diff options
Diffstat (limited to 'arch')
46 files changed, 1291 insertions, 516 deletions
diff --git a/arch/arm/mach-pxa/corgi.c b/arch/arm/mach-pxa/corgi.c index c5e28a46b29..a8d91b6c136 100644 --- a/arch/arm/mach-pxa/corgi.c +++ b/arch/arm/mach-pxa/corgi.c @@ -27,6 +27,7 @@ #include <linux/spi/spi.h> #include <linux/spi/ads7846.h> #include <linux/spi/corgi_lcd.h> +#include <linux/mtd/sharpsl.h> #include <video/w100fb.h> #include <asm/setup.h> @@ -542,6 +543,55 @@ err_free_1: static inline void corgi_init_spi(void) {} #endif +static struct mtd_partition sharpsl_nand_partitions[] = { + { + .name = "System Area", + .offset = 0, + .size = 7 * 1024 * 1024, + }, + { + .name = "Root Filesystem", + .offset = 7 * 1024 * 1024, + .size = 25 * 1024 * 1024, + }, + { + .name = "Home Filesystem", + .offset = MTDPART_OFS_APPEND, + .size = MTDPART_SIZ_FULL, + }, +}; + +static uint8_t scan_ff_pattern[] = { 0xff, 0xff }; + +static struct nand_bbt_descr sharpsl_bbt = { + .options = 0, + .offs = 4, + .len = 2, + .pattern = scan_ff_pattern +}; + +static struct sharpsl_nand_platform_data sharpsl_nand_platform_data = { + .badblock_pattern = &sharpsl_bbt, + .partitions = sharpsl_nand_partitions, + .nr_partitions = ARRAY_SIZE(sharpsl_nand_partitions), +}; + +static struct resource sharpsl_nand_resources[] = { + { + .start = 0x0C000000, + .end = 0x0C000FFF, + .flags = IORESOURCE_MEM, + }, +}; + +static struct platform_device sharpsl_nand_device = { + .name = "sharpsl-nand", + .id = -1, + .resource = sharpsl_nand_resources, + .num_resources = ARRAY_SIZE(sharpsl_nand_resources), + .dev.platform_data = &sharpsl_nand_platform_data, +}; + static struct mtd_partition sharpsl_rom_parts[] = { { .name ="Boot PROM Filesystem", @@ -577,6 +627,7 @@ static struct platform_device *devices[] __initdata = { &corgifb_device, &corgikbd_device, &corgiled_device, + &sharpsl_nand_device, &sharpsl_rom_device, }; @@ -617,6 +668,9 @@ static void __init corgi_init(void) platform_scoop_config = &corgi_pcmcia_config; + if (machine_is_husky()) + sharpsl_nand_partitions[1].size = 53 * 1024 * 1024; + platform_add_devices(devices, ARRAY_SIZE(devices)); } diff --git a/arch/arm/mach-pxa/poodle.c b/arch/arm/mach-pxa/poodle.c index ae88855bf97..f9093beba75 100644 --- a/arch/arm/mach-pxa/poodle.c +++ b/arch/arm/mach-pxa/poodle.c @@ -24,6 +24,7 @@ #include <linux/gpio.h> #include <linux/spi/spi.h> #include <linux/spi/ads7846.h> +#include <linux/mtd/sharpsl.h> #include <mach/hardware.h> #include <asm/mach-types.h> @@ -414,6 +415,55 @@ static struct pxafb_mach_info poodle_fb_info = { .lcd_conn = LCD_COLOR_TFT_16BPP, }; +static struct mtd_partition sharpsl_nand_partitions[] = { + { + .name = "System Area", + .offset = 0, + .size = 7 * 1024 * 1024, + }, + { + .name = "Root Filesystem", + .offset = 7 * 1024 * 1024, + .size = 22 * 1024 * 1024, + }, + { + .name = "Home Filesystem", + .offset = MTDPART_OFS_APPEND, + .size = MTDPART_SIZ_FULL, + }, +}; + +static uint8_t scan_ff_pattern[] = { 0xff, 0xff }; + +static struct nand_bbt_descr sharpsl_bbt = { + .options = 0, + .offs = 4, + .len = 2, + .pattern = scan_ff_pattern +}; + +static struct sharpsl_nand_platform_data sharpsl_nand_platform_data = { + .badblock_pattern = &sharpsl_bbt, + .partitions = sharpsl_nand_partitions, + .nr_partitions = ARRAY_SIZE(sharpsl_nand_partitions), +}; + +static struct resource sharpsl_nand_resources[] = { + { + .start = 0x0C000000, + .end = 0x0C000FFF, + .flags = IORESOURCE_MEM, + }, +}; + +static struct platform_device sharpsl_nand_device = { + .name = "sharpsl-nand", + .id = -1, + .resource = sharpsl_nand_resources, + .num_resources = ARRAY_SIZE(sharpsl_nand_resources), + .dev.platform_data = &sharpsl_nand_platform_data, +}; + static struct mtd_partition sharpsl_rom_parts[] = { { .name ="Boot PROM Filesystem", @@ -447,6 +497,7 @@ static struct platform_device sharpsl_rom_device = { static struct platform_device *devices[] __initdata = { &poodle_locomo_device, &poodle_scoop_device, + &sharpsl_nand_device, &sharpsl_rom_device, }; diff --git a/arch/arm/mach-pxa/spitz.c b/arch/arm/mach-pxa/spitz.c index 7299d87a1cb..6d447c9ce8a 100644 --- a/arch/arm/mach-pxa/spitz.c +++ b/arch/arm/mach-pxa/spitz.c @@ -31,6 +31,7 @@ #include <linux/spi/spi.h> #include <linux/spi/ads7846.h> #include <linux/spi/corgi_lcd.h> +#include <linux/mtd/sharpsl.h> #include <asm/setup.h> #include <asm/memory.h> @@ -613,6 +614,54 @@ static struct pxafb_mach_info spitz_pxafb_info = { .lcd_conn = LCD_COLOR_TFT_16BPP | LCD_ALTERNATE_MAPPING, }; +static struct mtd_partition sharpsl_nand_partitions[] = { + { + .name = "System Area", + .offset = 0, + .size = 7 * 1024 * 1024, + }, + { + .name = "Root Filesystem", + .offset = 7 * 1024 * 1024, + }, + { + .name = "Home Filesystem", + .offset = MTDPART_OFS_APPEND, + .size = MTDPART_SIZ_FULL, + }, +}; + +static uint8_t scan_ff_pattern[] = { 0xff, 0xff }; + +static struct nand_bbt_descr sharpsl_bbt = { + .options = 0, + .offs = 4, + .len = 2, + .pattern = scan_ff_pattern +}; + +static struct sharpsl_nand_platform_data sharpsl_nand_platform_data = { + .badblock_pattern = &sharpsl_bbt, + .partitions = sharpsl_nand_partitions, + .nr_partitions = ARRAY_SIZE(sharpsl_nand_partitions), +}; + +static struct resource sharpsl_nand_resources[] = { + { + .start = 0x0C000000, + .end = 0x0C000FFF, + .flags = IORESOURCE_MEM, + }, +}; + +static struct platform_device sharpsl_nand_device = { + .name = "sharpsl-nand", + .id = -1, + .resource = sharpsl_nand_resources, + .num_resources = ARRAY_SIZE(sharpsl_nand_resources), + .dev.platform_data = &sharpsl_nand_platform_data, +}; + static struct mtd_partition sharpsl_rom_parts[] = { { @@ -648,6 +697,7 @@ static struct platform_device *devices[] __initdata = { &spitzscoop_device, &spitzkbd_device, &spitzled_device, + &sharpsl_nand_device, &sharpsl_rom_device, }; @@ -671,6 +721,14 @@ static void __init common_init(void) pm_power_off = spitz_poweroff; arm_pm_restart = spitz_restart; + if (machine_is_spitz()) { + sharpsl_nand_partitions[1].size = 5 * 1024 * 1024; + } else if (machine_is_akita()) { + sharpsl_nand_partitions[1].size = 58 * 1024 * 1024; + } else if (machine_is_borzoi()) { + sharpsl_nand_partitions[1].size = 32 * 1024 * 1024; + } + PMCR = 0x00; /* Stop 3.6MHz and drive HIGH to PCMCIA and CS */ @@ -715,10 +773,29 @@ static struct i2c_board_info akita_i2c_board_info[] = { }, }; +static struct nand_bbt_descr sharpsl_akita_bbt = { + .options = 0, + .offs = 4, + .len = 1, + .pattern = scan_ff_pattern +}; + +static struct nand_ecclayout akita_oobinfo = { + .eccbytes = 24, + .eccpos = { + 0x5, 0x1, 0x2, 0x3, 0x6, 0x7, 0x15, 0x11, + 0x12, 0x13, 0x16, 0x17, 0x25, 0x21, 0x22, 0x23, + 0x26, 0x27, 0x35, 0x31, 0x32, 0x33, 0x36, 0x37}, + .oobfree = {{0x08, 0x09}} +}; + static void __init akita_init(void) { spitz_ficp_platform_data.transceiver_mode = akita_irda_transceiver_mode; + sharpsl_nand_platform_data.badblock_pattern = &sharpsl_akita_bbt; + sharpsl_nand_platform_data.ecc_layout = &akita_oobinfo; + /* We just pretend the second element of the array doesn't exist */ spitz_pcmcia_config.num_devs = 1; platform_scoop_config = &spitz_pcmcia_config; diff --git a/arch/avr32/mach-at32ap/at32ap700x.c b/arch/avr32/mach-at32ap/at32ap700x.c index ea7bc1e8562..3fbfd1e32a9 100644 --- a/arch/avr32/mach-at32ap/at32ap700x.c +++ b/arch/avr32/mach-at32ap/at32ap700x.c @@ -1305,7 +1305,7 @@ struct platform_device *__init at32_add_device_mci(unsigned int id, struct mci_platform_data *data) { struct platform_device *pdev; - struct dw_dma_slave *dws; + struct dw_dma_slave *dws = &data->dma_slave; u32 pioa_mask; u32 piob_mask; @@ -1324,22 +1324,13 @@ at32_add_device_mci(unsigned int id, struct mci_platform_data *data) ARRAY_SIZE(atmel_mci0_resource))) goto fail; - if (data->dma_slave) - dws = kmemdup(to_dw_dma_slave(data->dma_slave), - sizeof(struct dw_dma_slave), GFP_KERNEL); - else - dws = kzalloc(sizeof(struct dw_dma_slave), GFP_KERNEL); - - dws->slave.dev = &pdev->dev; - dws->slave.dma_dev = &dw_dmac0_device.dev; - dws->slave.reg_width = DMA_SLAVE_WIDTH_32BIT; + dws->dma_dev = &dw_dmac0_device.dev; + dws->reg_width = DW_DMA_SLAVE_WIDTH_32BIT; dws->cfg_hi = (DWC_CFGH_SRC_PER(0) | DWC_CFGH_DST_PER(1)); dws->cfg_lo &= ~(DWC_CFGL_HS_DST_POL | DWC_CFGL_HS_SRC_POL); - data->dma_slave = &dws->slave; - if (platform_device_add_data(pdev, data, sizeof(struct mci_platform_data))) goto fail; diff --git a/arch/ia64/include/asm/acpi-ext.h b/arch/ia64/include/asm/acpi-ext.h index 734d137dda6..7f8362b379e 100644 --- a/arch/ia64/include/asm/acpi-ext.h +++ b/arch/ia64/include/asm/acpi-ext.h @@ -14,7 +14,6 @@ #define _ASM_IA64_ACPI_EXT_H #include <linux/types.h> -#include <acpi/actypes.h> extern acpi_status hp_acpi_csr_space (acpi_handle, u64 *base, u64 *length); diff --git a/arch/ia64/include/asm/sn/acpi.h b/arch/ia64/include/asm/sn/acpi.h index 9ce2801cbd5..fd480db2556 100644 --- a/arch/ia64/include/asm/sn/acpi.h +++ b/arch/ia64/include/asm/sn/acpi.h @@ -9,8 +9,6 @@ #ifndef _ASM_IA64_SN_ACPI_H #define _ASM_IA64_SN_ACPI_H -#include "acpi/acglobal.h" - extern int sn_acpi_rev; #define SN_ACPI_BASE_SUPPORT() (sn_acpi_rev >= 0x20101) diff --git a/arch/ia64/kernel/acpi.c b/arch/ia64/kernel/acpi.c index 0553648b759..d541671caf4 100644 --- a/arch/ia64/kernel/acpi.c +++ b/arch/ia64/kernel/acpi.c @@ -65,6 +65,7 @@ EXPORT_SYMBOL(pm_idle); void (*pm_power_off) (void); EXPORT_SYMBOL(pm_power_off); +u32 acpi_rsdt_forced; unsigned int acpi_cpei_override; unsigned int acpi_cpei_phys_cpuid; diff --git a/arch/ia64/sn/kernel/io_acpi_init.c b/arch/ia64/sn/kernel/io_acpi_init.c index bc610a6c785..c5a214026a7 100644 --- a/arch/ia64/sn/kernel/io_acpi_init.c +++ b/arch/ia64/sn/kernel/io_acpi_init.c @@ -13,7 +13,6 @@ #include <asm/sn/sn_sal.h> #include "xtalk/hubdev.h" #include <linux/acpi.h> -#include <acpi/acnamesp.h> /* @@ -64,6 +63,7 @@ static acpi_status __init sn_acpi_hubdev_init(acpi_handle handle, u32 depth, void *context, void **ret) { struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL }; + struct acpi_buffer name_buffer = { ACPI_ALLOCATE_BUFFER, NULL }; u64 addr; struct hubdev_info *hubdev; struct hubdev_info *hubdev_ptr; @@ -77,11 +77,12 @@ sn_acpi_hubdev_init(acpi_handle handle, u32 depth, void *context, void **ret) status = acpi_get_vendor_resource(handle, METHOD_NAME__CRS, &sn_uuid, &buffer); if (ACPI_FAILURE(status)) { + acpi_get_name(handle, ACPI_FULL_PATHNAME, &name_buffer); printk(KERN_ERR "sn_acpi_hubdev_init: acpi_get_vendor_resource() " - "(0x%x) failed for: ", status); - acpi_ns_print_node_pathname(handle, NULL); - printk("\n"); + "(0x%x) failed for: %s\n", status, + (char *)name_buffer.pointer); + kfree(name_buffer.pointer); return AE_OK; /* Continue walking namespace */ } @@ -89,11 +90,12 @@ sn_acpi_hubdev_init(acpi_handle handle, u32 depth, void *context, void **ret) vendor = &resource->data.vendor_typed; if ((vendor->byte_length - sizeof(struct acpi_vendor_uuid)) != sizeof(struct hubdev_info *)) { + acpi_get_name(handle, ACPI_FULL_PATHNAME, &name_buffer); printk(KERN_ERR - "sn_acpi_hubdev_init: Invalid vendor data length: %d for: ", - vendor->byte_length); - acpi_ns_print_node_pathname(handle, NULL); - printk("\n"); + "sn_acpi_hubdev_init: Invalid vendor data length: " + "%d for: %s\n", + vendor->byte_length, (char *)name_buffer.pointer); + kfree(name_buffer.pointer); goto exit; } @@ -120,6 +122,7 @@ sn_get_bussoft_ptr(struct pci_bus *bus) { u64 addr; struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL }; + struct acpi_buffer name_buffer = { ACPI_ALLOCATE_BUFFER, NULL }; acpi_handle handle; struct pcibus_bussoft *prom_bussoft_ptr; struct acpi_resource *resource; @@ -131,11 +134,11 @@ sn_get_bussoft_ptr(struct pci_bus *bus) status = acpi_get_vendor_resource(handle, METHOD_NAME__CRS, &sn_uuid, &buffer); if (ACPI_FAILURE(status)) { + acpi_get_name(handle, ACPI_FULL_PATHNAME, &name_buffer); printk(KERN_ERR "%s: " - "acpi_get_vendor_resource() failed (0x%x) for: ", - __func__, status); - acpi_ns_print_node_pathname(handle, NULL); - printk("\n"); + "acpi_get_vendor_resource() failed (0x%x) for: %s\n", + __func__, status, (char *)name_buffer.pointer); + kfree(name_buffer.pointer); return NULL; } resource = buffer.pointer; @@ -168,6 +171,7 @@ sn_extract_device_info(acpi_handle handle, struct pcidev_info **pcidev_info, { u64 addr; struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL }; + struct acpi_buffer name_buffer = { ACPI_ALLOCATE_BUFFER, NULL }; struct sn_irq_info *irq_info, *irq_info_prom; struct pcidev_info *pcidev_ptr, *pcidev_prom_ptr; struct acpi_resource *resource; @@ -182,11 +186,11 @@ sn_extract_device_info(acpi_handle handle, struct pcidev_info **pcidev_info, status = acpi_get_vendor_resource(handle, METHOD_NAME__CRS, &sn_uuid, &buffer); if (ACPI_FAILURE(status)) { + acpi_get_name(handle, ACPI_FULL_PATHNAME, &name_buffer); printk(KERN_ERR - "%s: acpi_get_vendor_resource() failed (0x%x) for: ", - __func__, status); - acpi_ns_print_node_pathname(handle, NULL); - printk("\n"); + "%s: acpi_get_vendor_resource() failed (0x%x) for: %s\n", + __func__, status, (char *)name_buffer.pointer); + kfree(name_buffer.pointer); return 