate or mug an existing one: if it's a fresh one, - * repin gets set to 1. */ + /* + * If not, we allocate or mug an existing one: if it's a fresh one, + * repin gets set to 1. + */ if (newpgdir == ARRAY_SIZE(cpu->lg->pgdirs)) newpgdir = new_pgdir(cpu, pgtable, &repin); /* Change the current pgd index to the new one. */ @@ -637,9 +733,11 @@ void guest_new_pagetable(struct lg_cpu *cpu, unsigned long pgtable) pin_stack_pages(cpu); } -/*H:470 Finally, a routine which throws away everything: all PGD entries in all +/*H:470 + * Finally, a routine which throws away everything: all PGD entries in all * the shadow page tables, including the Guest's kernel mappings. This is used - * when we destroy the Guest. */ + * when we destroy the Guest. + */ static void release_all_pagetables(struct lguest *lg) { unsigned int i, j; @@ -656,8 +754,10 @@ static void release_all_pagetables(struct lguest *lg) spgd = lg->pgdirs[i].pgdir + SWITCHER_PGD_INDEX; pmdpage = __va(pgd_pfn(*spgd) << PAGE_SHIFT); - /* And release the pmd entries of that pmd page, - * except for the switcher pmd. */ + /* + * And release the pmd entries of that pmd page, + * except for the switcher pmd. + */ for (k = 0; k < SWITCHER_PMD_INDEX; k++) release_pmd(&pmdpage[k]); #endif @@ -667,10 +767,12 @@ static void release_all_pagetables(struct lguest *lg) } } -/* We also throw away everything when a Guest tells us it's changed a kernel +/* + * We also throw away everything when a Guest tells us it's changed a kernel * mapping. Since kernel mappings are in every page table, it's easiest to * throw them all away. This traps the Guest in amber for a while as - * everything faults back in, but it's rare. */ + * everything faults back in, but it's rare. + */ void guest_pagetable_clear_all(struct lg_cpu *cpu) { release_all_pagetables(cpu->lg); @@ -678,15 +780,19 @@ void guest_pagetable_clear_all(struct lg_cpu *cpu) pin_stack_pages(cpu); } /*:*/ -/*M:009 Since we throw away all mappings when a kernel mapping changes, our + +/*M:009 + * Since we throw away all mappings when a kernel mapping changes, our * performance sucks for guests using highmem. In fact, a guest with * PAGE_OFFSET 0xc0000000 (the default) and more than about 700MB of RAM is * usually slower than a Guest with less memory. * * This, of course, cannot be fixed. It would take some kind of... well, I - * don't know, but the term "puissant code-fu" comes to mind. :*/ + * don't know, but the term "puissant code-fu" comes to mind. +:*/ -/*H:420 This is the routine which actually sets the page table entry for then +/*H:420 + * This is the routine which actually sets the page table entry for then * "idx"'th shadow page table. * * Normally, we can just throw out the old entry and replace it with 0: if they @@ -715,31 +821,36 @@ static void do_set_pte(struct lg_cpu *cpu, int idx, spmd = spmd_addr(cpu, *spgd, vaddr); if (pmd_flags(*spmd) & _PAGE_PRESENT) { #endif - /* Otherwise, we start by releasing - * the existing entry. */ + /* Otherwise, start by releasing the existing entry. */ pte_t *spte = spte_addr(cpu, *spgd, vaddr); release_pte(*spte); - /* If they're setting this entry as dirty or accessed, - * we might as well put that entry they've given us - * in now. This shaves 10% off a - * copy-on-write micro-benchmark. */ + /* + * If they're setting this entry as dirty or accessed, + * we might as well put that entry they've given us in + * now. This shaves 10% off a copy-on-write + * micro-benchmark. + */ if (pte_flags(gpte) & (_PAGE_DIRTY | _PAGE_ACCESSED)) { check_gpte(cpu, gpte); native_set_pte(spte, gpte_to_spte(cpu, gpte, pte_flags(gpte) & _PAGE_DIRTY)); - } else - /* Otherwise kill it and we can demand_page() - * it in later. */ + } else { + /* + * Otherwise kill it and we can demand_page() + * it in later. + */ native_set_pte(spte, __pte(0)); + } #ifdef CONFIG_X86_PAE } #endif } } -/*H:410 Updating a PTE entry is a little trickier. +/*H:410 + * Updating a PTE entry is a little trickier. * * We keep track of several different page tables (the Guest uses one for each * process, so it makes sense to cache at least a few). Each of these have @@ -748,12 +859,15 @@ static void do_set_pte(struct lg_cpu *cpu, int idx, * all the page tables, not just the current one. This is rare. * * The benefit is that when we have to track a new page table, we can keep all - * the kernel mappings. This speeds up context switch immensely. */ + * the kernel mappings. This speeds up context switch immensely. + */ void guest_set_pte(struct lg_cpu *cpu, unsigned long gpgdir, unsigned long vaddr, pte_t gpte) { - /* Kernel mappings must be changed on all top levels. Slow, but doesn't - * happen often. */ + /* + * Kernel mappings must be changed on all top levels. Slow, but doesn't + * happen often. + */ if (vaddr >= cpu->lg->kernel_address) { unsigned int i; for (i = 0; i < ARRAY_SIZE(cpu->lg->pgdirs); i++) @@ -795,19 +909,25 @@ void guest_set_pgd(struct lguest *lg, unsigned long gpgdir, u32 idx) /* ... throw it away. */ release_pgd(lg->pgdirs[pgdir].pgdir + idx); } + #ifdef CONFIG_X86_PAE +/* For setting a mid-level, we just throw everything away. It's easy. */ void guest_set_pmd(struct lguest *lg, unsigned long pmdp, u32 idx) { guest_pagetable_clear_all(&lg->cpus[0]); } #endif -/* Once we know how much memory we have we can construct simple identity - * (which set virtual == physical) and linear mappings - * which will get the Guest far enough into the boot to create its own. +/*H:505 + * To get through boot, we construct simple identity page mappings (which + * set virtual == physical) and linear mappings which will get the Guest far + * enough into the boot to create its own. The linear mapping means we + * simplify the Guest boot, but it makes assumptions about their PAGE_OFFSET, + * as you'll see. * * We lay them out of the way, just below the initrd (which is why we need to - * know its size here). */ + * know its size here). + */ static unsigned long setup_pagetables(struct lguest *lg, unsigned long mem, unsigned long initrd_size) @@ -825,8 +945,10 @@ static unsigned long setup_pagetables(struct lguest *lg, unsigned int phys_linear; #endif - /* We have mapped_pages frames to map, so we need - * linear_pages page tables to map them. */ + /* + * We have mapped_pages frames to map, so we need linear_pages page + * tables to map them. + */ mapped_pages = mem / PAGE_SIZE; linear_pages = (mapped_pages + PTRS_PER_PTE - 1) / PTRS_PER_PTE; @@ -837,10 +959,16 @@ static unsigned long setup_pagetables(struct lguest *lg, linear = (void *)pgdir - linear_pages * PAGE_SIZE; #ifdef CONFIG_X86_PAE + /* + * And the single mid page goes below that. We only use one, but + * that's enough to map 1G, which definitely gets us through boot. + */ pmds = (void *)linear - PAGE_SIZE; #endif - /* Linear mapping is easy: put every page's address into the - * mapping in order. */ + /* + * Linear mapping is easy: put every page's address into the + * mapping in order. + */ for (i = 0; i < mapped_pages; i++) { pte_t pte; pte = pfn_pte(i, __pgprot(_PAGE_PRESENT|_PAGE_RW|_PAGE_USER)); @@ -848,11 +976,14 @@ static unsigned long setup_pagetables(struct lguest *lg, return -EFAULT; } - /* The top level points to the linear page table pages above. - * We setup the identity and linear mappings here. */ #ifdef CONFIG_X86_PAE + /* + * Make the Guest PMD entries point to the corresponding place in the + * linear mapping (up to one page worth of PMD). + */ for (i = j = 0; i < mapped_pages && j < PTRS_PER_PMD; i += PTRS_PER_PTE, j++) { + /* FIXME: native_set_pmd is overkill here. */ native_set_pmd(&pmd, __pmd(((unsigned long)(linear + i) - mem_base) | _PAGE_PRESENT | _PAGE_RW | _PAGE_USER)); @@ -860,18 +991,36 @@ static unsigned long setup_pagetables(struct lguest *lg, return -EFAULT; } + /* One PGD entry, pointing to that PMD page. */ set_pgd(&pgd, __pgd(((u32)pmds - mem_base) | _PAGE_PRESENT)); + /* Copy it in as the first PGD entry (ie. addresses 0-1G). */ if (copy_to_user(&pgdir[0], &pgd, sizeof(pgd)) != 0) return -EFAULT; + /* + * And the third PGD entry (ie. addresses 3G-4G). + * + * FIXME: This assumes that PAGE_OFFSET for the Guest is 0xC0000000. + */ if (copy_to_user(&pgdir[3], &pgd, sizeof(pgd)) != 0) return -EFAULT; #else + /* + * The top level points to the linear page table pages above. + * We setup the identity and linear mappings here. + */ phys_linear = (unsigned long)linear - mem_base; for (i = 0; i < mapped_pages; i += PTRS_PER_PTE) { pgd_t pgd; + /* + * Create a PGD entry which points to the right part of the + * linear PTE pages. + */ pgd = __pgd((phys_linear + i * sizeof(pte_t)) | (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER)); + /* + * Copy it into the PGD page at 0 and PAGE_OFFSET. + */ if (copy_to_user(&pgdir[i / PTRS_PER_PTE], &pgd, sizeof(pgd)) || copy_to_user(&pgdir[pgd_index(PAGE_OFFSET) + i / PTRS_PER_PTE], @@ -880,15 +1029,19 @@ static unsigned long setup_pagetables(struct lguest *lg, } #endif - /* We return the top level (guest-physical) address: remember where - * this is. */ + /* + * We return the top level (guest-physical) address: we remember where + * this is to write it into lguest_data when the Guest initializes. + */ return (unsigned long)pgdir - mem_base; } -/*H:500 (vii) Setting up the page tables initially. +/*H:500 + * (vii) Setting up the page tables initially. * * When a Guest is first created, the Launcher tells us where the toplevel of - * its first page table is. We set some things up here: */ + * its first page table is. We set some things up here: + */ int init_guest_pagetable(struct lguest *lg) { u64 mem; @@ -898,21 +1051,27 @@ int init_guest_pagetable(struct lguest *lg) pgd_t *pgd; pmd_t *pmd_table; #endif - /* Get the Guest memory size and the ramdisk size from the boot header - * located at lg->mem_base (Guest address 0). */ + /* + * Get the Guest memory size and the ramdisk size from the boot header + * located at lg->mem_base (Guest address 0). + */ if (copy_from_user(&mem, &boot->e820_map[0].size, sizeof(mem)) || get_user(initrd_size, &boot->hdr.ramdisk_size)) return -EFAULT; - /* We start on the first shadow page table, and give it a blank PGD - * page. */ + /* + * We start on the first shadow page table, and give it a blank PGD + * page. + */ lg->pgdirs[0].gpgdir = setup_pagetables(lg, mem, initrd_size); if (IS_ERR_VALUE(lg->pgdirs[0].gpgdir)) return lg->pgdirs[0].gpgdir; lg->pgdirs[0].pgdir = (pgd_t *)get_zeroed_page(GFP_KERNEL); if (!lg->pgdirs[0].pgdir) return -ENOMEM; + #ifdef CONFIG_X86_PAE + /* For PAE, we also create the initial mid-level. */ pgd = lg->pgdirs[0].pgdir; pmd_table = (pmd_t *) get_zeroed_page(GFP_KERNEL); if (!pmd_table) @@ -921,27 +1080,33 @@ int init_guest_pagetable(struct lguest *lg) set_pgd(pgd + SWITCHER_PGD_INDEX, __pgd(__pa(pmd_table) | _PAGE_PRESENT)); #endif + + /* This is the current page table. */ lg->cpus[0].cpu_pgd = 0; return 0; } -/* When the Guest calls LHCALL_LGUEST_INIT we do more setup. */ +/*H:508 When the Guest calls LHCALL_LGUEST_INIT we do more setup. */ void page_table_guest_data_init(struct lg_cpu *cpu) { /* We get the kernel address: above this is all kernel memory. */ if (get_user(cpu->lg->kernel_address, &cpu->lg->lguest_data->kernel_address) - /* We tell the Guest that it can't use the top 2 or 4 MB - * of virtual addresses used by the Switcher. */ + /* + * We tell the Guest that it can't use the top 2 or 4 MB + * of virtual addresses used by the Switcher. + */ || put_user(RESERVE_MEM * 1024 * 1024, &cpu->lg->lguest_data->reserve_mem) || put_user(cpu->lg->pgdirs[0].gpgdir, &cpu->lg->lguest_data->pgdir)) kill_guest(cpu, "bad guest page %p", cpu->lg->lguest_data); - /* In flush_user_mappings() we loop from 0 to + /* + * In flush_user_mappings() we loop from 0 to * "pgd_index(lg->kernel_address)". This assumes it won't hit the - * Switcher mappings, so check that now. */ + * Switcher mappings, so check that now. + */ #ifdef CONFIG_X86_PAE if (pgd_index(cpu->lg->kernel_address) == SWITCHER_PGD_INDEX && pmd_index(cpu->lg->kernel_address) == SWITCHER_PMD_INDEX) @@ -964,12 +1129,14 @@ void free_guest_pagetable(struct lguest *lg) free_page((long)lg->pgdirs[i].pgdir); } -/*H:480 (vi) Mapping the Switcher when the Guest is about to run. +/*H:480 + * (vi) Mapping the Switcher when the Guest is about to run. * * The Switcher and the two pages for this CPU need to be visible in the * Guest (and not the pages for other CPUs). We have the appropriate PTE pages * for each CPU already set up, we just need to hook them in now we know which - * Guest is about to run on this CPU. */ + * Guest is about to run on this CPU. + */ void map_switcher_in_guest(struct lg_cpu *cpu, struct lguest_pages *pages) { pte_t *switcher_pte_page = __get_cpu_var(switcher_pte_pages); @@ -980,30 +1147,38 @@ void map_switcher_in_guest(struct lg_cpu *cpu, struct lguest_pages *pages) pmd_t switcher_pmd; pmd_t *pmd_table; + /* FIXME: native_set_pmd is overkill here. */ native_set_pmd(&switcher_pmd, pfn_pmd(__pa(switcher_pte_page) >> PAGE_SHIFT, PAGE_KERNEL_EXEC)); + /* Figure out where the pmd page is, by reading the PGD, and converting + * it to a virtual address. */ pmd_table = __va(pgd_pfn(cpu->lg-> pgdirs[cpu->cpu_pgd].pgdir[SWITCHER_PGD_INDEX]) << PAGE_SHIFT); + /* Now write it into the shadow page table. */ native_set_pmd(&pmd_table[SWITCHER_PMD_INDEX], switcher_pmd); #else pgd_t switcher_pgd; - /* Make the last PGD entry for this Guest point to the Switcher's PTE - * page for this CPU (with appropriate flags). */ + /* + * Make the last PGD entry for this Guest point to the Switcher's PTE + * page for this CPU (with appropriate flags). + */ switcher_pgd = __pgd(__pa(switcher_pte_page) | __PAGE_KERNEL_EXEC); cpu->lg->pgdirs[cpu->cpu_pgd].pgdir[SWITCHER_PGD_INDEX] = switcher_pgd; #endif - /* We also change the Switcher PTE page. When we're running the Guest, + /* + * We also change the Switcher PTE page. When we're running the Guest, * we want the Guest's "regs" page to appear where the first Switcher * page for this CPU is. This is an optimization: when the Switcher * saves the Guest registers, it saves them into the first page of this * CPU's "struct lguest_pages": if we make sure the Guest's register * page is already mapped there, we don't have to copy them out - * again. */ + * again. + */ pfn = __pa(cpu->regs_page) >> PAGE_SHIFT; native_set_pte(®s_pte, pfn_pte(pfn, PAGE_KERNEL)); native_set_pte(&switcher_pte_page[pte_index((unsigned long)pages)], @@ -1019,10 +1194,12 @@ static void free_switcher_pte_pages(void) free_page((long)switcher_pte_page(i)); } -/*H:520 Setting up the Switcher PTE page for given CPU is fairly easy, given +/*H:520 + * Setting up the Switcher PTE page for given CPU is fairly easy, given * the CPU number and the "struct page"s for the Switcher code itself. * - * Currently the Switcher is less than a page long, so "pages" is always 1. */ + * Currently the Switcher is less than a page long, so "pages" is always 1. + */ static __init void populate_switcher_pte_page(unsigned int cpu, struct page *switcher_page[], unsigned int pages) @@ -1043,13 +1220,16 @@ static __init void populate_switcher_pte_page(unsigned int cpu, native_set_pte(&pte[i], pfn_pte(page_to_pfn(switcher_page[i]), __pgprot(_PAGE_PRESENT|_PAGE_ACCESSED|_PAGE_RW))); - /* The second page contains the "struct lguest_ro_state", and is - * read-only. */ + /* + * The second page contains the "struct lguest_ro_state", and is + * read-only. + */ native_set_pte(&pte[i+1], pfn_pte(page_to_pfn(switcher_page[i+1]), __pgprot(_PAGE_PRESENT|_PAGE_ACCESSED))); } -/* We've made it through the page table code. Perhaps our tired brains are +/* + * We've made it through the page table code. Perhaps our tired brains are * still processing the details, or perhaps we're simply glad it's over. * * If nothing else, note that all this complexity in juggling shadow page tables @@ -1058,10 +1238,13 @@ static __init void populate_switcher_pte_page(unsigned int cpu, * uses exotic direct Guest pagetable manipulation, and why both Intel and AMD * have implemented shadow page table support directly into hardware. * - * There is just one file remaining in the Host. */ + * There is just one file remaining in the Host. + */ -/*H:510 At boot or module load time, init_pagetables() allocates and populates - * the Switcher PTE page for each CPU. */ +/*H:510 + * At boot or module load time, init_pagetables() allocates and populates + * the Switcher PTE page for each CPU. + */ __init int init_pagetables(struct page **switcher_page, unsigned int pages) { unsigned int i; diff --git a/drivers/lguest/segments.c b/drivers/lguest/segments.c index 482ed5a1875..951c57b0a7e 100644 --- a/drivers/lguest/segments.c +++ b/drivers/lguest/segments.c @@ -1,4 +1,5 @@ -/*P:600 The x86 architecture has segments, which involve a table of descriptors +/*P:600 + * The x86 architecture has segments, which involve a table of descriptors * which can be used to do funky things with virtual address interpretation. * We originally used to use segments so the Guest couldn't alter the * Guest<->Host Switcher, and then we had to trim Guest segments, and restore @@ -8,7 +9,8 @@ * * In these modern times, the segment handling code consists of simple sanity * checks, and the worst you'll experience reading this code is butterfly-rash - * from frolicking through its parklike serenity. :*/ + * from frolicking through its parklike serenity. +:*/ #include "lg.h" /*H:600 @@ -41,10 +43,12 @@ * begin. */ -/* There are several entries we don't let the Guest set. The TSS entry is the +/* + * There are several entries we don't let the Guest set. The TSS entry is the * "Task State Segment" which controls all kinds of delicate things. The * LGUEST_CS and LGUEST_DS entries are reserved for the Switcher, and the - * the Guest can't be trusted to deal with double faults. */ + * the Guest can't be trusted to deal with double faults. + */ static bool ignored_gdt(unsigned int num) { return (num == GDT_ENTRY_TSS @@ -53,42 +57,52 @@ static bool ignored_gdt(unsigned int num) || num == GDT_ENTRY_DOUBLEFAULT_TSS); } -/*H:630 Once the Guest gave us new GDT entries, we fix them up a little. We +/*H:630 + * Once the Guest gave us new GDT entries, we fix them up a little. We * don't care if they're invalid: the worst that can happen is a General * Protection Fault in the Switcher when it restores a Guest segment register * which tries to use that entry. Then we kill the Guest for causing such a - * mess: the message will be "unhandled trap 256". */ + * mess: the message will be "unhandled trap 256". + */ static void fixup_gdt_table(struct lg_cpu *cpu, unsigned start, unsigned end) { unsigned int i; for (i = start; i < end; i++) { - /* We never copy these ones to real GDT, so we don't care what - * they say */ + /* + * We never copy these ones to real GDT, so we don't care what + * they say + */ if (ignored_gdt(i)) continue; - /* Segment descriptors contain a privilege level: the Guest is + /* + * Segment descriptors contain a privilege level: the Guest is * sometimes careless and leaves this as 0, even though it's - * running at privilege level 1. If so, we fix it here. */ + * running at privilege level 1. If so, we fix it here. + */ if ((cpu->arch.gdt[i].b & 0x00006000) == 0) cpu->arch.gdt[i].b |= (GUEST_PL << 13); - /* Each descriptor has an "accessed" bit. If we don't set it + /* + * Each descriptor has an "accessed" bit. If we don't set it * now, the CPU will try to set it when the Guest first loads * that entry into a segment register. But the GDT isn't - * writable by the Guest, so bad things can happen. */ + * writable by the Guest, so bad things can happen. + */ cpu->arch.gdt[i].b |= 0x00000100; } } -/*H:610 Like the IDT, we never simply use the GDT the Guest gives us. We keep +/*H:610 + * Like the IDT, we never simply use the GDT the Guest gives us. We keep * a GDT for each CPU, and copy across the Guest's entries each time we want to * run the Guest on that CPU. * * This routine is called at boot or modprobe time for each CPU to set up the * constant GDT entries: the ones which are the same no matter what Guest we're - * running. */ + * running. + */ void setup_default_gdt_entries(struct lguest_ro_state *state) { struct desc_struct *gdt = state->guest_gdt; @@ -98,30 +112,37 @@ void setup_default_gdt_entries(struct lguest_ro_state *state) gdt[GDT_ENTRY_LGUEST_CS] = FULL_EXEC_SEGMENT; gdt[GDT_ENTRY_LGUEST_DS] = FULL_SEGMENT; - /* The TSS segment refers to the TSS entry for this particular CPU. + /* + * The TSS segment refers to the TSS entry for this particular CPU. * Forgive the magic flags: the 0x8900 means the entry is Present, it's * privilege level 0 Available 386 TSS system segment, and the 0x67 - * means Saturn is eclipsed by Mercury in the twelfth house. */ + * means Saturn is eclipsed by Mercury in the twelfth house. + */ gdt[GDT_ENTRY_TSS].a = 0x00000067 | (tss << 16); gdt[GDT_ENTRY_TSS].b = 0x00008900 | (tss & 0xFF000000) | ((tss >> 16) & 0x000000FF); } -/* This routine sets up the initial Guest GDT for booting. All entries start - * as 0 (unusable). */ +/* + * This routine sets up the initial Guest GDT for booting. All entries start + * as 0 (unusable). + */ void setup_guest_gdt(struct lg_cpu *cpu) { - /* Start with full 0-4G segments... */ + /* + * Start with full 0-4G segments...except the Guest is allowed to use + * them, so set the privilege level appropriately in the flags. + */ cpu->arch.gdt[GDT_ENTRY_KERNEL_CS] = FULL_EXEC_SEGMENT; cpu->arch.gdt[GDT_ENTRY_KERNEL_DS] = FULL_SEGMENT; - /* ...except the Guest is allowed to use them, so set the privilege - * level appropriately in the flags. */ cpu->arch.gdt[GDT_ENTRY_KERNEL_CS].b |= (GUEST_PL << 13); cpu->arch.gdt[GDT_ENTRY_KERNEL_DS].b |= (GUEST_PL << 13); } -/*H:650 An optimization of copy_gdt(), for just the three "thead-local storage" - * entries. */ +/*H:650 + * An optimization of copy_gdt(), for just the three "thead-local storage" + * entries. + */ void copy_gdt_tls(const struct lg_cpu *cpu, struct desc_struct *gdt) { unsigned int i; @@ -130,26 +151,34 @@ void copy_gdt_tls(const struct lg_cpu *cpu, struct desc_struct *gdt) gdt[i] = cpu->arch.gdt[i]; } -/*H:640 When the Guest is run on a different CPU, or the GDT entries have - * changed, copy_gdt() is called to copy the Guest's GDT entries across to this - * CPU's GDT. */ +/*H:640 + * When the Guest is run on a different CPU, or the GDT entries have changed, + * copy_gdt() is called to copy the Guest's GDT entries across to this CPU's + * GDT. + */ void copy_gdt(const struct lg_cpu *cpu, struct desc_struct *gdt) { unsigned int i; - /* The default entries from setup_default_gdt_entries() are not - * replaced. See ignored_gdt() above. */ + /* + * The default entries from setup_default_gdt_entries() are not + * replaced. See ignored_gdt() above. + */ for (i = 0; i < GDT_ENTRIES; i++) if (!ignored_gdt(i)) gdt[i] = cpu->arch.gdt[i]; } -/*H:620 This is where the Guest asks us to load a new GDT entry - * (LHCALL_LOAD_GDT_ENTRY). We tweak the entry and copy it in. */ +/*H:620 + * This is where the Guest asks us to load a new GDT entry + * (LHCALL_LOAD_GDT_ENTRY). We tweak the entry and copy it in. + */ void load_guest_gdt_entry(struct lg_cpu *cpu, u32 num, u32 lo, u32 hi) { - /* We assume the Guest has the same number of GDT entries as the - * Host, otherwise we'd have to dynamically allocate the Guest GDT. */ + /* + * We assume the Guest has the same number of GDT entries as the + * Host, otherwise we'd have to dynamically allocate the Guest GDT. + */ if (num >= ARRAY_SIZE(cpu->arch.gdt)) kill_guest(cpu, "too many gdt entries %i", num); @@ -157,15 +186,19 @@ void load_guest_gdt_entry(struct lg_cpu *cpu, u32 num, u32 lo, u32 hi) cpu->arch.gdt[num].a = lo; cpu->arch.gdt[num].b = hi; fixup_gdt_table(cpu, num, num+1); - /* Mark that the GDT changed so the core knows it has to copy it again, - * even if the Guest is run on the same CPU. */ + /* + * Mark that the GDT changed so the core knows it has to copy it again, + * even if the Guest is run on the same CPU. + */ cpu->changed |= CHANGED_GDT; } -/* This is the fast-track version for just changing the three TLS entries. +/* + * This is the fast-track version for just changing the three TLS entries. * Remember that this happens on every context switch, so it's worth * optimizing. But wouldn't it be neater to have a single hypercall to cover - * both cases? */ + * both cases? + */ void guest_load_tls(struct lg_cpu *cpu, unsigned long gtls) { struct desc_struct *tls = &cpu->arch.gdt[GDT_ENTRY_TLS_MIN]; @@ -175,7 +208,6 @@ void guest_load_tls(struct lg_cpu *cpu, unsigned long gtls) /* Note that just the TLS entries have changed. */ cpu->changed |= CHANGED_GDT_TLS; } -/*:*/ /*H:660 * With this, we have finished the Host. diff --git a/drivers/lguest/x86/core.c b/drivers/lguest/x86/core.c index eaf722fe309..6ae388849a3 100644 --- a/drivers/lguest/x86/core.c +++ b/drivers/lguest/x86/core.c @@ -17,13 +17,15 @@ * along with this program; if not, write to the Free Software * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ -/*P:450 This file contains the x86-specific lguest code. It used to be all +/*P:450 + * This file contains the x86-specific lguest code. It used to be all * mixed in with drivers/lguest/core.c but several foolhardy code slashers * wrestled most of the dependencies out to here in preparation for porting * lguest to other architectures (see what I mean by foolhardy?). * * This also contains a couple of non-obvious setup and teardown pieces which - * were implemented after days of debugging pain. :*/ + * were implemented after days of debugging pain. +:*/ #include <linux/kernel.h> #include <linux/start_kernel.h> #include <linux/string.h> @@ -82,25 +84,33 @@ static DEFINE_PER_CPU(struct lg_cpu *, last_cpu); */ static void copy_in_guest_info(struct lg_cpu *cpu, struct lguest_pages *pages) { - /* Copying all this data can be quite expensive. We usually run the + /* + * Copying all this data can be quite expensive. We usually run the * same Guest we ran last time (and that Guest hasn't run anywhere else * meanwhile). If that's not the case, we pretend everything in the - * Guest has changed. */ + * Guest has changed. + */ if (__get_cpu_var(last_cpu) != cpu || cpu->last_pages != pages) { __get_cpu_var(last_cpu) = cpu; cpu->last_pages = pages; cpu->changed = CHANGED_ALL; } - /* These copies are pretty cheap, so we do them unconditionally: */ - /* Save the current Host top-level page directory. */ + /* + * These copies are pretty cheap, so we do them unconditionally: */ + /* Save the current Host top-level page directory. + */ pages->state.host_cr3 = __pa(current->mm->pgd); - /* Set up the Guest's page tables to see this CPU's pages (and no - * other CPU's pages). */ + /* + * Set up the Guest's page tables to see this CPU's pages (and no + * other CPU's pages). + */ map_switcher_in_guest(cpu, pages); - /* Set up the two "TSS" members which tell the CPU what stack to use + /* + * Set up the two "TSS" members which tell the CPU what stack to use * for traps which do directly into the Guest (ie. traps at privilege - * level 1). */ + * level 1). + */ pages->state.guest_tss.sp1 = cpu->esp1; pages->state.guest_tss.ss1 = cpu->ss1; @@ -125,97 +135,126 @@ static void run_guest_once(struct lg_cpu *cpu, struct lguest_pages *pages) /* This is a dummy value we need for GCC's sake. */ unsigned int clobber; - /* Copy the guest-specific information into this CPU's "struct - * lguest_pages". */ + /* + * Copy the guest-specific information into this CPU's "struct + * lguest_pages". + */ copy_in_guest_info(cpu, pages); - /* Set the trap number to 256 (impossible value). If we fault while + /* + * Set the trap number to 256 (impossible value). If we fault while * switching to the Guest (bad segment registers or bug), this will - * cause us to abort the Guest. */ + * cause us to abort the Guest. + */ cpu->regs->trapnum = 256; - /* Now: we push the "eflags" register on the stack, then do an "lcall". + /* + * Now: we push the "eflags" register on the stack, then do an "lcall". * This is how we change from using the kernel code segment to using * the dedicated lguest code segment, as well as jumping into the * Switcher. * * The lcall also pushes the old code segment (KERNEL_CS) onto the * stack, then the address of this call. This stack layout happens to - * exactly match the stack layout created by an interrupt... */ + * exactly match the stack layout created by an interrupt... + */ asm volatile("pushf; lcall *lguest_entry" - /* This is how we tell GCC that %eax ("a") and %ebx ("b") - * are changed by this routine. The "=" means output. */ + /* + * This is how we tell GCC that %eax ("a") and %ebx ("b") + * are changed by this routine. The "=" means output. + */ : "=a"(clobber), "=b"(clobber) - /* %eax contains the pages pointer. ("0" refers to the + /* + * %eax contains the pages pointer. ("0" refers to the * 0-th argument above, ie "a"). %ebx contains the * physical address of the Guest's top-level page - * directory. */ + * directory. + */ : "0"(pages), "1"(__pa(cpu->lg->pgdirs[cpu->cpu_pgd].pgdir)) - /* We tell gcc that all these registers could change, + /* + * We tell gcc that all these registers could change, * which means we don't have to save and restore them in - * the Switcher. */ + * the Switcher. + */ : "memory", "%edx", "%ecx", "%edi", "%esi"); } /*:*/ -/*M:002 There are hooks in the scheduler which we can register to tell when we +/*M:002 + * There are hooks in the scheduler which we can register to tell when we * get kicked off the CPU (preempt_notifier_register()). This would allow us * to lazily disable SYSENTER which would regain some performance, and should * also simplify copy_in_guest_info(). Note that we'd still need to restore * things when we exit to Launcher userspace, but that's fairly easy. * - * We could also try using this hooks for PGE, but that might be too expensive. + * We could also try using these hooks for PGE, but that might be too expensive. * - * The hooks were designed for KVM, but we can also put them to good use. :*/ + * The hooks were designed for KVM, but we can also put them to good use. +:*/ -/*H:040 This is the i386-specific code to setup and run the Guest. Interrupts - * are disabled: we own the CPU. */ +/*H:040 + * This is the i386-specific code to setup and run the Guest. Interrupts + * are disabled: we own the CPU. + */ void lguest_arch_run_guest(struct lg_cpu *cpu) { - /* Remember the awfully-named TS bit? If the Guest has asked to set it + /* + * Remember the awfully-named TS bit? If the Guest has asked to set it * we set it now, so we can trap and pass that trap to the Guest if it - * uses the FPU. */ + * uses the FPU. + */ if (cpu->ts) unlazy_fpu(current); - /* SYSENTER is an optimized way of doing system calls. We can't allow + /* + * SYSENTER is an optimized way of doing system calls. We can't allow * it because it always jumps to privilege level 0. A normal Guest * won't try it because we don't advertise it in CPUID, but a malicious * Guest (or malicious Guest userspace program) could, so we tell the - * CPU to disable it before running the Guest. */ + * CPU to disable it before running the Guest. + */ if (boot_cpu_has(X86_FEATURE_SEP)) wrmsr(MSR_IA32_SYSENTER_CS, 0, 0); - /* Now we actually run the Guest. It will return when something + /* + * Now we actually run the Guest. It will return when something * interesting happens, and we can examine its registers to see what it - * was doing. */ + * was doing. + */ run_guest_once(cpu, lguest_pages(raw_smp_processor_id())); - /* Note that the "regs" structure contains two extra entries which are + /* + * Note that the "regs" structure contains two extra entries which are * not really registers: a trap number which says what interrupt or * trap made the switcher code come back, and an error code which some - * traps set. */ + * traps set. + */ /* Restore SYSENTER if it's supposed to be on. */ if (boot_cpu_has(X86_FEATURE_SEP)) wrmsr(MSR_IA32_SYSENTER_CS, __KERNEL_CS, 0); - /* If the Guest page faulted, then the cr2 register will tell us the + /* + * If the Guest page faulted, then the cr2 register will tell us the * bad virtual address. We have to grab this now, because once we * re-enable interrupts an interrupt could fault and thus overwrite - * cr2, or we could even move off to a different CPU. */ + * cr2, or we could even move off to a different CPU. + */ if (cpu->regs->trapnum == 14) cpu->arch.last_pagefault = read_cr2(); - /* Similarly, if we took a trap because the Guest used the FPU, + /* + * Similarly, if we took a trap because the Guest used the FPU, * we have to restore the FPU it expects to see. * math_state_restore() may sleep and we may even move off to * a different CPU. So all the critical stuff should be done - * before this. */ + * before this. + */ else if (cpu->regs->trapnum == 7) math_state_restore(); } -/*H:130 Now we've examined the hypercall code; our Guest can make requests. +/*H:130 + * Now we've examined the hypercall code; our Guest can make requests. * Our Guest is usually so well behaved; it never tries to do things it isn't * allowed to, and uses hypercalls instead. Unfortunately, Linux's paravirtual * infrastructure isn't quite complete, because it doesn't contain replacements @@ -225,26 +264,33 @@ void lguest_arch_run_guest(struct lg_cpu *cpu) * * When the Guest uses one of these instructions, we get a trap (General * Protection Fault) and come here. We see if it's one of those troublesome - * instructions and skip over it. We return true if we did. */ + * instructions and skip over it. We return true if we did. + */ static int emulate_insn(struct lg_cpu *cpu) { u8 insn; unsigned int insnlen = 0, in = 0, shift = 0; - /* The eip contains the *virtual* address of the Guest's instruction: - * guest_pa just subtracts the Guest's page_offset. */ + /* + * The eip contains the *virtual* address of the Guest's instruction: + * guest_pa just subtracts the Guest's page_offset. + */ unsigned long physaddr = guest_pa(cpu, cpu->regs->eip); - /* This must be the Guest kernel trying to do something, not userspace! + /* + * This must be the Guest kernel trying to do something, not userspace! * The bottom two bits of the CS segment register are the privilege - * level. */ + * level. + */ if ((cpu->regs->cs & 3) != GUEST_PL) return 0; /* Decoding x86 instructions is icky. */ insn = lgread(cpu, physaddr, u8); - /* 0x66 is an "operand prefix". It means it's using the upper 16 bits - of the eax register. */ + /* + * 0x66 is an "operand prefix". It means it's using the upper 16 bits + * of the eax register. + */ if (insn == 0x66) { shift = 16; /* The instruction is 1 byte so far, read the next byte. */ @@ -252,8 +298,10 @@ static int emulate_insn(struct lg_cpu *cpu) insn = lgread(cpu, physaddr + insnlen, u8); } - /* We can ignore the lower bit for the moment and decode the 4 opcodes - * we need to emulate. */ + /* + * We can ignore the lower bit for the moment and decode the 4 opcodes + * we need to emulate. + */ switch (insn & 0xFE) { case 0xE4: /* in <next byte>,%al */ insnlen += 2; @@ -274,9 +322,11 @@ static int emulate_insn(struct lg_cpu *cpu) return 0; } - /* If it was an "IN" instruction, they expect the result to be read + /* + * If it was an "IN" instruction, they expect the result to be read * into %eax, so we change %eax. We always return all-ones, which - * traditionally means "there's nothing there". */ + * traditionally means "there's nothing there". + */ if (in) { /* Lower bit tells is whether it's a 16 or 32 bit access */ if (insn & 0x1) @@ -290,7 +340,8 @@ static int emulate_insn(struct lg_cpu *cpu) return 1; } -/* Our hypercalls mechanism used to be based on direct software interrupts. +/* + * Our hypercalls mechanism used to be based on direct software interrupts. * After Anthony's "Refactor hypercall infrastructure" kvm patch, we decided to * change over to using kvm hypercalls. * @@ -318,16 +369,20 @@ static int emulate_insn(struct lg_cpu *cpu) */ static void rewrite_hypercall(struct lg_cpu *cpu) { - /* This are the opcodes we use to patch the Guest. The opcode for "int + /* + * This are the opcodes we use to patch the Guest. The opcode for "int * $0x1f" is "0xcd 0x1f" but vmcall instruction is 3 bytes long, so we - * complete the sequence with a NOP (0x90). */ + * complete the sequence with a NOP (0x90). + */ u8 insn[3] = {0xcd, 0x1f, 0x90}; __lgwrite(cpu, guest_pa(cpu, cpu->regs->eip), insn, sizeof(insn)); - /* The above write might have caused a copy of that page to be made + /* + * The above write might have caused a copy of that page to be made * (if it was read-only). We need to make sure the Guest has * up-to-date pagetables. As this doesn't happen often, we can just - * drop them all. */ + * drop them all. + */ guest_pagetable_clear_all(cpu); } @@ -335,9 +390,11 @@ static bool is_hypercall(struct lg_cpu *cpu) { u8 insn[3]; - /* This must be the Guest kernel trying to do something. + /* + * This must be the Guest kernel trying to do something. * The bottom two bits of the CS segment register are the privilege - * level. */ + * level. + */ if ((cpu->regs->cs & 3) != GUEST_PL) return false; @@ -351,86 +408,105 @@ void lguest_arch_handle_trap(struct lg_cpu *cpu) { switch (cpu->regs->trapnum) { case 13: /* We've intercepted a General Protection Fault. */ - /* Check if this was one of those annoying IN or OUT + /* + * Check if this was one of those annoying IN or OUT * instructions which we need to emulate. If so, we just go - * back into the Guest after we've done it. */ + * back into the Guest after we've done it. + */ if (cpu->regs->errcode == 0) { if (emulate_insn(cpu)) return; } - /* If KVM is active, the vmcall instruction triggers a - * General Protection Fault. Normally it triggers an - * invalid opcode fault (6): */ + /* + * If KVM is active, the vmcall instruction triggers a General + * Protection Fault. Normally it triggers an invalid opcode + * fault (6): + */ case 6: - /* We need to check if ring == GUEST_PL and - * faulting instruction == vmcall. */ + /* + * We need to check if ring == GUEST_PL and faulting + * instruction == vmcall. + */ if (is_hypercall(cpu)) { rewrite_hypercall(cpu); return; } break; case 14: /* We've intercepted a Page Fault. */ - /* The Guest accessed a virtual address that wasn't mapped. + /* + * The Guest accessed a virtual address that wasn't mapped. * This happens a lot: we don't actually set up most of the page * tables for the Guest at all when we start: as it runs it asks * for more and more, and we set them up as required. In this * case, we don't even tell the Guest that the fault happened. * * The errcode tells whether this was a read or a write, and - * whether kernel or userspace code. */ + * whether kernel or userspace code. + */ if (demand_page(cpu, cpu->arch.last_pagefault, cpu->regs->errcode)) return; - /* OK, it's really not there (or not OK): the Guest needs to + /* + * OK, it's really not there (or not OK): the Guest needs to * know. We write out the cr2 value so it knows where the * fault occurred. * * Note that if the Guest were really messed up, this could * happen before it's done the LHCALL_LGUEST_INIT hypercall, so - * lg->lguest_data could be NULL */ + * lg->lguest_data could be NULL + */ if (cpu->lg->lguest_data && put_user(cpu->arch.last_pagefault, &cpu->lg->lguest_data->cr2)) kill_guest(cpu, "Writing cr2"); break; case 7: /* We've intercepted a Device Not Available fault. */ - /* If the Guest doesn't want to know, we already restored the - * Floating Point Unit, so we just continue without telling - * it. */ + /* + * If the Guest doesn't want to know, we already restored the + * Floating Point Unit, so we just continue without telling it. + */ if (!cpu->ts) return; break; case 32 ... 255: - /* These values mean a real interrupt occurred, in which case + /* + * These values mean a real interrupt occurred, in which case * the Host handler has already been run. We just do a * friendly check if another process should now be run, then - * return to run the Guest again */ + * return to run the Guest again + */ cond_resched(); return; case LGUEST_TRAP_ENTRY: - /* Our 'struct hcall_args' maps directly over our regs: we set - * up the pointer now to indicate a hypercall is pending. */ + /* + * Our 'struct hcall_args' maps directly over our regs: we set + * up the pointer now to indicate a hypercall is pending. + */ cpu->hcall = (struct hcall_args *)cpu->regs; return; } /* We didn't handle the trap, so it needs to go to the Guest. */ if (!deliver_trap(cpu, cpu->regs->trapnum)) - /* If the Guest doesn't have a handler (either it hasn't + /* + * If the Guest doesn't have a handler (either it hasn't * registered any yet, or it's one of the faults we don't let - * it handle), it dies with this cryptic error message. */ + * it handle), it dies with this cryptic error message. + */ kill_guest(cpu, "unhandled trap %li at %#lx (%#lx)", cpu->regs->trapnum, cpu->regs->eip, cpu->regs->trapnum == 14 ? cpu->arch.last_pagefault : cpu->regs->errcode); } -/* Now we can look at each of the routines this calls, in increasing order of +/* + * Now we can look at each of the routines this calls, in increasing order of * complexity: do_hypercalls(), emulate_insn(), maybe_do_interrupt(), * deliver_trap() and demand_page(). After all those, we'll be ready to * examine the Switcher, and our philosophical understanding of the Host/Guest - * duality will be complete. :*/ + * duality will be complete. +:*/ static void adjust_pge(void *on) { if (on) @@ -439,13 +515,16 @@ static void adjust_pge(void *on) write_cr4(read_cr4() & ~X86_CR4_PGE); } -/*H:020 Now the Switcher is mapped and every thing else is ready, we need to do - * some more i386-specific initialization. */ +/*H:020 + * Now the Switcher is mapped and every thing else is ready, we need to do + * some more i386-specific initialization. + */ void __init lguest_arch_host_init(void) { int i; - /* Most of the i386/switcher.S doesn't care that it's been moved; on + /* + * Most of the i386/switcher.S doesn't care that it's been moved; on * Intel, jumps are relative, and it doesn't access any references to * external code or data. * @@ -453,7 +532,8 @@ void __init lguest_arch_host_init(void) * addresses are placed in a table (default_idt_entries), so we need to * update the table with the new addresses. switcher_offset() is a * convenience function which returns the distance between the - * compiled-in switcher code and the high-mapped copy we just made. */ + * compiled-in switcher code and the high-mapped copy we just made. + */ for (i = 0; i < IDT_ENTRIES; i++) default_idt_entries[i] += switcher_offset(); @@ -468,63 +548,81 @@ void __init lguest_arch_host_init(void) for_each_possible_cpu(i) { /* lguest_pages() returns this CPU's two pages. */ struct lguest_pages *pages = lguest_pages(i); - /* This is a convenience pointer to make the code fit one - * statement to a line. */ + /* This is a convenience pointer to make the code neater. */ struct lguest_ro_state *state = &pages->state; - /* The Global Descriptor Table: the Host has a different one + /* + * The Global Descriptor Table: the Host has a different one * for each CPU. We keep a descriptor for the GDT which says * where it is and how big it is (the size is actually the last - * byte, not the size, hence the "-1"). */ + * byte, not the size, hence the "-1"). + */ state->host_gdt_desc.size = GDT_SIZE-1; state->host_gdt_desc.address = (long)get_cpu_gdt_table(i); - /* All CPUs on the Host use the same Interrupt Descriptor + /* + * All CPUs on the Host use the same Interrupt Descriptor * Table, so we just use store_idt(), which gets this CPU's IDT - * descriptor. */ + * descriptor. + */ store_idt(&state->host_idt_desc); - /* The descriptors for the Guest's GDT and IDT can be filled + /* + * The descriptors for the Guest's GDT and IDT can be filled * out now, too. We copy the GDT & IDT into ->guest_gdt and - * ->guest_idt before actually running the Guest. */ + * ->guest_idt before actually running the Guest. + */ state->guest_idt_desc.size = sizeof(state->guest_idt)-1; state->guest_idt_desc.address = (long)&state->guest_idt; state->guest_gdt_desc.size = sizeof(state->guest_gdt)-1; state->guest_gdt_desc.address = (long)&state->guest_gdt; - /* We know where we want the stack to be when the Guest enters + /* + * We know where we want the stack to be when the Guest enters * the Switcher: in pages->regs. The stack grows upwards, so - * we start it at the end of that structure. */ + * we start it at the end of that structure. + */ state->guest_tss.sp0 = (long)(&pages->regs + 1); - /* And this is the GDT entry to use for the stack: we keep a - * couple of special LGUEST entries. */ + /* + * And this is the GDT entry to use for the stack: we keep a + * couple of special LGUEST entries. + */ state->guest_tss.ss0 = LGUEST_DS; - /* x86 can have a finegrained bitmap which indicates what I/O + /* + * x86 can have a finegrained bitmap which indicates what I/O * ports the process can use. We set it to the end of our - * structure, meaning "none". */ + * structure, meaning "none". + */ state->guest_tss.io_bitmap_base = sizeof(state->guest_tss); - /* Some GDT entries are the same across all Guests, so we can - * set them up now. */ + /* + * Some GDT entries are the same across all Guests, so we can + * set them up now. + */ setup_default_gdt_entries(state); /* Most IDT entries are the same for all Guests, too.*/ setup_default_idt_entries(state, default_idt_entries); - /* The Host needs to be able to use the LGUEST segments on this - * CPU, too, so put them in the Host GDT. */ + /* + * The Host needs to be able to use the LGUEST segments on this + * CPU, too, so put them in the Host GDT. + */ get_cpu_gdt_table(i)[GDT_ENTRY_LGUEST_CS] = FULL_EXEC_SEGMENT; get_cpu_gdt_table(i)[GDT_ENTRY_LGUEST_DS] = FULL_SEGMENT; } - /* In the Switcher, we want the %cs segment register to use the + /* + * In the Switcher, we want the %cs segment register to use the * LGUEST_CS GDT entry: we've put that in the Host and Guest GDTs, so * it will be undisturbed when we switch. To change %cs and jump we - * need this structure to feed to Intel's "lcall" instruction. */ + * need this structure to feed to Intel's "lcall" instruction. + */ lguest_entry.offset = (long)switch_to_guest + switcher_offset(); lguest_entry.segment = LGUEST_CS; - /* Finally, we need to turn off "Page Global Enable". PGE is an + /* + * Finally, we need to turn off "Page Global Enable". PGE is an * optimization where page table entries are specially marked to show * they never change. The Host kernel marks all the kernel pages this * way because it's always present, even when userspace is running. @@ -534,16 +632,21 @@ void __init lguest_arch_host_init(void) * you'll get really weird bugs that you'll chase for two days. * * I used to turn PGE off every time we switched to the Guest and back - * on when we return, but that slowed the Switcher down noticibly. */ + * on when we return, but that slowed the Switcher down noticibly. + */ - /* We don't need the complexity of CPUs coming and going while we're - * doing this. */ + /* + * We don't need the complexity of CPUs coming and going while we're + * doing this. + */ get_online_cpus(); if (cpu_has_pge) { /* We have a broader idea of "global". */ /* Remember that this was originally set (for cleanup). */ cpu_had_pge = 1; - /* adjust_pge is a helper function which sets or unsets the PGE - * bit on its CPU, depending on the argument (0 == unset). */ + /* + * adjust_pge is a helper function which sets or unsets the PGE + * bit on its CPU, depending on the argument (0 == unset). + */ on_each_cpu(adjust_pge, (void *)0, 1); /* Turn off the feature in the global feature set. */ clear_cpu_cap(&boot_cpu_data, X86_FEATURE_PGE); @@ -590,26 +693,32 @@ int lguest_arch_init_hypercalls(struct lg_cpu *cpu) { u32 tsc_speed; - /* The pointer to the Guest's "struct lguest_data" is the only argument. - * We check that address now. */ + /* + * The pointer to the Guest's "struct lguest_data" is the only argument. + * We check that address now. + */ if (!lguest_address_ok(cpu->lg, cpu->hcall->arg1, sizeof(*cpu->lg->lguest_data))) return -EFAULT; - /* Having checked it, we simply set lg->lguest_data to point straight + /* + * Having checked it, we simply set lg->lguest_data to point straight * into the Launcher's memory at the right place and then use * copy_to_user/from_user from now on, instead of lgread/write. I put * this in to show that I'm not immune to writing stupid - * optimizations. */ + * optimizations. + */ cpu->lg->lguest_data = cpu->lg->mem_base + cpu->hcall->arg1; - /* We insist that the Time Stamp Counter exist and doesn't change with + /* + * We insist that the Time Stamp Counter exist and doesn't change with * cpu frequency. Some devious chip manufacturers decided that TSC * changes could be handled in software. I decided that time going * backwards might be good for benchmarks, but it's bad for users. * * We also insist that the TSC be stable: the kernel detects unreliable - * TSCs for its own purposes, and we use that here. */ + * TSCs for its own purposes, and we use that here. + */ if (boot_cpu_has(X86_FEATURE_CONSTANT_TSC) && !check_tsc_unstable()) tsc_speed = tsc_khz; else @@ -625,38 +734,47 @@ int lguest_arch_init_hypercalls(struct lg_cpu *cpu) } /*:*/ -/*L:030 lguest_arch_setup_regs() +/*L:030 + * lguest_arch_setup_regs() * * Most of the Guest's registers are left alone: we used get_zeroed_page() to - * allocate the structure, so they will be 0. */ + * allocate the structure, so they will be 0. + */ void lguest_arch_setup_regs(struct lg_cpu *cpu, unsigned long start) { struct lguest_regs *regs = cpu->regs; - /* There are four "segment" registers which the Guest needs to boot: + /* + * There are four "segment" registers which the Guest needs to boot: * The "code segment" register (cs) refers to the kernel code segment * __KERNEL_CS, and the "data", "extra" and "stack" segment registers * refer to the kernel data segment __KERNEL_DS. * * The privilege level is packed into the lower bits. The Guest runs - * at privilege level 1 (GUEST_PL).*/ + * at privilege level 1 (GUEST_PL). + */ regs->ds = regs->es = regs->ss = __KERNEL_DS|GUEST_PL; regs->cs = __KERNEL_CS|GUEST_PL; - /* The "eflags" register contains miscellaneous flags. Bit 1 (0x002) + /* + * The "eflags" register contains miscellaneous flags. Bit 1 (0x002) * is supposed to always be "1". Bit 9 (0x200) controls whether * interrupts are enabled. We always leave interrupts enabled while - * running the Guest. */ + * running the Guest. + */ regs->eflags = X86_EFLAGS_IF | 0x2; - /* The "Extended Instruction Pointer" register says where the Guest is - * running. */ + /* + * The "Extended Instruction Pointer" register says where the Guest is + * running. + */ regs->eip = start; - /* %esi points to our boot information, at physical address 0, so don't - * touch it. */ + /* + * %esi points to our boot information, at physical address 0, so don't + * touch it. + */ - /* There are a couple of GDT entries the Guest expects when first - * booting. */ + /* There are a couple of GDT entries the Guest expects at boot. */ setup_guest_gdt(cpu); } diff --git a/drivers/lguest/x86/switcher_32.S b/drivers/lguest/x86/switcher_32.S index 3fc15318a80..40634b0db9f 100644 --- a/drivers/lguest/x86/switcher_32.S +++ b/drivers/lguest/x86/switcher_32.S @@ -1,12 +1,15 @@ -/*P:900 This is the Switcher: code which sits at 0xFFC00000 astride both the - * Host and Guest to do the low-level Guest<->Host switch. It is as simple as - * it can be made, but it's naturally very specific to x86. +/*P:900 + * This is the Switcher: code which sits at 0xFFC00000 (or 0xFFE00000) astride + * both the Host and Guest to do the low-level Guest<->Host switch. It is as + * simple as it can be made, but it's naturally very specific to x86. * * You have now completed Preparation. If this has whet your appetite; if you * are feeling invigorated and refreshed then the next, more challenging stage - * can be found in "make Guest". :*/ + * can be found in "make Guest". + :*/ -/*M:012 Lguest is meant to be simple: my rule of thumb is that 1% more LOC must +/*M:012 + * Lguest is meant to be simple: my rule of thumb is that 1% more LOC must * gain at least 1% more performance. Since neither LOC nor performance can be * measured beforehand, it generally means implementing a feature then deciding * if it's worth it. And once it's implemented, who can say no? @@ -31,11 +34,14 @@ * Host (which is actually really easy). * * Two questions remain. Would the performance gain outweigh the complexity? - * And who would write the verse documenting it? :*/ + * And who would write the verse documenting it? +:*/ -/*M:011 Lguest64 handles NMI. This gave me NMI envy (until I looked at their +/*M:011 + * Lguest64 handles NMI. This gave me NMI envy (until I looked at their * code). It's worth doing though, since it would let us use oprofile in the - * Host when a Guest is running. :*/ + * Host when a Guest is running. +:*/ /*S:100 * Welcome to the Switcher itself! diff --git a/drivers/misc/eeprom/at25.c b/drivers/misc/eeprom/at25.c index b34cb5f79ee..2e535a0ccd5 100644 --- a/drivers/misc/eeprom/at25.c +++ b/drivers/misc/eeprom/at25.c @@ -173,6 +173,7 @@ at25_ee_write(struct at25_data *at25, const char *buf, loff_t off, unsigned segment; unsigned offset = (unsigned) off; u8 *cp = bounce + 1; + int sr; *cp = AT25_WREN; status = spi_write(at25->spi, cp, 1); @@ -214,7 +215,6 @@ at25_ee_write(struct at25_data *at25, const char *buf, loff_t off, timeout = jiffies + msecs_to_jiffies(EE_TIMEOUT); retries = 0; do { - int sr; sr = spi_w8r8(at25->spi, AT25_RDSR); if (sr < 0 || (sr & AT25_SR_nRDY)) { @@ -228,7 +228,7 @@ at25_ee_write(struct at25_data *at25, const char *buf, loff_t off, break; } while (retries++ < 3 || time_before_eq(jiffies, timeout)); - if (time_after(jiffies, timeout)) { + if ((sr < 0) || (sr & AT25_SR_nRDY)) { dev_err(&at25->spi->dev, "write %d bytes offset %d, " "timeout after %u msecs\n", diff --git a/drivers/mmc/host/sdhci-of.c b/drivers/mmc/host/sdhci-of.c index d79fa55c3b8..908844327db 100644 --- a/drivers/mmc/host/sdhci-of.c +++ b/drivers/mmc/host/sdhci-of.c @@ -158,6 +158,13 @@ static unsigned int esdhc_get_max_clock(struct sdhci_host *host) return of_host->clock; } +static unsigned int esdhc_get_min_clock(struct sdhci_host *host) +{ + struct sdhci_of_host *of_host = sdhci_priv(host); + + return of_host->clock / 256 / 16; +} + static unsigned int esdhc_get_timeout_clock(struct sdhci_host *host) { struct sdhci_of_host *of_host = sdhci_priv(host); @@ -184,6 +191,7 @@ static struct sdhci_of_data sdhci_esdhc = { .set_clock = esdhc_set_clock, .enable_dma = esdhc_enable_dma, .get_max_clock = esdhc_get_max_clock, + .get_min_clock = esdhc_get_min_clock, .get_timeout_clock = esdhc_get_timeout_clock, }, }; diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c index 6779b4ecab1..62041c7e924 100644 --- a/drivers/mmc/host/sdhci.c +++ b/drivers/mmc/host/sdhci.c @@ -1766,7 +1766,10 @@ int sdhci_add_host(struct sdhci_host *host) * Set host parameters. */ mmc->ops = &sdhci_ops; - mmc->f_min = host->max_clk / 256; + if (host->ops->get_min_clock) + mmc->f_min = host->ops->get_min_clock(host); + else + mmc->f_min = host->max_clk / 256; mmc->f_max = host->max_clk; mmc->caps = MMC_CAP_SDIO_IRQ; diff --git a/drivers/mmc/host/sdhci.h b/drivers/mmc/host/sdhci.h index 831ddf7dcb4..c77e9ff3022 100644 --- a/drivers/mmc/host/sdhci.h +++ b/drivers/mmc/host/sdhci.h @@ -302,6 +302,7 @@ struct sdhci_ops { int (*enable_dma)(struct sdhci_host *host); unsigned int (*get_max_clock)(struct sdhci_host *host); + unsigned int (*get_min_clock)(struct sdhci_host *host); unsigned int (*get_timeout_clock)(struct sdhci_host *host); }; diff --git a/drivers/rtc/rtc-cmos.c b/drivers/rtc/rtc-cmos.c index 23e10b6263d..f7a4701bf86 100644 --- a/drivers/rtc/rtc-cmos.c +++ b/drivers/rtc/rtc-cmos.c @@ -1174,23 +1174,34 @@ static struct platform_driver cmos_platform_driver = { } }; +#ifdef CONFIG_PNP +static bool pnp_driver_registered; +#endif +static bool platform_driver_registered; + static int __init cmos_init(void) { int retval = 0; #ifdef CONFIG_PNP - pnp_register_driver(&cmos_pnp_driver); + retval = pnp_register_driver(&cmos_pnp_driver); + if (retval == 0) + pnp_driver_registered = true; #endif - if (!cmos_rtc.dev) + if (!cmos_rtc.dev) { retval = platform_driver_probe(&cmos_platform_driver, cmos_platform_probe); + if (retval == 0) + platform_driver_registered = true; + } if (retval == 0) return 0; #ifdef CONFIG_PNP - pnp_unregister_driver(&cmos_pnp_driver); + if (pnp_driver_registered) + pnp_unregister_driver(&cmos_pnp_driver); #endif return retval; } @@ -1199,9 +1210,11 @@ module_init(cmos_init); static void __exit cmos_exit(void) { #ifdef CONFIG_PNP - pnp_unregister_driver(&cmos_pnp_driver); + if (pnp_driver_registered) + pnp_unregister_driver(&cmos_pnp_driver); #endif - platform_driver_unregister(&cmos_platform_driver); + if (platform_driver_registered) + platform_driver_unregister(&cmos_platform_driver); } module_exit(cmos_exit); diff --git a/drivers/serial/atmel_serial.c b/drivers/serial/atmel_serial.c index 338b15c0a54..607d43a3104 100644 --- a/drivers/serial/atmel_serial.c +++ b/drivers/serial/atmel_serial.c @@ -1551,6 +1551,7 @@ static int __devinit atmel_serial_probe(struct platform_device *pdev) if (ret) goto err_add_port; +#ifdef CONFIG_SERIAL_ATMEL_CONSOLE if (atmel_is_console_port(&port->uart) && ATMEL_CONSOLE_DEVICE->flags & CON_ENABLED) { /* @@ -1559,6 +1560,7 @@ static int __devinit atmel_serial_probe(struct platform_device *pdev) */ clk_disable(port->clk); } +#endif device_init_wakeup(&pdev->dev, 1); platform_set_drvdata(pdev, port); diff --git a/drivers/spi/omap2_mcspi.c b/drivers/spi/omap2_mcspi.c index eee4b6e0af2..9b80ad36dbb 100644 --- a/drivers/spi/omap2_mcspi.c +++ b/drivers/spi/omap2_mcspi.c @@ -59,6 +59,8 @@ /* per-register bitmasks: */ +#define OMAP2_MCSPI_SYSCONFIG_SMARTIDLE (2 << 3) +#define OMAP2_MCSPI_SYSCONFIG_ENAWAKEUP (1 << 2) #define OMAP2_MCSPI_SYSCONFIG_AUTOIDLE (1 << 0) #define OMAP2_MCSPI_SYSCONFIG_SOFTRESET (1 << 1) @@ -90,6 +92,7 @@ #define OMAP2_MCSPI_CHCTRL_EN (1 << 0) +#define OMAP2_MCSPI_WAKEUPENABLE_WKEN (1 << 0) /* We have 2 DMA channels per CS, one for RX and one for TX */ struct omap2_mcspi_dma { @@ -269,7 +272,7 @@ omap2_mcspi_txrx_dma(struct spi_device *spi, struct spi_transfer *xfer) if (rx != NULL) { omap_set_dma_transfer_params(mcspi_dma->dma_rx_channel, - data_type, element_count, 1, + data_type, element_count - 1, 1, OMAP_DMA_SYNC_ELEMENT, mcspi_dma->dma_rx_sync_dev, 1); @@ -300,6 +303,25 @@ omap2_mcspi_txrx_dma(struct spi_device *spi, struct spi_transfer *xfer) if (rx != NULL) { wait_for_completion(&mcspi_dma->dma_rx_completion); dma_unmap_single(NULL, xfer->rx_dma, count, DMA_FROM_DEVICE); + omap2_mcspi_set_enable(spi, 0); + if (likely(mcspi_read_cs_reg(spi, OMAP2_MCSPI_CHSTAT0) + & OMAP2_MCSPI_CHSTAT_RXS)) { + u32 w; + + w = mcspi_read_cs_reg(spi, OMAP2_MCSPI_RX0); + if (word_len <= 8) + ((u8 *)xfer->rx_buf)[element_count - 1] = w; + else if (word_len <= 16) + ((u16 *)xfer->rx_buf)[element_count - 1] = w; + else /* word_len <= 32 */ + ((u32 *)xfer->rx_buf)[element_count - 1] = w; + } else { + dev_err(&spi->dev, "DMA RX last word empty"); + count -= (word_len <= 8) ? 1 : + (word_len <= 16) ? 2 : + /* word_len <= 32 */ 4; + } + omap2_mcspi_set_enable(spi, 1); } return count; } @@ -873,8 +895,12 @@ static int __init omap2_mcspi_reset(struct omap2_mcspi *mcspi) } while (!(tmp & OMAP2_MCSPI_SYSSTATUS_RESETDONE)); mcspi_write_reg(master, OMAP2_MCSPI_SYSCONFIG, - /* (3 << 8) | (2 << 3) | */ - OMAP2_MCSPI_SYSCONFIG_AUTOIDLE); + OMAP2_MCSPI_SYSCONFIG_AUTOIDLE | + OMAP2_MCSPI_SYSCONFIG_ENAWAKEUP | + OMAP2_MCSPI_SYSCONFIG_SMARTIDLE); + + mcspi_write_reg(master, OMAP2_MCSPI_WAKEUPENABLE, + OMAP2_MCSPI_WAKEUPENABLE_WKEN); omap2_mcspi_set_master_mode(master); diff --git a/drivers/staging/Kconfig b/drivers/staging/Kconfig index 348bf61a8fe..975ecddbce3 100644 --- a/drivers/staging/Kconfig +++ b/drivers/staging/Kconfig @@ -103,8 +103,6 @@ source "drivers/staging/pohmelfs/Kconfig" source "drivers/staging/stlc45xx/Kconfig" -source "drivers/staging/uc2322/Kconfig" - source "drivers/staging/b3dfg/Kconfig" source "drivers/staging/phison/Kconfig" diff --git a/drivers/staging/Makefile b/drivers/staging/Makefile index 8d61d7b4deb..2241ae1b21e 100644 --- a/drivers/staging/Makefile +++ b/drivers/staging/Makefile @@ -34,7 +34,6 @@ obj-$(CONFIG_ANDROID) += android/ obj-$(CONFIG_DST) += dst/ obj-$(CONFIG_POHMELFS) += pohmelfs/ obj-$(CONFIG_STLC45XX) += stlc45xx/ -obj-$(CONFIG_USB_SERIAL_ATEN2011) += uc2322/ obj-$(CONFIG_B3DFG) += b3dfg/ obj-$(CONFIG_IDE_PHISON) += phison/ obj-$(CONFIG_PLAN9AUTH) += p9auth/ diff --git a/drivers/staging/android/lowmemorykiller.c b/drivers/staging/android/lowmemorykiller.c index fe72240f5a9..f934393f395 100644 --- a/drivers/staging/android/lowmemorykiller.c +++ b/drivers/staging/android/lowmemorykiller.c @@ -96,19 +96,21 @@ static int lowmem_shrink(int nr_to_scan, gfp_t gfp_mask) read_lock(&tasklist_lock); for_each_process(p) { + struct mm_struct *mm; int oom_adj; task_lock(p); - if (!p->mm) { + mm = p->mm; + if (!mm) { task_unlock(p); continue; } - oom_adj = p->oomkilladj; + oom_adj = mm->oom_adj; if (oom_adj < min_adj) { task_unlock(p); continue; } - tasksize = get_mm_rss(p->mm); + tasksize = get_mm_rss(mm); task_unlock(p); if (tasksize <= 0) continue; diff --git a/drivers/staging/serqt_usb2/serqt_usb2.c b/drivers/staging/serqt_usb2/serqt_usb2.c index a9bd4106beb..0fdf8c6dc64 100644 --- a/drivers/staging/serqt_usb2/serqt_usb2.c +++ b/drivers/staging/serqt_usb2/serqt_usb2.c @@ -360,18 +360,18 @@ static void qt_read_bulk_callback(struct urb *urb) if (port_paranoia_check(port, __func__) != 0) { dbg("%s - port_paranoia_check, exiting\n", __func__); qt_port->ReadBulkStopped = 1; - return; + goto exit; } if (!serial) { dbg("%s - bad serial pointer, exiting\n", __func__); - return; + goto exit; } if (qt_port->closePending == 1) { /* Were closing , stop reading */ dbg("%s - (qt_port->closepending == 1\n", __func__); qt_port->ReadBulkStopped = 1; - return; + goto exit; } /* @@ -381,7 +381,7 @@ static void qt_read_bulk_callback(struct urb *urb) */ if (qt_port->RxHolding == 1) { qt_port->ReadBulkStopped = 1; - return; + goto exit; } if (urb->status) { @@ -389,7 +389,7 @@ static void qt_read_bulk_callback(struct urb *urb) dbg("%s - nonzero read bulk status received: %d\n", __func__, urb->status); - return; + goto exit; } if (tty && RxCount) { @@ -463,6 +463,8 @@ static void qt_read_bulk_callback(struct urb *urb) } schedule_work(&port->work); +exit: + tty_kref_put(tty); } /* @@ -736,6 +738,11 @@ static int qt_startup(struct usb_serial *serial) if (!qt_port) { dbg("%s: kmalloc for quatech_port (%d) failed!.", __func__, i); + for(--i; i >= 0; i--) { + port = serial->port[i]; + kfree(usb_get_serial_port_data(port)); + usb_set_serial_port_data(port, NULL); + } return -ENOMEM; } spin_lock_init(&qt_port->lock); @@ -1041,7 +1048,7 @@ static void qt_block_until_empty(struct tty_struct *tty, } } -static void qt_close( struct usb_serial_port *port) +static void qt_close(struct usb_serial_port *port) { struct usb_serial *serial = port->serial; struct quatech_port *qt_port; @@ -1068,6 +1075,7 @@ static void qt_close( struct usb_serial_port *port) /* wait up to for transmitter to empty */ if (serial->dev) qt_block_until_empty(tty, qt_port); + tty_kref_put(tty); /* Close uart channel */ status = qt_close_channel(serial, index); diff --git a/drivers/staging/uc2322/Kconfig b/drivers/staging/uc2322/Kconfig deleted file mode 100644 index 2e0c6e79df2..00000000000 --- a/drivers/staging/uc2322/Kconfig +++ /dev/null @@ -1,10 +0,0 @@ -config USB_SERIAL_ATEN2011 - tristate "ATEN 2011 USB to serial device support" - depends on USB_SERIAL - default N - ---help--- - Say Y here if you want to use a ATEN 2011 dual port USB to serial - adapter. - - To compile this driver as a module, choose M here: the module will be - called aten2011. diff --git a/drivers/staging/uc2322/Makefile b/drivers/staging/uc2322/Makefile deleted file mode 100644 index 49c18d6e579..00000000000 --- a/drivers/staging/uc2322/Makefile +++ /dev/null @@ -1 +0,0 @@ -obj-$(CONFIG_USB_SERIAL_ATEN2011) += aten2011.o diff --git a/drivers/staging/uc2322/TODO b/drivers/staging/uc2322/TODO deleted file mode 100644 index c189a64c418..00000000000 --- a/drivers/staging/uc2322/TODO +++ /dev/null @@ -1,7 +0,0 @@ -TODO: - - checkpatch.pl cleanups - - remove dead and useless code (auditing the tty ioctls to - verify that they really are correct and needed.) - -Please send any patches to Greg Kroah-Hartman <greg@kroah.com> and -Russell Lang <gsview@ghostgum.com.au>. diff --git a/drivers/staging/uc2322/aten2011.c b/drivers/staging/uc2322/aten2011.c deleted file mode 100644 index 39d0926d1a9..00000000000 --- a/drivers/staging/uc2322/aten2011.c +++ /dev/null @@ -1,2430 +0,0 @@ -/* - * Aten 2011 USB serial driver for 4 port devices - * - * Copyright (C) 2000 Inside Out Networks - * Copyright (C) 2001-2002, 2009 Greg Kroah-Hartman <greg@kroah.com> - * Copyright (C) 2009 Novell Inc. - * - * This program is free software; you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation; either version 2 of the License, or - * (at your option) any later version. - * - */ - -#include <linux/kernel.h> -#include <linux/errno.h> -#include <linux/init.h> -#include <linux/slab.h> -#include <linux/tty.h> -#include <linux/tty_driver.h> -#include <linux/tty_flip.h> -#include <linux/module.h> -#include <linux/serial.h> -#include <linux/uaccess.h> -#include <linux/usb.h> -#include <linux/usb/serial.h> - - -#define ZLP_REG1 0x3A /* Zero_Flag_Reg1 58 */ -#define ZLP_REG2 0x3B /* Zero_Flag_Reg2 59 */ -#define ZLP_REG3 0x3C /* Zero_Flag_Reg3 60 */ -#define ZLP_REG4 0x3D /* Zero_Flag_Reg4 61 */ -#define ZLP_REG5 0x3E /* Zero_Flag_Reg5 62 */ - -/* Interrupt Rotinue Defines */ -#define SERIAL_IIR_RLS 0x06 -#define SERIAL_IIR_RDA 0x04 -#define SERIAL_IIR_CTI 0x0c -#define SERIAL_IIR_THR 0x02 -#define SERIAL_IIR_MS 0x00 - -/* Emulation of the bit mask on the LINE STATUS REGISTER. */ -#define SERIAL_LSR_DR 0x0001 -#define SERIAL_LSR_OE 0x0002 -#define SERIAL_LSR_PE 0x0004 -#define SERIAL_LSR_FE 0x0008 -#define SERIAL_LSR_BI 0x0010 -#define SERIAL_LSR_THRE 0x0020 -#define SERIAL_LSR_TEMT 0x0040 -#define SERIAL_LSR_FIFOERR 0x0080 - -/* MSR bit defines(place holders) */ -#define ATEN_MSR_DELTA_CTS 0x10 -#define ATEN_MSR_DELTA_DSR 0x20 -#define ATEN_MSR_DELTA_RI 0x40 -#define ATEN_MSR_DELTA_CD 0x80 - -/* Serial Port register Address */ -#define RECEIVE_BUFFER_REGISTER ((__u16)(0x00)) -#define TRANSMIT_HOLDING_REGISTER ((__u16)(0x00)) -#define INTERRUPT_ENABLE_REGISTER ((__u16)(0x01)) -#define INTERRUPT_IDENT_REGISTER ((__u16)(0x02)) -#define FIFO_CONTROL_REGISTER ((__u16)(0x02)) -#define LINE_CONTROL_REGISTER ((__u16)(0x03)) -#define MODEM_CONTROL_REGISTER ((__u16)(0x04)) -#define LINE_STATUS_REGISTER ((__u16)(0x05)) -#define MODEM_STATUS_REGISTER ((__u16)(0x06)) -#define SCRATCH_PAD_REGISTER ((__u16)(0x07)) -#define DIVISOR_LATCH_LSB ((__u16)(0x00)) -#define DIVISOR_LATCH_MSB ((__u16)(0x01)) - -#define SP1_REGISTER ((__u16)(0x00)) -#define CONTROL1_REGISTER ((__u16)(0x01)) -#define CLK_MULTI_REGISTER ((__u16)(0x02)) -#define CLK_START_VALUE_REGISTER ((__u16)(0x03)) -#define DCR1_REGISTER ((__u16)(0x04)) -#define GPIO_REGISTER ((__u16)(0x07)) - -#define SERIAL_LCR_DLAB ((__u16)(0x0080)) - -/* - * URB POOL related defines - */ -#define NUM_URBS 16 /* URB Count */ -#define URB_TRANSFER_BUFFER_SIZE 32 /* URB Size */ - -#define USB_VENDOR_ID_ATENINTL 0x0557 -#define ATENINTL_DEVICE_ID_2011 0x2011 -#define ATENINTL_DEVICE_ID_7820 0x7820 - -static struct usb_device_id id_table[] = { - { USB_DEVICE(USB_VENDOR_ID_ATENINTL, ATENINTL_DEVICE_ID_2011) }, - { USB_DEVICE(USB_VENDOR_ID_ATENINTL, ATENINTL_DEVICE_ID_7820) }, - { } /* terminating entry */ -}; -MODULE_DEVICE_TABLE(usb, id_table); - -/* This structure holds all of the local port information */ -struct ATENINTL_port { - int port_num; /*Actual port number in the device(1,2,etc)*/ - __u8 bulk_out_endpoint; /* the bulk out endpoint handle */ - unsigned char *bulk_out_buffer; /* buffer used for the bulk out endpoint */ - struct urb *write_urb; /* write URB for this port */ - __u8 bulk_in_endpoint; /* the bulk in endpoint handle */ - unsigned char *bulk_in_buffer; /* the buffer we use for the bulk in endpoint */ - struct urb *read_urb; /* read URB for this port */ - __u8 shadowLCR; /* last LCR value received */ - __u8 shadowMCR; /* last MCR value received */ - char open; - char chaseResponsePending; - wait_queue_head_t wait_chase; /* for handling sleeping while waiting for chase to finish */ - wait_queue_head_t wait_command; /* for handling sleeping while waiting for command to finish */ - struct async_icount icount; - struct usb_serial_port *port; /* loop back to the owner of this object */ - /*Offsets*/ - __u8 SpRegOffset; - __u8 ControlRegOffset; - __u8 DcrRegOffset; - /* for processing control URBS in interrupt context */ - struct urb *control_urb; - char *ctrl_buf; - int MsrLsr; - - struct urb *write_urb_pool[NUM_URBS]; - /* we pass a pointer to this as the arguement sent to cypress_set_termios old_termios */ - struct ktermios tmp_termios; /* stores the old termios settings */ - spinlock_t lock; /* private lock */ -}; - -/* This structure holds all of the individual serial device information */ -struct ATENINTL_serial { - __u8 interrupt_in_endpoint; /* the interrupt endpoint handle */ - unsigned char *interrupt_in_buffer; /* the buffer we use for the interrupt endpoint */ - struct urb *interrupt_read_urb; /* our interrupt urb */ - __u8 bulk_in_endpoint; /* the bulk in endpoint handle */ - unsigned char *bulk_in_buffer; /* the buffer we use for the bulk in endpoint */ - struct urb *read_urb; /* our bulk read urb */ - __u8 bulk_out_endpoint; /* the bulk out endpoint handle */ - struct usb_serial *serial; /* loop back to the owner of this object */ - int ATEN2011_spectrum_2or4ports; /* this says the number of ports in the device */ - /* Indicates about the no.of opened ports of an individual USB-serial adapater. */ - unsigned int NoOfOpenPorts; - /* a flag for Status endpoint polling */ - unsigned char status_polling_started; -}; - -static void ATEN2011_set_termios(struct tty_struct *tty, - struct usb_serial_port *port, - struct ktermios *old_termios); -static void ATEN2011_change_port_settings(struct tty_struct *tty, - struct ATENINTL_port *ATEN2011_port, - struct ktermios *old_termios); - -/************************************* - * Bit definitions for each register * - *************************************/ -#define LCR_BITS_5 0x00 /* 5 bits/char */ -#define LCR_BITS_6 0x01 /* 6 bits/char */ -#define LCR_BITS_7 0x02 /* 7 bits/char */ -#define LCR_BITS_8 0x03 /* 8 bits/char */ -#define LCR_BITS_MASK 0x03 /* Mask for bits/char field */ - -#define LCR_STOP_1 0x00 /* 1 stop bit */ -#define LCR_STOP_1_5 0x04 /* 1.5 stop bits (if 5 bits/char) */ -#define LCR_STOP_2 0x04 /* 2 stop bits (if 6-8 bits/char) */ -#define LCR_STOP_MASK 0x04 /* Mask for stop bits field */ - -#define LCR_PAR_NONE 0x00 /* No parity */ -#define LCR_PAR_ODD 0x08 /* Odd parity */ -#define LCR_PAR_EVEN 0x18 /* Even parity */ -#define LCR_PAR_MARK 0x28 /* Force parity bit to 1 */ -#define LCR_PAR_SPACE 0x38 /* Force parity bit to 0 */ -#define LCR_PAR_MASK 0x38 /* Mask for parity field */ - -#define LCR_SET_BREAK 0x40 /* Set Break condition */ -#define LCR_DL_ENABLE 0x80 /* Enable access to divisor latch */ - -#define MCR_DTR 0x01 /* Assert DTR */ -#define MCR_RTS 0x02 /* Assert RTS */ -#define MCR_OUT1 0x04 /* Loopback only: Sets state of RI */ -#define MCR_MASTER_IE 0x08 /* Enable interrupt outputs */ -#define MCR_LOOPBACK 0x10 /* Set internal (digital) loopback mode */ -#define MCR_XON_ANY 0x20 /* Enable any char to exit XOFF mode */ - -#define ATEN2011_MSR_CTS 0x10 /* Current state of CTS */ -#define ATEN2011_MSR_DSR 0x20 /* Current state of DSR */ -#define ATEN2011_MSR_RI 0x40 /* Current state of RI */ -#define ATEN2011_MSR_CD 0x80 /* Current state of CD */ - - -static int debug; - -/* - * Version Information - */ -#define DRIVER_VERSION "2.0" -#define DRIVER_DESC "ATENINTL 2011 USB Serial Adapter" - -/* - * Defines used for sending commands to port - */ - -#define ATEN_WDR_TIMEOUT (50) /* default urb timeout */ - -/* Requests */ -#define ATEN_RD_RTYPE 0xC0 -#define ATEN_WR_RTYPE 0x40 -#define ATEN_RDREQ 0x0D -#define ATEN_WRREQ 0x0E -#define ATEN_CTRL_TIMEOUT 500 -#define VENDOR_READ_LENGTH (0x01) - -/* set to 1 for RS485 mode and 0 for RS232 mode */ -/* FIXME make this somehow dynamic and not build time specific */ -static int RS485mode; - -static int set_reg_sync(struct usb_serial_port *port, __u16 reg, __u16 val) -{ - struct usb_device *dev = port->serial->dev; - val = val & 0x00ff; - - dbg("%s: is %x, value %x", __func__, reg, val); - - return usb_control_msg(dev, usb_sndctrlpipe(dev, 0), ATEN_WRREQ, - ATEN_WR_RTYPE, val, reg, NULL, 0, - ATEN_WDR_TIMEOUT); -} - -static int get_reg_sync(struct usb_serial_port *port, __u16 reg, __u16 *val) -{ - struct usb_device *dev = port->serial->dev; - int ret; - - ret = usb_control_msg(dev, usb_rcvctrlpipe(dev, 0), ATEN_RDREQ, - ATEN_RD_RTYPE, 0, reg, val, VENDOR_READ_LENGTH, - ATEN_WDR_TIMEOUT); - dbg("%s: offset is %x, return val %x", __func__, reg, *val); - *val = (*val) & 0x00ff; - return ret; -} - -static int set_uart_reg(struct usb_serial_port *port, __u16 reg, __u16 val) -{ - struct usb_device *dev = port->serial->dev; - struct ATENINTL_serial *a_serial; - __u16 minor; - - a_serial = usb_get_serial_data(port->serial); - minor = port->serial->minor; - if (minor == SERIAL_TTY_NO_MINOR) - minor = 0; - val = val & 0x00ff; - - /* - * For the UART control registers, - * the application number need to be Or'ed - */ - if (a_serial->ATEN2011_spectrum_2or4ports == 4) - val |= (((__u16)port->number - minor) + 1) << 8; - else { - if (((__u16) port->number - minor) == 0) - val |= (((__u16)port->number - minor) + 1) << 8; - else - val |= (((__u16)port->number - minor) + 2) << 8; - } - dbg("%s: application number is %x", __func__, val); - - return usb_control_msg(dev, usb_sndctrlpipe(dev, 0), ATEN_WRREQ, - ATEN_WR_RTYPE, val, reg, NULL, 0, - ATEN_WDR_TIMEOUT); -} - -static int get_uart_reg(struct usb_serial_port *port, __u16 reg, __u16 *val) -{ - struct usb_device *dev = port->serial->dev; - int ret = 0; - __u16 wval; - struct ATENINTL_serial *a_serial; - __u16 minor = port->serial->minor; - - a_serial = usb_get_serial_data(port->serial); - if (minor == SERIAL_TTY_NO_MINOR) - minor = 0; - - /* wval is same as application number */ - if (a_serial->ATEN2011_spectrum_2or4ports == 4) - wval = (((__u16)port->number - minor) + 1) << 8; - else { - if (((__u16) port->number - minor) == 0) - wval = (((__u16) port->number - minor) + 1) << 8; - else - wval = (((__u16) port->number - minor) + 2) << 8; - } - dbg("%s: application number is %x", __func__, wval); - ret = usb_control_msg(dev, usb_rcvctrlpipe(dev, 0), ATEN_RDREQ, - ATEN_RD_RTYPE, wval, reg, val, VENDOR_READ_LENGTH, - ATEN_WDR_TIMEOUT); - *val = (*val) & 0x00ff; - return ret; -} - -static int handle_newMsr(struct ATENINTL_port *port, __u8 newMsr) -{ - struct ATENINTL_port *ATEN2011_port; - struct async_icount *icount; - ATEN2011_port = port; - icount = &ATEN2011_port->icount; - if (newMsr & - (ATEN_MSR_DELTA_CTS | ATEN_MSR_DELTA_DSR | ATEN_MSR_DELTA_RI | - ATEN_MSR_DELTA_CD)) { - icount = &ATEN2011_port->icount; - - /* update input line counters */ - if (newMsr & ATEN_MSR_DELTA_CTS) - icount->cts++; - if (newMsr & ATEN_MSR_DELTA_DSR) - icount->dsr++; - if (newMsr & ATEN_MSR_DELTA_CD) - icount->dcd++; - if (newMsr & ATEN_MSR_DELTA_RI) - icount->rng++; - } - - return 0; -} - -static int handle_newLsr(struct ATENINTL_port *port, __u8 newLsr) -{ - struct async_icount *icount; - - dbg("%s - %02x", __func__, newLsr); - - if (newLsr & SERIAL_LSR_BI) { - /* - * Parity and Framing errors only count if they occur exclusive - * of a break being received. - */ - newLsr &= (__u8) (SERIAL_LSR_OE | SERIAL_LSR_BI); - } - - /* update input line counters */ - icount = &port->icount; - if (newLsr & SERIAL_LSR_BI) - icount->brk++; - if (newLsr & SERIAL_LSR_OE) - icount->overrun++; - if (newLsr & SERIAL_LSR_PE) - icount->parity++; - if (newLsr & SERIAL_LSR_FE) - icount->frame++; - - return 0; -} - -static void ATEN2011_control_callback(struct urb *urb) -{ - unsigned char *data; - struct ATENINTL_port *ATEN2011_port; - __u8 regval = 0x0; - - switch (urb->status) { - case 0: - /* success */ - break; - case -ECONNRESET: - case -ENOENT: - case -ESHUTDOWN: - /* this urb is terminated, clean up */ - dbg("%s - urb shutting down with status: %d", __func__, - urb->status); - return; - default: - dbg("%s - nonzero urb status received: %d", __func__, - urb->status); - goto exit; - } - - ATEN2011_port = (struct ATENINTL_port *)urb->context; - - dbg("%s urb buffer size is %d", __func__, urb->actual_length); - dbg("%s ATEN2011_port->MsrLsr is %d port %d", __func__, - ATEN2011_port->MsrLsr, ATEN2011_port->port_num); - data = urb->transfer_buffer; - regval = (__u8) data[0]; - dbg("%s data is %x", __func__, regval); - if (ATEN2011_port->MsrLsr == 0) - handle_newMsr(ATEN2011_port, regval); - else if (ATEN2011_port->MsrLsr == 1) - handle_newLsr(ATEN2011_port, regval); - -exit: - return; -} - -static int ATEN2011_get_reg(struct ATENINTL_port *ATEN, __u16 Wval, __u16 reg, - __u16 *val) -{ - struct usb_device *dev = ATEN->port->serial->dev; - struct usb_ctrlrequest *dr = NULL; - unsigned char *buffer = NULL; - int ret = 0; - buffer = (__u8 *) ATEN->ctrl_buf; - - dr = (void *)(buffer + 2); - dr->bRequestType = ATEN_RD_RTYPE; - dr->bRequest = ATEN_RDREQ; - dr->wValue = cpu_to_le16(Wval); - dr->wIndex = cpu_to_le16(reg); - dr->wLength = cpu_to_le16(2); - - usb_fill_control_urb(ATEN->control_urb, dev, usb_rcvctrlpipe(dev, 0), - (unsigned char *)dr, buffer, 2, - ATEN2011_control_callback, ATEN); - ATEN->control_urb->transfer_buffer_length = 2; - ret = usb_submit_urb(ATEN->control_urb, GFP_ATOMIC); - return ret; -} - -static void ATEN2011_interrupt_callback(struct urb *urb) -{ - int result; - int length; - struct ATENINTL_port *ATEN2011_port; - struct ATENINTL_serial *ATEN2011_serial; - struct usb_serial *serial; - __u16 Data; - unsigned char *data; - __u8 sp[5], st; - int i; - __u16 wval; - int minor; - - dbg("%s", " : Entering"); - - ATEN2011_serial = (struct ATENINTL_serial *)urb->context; - - switch (urb->status) { - case 0: - /* success */ - break; - case -ECONNRESET: - case -ENOENT: - case -ESHUTDOWN: - /* this urb is terminated, clean up */ - dbg("%s - urb shutting down with status: %d", __func__, - urb->status); - return; - default: - dbg("%s - nonzero urb status received: %d", __func__, - urb->status); - goto exit; - } - length = urb->actual_length; - data = urb->transfer_buffer; - - serial = ATEN2011_serial->serial; - - /* ATENINTL get 5 bytes - * Byte 1 IIR Port 1 (port.number is 0) - * Byte 2 IIR Port 2 (port.number is 1) - * Byte 3 IIR Port 3 (port.number is 2) - * Byte 4 IIR Port 4 (port.number is 3) - * Byte 5 FIFO status for both */ - - if (length && length > 5) { - dbg("%s", "Wrong data !!!"); - return; - } - - /* MATRIX */ - if (ATEN2011_serial->ATEN2011_spectrum_2or4ports == 4) { - sp[0] = (__u8) data[0]; - sp[1] = (__u8) data[1]; - sp[2] = (__u8) data[2]; - sp[3] = (__u8) data[3]; - st = (__u8) data[4]; - } else { - sp[0] = (__u8) data[0]; - sp[1] = (__u8) data[2]; - /* sp[2]=(__u8)data[2]; */ - /* sp[3]=(__u8)data[3]; */ - st = (__u8) data[4]; - - } - for (i = 0; i < serial->num_ports; i++) { - ATEN2011_port = usb_get_serial_port_data(serial->port[i]); - minor = serial->minor; - if (minor == SERIAL_TTY_NO_MINOR) - minor = 0; - if ((ATEN2011_serial->ATEN2011_spectrum_2or4ports == 2) - && (i != 0)) - wval = - (((__u16) serial->port[i]->number - - (__u16) (minor)) + 2) << 8; - else - wval = - (((__u16) serial->port[i]->number - - (__u16) (minor)) + 1) << 8; - if (ATEN2011_port->open != 0) { - if (sp[i] & 0x01) { - dbg("SP%d No Interrupt !!!", i); - } else { - switch (sp[i] & 0x0f) { - case SERIAL_IIR_RLS: - dbg("Serial Port %d: Receiver status error or address bit detected in 9-bit mode", i); - ATEN2011_port->MsrLsr = 1; - ATEN2011_get_reg(ATEN2011_port, wval, - LINE_STATUS_REGISTER, - &Data); - break; - case SERIAL_IIR_MS: - dbg("Serial Port %d: Modem status change", i); - ATEN2011_port->MsrLsr = 0; - ATEN2011_get_reg(ATEN2011_port, wval, - MODEM_STATUS_REGISTER, - &Data); - break; - } - } - } - - } -exit: - if (ATEN2011_serial->status_polling_started == 0) - return; - - result = usb_submit_urb(urb, GFP_ATOMIC); - if (result) { - dev_err(&urb->dev->dev, - "%s - Error %d submitting interrupt urb\n", - __func__, result); - } - - return; -} - -static void ATEN2011_bulk_in_callback(struct urb *urb) -{ - int status; - unsigned char *data; - struct usb_serial *serial; - struct usb_serial_port *port; - struct ATENINTL_serial *ATEN2011_serial; - struct ATENINTL_port *ATEN2011_port; - struct tty_struct *tty; - - if (urb->status) { - dbg("nonzero read bulk status received: %d", urb->status); - return; - } - - ATEN2011_port = (struct ATENINTL_port *)urb->context; - - port = (struct usb_serial_port *)ATEN2011_port->port; - serial = port->serial; - - dbg("%s", "Entering..."); - - data = urb->transfer_buffer; - ATEN2011_serial = usb_get_serial_data(serial); - - if (urb->actual_length) { - tty = tty_port_tty_get(&ATEN2011_port->port->port); - if (tty) { - tty_buffer_request_room(tty, urb->actual_length); - tty_insert_flip_string(tty, data, urb->actual_length); - tty_flip_buffer_push(tty); - tty_kref_put(tty); - } - - ATEN2011_port->icount.rx += urb->actual_length; - dbg("ATEN2011_port->icount.rx is %d:", - ATEN2011_port->icount.rx); - } - - if (!ATEN2011_port->read_urb) { - dbg("%s", "URB KILLED !!!"); - return; - } - - if (ATEN2011_port->read_urb->status != -EINPROGRESS) { - ATEN2011_port->read_urb->dev = serial->dev; - - status = usb_submit_urb(ATEN2011_port->read_urb, GFP_ATOMIC); - if (status) - dbg("usb_submit_urb(read bulk) failed, status = %d", status); - } -} - -static void ATEN2011_bulk_out_data_callback(struct urb *urb) -{ - struct ATENINTL_port *ATEN2011_port; - struct tty_struct *tty; - - if (urb->status) { - dbg("nonzero write bulk status received:%d", urb->status); - return; - } - - ATEN2011_port = (struct ATENINTL_port *)urb->context; - - dbg("%s", "Entering ........."); - - tty = tty_port_tty_get(&ATEN2011_port->port->port); - - if (tty && ATEN2011_port->open) - /* tell the tty driver that something has changed */ - tty_wakeup(tty); - - /* schedule_work(&ATEN2011_port->port->work); */ - tty_kref_put(tty); - -} - -#ifdef ATENSerialProbe -static int ATEN2011_serial_probe(struct usb_serial *serial, - const struct usb_device_id *id) -{ - - /*need to implement the mode_reg reading and updating\ - structures usb_serial_ device_type\ - (i.e num_ports, num_bulkin,bulkout etc) */ - /* Also we can update the changes attach */ - return 1; -} -#endif - -static int ATEN2011_open(struct tty_struct *tty, struct usb_serial_port *port, - struct file *filp) -{ - int response; - int j; - struct usb_serial *serial; - struct urb *urb; - __u16 Data; - int status; - struct ATENINTL_serial *ATEN2011_serial; - struct ATENINTL_port *ATEN2011_port; - struct ktermios tmp_termios; - int minor; - - serial = port->serial; - - ATEN2011_port = usb_get_serial_port_data(port); - - if (ATEN2011_port == NULL) - return -ENODEV; - - ATEN2011_serial = usb_get_serial_data(serial); - if (ATEN2011_serial == NULL) - return -ENODEV; - - /* increment the number of opened ports counter here */ - ATEN2011_serial->NoOfOpenPorts++; - - usb_clear_halt(serial->dev, port->write_urb->pipe); - usb_clear_halt(serial->dev, port->read_urb->pipe); - - /* Initialising the write urb pool */ - for (j = 0; j < NUM_URBS; ++j) { - urb = usb_alloc_urb(0, GFP_ATOMIC); - ATEN2011_port->write_urb_pool[j] = urb; - - if (urb == NULL) { - err("No more urbs???"); - continue; - } - - urb->transfer_buffer = NULL; - urb->transfer_buffer = - kmalloc(URB_TRANSFER_BUFFER_SIZE, GFP_KERNEL); - if (!urb->transfer_buffer) { - err("%s-out of memory for urb buffers.", __func__); - continue; - } - } - -/***************************************************************************** - * Initialize ATEN2011 -- Write Init values to corresponding Registers - * - * Register Index - * 1 : IER - * 2 : FCR - * 3 : LCR - * 4 : MCR - * - * 0x08 : SP1/2 Control Reg - *****************************************************************************/ - -/* NEED to check the fallowing Block */ - - Data = 0x0; - status = get_reg_sync(port, ATEN2011_port->SpRegOffset, &Data); - if (status < 0) { - dbg("Reading Spreg failed"); - return -1; - } - Data |= 0x80; - status = set_reg_sync(port, ATEN2011_port->SpRegOffset, Data); - if (status < 0) { - dbg("writing Spreg failed"); - return -1; - } - - Data &= ~0x80; - status = set_reg_sync(port, ATEN2011_port->SpRegOffset, Data); - if (status < 0) { - dbg("writing Spreg failed"); - return -1; - } - -/* End of block to be checked */ -/**************************CHECK***************************/ - - if (RS485mode == 0) - Data = 0xC0; - else - Data = 0x00; - status = set_uart_reg(port, SCRATCH_PAD_REGISTER, Data); - if (status < 0) { - dbg("Writing SCRATCH_PAD_REGISTER failed status-0x%x", status); - return -1; - } else - dbg("SCRATCH_PAD_REGISTER Writing success status%d", status); - -/**************************CHECK***************************/ - - Data = 0x0; - status = get_reg_sync(port, ATEN2011_port->ControlRegOffset, &Data); - if (status < 0) { - dbg("Reading Controlreg failed"); - return -1; - } - Data |= 0x08; /* Driver done bit */ - Data |= 0x20; /* rx_disable */ - status = 0; - status = - set_reg_sync(port, ATEN2011_port->ControlRegOffset, Data); - if (status < 0) { - dbg("writing Controlreg failed"); - return -1; - } - /* - * do register settings here - * Set all regs to the device default values. - * First Disable all interrupts. - */ - - Data = 0x00; - status = set_uart_reg(port, INTERRUPT_ENABLE_REGISTER, Data); - if (status < 0) { - dbg("disableing interrupts failed"); - return -1; - } - /* Set FIFO_CONTROL_REGISTER to the default value */ - Data = 0x00; - status = set_uart_reg(port, FIFO_CONTROL_REGISTER, Data); - if (status < 0) { - dbg("Writing FIFO_CONTROL_REGISTER failed"); - return -1; - } - - Data = 0xcf; /* chk */ - status = set_uart_reg(port, FIFO_CONTROL_REGISTER, Data); - if (status < 0) { - dbg("Writing FIFO_CONTROL_REGISTER failed"); - return -1; - } - - Data = 0x03; /* LCR_BITS_8 */ - status = set_uart_reg(port, LINE_CONTROL_REGISTER, Data); - ATEN2011_port->shadowLCR = Data; - - Data = 0x0b; /* MCR_DTR|MCR_RTS|MCR_MASTER_IE */ - status = set_uart_reg(port, MODEM_CONTROL_REGISTER, Data); - ATEN2011_port->shadowMCR = Data; - -#ifdef Check - Data = 0x00; - status = get_uart_reg(port, LINE_CONTROL_REGISTER, &Data); - ATEN2011_port->shadowLCR = Data; - - Data |= SERIAL_LCR_DLAB; /* data latch enable in LCR 0x80 */ - status = set_uart_reg(port, LINE_CONTROL_REGISTER, Data); - - Data = 0x0c; - status = set_uart_reg(port, DIVISOR_LATCH_LSB, Data); - - Data = 0x0; - status = set_uart_reg(port, DIVISOR_LATCH_MSB, Data); - - Data = 0x00; - status = get_uart_reg(port, LINE_CONTROL_REGISTER, &Data); - -/* Data = ATEN2011_port->shadowLCR; */ /* data latch disable */ - Data = Data & ~SERIAL_LCR_DLAB; - status = set_uart_reg(port, LINE_CONTROL_REGISTER, Data); - ATEN2011_port->shadowLCR = Data; -#endif - /* clearing Bulkin and Bulkout Fifo */ - Data = 0x0; - status = get_reg_sync(port, ATEN2011_port->SpRegOffset, &Data); - - Data = Data | 0x0c; - status = set_reg_sync(port, ATEN2011_port->SpRegOffset, Data); - - Data = Data & ~0x0c; - status = set_reg_sync(port, ATEN2011_port->SpRegOffset, Data); - /* Finally enable all interrupts */ - Data = 0x0; - Data = 0x0c; - status = set_uart_reg(port, INTERRUPT_ENABLE_REGISTER, Data); - - /* clearing rx_disable */ - Data = 0x0; - status = get_reg_sync(port, ATEN2011_port->ControlRegOffset, &Data); - Data = Data & ~0x20; - status = set_reg_sync(port, ATEN2011_port->ControlRegOffset, Data); - - /* rx_negate */ - Data = 0x0; - status = get_reg_sync(port, ATEN2011_port->ControlRegOffset, &Data); - Data = Data | 0x10; - status = 0; - status = set_reg_sync(port, ATEN2011_port->ControlRegOffset, Data); - - /* - * Check to see if we've set up our endpoint info yet - * (can't set it up in ATEN2011_startup as the structures - * were not set up at that time.) - */ - if (ATEN2011_serial->NoOfOpenPorts == 1) { - /* start the status polling here */ - ATEN2011_serial->status_polling_started = 1; - /* If not yet set, Set here */ - ATEN2011_serial->interrupt_in_buffer = - serial->port[0]->interrupt_in_buffer; - ATEN2011_serial->interrupt_in_endpoint = - serial->port[0]->interrupt_in_endpointAddress; - ATEN2011_serial->interrupt_read_urb = - serial->port[0]->interrupt_in_urb; - - /* set up interrupt urb */ - usb_fill_int_urb(ATEN2011_serial->interrupt_read_urb, - serial->dev, - usb_rcvintpipe(serial->dev, - ATEN2011_serial-> - interrupt_in_endpoint), - ATEN2011_serial->interrupt_in_buffer, - ATEN2011_serial->interrupt_read_urb-> - transfer_buffer_length, - ATEN2011_interrupt_callback, ATEN2011_serial, - ATEN2011_serial->interrupt_read_urb->interval); - - /* start interrupt read for ATEN2011 * - * will continue as long as ATEN2011 is connected */ - - response = - usb_submit_urb(ATEN2011_serial->interrupt_read_urb, - GFP_KERNEL); - if (response) { - dbg("%s - Error %d submitting interrupt urb", - __func__, response); - } - - } - - /* - * See if we've set up our endpoint info yet - * (can't set it up in ATEN2011_startup as the - * structures were not set up at that time.) - */ - - dbg("port number is %d", port->number); - dbg("serial number is %d", port->serial->minor); - dbg("Bulkin endpoint is %d", port->bulk_in_endpointAddress); - dbg("BulkOut endpoint is %d", port->bulk_out_endpointAddress); - dbg("Interrupt endpoint is %d", - port->interrupt_in_endpointAddress); - dbg("port's number in the device is %d", ATEN2011_port->port_num); - ATEN2011_port->bulk_in_buffer = port->bulk_in_buffer; - ATEN2011_port->bulk_in_endpoint = port->bulk_in_endpointAddress; - ATEN2011_port->read_urb = port->read_urb; - ATEN2011_port->bulk_out_endpoint = port->bulk_out_endpointAddress; - - minor = port->serial->minor; - if (minor == SERIAL_TTY_NO_MINOR) - minor = 0; - - /* set up our bulk in urb */ - if ((ATEN2011_serial->ATEN2011_spectrum_2or4ports == 2) - && (((__u16) port->number - (__u16) (minor)) != 0)) { - usb_fill_bulk_urb(ATEN2011_port->read_urb, serial->dev, - usb_rcvbulkpipe(serial->dev, - (port-> - bulk_in_endpointAddress + - 2)), port->bulk_in_buffer, - ATEN2011_port->read_urb-> - transfer_buffer_length, - ATEN2011_bulk_in_callback, ATEN2011_port); - } else - usb_fill_bulk_urb(ATEN2011_port->read_urb, - serial->dev, - usb_rcvbulkpipe(serial->dev, - port-> - bulk_in_endpointAddress), - port->bulk_in_buffer, - ATEN2011_port->read_urb-> - transfer_buffer_length, - ATEN2011_bulk_in_callback, ATEN2011_port); - - dbg("ATEN2011_open: bulkin endpoint is %d", - port->bulk_in_endpointAddress); - response = usb_submit_urb(ATEN2011_port->read_urb, GFP_KERNEL); - if (response) { - err("%s - Error %d submitting control urb", __func__, - response); - } - - /* initialize our wait queues */ - init_waitqueue_head(&ATEN2011_port->wait_chase); - init_waitqueue_head(&ATEN2011_port->wait_command); - - /* initialize our icount structure */ - memset(&(ATEN2011_port->icount), 0x00, sizeof(ATEN2011_port->icount)); - - /* initialize our port settings */ - ATEN2011_port->shadowMCR = MCR_MASTER_IE; /* Must set to enable ints! */ - ATEN2011_port->chaseResponsePending = 0; - /* send a open port command */ - ATEN2011_port->open = 1; - /* ATEN2011_change_port_settings(ATEN2011_port,old_termios); */ - /* Setup termios */ - ATEN2011_set_termios(tty, port, &tmp_termios); - ATEN2011_port->icount.tx = 0; - ATEN2011_port->icount.rx = 0; - - dbg("usb_serial serial:%x ATEN2011_port:%x\nATEN2011_serial:%x usb_serial_port port:%x", - (unsigned int)serial, (unsigned int)ATEN2011_port, - (unsigned int)ATEN2011_serial, (unsigned int)port); - - return 0; - -} - -static int ATEN2011_chars_in_buffer(struct tty_struct *tty) -{ - struct usb_serial_port *port = tty->driver_data; - int i; - int chars = 0; - struct ATENINTL_port *ATEN2011_port; - - /* dbg("%s"," ATEN2011_chars_in_buffer:entering ..........."); */ - - ATEN2011_port = usb_get_serial_port_data(port); - if (ATEN2011_port == NULL) { - dbg("%s", "ATEN2011_break:leaving ..........."); - return -1; - } - - for (i = 0; i < NUM_URBS; ++i) - if (ATEN2011_port->write_urb_pool[i]->status == -EINPROGRESS) - chars += URB_TRANSFER_BUFFER_SIZE; - - dbg("%s - returns %d", __func__, chars); - return chars; - -} - -static void ATEN2011_block_until_tx_empty(struct tty_struct *tty, - struct ATENINTL_port *ATEN2011_port) -{ - int timeout = HZ / 10; - int wait = 30; - int count; - - while (1) { - count = ATEN2011_chars_in_buffer(tty); - - /* Check for Buffer status */ - if (count <= 0) - return; - - /* Block the thread for a while */ - interruptible_sleep_on_timeout(&ATEN2011_port->wait_chase, - timeout); - - /* No activity.. count down section */ - wait--; - if (wait == 0) { - dbg("%s - TIMEOUT", __func__); - return; - } else { - /* Reset timout value back to seconds */ - wait = 30; - } - } -} - -static void ATEN2011_close(struct tty_struct *tty, struct usb_serial_port *port, - struct file *filp) -{ - struct usb_serial *serial; - struct ATENINTL_serial *ATEN2011_serial; - struct ATENINTL_port *ATEN2011_port; - int no_urbs; - __u16 Data; - - dbg("%s", "ATEN2011_close:entering..."); - serial = port->serial; - - /* take the Adpater and port's private data */ - ATEN2011_serial = usb_get_serial_data(serial); - ATEN2011_port = usb_get_serial_port_data(port); - if ((ATEN2011_serial == NULL) || (ATEN2011_port == NULL)) - return; - - if (serial->dev) { - /* flush and block(wait) until tx is empty */ - ATEN2011_block_until_tx_empty(tty, ATEN2011_port); - } - /* kill