1; } @@ -194,11 +198,12 @@ sn_extract_device_info(acpi_handle handle, struct pcidev_info **pcidev_info, vendor = &resource->data.vendor_typed; if ((vendor->byte_length - sizeof(struct acpi_vendor_uuid)) != sizeof(struct pci_devdev_info *)) { + acpi_get_name(handle, ACPI_FULL_PATHNAME, &name_buffer); printk(KERN_ERR - "%s: Invalid vendor data length: %d for: ", - __func__, vendor->byte_length); - acpi_ns_print_node_pathname(handle, NULL); - printk("\n"); + "%s: Invalid vendor data length: %d for: %s\n", + __func__, vendor->byte_length, + (char *)name_buffer.pointer); + kfree(name_buffer.pointer); ret = 1; goto exit; } @@ -239,6 +244,9 @@ get_host_devfn(acpi_handle device_handle, acpi_handle rootbus_handle) acpi_handle parent; int slot; acpi_status status; + struct acpi_buffer name_buffer = { ACPI_ALLOCATE_BUFFER, NULL }; + + acpi_get_name(device_handle, ACPI_FULL_PATHNAME, &name_buffer); /* * Do an upward search to find the root bus device, and @@ -249,9 +257,8 @@ get_host_devfn(acpi_handle device_handle, acpi_handle rootbus_handle) status = acpi_get_parent(child, &parent); if (ACPI_FAILURE(status)) { printk(KERN_ERR "%s: acpi_get_parent() failed " - "(0x%x) for: ", __func__, status); - acpi_ns_print_node_pathname(child, NULL); - printk("\n"); + "(0x%x) for: %s\n", __func__, status, + (char *)name_buffer.pointer); panic("%s: Unable to find host devfn\n", __func__); } if (parent == rootbus_handle) @@ -259,22 +266,20 @@ get_host_devfn(acpi_handle device_handle, acpi_handle rootbus_handle) child = parent; } if (!child) { - printk(KERN_ERR "%s: Unable to find root bus for: ", - __func__); - acpi_ns_print_node_pathname(device_handle, NULL); - printk("\n"); + printk(KERN_ERR "%s: Unable to find root bus for: %s\n", + __func__, (char *)name_buffer.pointer); BUG(); } status = acpi_evaluate_integer(child, METHOD_NAME__ADR, NULL, &adr); if (ACPI_FAILURE(status)) { - printk(KERN_ERR "%s: Unable to get _ADR (0x%x) for: ", - __func__, status); - acpi_ns_print_node_pathname(child, NULL); - printk("\n"); + printk(KERN_ERR "%s: Unable to get _ADR (0x%x) for: %s\n", + __func__, status, (char *)name_buffer.pointer); panic("%s: Unable to find host devfn\n", __func__); } + kfree(name_buffer.pointer); + slot = (adr >> 16) & 0xffff; function = adr & 0xffff; devfn = PCI_DEVFN(slot, function); @@ -300,27 +305,28 @@ find_matching_device(acpi_handle handle, u32 lvl, void *context, void **rv) int function; int slot; struct sn_pcidev_match *info = context; + struct acpi_buffer name_buffer = { ACPI_ALLOCATE_BUFFER, NULL }; status = acpi_evaluate_integer(handle, METHOD_NAME__ADR, NULL, &adr); if (ACPI_SUCCESS(status)) { status = acpi_get_parent(handle, &parent); if (ACPI_FAILURE(status)) { + acpi_get_name(handle, ACPI_FULL_PATHNAME, &name_buffer); printk(KERN_ERR - "%s: acpi_get_parent() failed (0x%x) for: ", - __func__, status); - acpi_ns_print_node_pathname(handle, NULL); - printk("\n"); + "%s: acpi_get_parent() failed (0x%x) for: %s\n", + __func__, status, (char *)name_buffer.pointer); + kfree(name_buffer.pointer); return AE_OK; } status = acpi_evaluate_integer(parent, METHOD_NAME__BBN, NULL, &bbn); if (ACPI_FAILURE(status)) { + acpi_get_name(handle, ACPI_FULL_PATHNAME, &name_buffer); printk(KERN_ERR - "%s: Failed to find _BBN in parent of: ", - __func__); - acpi_ns_print_node_pathname(handle, NULL); - printk("\n"); + "%s: Failed to find _BBN in parent of: %s\n", + __func__, (char *)name_buffer.pointer); + kfree(name_buffer.pointer); return AE_OK; } @@ -350,24 +356,27 @@ sn_acpi_get_pcidev_info(struct pci_dev *dev, struct pcidev_info **pcidev_info, acpi_handle rootbus_handle; unsigned long long segment; acpi_status status; + struct acpi_buffer name_buffer = { ACPI_ALLOCATE_BUFFER, NULL }; rootbus_handle = PCI_CONTROLLER(dev)->acpi_handle; status = acpi_evaluate_integer(rootbus_handle, METHOD_NAME__SEG, NULL, &segment); if (ACPI_SUCCESS(status)) { if (segment != pci_domain_nr(dev)) { + acpi_get_name(rootbus_handle, ACPI_FULL_PATHNAME, + &name_buffer); printk(KERN_ERR - "%s: Segment number mismatch, 0x%llx vs 0x%x for: ", - __func__, segment, pci_domain_nr(dev)); - acpi_ns_print_node_pathname(rootbus_handle, NULL); - printk("\n"); + "%s: Segment number mismatch, 0x%llx vs 0x%x for: %s\n", + __func__, segment, pci_domain_nr(dev), + (char *)name_buffer.pointer); + kfree(name_buffer.pointer); return 1; } } else { - printk(KERN_ERR "%s: Unable to get __SEG from: ", - __func__); - acpi_ns_print_node_pathname(rootbus_handle, NULL); - printk("\n"); + acpi_get_name(rootbus_handle, ACPI_FULL_PATHNAME, &name_buffer); + printk(KERN_ERR "%s: Unable to get __SEG from: %s\n", + __func__, (char *)name_buffer.pointer); + kfree(name_buffer.pointer); return 1; } diff --git a/arch/ia64/sn/kernel/io_common.c b/arch/ia64/sn/kernel/io_common.c index 8a924a5661d..0d4ffa4da1d 100644 --- a/arch/ia64/sn/kernel/io_common.c +++ b/arch/ia64/sn/kernel/io_common.c @@ -26,7 +26,6 @@ #include <linux/acpi.h> #include <asm/sn/sn2/sn_hwperf.h> #include <asm/sn/acpi.h> -#include "acpi/acglobal.h" extern void sn_init_cpei_timer(void); extern void register_sn_procfs(void); @@ -473,7 +472,7 @@ sn_io_early_init(void) { struct acpi_table_header *header = NULL; - acpi_get_table_by_index(ACPI_TABLE_INDEX_DSDT, &header); + acpi_get_table(ACPI_SIG_DSDT, 1, &header); BUG_ON(header == NULL); sn_acpi_rev = header->oem_revision; } @@ -505,7 +504,7 @@ sn_io_early_init(void) { struct acpi_table_header *header; - (void)acpi_get_table_by_index(ACPI_TABLE_INDEX_DSDT, &header); + (void)acpi_get_table(ACPI_SIG_DSDT, 1, &header); printk(KERN_INFO "ACPI DSDT OEM Rev 0x%x\n", header->oem_revision); } diff --git a/arch/parisc/Makefile b/arch/parisc/Makefile index 5ddad7bd60a..0d428278356 100644 --- a/arch/parisc/Makefile +++ b/arch/parisc/Makefile @@ -77,7 +77,7 @@ libs-y += arch/parisc/lib/ `$(CC) -print-libgcc-file-name` drivers-$(CONFIG_OPROFILE) += arch/parisc/oprofile/ -PALO := $(shell if which palo; then : ; \ +PALO := $(shell if (which palo 2>&1); then : ; \ elif [ -x /sbin/palo ]; then echo /sbin/palo; \ fi) diff --git a/arch/parisc/include/asm/Kbuild b/arch/parisc/include/asm/Kbuild index f88b252e419..2121d99f836 100644 --- a/arch/parisc/include/asm/Kbuild +++ b/arch/parisc/include/asm/Kbuild @@ -1,3 +1,4 @@ include include/asm-generic/Kbuild.asm unifdef-y += pdc.h +unifdef-y += swab.h diff --git a/arch/parisc/include/asm/byteorder.h b/arch/parisc/include/asm/byteorder.h index db148313de5..da66029c4cb 100644 --- a/arch/parisc/include/asm/byteorder.h +++ b/arch/parisc/include/asm/byteorder.h @@ -1,82 +1,7 @@ #ifndef _PARISC_BYTEORDER_H #define _PARISC_BYTEORDER_H -#include <asm/types.h> -#include <linux/compiler.h> - -#ifdef __GNUC__ - -static __inline__ __attribute_const__ __u16 ___arch__swab16(__u16 x) -{ - __asm__("dep %0, 15, 8, %0\n\t" /* deposit 00ab -> 0bab */ - "shd %%r0, %0, 8, %0" /* shift 000000ab -> 00ba */ - : "=r" (x) - : "0" (x)); - return x; -} - -static __inline__ __attribute_const__ __u32 ___arch__swab24(__u32 x) -{ - __asm__("shd %0, %0, 8, %0\n\t" /* shift xabcxabc -> cxab */ - "dep %0, 15, 8, %0\n\t" /* deposit cxab -> cbab */ - "shd %%r0, %0, 8, %0" /* shift 0000cbab -> 0cba */ - : "=r" (x) - : "0" (x)); - return x; -} - -static __inline__ __attribute_const__ __u32 ___arch__swab32(__u32 x) -{ - unsigned int temp; - __asm__("shd %0, %0, 16, %1\n\t" /* shift abcdabcd -> cdab */ - "dep %1, 15, 8, %1\n\t" /* deposit cdab -> cbab */ - "shd %0, %1, 8, %0" /* shift abcdcbab -> dcba */ - : "=r" (x), "=&r" (temp) - : "0" (x)); - return x; -} - - -#if BITS_PER_LONG > 32 -/* -** From "PA-RISC 2.0 Architecture", HP Professional Books. -** See Appendix I page 8 , "Endian Byte Swapping". -** -** Pretty cool algorithm: (* == zero'd bits) -** PERMH 01234567 -> 67452301 into %0 -** HSHL 67452301 -> 7*5*3*1* into %1 -** HSHR 67452301 -> *6*4*2*0 into %0 -** OR %0 | %1 -> 76543210 into %0 (all done!) -*/ -static __inline__ __attribute_const__ __u64 ___arch__swab64(__u64 x) { - __u64 temp; - __asm__("permh,3210 %0, %0\n\t" - "hshl %0, 8, %1\n\t" - "hshr,u %0, 8, %0\n\t" - "or %1, %0, %0" - : "=r" (x), "=&r" (temp) - : "0" (x)); - return x; -} -#define __arch__swab64(x) ___arch__swab64(x) -#define __BYTEORDER_HAS_U64__ -#elif !defined(__STRICT_ANSI__) -static __inline__ __attribute_const__ __u64 ___arch__swab64(__u64 x) -{ - __u32 t1 = ___arch__swab32((__u32) x); - __u32 t2 = ___arch__swab32((__u32) (x >> 32)); - return (((__u64) t1 << 32) | t2); -} -#define __arch__swab64(x) ___arch__swab64(x) -#define __BYTEORDER_HAS_U64__ -#endif - -#define __arch__swab16(x) ___arch__swab16(x) -#define __arch__swab24(x) ___arch__swab24(x) -#define __arch__swab32(x) ___arch__swab32(x) - -#endif /* __GNUC__ */ - +#include <asm/swab.h> #include <linux/byteorder/big_endian.h> #endif /* _PARISC_BYTEORDER_H */ diff --git a/arch/parisc/include/asm/checksum.h b/arch/parisc/include/asm/checksum.h index e9639ccc3fc..c84b2fcb18a 100644 --- a/arch/parisc/include/asm/checksum.h +++ b/arch/parisc/include/asm/checksum.h @@ -182,7 +182,7 @@ static __inline__ __sum16 csum_ipv6_magic(const struct in6_addr *saddr, #endif : "=r" (sum), "=r" (saddr), "=r" (daddr), "=r" (len) : "0" (sum), "1" (saddr), "2" (daddr), "3" (len), "r" (proto) - : "r19", "r20", "r21", "r22"); + : "r19", "r20", "r21", "r22", "memory"); return csum_fold(sum); } diff --git a/arch/parisc/include/asm/io.h b/arch/parisc/include/asm/io.h index 55ddb184210..d3031d1f9d0 100644 --- a/arch/parisc/include/asm/io.h +++ b/arch/parisc/include/asm/io.h @@ -4,12 +4,6 @@ #include <linux/types.h> #include <asm/pgtable.h> -extern unsigned long parisc_vmerge_boundary; -extern unsigned long parisc_vmerge_max_size; - -#define BIO_VMERGE_BOUNDARY parisc_vmerge_boundary -#define BIO_VMERGE_MAX_SIZE parisc_vmerge_max_size - #define virt_to_phys(a) ((unsigned long)__pa(a)) #define phys_to_virt(a) __va(a) #define virt_to_bus virt_to_phys @@ -182,9 +176,9 @@ static inline void __raw_writeq(unsigned long long b, volatile void __iomem *add /* readb can never be const, so use __fswab instead of le*_to_cpu */ #define readb(addr) __raw_readb(addr) -#define readw(addr) __fswab16(__raw_readw(addr)) -#define readl(addr) __fswab32(__raw_readl(addr)) -#define readq(addr) __fswab64(__raw_readq(addr)) +#define readw(addr) le16_to_cpu(__raw_readw(addr)) +#define readl(addr) le32_to_cpu(__raw_readl(addr)) +#define readq(addr) le64_to_cpu(__raw_readq(addr)) #define writeb(b, addr) __raw_writeb(b, addr) #define writew(b, addr) __raw_writew(cpu_to_le16(b), addr) #define writel(b, addr) __raw_writel(cpu_to_le32(b), addr) diff --git a/arch/parisc/include/asm/mmu_context.h b/arch/parisc/include/asm/mmu_context.h index 85856c74ad1..354b2aca990 100644 --- a/arch/parisc/include/asm/mmu_context.h +++ b/arch/parisc/include/asm/mmu_context.h @@ -34,16 +34,21 @@ destroy_context(struct mm_struct *mm) mm->context = 0; } -static inline void load_context(mm_context_t context) +static inline unsigned long __space_to_prot(mm_context_t context) { - mtsp(context, 3); #if SPACEID_SHIFT == 0 - mtctl(context << 1,8); + return context << 1; #else - mtctl(context >> (SPACEID_SHIFT - 1),8); + return context >> (SPACEID_SHIFT - 1); #endif } +static inline void load_context(mm_context_t context) +{ + mtsp(context, 3); + mtctl(__space_to_prot(context), 8); +} + static inline void switch_mm(struct mm_struct *prev, struct mm_struct *next, struct task_struct *tsk) { diff --git a/arch/parisc/include/asm/processor.h b/arch/parisc/include/asm/processor.h index 3c9d34844c8..9d64df8754b 100644 --- a/arch/parisc/include/asm/processor.h +++ b/arch/parisc/include/asm/processor.h @@ -17,6 +17,7 @@ #include <asm/ptrace.h> #include <asm/types.h> #include <asm/system.h> +#include <asm/percpu.h> #endif /* __ASSEMBLY__ */ #define KERNEL_STACK_SIZE (4*PAGE_SIZE) @@ -109,8 +110,7 @@ struct cpuinfo_parisc { }; extern struct system_cpuinfo_parisc boot_cpu_data; -extern struct cpuinfo_parisc cpu_data[NR_CPUS]; -#define current_cpu_data cpu_data[smp_processor_id()] +DECLARE_PER_CPU(struct cpuinfo_parisc, cpu_data); #define CPU_HVERSION ((boot_cpu_data.hversion >> 4) & 0x0FFF) diff --git a/arch/parisc/include/asm/swab.h b/arch/parisc/include/asm/swab.h new file mode 100644 index 00000000000..3ff16c5a335 --- /dev/null +++ b/arch/parisc/include/asm/swab.h @@ -0,0 +1,66 @@ +#ifndef _PARISC_SWAB_H +#define _PARISC_SWAB_H + +#include <asm/types.h> +#include <linux/compiler.h> + +#define __SWAB_64_THRU_32__ + +static inline __attribute_const__ __u16 __arch_swab16(__u16 x) +{ + __asm__("dep %0, 15, 8, %0\n\t" /* deposit 00ab -> 0bab */ + "shd %%r0, %0, 8, %0" /* shift 000000ab -> 00ba */ + : "=r" (x) + : "0" (x)); + return x; +} +#define __arch_swab16 __arch_swab16 + +static inline __attribute_const__ __u32 __arch_swab24(__u32 x) +{ + __asm__("shd %0, %0, 8, %0\n\t" /* shift xabcxabc -> cxab */ + "dep %0, 15, 8, %0\n\t" /* deposit cxab -> cbab */ + "shd %%r0, %0, 8, %0" /* shift 0000cbab -> 0cba */ + : "=r" (x) + : "0" (x)); + return x; +} + +static inline __attribute_const__ __u32 __arch_swab32(__u32 x) +{ + unsigned int temp; + __asm__("shd %0, %0, 16, %1\n\t" /* shift abcdabcd -> cdab */ + "dep %1, 15, 8, %1\n\t" /* deposit cdab -> cbab */ + "shd %0, %1, 8, %0" /* shift abcdcbab -> dcba */ + : "=r" (x), "=&r" (temp) + : "0" (x)); + return x; +} +#define __arch_swab32 __arch_swab32 + +#if BITS_PER_LONG > 32 +/* +** From "PA-RISC 2.0 Architecture", HP Professional Books. +** See Appendix I page 8 , "Endian Byte Swapping". +** +** Pretty cool algorithm: (* == zero'd bits) +** PERMH 01234567 -> 67452301 into %0 +** HSHL 67452301 -> 7*5*3*1* into %1 +** HSHR 67452301 -> *6*4*2*0 into %0 +** OR %0 | %1 -> 76543210 into %0 (all done!) +*/ +static inline __attribute_const__ __u64 __arch_swab64(__u64 x) +{ + __u64 temp; + __asm__("permh,3210 %0, %0\n\t" + "hshl %0, 8, %1\n\t" + "hshr,u %0, 8, %0\n\t" + "or %1, %0, %0" + : "=r" (x), "=&r" (temp) + : "0" (x)); + return x; +} +#define __arch_swab64 __arch_swab64 +#endif /* BITS_PER_LONG > 32 */ + +#endif /* _PARISC_SWAB_H */ diff --git a/arch/parisc/include/asm/uaccess.h b/arch/parisc/include/asm/uaccess.h index 4878b9501f2..1c6dbb6f6e5 100644 --- a/arch/parisc/include/asm/uaccess.h +++ b/arch/parisc/include/asm/uaccess.h @@ -241,4 +241,6 @@ unsigned long copy_in_user(void __user *dst, const void __user *src, unsigned lo #define __copy_to_user_inatomic __copy_to_user #define __copy_from_user_inatomic __copy_from_user +int fixup_exception(struct pt_regs *regs); + #endif /* __PARISC_UACCESS_H */ diff --git a/arch/parisc/kernel/drivers.c b/arch/parisc/kernel/drivers.c index 884b7ce16a3..994bcd98090 100644 --- a/arch/parisc/kernel/drivers.c +++ b/arch/parisc/kernel/drivers.c @@ -549,6 +549,38 @@ static int parisc_generic_match(struct device *dev, struct device_driver *drv) return match_device(to_parisc_driver(drv), to_parisc_device(dev)); } +static ssize_t make_modalias(struct device *dev, char *buf) +{ + const struct parisc_device *padev = to_parisc_device(dev); + const struct parisc_device_id *id = &padev->id; + + return sprintf(buf, "parisc:t%02Xhv%04Xrev%02Xsv%08X\n", + (u8)id->hw_type, (u16)id->hversion, (u8)id->hversion_rev, + (u32)id->sversion); +} + +static int parisc_uevent(struct device *dev, struct kobj_uevent_env *env) +{ + const struct parisc_device *padev; + char modalias[40]; + + if (!dev) + return -ENODEV; + + padev = to_parisc_device(dev); + if (!padev) + return -ENODEV; + + if (add_uevent_var(env, "PARISC_NAME=%s", padev->name)) + return -ENOMEM; + + make_modalias(dev, modalias); + if (add_uevent_var(env, "MODALIAS=%s", modalias)) + return -ENOMEM; + + return 0; +} + #define pa_dev_attr(name, field, format_string) \ static ssize_t name##_show(struct device *dev, struct device_attribute *attr, char *buf) \ { \ @@ -566,12 +598,7 @@ pa_dev_attr_id(sversion, "0x%05x\n"); static ssize_t modalias_show(struct device *dev, struct device_attribute *attr, char *buf) { - struct parisc_device *padev = to_parisc_device(dev); - struct parisc_device_id *id = &padev->id; - - return sprintf(buf, "parisc:t%02Xhv%04Xrev%02Xsv%08X\n", - (u8)id->hw_type, (u16)id->hversion, (u8)id->hversion_rev, - (u32)id->sversion); + return make_modalias(dev, buf); } static struct device_attribute parisc_device_attrs[] = { @@ -587,6 +614,7 @@ static struct device_attribute parisc_device_attrs[] = { struct bus_type parisc_bus_type = { .name = "parisc", .match = parisc_generic_match, + .uevent = parisc_uevent, .dev_attrs = parisc_device_attrs, .probe = parisc_driver_probe, .remove = parisc_driver_remove, diff --git a/arch/parisc/kernel/hpmc.S b/arch/parisc/kernel/hpmc.S index 2cbf13b3ef1..5595a2f3118 100644 --- a/arch/parisc/kernel/hpmc.S +++ b/arch/parisc/kernel/hpmc.S @@ -80,6 +80,7 @@ END(hpmc_pim_data) .import intr_save, code ENTRY(os_hpmc) +.os_hpmc: /* * registers modified: @@ -295,5 +296,10 @@ os_hpmc_6: b . nop ENDPROC(os_hpmc) -ENTRY(os_hpmc_end) /* this label used to compute os_hpmc checksum */ +.os_hpmc_end: nop +.data +.align 4 + .export os_hpmc_size +os_hpmc_size: + .word .os_hpmc_end-.os_hpmc diff --git a/arch/parisc/kernel/irq.c b/arch/parisc/kernel/irq.c index 4cea935e2f9..ac2c822928c 100644 --- a/arch/parisc/kernel/irq.c +++ b/arch/parisc/kernel/irq.c @@ -298,7 +298,7 @@ unsigned long txn_affinity_addr(unsigned int irq, int cpu) irq_desc[irq].affinity = cpumask_of_cpu(cpu); #endif - return cpu_data[cpu].txn_addr; + return per_cpu(cpu_data, cpu).txn_addr; } @@ -309,8 +309,9 @@ unsigned long txn_alloc_addr(unsigned int virt_irq) next_cpu++; /* assign