the ports URB's */ - for (no_urbs = 0; no_urbs < NUM_URBS; no_urbs++) - usb_kill_urb(ATEN2011_port->write_urb_pool[no_urbs]); - /* Freeing Write URBs */ - for (no_urbs = 0; no_urbs < NUM_URBS; ++no_urbs) { - kfree(ATEN2011_port->write_urb_pool[no_urbs]->transfer_buffer); - usb_free_urb(ATEN2011_port->write_urb_pool[no_urbs]); - } - /* While closing port, shutdown all bulk read, write * - * and interrupt read if they exists */ - if (serial->dev) { - if (ATEN2011_port->write_urb) { - dbg("%s", "Shutdown bulk write"); - usb_kill_urb(ATEN2011_port->write_urb); - } - if (ATEN2011_port->read_urb) { - dbg("%s", "Shutdown bulk read"); - usb_kill_urb(ATEN2011_port->read_urb); - } - if ((&ATEN2011_port->control_urb)) { - dbg("%s", "Shutdown control read"); - /* usb_kill_urb (ATEN2011_port->control_urb); */ - - } - } - /* if(ATEN2011_port->ctrl_buf != NULL) */ - /* kfree(ATEN2011_port->ctrl_buf); */ - /* decrement the no.of open ports counter of an individual USB-serial adapter. */ - ATEN2011_serial->NoOfOpenPorts--; - dbg("NoOfOpenPorts in close%d:in port%d", - ATEN2011_serial->NoOfOpenPorts, port->number); - if (ATEN2011_serial->NoOfOpenPorts == 0) { - /* stop the stus polling here */ - ATEN2011_serial->status_polling_started = 0; - if (ATEN2011_serial->interrupt_read_urb) { - dbg("%s", "Shutdown interrupt_read_urb"); - /* ATEN2011_serial->interrupt_in_buffer=NULL; */ - /* usb_kill_urb (ATEN2011_serial->interrupt_read_urb); */ - } - } - if (ATEN2011_port->write_urb) { - /* if this urb had a transfer buffer already (old tx) free it */ - kfree(ATEN2011_port->write_urb->transfer_buffer); - usb_free_urb(ATEN2011_port->write_urb); - } - - /* clear the MCR & IER */ - Data = 0x00; - set_uart_reg(port, MODEM_CONTROL_REGISTER, Data); - Data = 0x00; - set_uart_reg(port, INTERRUPT_ENABLE_REGISTER, Data); - - ATEN2011_port->open = 0; - dbg("%s", "Leaving ............"); - -} - -static void ATEN2011_block_until_chase_response(struct tty_struct *tty, - struct ATENINTL_port - *ATEN2011_port) -{ - int timeout = 1 * HZ; - int wait = 10; - int count; - - while (1) { - count = ATEN2011_chars_in_buffer(tty); - - /* Check for Buffer status */ - if (count <= 0) { - ATEN2011_port->chaseResponsePending = 0; - return; - } - - /* Block the thread for a while */ - interruptible_sleep_on_timeout(&ATEN2011_port->wait_chase, - timeout); - /* No activity.. count down section */ - wait--; - if (wait == 0) { - dbg("%s - TIMEOUT", __func__); - return; - } else { - /* Reset timout value back to seconds */ - wait = 10; - } - } - -} - -static void ATEN2011_break(struct tty_struct *tty, int break_state) -{ - struct usb_serial_port *port = tty->driver_data; - unsigned char data; - struct usb_serial *serial; - struct ATENINTL_serial *ATEN2011_serial; - struct ATENINTL_port *ATEN2011_port; - - dbg("%s", "Entering ..........."); - dbg("ATEN2011_break: Start"); - - serial = port->serial; - - ATEN2011_serial = usb_get_serial_data(serial); - ATEN2011_port = usb_get_serial_port_data(port); - - if ((ATEN2011_serial == NULL) || (ATEN2011_port == NULL)) - return; - - /* flush and chase */ - ATEN2011_port->chaseResponsePending = 1; - - if (serial->dev) { - /* flush and block until tx is empty */ - ATEN2011_block_until_chase_response(tty, ATEN2011_port); - } - - if (break_state == -1) - data = ATEN2011_port->shadowLCR | LCR_SET_BREAK; - else - data = ATEN2011_port->shadowLCR & ~LCR_SET_BREAK; - - ATEN2011_port->shadowLCR = data; - dbg("ATEN2011_break ATEN2011_port->shadowLCR is %x", - ATEN2011_port->shadowLCR); - set_uart_reg(port, LINE_CONTROL_REGISTER, ATEN2011_port->shadowLCR); - - return; -} - -static int ATEN2011_write_room(struct tty_struct *tty) -{ - struct usb_serial_port *port = tty->driver_data; - int i; - int room = 0; - struct ATENINTL_port *ATEN2011_port; - - ATEN2011_port = usb_get_serial_port_data(port); - if (ATEN2011_port == NULL) { - dbg("%s", "ATEN2011_break:leaving ..........."); - return -1; - } - - for (i = 0; i < NUM_URBS; ++i) - if (ATEN2011_port->write_urb_pool[i]->status != -EINPROGRESS) - room += URB_TRANSFER_BUFFER_SIZE; - - dbg("%s - returns %d", __func__, room); - return room; - -} - -static int ATEN2011_write(struct tty_struct *tty, struct usb_serial_port *port, - const unsigned char *data, int count) -{ - int status; - int i; - int bytes_sent = 0; - int transfer_size; - int minor; - - struct ATENINTL_port *ATEN2011_port; - struct usb_serial *serial; - struct ATENINTL_serial *ATEN2011_serial; - struct urb *urb; - const unsigned char *current_position = data; - unsigned char *data1; - dbg("%s", "entering ..........."); - - serial = port->serial; - - ATEN2011_port = usb_get_serial_port_data(port); - if (ATEN2011_port == NULL) { - dbg("%s", "ATEN2011_port is NULL"); - return -1; - } - - ATEN2011_serial = usb_get_serial_data(serial); - if (ATEN2011_serial == NULL) { - dbg("%s", "ATEN2011_serial is NULL"); - return -1; - } - - /* try to find a free urb in the list */ - urb = NULL; - - for (i = 0; i < NUM_URBS; ++i) { - if (ATEN2011_port->write_urb_pool[i]->status != -EINPROGRESS) { - urb = ATEN2011_port->write_urb_pool[i]; - dbg("URB:%d", i); - break; - } - } - - if (urb == NULL) { - dbg("%s - no more free urbs", __func__); - goto exit; - } - - if (urb->transfer_buffer == NULL) { - urb->transfer_buffer = - kmalloc(URB_TRANSFER_BUFFER_SIZE, GFP_KERNEL); - - if (urb->transfer_buffer == NULL) { - err("%s no more kernel memory...", __func__); - goto exit; - } - } - transfer_size = min(count, URB_TRANSFER_BUFFER_SIZE); - - memcpy(urb->transfer_buffer, current_position, transfer_size); - /* usb_serial_debug_data (__FILE__, __func__, transfer_size, urb->transfer_buffer); */ - - /* fill urb with data and submit */ - minor = port->serial->minor; - if (minor == SERIAL_TTY_NO_MINOR) - minor = 0; - if ((ATEN2011_serial->ATEN2011_spectrum_2or4ports == 2) - && (((__u16) port->number - (__u16) (minor)) != 0)) { - usb_fill_bulk_urb(urb, ATEN2011_serial->serial->dev, - usb_sndbulkpipe(ATEN2011_serial->serial->dev, - (port-> - bulk_out_endpointAddress) + - 2), urb->transfer_buffer, - transfer_size, - ATEN2011_bulk_out_data_callback, - ATEN2011_port); - } else - - usb_fill_bulk_urb(urb, - ATEN2011_serial->serial->dev, - usb_sndbulkpipe(ATEN2011_serial->serial->dev, - port-> - bulk_out_endpointAddress), - urb->transfer_buffer, transfer_size, - ATEN2011_bulk_out_data_callback, - ATEN2011_port); - - data1 = urb->transfer_buffer; - dbg("bulkout endpoint is %d", port->bulk_out_endpointAddress); - /* for(i=0;i < urb->actual_length;i++) */ - /* dbg("Data is %c ",data1[i]); */ - - /* send it down the pipe */ - status = usb_submit_urb(urb, GFP_ATOMIC); - - if (status) { - err("%s - usb_submit_urb(write bulk) failed with status = %d", - __func__, status); - bytes_sent = status; - goto exit; - } - bytes_sent = transfer_size; - ATEN2011_port->icount.tx += transfer_size; - dbg("ATEN2011_port->icount.tx is %d:", ATEN2011_port->icount.tx); - -exit: - return bytes_sent; -} - -static void ATEN2011_throttle(struct tty_struct *tty) -{ - struct usb_serial_port *port = tty->driver_data; - struct ATENINTL_port *ATEN2011_port; - int status; - - dbg("- port %d", port->number); - - ATEN2011_port = usb_get_serial_port_data(port); - - if (ATEN2011_port == NULL) - return; - - if (!ATEN2011_port->open) { - dbg("%s", "port not opened"); - return; - } - - dbg("%s", "Entering .......... "); - - if (!tty) { - dbg("%s - no tty available", __func__); - return; - } - - /* if we are implementing XON/XOFF, send the stop character */ - if (I_IXOFF(tty)) { - unsigned char stop_char = STOP_CHAR(tty); - status = ATEN2011_write(tty, port, &stop_char, 1); - if (status <= 0) - return; - } - - /* if we are implementing RTS/CTS, toggle that line */ - if (tty->termios->c_cflag & CRTSCTS) { - ATEN2011_port->shadowMCR &= ~MCR_RTS; - status = set_uart_reg(port, MODEM_CONTROL_REGISTER, - ATEN2011_port->shadowMCR); - if (status < 0) - return; - } - - return; -} - -static void ATEN2011_unthrottle(struct tty_struct *tty) -{ - struct usb_serial_port *port = tty->driver_data; - int status; - struct ATENINTL_port *ATEN2011_port = usb_get_serial_port_data(port); - - if (ATEN2011_port == NULL) - return; - - if (!ATEN2011_port->open) { - dbg("%s - port not opened", __func__); - return; - } - - dbg("%s", "Entering .......... "); - - if (!tty) { - dbg("%s - no tty available", __func__); - return; - } - - /* if we are implementing XON/XOFF, send the start character */ - if (I_IXOFF(tty)) { - unsigned char start_char = START_CHAR(tty); - status = ATEN2011_write(tty, port, &start_char, 1); - if (status <= 0) - return; - } - - /* if we are implementing RTS/CTS, toggle that line */ - if (tty->termios->c_cflag & CRTSCTS) { - ATEN2011_port->shadowMCR |= MCR_RTS; - status = set_uart_reg(port, MODEM_CONTROL_REGISTER, - ATEN2011_port->shadowMCR); - if (status < 0) - return; - } - - return; -} - -static int ATEN2011_tiocmget(struct tty_struct *tty, struct file *file) -{ - struct usb_serial_port *port = tty->driver_data; - struct ATENINTL_port *ATEN2011_port; - unsigned int result; - __u16 msr; - __u16 mcr; - /* unsigned int mcr; */ - int status = 0; - ATEN2011_port = usb_get_serial_port_data(port); - - dbg("%s - port %d", __func__, port->number); - - if (ATEN2011_port == NULL) - return -ENODEV; - - status = get_uart_reg(port, MODEM_STATUS_REGISTER, &msr); - status = get_uart_reg(port, MODEM_CONTROL_REGISTER, &mcr); - /* mcr = ATEN2011_port->shadowMCR; */ - /* COMMENT2: the Fallowing three line are commented for updating only MSR values */ - result = ((mcr & MCR_DTR) ? TIOCM_DTR : 0) - | ((mcr & MCR_RTS) ? TIOCM_RTS : 0) - | ((mcr & MCR_LOOPBACK) ? TIOCM_LOOP : 0) - | ((msr & ATEN2011_MSR_CTS) ? TIOCM_CTS : 0) - | ((msr & ATEN2011_MSR_CD) ? TIOCM_CAR : 0) - | ((msr & ATEN2011_MSR_RI) ? TIOCM_RI : 0) - | ((msr & ATEN2011_MSR_DSR) ? TIOCM_DSR : 0); - - dbg("%s - 0x%04X", __func__, result); - - return result; -} - -static int ATEN2011_tiocmset(struct tty_struct *tty, struct file *file, - unsigned int set, unsigned int clear) -{ - struct usb_serial_port *port = tty->driver_data; - struct ATENINTL_port *ATEN2011_port; - unsigned int mcr; - unsigned int status; - - dbg("%s - port %d", __func__, port->number); - - ATEN2011_port = usb_get_serial_port_data(port); - - if (ATEN2011_port == NULL) - return -ENODEV; - - mcr = ATEN2011_port->shadowMCR; - if (clear & TIOCM_RTS) - mcr &= ~MCR_RTS; - if (clear & TIOCM_DTR) - mcr &= ~MCR_DTR; - if (clear & TIOCM_LOOP) - mcr &= ~MCR_LOOPBACK; - - if (set & TIOCM_RTS) - mcr |= MCR_RTS; - if (set & TIOCM_DTR) - mcr |= MCR_DTR; - if (set & TIOCM_LOOP) - mcr |= MCR_LOOPBACK; - - ATEN2011_port->shadowMCR = mcr; - - status = set_uart_reg(port, MODEM_CONTROL_REGISTER, mcr); - if (status < 0) { - dbg("setting MODEM_CONTROL_REGISTER Failed"); - return -1; - } - - return 0; -} - -static void ATEN2011_set_termios(struct tty_struct *tty, - struct usb_serial_port *port, - struct ktermios *old_termios) -{ - int status; - unsigned int cflag; - struct usb_serial *serial; - struct ATENINTL_port *ATEN2011_port; - - dbg("ATEN2011_set_termios: START"); - - serial = port->serial; - - ATEN2011_port = usb_get_serial_port_data(port); - - if (ATEN2011_port == NULL) - return; - - if (!ATEN2011_port->open) { - dbg("%s - port not opened", __func__); - return; - } - - dbg("%s", "setting termios - "); - - cflag = tty->termios->c_cflag; - - dbg("%s - cflag %08x iflag %08x", __func__, - tty->termios->c_cflag, RELEVANT_IFLAG(tty->termios->c_iflag)); - - if (old_termios) { - dbg("%s - old clfag %08x old iflag %08x", __func__, - old_termios->c_cflag, RELEVANT_IFLAG(old_termios->c_iflag)); - } - - dbg("%s - port %d", __func__, port->number); - - /* change the port settings to the new ones specified */ - - ATEN2011_change_port_settings(tty, ATEN2011_port, old_termios); - - if (!ATEN2011_port->read_urb) { - dbg("%s", "URB KILLED !!!!!"); - return; - } - - if (ATEN2011_port->read_urb->status != -EINPROGRESS) { - ATEN2011_port->read_urb->dev = serial->dev; - status = usb_submit_urb(ATEN2011_port->read_urb, GFP_ATOMIC); - if (status) { - dbg - (" usb_submit_urb(read bulk) failed, status = %d", - status); - } - } - return; -} - -static int get_lsr_info(struct tty_struct *tty, - struct ATENINTL_port *ATEN2011_port, - unsigned int __user *value) -{ - int count; - unsigned int result = 0; - - count = ATEN2011_chars_in_buffer(tty); - if (count == 0) { - dbg("%s -- Empty", __func__); - result = TIOCSER_TEMT; - } - - if (copy_to_user(value, &result, sizeof(int))) - return -EFAULT; - return 0; -} - -static int get_number_bytes_avail(struct tty_struct *tty, - struct ATENINTL_port *ATEN2011_port, - unsigned int __user *value) -{ - unsigned int result = 0; - - if (!tty) - return -ENOIOCTLCMD; - - result = tty->read_cnt; - - dbg("%s(%d) = %d", __func__, ATEN2011_port->port->number, result); - if (copy_to_user(value, &result, sizeof(int))) - return -EFAULT; - - return -ENOIOCTLCMD; -} - -static int set_modem_info(struct ATENINTL_port *ATEN2011_port, unsigned int cmd, - unsigned int __user *value) -{ - unsigned int mcr; - unsigned int arg; - __u16 Data; - int status; - struct usb_serial_port *port; - - if (ATEN2011_port == NULL) - return -1; - - port = (struct usb_serial_port *)ATEN2011_port->port; - - mcr = ATEN2011_port->shadowMCR; - - if (copy_from_user(&arg, value, sizeof(int))) - return -EFAULT; - - switch (cmd) { - case TIOCMBIS: - if (arg & TIOCM_RTS) - mcr |= MCR_RTS; - if (arg & TIOCM_DTR) - mcr |= MCR_RTS; - if (arg & TIOCM_LOOP) - mcr |= MCR_LOOPBACK; - break; - - case TIOCMBIC: - if (arg & TIOCM_RTS) - mcr &= ~MCR_RTS; - if (arg & TIOCM_DTR) - mcr &= ~MCR_RTS; - if (arg & TIOCM_LOOP) - mcr &= ~MCR_LOOPBACK; - break; - - case TIOCMSET: - /* turn off the RTS and DTR and LOOPBACK - * and then only turn on what was asked to */ - mcr &= ~(MCR_RTS | MCR_DTR | MCR_LOOPBACK); - mcr |= ((arg & TIOCM_RTS) ? MCR_RTS : 0); - mcr |= ((arg & TIOCM_DTR) ? MCR_DTR : 0); - mcr |= ((arg & TIOCM_LOOP) ? MCR_LOOPBACK : 0); - break; - } - - ATEN2011_port->shadowMCR = mcr; - - Data = ATEN2011_port->shadowMCR; - status = set_uart_reg(port, MODEM_CONTROL_REGISTER, Data); - if (status < 0) { - dbg("setting MODEM_CONTROL_REGISTER Failed"); - return -1; - } - - return 0; -} - -static int get_modem_info(struct ATENINTL_port *ATEN2011_port, - unsigned int __user *value) -{ - unsigned int result = 0; - __u16 msr; - unsigned int mcr = ATEN2011_port->shadowMCR; - int status; - - status = get_uart_reg(ATEN2011_port->port, MODEM_STATUS_REGISTER, &msr); - result = ((mcr & MCR_DTR) ? TIOCM_DTR : 0) /* 0x002 */ - |((mcr & MCR_RTS) ? TIOCM_RTS : 0) /* 0x004 */ - |((msr & ATEN2011_MSR_CTS) ? TIOCM_CTS : 0) /* 0x020 */ - |((msr & ATEN2011_MSR_CD) ? TIOCM_CAR : 0) /* 0x040 */ - |((msr & ATEN2011_MSR_RI) ? TIOCM_RI : 0) /* 0x080 */ - |((msr & ATEN2011_MSR_DSR) ? TIOCM_DSR : 0); /* 0x100 */ - - dbg("%s -- %x", __func__, result); - - if (copy_to_user(value, &result, sizeof(int))) - return -EFAULT; - return 0; -} - -static int get_serial_info(struct ATENINTL_port *ATEN2011_port, - struct serial_struct __user *retinfo) -{ - struct serial_struct tmp; - - if (ATEN2011_port == NULL) - return -1; - - if (!retinfo) - return -EFAULT; - - memset(&tmp, 0, sizeof(tmp)); - - tmp.type = PORT_16550A; - tmp.line = ATEN2011_port->port->serial->minor; - if (tmp.line == SERIAL_TTY_NO_MINOR) - tmp.line = 0; - tmp.port = ATEN2011_port->port->number; - tmp.irq = 0; - tmp.flags = ASYNC_SKIP_TEST | ASYNC_AUTO_IRQ; - tmp.xmit_fifo_size = NUM_URBS * URB_TRANSFER_BUFFER_SIZE; - tmp.baud_base = 9600; - tmp.close_delay = 5 * HZ; - tmp.closing_wait = 30 * HZ; - - if (copy_to_user(retinfo, &tmp, sizeof(*retinfo))) - return -EFAULT; - return 0; -} - -static int ATEN2011_ioctl(struct tty_struct *tty, struct file *file, - unsigned int cmd, unsigned long arg) -{ - struct usb_serial_port *port = tty->driver_data; - struct ATENINTL_port *ATEN2011_port; - struct async_icount cnow; - struct async_icount cprev; - struct serial_icounter_struct icount; - int ATENret = 0; - unsigned int __user *user_arg = (unsigned int __user *)arg; - - ATEN2011_port = usb_get_serial_port_data(port); - - if (ATEN2011_port == NULL) - return -1; - - dbg("%s - port %d, cmd = 0x%x", __func__, port->number, cmd); - - switch (cmd) { - /* return number of bytes available */ - - case TIOCINQ: - dbg("%s (%d) TIOCINQ", __func__, port->number); - return get_number_bytes_avail(tty, ATEN2011_port, user_arg); - break; - - case TIOCOUTQ: - dbg("%s (%d) TIOCOUTQ", __func__, port->number); - return put_user(ATEN2011_chars_in_buffer(tty), user_arg); - break; - - case TIOCSERGETLSR: - dbg("%s (%d) TIOCSERGETLSR", __func__, port->number); - return get_lsr_info(tty, ATEN2011_port, user_arg); - return 0; - - case TIOCMBIS: - case TIOCMBIC: - case TIOCMSET: - dbg("%s (%d) TIOCMSET/TIOCMBIC/TIOCMSET", __func__, - port->number); - ATENret = set_modem_info(ATEN2011_port, cmd, user_arg); - return ATENret; - - case TIOCMGET: - dbg("%s (%d) TIOCMGET", __func__, port->number); - return get_modem_info(ATEN2011_port, user_arg); - - case TIOCGSERIAL: - dbg("%s (%d) TIOCGSERIAL", __func__, port->number); - return get_serial_info(ATEN2011_port, - (struct serial_struct __user *)arg); - - case TIOCSSERIAL: - dbg("%s (%d) TIOCSSERIAL", __func__, port->number); - break; - - case TIOCMIWAIT: - dbg("%s (%d) TIOCMIWAIT", __func__, port->number); - cprev = ATEN2011_port->icount; - while (1) { - /* see if a signal did it */ - if (signal_pending(current)) - return -ERESTARTSYS; - cnow = ATEN2011_port->icount; - if (cnow.rng == cprev.rng && cnow.dsr == cprev.dsr && - cnow.dcd == cprev.dcd && cnow.cts == cprev.cts) - return -EIO; /* no change => error */ - if (((arg & TIOCM_RNG) && (cnow.rng != cprev.rng)) || - ((arg & TIOCM_DSR) && (cnow.dsr != cprev.dsr)) || - ((arg & TIOCM_CD) && (cnow.dcd != cprev.dcd)) || - ((arg & TIOCM_CTS) && (cnow.cts != cprev.cts))) { - return 0; - } - cprev = cnow; - } - /* NOTREACHED */ - break; - - case TIOCGICOUNT: - cnow = ATEN2011_port->icount; - icount.cts = cnow.cts; - icount.dsr = cnow.dsr; - icount.rng = cnow.rng; - icount.dcd = cnow.dcd; - icount.rx = cnow.rx; - icount.tx = cnow.tx; - icount.frame = cnow.frame; - icount.overrun = cnow.overrun; - icount.parity = cnow.parity; - icount.brk = cnow.brk; - icount.buf_overrun = cnow.buf_overrun; - - dbg("%s (%d) TIOCGICOUNT RX=%d, TX=%d", __func__, - port->number, icount.rx, icount.tx); - if (copy_to_user((void __user *)arg, &icount, sizeof(icount))) - return -EFAULT; - return 0; - - default: - break; - } - - return -ENOIOCTLCMD; -} - -static int ATEN2011_calc_baud_rate_divisor(int baudRate, int *divisor, - __u16 *clk_sel_val) -{ - dbg("%s - %d", __func__, baudRate); - - if (baudRate <= 115200) { - *divisor = 115200 / baudRate; - *clk_sel_val = 0x0; - } - if ((baudRate > 115200) && (baudRate <= 230400)) { - *divisor = 230400 / baudRate; - *clk_sel_val = 0x10; - } else if ((baudRate > 230400) && (baudRate <= 403200)) { - *divisor = 403200 / baudRate; - *clk_sel_val = 0x20; - } else if ((baudRate > 403200) && (baudRate <= 460800)) { - *divisor = 460800 / baudRate; - *clk_sel_val = 0x30; - } else if ((baudRate > 460800) && (baudRate <= 806400)) { - *divisor = 806400 / baudRate; - *clk_sel_val = 0x40; - } else if ((baudRate > 806400) && (baudRate <= 921600)) { - *divisor = 921600 / baudRate; - *clk_sel_val = 0x50; - } else if ((baudRate > 921600) && (baudRate <= 1572864)) { - *divisor = 1572864 / baudRate; - *clk_sel_val = 0x60; - } else if ((baudRate > 1572864) && (baudRate <= 3145728)) { - *divisor = 3145728 / baudRate; - *clk_sel_val = 0x70; - } - return 0; -} - -static int ATEN2011_send_cmd_write_baud_rate(struct ATENINTL_port - *ATEN2011_port, int baudRate) -{ - int divisor = 0; - int status; - __u16 Data; - unsigned char number; - __u16 clk_sel_val; - struct usb_serial_port *port; - int minor; - - if (ATEN2011_port == NULL) - return -1; - - port = (struct usb_serial_port *)ATEN2011_port->port; - - dbg("%s", "Entering .......... "); - - minor = ATEN2011_port->port->serial->minor; - if (minor == SERIAL_TTY_NO_MINOR) - minor = 0; - number = ATEN2011_port->port->number - minor; - - dbg("%s - port = %d, baud = %d", __func__, - ATEN2011_port->port->number, baudRate); - /* reset clk_uart_sel in spregOffset */ - if (baudRate > 115200) { -#ifdef HW_flow_control - /* - * NOTE: need to see the pther register to modify - * setting h/w flow control bit to 1; - */ - /* Data = ATEN2011_port->shadowMCR; */ - Data = 0x2b; - ATEN2011_port->shadowMCR = Data; - status = set_uart_reg(port, MODEM_CONTROL_REGISTER, Data); - if (status < 0) { - dbg("Writing spreg failed in set_serial_baud"); - return -1; - } -#endif - - } else { -#ifdef HW_flow_control - /* setting h/w flow control bit to 0; */ - /* Data = ATEN2011_port->shadowMCR; */ - Data = 0xb; - ATEN2011_port->shadowMCR = Data; - status = set_uart_reg(port, MODEM_CONTROL_REGISTER, Data); - if (status < 0) { - dbg("Writing spreg failed in set_serial_baud"); - return -1; - } -#endif - - } - - if (1) /* baudRate <= 115200) */ { - clk_sel_val = 0x0; - Data = 0x0; - status = - ATEN2011_calc_baud_rate_divisor(baudRate, &divisor, - &clk_sel_val); - status = get_reg_sync(port, ATEN2011_port->SpRegOffset, &Data); - if (status < 0) { - dbg("reading spreg failed in set_serial_baud"); - return -1; - } - Data = (Data & 0x8f) | clk_sel_val; - status = set_reg_sync(port, ATEN2011_port->SpRegOffset, Data); - if (status < 0) { - dbg("Writing spreg failed in set_serial_baud"); - return -1; - } - /* Calculate the Divisor */ - - if (status) { - err("%s - bad baud rate", __func__); - dbg("%s", "bad baud rate"); - return status; - } - /* Enable access to divisor latch */ - Data = ATEN2011_port->shadowLCR | SERIAL_LCR_DLAB; - ATEN2011_port->shadowLCR = Data; - set_uart_reg(port, LINE_CONTROL_REGISTER, Data); - - /* Write the divisor */ - Data = (unsigned char)(divisor & 0xff); - dbg("set_serial_baud Value to write DLL is %x", Data); - set_uart_reg(port, DIVISOR_LATCH_LSB, Data); - - Data = (unsigned char)((divisor & 0xff00) >> 8); - dbg("set_serial_baud Value to write DLM is %x", Data); - set_uart_reg(port, DIVISOR_LATCH_MSB, Data); - - /* Disable access to divisor latch */ - Data = ATEN2011_port->shadowLCR & ~SERIAL_LCR_DLAB; - ATEN2011_port->shadowLCR = Data; - set_uart_reg(port, LINE_CONTROL_REGISTER, Data); - - } - - return status; -} - -static void ATEN2011_change_port_settings(struct tty_struct *tty, - struct ATENINTL_port *ATEN2011_port, - struct ktermios *old_termios) -{ - int baud; - unsigned cflag; - unsigned iflag; - __u8 lData; - __u8 lParity; - __u8 lStop; - int status; - __u16 Data; - struct usb_serial_port *port; - struct usb_serial *serial; - - if (ATEN2011_port == NULL) - return; - - port = (struct usb_serial_port *)ATEN2011_port->port; - - serial = port->serial; - - dbg("%s - port %d", __func__, ATEN2011_port->port->number); - - if (!ATEN2011_port->open) { - dbg("%s - port not opened", __func__); - return; - } - - if ((!tty) || (!tty->termios)) { - dbg("%s - no tty structures", __func__); - return; - } - - dbg("%s", "Entering .......... "); - - lData = LCR_BITS_8; - lStop = LCR_STOP_1; - lParity = LCR_PAR_NONE; - - cflag = tty->termios->c_cflag; - iflag = tty->termios->c_iflag; - - /* Change the number of bits */ - - /* COMMENT1: the below Line"if(cflag & CSIZE)" is added for the errors we get for serial loop data test i.e serial_loopback.pl -v */ - /* if(cflag & CSIZE) */ - { - switch (cflag & CSIZE) { - case CS5: - lData = LCR_BITS_5; - break; - - case CS6: - lData = LCR_BITS_6; - break; - - case CS7: - lData = LCR_BITS_7; - break; - default: - case CS8: - lData = LCR_BITS_8; - break; - } - } - /* Change the Parity bit */ - if (cflag & PARENB) { - if (cflag & PARODD) { - lParity = LCR_PAR_ODD; - dbg("%s - parity = odd", __func__); - } else { - lParity = LCR_PAR_EVEN; - dbg("%s - parity = even", __func__); - } - - } else { - dbg("%s - parity = none", __func__); - } - - if (cflag & CMSPAR) - lParity = lParity | 0x20; - - /* Change the Stop bit */ - if (cflag & CSTOPB) { - lStop = LCR_STOP_2; - dbg("%s - stop bits = 2", __func__); - } else { - lStop = LCR_STOP_1; - dbg("%s - stop bits = 1", __func__); - } - - /* Update the LCR with the correct value */ - ATEN2011_port->shadowLCR &= - ~(LCR_BITS_MASK | LCR_STOP_MASK | LCR_PAR_MASK); - ATEN2011_port->shadowLCR |= (lData | lParity | lStop); - - dbg - ("ATEN2011_change_port_settings ATEN2011_port->shadowLCR is %x", - ATEN2011_port->shadowLCR); - /* Disable Interrupts */ - Data = 0x00; - set_uart_reg(port, INTERRUPT_ENABLE_REGISTER, Data); - - Data = 0x00; - set_uart_reg(port, FIFO_CONTROL_REGISTER, Data); - - Data = 0xcf; - set_uart_reg(port, FIFO_CONTROL_REGISTER, Data); - - /* Send the updated LCR value to the ATEN2011 */ - Data = ATEN2011_port->shadowLCR; - - set_uart_reg(port, LINE_CONTROL_REGISTER, Data); - - Data = 0x00b; - ATEN2011_port->shadowMCR = Data; - set_uart_reg(port, MODEM_CONTROL_REGISTER, Data); - Data = 0x00b; - set_uart_reg(port, MODEM_CONTROL_REGISTER, Data); - - /* set up the MCR register and send it to the ATEN2011 */ - - ATEN2011_port->shadowMCR = MCR_MASTER_IE; - if (cflag & CBAUD) - ATEN2011_port->shadowMCR |= (MCR_DTR | MCR_RTS); - - if (cflag & CRTSCTS) - ATEN2011_port->shadowMCR |= (MCR_XON_ANY); - else - ATEN2011_port->shadowMCR &= ~(MCR_XON_ANY); - - Data = ATEN2011_port->shadowMCR; - set_uart_reg(port, MODEM_CONTROL_REGISTER, Data); - - /* Determine divisor based on baud rate */ - baud = tty_get_baud_rate(tty); - - if (!baud) { - /* pick a default, any default... */ - dbg("%s", "Picked default baud..."); - baud = 9600; - } - - dbg("%s - baud rate = %d", __func__, baud); - status = ATEN2011_send_cmd_write_baud_rate(ATEN2011_port, baud); - - /* Enable Interrupts */ - Data = 0x0c; - set_uart_reg(port, INTERRUPT_ENABLE_REGISTER, Data); - - if (ATEN2011_port->read_urb->status != -EINPROGRESS) { - ATEN2011_port->read_urb->dev = serial->dev; - - status = usb_submit_urb(ATEN2011_port->read_urb, GFP_ATOMIC); - - if (status) { - dbg - (" usb_submit_urb(read bulk) failed, status = %d", - status); - } - } - dbg - ("ATEN2011_change_port_settings ATEN2011_port->shadowLCR is End %x", - ATEN2011_port->shadowLCR); - - return; -} - -static int ATEN2011_calc_num_ports(struct usb_serial *serial) -{ - - __u16 Data = 0x00; - int ret = 0; - int ATEN2011_2or4ports; - ret = usb_control_msg(serial->dev, usb_rcvctrlpipe(serial->dev, 0), - ATEN_RDREQ, ATEN_RD_RTYPE, 0, GPIO_REGISTER, - &Data, VENDOR_READ_LENGTH, ATEN_WDR_TIMEOUT); - -/* ghostgum: here is where the problem appears to bet */ -/* Which of the following are needed? */ -/* Greg used the serial->type->num_ports=2 */ -/* But the code in the ATEN2011_open relies on serial->num_ports=2 */ - if ((Data & 0x01) == 0) { - ATEN2011_2or4ports = 2; - serial->type->num_ports = 2; - serial->num_ports = 2; - } - /* else if(serial->interface->cur_altsetting->desc.bNumEndpoints == 9) */ - else { - ATEN2011_2or4ports = 4; - serial->type->num_ports = 4; - serial->num_ports = 4; - - } - - return ATEN2011_2or4ports; -} - -static int ATEN2011_startup(struct usb_serial *serial) -{ - struct ATENINTL_serial *ATEN2011_serial; - struct ATENINTL_port *ATEN2011_port; - struct usb_device *dev; - int i, status; - int minor; - - __u16 Data; - dbg("%s", " ATEN2011_startup :entering.........."); - - if (!serial) { - dbg("%s", "Invalid Handler"); - return -1; - } - - dev = serial->dev; - - dbg("%s", "Entering..."); - - /* create our private serial structure */ - ATEN2011_serial = kzalloc(sizeof(struct ATENINTL_serial), GFP_KERNEL); - if (ATEN2011_serial == NULL) { - err("%s - Out of memory", __func__); - return -ENOMEM; - } - - /* resetting the private structure field values to zero */ - memset(ATEN2011_serial, 0, sizeof(struct ATENINTL_serial)); - - ATEN2011_serial->serial = serial; - /* initilize status polling flag to 0 */ - ATEN2011_serial->status_polling_started = 0; - - usb_set_serial_data(serial, ATEN2011_serial); - ATEN2011_serial->ATEN2011_spectrum_2or4ports = - ATEN2011_calc_num_ports(serial); - /* we set up the pointers to the endpoints in the ATEN2011_open * - * function, as the structures aren't created yet. */ - - /* set up port private structures */ - for (i = 0; i < serial->num_ports; ++i) { - ATEN2011_port = - kmalloc(sizeof(struct ATENINTL_port), GFP_KERNEL); - if (ATEN2011_port == NULL) { - err("%s - Out of memory", __func__); - usb_set_serial_data(serial, NULL); - kfree(ATEN2011_serial); - return -ENOMEM; - } - memset(ATEN2011_port, 0, sizeof(struct ATENINTL_port)); - - /* - * Initialize all port interrupt end point to port 0 - * int endpoint. Our device has only one interrupt end point - * comman to all port - */ - /* serial->port[i]->interrupt_in_endpointAddress = serial->port[0]->interrupt_in_endpointAddress; */ - - ATEN2011_port->port = serial->port[i]; - usb_set_serial_port_data(serial->port[i], ATEN2011_port); - - minor = serial->port[i]->serial->minor; - if (minor == SERIAL_TTY_NO_MINOR) - minor = 0; - ATEN2011_port->port_num = - ((serial->port[i]->number - minor) + 1); - - if (ATEN2011_port->port_num == 1) { - ATEN2011_port->SpRegOffset = 0x0; - ATEN2011_port->ControlRegOffset = 0x1; - ATEN2011_port->DcrRegOffset = 0x4; - } else if ((ATEN2011_port->port_num == 2) - && (ATEN2011_serial->ATEN2011_spectrum_2or4ports == - 4)) { - ATEN2011_port->SpRegOffset = 0x8; - ATEN2011_port->ControlRegOffset = 0x9; - ATEN2011_port->DcrRegOffset = 0x16; - } else if ((ATEN2011_port->port_num == 2) - && (ATEN2011_serial->ATEN2011_spectrum_2or4ports == - 2)) { - ATEN2011_port->SpRegOffset = 0xa; - ATEN2011_port->ControlRegOffset = 0xb; - ATEN2011_port->DcrRegOffset = 0x19; - } else if ((ATEN2011_port->port_num == 3) - && (ATEN2011_serial->ATEN2011_spectrum_2or4ports == - 4)) { - ATEN2011_port->SpRegOffset = 0xa; - ATEN2011_port->ControlRegOffset = 0xb; - ATEN2011_port->DcrRegOffset = 0x19; - } else if ((ATEN2011_port->port_num == 4) - && (ATEN2011_serial->ATEN2011_spectrum_2or4ports == - 4)) { - ATEN2011_port->SpRegOffset = 0xc; - ATEN2011_port->ControlRegOffset = 0xd; - ATEN2011_port->DcrRegOffset = 0x1c; - } - - usb_set_serial_port_data(serial->port[i], ATEN2011_port); - - /* enable rx_disable bit in control register */ - - status = get_reg_sync(serial->port[i], - ATEN2011_port->ControlRegOffset, &Data); - if (status < 0) { - dbg("Reading ControlReg failed status-0x%x", - status); - break; - } else - dbg - ("ControlReg Reading success val is %x, status%d", - Data, status); - Data |= 0x08; /* setting driver done bit */ - Data |= 0x04; /* sp1_bit to have cts change reflect in modem status reg */ - - /* Data |= 0x20; */ /* rx_disable bit */ - status = set_reg_sync(serial->port[i], - ATEN2011_port->ControlRegOffset, Data); - if (status < 0) { - dbg - ("Writing ControlReg failed(rx_disable) status-0x%x", - status); - break; - } else - dbg - ("ControlReg Writing success(rx_disable) status%d", - status); - - /* - * Write default values in DCR (i.e 0x01 in DCR0, 0x05 in DCR2 - * and 0x24 in DCR3 - */ - Data = 0x01; - status = set_reg_sync(serial->port[i], - (__u16)(ATEN2011_port->DcrRegOffset + 0), - Data); - if (status < 0) { - dbg("Writing DCR0 failed status-0x%x", status); - break; - } else - dbg("DCR0 Writing success status%d", status); - - Data = 0x05; - status = set_reg_sync(serial->port[i], - (__u16)(ATEN2011_port->DcrRegOffset + 1), - Data); - if (status < 0) { - dbg("Writing DCR1 failed status-0x%x", status); - break; - } else - dbg("DCR1 Writing success status%d", status); - - Data = 0x24; - status = set_reg_sync(serial->port[i], - (__u16)(ATEN2011_port->DcrRegOffset + 2), - Data); - if (status < 0) { - dbg("Writing DCR2 failed status-0x%x", status); - break; - } else - dbg("DCR2 Writing success status%d", status); - - /* write values in clkstart0x0 and clkmulti 0x20 */ - Data = 0x0; - status = set_reg_sync(serial->port[i], CLK_START_VALUE_REGISTER, - Data); - if (status < 0) { - dbg - ("Writing CLK_START_VALUE_REGISTER failed status-0x%x", - status); - break; - } else - dbg - ("CLK_START_VALUE_REGISTER Writing success status%d", - status); - - Data = 0x20; - status = set_reg_sync(serial->port[i], CLK_MULTI_REGISTER, - Data); - if (status < 0) { - dbg - ("Writing CLK_MULTI_REGISTER failed status-0x%x", - status); - break; - } else - dbg("CLK_MULTI_REGISTER Writing success status%d", - status); - - /* Zero Length flag register */ - if ((ATEN2011_port->port_num != 1) - && (ATEN2011_serial->ATEN2011_spectrum_2or4ports == 2)) { - - Data = 0xff; - status = set_reg_sync(serial->port[i], - (__u16)(ZLP_REG1 + ((__u16)ATEN2011_port->port_num)), - Data); - dbg("ZLIP offset%x", - (__u16) (ZLP_REG1 + - ((__u16) ATEN2011_port->port_num))); - if (status < 0) { - dbg - ("Writing ZLP_REG%d failed status-0x%x", - i + 2, status); - break; - } else - dbg("ZLP_REG%d Writing success status%d", - i + 2, status); - } else { - Data = 0xff; - status = set_reg_sync(serial->port[i], - (__u16)(ZLP_REG1 + ((__u16)ATEN2011_port->port_num) - 0x1), - Data); - dbg("ZLIP offset%x", - (__u16) (ZLP_REG1 + - ((__u16) ATEN2011_port->port_num) - - 0x1)); - if (status < 0) { - dbg - ("Writing ZLP_REG%d failed status-0x%x", - i + 1, status); - break; - } else - dbg("ZLP_REG%d Writing success status%d", - i + 1, status); - - } - ATEN2011_port->control_urb = usb_alloc_urb(0, GFP_ATOMIC); - ATEN2011_port->ctrl_buf = kmalloc(16, GFP_KERNEL); - - } - - /* Zero Length flag enable */ - Data = 0x0f; - status = set_reg_sync(serial->port[0], ZLP_REG5, Data); - if (status < 0) { - dbg("Writing ZLP_REG5 failed status-0x%x", status); - return -1; - } else - dbg("ZLP_REG5 Writing success status%d", status); - - /* setting configuration feature to one */ - usb_control_msg(serial->dev, usb_sndctrlpipe(serial->dev, 0), - (__u8) 0x03, 0x00, 0x01, 0x00, NULL, 0x00, 5 * HZ); - return 0; -} - -static void ATEN2011_release(struct usb_serial *serial) -{ - int i; - struct ATENINTL_port *ATEN2011_port; - - /* check for the ports to be closed,close the ports and disconnect */ - - /* free private structure allocated for serial port * - * stop reads and writes on all ports */ - - for (i = 0; i < serial->num_ports; ++i) { - ATEN2011_port = usb_get_serial_port_data(serial->port[i]); - kfree(ATEN2011_port->ctrl_buf); - usb_kill_urb(ATEN2011_port->control_urb); - kfree(ATEN2011_port); - usb_set_serial_port_data(serial->port[i], NULL); - } - - /* free private structure allocated for serial device */ - - kfree(usb_get_serial_data(serial)); - usb_set_serial_data(serial, NULL); -} - -static struct usb_serial_driver aten_serial_driver = { - .driver = { - .owner = THIS_MODULE, - .name = "aten2011", - }, - .description = DRIVER_DESC, - .id_table = id_table, - .open = ATEN2011_open, - .close = ATEN2011_close, - .write = ATEN2011_write, - .write_room = ATEN2011_write_room, - .chars_in_buffer = ATEN2011_chars_in_buffer, - .throttle = ATEN2011_throttle, - .unthrottle = ATEN2011_unthrottle, - .calc_num_ports = ATEN2011_calc_num_ports, - - .ioctl = ATEN2011_ioctl, - .set_termios = ATEN2011_set_termios, - .break_ctl = ATEN2011_break, - .tiocmget = ATEN2011_tiocmget, - .tiocmset = ATEN2011_tiocmset, - .attach = ATEN2011_startup, - .release = ATEN2011_release, - .read_bulk_callback = ATEN2011_bulk_in_callback, - .read_int_callback = ATEN2011_interrupt_callback, -}; - -static struct usb_driver aten_driver = { - .name = "aten2011", - .probe = usb_serial_probe, - .disconnect = usb_serial_disconnect, - .id_table = id_table, -}; - -static int __init aten_init(void) -{ - int retval; - - /* Register with the usb serial */ - retval = usb_serial_register(&aten_serial_driver); - if (retval) - return retval; - - printk(KERN_INFO KBUILD_MODNAME ":" - DRIVER_DESC " " DRIVER_VERSION "\n"); - - /* Register with the usb */ - retval = usb_register(&aten_driver); - if (retval) - usb_serial_deregister(&aten_serial_driver); - - return retval; -} - -static void __exit aten_exit(void) -{ - usb_deregister(&aten_driver); - usb_serial_deregister(&aten_serial_driver); -} - -module_init(aten_init); -module_exit(aten_exit); - -/* Module information */ -MODULE_DESCRIPTION(DRIVER_DESC); -MODULE_LICENSE("GPL"); - -MODULE_PARM_DESC(debug, "Debug enabled or not"); diff --git a/drivers/staging/udlfb/udlfb.c b/drivers/staging/udlfb/udlfb.c index 0ab9d15f343..f5416af1e90 100644 --- a/drivers/staging/udlfb/udlfb.c +++ b/drivers/staging/udlfb/udlfb.c @@ -21,6 +21,7 @@ #include <linux/mm.h> #include <linux/fb.h> #include <linux/mutex.h> +#include <linux/vmalloc.h> #include "udlfb.h" diff --git a/drivers/usb/core/config.c b/drivers/usb/core/config.c index 24dfb33f90c..a16c538d013 100644 --- a/drivers/usb/core/config.c +++ b/drivers/usb/core/config.c @@ -80,38 +80,18 @@ static int usb_parse_ss_endpoint_companion(struct device *ddev, int cfgno, int max_tx; int i; - /* Allocate space for the SS endpoint companion descriptor */ - ep->ss_ep_comp = kzalloc(sizeof(struct usb_host_ss_ep_comp), - GFP_KERNEL); - if (!ep->ss_ep_comp) - return -ENOMEM; desc = (struct usb_ss_ep_comp_descriptor *) buffer; if (desc->bDescriptorType != USB_DT_SS_ENDPOINT_COMP) { dev_warn(ddev, "No SuperSpeed endpoint companion for config %d " " interface %d altsetting %d ep %d: " "using minimum values\n", cfgno, inum, asnum, ep->desc.bEndpointAddress); - ep->ss_ep_comp->desc.bLength = USB_DT_SS_EP_COMP_SIZE; - ep->ss_ep_comp->desc.bDescriptorType = USB_DT_SS_ENDPOINT_COMP; - ep->ss_ep_comp->desc.bMaxBurst = 0; - /* - * Leave bmAttributes as zero, which will mean no streams for - * bulk, and isoc won't support multiple bursts of packets. - * With bursts of only one packet, and a Mult of 1, the max - * amount of data moved per endpoint service interval is one - * packet. - */ - if (usb_endpoint_xfer_isoc(&ep->desc) || - usb_endpoint_xfer_int(&ep->desc)) - ep->ss_ep_comp->desc.wBytesPerInterval = - ep->desc.wMaxPacketSize; /* * The next descriptor is for an Endpoint or Interface, * no extra descriptors to copy into the companion structure, * and we didn't eat up any of the buffer. */ - retval = 0; - goto valid; + return 0; } memcpy(&ep->ss_ep_comp->desc, desc, USB_DT_SS_EP_COMP_SIZE); desc = &ep->ss_ep_comp->desc; @@ -320,6 +300,28 @@ static int usb_parse_endpoint(struct device *ddev, int cfgno, int inum, buffer += i; size -= i; + /* Allocate space for the SS endpoint companion descriptor */ + endpoint->ss_ep_comp = kzalloc(sizeof(struct usb_host_ss_ep_comp), + GFP_KERNEL); + if (!endpoint->ss_ep_comp) + return -ENOMEM; + + /* Fill in some default values (may be overwritten later) */ + endpoint->ss_ep_comp->desc.bLength = USB_DT_SS_EP_COMP_SIZE; + endpoint->ss_ep_comp->desc.bDescriptorType = USB_DT_SS_ENDPOINT_COMP; + endpoint->ss_ep_comp->desc.bMaxBurst = 0; + /* + * Leave bmAttributes as zero, which will mean no streams for + * bulk, and isoc won't support multiple bursts of packets. + * With bursts of only one packet, and a Mult of 1, the max + * amount of data moved per endpoint service interval is one + * packet. + */ + if (usb_endpoint_xfer_isoc(&endpoint->desc) || + usb_endpoint_xfer_int(&endpoint->desc)) + endpoint->ss_ep_comp->desc.wBytesPerInterval = + endpoint->desc.wMaxPacketSize; + if (size > 0) { retval = usb_parse_ss_endpoint_companion(ddev, cfgno, inum, asnum, endpoint, num_ep, buffer, @@ -329,6 +331,10 @@ static int usb_parse_endpoint(struct device *ddev, int cfgno, int inum, retval = buffer - buffer0; } } else { + dev_warn(ddev, "config %d interface %d altsetting %d " + "endpoint 0x%X has no " + "SuperSpeed companion descriptor\n", + cfgno, inum, asnum, d->bEndpointAddress); retval = buffer - buffer0; } } else { diff --git a/drivers/usb/host/ehci-orion.c b/drivers/usb/host/ehci-orion.c index dc2ac613a9d..1d283e1b2b8 100644 --- a/drivers/usb/host/ehci-orion.c +++ b/drivers/usb/host/ehci-orion.c @@ -105,6 +105,7 @@ static int ehci_orion_setup(struct usb_hcd *hcd) struct ehci_hcd *ehci = hcd_to_ehci(hcd); int retval; + ehci_reset(ehci); retval = ehci_halt(ehci); if (retval) return retval; @@ -118,7 +119,6 @@ static int ehci_orion_setup(struct usb_hcd *hcd) hcd->has_tt = 1; - ehci_reset(ehci); ehci_port_power(ehci, 0); return retval; diff --git a/drivers/usb/host/ohci-omap.c b/drivers/usb/host/ohci-omap.c index f3aaba35e91..83cbecd2a1e 100644 --- a/drivers/usb/host/ohci-omap.c +++ b/drivers/usb/host/ohci-omap.c @@ -282,6 +282,7 @@ static int ohci_omap_init(struct usb_hcd *hcd) static void ohci_omap_stop(struct usb_hcd *hcd) { dev_dbg(hcd->self.controller, "stopping USB Controller\n"); + ohci_stop(hcd); omap_ohci_clock_power(0); } diff --git a/drivers/usb/host/xhci-dbg.c b/drivers/usb/host/xhci-dbg.c index 2501c571f85..705e3432415 100644 --- a/drivers/usb/host/xhci-dbg.c +++ b/drivers/usb/host/xhci-dbg.c @@ -173,6 +173,7 @@ void xhci_print_ir_set(struct xhci_hcd *xhci, struct xhci_intr_reg *ir_set, int { void *addr; u32 temp; + u64 temp_64; addr = &ir_set->irq_pending; temp = xhci_readl(xhci, addr); @@ -200,25 +201,15 @@ void xhci_print_ir_set(struct xhci_hcd *xhci, struct xhci_intr_reg *ir_set, int xhci_dbg(xhci, " WARN: %p: ir_set.rsvd = 0x%x\n", addr, (unsigned int)temp); - addr = &ir_set->erst_base[0]; - temp = xhci_readl(xhci, addr); - xhci_dbg(xhci, " %p: ir_set.erst_base[0] = 0x%x\n", - addr, (unsigned int) temp); - - addr = &ir_set->erst_base[1]; - temp = xhci_readl(xhci, addr); - xhci_dbg(xhci, " %p: ir_set.erst_base[1] = 0x%x\n", - addr, (unsigned int) temp); + addr = &ir_set->erst_base; + temp_64 = xhci_read_64(xhci, addr); + xhci_dbg(xhci, " %p: ir_set.erst_base = @%08llx\n", + addr, temp_64); - addr = &ir_set->erst_dequeue[0]; - temp = xhci_readl(xhci, addr); - xhci_dbg(xhci, " %p: ir_set.erst_dequeue[0] = 0x%x\n", - addr, (unsigned int) temp); - - addr = &ir_set->erst_dequeue[1]; - temp = xhci_readl(xhci, addr); - xhci_dbg(xhci, " %p: ir_set.erst_dequeue[1] = 0x%x\n", - addr, (unsigned int) temp); + addr = &ir_set->erst_dequeue; + temp_64 = xhci_read_64(xhci, addr); + xhci_dbg(xhci, " %p: ir_set.erst_dequeue = @%08llx\n", + addr, temp_64); } void xhci_print_run_regs(struct xhci_hcd *xhci) @@ -268,8 +259,7 @@ void xhci_debug_trb(struct xhci_hcd *xhci, union xhci_trb *trb) xhci_dbg(xhci, "Link TRB:\n"); xhci_print_trb_offsets(xhci, trb); - address = trb->link.segment_ptr[0] + - (((u64) trb->link.segment_ptr[1]) << 32); + address = trb->link.segment_ptr; xhci_dbg(xhci, "Next ring segment DMA address = 0x%llx\n", address); xhci_dbg(xhci, "Interrupter target = 0x%x\n", @@ -282,8 +272,7 @@ void xhci_debug_trb(struct xhci_hcd *xhci, union xhci_trb *trb) (unsigned int) (trb->link.control & TRB_NO_SNOOP)); break; case TRB_TYPE(TRB_TRANSFER): - address = trb->trans_event.buffer[0] + - (((u64) trb->trans_event.buffer[1]) << 32); + address = trb->trans_event.buffer; /* * FIXME: look at flags to figure out if it's an address or if * the data is directly in the buffer field. @@ -291,8 +280,7 @@ void xhci_debug_trb(struct xhci_hcd *xhci, union xhci_trb *trb) xhci_dbg(xhci, "DMA address or buffer contents= %llu\n", address); break; case TRB_TYPE(TRB_COMPLETION): - address = trb->event_cmd.cmd_trb[0] + - (((u64) trb->event_cmd.cmd_trb[1]) << 32); + address = trb->event_cmd.cmd_trb; xhci_dbg(xhci, "Command TRB pointer = %llu\n", address); xhci_dbg(xhci, "Completion status = %u\n", (unsigned int) GET_COMP_CODE(trb->event_cmd.status)); @@ -328,8 +316,8 @@ void xhci_debug_segment(struct xhci_hcd *xhci, struct xhci_segment *seg) for (i = 0; i < TRBS_PER_SEGMENT; ++i) { trb = &seg->trbs[i]; xhci_dbg(xhci, "@%08x %08x %08x %08x %08x\n", addr, - (unsigned int) trb->link.segment_ptr[0], - (unsigned int) trb->link.segment_ptr[1], + lower_32_bits(trb->link.segment_ptr), + upper_32_bits(trb->link.segment_ptr), (unsigned int) trb->link.intr_target, (unsigned int) trb->link.control); addr += sizeof(*trb); @@ -386,8 +374,8 @@ void xhci_dbg_erst(struct xhci_hcd *xhci, struct xhci_erst *erst) entry = &erst->entries[i]; xhci_dbg(xhci, "@%08x %08x %08x %08x %08x\n", (unsigned int) addr, - (unsigned int) entry->seg_addr[0], - (unsigned int) entry->seg_addr[1], + lower_32_bits(entry->seg_addr), + upper_32_bits(entry->seg_addr), (unsigned int) entry->seg_size, (unsigned int) entry->rsvd); addr += sizeof(*entry); @@ -396,90 +384,147 @@ void xhci_dbg_erst(struct xhci_hcd *xhci, struct xhci_erst *erst) void xhci_dbg_cmd_ptrs(struct xhci_hcd *xhci) { - u32 val; + u64 val; - val = xhci_readl(xhci, &xhci->op_regs->cmd_ring[0]); - xhci_dbg(xhci, "// xHC command ring deq ptr low bits + flags = 0x%x\n", val); - val = xhci_readl(xhci, &xhci->op_regs->cmd_ring[1]); - xhci_dbg(xhci, "// xHC command ring deq ptr high bits = 0x%x\n", val); + val = xhci_read_64(xhci, &xhci->op_regs->cmd_ring); + xhci_dbg(xhci, "// xHC command ring deq ptr low bits + flags = @%08x\n", + lower_32_bits(val)); + xhci_dbg(xhci, "// xHC command ring deq ptr high bits = @%08x\n", + upper_32_bits(val)); } -void xhci_dbg_ctx(struct xhci_hcd *xhci, struct xhci_device_control *ctx, dma_addr_t dma, unsigned int last_ep) +/* Print the last 32 bytes for 64-byte contexts */ +static void dbg_rsvd64(struct xhci_hcd *xhci, u64 *ctx, dma_addr_t dma) +{ + int i; + for (i = 0; i < 4; ++i) { + xhci_dbg(xhci, "@%p (virt) @%08llx " + "(dma) %#08llx - rsvd64[%d]\n", + &ctx[4 + i], (unsigned long long)dma, + ctx[4 + i], i); + dma += 8; + } +} + +void xhci_dbg_slot_ctx(struct xhci_hcd *xhci, struct xhci_container_ctx *ctx) { - int i, j; - int last_ep_ctx = 31; /* Fields are 32 bits wide, DMA addresses are in bytes */ int field_size = 32 / 8; + int i; - xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - drop flags\n", - &ctx->drop_flags, (unsigned long long)dma, - ctx->drop_flags); - dma += field_size; - xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - add flags\n", - &ctx->add_flags, (unsigned long long)dma, - ctx->add_flags); - dma += field_size; - for (i = 0; i > 6; ++i) { - xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - rsvd[%d]\n", - &ctx->rsvd[i], (unsigned long long)dma, - ctx->rsvd[i], i); - dma += field_size; - } + struct xhci_slot_ctx *slot_ctx = xhci_get_slot_ctx(xhci, ctx); + dma_addr_t dma = ctx->dma + ((unsigned long)slot_ctx - (unsigned long)ctx); + int csz = HCC_64BYTE_CONTEXT(xhci->hcc_params); xhci_dbg(xhci, "Slot Context:\n"); xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - dev_info\n", - &ctx->slot.dev_info, - (unsigned long long)dma, ctx->slot.dev_info); + &slot_ctx->dev_info, + (unsigned long long)dma, slot_ctx->dev_info); dma += field_size; xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - dev_info2\n", - &ctx->slot.dev_info2, - (unsigned long long)dma, ctx->slot.dev_info2); + &slot_ctx->dev_info2, + (unsigned long long)dma, slot_ctx->dev_info2); dma += field_size; xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - tt_info\n", - &ctx->slot.tt_info, - (unsigned long long)dma, ctx->slot.tt_info); + &slot_ctx->tt_info, + (unsigned long long)dma, slot_ctx->tt_info); dma += field_size; xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - dev_state\n", - &ctx->slot.dev_state, - (unsigned long long)dma, ctx->slot.dev_state); + &slot_ctx->dev_state, + (unsigned long long)dma, slot_ctx->dev_state); dma += field_size; - for (i = 0; i > 4; ++i) { + for (i = 0; i < 4; ++i) { xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - rsvd[%d]\n", - &ctx->slot.reserved[i], (unsigned long long)dma, - ctx->slot.reserved[i], i); + &slot_ctx->reserved[i], (unsigned long long)dma, + slot_ctx->reserved[i], i); dma += field_size; } + if (csz) + dbg_rsvd64(xhci, (u64 *)slot_ctx, dma); +} + +void xhci_dbg_ep_ctx(struct xhci_hcd *xhci, + struct xhci_container_ctx *ctx, + unsigned int last_ep) +{ + int i, j; + int last_ep_ctx = 31; + /* Fields are 32 bits wide, DMA addresses are in bytes */ + int field_size = 32 / 8; + int csz = HCC_64BYTE_CONTEXT(xhci->hcc_params); + if (last_ep < 31) last_ep_ctx = last_ep + 1; for (i = 0; i < last_ep_ctx; ++i) { + struct xhci_ep_ctx *ep_ctx = xhci_get_ep_ctx(xhci, ctx, i); + dma_addr_t dma = ctx->dma + + ((unsigned long)ep_ctx - (unsigned long)ctx); + xhci_dbg(xhci, "Endpoint %02d Context:\n", i); xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - ep_info\n", - &ctx->ep[i].ep_info, - (unsigned long long)dma, ctx->ep[i].ep_info); + &ep_ctx->ep_info, + (unsigned long long)dma, ep_ctx->ep_info); dma += field_size; xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - ep_info2\n", - &ctx->ep[i].ep_info2, - (unsigned long long)dma, ctx->ep[i].ep_info2); - dma += field_size; - xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - deq[0]\n", - &ctx->ep[i].deq[0], - (unsigned long long)dma, ctx->ep[i].deq[0]); - dma += field_size; - xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - deq[1]\n", - &ctx->ep[i].deq[1], - (unsigned long long)dma, ctx->ep[i].deq[1]); + &ep_ctx->ep_info2, + (unsigned long long)dma, ep_ctx->ep_info2); dma += field_size; + xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08llx - deq\n", + &ep_ctx->deq, + (unsigned long long)dma, ep_ctx->deq); + dma += 2*field_size; xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - tx_info\n", - &ctx->ep[i].tx_info, - (unsigned long long)dma, ctx->ep[i].tx_info); + &ep_ctx->tx_info, + (unsigned long long)dma, ep_ctx->tx_info); dma += field_size; for (j = 0; j < 3; ++j) { xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - rsvd[%d]\n", - &ctx->ep[i].reserved[j], + &ep_ctx->reserved[j], (unsigned long long)dma, - ctx->ep[i].reserved[j], j); + ep_ctx->reserved[j], j); + dma += field_size; + } + + if (csz) + dbg_rsvd64(xhci, (u64 *)ep_ctx, dma); + } +} + +void xhci_dbg_ctx(struct xhci_hcd *xhci, + struct xhci_container_ctx *ctx, + unsigned int last_ep) +{ + int i; + /* Fields are 32 bits wide, DMA addresses are in bytes */ + int field_size = 32 / 8; + struct xhci_slot_ctx *slot_ctx; + dma_addr_t dma = ctx->dma; + int csz = HCC_64BYTE_CONTEXT(xhci->hcc_params); + + if (ctx->type == XHCI_CTX_TYPE_INPUT) { + struct xhci_input_control_ctx *ctrl_ctx = + xhci_get_input_control_ctx(xhci, ctx); + xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - drop flags\n", + &ctrl_ctx->drop_flags, (unsigned long long)dma, + ctrl_ctx->drop_flags); + dma += field_size; + xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - add flags\n", + &ctrl_ctx->add_flags, (unsigned long long)dma, + ctrl_ctx->add_flags); + dma += field_size; + for (i = 0; i < 6; ++i) { + xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - rsvd2[%d]\n", + &ctrl_ctx->rsvd2[i], (unsigned long long)dma, + ctrl_ctx->rsvd2[i], i); dma += field_size; } + + if (csz) + dbg_rsvd64(xhci, (u64 *)ctrl_ctx, dma); } + + slot_ctx = xhci_get_slot_ctx(xhci, ctx); + xhci_dbg_slot_ctx(xhci, ctx); + xhci_dbg_ep_ctx(xhci, ctx, last_ep); } diff --git a/drivers/usb/host/xhci-hcd.c b/drivers/usb/host/xhci-hcd.c index dba3e07ccd0..816c39caca1 100644 --- a/drivers/usb/host/xhci-hcd.c +++ b/drivers/usb/host/xhci-hcd.c @@ -103,7 +103,10 @@ int xhci_reset(struct xhci_hcd *xhci) u32 state; state = xhci_readl(xhci, &xhci->op_regs->status); - BUG_ON((state & STS_HALT) == 0); + if ((state & STS_HALT) == 0) { + xhci_warn(xhci, "Host controller not halted, aborting reset.\n"); + return 0; + } xhci_dbg(xhci, "// Reset the HC\n"); command = xhci_readl(xhci, &xhci->op_regs->command); @@ -226,6 +229,7 @@ int xhci_init(struct usb_hcd *hcd) static void xhci_work(struct xhci_hcd *xhci) { u32 temp; + u64 temp_64; /* * Clear the op reg interrupt status first, @@ -248,9 +252,9 @@ static void xhci_work(struct xhci_hcd *xhci) /* FIXME this should be a delayed service routine that clears the EHB */ xhci_handle_event(xhci); - /* Clear the event handler busy flag; the event ring should be empty. */ - temp = xhci_readl(xhci, &xhci->ir_set->erst_dequeue[0]); - xhci_writel(xhci, temp & ~ERST_EHB, &xhci->ir_set->erst_dequeue[0]); + /* Clear the event handler busy flag (RW1C); the event ring should be empty. */ + temp_64 = xhci_read_64(xhci, &xhci->ir_set->erst_dequeue); + xhci_write_64(xhci, temp_64 | ERST_EHB, &xhci->ir_set->erst_dequeue); /* Flush posted writes -- FIXME is this necessary? */ xhci_readl(xhci, &xhci->ir_set->irq_pending); } @@ -266,19 +270,34 @@ irqreturn_t xhci_irq(struct usb_hcd *hcd) { struct xhci_hcd *xhci = hcd_to_xhci(hcd); u32 temp, temp2; + union xhci_trb *trb; spin_lock(&xhci->lock); + trb = xhci->event_ring->dequeue; /* Check if the xHC generated the interrupt, or the irq is shared */ temp = xhci_readl(xhci, &xhci->op_regs->status); temp2 = xhci_readl(xhci, &xhci->ir_set->irq_pending); + if (temp == 0xffffffff && temp2 == 0xffffffff) + goto hw_died; + if (!(temp & STS_EINT) && !ER_IRQ_PENDING(temp2)) { spin_unlock(&xhci->lock); return IRQ_NONE; } + xhci_dbg(xhci, "op reg status = %08x\n", temp); + xhci_dbg(xhci, "ir set irq_pending = %08x\n", temp2); + xhci_dbg(xhci, "Event ring dequeue ptr:\n"); + xhci_dbg(xhci, "@%llx %08x %08x %08x %08x\n", + (unsigned long long)xhci_trb_virt_to_dma(xhci->event_ring->deq_seg, trb), + lower_32_bits(trb->link.segment_ptr), + upper_32_bits(trb->link.segment_ptr), + (unsigned int) trb->link.intr_target, + (unsigned int) trb->link.control); if (temp & STS_FATAL) { xhci_warn(xhci, "WARNING: Host System Error\n"); xhci_halt(xhci); +hw_died: xhci_to_hcd(xhci)->state = HC_STATE_HALT; spin_unlock(&xhci->lock); return -ESHUTDOWN; @@ -295,6 +314,7 @@ void xhci_event_ring_work(unsigned long arg) { unsigned long flags; int temp; + u64 temp_64; struct xhci_hcd *xhci = (struct xhci_hcd *) arg; int i, j; @@ -311,9 +331,9 @@ void xhci_event_ring_work(unsigned long arg) xhci_dbg(xhci, "Event ring:\n"); xhci_debug_segment(xhci, xhci->event_ring->deq_seg); xhci_dbg_ring_ptrs(xhci, xhci->event_ring); - temp = xhci_readl(xhci, &xhci->ir_set->erst_dequeue[0]); - temp &= ERST_PTR_MASK; - xhci_dbg(xhci, "ERST deq = 0x%x\n", temp); + temp_64 = xhci_read_64(xhci, &xhci->ir_set->erst_dequeue); + temp_64 &= ~ERST_PTR_MASK; + xhci_dbg(xhci, "ERST deq = 64'h%0lx\n", (long unsigned int) temp_64); xhci_dbg(xhci, "Command ring:\n"); xhci_debug_segment(xhci, xhci->cmd_ring->deq_seg); xhci_dbg_ring_ptrs(xhci, xhci->cmd_ring); @@ -356,6 +376,7 @@ void xhci_event_ring_work(unsigned long arg) int xhci_run(struct usb_hcd *hcd) { u32 temp; + u64 temp_64; struct xhci_hcd *xhci = hcd_to_xhci(hcd); void (*doorbell)(struct xhci_hcd *) = NULL; @@ -382,6 +403,20 @@ int xhci_run(struct usb_hcd *hcd) add_timer(&xhci->event_ring_timer); #endif + xhci_dbg(xhci, "Command ring memory map follows:\n"); + xhci_debug_ring(xhci, xhci->cmd_ring); + xhci_dbg_ring_ptrs(xhci, xhci->cmd_ring); + xhci_dbg_cmd_ptrs(xhci); + + xhci_dbg(xhci, "ERST memory map follows:\n"); + xhci_dbg_erst(xhci, &xhci->erst); + xhci_dbg(xhci, "Event ring:\n"); + xhci_debug_ring(xhci, xhci->event_ring); + xhci_dbg_ring_ptrs(xhci, xhci->event_ring); + temp_64 = xhci_read_64(xhci, &xhci->ir_set->erst_dequeue); + temp_64 &= ~ERST_PTR_MASK; + xhci_dbg(xhci, "ERST deq = 64'h%0lx\n", (long unsigned int) temp_64); + xhci_dbg(xhci, "// Set the interrupt modulation register\n"); temp = xhci_readl(xhci, &xhci->ir_set->irq_control); temp &= ~ER_IRQ_INTERVAL_MASK; @@ -406,22 +441,6 @@ int xhci_run(struct usb_hcd *hcd) if (NUM_TEST_NOOPS > 0) doorbell = xhci_setup_one_noop(xhci); - xhci_dbg(xhci, "Command ring memory map follows:\n"); - xhci_debug_ring(xhci, xhci->cmd_ring); - xhci_dbg_ring_ptrs(xhci, xhci->cmd_ring); - xhci_dbg_cmd_ptrs(xhci); - - xhci_dbg(xhci, "ERST memory map follows:\n"); - xhci_dbg_erst(xhci, &xhci->erst); - xhci_dbg(xhci, "Event ring:\n"); - xhci_debug_ring(xhci, xhci->event_ring); - xhci_dbg_ring_ptrs(xhci, xhci->event_ring); - temp = xhci_readl(xhci, &xhci->ir_set->erst_dequeue[0]); - temp &= ERST_PTR_MASK; - xhci_dbg(xhci, "ERST deq = 0x%x\n", temp); - temp = xhci_readl(xhci, &xhci->ir_set->erst_dequeue[1]); - xhci_dbg(xhci, "ERST deq upper = 0x%x\n", temp); - temp = xhci_readl(xhci, &xhci->op_regs->command); temp |= (CMD_RUN); xhci_dbg(xhci, "// Turn on HC, cmd = 0x%x.\n", @@ -601,10 +620,13 @@ int xhci_urb_enqueue(struct usb_hcd *hcd, struct urb *urb, gfp_t mem_flags) goto exit; } if (usb_endpoint_xfer_control(&urb->ep->desc)) - ret = xhci_queue_ctrl_tx(xhci, mem_flags, urb, + /* We have a spinlock and interrupts disabled, so we must pass + * atomic context to this function, which may allocate memory. + */ + ret = xhci_queue_ctrl_tx(xhci, GFP_ATOMIC, urb, slot_id, ep_index); else if (usb_endpoint_xfer_bulk(&urb->ep->desc)) - ret = xhci_queue_bulk_tx(xhci, mem_flags, urb, + ret = xhci_queue_bulk_tx(xhci, GFP_ATOMIC, urb, slot_id, ep_index); else ret = -EINVAL; @@ -661,8 +683,12 @@ int xhci_urb_dequeue(struct usb_hcd *hcd, struct urb *urb, int status) goto done; xhci_dbg(xhci, "Cancel URB %p\n", urb); + xhci_dbg(xhci, "Event ring:\n"); + xhci_debug_ring(xhci, xhci->event_ring); ep_index = xhci_get_endpoint_index(&urb->ep->desc); ep_ring = xhci->devs[urb->dev->slot_id]->ep_rings[ep_index]; + xhci_dbg(xhci, "Endpoint ring:\n"); + xhci_debug_ring(xhci, ep_ring); td = (struct xhci_td *) urb->hcpriv; ep_ring->cancels_pending++; @@ -696,7 +722,9 @@ int xhci_drop_endpoint(struct usb_hcd *hcd, struct usb_device *udev, struct usb_host_endpoint *ep) { struct xhci_hcd *xhci; - struct xhci_device_control *in_ctx; + struct xhci_container_ctx *in_ctx, *out_ctx; + struct xhci_input_control_ctx *ctrl_ctx; + struct xhci_slot_ctx *slot_ctx; unsigned int last_ctx; unsigned int ep_index; struct xhci_ep_ctx *ep_ctx; @@ -724,31 +752,34 @@ int xhci_drop_endpoint(struct usb_hcd *hcd, struct usb_device *udev, } in_ctx = xhci->devs[udev->slot_id]->in_ctx; + out_ctx = xhci->devs[udev->slot_id]->out_ctx; + ctrl_ctx = xhci_get_input_control_ctx(xhci, in_ctx); ep_index = xhci_get_endpoint_index(&ep->desc); - ep_ctx = &xhci->devs[udev->slot_id]->out_ctx->ep[ep_index]; + ep_ctx = xhci_get_ep_ctx(xhci, out_ctx, ep_index); /* If the HC already knows the endpoint is disabled, * or the HCD has noted it is disabled, ignore this request */ if ((ep_ctx->ep_info & EP_STATE_MASK) == EP_STATE_DISABLED || - in_ctx->drop_flags & xhci_get_endpoint_flag(&ep->desc)) { + ctrl_ctx->drop_flags & xhci_get_endpoint_flag(&ep->desc)) { xhci_warn(xhci, "xHCI %s called with disabled ep %p\n", __func__, ep); return 0; } - in_ctx->drop_flags |= drop_flag; - new_drop_flags = in_ctx->drop_flags; + ctrl_ctx->drop_flags |= drop_flag; + new_drop_flags = ctrl_ctx->drop_flags; - in_ctx->add_flags = ~drop_flag; - new_add_flags = in_ctx->add_flags; + ctrl_ctx->add_flags = ~drop_flag; + new_add_flags = ctrl_ctx->add_flags; - last_ctx = xhci_last_valid_endpoint(in_ctx->add_flags); + last_ctx = xhci_last_valid_endpoint(ctrl_ctx->add_flags); + slot_ctx = xhci_get_slot_ctx(xhci, in_ctx); /* Update the last valid endpoint context, if we deleted the last one */ - if ((in_ctx->slot.dev_info & LAST_CTX_MASK) > LAST_CTX(last_ctx)) { - in_ctx->slot.dev_info &= ~LAST_CTX_MASK; - in_ctx->slot.dev_info |= LAST_CTX(last_ctx); + if ((slot_ctx->dev_info & LAST_CTX_MASK) > LAST_CTX(last_ctx)) { + slot_ctx->dev_info &= ~LAST_CTX_MASK; + slot_ctx->dev_info |= LAST_CTX(last_ctx); } - new_slot_info = in_ctx->slot.dev_info; + new_slot_info = slot_ctx->dev_info; xhci_endpoint_zero(xhci, xhci->devs[udev->slot_id], ep); @@ -778,17 +809,22 @@ int xhci_add_endpoint(struct usb_hcd *hcd, struct usb_device *udev, struct usb_host_endpoint *ep) { struct xhci_hcd *xhci; - struct xhci_device_control *in_ctx; + struct xhci_container_ctx *in_ctx, *out_ctx; unsigned int ep_index; struct xhci_ep_ctx *ep_ctx; + struct xhci_slot_ctx *slot_ctx; + struct xhci_input_control_ctx *ctrl_ctx; u32 added_ctxs; unsigned int last_ctx; u32 new_add_flags, new_drop_flags, new_slot_info; int ret = 0; ret = xhci_check_args(hcd, udev, ep, 1, __func__); - if (ret <= 0) + if (ret <= 0) { + /* So we won't queue a reset ep command for a root hub */ + ep->hcpriv = NULL; return ret; + } xhci = hcd_to_xhci(hcd); added_ctxs = xhci_get_endpoint_flag(&ep->desc); @@ -810,12 +846,14 @@ int xhci_add_endpoint(struct usb_hcd *hcd, struct usb_device *udev, } in_ctx = xhci->devs[udev->slot_id]->in_ctx; + out_ctx = xhci->devs[udev->slot_id]->out_ctx; + ctrl_ctx = xhci_get_input_control_ctx(xhci, in_ctx); ep_index = xhci_get_endpoint_index(&ep->desc); - ep_ctx = &xhci->devs[udev->slot_id]->out_ctx->ep[ep_index]; + ep_ctx = xhci_get_ep_ctx(xhci, out_ctx, ep_index); /* If the HCD has already noted the endpoint is enabled, * ignore this request. */ - if (in_ctx->add_flags & xhci_get_endpoint_flag(&ep->desc)) { + if (ctrl_ctx->add_flags & xhci_get_endpoint_flag(&ep->desc)) { xhci_warn(xhci, "xHCI %s called with enabled ep %p\n", __func__, ep); return 0; @@ -833,8 +871,8 @@ int xhci_add_endpoint(struct usb_hcd *hcd, struct usb_device *udev, return -ENOMEM; } - in_ctx->add_flags |= added_ctxs; - new_add_flags = in_ctx->add_flags; + ctrl_ctx->add_flags |= added_ctxs; + new_add_flags = ctrl_ctx->add_flags; /* If xhci_endpoint_disable() was called for this endpoint, but the * xHC hasn't been notified yet through the check_bandwidth() call, @@ -842,14 +880,18 @@ int xhci_add_endpoint(struct usb_hcd *hcd, struct usb_device *udev, * descriptors. We must drop and re-add this endpoint, so we leave the * drop flags alone. */ - new_drop_flags = in_ctx->drop_flags; + new_drop_flags = ctrl_ctx->drop_flags; + slot_ctx = xhci_get_slot_ctx(xhci, in_ctx); /* Update the last valid endpoint context, if we just added one past */ - if ((in_ctx->slot.dev_info & LAST_CTX_MASK) < LAST_CTX(last_ctx)) { - in_ctx->slot.dev_info &= ~LAST_CTX_MASK; - in_ctx->slot.dev_info |= LAST_CTX(last_ctx); + if ((slot_ctx->dev_info & LAST_CTX_MASK) < LAST_CTX(last_ctx)) { + slot_ctx->dev_info &= ~LAST_CTX_MASK; + slot_ctx->dev_info |= LAST_CTX(last_ctx); } - new_slot_info = in_ctx->slot.dev_info; + new_slot_info = slot_ctx->dev_info; + + /* Store the usb_device pointer for later use */ + ep->hcpriv = udev; xhci_dbg(xhci, "add ep 0x%x, slot id %d, new drop flags = %#x, new add flags = %#x, new slot info = %#x\n", (unsigned int) ep->desc.bEndpointAddress, @@ -860,9 +902,11 @@ int xhci_add_endpoint(struct usb_hcd *hcd, struct usb_device *udev, return 0; } -static void xhci_zero_in_ctx(struct xhci_virt_device *virt_dev) +static void xhci_zero_in_ctx(struct xhci_hcd *xhci, struct xhci_virt_device *virt_dev) { + struct xhci_input_control_ctx *ctrl_ctx; struct xhci_ep_ctx *ep_ctx; + struct xhci_slot_ctx *slot_ctx; int i; /* When a device's add flag and drop flag are zero, any subsequent @@ -870,17 +914,18 @@ static void xhci_zero_in_ctx(struct xhci_virt_device *virt_dev) * untouched. Make sure we don't leave any old state in the input * endpoint contexts. */ - virt_dev->in_ctx->drop_flags = 0; - virt_dev->in_ctx->add_flags = 0; - virt_dev->in_ctx->slot.dev_info &= ~LAST_CTX_MASK; + ctrl_ctx = xhci_get_input_control_ctx(xhci, virt_dev->in_ctx); + ctrl_ctx->drop_flags = 0; + ctrl_ctx->add_flags = 0; + slot_ctx = xhci_get_slot_ctx(xhci, virt_dev->in_ctx); + slot_ctx->dev_info &= ~LAST_CTX_MASK; /* Endpoint 0 is always valid */ - virt_dev->in_ctx->slot.dev_info |= LAST_CTX(1); + slot_ctx->dev_info |= LAST_CTX(1); for (i = 1; i < 31; ++i) { - ep_ctx = &virt_dev->in_ctx->ep[i]; + ep_ctx = xhci_get_ep_ctx(xhci, virt_dev->in_ctx, i); ep_ctx->ep_info = 0; ep_ctx->ep_info2 = 0; - ep_ctx->deq[0] = 0; - ep_ctx->deq[1] = 0; + ep_ctx->deq = 0; ep_ctx->tx_info = 0; } } @@ -903,6 +948,8 @@ int xhci_check_bandwidth(struct usb_hcd *hcd, struct usb_device *udev) unsigned long flags; struct xhci_hcd *xhci; struct xhci_virt_device *virt_dev; + struct xhci_input_control_ctx *ctrl_ctx; + struct xhci_slot_ctx *slot_ctx; ret = xhci_check_args(hcd, udev, NULL, 0, __func__); if (ret <= 0) @@ -918,16 +965,18 @@ int xhci_check_bandwidth(struct usb_hcd *hcd, struct usb_device *udev) virt_dev = xhci->devs[udev->slot_id]; /* See section 4.6.6 - A0 = 1; A1 = D0 = D1 = 0 */ - virt_dev->in_ctx->add_flags |= SLOT_FLAG; - virt_dev->in_ctx->add_flags &= ~EP0_FLAG; - virt_dev->in_ctx->drop_flags &= ~SLOT_FLAG; - virt_dev->in_ctx->drop_flags &= ~EP0_FLAG; + ctrl_ctx = xhci_get_input_control_ctx(xhci, virt_dev->in_ctx); + ctrl_ctx->add_flags |= SLOT_FLAG; + ctrl_ctx->add_flags &= ~EP0_FLAG; + ctrl_ctx->drop_flags &= ~SLOT_FLAG; + ctrl_ctx->drop_flags &= ~EP0_FLAG; xhci_dbg(xhci, "New Input Control Context:\n"); - xhci_dbg_ctx(xhci, virt_dev->in_ctx, virt_dev->in_ctx_dma, - LAST_CTX_TO_EP_NUM(virt_dev->in_ctx->slot.dev_info)); + slot_ctx = xhci_get_slot_ctx(xhci, virt_dev->in_ctx); + xhci_dbg_ctx(xhci, virt_dev->in_ctx, + LAST_CTX_TO_EP_NUM(slot_ctx->dev_info)); spin_lock_irqsave(&xhci->lock, flags); - ret = xhci_queue_configure_endpoint(xhci, virt_dev->in_ctx_dma, + ret = xhci_queue_configure_endpoint(xhci, virt_dev->in_ctx->dma, udev->slot_id); if (ret < 0) { spin_unlock_irqrestore(&xhci->lock, flags); @@ -982,10 +1031,10 @@ int xhci_check_bandwidth(struct usb_hcd *hcd, struct usb_device *udev) } xhci_dbg(xhci, "Output context after successful config ep cmd:\n"); - xhci_dbg_ctx(xhci, virt_dev->out_ctx, virt_dev->out_ctx_dma, - LAST_CTX_TO_EP_NUM(virt_dev->in_ctx->slot.dev_info)); + xhci_dbg_ctx(xhci, virt_dev->out_ctx, + LAST_CTX_TO_EP_NUM(slot_ctx->dev_info)); - xhci_zero_in_ctx(virt_dev); + xhci_zero_in_ctx(xhci, virt_dev); /* Free any old rings */ for (i = 1; i < 31; ++i) { if (virt_dev->new_ep_rings[i]) { @@ -1023,7 +1072,67 @@ void xhci_reset_bandwidth(struct usb_hcd *hcd, struct usb_device *udev) virt_dev->new_ep_rings[i] = NULL; } } - xhci_zero_in_ctx(virt_dev); + xhci_zero_in_ctx(xhci, virt_dev); +} + +/* Deal with stalled endpoints. The core should have sent the control message + * to clear the halt condition. However, we need to make the xHCI hardware + * reset its sequence number, since a device will expect a sequence number of + * zero after the halt condition is cleared. + * Context: in_interrupt + */ +void xhci_endpoint_reset(struct usb_hcd *hcd, + struct usb_host_endpoint *ep) +{ + struct xhci_hcd *xhci; + struct usb_device *udev; + unsigned int ep_index; + unsigned long flags; + int ret; + struct xhci_dequeue_state deq_state; + struct xhci_ring *ep_ring; + + xhci = hcd_to_xhci(hcd); + udev = (struct usb_device *) ep->hcpriv; + /* Called with a root hub endpoint (or an endpoint that wasn't added + * with xhci_add_endpoint() + */ + if (!ep->hcpriv) + return; + ep_index = xhci_get_endpoint_index(&ep->desc); + ep_ring = xhci->devs[udev->slot_id]->ep_rings[ep_index]; + if (!ep_ring->stopped_td) { + xhci_dbg(xhci, "Endpoint 0x%x not halted, refusing to reset.\n", + ep->desc.bEndpointAddress); + return; + } + + xhci_dbg(xhci, "Queueing reset endpoint command\n"); + spin_lock_irqsave(&xhci->lock, flags); + ret = xhci_queue_reset_ep(xhci, udev->slot_id, ep_index); + /* + * Can't change the ring dequeue pointer until it's transitioned to the + * stopped state, which is only upon a successful reset endpoint + * command. Better hope that last command worked! + */ + if (!ret) { + xhci_dbg(xhci, "Cleaning up stalled endpoint ring\n"); + /* We need to move the HW's dequeue pointer past this TD, + * or it will attempt to resend it on the next doorbell ring. + */ + xhci_find_new_dequeue_state(xhci, udev->slot_id, + ep_index, ep_ring->stopped_td, &deq_state); + xhci_dbg(xhci, "Queueing new dequeue state\n"); + xhci_queue_new_dequeue_state(xhci, ep_ring, + udev->slot_id, + ep_index, &deq_state); + kfree(ep_ring->stopped_td); + xhci_ring_cmd_db(xhci); + } + spin_unlock_irqrestore(&xhci->lock, flags); + + if (ret) + xhci_warn(xhci, "FIXME allocate a new ring segment\n"); } /* @@ -1120,7 +1229,9 @@ int xhci_address_device(struct usb_hcd *hcd, struct usb_device *udev) struct xhci_virt_device *virt_dev; int ret = 0; struct xhci_hcd *xhci = hcd_to_xhci(hcd); - u32 temp; + struct xhci_slot_ctx *slot_ctx; + struct xhci_input_control_ctx *ctrl_ctx; + u64 temp_64; if (!udev->slot_id) { xhci_dbg(xhci, "Bad Slot ID %d\n", udev->slot_id); @@ -1133,10 +1244,12 @@ int xhci_address_device(struct usb_hcd *hcd, struct usb_device *udev) if (!udev->config) xhci_setup_addressable_virt_dev(xhci, udev); /* Otherwise, assume the core has the device configured how it wants */ + xhci_dbg(xhci, "Slot ID %d Input Context:\n", udev->slot_id); + xhci_dbg_ctx(xhci, virt_dev->in_ctx, 2); spin_lock_irqsave(&xhci->lock, flags); - ret = xhci_queue_address_device(xhci, virt_dev->in_ctx_dma, - udev->slot_id); + ret = xhci_queue_address_device(xhci, virt_dev->in_ctx->dma, + udev->slot_id); if (ret) { spin_unlock_irqrestore(&xhci->lock, flags); xhci_dbg(xhci, "FIXME: allocate a command ring segment\n"); @@ -1176,41 +1289,37 @@ int xhci_address_device(struct usb_hcd *hcd, struct usb_device *udev) default: xhci_err(xhci, "ERROR: unexpected command completion " "code 0x%x.\n", virt_dev->cmd_status); + xhci_dbg(xhci, "Slot ID %d Output Context:\n", udev->slot_id); + xhci_dbg_ctx(xhci, virt_dev->out_ctx, 2); ret = -EINVAL; break; } if (ret) { return ret; } - temp = xhci_readl(xhci, &xhci->op_regs->dcbaa_ptr[0]); - xhci_dbg(xhci, "Op regs DCBAA ptr[0] = %#08x\n", temp); - temp = xhci_readl(xhci, &xhci->op_regs->dcbaa_ptr[1]); - xhci_dbg(xhci, "Op regs DCBAA ptr[1] = %#08x\n", temp); - xhci_dbg(xhci, "Slot ID %d dcbaa entry[0] @%p = %#08x\n", - udev->slot_id, - &xhci->dcbaa->dev_context_ptrs[2*udev->slot_id], - xhci->dcbaa->dev_context_ptrs[2*udev->slot_id]); - xhci_dbg(xhci, "Slot ID %d dcbaa entry[1] @%p = %#08x\n", + temp_64 = xhci_read_64(xhci, &xhci->op_regs->dcbaa_ptr); + xhci_dbg(xhci, "Op regs DCBAA ptr = %#016llx\n", temp_64); + xhci_dbg(xhci, "Slot ID %d dcbaa entry @%p = %#016llx\n", udev->slot_id, - &xhci->dcbaa->dev_context_ptrs[2*udev->slot_id+1], - xhci->dcbaa->dev_context_ptrs[2*udev->slot_id+1]); + &xhci->dcbaa->dev_context_ptrs[udev->slot_id], + (unsigned long long) + xhci->dcbaa->dev_context_ptrs[udev->slot_id]); xhci_dbg(xhci, "Output Context DMA address = %#08llx\n", - (unsigned long long)virt_dev->out_ctx_dma); + (unsigned long long)virt_dev->out_ctx->dma); xhci_dbg(xhci, "Slot ID %d Input Context:\n", udev->slot_id); - xhci_dbg_ctx(xhci, virt_dev->in_ctx, virt_dev->in_ctx_dma, 2); + xhci_dbg_ctx(xhci, virt_dev->in_ctx, 2); xhci_dbg(xhci, "Slot ID %d Output Context:\n", udev->slot_id); - xhci_dbg_ctx(xhci, virt_dev->out_ctx, virt_dev->out_ctx_dma, 2); + xhci_dbg_ctx(xhci, virt_dev->out_ctx, 2); /* * USB core uses address 1 for the roothubs, so we add one to the * address given back to us by the HC. */ - udev->devnum = (virt_dev->out_ctx->slot.dev_state & DEV_ADDR_MASK) + 1; + slot_ctx = xhci_get_slot_ctx(xhci, virt_dev->out_ctx); + udev->devnum = (slot_ctx->dev_state & DEV_ADDR_MASK) + 1; /* Zero the input context control for later use */ - virt_dev->in_ctx->add_flags = 0; - virt_dev->in_ctx->drop_flags = 0; - /* Mirror flags in the output context for future ep enable/disable */ - virt_dev->out_ctx->add_flags = SLOT_FLAG | EP0_FLAG; - virt_dev->out_ctx->drop_flags = 0; + ctrl_ctx = xhci_get_input_control_ctx(xhci, virt_dev->in_ctx); + ctrl_ctx->add_flags = 0; + ctrl_ctx->drop_flags = 0; xhci_dbg(xhci, "Device address = %d\n", udev->devnum); /* XXX Meh, not sure if anyone else but choose_address uses this. */ @@ -1252,7 +1361,6 @@ static int __init xhci_hcd_init(void) /* xhci_device_control has eight fields, and also * embeds one xhci_slot_ctx and 31 xhci_ep_ctx */ - BUILD_BUG_ON(sizeof(struct xhci_device_control) != (8+8+8*31)*32/8); BUILD_BUG_ON(sizeof(struct xhci_stream_ctx) != 4*32/8); BUILD_BUG_ON(sizeof(union xhci_trb) != 4*32/8); BUILD_BUG_ON(sizeof(struct xhci_erst_entry) != 4*32/8); diff --git a/drivers/usb/host/xhci-mem.c b/drivers/usb/host/xhci-mem.c index c8a72de1c50..e6b9a1c6002 100644 --- a/drivers/usb/host/xhci-mem.c +++ b/drivers/usb/host/xhci-mem.c @@ -88,7 +88,7 @@ static void xhci_link_segments(struct xhci_hcd *xhci, struct xhci_segment *prev, return; prev->next = next; if (link_trbs) { - prev->trbs[TRBS_PER_SEGMENT-1].link.segment_ptr[0] = next->dma; + prev->trbs[TRBS_PER_SEGMENT-1].link.segment_ptr = next->dma; /* Set the last TRB in the segment to have a TRB type ID of Link TRB */ val = prev->trbs[TRBS_PER_SEGMENT-1].link.control; @@ -189,6 +189,63 @@ fail: return 0; } +#define CTX_SIZE(_hcc) (HCC_64BYTE_CONTEXT(_hcc) ? 64 : 32) + +struct xhci_container_ctx *xhci_alloc_container_ctx(struct xhci_hcd *xhci, + int type, gfp_t flags) +{ + struct xhci_container_ctx *ctx = kzalloc(sizeof(*ctx), flags); + if (!ctx) + return NULL; + + BUG_ON((type != XHCI_CTX_TYPE_DEVICE) && (type != XHCI_CTX_TYPE_INPUT)); + ctx->type = type; + ctx->size = HCC_64BYTE_CONTEXT(xhci->hcc_params) ? 2048 : 1024; + if (type == XHCI_CTX_TYPE_INPUT) + ctx->size += CTX_SIZE(xhci->hcc_params); + + ctx->bytes = dma_pool_alloc(xhci->device_pool, flags, &ctx->dma); + memset(ctx->bytes, 0, ctx->size); + return ctx; +} + +void xhci_free_container_ctx(struct xhci_hcd *xhci, + struct xhci_container_ctx *ctx) +{ + dma_pool_free(xhci->device_pool, ctx->bytes, ctx->dma); + kfree(ctx); +} + +struct xhci_input_control_ctx *xhci_get_input_control_ctx(struct xhci_hcd *xhci, + struct xhci_container_ctx *ctx) +{ + BUG_ON(ctx->type != XHCI_CTX_TYPE_INPUT); + return (struct xhci_input_control_ctx *)ctx->bytes; +} + +struct xhci_slot_ctx *xhci_get_slot_ctx(struct xhci_hcd *xhci, + struct xhci_container_ctx *ctx) +{ + if (ctx->type == XHCI_CTX_TYPE_DEVICE) + return (struct xhci_slot_ctx *)ctx->bytes; + + return (struct xhci_slot_ctx *) + (ctx->bytes + CTX_SIZE(xhci->hcc_params)); +} + +struct xhci_ep_ctx *xhci_get_ep_ctx(struct xhci_hcd *xhci, + struct xhci_container_ctx *ctx, + unsigned int ep_index) +{ + /* increment ep index by offset of start of ep ctx array */ + ep_index++; + if (ctx->type == XHCI_CTX_TYPE_INPUT) + ep_index++; + + return (struct xhci_ep_ctx *) + (ctx->bytes + (ep_index * CTX_SIZE(xhci->hcc_params))); +} + /* All the xhci_tds in the ring's TD list should be freed at this point */ void xhci_free_virt_device(struct xhci_hcd *xhci, int slot_id) { @@ -200,8 +257,7 @@ void xhci_free_virt_device(struct xhci_hcd *xhci, int slot_id) return; dev = xhci->devs[slot_id]; - xhci->dcbaa->dev_context_ptrs[2*slot_id] = 0; - xhci->dcbaa->dev_context_ptrs[2*slot_id + 1] = 0; + xhci->dcbaa->dev_context_ptrs[slot_id] = 0; if (!dev) return; @@ -210,11 +266,10 @@ void xhci_free_virt_device(struct xhci_hcd *xhci, int slot_id) xhci_ring_free(xhci, dev->ep_rings[i]); if (dev->in_ctx) - dma_pool_free(xhci->device_pool, - dev->in_ctx, dev->in_ctx_dma); + xhci_free_container_ctx(xhci, dev->in_ctx); if (dev->out_ctx) - dma_pool_free(xhci->device_pool, - dev->out_ctx, dev->out_ctx_dma); + xhci_free_container_ctx(xhci, dev->out_ctx); + kfree(xhci->devs[slot_id]); xhci->devs[slot_id] = 0; } @@ -222,7 +277,6 @@ void xhci_free_virt_device(struct xhci_hcd *xhci, int slot_id) int xhci_alloc_virt_device(struct xhci_hcd *xhci, int slot_id, struct usb_device *udev, gfp_t flags) { - dma_addr_t dma; struct xhci_virt_device *dev; /* Slot ID 0 is reserved */ @@ -236,23 +290,21 @@ int xhci_alloc_virt_device(struct xhci_hcd *xhci, int slot_id, return 0; dev = xhci->devs[slot_id]; - /* Allocate the (output) device context that will be used in the HC */ - dev->out_ctx = dma_pool_alloc(xhci->device_pool, flags, &dma); + /* Allocate the (output) device context that will be used in the HC. */ + dev->out_ctx = xhci_alloc_container_ctx(xhci, XHCI_CTX_TYPE_DEVICE, flags); if (!dev->out_ctx) goto fail; - dev->out_ctx_dma = dma; + xhci_dbg(xhci, "Slot %d output ctx = 0x%llx (dma)\n", slot_id, - (unsigned long long)dma); - memset(dev->out_ctx, 0, sizeof(*dev->out_ctx)); + (unsigned long long)dev->out_ctx->dma); /* Allocate the (input) device context for address device command */ - dev->in_ctx = dma_pool_alloc(xhci->device_pool, flags, &dma); + dev->in_ctx = xhci_alloc_container_ctx(xhci, XHCI_CTX_TYPE_INPUT, flags); if (!dev->in_ctx) goto fail; - dev->in_ctx_dma = dma; + xhci_dbg(xhci, "Slot %d input ctx = 0x%llx (dma)\n", slot_id, - (unsigned long long)dma); - memset(dev->in_ctx, 0, sizeof(*dev->in_ctx)); + (unsigned long long)dev->in_ctx->dma); /* Allocate endpoint 0 ring */ dev->ep_rings[0] = xhci_ring_alloc(xhci, 1, true, flags); @@ -261,17 +313,12 @@ int xhci_alloc_virt_device(struct xhci_hcd *xhci, int slot_id, init_completion(&dev->cmd_completion); - /* - * Point to output device context in dcbaa; skip the output control - * context, which is eight 32 bit fields (or 32 bytes long) - */ - xhci->dcbaa->dev_context_ptrs[2*slot_id] = - (u32) dev->out_ctx_dma + (32); + /* Point to output device context in dcbaa. */ + xhci->dcbaa->dev_context_ptrs[slot_id] = dev->out_ctx->dma; xhci_dbg(xhci, "Set slot id %d dcbaa entry %p to 0x%llx\n", slot_id, - &xhci->dcbaa->dev_context_ptrs[2*slot_id], - (unsigned long long)dev->out_ctx_dma); - xhci->dcbaa->dev_context_ptrs[2*slot_id + 1] = 0; + &xhci->dcbaa->dev_context_ptrs[slot_id], + (unsigned long long) xhci->dcbaa->dev_context_ptrs[slot_id]); return 1; fail: @@ -285,6 +332,8 @@ int xhci_setup_addressable_virt_dev(struct xhci_hcd *xhci, struct usb_device *ud struct xhci_virt_device *dev; struct xhci_ep_ctx *ep0_ctx; struct usb_device *top_dev; + struct xhci_slot_ctx *slot_ctx; + struct xhci_input_control_ctx *ctrl_ctx; dev = xhci->devs[udev->slot_id]; /* Slot ID 0 is reserved */ @@ -293,27 +342,29 @@ int xhci_setup_addressable_virt_dev(struct xhci_hcd *xhci, struct usb_device *ud udev->slot_id); return -EINVAL; } - ep0_ctx = &dev->in_ctx->ep[0]; + ep0_ctx = xhci_get_ep_ctx(xhci, dev->in_ctx, 0); + ctrl_ctx = xhci_get_input_control_ctx(xhci, dev->in_ctx); + slot_ctx = xhci_get_slot_ctx(xhci, dev->in_ctx); /* 2) New slot context and endpoint 0 context are valid*/ - dev->in_ctx->add_flags = SLOT_FLAG | EP0_FLAG; + ctrl_ctx->add_flags = SLOT_FLAG | EP0_FLAG; /* 3) Only the control endpoint is valid - one endpoint context */ - dev->in_ctx->slot.dev_info |= LAST_CTX(1); + slot_ctx->dev_info |= LAST_CTX(1); switch (udev->speed) { case USB_SPEED_SUPER: - dev->in_ctx->slot.dev_info |= (u32) udev->route; - dev->in_ctx->slot.dev_info |= (u32) SLOT_SPEED_SS; + slot_ctx->dev_info |= (u32) udev->route; + slot_ctx->dev_info |= (u32) SLOT_SPEED_SS; break; case USB_SPEED_HIGH: - dev->in_ctx->slot.dev_info |= (u32) SLOT_SPEED_HS; + slot_ctx->dev_info |= (u32) SLOT_SPEED_HS; break; case USB_SPEED_FULL: - dev->in_ctx->slot.dev_info |= (u32) SLOT_SPEED_FS; + slot_ctx->dev_info |= (u32) SLOT_SPEED_FS; break; case USB_SPEED_LOW: - dev->in_ctx->slot.dev_info |= (u32) SLOT_SPEED_LS; + slot_ctx->dev_info |= (u32) SLOT_SPEED_LS; break; case USB_SPEED_VARIABLE: xhci_dbg(xhci, "FIXME xHCI doesn't support wireless speeds\n"); @@ -327,7 +378,7 @@ int xhci_setup_addressable_virt_dev(struct xhci_hcd *xhci, struct usb_device *ud for (top_dev = udev; top_dev->parent && top_dev->parent->parent; top_dev = top_dev->parent) /* Found device below root hub */; - dev->in_ctx->slot.dev_info2 |= (u32) ROOT_HUB_PORT(top_dev->portnum); + slot_ctx->dev_info2 |= (u32) ROOT_HUB_PORT(top_dev->portnum); xhci_dbg(xhci, "Set root hub portnum to %d\n", top_dev->portnum); /* Is this a LS/FS device under a HS hub? */ @@ -337,8 +388,8 @@ int xhci_setup_addressable_virt_dev(struct xhci_hcd *xhci, struct usb_device *ud */ if ((udev->speed == USB_SPEED_LOW || udev->speed == USB_SPEED_FULL) && udev->tt) { - dev->in_ctx->slot.tt_info = udev->tt->hub->slot_id; - dev->in_ctx->slot.tt_info |= udev->ttport << 8; + slot_ctx->tt_info = udev->tt->hub->slot_id; + slot_ctx->tt_info |= udev->ttport << 8; } xhci_dbg(xhci, "udev->tt = %p\n", udev->tt); xhci_dbg(xhci, "udev->ttport = 0x%x\n", udev->ttport); @@ -360,10 +411,9 @@ int xhci_setup_addressable_virt_dev(struct xhci_hcd *xhci, struct usb_device *ud ep0_ctx->ep_info2 |= MAX_BURST(0); ep0_ctx->ep_info2 |= ERROR_COUNT(3); - ep0_ctx->deq[0] = + ep0_ctx->deq = dev->ep_rings[0]->first_seg->dma; - ep0_ctx->deq[0] |= dev->ep_rings[0]->cycle_state; - ep0_ctx->deq[1] = 0; + ep0_ctx->deq |= dev->ep_rings[0]->cycle_state; /* Steps 7 and 8 were done in xhci_alloc_virt_device() */ @@ -470,25 +520,26 @@ int xhci_endpoint_init(struct xhci_hcd *xhci, unsigned int max_burst; ep_index = xhci_get_endpoint_index(&ep->desc); - ep_ctx = &virt_dev->in_ctx->ep[ep_index]; + ep_ctx = xhci_get_ep_ctx(xhci, virt_dev->in_ctx, ep_index); /* Set up the endpoint ring */ virt_dev->new_ep_rings[ep_index] = xhci_ring_alloc(xhci, 1, true, mem_flags); if (!virt_dev->new_ep_rings[ep_index]) return -ENOMEM; ep_ring = virt_dev->new_ep_rings[ep_index]; - ep_ctx->deq[0] = ep_ring->first_seg->dma | ep_ring->cycle_state; - ep_ctx->deq[1] = 0; + ep_ctx->deq = ep_ring->first_seg->dma | ep_ring->cycle_state; ep_ctx->ep_info = xhci_get_endpoint_interval(udev, ep); /* FIXME dig Mult and streams info out of ep companion desc */ - /* Allow 3 retries for everything but isoc */ + /* Allow 3 retries for everything but isoc; + * error count = 0 means infinite retries. + */ if (!usb_endpoint_xfer_isoc(&ep->desc)) ep_ctx->ep_info2 = ERROR_COUNT(3); else - ep_ctx->ep_info2 = ERROR_COUNT(0); + ep_ctx->ep_info2 = ERROR_COUNT(1); ep_ctx->ep_info2 |= xhci_get_endpoint_type(udev, ep); @@ -498,7 +549,12 @@ int xhci_endpoint_init(struct xhci_hcd *xhci, max_packet = ep->desc.wMaxPacketSize; ep_ctx->ep_info2 |= MAX_PACKET(max_packet); /* dig out max burst from ep companion desc */ - max_packet = ep->ss_ep_comp->desc.bMaxBurst; + if (!ep->ss_ep_comp) { + xhci_warn(xhci, "WARN no SS endpoint companion descriptor.\n"); + max_packet = 0; + } else { + max_packet = ep->ss_ep_comp->desc.bMaxBurst; + } ep_ctx->ep_info2 |= MAX_BURST(max_packet); break; case USB_SPEED_HIGH: @@ -531,18 +587,114 @@ void xhci_endpoint_zero(struct xhci_hcd *xhci, struct xhci_ep_ctx *ep_ctx; ep_index = xhci_get_endpoint_index(&ep->desc); - ep_ctx = &virt_dev->in_ctx->ep[ep_index]; + ep_ctx = xhci_get_ep_ctx(xhci, virt_dev->in_ctx, ep_index); ep_ctx->ep_info = 0; ep_ctx->ep_info2 = 0; - ep_ctx->deq[0] = 0; - ep_ctx->deq[1] = 0; + ep_ctx->deq = 0; ep_ctx->tx_info = 0; /* Don't free the endpoint ring until the set interface or configuration * request succeeds. */ } +/* Set up the scratchpad buffer array and scratchpad buffers, if needed. */ +static int scratchpad_alloc(struct xhci_hcd *xhci, gfp_t flags) +{ + int i; + struct device *dev = xhci_to_hcd(xhci)->self.controller; + int num_sp = HCS_MAX_SCRATCHPAD(xhci->hcs_params2); + + xhci_dbg(xhci, "Allocating %d scratchpad buffers\n", num_sp); + + if (!num_sp) + return 0; + + xhci->scratchpad = kzalloc(sizeof(*xhci->scratchpad), flags); + if (!xhci->scratchpad) + goto fail_sp; + + xhci->scratchpad->sp_array = + pci_alloc_consistent(to_pci_dev(dev), + num_sp * sizeof(u64), + &xhci->scratchpad->sp_dma); + if (!xhci->scratchpad->sp_array) + goto fail_sp2; + + xhci->scratchpad->sp_buffers = kzalloc(sizeof(void *) * num_sp, flags); + if (!xhci->scratchpad->sp_buffers) + goto fail_sp3; + + xhci->scratchpad->sp_dma_buffers = + kzalloc(sizeof(dma_addr_t) * num_sp, flags); + + if (!xhci->scratchpad->sp_dma_buffers) + goto fail_sp4; + + xhci->dcbaa->dev_context_ptrs[0] = xhci->scratchpad->sp_dma; + for (i = 0; i < num_sp; i++) { + dma_addr_t dma; + void *buf = pci_alloc_consistent(to_pci_dev(dev), + xhci->page_size, &dma); + if (!buf) + goto fail_sp5; + + xhci->scratchpad->sp_array[i] = dma; + xhci->scratchpad->sp_buffers[i] = buf; + xhci->scratchpad->sp_dma_buffers[i] = dma; + } + + return 0; + + fail_sp5: + for (i = i - 1; i >= 0; i--) { + pci_free_consistent(to_pci_dev(dev), xhci->page_size, + xhci->scratchpad->sp_buffers[i], + xhci->scratchpad->sp_dma_buffers[i]); + } + kfree(xhci->scratchpad->sp_dma_buffers); + + fail_sp4: + kfree(xhci->scratchpad->sp_buffers); + + fail_sp3: + pci_free_consistent(to_pci_dev(dev), num_sp * sizeof(u64), + xhci->scratchpad->sp_array, + xhci->scratchpad->sp_dma); + + fail_sp2: + kfree(xhci->scratchpad); + xhci->scratchpad = NULL; + + fail_sp: + return -ENOMEM; +} + +static void scratchpad_free(struct xhci_hcd *xhci) +{ + int num_sp; + int i; + struct pci_dev *pdev = to_pci_dev(xhci_to_hcd(xhci)->self.controller); + + if (!xhci->scratchpad) + return; + + num_sp = HCS_MAX_SCRATCHPAD(xhci->hcs_params2); + + for (i = 0; i < num_sp; i++) { + pci_free_consistent(pdev, xhci->page_size, + xhci->scratchpad->sp_buffers[i], + xhci->scratchpad->sp_dma_buffers[i]); + } + kfree(xhci->scratchpad->sp_dma_buffers); + kfree(xhci->scratchpad->sp_buffers); + pci_free_consistent(pdev, num_sp * sizeof(u64), + xhci->scratchpad->sp_array, + xhci->scratchpad->sp_dma); + kfree(xhci->scratchpad); + xhci->scratchpad = NULL; +} + void xhci_mem_cleanup(struct xhci_hcd *xhci) { struct pci_dev *pdev = to_pci_dev(xhci_to_hcd(xhci)->self.controller); @@ -551,10 +703,8 @@ void xhci_mem_cleanup(struct xhci_hcd *xhci) /* Free the Event Ring Segment Table and the actual Event Ring */ xhci_writel(xhci, 0, &xhci->ir_set->erst_size); - xhci_writel(xhci, 0, &xhci->ir_set->erst_base[0]); - xhci_writel(xhci, 0, &xhci->ir_set->erst_base[1]); - xhci_writel(xhci, 0, &xhci->ir_set->erst_dequeue[0]); - xhci_writel(xhci, 0, &xhci->ir_set->erst_dequeue[1]); + xhci_write_64(xhci, 0, &xhci->ir_set->erst_base); + xhci_write_64(xhci, 0, &xhci->ir_set->erst_dequeue); size = sizeof(struct xhci_erst_entry)*(xhci->erst.num_entries); if (xhci->erst.entries) pci_free_consistent(pdev, size, @@ -566,8 +716,7 @@ void xhci_mem_cleanup(struct xhci_hcd *xhci) xhci->event_ring = NULL; xhci_dbg(xhci, "Freed event ring\n"); - xhci_writel(xhci, 0, &xhci->op_regs->cmd_ring[0]); - xhci_writel(xhci, 0, &xhci->op_regs->cmd_ring[1]); + xhci_write_64(xhci, 0, &xhci->op_regs->cmd_ring); if (xhci->cmd_ring) xhci_ring_free(xhci, xhci->cmd_ring); xhci->cmd_ring = NULL; @@ -586,8 +735,7 @@ void xhci_mem_cleanup(struct xhci_hcd *xhci) xhci->device_pool = NULL; xhci_dbg(xhci, "Freed device context pool\n"); - xhci_writel(xhci, 0, &xhci->op_regs->dcbaa_ptr[0]); - xhci_writel(xhci, 0, &xhci->op_regs->dcbaa_ptr[1]); + xhci_write_64(xhci, 0, &xhci->op_regs->dcbaa_ptr); if (xhci->dcbaa) pci_free_consistent(pdev, sizeof(*xhci->dcbaa), xhci->dcbaa, xhci->dcbaa->dma); @@ -595,6 +743,7 @@ void xhci_mem_cleanup(struct xhci_hcd *xhci) xhci->page_size = 0; xhci->page_shift = 0; + scratchpad_free(xhci); } int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags) @@ -602,6 +751,7 @@ int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags) dma_addr_t dma; struct device *dev = xhci_to_hcd(xhci)->self.controller; unsigned int val, val2; + u64 val_64; struct xhci_segment *seg; u32 page_size; int i; @@ -647,8 +797,7 @@ int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags) xhci->dcbaa->dma = dma; xhci_dbg(xhci, "// Device context base array address = 0x%llx (DMA), %p (virt)\n", (unsigned long long)xhci->dcbaa->dma, xhci->dcbaa); - xhci_writel(xhci, dma, &xhci->op_regs->dcbaa_ptr[0]); - xhci_writel(xhci, (u32) 0, &xhci->op_regs->dcbaa_ptr[1]); + xhci_write_64(xhci, dma, &xhci->op_regs->dcbaa_ptr); /* * Initialize the ring segment pool. The ring must be a contiguous @@ -658,11 +807,10 @@ int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags) */ xhci->segment_pool = dma_pool_create("xHCI ring segments", dev, SEGMENT_SIZE, 64, xhci->page_size); + /* See Table 46 and Note on Figure 55 */ - /* FIXME support 64-byte contexts */ xhci->device_pool = dma_pool_create("xHCI input/output contexts", dev, - sizeof(struct xhci_device_control), - 64, xhci->page_size); + 2112, 64, xhci->page_size); if (!xhci->segment_pool || !xhci->device_pool) goto fail; @@ -675,14 +823,12 @@ int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags) (unsigned long long)xhci->cmd_ring->first_seg->dma); /* Set the address in the Command Ring Control register */ - val = xhci_readl(xhci, &xhci->op_regs->cmd_ring[0]); - val = (val & ~CMD_RING_ADDR_MASK) | - (xhci->cmd_ring->first_seg->dma & CMD_RING_ADDR_MASK) | + val_64 = xhci_read_64(xhci, &xhci->op_regs->cmd_ring); + val_64 = (val_64 & (u64) CMD_RING_RSVD_BITS) | + (xhci->cmd_ring->first_seg->dma & (u64) ~CMD_RING_RSVD_BITS) | xhci->cmd_ring->cycle_state; - xhci_dbg(xhci, "// Setting command ring address low bits to 0x%x\n", val); - xhci_writel(xhci, val, &xhci->op_regs->cmd_ring[0]); - xhci_dbg(xhci, "// Setting command ring address high bits to 0x0\n"); - xhci_writel(xhci, (u32) 0, &xhci->op_regs->cmd_ring[1]); + xhci_dbg(xhci, "// Setting command ring address to 0x%x\n", val); + xhci_write_64(xhci, val_64, &xhci->op_regs->cmd_ring); xhci_dbg_cmd_ptrs(xhci); val = xhci_readl(xhci, &xhci->cap_regs->db_off); @@ -722,8 +868,7 @@ int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags) /* set ring base address and size for each segment table entry */ for (val = 0, seg = xhci->event_ring->first_seg; val < ERST_NUM_SEGS; val++) { struct xhci_erst_entry *entry = &xhci->erst.entries[val]; - entry->seg_addr[0] = seg->dma; - entry->seg_addr[1] = 0; + entry->seg_addr = seg->dma; entry->seg_size = TRBS_PER_SEGMENT; entry->rsvd = 0; seg = seg->next; @@ -741,11 +886,10 @@ int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags) /* set the segment table base address */ xhci_dbg(xhci, "// Set ERST base address for ir_set 0 = 0x%llx\n", (unsigned long long)xhci->erst.erst_dma_addr); - val = xhci_readl(xhci, &xhci->ir_set->erst_base[0]); - val &= ERST_PTR_MASK; - val |= (xhci->erst.erst_dma_addr & ~ERST_PTR_MASK); - xhci_writel(xhci, val, &xhci->ir_set->erst_base[0]); - xhci_writel(xhci, 0, &xhci->ir_set->erst_base[1]); + val_64 = xhci_read_64(xhci, &xhci->ir_set->erst_base); + val_64 &= ERST_PTR_MASK; + val_64 |= (xhci->erst.erst_dma_addr & (u64) ~ERST_PTR_MASK); + xhci_write_64(xhci, val_64, &xhci->ir_set->erst_base); /* Set the event ring dequeue address */ xhci_set_hc_event_deq(xhci); @@ -761,7 +905,11 @@ int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags) for (i = 0; i < MAX_HC_SLOTS; ++i) xhci->devs[i] = 0; + if (scratchpad_alloc(xhci, flags)) + goto fail; + return 0; + fail: xhci_warn(xhci, "Couldn't initialize memory\n"); xhci_mem_cleanup(xhci); diff --git a/drivers/usb/host/xhci-pci.c b/drivers/usb/host/xhci-pci.c index 1462709e26c..592fe7e623f 100644 --- a/drivers/usb/host/xhci-pci.c +++ b/drivers/usb/host/xhci-pci.c @@ -117,6 +117,7 @@ static const struct hc_driver xhci_pci_hc_driver = { .free_dev = xhci_free_dev, .add_endpoint = xhci_add_endpoint, .drop_endpoint = xhci_drop_endpoint, + .endpoint_reset = xhci_endpoint_reset, .check_bandwidth = xhci_check_bandwidth, .reset_bandwidth = xhci_reset_bandwidth, .address_device = xhci_address_device, diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c index 02d81985c45..aa88a067148 100644 --- a/drivers/usb/host/xhci-ring.c +++ b/drivers/usb/host/xhci-ring.c @@ -135,6 +135,7 @@ static void next_trb(struct xhci_hcd *xhci, static void inc_deq(struct xhci_hcd *xhci, struct xhci_ring *ring, bool consumer) { union xhci_trb *next = ++(ring->dequeue); + unsigned long long addr; ring->deq_updates++; /* Update the dequeue pointer further if that was a link TRB or we're at @@ -152,6 +153,13 @@ static void inc_deq(struct xhci_hcd *xhci, struct xhci_ring *ring, bool consumer ring->dequeue = ring->deq_seg->trbs; next = ring->dequeue; } + addr = (unsigned long long) xhci_trb_virt_to_dma(ring->deq_seg, ring->dequeue); + if (ring == xhci->event_ring) + xhci_dbg(xhci, "Event ring deq = 0x%llx (DMA)\n", addr); + else if (ring == xhci->cmd_ring) + xhci_dbg(xhci, "Command ring deq = 0x%llx (DMA)\n", addr); + else + xhci_dbg(xhci, "Ring deq = 0x%llx (DMA)\n", addr); } /* @@ -171,6 +179,7 @@ static void inc_enq(struct xhci_hcd *xhci, struct xhci_ring *ring, bool consumer { u32 chain; union xhci_trb *next; + unsigned long long addr; chain = ring->enqueue->generic.field[3] & TRB_CHAIN; next = ++(ring->enqueue); @@ -204,6 +213,13 @@ static void inc_enq(struct xhci_hcd *xhci, struct xhci_ring *ring, bool consumer ring->enqueue = ring->enq_seg->trbs; next = ring->enqueue; } + addr = (unsigned long long) xhci_trb_virt_to_dma(ring->enq_seg, ring->enqueue); + if (ring == xhci->event_ring) + xhci_dbg(xhci, "Event ring enq = 0x%llx (DMA)\n", addr); + else if (ring == xhci->cmd_ring) + xhci_dbg(xhci, "Command ring enq = 0x%llx (DMA)\n", addr); + else + xhci_dbg(xhci, "Ring enq = 0x%llx (DMA)\n", addr); } /* @@ -237,7 +253,7 @@ static int room_on_ring(struct xhci_hcd *xhci, struct xhci_ring *ring, void xhci_set_hc_event_deq(struct xhci_hcd *xhci) { - u32 temp; + u64 temp; dma_addr_t deq; deq = xhci_trb_virt_to_dma(xhci->event_ring->deq_seg, @@ -246,13 +262,15 @@ void xhci_set_hc_event_deq(struct xhci_hcd *xhci) xhci_warn(xhci, "WARN something wrong with SW event ring " "dequeue ptr.\n"); /* Update HC event ring dequeue pointer */ - temp = xhci_readl(xhci, &xhci->ir_set->erst_dequeue[0]); + temp = xhci_read_64(xhci, &xhci->ir_set->erst_dequeue); temp &= ERST_PTR_MASK; - if (!in_interrupt()) - xhci_dbg(xhci, "// Write event ring dequeue pointer\n"); - xhci_writel(xhci, 0, &xhci->ir_set->erst_dequeue[1]); - xhci_writel(xhci, (deq & ~ERST_PTR_MASK) | temp, - &xhci->ir_set->erst_dequeue[0]); + /* Don't clear the EHB bit (which is RW1C) because + * there might be more events to service. + */ + temp &= ~ERST_EHB; + xhci_dbg(xhci, "// Write event ring dequeue pointer, preserving EHB bit\n"); + xhci_write_64(xhci, ((u64) deq & (u64) ~ERST_PTR_MASK) | temp, + &xhci->ir_set->erst_dequeue); } /* Ring the host controller doorbell after placing a command on the ring */ @@ -279,7 +297,8 @@ static void ring_ep_doorbell(struct xhci_hcd *xhci, /* Don't ring the doorbell for this endpoint if there are pending * cancellations because the we don't want to interrupt processing. */ - if (!ep_ring->cancels_pending && !(ep_ring->state & SET_DEQ_PENDING)) { + if (!ep_ring->cancels_pending && !(ep_ring->state & SET_DEQ_PENDING) + && !(ep_ring->state & EP_HALTED)) { field = xhci_readl(xhci, db_addr) & DB_MASK; xhci_writel(xhci, field | EPI_TO_DB(ep_index), db_addr); /* Flush PCI posted writes - FIXME Matthew Wilcox says this @@ -316,12 +335,6 @@ static struct xhci_segment *find_trb_seg( return cur_seg; } -struct dequeue_state { - struct xhci_segment *new_deq_seg; - union xhci_trb *new_deq_ptr; - int new_cycle_state; -}; - /* * Move the xHC's endpoint ring dequeue pointer past cur_td. * Record the new state of the xHC's endpoint ring dequeue segment, @@ -336,24 +349,30 @@ struct dequeue_state { * - Finally we move the dequeue state one TRB further, toggling the cycle bit * if we've moved it past a link TRB with the toggle cycle bit set. */ -static void find_new_dequeue_state(struct xhci_hcd *xhci, +void xhci_find_new_dequeue_state(struct xhci_hcd *xhci, unsigned int slot_id, unsigned int ep_index, - struct xhci_td *cur_td, struct dequeue_state *state) + struct xhci_td *cur_td, struct xhci_dequeue_state *state) { struct xhci_virt_device *dev = xhci->devs[slot_id]; struct xhci_ring *ep_ring = dev->ep_rings[ep_index]; struct xhci_generic_trb *trb; + struct xhci_ep_ctx *ep_ctx; + dma_addr_t addr; state->new_cycle_state = 0; + xhci_dbg(xhci, "Finding segment containing stopped TRB.\n"); state->new_deq_seg = find_trb_seg(cur_td->start_seg, ep_ring->stopped_trb, &state->new_cycle_state); if (!state->new_deq_seg) BUG(); /* Dig out the cycle state saved by the xHC during the stop ep cmd */ - state->new_cycle_state = 0x1 & dev->out_ctx->ep[ep_index].deq[0]; + xhci_dbg(xhci, "Finding endpoint context\n"); + ep_ctx = xhci_get_ep_ctx(xhci, dev->out_ctx, ep_index); + state->new_cycle_state = 0x1 & ep_ctx->deq; state->new_deq_ptr = cur_td->last_trb; + xhci_dbg(xhci, "Finding segment containing last TRB in TD.\n"); state->new_deq_seg = find_trb_seg(state->new_deq_seg, state->new_deq_ptr, &state->new_cycle_state); @@ -367,6 +386,12 @@ static void find_new_dequeue_state(struct xhci_hcd *xhci, next_trb(xhci, ep_ring, &state->new_deq_seg, &state->new_deq_ptr); /* Don't update the ring cycle state for the producer (us). */ + xhci_dbg(xhci, "New dequeue segment = %p (virtual)\n", + state->new_deq_seg); + addr = xhci_trb_virt_to_dma(state->new_deq_seg, state->new_deq_ptr); + xhci_dbg(xhci, "New dequeue pointer = 0x%llx (DMA)\n", + (unsigned long long) addr); + xhci_dbg(xhci, "Setting dequeue pointer in internal ring state.\n"); ep_ring->dequeue = state->new_deq_ptr; ep_ring->deq_seg = state->new_deq_seg; } @@ -416,6 +441,30 @@ static int queue_set_tr_deq(struct xhci_hcd *xhci, int slot_id, unsigned int ep_index, struct xhci_segment *deq_seg, union xhci_trb *deq_ptr, u32 cycle_state); +void xhci_queue_new_dequeue_state(struct xhci_hcd *xhci, + struct xhci_ring *ep_ring, unsigned int slot_id, + unsigned int ep_index, struct xhci_dequeue_state *deq_state) +{ + xhci_dbg(xhci, "Set TR Deq Ptr cmd, new deq seg = %p (0x%llx dma), " + "new deq ptr = %p (0x%llx dma), new cycle = %u\n", + deq_state->new_deq_seg, + (unsigned long long)deq_state->new_deq_seg->dma, + deq_state->new_deq_ptr, + (unsigned long long)xhci_trb_virt_to_dma(deq_state->new_deq_seg, deq_state->new_deq_ptr), + deq_state->new_cycle_state); + queue_set_tr_deq(xhci, slot_id, ep_index, + deq_state->new_deq_seg, + deq_state->new_deq_ptr, + (u32) deq_state->new_cycle_state); + /* Stop the TD queueing code from ringing the doorbell until + * this command completes. The HC won't set the dequeue pointer + * if the ring is running, and ringing the doorbell starts the + * ring running. + */ + ep_ring->state |= SET_DEQ_PENDING; + xhci_ring_cmd_db(xhci); +} + /* * When we get a command completion for a Stop Endpoint Command, we need to * unlink any cancelled TDs from the ring. There are two ways to do that: @@ -436,7 +485,7 @@ static void handle_stopped_endpoint(struct xhci_hcd *xhci, struct xhci_td *cur_td = 0; struct xhci_td *last_unlinked_td; - struct dequeue_state deq_state; + struct xhci_dequeue_state deq_state; #ifdef CONFIG_USB_HCD_STAT ktime_t stop_time = ktime_get(); #endif @@ -464,7 +513,7 @@ static void handle_stopped_endpoint(struct xhci_hcd *xhci, * move the xHC endpoint ring dequeue pointer past this TD. */ if (cur_td == ep_ring->stopped_td) - find_new_dequeue_state(xhci, slot_id, ep_index, cur_td, + xhci_find_new_dequeue_state(xhci, slot_id, ep_index, cur_td, &deq_state); else td_to_noop(xhci, ep_ring, cur_td); @@ -480,24 +529,8 @@ static void handle_stopped_endpoint(struct xhci_hcd *xhci, /* If necessary, queue a Set Transfer Ring Dequeue Pointer command */ if (deq_state.new_deq_ptr && deq_state.new_deq_seg) { - xhci_dbg(xhci, "Set TR Deq Ptr cmd, new deq seg = %p (0x%llx dma), " - "new deq ptr = %p (0x%llx dma), new cycle = %u\n", - deq_state.new_deq_seg, - (unsigned long long)deq_state.new_deq_seg->dma, - deq_state.new_deq_ptr, - (unsigned long long)xhci_trb_virt_to_dma(deq_state.new_deq_seg, deq_state.new_deq_ptr), - deq_state.new_cycle_state); - queue_set_tr_deq(xhci, slot_id, ep_index, - deq_state.new_deq_seg, - deq_state.new_deq_ptr, - (u32) deq_state.new_cycle_state); - /* Stop the TD queueing code from ringing the doorbell until - * this command completes. The HC won't set the dequeue pointer - * if the ring is running, and ringing the doorbell starts the - * ring running. - */ - ep_ring->state |= SET_DEQ_PENDING; - xhci_ring_cmd_db(xhci); + xhci_queue_new_dequeue_state(xhci, ep_ring, + slot_id, ep_index, &deq_state); } else { /* Otherwise just ring the doorbell to restart the ring */ ring_ep_doorbell(xhci, slot_id, ep_index); @@ -551,11 +584,15 @@ static void handle_set_deq_completion(struct xhci_hcd *xhci, unsigned int ep_index; struct xhci_ring *ep_ring; struct xhci_virt_device *dev; + struct xhci_ep_ctx *ep_ctx; + struct xhci_slot_ctx *slot_ctx; slot_id = TRB_TO_SLOT_ID(trb->generic.field[3]); ep_index = TRB_TO_EP_INDEX(trb->generic.field[3]); dev = xhci->devs[slot_id]; ep_ring = dev->ep_rings[ep_index]; + ep_ctx = xhci_get_ep_ctx(xhci, dev->out_ctx, ep_index); + slot_ctx = xhci_get_slot_ctx(xhci, dev->out_ctx); if (GET_COMP_CODE(event->status) != COMP_SUCCESS) { unsigned int ep_state; @@ -569,9 +606,9 @@ static void handle_set_deq_completion(struct xhci_hcd *xhci, case COMP_CTX_STATE: xhci_warn(xhci, "WARN Set TR Deq Ptr cmd failed due " "to incorrect slot or ep state.\n"); - ep_state = dev->out_ctx->ep[ep_index].ep_info; + ep_state = ep_ctx->ep_info; ep_state &= EP_STATE_MASK; - slot_state = dev->out_ctx->slot.dev_state; + slot_state = slot_ctx->dev_state; slot_state = GET_SLOT_STATE(slot_state); xhci_dbg(xhci, "Slot state = %u, EP state = %u\n", slot_state, ep_state); @@ -593,16 +630,33 @@ static void handle_set_deq_completion(struct xhci_hcd *xhci, * cancelling URBs, which might not be an error... */ } else { - xhci_dbg(xhci, "Successful Set TR Deq Ptr cmd, deq[0] = 0x%x, " - "deq[1] = 0x%x.\n", - dev->out_ctx->ep[ep_index].deq[0], - dev->out_ctx->ep[ep_index].deq[1]); + xhci_dbg(xhci, "Successful Set TR Deq Ptr cmd, deq = @%08llx\n", + ep_ctx->deq); } ep_ring->state &= ~SET_DEQ_PENDING; ring_ep_doorbell(xhci, slot_id, ep_index); } +static void handle_reset_ep_completion(struct xhci_hcd *xhci, + struct xhci_event_cmd *event, + union xhci_trb *trb) +{ + int slot_id; + unsigned int ep_index; + + slot_id = TRB_TO_SLOT_ID(trb->generic.field[3]); + ep_index = TRB_TO_EP_INDEX(trb->generic.field[3]); + /* This command will only fail if the endpoint wasn't halted, + * but we don't care. + */ + xhci_dbg(xhci, "Ignoring reset ep completion code of %u\n", + (unsigned int) GET_COMP_CODE(event->status)); + + /* Clear our internal halted state and restart the ring */ + xhci->devs[slot_id]->ep_rings[ep_index]->state &= ~EP_HALTED; + ring_ep_doorbell(xhci, slot_id, ep_index); +} static void handle_cmd_completion(struct xhci_hcd *xhci, struct xhci_event_cmd *event) @@ -611,7 +665,7 @@ static void handle_cmd_completion(struct xhci_hcd *xhci, u64 cmd_dma; dma_addr_t cmd_dequeue_dma; - cmd_dma = (((u64) event->cmd_trb[1]) << 32) + event->cmd_trb[0]; + cmd_dma = event->cmd_trb; cmd_dequeue_dma = xhci_trb_virt_to_dma(xhci->cmd_ring->deq_seg, xhci->cmd_ring->dequeue); /* Is the command ring deq ptr out of sync with the deq seg ptr? */ @@ -653,6 +707,9 @@ static void handle_cmd_completion(struct xhci_hcd *xhci, case TRB_TYPE(TRB_CMD_NOOP): ++xhci->noops_handled; break; + case TRB_TYPE(TRB_RESET_EP): + handle_reset_ep_completion(xhci, event, xhci->cmd_ring->dequeue); + break; default: /* Skip over unknown commands on the event ring */ xhci->error_bitmask |= 1 << 6; @@ -756,7 +813,9 @@ static int handle_tx_event(struct xhci_hcd *xhci, union xhci_trb *event_trb; struct urb *urb = 0; int status = -EINPROGRESS; + struct xhci_ep_ctx *ep_ctx; + xhci_dbg(xhci, "In %s\n", __func__); xdev = xhci->devs[TRB_TO_SLOT_ID(event->flags)]; if (!xdev) { xhci_err(xhci, "ERROR Transfer event pointed to bad slot\n"); @@ -765,17 +824,17 @@ static int handle_tx_event(struct xhci_hcd *xhci, /* Endpoint ID is 1 based, our index is zero based */ ep_index = TRB_TO_EP_ID(event->flags) - 1; + xhci_dbg(xhci, "%s - ep index = %d\n", __func__, ep_index); ep_ring = xdev->ep_rings[ep_index]; - if (!ep_ring || (xdev->out_ctx->ep[ep_index].ep_info & EP_STATE_MASK) == EP_STATE_DISABLED) { + ep_ctx = xhci_get_ep_ctx(xhci, xdev->out_ctx, ep_index); + if (!ep_ring || (ep_ctx->ep_info & EP_STATE_MASK) == EP_STATE_DISABLED) { xhci_err(xhci, "ERROR Transfer event pointed to disabled endpoint\n"); return -ENODEV; } - event_dma = event->buffer[0]; - if (event->buffer[1] != 0) - xhci_warn(xhci, "WARN ignoring upper 32-bits of 64-bit TRB dma address\n"); - + event_dma = event->buffer; /* This TRB should be in the TD at the head of this ring's TD list */ + xhci_dbg(xhci, "%s - checking for list empty\n", __func__); if (list_empty(&ep_ring->td_list)) { xhci_warn(xhci, "WARN Event TRB for slot %d ep %d with no TDs queued?\n", TRB_TO_SLOT_ID(event->flags), ep_index); @@ -785,11 +844,14 @@ static int handle_tx_event(struct xhci_hcd *xhci, urb = NULL; goto cleanup; } + xhci_dbg(xhci, "%s - getting list entry\n", __func__); td = list_entry(ep_ring->td_list.next, struct xhci_td, td_list); /* Is this a TRB in the currently executing TD? */ + xhci_dbg(xhci, "%s - looking for TD\n", __func__); event_seg = trb_in_td(ep_ring->deq_seg, ep_ring->dequeue, td->last_trb, event_dma); + xhci_dbg(xhci, "%s - found event_seg = %p\n", __func__, event_seg); if (!event_seg) { /* HC is busted, give up! */ xhci_err(xhci, "ERROR Transfer event TRB DMA ptr not part of current TD\n"); @@ -798,10 +860,10 @@ static int handle_tx_event(struct xhci_hcd *xhci, event_trb = &event_seg->trbs[(event_dma - event_seg->dma) / sizeof(*event_trb)]; xhci_dbg(xhci, "Event TRB with TRB type ID %u\n", (unsigned int) (event->flags & TRB_TYPE_BITMASK)>>10); - xhci_dbg(xhci, "Offset 0x00 (buffer[0]) = 0x%x\n", - (unsigned int) event->buffer[0]); - xhci_dbg(xhci, "Offset 0x04 (buffer[0]) = 0x%x\n", - (unsigned int) event->buffer[1]); + xhci_dbg(xhci, "Offset 0x00 (buffer lo) = 0x%x\n", + lower_32_bits(event->buffer)); + xhci_dbg(xhci, "Offset 0x04 (buffer hi) = 0x%x\n", + upper_32_bits(event->buffer)); xhci_dbg(xhci, "Offset 0x08 (transfer length) = 0x%x\n", (unsigned int) event->transfer_len); xhci_dbg(xhci, "Offset 0x0C (flags) = 0x%x\n", @@ -823,6 +885,7 @@ static int handle_tx_event(struct xhci_hcd *xhci, break; case COMP_STALL: xhci_warn(xhci, "WARN: Stalled endpoint\n"); + ep_ring->state |= EP_HALTED; status = -EPIPE; break; case COMP_TRB_ERR: @@ -833,6 +896,10 @@ static int handle_tx_event(struct xhci_hcd *xhci, xhci_warn(xhci, "WARN: transfer error on endpoint\n"); status = -EPROTO; break; + case COMP_BABBLE: + xhci_warn(xhci, "WARN: babble error on endpoint\n"); + status = -EOVERFLOW; + break; case COMP_DB_ERR: xhci_warn(xhci, "WARN: HC couldn't access mem fast enough\n"); status = -ENOSR; @@ -874,15 +941,26 @@ static int handle_tx_event(struct xhci_hcd *xhci, if (event_trb != ep_ring->dequeue) { /* The event was for the status stage */ if (event_trb == td->last_trb) { - td->urb->actual_length = - td->urb->transfer_buffer_length; + if (td->urb->actual_length != 0) { + /* Don't overwrite a previously set error code */ + if (status == -EINPROGRESS || status == 0) + /* Did we already see a short data stage? */ + status = -EREMOTEIO; + } else { + td->urb->actual_length = + td->urb->transfer_buffer_length; + } } else { /* Maybe the event was for the data stage? */ - if (GET_COMP_CODE(event->transfer_len) != COMP_STOP_INVAL) + if (GET_COMP_CODE(event->transfer_len) != COMP_STOP_INVAL) { /* We didn't stop on a link TRB in the middle */ td->urb->actual_length = td->urb->transfer_buffer_length - TRB_LEN(event->transfer_len); + xhci_dbg(xhci, "Waiting for status stage event\n"); + urb = NULL; + goto cleanup; + } } } } else { @@ -929,16 +1007,20 @@ static int handle_tx_event(struct xhci_hcd *xhci, TRB_LEN(event->transfer_len)); td->urb->actual_length = 0; } - if (td->urb->transfer_flags & URB_SHORT_NOT_OK) - status = -EREMOTEIO; - else - status = 0; + /* Don't overwrite a previously set error code */ + if (status == -EINPROGRESS) { + if (td->urb->transfer_flags & URB_SHORT_NOT_OK) + status = -EREMOTEIO; + else + status = 0; + } } else { td->urb->actual_length = td->urb->transfer_buffer_length; /* Ignore a short packet completion if the * untransferred length was zero. */ - status = 0; + if (status == -EREMOTEIO) + status = 0; } } else { /* Slow path - walk the list, starting from the dequeue @@ -965,19 +1047,30 @@ static int handle_tx_event(struct xhci_hcd *xhci, TRB_LEN(event->transfer_len); } } - /* The Endpoint Stop Command completion will take care of - * any stopped TDs. A stopped TD may be restarted, so don't update the - * ring dequeue pointer or take this TD off any lists yet. - */ if (GET_COMP_CODE(event->transfer_len) == COMP_STOP_INVAL || GET_COMP_CODE(event->transfer_len) == COMP_STOP) { + /* The Endpoint Stop Command completion will take care of any + * stopped TDs. A stopped TD may be restarted, so don't update + * the ring dequeue pointer or take this TD off any lists yet. + */ ep_ring->stopped_td = td; ep_ring->stopped_trb = event_trb; } else { - /* Update ring dequeue pointer */ - while (ep_ring->dequeue != td->last_trb) + if (GET_COMP_CODE(event->transfer_len) == COMP_STALL) { + /* The transfer is completed from the driver's + * perspective, but we need to issue a set dequeue + * command for this stalled endpoint to move the dequeue + * pointer past the TD. We can't do that here because + * the halt condition must be cleared first. + */ + ep_ring->stopped_td = td; + ep_ring->stopped_trb = event_trb; + } else { + /* Update ring dequeue pointer */ + while (ep_ring->dequeue != td->last_trb) + inc_deq(xhci, ep_ring, false); inc_deq(xhci, ep_ring, false); - inc_deq(xhci, ep_ring, false); + } /* Clean up the endpoint's TD list */ urb = td->urb; @@ -987,7 +1080,10 @@ static int handle_tx_event(struct xhci_hcd *xhci, list_del(&td->cancelled_td_list); ep_ring->cancels_pending--; } - kfree(td); + /* Leave the TD around for the reset endpoint function to use */ + if (GET_COMP_CODE(event->transfer_len) != COMP_STALL) { + kfree(td); + } urb->hcpriv = NULL; } cleanup: @@ -997,6 +1093,8 @@ cleanup: /* FIXME for multi-TD URBs (who have buffers bigger than 64MB) */ if (urb) { usb_hcd_unlink_urb_from_ep(xhci_to_hcd(xhci), urb); + xhci_dbg(xhci, "Giveback URB %p, len = %d, status = %d\n", + urb, td->urb->actual_length, status); spin_unlock(&xhci->lock); usb_hcd_giveback_urb(xhci_to_hcd(xhci), urb, status); spin_lock(&xhci->lock); @@ -1014,6 +1112,7 @@ void xhci_handle_event(struct xhci_hcd *xhci) int update_ptrs = 1; int ret; + xhci_dbg(xhci, "In %s\n", __func__); if (!xhci->event_ring || !xhci->event_ring->dequeue) { xhci->error_bitmask |= 1 << 1; return; @@ -1026,18 +1125,25 @@ void xhci_handle_event(struct xhci_hcd *xhci) xhci->error_bitmask |= 1 << 2; return; } + xhci_dbg(xhci, "%s - OS owns TRB\n", __func__); /* FIXME: Handle more event types. */ switch ((event->event_cmd.flags & TRB_TYPE_BITMASK)) { case TRB_TYPE(TRB_COMPLETION): + xhci_dbg(xhci, "%s - calling handle_cmd_completion\n", __func__); handle_cmd_completion(xhci, &event->event_cmd); + xhci_dbg(xhci, "%s - returned from handle_cmd_completion\n", __func__); break; case TRB_TYPE(TRB_PORT_STATUS): + xhci_dbg(xhci, "%s - calling handle_port_status\n", __func__); handle_port_status(xhci, event); + xhci_dbg(xhci, "%s - returned from handle_port_status\n", __func__); update_ptrs = 0; break; case TRB_TYPE(TRB_TRANSFER): + xhci_dbg(xhci, "%s - calling handle_tx_event\n", __func__); ret = handle_tx_event(xhci, &event->trans_event); + xhci_dbg(xhci, "%s - returned from handle_tx_event\n", __func__); if (ret < 0) xhci->error_bitmask |= 1 << 9; else @@ -1093,13 +1199,13 @@ static int prepare_ring(struct xhci_hcd *xhci, struct xhci_ring *ep_ring, */ xhci_warn(xhci, "WARN urb submitted to disabled ep\n"); return -ENOENT; - case EP_STATE_HALTED: case EP_STATE_ERROR: - xhci_warn(xhci, "WARN waiting for halt or error on ep " - "to be cleared\n"); + xhci_warn(xhci, "WARN waiting for error on ep to be cleared\n"); /* FIXME event handling code for error needs to clear it */ /* XXX not sure if this should be -ENOENT or not */ return -EINVAL; + case EP_STATE_HALTED: + xhci_dbg(xhci, "WARN halted endpoint, queueing URB anyway.\n"); case EP_STATE_STOPPED: case EP_STATE_RUNNING: break; @@ -1128,9 +1234,9 @@ static int prepare_transfer(struct xhci_hcd *xhci, gfp_t mem_flags) { int ret; - + struct xhci_ep_ctx *ep_ctx = xhci_get_ep_ctx(xhci, xdev->out_ctx, ep_index); ret = prepare_ring(xhci, xdev->ep_rings[ep_index], - xdev->out_ctx->ep[ep_index].ep_info & EP_STATE_MASK, + ep_ctx->ep_info & EP_STATE_MASK, num_trbs, mem_flags); if (ret) return ret; @@ -1285,6 +1391,7 @@ static int queue_bulk_sg_tx(struct xhci_hcd *xhci, gfp_t mem_flags, /* Queue the first TRB, even if it's zero-length */ do { u32 field = 0; + u32 length_field = 0; /* Don't change the cycle bit of the first TRB until later */ if (first_trb) @@ -1314,10 +1421,13 @@ static int queue_bulk_sg_tx(struct xhci_hcd *xhci, gfp_t mem_flags, (unsigned int) (addr + TRB_MAX_BUFF_SIZE) & ~(TRB_MAX_BUFF_SIZE - 1), (unsigned int) addr + trb_buff_len); } + length_field = TRB_LEN(trb_buff_len) | + TD_REMAINDER(urb->transfer_buffer_length - running_total) | + TRB_INTR_TARGET(0); queue_trb(xhci, ep_ring, false, - (u32) addr, - (u32) ((u64) addr >> 32), - TRB_LEN(trb_buff_len) | TRB_INTR_TARGET(0), + lower_32_bits(addr), + upper_32_bits(addr), + length_field, /* We always want to know if the TRB was short, * or we won't get an event when it completes. * (Unless we use event data TRBs, which are a @@ -1365,7 +1475,7 @@ int xhci_queue_bulk_tx(struct xhci_hcd *xhci, gfp_t mem_flags, struct xhci_generic_trb *start_trb; bool first_trb; int start_cycle; - u32 field; + u32 field, length_field; int running_total, trb_buff_len, ret; u64 addr; @@ -1443,10 +1553,13 @@ int xhci_queue_bulk_tx(struct xhci_hcd *xhci, gfp_t mem_flags, td->last_trb = ep_ring->enqueue; field |= TRB_IOC; } + length_field = TRB_LEN(trb_buff_len) | + TD_REMAINDER(urb->transfer_buffer_length - running_total) | + TRB_INTR_TARGET(0); queue_trb(xhci, ep_ring, false, - (u32) addr, - (u32) ((u64) addr >> 32), - TRB_LEN(trb_buff_len) | TRB_INTR_TARGET(0), + lower_32_bits(addr), + upper_32_bits(addr), + length_field, /* We always want to know if the TRB was short, * or we won't get an event when it completes. * (Unless we use event data TRBs, which are a @@ -1478,7 +1591,7 @@ int xhci_queue_ctrl_tx(struct xhci_hcd *xhci, gfp_t mem_flags, struct usb_ctrlrequest *setup; struct xhci_generic_trb *start_trb; int start_cycle; - u32 field; + u32 field, length_field; struct xhci_td *td; ep_ring = xhci->devs[slot_id]->ep_rings[ep_index]; @@ -1528,13 +1641,16 @@ int xhci_queue_ctrl_tx(struct xhci_hcd *xhci, gfp_t mem_flags, /* If there's data, queue data TRBs */ field = 0; + length_field = TRB_LEN(urb->transfer_buffer_length) | + TD_REMAINDER(urb->transfer_buffer_length) | + TRB_INTR_TARGET(0); if (urb->transfer_buffer_length > 0) { if (setup->bRequestType & USB_DIR_IN) field |= TRB_DIR_IN; queue_trb(xhci, ep_ring, false, lower_32_bits(urb->transfer_dma), upper_32_bits(urb->transfer_dma), - TRB_LEN(urb->transfer_buffer_length) | TRB_INTR_TARGET(0), + length_field, /* Event on short tx */ field | TRB_ISP | TRB_TYPE(TRB_DATA) | ep_ring->cycle_state); } @@ -1603,7 +1719,8 @@ int xhci_queue_slot_control(struct xhci_hcd *xhci, u32 trb_type, u32 slot_id) int xhci_queue_address_device(struct xhci_hcd *xhci, dma_addr_t in_ctx_ptr, u32 slot_id) { - return queue_command(xhci, in_ctx_ptr, 0, 0, + return queue_command(xhci, lower_32_bits(in_ctx_ptr), + upper_32_bits(in_ctx_ptr), 0, TRB_TYPE(TRB_ADDR_DEV) | SLOT_ID_FOR_TRB(slot_id)); } @@ -1611,7 +1728,8 @@ int xhci_queue_address_device(struct xhci_hcd *xhci, dma_addr_t in_ctx_ptr, int xhci_queue_configure_endpoint(struct xhci_hcd *xhci, dma_addr_t in_ctx_ptr, u32 slot_id) { - return queue_command(xhci, in_ctx_ptr, 0, 0, + return queue_command(xhci, lower_32_bits(in_ctx_ptr), + upper_32_bits(in_ctx_ptr), 0, TRB_TYPE(TRB_CONFIG_EP) | SLOT_ID_FOR_TRB(slot_id)); } @@ -1639,10 +1757,23 @@ static int queue_set_tr_deq(struct xhci_hcd *xhci, int slot_id, u32 type = TRB_TYPE(TRB_SET_DEQ); addr = xhci_trb_virt_to_dma(deq_seg, deq_ptr); - if (addr == 0) + if (addr == 0) { xhci_warn(xhci, "WARN Cannot submit Set TR Deq Ptr\n"); xhci_warn(xhci, "WARN deq seg = %p, deq pt = %p\n", deq_seg, deq_ptr); - return queue_command(xhci, (u32) addr | cycle_state, 0, 0, + return 0; + } + return queue_command(xhci, lower_32_bits(addr) | cycle_state, + upper_32_bits(addr), 0, trb_slot_id | trb_ep_index | type); } + +int xhci_queue_reset_ep(struct xhci_hcd *xhci, int slot_id, + unsigned int ep_index) +{ + u32 trb_slot_id = SLOT_ID_FOR_TRB(slot_id); + u32 trb_ep_index = EP_ID_FOR_TRB(ep_index); + u32 type = TRB_TYPE(TRB_RESET_EP); + + return queue_command(xhci, 0, 0, 0, trb_slot_id | trb_ep_index | type); +} diff --git a/drivers/usb/host/xhci.h b/drivers/usb/host/xhci.h index 8936eeb5588..d31d32206ba 100644 --- a/drivers/usb/host/xhci.h +++ b/drivers/usb/host/xhci.h @@ -25,6 +25,7 @@ #include <linux/usb.h> #include <linux/timer.h> +#include <linux/kernel.h> #include "../core/hcd.h" /* Code sharing between pci-quirks and xhci hcd */ @@ -42,14 +43,6 @@ * xHCI register interface. * This corresponds to the eXtensible Host Controller Interface (xHCI) * Revision 0.95 specification - * - * Registers should always be accessed with double word or quad word accesses. - * - * Some xHCI implementations may support 64-bit address pointers. Registers - * with 64-bit address pointers should be written to with dword accesses by - * writing the low dword first (ptr[0]), then the high dword (ptr[1]) second. - * xHCI implementations that do not support 64-bit address pointers will ignore - * the high dword, and write order is irrelevant. */ /** @@ -96,6 +89,7 @@ struct xhci_cap_regs { #define HCS_ERST_MAX(p) (((p) >> 4) & 0xf) /* bit 26 Scratchpad restore - for save/restore HW state - not used yet */ /* bits 27:31 number of Scratchpad buffers SW must allocate for the HW */ +#define HCS_MAX_SCRATCHPAD(p) (((p) >> 27) & 0x1f) /* HCSPARAMS3 - hcs_params3 - bitmasks */ /* bits 0:7, Max U1 to U0 latency for the roothub ports */ @@ -166,10 +160,10 @@ struct xhci_op_regs { u32 reserved1; u32 reserved2; u32 dev_notification; - u32 cmd_ring[2]; + u64 cmd_ring; /* rsvd: offset 0x20-2F */ u32 reserved3[4]; - u32 dcbaa_ptr[2]; + u64 dcbaa_ptr; u32 config_reg; /* rsvd: offset 0x3C-3FF */ u32 reserved4[241]; @@ -254,7 +248,7 @@ struct xhci_op_regs { #define CMD_RING_RUNNING (1 << 3) /* bits 4:5 reserved and should be preserved */ /* Command Ring pointer - bit mask for the lower 32 bits. */ -#define CMD_RING_ADDR_MASK (0xffffffc0) +#define CMD_RING_RSVD_BITS (0x3f) /* CONFIG - Configure Register - config_reg bitmasks */ /* bits 0:7 - maximum number of device slots enabled (NumSlotsEn) */ @@ -382,8 +376,8 @@ struct xhci_intr_reg { u32 irq_control; u32 erst_size; u32 rsvd; - u32 erst_base[2]; - u32 erst_dequeue[2]; + u64 erst_base; + u64 erst_dequeue; }; /* irq_pending bitmasks */ @@ -453,6 +447,27 @@ struct xhci_doorbell_array { /** + * struct xhci_container_ctx + * @type: Type of context. Used to calculated offsets to contained contexts. + * @size: Size of the context data + * @bytes: The raw context data given to HW + * @dma: dma address of the bytes + * + * Represents either a Device or Input context. Holds a pointer to the raw + * memory used for the context (bytes) and dma address of it (dma). + */ +struct xhci_container_ctx { + unsigned type; +#define XHCI_CTX_TYPE_DEVICE 0x1 +#define XHCI_CTX_TYPE_INPUT 0x2 + + int size; + + u8 *bytes; + dma_addr_t dma; +}; + +/** * struct xhci_slot_ctx * @dev_info: Route string, device speed, hub info, and last valid endpoint * @dev_info2: Max exit latency for device number, root hub port number @@ -538,7 +553,7 @@ struct xhci_slot_ctx { struct xhci_ep_ctx { u32 ep_info; u32 ep_info2; - u32 deq[2]; + u64 deq; u32 tx_info; /* offset 0x14 - 0x1f reserved for HC internal use */ u32 reserved[3]; @@ -589,18 +604,16 @@ struct xhci_ep_ctx { /** - * struct xhci_device_control - * Input/Output context; see section 6.2.5. + * struct xhci_input_control_context + * Input control context; see section 6.2.5. * * @drop_context: set the bit of the endpoint context you want to disable * @add_context: set the bit of the endpoint context you want to enable */ -struct xhci_device_control { +struct xhci_input_control_ctx { u32 drop_flags; u32 add_flags; - u32 rsvd[6]; - struct xhci_slot_ctx slot; - struct xhci_ep_ctx ep[31]; + u32 rsvd2[6]; }; /* drop context bitmasks */ @@ -608,7 +621,6 @@ struct xhci_device_control { /* add context bitmasks */ #define ADD_EP(x) (0x1 << x) - struct xhci_virt_device { /* * Commands to the hardware are passed an "input context" that @@ -618,11 +630,10 @@ struct xhci_virt_device { * track of input and output contexts separately because * these commands might fail and we don't trust the hardware. */ - struct xhci_device_control *out_ctx; - dma_addr_t out_ctx_dma; + struct xhci_container_ctx *out_ctx; /* Used for addressing devices and configuration changes */ - struct xhci_device_control *in_ctx; - dma_addr_t in_ctx_dma; + struct xhci_container_ctx *in_ctx; + /* FIXME when stream support is added */ struct xhci_ring *ep_rings[31]; /* Temporary storage in case the configure endpoint command fails and we @@ -641,7 +652,7 @@ struct xhci_virt_device { */ struct xhci_device_context_array { /* 64-bit device addresses; we only write 32-bit addresses */ - u32 dev_context_ptrs[2*MAX_HC_SLOTS]; + u64 dev_context_ptrs[MAX_HC_SLOTS]; /* private xHCD pointers */ dma_addr_t dma; }; @@ -654,7 +665,7 @@ struct xhci_device_context_array { struct xhci_stream_ctx { /* 64-bit stream ring address, cycle state, and stream type */ - u32 stream_ring[2]; + u64 stream_ring; /* offset 0x14 - 0x1f reserved for HC internal use */ u32 reserved[2]; }; @@ -662,7 +673,7 @@ struct xhci_stream_ctx { struct xhci_transfer_event { /* 64-bit buffer address, or immediate data */ - u32 buffer[2]; + u64 buffer; u32 transfer_len; /* This field is interpreted differently based on the type of TRB */ u32 flags; @@ -744,7 +755,7 @@ struct xhci_transfer_event { struct xhci_link_trb { /* 64-bit segment pointer*/ - u32 segment_ptr[2]; + u64 segment_ptr; u32 intr_target; u32 control; }; @@ -755,7 +766,7 @@ struct xhci_link_trb { /* Command completion event TRB */ struct xhci_event_cmd { /* Pointer to command TRB, or the value passed by the event data trb */ - u32 cmd_trb[2]; + u64 cmd_trb; u32 status; u32 flags; }; @@ -848,8 +859,8 @@ union xhci_trb { #define TRB_CONFIG_EP 12 /* Evaluate Context Command */ #define TRB_EVAL_CONTEXT 13 -/* Reset Transfer Ring Command */ -#define TRB_RESET_RING 14 +/* Reset Endpoint Command */ +#define TRB_RESET_EP 14 /* Stop Transfer Ring Command */ #define TRB_STOP_RING 15 /* Set Transfer Ring Dequeue Pointer Command */ @@ -929,6 +940,7 @@ struct xhci_ring { unsigned int cancels_pending; unsigned int state; #define SET_DEQ_PENDING (1 << 0) +#define EP_HALTED (1 << 1) /* The TRB that was last reported in a stopped endpoint ring */ union xhci_trb *stopped_trb; struct xhci_td *stopped_td; @@ -940,9 +952,15 @@ struct xhci_ring { u32 cycle_state; }; +struct xhci_dequeue_state { + struct xhci_segment *new_deq_seg; + union xhci_trb *new_deq_ptr; + int new_cycle_state; +}; + struct xhci_erst_entry { /* 64-bit event ring segment address */ - u32 seg_addr[2]; + u64 seg_addr; u32 seg_size; /* Set to zero */ u32 rsvd; @@ -957,6 +975,13 @@ struct xhci_erst { unsigned int erst_size; }; +struct xhci_scratchpad { + u64 *sp_array; + dma_addr_t sp_dma; + void **sp_buffers; + dma_addr_t *sp_dma_buffers; +}; + /* * Each segment table entry is 4*32bits long. 1K seems like an ok size: * (1K bytes * 8bytes/bit) / (4*32 bits) = 64 segment entries in the table, @@ -1011,6 +1036,9 @@ struct xhci_hcd { struct xhci_ring *cmd_ring; struct xhci_ring *event_ring; struct xhci_erst erst; + /* Scratchpad */ + struct xhci_scratchpad *scratchpad; + /* slot enabling and address device helpers */ struct completion addr_dev; int slot_id; @@ -1071,13 +1099,43 @@ static inline unsigned int xhci_readl(const struct xhci_hcd *xhci, static inline void xhci_writel(struct xhci_hcd *xhci, const unsigned int val, __u32 __iomem *regs) { - if (!in_interrupt()) - xhci_dbg(xhci, - "`MEM_WRITE_DWORD(3'b000, 32'h%p, 32'h%0x, 4'hf);\n", - regs, val); + xhci_dbg(xhci, + "`MEM_WRITE_DWORD(3'b000, 32'h%p, 32'h%0x, 4'hf);\n", + regs, val); writel(val, regs); } +/* + * Registers should always be accessed with double word or quad word accesses. + * + * Some xHCI implementations may support 64-bit address pointers. Registers + * with 64-bit address pointers should be written to with dword accesses by + * writing the low dword first (ptr[0]), then the high dword (ptr[1]) second. + * xHCI implementations that do not support 64-bit address pointers will ignore + * the high dword, and write order is irrelevant. + */ +static inline u64 xhci_read_64(const struct xhci_hcd *xhci, + __u64 __iomem *regs) +{ + __u32 __iomem *ptr = (__u32 __iomem *) regs; + u64 val_lo = readl(ptr); + u64 val_hi = readl(ptr + 1); + return val_lo + (val_hi << 32); +} +static inline void xhci_write_64(struct xhci_hcd *xhci, + const u64 val, __u64 __iomem *regs) +{ + __u32 __iomem *ptr = (__u32 __iomem *) regs; + u32 val_lo = lower_32_bits(val); + u32 val_hi = upper_32_bits(val); + + xhci_dbg(xhci, + "`MEM_WRITE_DWORD(3'b000, 64'h%p, 64'h%0lx, 4'hf);\n", + regs, (long unsigned int) val); + writel(val_lo, ptr); + writel(val_hi, ptr + 1); +} + /* xHCI debugging */ void xhci_print_ir_set(struct xhci_hcd *xhci, struct xhci_intr_reg *ir_set, int set_num); void xhci_print_registers(struct xhci_hcd *xhci); @@ -1090,7 +1148,7 @@ void xhci_debug_ring(struct xhci_hcd *xhci, struct xhci_ring *ring); void xhci_dbg_erst(struct xhci_hcd *xhci, struct xhci_erst *erst); void xhci_dbg_cmd_ptrs(struct xhci_hcd *xhci); void xhci_dbg_ring_ptrs(struct xhci_hcd *xhci, struct xhci_ring *ring); -void xhci_dbg_ctx(struct xhci_hcd *xhci, struct xhci_device_control *ctx, dma_addr_t dma, unsigned int last_ep); +void xhci_dbg_ctx(struct xhci_hcd *xhci, struct xhci_container_ctx *ctx, unsigned int last_ep); /* xHCI memory managment */ void xhci_mem_cleanup(struct xhci_hcd *xhci); @@ -1128,6 +1186,7 @@ int xhci_urb_enqueue(struct usb_hcd *hcd, struct urb *urb, gfp_t mem_flags); int xhci_urb_dequeue(struct usb_hcd *hcd, struct urb *urb, int status); int xhci_add_endpoint(struct usb_hcd *hcd, struct usb_device *udev, struct usb_host_endpoint *ep); int xhci_drop_endpoint(struct usb_hcd *hcd, struct usb_device *udev, struct usb_host_endpoint *ep); +void xhci_endpoint_reset(struct usb_hcd *hcd, struct usb_host_endpoint *ep); int xhci_check_bandwidth(struct usb_hcd *hcd, struct usb_device *udev); void xhci_reset_bandwidth(struct usb_hcd *hcd, struct usb_device *udev); @@ -1148,10 +1207,23 @@ int xhci_queue_bulk_tx(struct xhci_hcd *xhci, gfp_t mem_flags, struct urb *urb, int slot_id, unsigned int ep_index); int xhci_queue_configure_endpoint(struct xhci_hcd *xhci, dma_addr_t in_ctx_ptr, u32 slot_id); +int xhci_queue_reset_ep(struct xhci_hcd *xhci, int slot_id, + unsigned int ep_index); +void xhci_find_new_dequeue_state(struct xhci_hcd *xhci, + unsigned int slot_id, unsigned int ep_index, + struct xhci_td *cur_td, struct xhci_dequeue_state *state); +void xhci_queue_new_dequeue_state(struct xhci_hcd *xhci, + struct xhci_ring *ep_ring, unsigned int slot_id, + unsigned int ep_index, struct xhci_dequeue_state *deq_state); /* xHCI roothub code */ int xhci_hub_control(struct usb_hcd *hcd, u16 typeReq, u16 wValue, u16 wIndex, char *buf, u16 wLength); int xhci_hub_status_data(struct usb_hcd *hcd, char *buf); +/* xHCI contexts */ +struct xhci_input_control_ctx *xhci_get_input_control_ctx(struct xhci_hcd *xhci, struct xhci_container_ctx *ctx); +struct xhci_slot_ctx *xhci_get_slot_ctx(struct xhci_hcd *xhci, struct xhci_container_ctx *ctx); +struct xhci_ep_ctx *xhci_get_ep_ctx(struct xhci_hcd *xhci, struct xhci_container_ctx *ctx, unsigned int ep_index); + #endif /* __LINUX_XHCI_HCD_H */ diff --git a/drivers/usb/misc/Kconfig b/drivers/usb/misc/Kconfig index a68d91a11be..abe3aa67ed0 100644 --- a/drivers/usb/misc/Kconfig +++ b/drivers/usb/misc/Kconfig @@ -220,7 +220,7 @@ config USB_IOWARRIOR config USB_TEST tristate "USB testing driver" - depends on USB && USB_DEVICEFS + depends on USB help This driver is for testing host controller software. It is used with specialized device firmware for regression and stress testing, diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c index 554a414f65d..c7c1ca0494c 100644 --- a/drivers/usb/musb/musb_core.c +++ b/drivers/usb/musb/musb_core.c @@ -1326,7 +1326,6 @@ static int __init musb_core_init(u16 musb_type, struct musb *musb) int i; /* log core options (read using indexed model) */ - musb_ep_select(mbase, 0); reg = musb_read_configdata(mbase); strcpy(aInfo, (reg & MUSB_CONFIGDATA_UTMIDW) ? "UTMI-16" : "UTMI-8"); @@ -1990,7 +1989,7 @@ bad_config: if (status < 0) goto fail2; -#ifdef CONFIG_USB_OTG +#ifdef CONFIG_USB_MUSB_OTG setup_timer(&musb->otg_timer, musb_otg_timer_func, (unsigned long) musb); #endif diff --git a/drivers/usb/musb/musb_gadget_ep0.c b/drivers/usb/musb/musb_gadget_ep0.c index 40ed50ecedf..7a6778675ad 100644 --- a/drivers/usb/musb/musb_gadget_ep0.c +++ b/drivers/usb/musb/musb_gadget_ep0.c @@ -407,7 +407,7 @@ stall: csr |= MUSB_RXCSR_P_SENDSTALL | MUSB_RXCSR_FLUSHFIFO | MUSB_RXCSR_CLRDATATOG - | MUSB_TXCSR_P_WZC_BITS; + | MUSB_RXCSR_P_WZC_BITS; musb_writew(regs, MUSB_RXCSR, csr); } diff --git a/drivers/usb/musb/musb_regs.h b/drivers/usb/musb/musb_regs.h index de3b2f18db4..fbfd3fd9ce1 100644 --- a/drivers/usb/musb/musb_regs.h +++ b/drivers/usb/musb/musb_regs.h @@ -323,6 +323,7 @@ static inline void musb_write_rxfifoadd(void __iomem *mbase, u16 c_off) static inline u8 musb_read_configdata(void __iomem *mbase) { + musb_writeb(mbase, MUSB_INDEX, 0); return musb_readb(mbase, 0x10 + MUSB_CONFIGDATA); } diff --git a/drivers/usb/serial/cp210x.c b/drivers/usb/serial/cp210x.c index e9a40b820fd..985cbcf48bd 100644 --- a/drivers/usb/serial/cp210x.c +++ b/drivers/usb/serial/cp210x.c @@ -80,6 +80,7 @@ static struct usb_device_id id_table [] = { { USB_DEVICE(0x10C4, 0x80F6) }, /* Suunto sports instrument */ { USB_DEVICE(0x10C4, 0x8115) }, /* Arygon NFC/Mifare Reader */ { USB_DEVICE(0x10C4, 0x813D) }, /* Burnside Telecom Deskmobile */ + { USB_DEVICE(0x10C4, 0x813F) }, /* Tams Master Easy Control */ { USB_DEVICE(0x10C4, 0x814A) }, /* West Mountain Radio RIGblaster P&P */ { USB_DEVICE(0x10C4, 0x814B) }, /* West Mountain Radio RIGtalk */ { USB_DEVICE(0x10C4, 0x815E) }, /* Helicomm IP-Link 1220-DVM */ @@ -96,7 +97,9 @@ static struct usb_device_id id_table [] = { { USB_DEVICE(0x10c4, 0x8293) }, /* Telegesys ETRX2USB */ { USB_DEVICE(0x10C4, 0x82F9) }, /* Procyon AVS */ { USB_DEVICE(0x10C4, 0x8341) }, /* Siemens MC35PU GPRS Modem */ + { USB_DEVICE(0x10C4, 0x8382) }, /* Cygnal Integrated Products, Inc. */ { USB_DEVICE(0x10C4, 0x83A8) }, /* Amber Wireless AMB2560 */ + { USB_DEVICE(0x10C4, 0x8411) }, /* Kyocera GPS Module */ { USB_DEVICE(0x10C4, 0x846E) }, /* BEI USB Sensor Interface (VCP) */ { USB_DEVICE(0x10C4, 0xEA60) }, /* Silicon Labs factory default */ { USB_DEVICE(0x10C4, 0xEA61) }, /* Silicon Labs factory default */ diff --git a/drivers/usb/serial/ftdi_sio.c b/drivers/usb/serial/ftdi_sio.c index 60c64cc5be2..b574878c78b 100644 --- a/drivers/usb/serial/ftdi_sio.c +++ b/drivers/usb/serial/ftdi_sio.c @@ -698,6 +698,7 @@ static struct usb_device_id id_table_combined [] = { { USB_DEVICE(MARVELL_VID, MARVELL_SHEEVAPLUG_PID), .driver_info = (kernel_ulong_t)&ftdi_jtag_quirk }, { USB_DEVICE(LARSENBRUSGAARD_VID, LB_ALTITRACK_PID) }, + { USB_DEVICE(GN_OTOMETRICS_VID, AURICAL_USB_PID) }, { }, /* Optional parameter entry */ { } /* Terminating entry */ }; diff --git a/drivers/usb/serial/ftdi_sio.h b/drivers/usb/serial/ftdi_sio.h index c9fbd741509..24dbd99e87d 100644 --- a/drivers/usb/serial/ftdi_sio.h +++ b/drivers/usb/serial/ftdi_sio.h @@ -947,6 +947,13 @@ #define FTDI_TURTELIZER_PID 0xBDC8 /* JTAG/RS-232 adapter by egnite GmBH */ /* + * GN Otometrics (http://www.otometrics.com) + * Submitted by Ville Sundberg. + */ +#define GN_OTOMETRICS_VID 0x0c33 /* Vendor ID */ +#define AURICAL_USB_PID 0x0010 /* Aurical USB Audiometer */ + +/* * BmRequestType: 1100 0000b * bRequest: FTDI_E2_READ * wValue: 0 diff --git a/drivers/usb/serial/mos7840.c b/drivers/usb/serial/mos7840.c index c31940a307f..270009afdf7 100644 --- a/drivers/usb/serial/mos7840.c +++ b/drivers/usb/serial/mos7840.c @@ -124,10 +124,13 @@ #define BANDB_DEVICE_ID_USOPTL4_4 0xAC44 #define BANDB_DEVICE_ID_USOPTL4_2 0xAC42 -/* This driver also supports the ATEN UC2324 device since it is mos7840 based - * - if I knew the device id it would also support the ATEN UC2322 */ +/* This driver also supports + * ATEN UC2324 device using Moschip MCS7840 + * ATEN UC2322 device using Moschip MCS7820 + */ #define USB_VENDOR_ID_ATENINTL 0x0557 #define ATENINTL_DEVICE_ID_UC2324 0x2011 +#define ATENINTL_DEVICE_ID_UC2322 0x7820 /* Interrupt Routine Defines */ @@ -177,6 +180,7 @@ static struct usb_device_id moschip_port_id_table[] = { {USB_DEVICE(USB_VENDOR_ID_BANDB, BANDB_DEVICE_ID_USOPTL4_4)}, {USB_DEVICE(USB_VENDOR_ID_BANDB, BANDB_DEVICE_ID_USOPTL4_2)}, {USB_DEVICE(USB_VENDOR_ID_ATENINTL, ATENINTL_DEVICE_ID_UC2324)}, + {USB_DEVICE(USB_VENDOR_ID_ATENINTL, ATENINTL_DEVICE_ID_UC2322)}, {} /* terminating entry */ }; @@ -186,6 +190,7 @@ static __devinitdata struct usb_device_id moschip_id_table_combined[] = { {USB_DEVICE(USB_VENDOR_ID_BANDB, BANDB_DEVICE_ID_USOPTL4_4)}, {USB_DEVICE(USB_VENDOR_ID_BANDB, BANDB_DEVICE_ID_USOPTL4_2)}, {USB_DEVICE(USB_VENDOR_ID_ATENINTL, ATENINTL_DEVICE_ID_UC2324)}, + {USB_DEVICE(USB_VENDOR_ID_ATENINTL, ATENINTL_DEVICE_ID_UC2322)}, {} /* terminating entry */ }; diff --git a/drivers/usb/serial/option.c b/drivers/usb/serial/option.c index 98262dd552b..c784ddbe7b6 100644 --- a/drivers/usb/serial/option.c +++ b/drivers/usb/serial/option.c @@ -66,8 +66,10 @@ static int option_tiocmget(struct tty_struct *tty, struct file *file); static int option_tiocmset(struct tty_struct *tty, struct file *file, unsigned int set, unsigned int clear); static int option_send_setup(struct usb_serial_port *port); +#ifdef CONFIG_PM static int option_suspend(struct usb_serial *serial, pm_message_t message); static int option_resume(struct usb_serial *serial); +#endif /* Vendor and product IDs */ #define OPTION_VENDOR_ID 0x0AF0 @@ -205,6 +207,7 @@ static int option_resume(struct usb_serial *serial); #define NOVATELWIRELESS_PRODUCT_MC727 0x4100 #define NOVATELWIRELESS_PRODUCT_MC950D 0x4400 #define NOVATELWIRELESS_PRODUCT_U727 0x5010 +#define NOVATELWIRELESS_PRODUCT_MC727_NEW 0x5100 #define NOVATELWIRELESS_PRODUCT_MC760 0x6000 #define NOVATELWIRELESS_PRODUCT_OVMC760 0x6002 @@ -259,11 +262,6 @@ static int option_resume(struct usb_serial *serial); #define AXESSTEL_VENDOR_ID 0x1726 #define AXESSTEL_PRODUCT_MV110H 0x1000 -#define ONDA_VENDOR_ID 0x19d2 -#define ONDA_PRODUCT_MSA501HS 0x0001 -#define ONDA_PRODUCT_ET502HS 0x0002 -#define ONDA_PRODUCT_MT503HS 0x2000 - #define BANDRICH_VENDOR_ID 0x1A8D #define BANDRICH_PRODUCT_C100_1 0x1002 #define BANDRICH_PRODUCT_C100_2 0x1003 @@ -301,6 +299,7 @@ static int option_resume(struct usb_serial *serial); #define ZTE_PRODUCT_MF628 0x0015 #define ZTE_PRODUCT_MF626 0x0031 #define ZTE_PRODUCT_CDMA_TECH 0xfffe +#define ZTE_PRODUCT_AC8710 0xfff1 #define BENQ_VENDOR_ID 0x04a5 #define BENQ_PRODUCT_H10 0x4068 @@ -322,6 +321,11 @@ static int option_resume(struct usb_serial *serial); #define ALINK_VENDOR_ID 0x1e0e #define ALINK_PRODUCT_3GU 0x9200 +/* ALCATEL PRODUCTS */ +#define ALCATEL_VENDOR_ID 0x1bbb +#define ALCATEL_PRODUCT_X060S 0x0000 + + static struct usb_device_id option_ids[] = { { USB_DEVICE(OPTION_VENDOR_ID, OPTION_PRODUCT_COLT) }, { USB_DEVICE(OPTION_VENDOR_ID, OPTION_PRODUCT_RICOLA) }, @@ -438,6 +442,7 @@ static struct usb_device_id option_ids[] = { { USB_DEVICE(NOVATELWIRELESS_VENDOR_ID, NOVATELWIRELESS_PRODUCT_EU870D) }, /* Novatel EU850D/EU860D/EU870D */ { USB_DEVICE(NOVATELWIRELESS_VENDOR_ID, NOVATELWIRELESS_PRODUCT_MC950D) }, /* Novatel MC930D/MC950D */ { USB_DEVICE(NOVATELWIRELESS_VENDOR_ID, NOVATELWIRELESS_PRODUCT_MC727) }, /* Novatel MC727/U727/USB727 */ + { USB_DEVICE(NOVATELWIRELESS_VENDOR_ID, NOVATELWIRELESS_PRODUCT_MC727_NEW) }, /* Novatel MC727/U727/USB727 refresh */ { USB_DEVICE(NOVATELWIRELESS_VENDOR_ID, NOVATELWIRELESS_PRODUCT_U727) }, /* Novatel MC727/U727/USB727 */ { USB_DEVICE(NOVATELWIRELESS_VENDOR_ID, NOVATELWIRELESS_PRODUCT_MC760) }, /* Novatel MC760/U760/USB760 */ { USB_DEVICE(NOVATELWIRELESS_VENDOR_ID, NOVATELWIRELESS_PRODUCT_OVMC760) }, /* Novatel Ovation MC760 */ @@ -474,42 +479,6 @@ static struct usb_device_id option_ids[] = { { USB_DEVICE(ANYDATA_VENDOR_ID, ANYDATA_PRODUCT_ADU_500A) }, { USB_DEVICE(ANYDATA_VENDOR_ID, ANYDATA_PRODUCT_ADU_620UW) }, { USB_DEVICE(AXESSTEL_VENDOR_ID, AXESSTEL_PRODUCT_MV110H) }, - { USB_DEVICE(ONDA_VENDOR_ID, ONDA_PRODUCT_MSA501HS) }, - { USB_DEVICE(ONDA_VENDOR_ID, ONDA_PRODUCT_ET502HS) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0003) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0004) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0005) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0006) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0007) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0008) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0009) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x000a) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x000b) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x000c) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x000d) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x000e) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x000f) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0010) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0011) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0012) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0013) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0014) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0015) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0016) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0017) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0018) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0019) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0020) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0021) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0022) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0023) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0024) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0025) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0026) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0027) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0028) }, - { USB_DEVICE(ONDA_VENDOR_ID, 0x0029) }, - { USB_DEVICE(ONDA_VENDOR_ID, ONDA_PRODUCT_MT503HS) }, { USB_DEVICE(YISO_VENDOR_ID, YISO_PRODUCT_U893) }, { USB_DEVICE(BANDRICH_VENDOR_ID, BANDRICH_PRODUCT_C100_1) }, { USB_DEVICE(BANDRICH_VENDOR_ID, BANDRICH_PRODUCT_C100_2) }, @@ -534,10 +503,75 @@ static struct usb_device_id option_ids[] = { { USB_DEVICE(QUALCOMM_VENDOR_ID, 0x6613)}, /* Onda H600/ZTE MF330 */ { USB_DEVICE(MAXON_VENDOR_ID, 0x6280) }, /* BP3-USB & BP3-EXT HSDPA */ { USB_DEVICE(TELIT_VENDOR_ID, TELIT_PRODUCT_UC864E) }, - { USB_DEVICE(ZTE_VENDOR_ID, ZTE_PRODUCT_MF622) }, - { USB_DEVICE(ZTE_VENDOR_ID, ZTE_PRODUCT_MF626) }, - { USB_DEVICE(ZTE_VENDOR_ID, ZTE_PRODUCT_MF628) }, - { USB_DEVICE(ZTE_VENDOR_ID, ZTE_PRODUCT_CDMA_TECH) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, ZTE_PRODUCT_MF622, 0xff, 0xff, 0xff) }, /* ZTE WCDMA products */ + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0002, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0003, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0004, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0005, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0006, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0007, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0008, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0009, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x000a, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x000b, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x000c, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x000d, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x000e, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x000f, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0010, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0011, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0012, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0013, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, ZTE_PRODUCT_MF628, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0016, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0017, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0018, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0019, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0020, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0021, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0022, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0023, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0024, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0025, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0026, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0028, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0029, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0030, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, ZTE_PRODUCT_MF626, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0032, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0033, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0037, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0039, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0042, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0043, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0048, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0049, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0051, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0052, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0054, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0055, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0057, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0058, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0061, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0062, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0063, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0064, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0066, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0069, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0076, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0078, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0082, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0086, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x2002, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x2003, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0014, 0xff, 0xff, 0xff) }, /* ZTE CDMA products */ + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0027, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0059, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0060, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0070, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0073, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, ZTE_PRODUCT_CDMA_TECH, 0xff, 0xff, 0xff) }, + { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, ZTE_PRODUCT_AC8710, 0xff, 0xff, 0xff) }, { USB_DEVICE(BENQ_VENDOR_ID, BENQ_PRODUCT_H10) }, { USB_DEVICE(DLINK_VENDOR_ID, DLINK_PRODUCT_DWM_652) }, { USB_DEVICE(QISDA_VENDOR_ID, QISDA_PRODUCT_H21_4512) }, @@ -547,6 +581,7 @@ static struct usb_device_id option_ids[] = { { USB_DEVICE(TOSHIBA_VENDOR_ID, TOSHIBA_PRODUCT_HSDPA_MINICARD ) }, /* Toshiba 3G HSDPA == Novatel Expedite EU870D MiniCard */ { USB_DEVICE(ALINK_VENDOR_ID, 0x9000) }, { USB_DEVICE_AND_INTERFACE_INFO(ALINK_VENDOR_ID, ALINK_PRODUCT_3GU, 0xff, 0xff, 0xff) }, + { USB_DEVICE(ALCATEL_VENDOR_ID, ALCATEL_PRODUCT_X060S) }, { } /* Terminating entry */ }; MODULE_DEVICE_TABLE(usb, option_ids); @@ -555,8 +590,10 @@ static struct usb_driver option_driver = { .name = "option", .probe = usb_serial_probe, .disconnect = usb_serial_disconnect, +#ifdef CONFIG_PM .suspend = usb_serial_suspend, .resume = usb_serial_resume, +#endif .id_table = option_ids, .no_dynamic_id = 1, }; @@ -588,8 +625,10 @@ static struct usb_serial_driver option_1port_device = { .disconnect = option_disconnect, .release = option_release, .read_int_callback = option_instat_callback, +#ifdef CONFIG_PM .suspend = option_suspend, .resume = option_resume, +#endif }; static int debug; @@ -831,7 +870,6 @@ static void option_instat_callback(struct urb *urb) int status = urb->status; struct usb_serial_port *port = urb->context; struct option_port_private *portdata = usb_get_serial_port_data(port); - struct usb_serial *serial = port->serial; dbg("%s", __func__); dbg("%s: urb %p port %p has data %p", __func__, urb, port, portdata); @@ -927,7 +965,6 @@ static int option_open(struct tty_struct *tty, struct usb_serial_port *port, struct file *filp) { struct option_port_private *portdata; - struct usb_serial *serial = port->serial; int i, err; struct urb *urb; @@ -1187,6 +1224,7 @@ static void option_release(struct usb_serial *serial) } } +#ifdef CONFIG_PM static int option_suspend(struct usb_serial *serial, pm_message_t message) { dbg("%s entered", __func__); @@ -1245,6 +1283,7 @@ static int option_resume(struct usb_serial *serial) } return 0; } +#endif MODULE_AUTHOR(DRIVER_AUTHOR); MODULE_DESCRIPTION(DRIVER_DESC); diff --git a/drivers/usb/storage/transport.c b/drivers/usb/storage/transport.c index fcb32021721..e20dc525d17 100644 --- a/drivers/usb/storage/transport.c +++ b/drivers/usb/storage/transport.c @@ -961,7 +961,7 @@ int usb_stor_Bulk_max_lun(struct us_data *us) US_BULK_GET_MAX_LUN, USB_DIR_IN | USB_TYPE_CLASS | USB_RECIP_INTERFACE, - 0, us->ifnum, us->iobuf, 1, HZ); + 0, us->ifnum, us->iobuf, 1, 10*HZ); US_DEBUGP("GetMaxLUN command result is %d, data is %d\n", result, us->iobuf[0]); diff --git a/drivers/video/backlight/jornada720_bl.c b/drivers/video/backlight/jornada720_bl.c index c3ebb6b41ce..7aed2565c1b 100644 --- a/drivers/video/backlight/jornada720_bl.c +++ b/drivers/video/backlight/jornada720_bl.c @@ -72,7 +72,7 @@ static int jornada_bl_update_status(struct backlight_device *bd) if (jornada_ssp_byte(SETBRIGHTNESS) != TXDUMMY) { printk(KERN_INFO "bl : failed to set brightness\n"); ret = -ETIMEDOUT; - goto out + goto out; } /* at this point we expect that the mcu has accepted diff --git a/drivers/video/s3c-fb.c b/drivers/video/s3c-fb.c index bb63c07e13d..5a72083dc67 100644 --- a/drivers/video/s3c-fb.c +++ b/drivers/video/s3c-fb.c @@ -964,7 +964,7 @@ static int __devexit s3c_fb_remove(struct platform_device *pdev) struct s3c_fb *sfb = platform_get_drvdata(pdev); int win; - for (win = 0; win <= S3C_FB_MAX_WIN; win++) + for (win = 0; win < S3C_FB_MAX_WIN; win++) if (sfb->windows[win]) s3c_fb_release_win(sfb, sfb->windows[win]); @@ -988,7 +988,7 @@ static int s3c_fb_suspend(struct platform_device *pdev, pm_message_t state) struct s3c_fb_win *win; int win_no; - for (win_no = S3C_FB_MAX_WIN; win_no >= 0; win_no--) { + for (win_no = S3C_FB_MAX_WIN - 1; win_no >= 0; win_no--) { win = sfb->windows[win_no]; if (!win) continue; diff --git a/drivers/virtio/virtio_pci.c b/drivers/virtio/virtio_pci.c index bcec78ffc76..248e00ec4dc 100644 --- a/drivers/virtio/virtio_pci.c +++ b/drivers/virtio/virtio_pci.c @@ -52,8 +52,10 @@ struct virtio_pci_device char (*msix_names)[256]; /* Number of available vectors */ unsigned msix_vectors; - /* Vectors allocated */ + /* Vectors allocated, excluding per-vq vectors if any */ unsigned msix_used_vectors; + /* Whether we have vector per vq */ + bool per_vq_vectors; }; /* Constants for MSI-X */ @@ -258,7 +260,6 @@ static void vp_free_vectors(struct virtio_device *vdev) for (i = 0; i < vp_dev->msix_used_vectors; ++i) free_irq(vp_dev->msix_entries[i].vector, vp_dev); - vp_dev->msix_used_vectors = 0; if (vp_dev->msix_enabled) { /* Disable the vector used for configuration */ @@ -267,80 +268,77 @@ static void vp_free_vectors(struct virtio_device *vdev) /* Flush the write out to device */ ioread16(vp_dev->ioaddr + VIRTIO_MSI_CONFIG_VECTOR); - vp_dev->msix_enabled = 0; pci_disable_msix(vp_dev->pci_dev); + vp_dev->msix_enabled = 0; + vp_dev->msix_vectors = 0; } -} -static int vp_enable_msix(struct pci_dev *dev, struct msix_entry *entries, - int *options, int noptions) -{ - int i; - for (i = 0; i < noptions; ++i) - if (!pci_enable_msix(dev, entries, options[i])) - return options[i]; - return -EBUSY; + vp_dev->msix_used_vectors = 0; + kfree(vp_dev->msix_names); + vp_dev->msix_names = NULL; + kfree(vp_dev->msix_entries); + vp_dev->msix_entries = NULL; } -static int vp_request_vectors(struct virtio_device *vdev, unsigned max_vqs) +static int vp_request_vectors(struct virtio_device *vdev, int nvectors, + bool per_vq_vectors) { struct virtio_pci_device *vp_dev = to_vp_device(vdev); const char *name = dev_name(&vp_dev->vdev.dev); unsigned i, v; int err = -ENOMEM; - /* We want at most one vector per queue and one for config changes. - * Fallback to separate vectors for config and a shared for queues. - * Finally fall back to regular interrupts. */ - int options[] = { max_vqs + 1, 2 }; - int nvectors = max(options[0], options[1]); + + if (!nvectors) { + /* Can't allocate MSI-X vectors, use regular interrupt */ + vp_dev->msix_vectors = 0; + err = request_irq(vp_dev->pci_dev->irq, vp_interrupt, + IRQF_SHARED, name, vp_dev); + if (err) + return err; + vp_dev->intx_enabled = 1; + return 0; + } vp_dev->msix_entries = kmalloc(nvectors * sizeof *vp_dev->msix_entries, GFP_KERNEL); if (!vp_dev->msix_entries) - goto error_entries; + goto error; vp_dev->msix_names = kmalloc(nvectors * sizeof *vp_dev->msix_names, GFP_KERNEL); if (!vp_dev->msix_names) - goto error_names; + goto error; for (i = 0; i < nvectors; ++i) vp_dev->msix_entries[i].entry = i; - err = vp_enable_msix(vp_dev->pci_dev, vp_dev->msix_entries, - options, ARRAY_SIZE(options)); - if (err < 0) { - /* Can't allocate enough MSI-X vectors, use regular interrupt */ - vp_dev->msix_vectors = 0; - err = request_irq(vp_dev->pci_dev->irq, vp_interrupt, - IRQF_SHARED, name, vp_dev); - if (err) - goto error_irq; - vp_dev->intx_enabled = 1; - } else { - vp_dev->msix_vectors = err; - vp_dev->msix_enabled = 1; - - /* Set the vector used for configuration */ - v = vp_dev->msix_used_vectors; - snprintf(vp_dev->msix_names[v], sizeof *vp_dev->msix_names, - "%s-config", name); - err = request_irq(vp_dev->msix_entries[v].vector, - vp_config_changed, 0, vp_dev->msix_names[v], - vp_dev); - if (err) - goto error_irq; - ++vp_dev->msix_used_vectors; - - iowrite16(v, vp_dev->ioaddr + VIRTIO_MSI_CONFIG_VECTOR); - /* Verify we had enough resources to assign the vector */ - v = ioread16(vp_dev->ioaddr + VIRTIO_MSI_CONFIG_VECTOR); - if (v == VIRTIO_MSI_NO_VECTOR) { - err = -EBUSY; - goto error_irq; - } + err = pci_enable_msix(vp_dev->pci_dev, vp_dev->msix_entries, nvectors); + if (err > 0) + err = -ENOSPC; + if (err) + goto error; + vp_dev->msix_vectors = nvectors; + vp_dev->msix_enabled = 1; + + /* Set the vector used for configuration */ + v = vp_dev->msix_used_vectors; + snprintf(vp_dev->msix_names[v], sizeof *vp_dev->msix_names, + "%s-config", name); + err = request_irq(vp_dev->msix_entries[v].vector, + vp_config_changed, 0, vp_dev->msix_names[v], + vp_dev); + if (err) + goto error; + ++vp_dev->msix_used_vectors; + + iowrite16(v, vp_dev->ioaddr + VIRTIO_MSI_CONFIG_VECTOR); + /* Verify we had enough resources to assign the vector */ + v = ioread16(vp_dev->ioaddr + VIRTIO_MSI_CONFIG_VECTOR); + if (v == VIRTIO_MSI_NO_VECTOR) { + err = -EBUSY; + goto error; } - if (vp_dev->msix_vectors && vp_dev->msix_vectors != max_vqs + 1) { + if (!per_vq_vectors) { /* Shared vector for all VQs */ v = vp_dev->msix_used_vectors; snprintf(vp_dev->msix_names[v], sizeof *vp_dev->msix_names, @@ -349,28 +347,25 @@ static int vp_request_vectors(struct virtio_device *vdev, unsigned max_vqs) vp_vring_interrupt, 0, vp_dev->msix_names[v], vp_dev); if (err) - goto error_irq; + goto error; ++vp_dev->msix_used_vectors; } return 0; -error_irq: +error: vp_free_vectors(vdev); - kfree(vp_dev->msix_names); -error_names: - kfree(vp_dev->msix_entries); -error_entries: return err; } static struct virtqueue *vp_find_vq(struct virtio_device *vdev, unsigned index, void (*callback)(struct virtqueue *vq), - const char *name) + const char *name, + u16 vector) { struct virtio_pci_device *vp_dev = to_vp_device(vdev); struct virtio_pci_vq_info *info; struct virtqueue *vq; unsigned long flags, size; - u16 num, vector; + u16 num; int err; /* Select the queue we're interested in */ @@ -389,7 +384,7 @@ static struct virtqueue *vp_find_vq(struct virtio_device *vdev, unsigned index, info->queue_index = index; info->num = num; - info->vector = VIRTIO_MSI_NO_VECTOR; + info->vector = vector; size = PAGE_ALIGN(vring_size(num, VIRTIO_PCI_VRING_ALIGN)); info->queue = alloc_pages_exact(size, GFP_KERNEL|__GFP_ZERO); @@ -413,22 +408,7 @@ static struct virtqueue *vp_find_vq(struct virtio_device *vdev, unsigned index, vq->priv = info; info->vq = vq; - /* allocate per-vq vector if available and necessary */ - if (callback && vp_dev->msix_used_vectors < vp_dev->msix_vectors) { - vector = vp_dev->msix_used_vectors; - snprintf(vp_dev->msix_names[vector], sizeof *vp_dev->msix_names, - "%s-%s", dev_name(&vp_dev->vdev.dev), name); - err = request_irq(vp_dev->msix_entries[vector].vector, - vring_interrupt, 0, - vp_dev->msix_names[vector], vq); - if (err) - goto out_request_irq; - info->vector = vector; - ++vp_dev->msix_used_vectors; - } else - vector = VP_MSIX_VQ_VECTOR; - - if (callback && vp_dev->msix_enabled) { + if (vector != VIRTIO_MSI_NO_VECTOR) { iowrite16(vector, vp_dev->ioaddr + VIRTIO_MSI_QUEUE_VECTOR); vector = ioread16(vp_dev->ioaddr + VIRTIO_MSI_QUEUE_VECTOR); if (vector == VIRTIO_MSI_NO_VECTOR) { @@ -444,11 +424,6 @@ static struct virtqueue *vp_find_vq(struct virtio_device *vdev, unsigned index, return vq; out_assign: - if (info->vector != VIRTIO_MSI_NO_VECTOR) { - free_irq(vp_dev->msix_entries[info->vector].vector, vq); - --vp_dev->msix_used_vectors; - } -out_request_irq: vring_del_virtqueue(vq); out_activate_queue: iowrite32(0, vp_dev->ioaddr + VIRTIO_PCI_QUEUE_PFN); @@ -462,12 +437,13 @@ static void vp_del_vq(struct virtqueue *vq) { struct virtio_pci_device *vp_dev = to_vp_device(vq->vdev); struct virtio_pci_vq_info *info = vq->priv; - unsigned long size; + unsigned long flags, size; - iowrite16(info->queue_index, vp_dev->ioaddr + VIRTIO_PCI_QUEUE_SEL); + spin_lock_irqsave(&vp_dev->lock, flags); + list_del(&info->node); + spin_unlock_irqrestore(&vp_dev->lock, flags); - if (info->vector != VIRTIO_MSI_NO_VECTOR) - free_irq(vp_dev->msix_entries[info->vector].vector, vq); + iowrite16(info->queue_index, vp_dev->ioaddr + VIRTIO_PCI_QUEUE_SEL); if (vp_dev->msix_enabled) { iowrite16(VIRTIO_MSI_NO_VECTOR, @@ -489,36 +465,62 @@ static void vp_del_vq(struct virtqueue *vq) /* the config->del_vqs() implementation */ static void vp_del_vqs(struct virtio_device *vdev) { + struct virtio_pci_device *vp_dev = to_vp_device(vdev); struct virtqueue *vq, *n; + struct virtio_pci_vq_info *info; - list_for_each_entry_safe(vq, n, &vdev->vqs, list) + list_for_each_entry_safe(vq, n, &vdev->vqs, list) { + info = vq->priv; + if (vp_dev->per_vq_vectors) + free_irq(vp_dev->msix_entries[info->vector].vector, vq); vp_del_vq(vq); + } + vp_dev->per_vq_vectors = false; vp_free_vectors(vdev); } -/* the config->find_vqs() implementation */ -static int vp_find_vqs(struct virtio_device *vdev, unsigned nvqs, - struct virtqueue *vqs[], - vq_callback_t *callbacks[], - const char *names[]) +static int vp_try_to_find_vqs(struct virtio_device *vdev, unsigned nvqs, + struct virtqueue *vqs[], + vq_callback_t *callbacks[], + const char *names[], + int nvectors, + bool per_vq_vectors) { - int vectors = 0; - int i, err; - - /* How many vectors would we like? */ - for (i = 0; i < nvqs; ++i) - if (callbacks[i]) - ++vectors; + struct virtio_pci_device *vp_dev = to_vp_device(vdev); + u16 vector; + int i, err, allocated_vectors; - err = vp_request_vectors(vdev, vectors); + err = vp_request_vectors(vdev, nvectors, per_vq_vectors); if (err) goto error_request; + vp_dev->per_vq_vectors = per_vq_vectors; + allocated_vectors = vp_dev->msix_used_vectors; for (i = 0; i < nvqs; ++i) { - vqs[i] = vp_find_vq(vdev, i, callbacks[i], names[i]); - if (IS_ERR(vqs[i])) + if (!callbacks[i] || !vp_dev->msix_enabled) + vector = VIRTIO_MSI_NO_VECTOR; + else if (vp_dev->per_vq_vectors) + vector = allocated_vectors++; + else + vector = VP_MSIX_VQ_VECTOR; + vqs[i] = vp_find_vq(vdev, i, callbacks[i], names[i], vector); + if (IS_ERR(vqs[i])) { + err = PTR_ERR(vqs[i]); goto error_find; + } + /* allocate per-vq irq if available and necessary */ + if (vp_dev->per_vq_vectors && vector != VIRTIO_MSI_NO_VECTOR) { + snprintf(vp_dev->msix_names[vector], sizeof *vp_dev->msix_names, + "%s-%s", dev_name(&vp_dev->vdev.dev), names[i]); + err = request_irq(vp_dev->msix_entries[vector].vector, + vring_interrupt, 0, + vp_dev->msix_names[vector], vqs[i]); + if (err) { + vp_del_vq(vqs[i]); + goto error_find; + } + } } return 0; @@ -526,7 +528,37 @@ error_find: vp_del_vqs(vdev); error_request: - return PTR_ERR(vqs[i]); + return err; +} + +/* the config->find_vqs() implementation */ +static int vp_find_vqs(struct virtio_device *vdev, unsigned nvqs, + struct virtqueue *vqs[], + vq_callback_t *callbacks[], + const char *names[]) +{ + int vectors = 0; + int i, uninitialized_var(err); + + /* How many vectors would we like? */ + for (i = 0; i < nvqs; ++i) + if (callbacks[i]) + ++vectors; + + /* We want at most one vector per queue and one for config changes. */ + err = vp_try_to_find_vqs(vdev, nvqs, vqs, callbacks, names, + vectors + 1, true); + if (!err) + return 0; + /* Fallback to separate vectors for config and a shared for queues. */ + err = vp_try_to_find_vqs(vdev, nvqs, vqs, callbacks, names, + 2, false); + if (!err) + return 0; + /* Finally fall back to regular interrupts. */ + err = vp_try_to_find_vqs(vdev, nvqs, vqs, callbacks, names, + 0, false); + return err; } static struct virtio_config_ops virtio_pci_config_ops = { |