to "next" CPU we want this bugger on */ /* validate entry */ - while ((next_cpu < NR_CPUS) && (!cpu_data[next_cpu].txn_addr || - !cpu_online(next_cpu))) + while ((next_cpu < NR_CPUS) && + (!per_cpu(cpu_data, next_cpu).txn_addr || + !cpu_online(next_cpu))) next_cpu++; if (next_cpu >= NR_CPUS) @@ -359,7 +360,7 @@ void do_cpu_irq_mask(struct pt_regs *regs) printk(KERN_DEBUG "redirecting irq %d from CPU %d to %d\n", irq, smp_processor_id(), cpu); gsc_writel(irq + CPU_IRQ_BASE, - cpu_data[cpu].hpa); + per_cpu(cpu_data, cpu).hpa); goto set_out; } #endif @@ -421,5 +422,5 @@ void __init init_IRQ(void) void ack_bad_irq(unsigned int irq) { - printk("unexpected IRQ %d\n", irq); + printk(KERN_WARNING "unexpected IRQ %d\n", irq); } diff --git a/arch/parisc/kernel/pdc_cons.c b/arch/parisc/kernel/pdc_cons.c index ccb68090781..1ff366cb968 100644 --- a/arch/parisc/kernel/pdc_cons.c +++ b/arch/parisc/kernel/pdc_cons.c @@ -52,7 +52,7 @@ #include <linux/tty.h> #include <asm/pdc.h> /* for iodc_call() proto and friends */ -static spinlock_t pdc_console_lock = SPIN_LOCK_UNLOCKED; +static DEFINE_SPINLOCK(pdc_console_lock); static void pdc_console_write(struct console *co, const char *s, unsigned count) { diff --git a/arch/parisc/kernel/perf.c b/arch/parisc/kernel/perf.c index f696f57faa1..75099efb3bf 100644 --- a/arch/parisc/kernel/perf.c +++ b/arch/parisc/kernel/perf.c @@ -541,9 +541,9 @@ static int __init perf_init(void) spin_lock_init(&perf_lock); /* TODO: this only lets us access the first cpu.. what to do for SMP? */ - cpu_device = cpu_data[0].dev; + cpu_device = per_cpu(cpu_data, 0).dev; printk("Performance monitoring counters enabled for %s\n", - cpu_data[0].dev->name); + per_cpu(cpu_data, 0).dev->name); return 0; } diff --git a/arch/parisc/kernel/processor.c b/arch/parisc/kernel/processor.c index 370086fb833..ecb609342fe 100644 --- a/arch/parisc/kernel/processor.c +++ b/arch/parisc/kernel/processor.c @@ -3,7 +3,7 @@ * Initial setup-routines for HP 9000 based hardware. * * Copyright (C) 1991, 1992, 1995 Linus Torvalds - * Modifications for PA-RISC (C) 1999 Helge Deller <deller@gmx.de> + * Modifications for PA-RISC (C) 1999-2008 Helge Deller <deller@gmx.de> * Modifications copyright 1999 SuSE GmbH (Philipp Rumpf) * Modifications copyright 2000 Martin K. Petersen <mkp@mkp.net> * Modifications copyright 2000 Philipp Rumpf <prumpf@tux.org> @@ -46,7 +46,7 @@ struct system_cpuinfo_parisc boot_cpu_data __read_mostly; EXPORT_SYMBOL(boot_cpu_data); -struct cpuinfo_parisc cpu_data[NR_CPUS] __read_mostly; +DEFINE_PER_CPU(struct cpuinfo_parisc, cpu_data); extern int update_cr16_clocksource(void); /* from time.c */ @@ -69,6 +69,23 @@ extern int update_cr16_clocksource(void); /* from time.c */ */ /** + * init_cpu_profiler - enable/setup per cpu profiling hooks. + * @cpunum: The processor instance. + * + * FIXME: doesn't do much yet... + */ +static void __cpuinit +init_percpu_prof(unsigned long cpunum) +{ + struct cpuinfo_parisc *p; + + p = &per_cpu(cpu_data, cpunum); + p->prof_counter = 1; + p->prof_multiplier = 1; +} + + +/** * processor_probe - Determine if processor driver should claim this device. * @dev: The device which has been found. * @@ -147,7 +164,7 @@ static int __cpuinit processor_probe(struct parisc_device *dev) } #endif - p = &cpu_data[cpuid]; + p = &per_cpu(cpu_data, cpuid); boot_cpu_data.cpu_count++; /* initialize counters - CPU 0 gets it_value set in time_init() */ @@ -162,12 +179,9 @@ static int __cpuinit processor_probe(struct parisc_device *dev) #ifdef CONFIG_SMP /* ** FIXME: review if any other initialization is clobbered - ** for boot_cpu by the above memset(). + ** for boot_cpu by the above memset(). */ - - /* stolen from init_percpu_prof() */ - cpu_data[cpuid].prof_counter = 1; - cpu_data[cpuid].prof_multiplier = 1; + init_percpu_prof(cpuid); #endif /* @@ -261,19 +275,6 @@ void __init collect_boot_cpu_data(void) } -/** - * init_cpu_profiler - enable/setup per cpu profiling hooks. - * @cpunum: The processor instance. - * - * FIXME: doesn't do much yet... - */ -static inline void __init -init_percpu_prof(int cpunum) -{ - cpu_data[cpunum].prof_counter = 1; - cpu_data[cpunum].prof_multiplier = 1; -} - /** * init_per_cpu - Handle individual processor initializations. @@ -293,7 +294,7 @@ init_percpu_prof(int cpunum) * * o Enable CPU profiling hooks. */ -int __init init_per_cpu(int cpunum) +int __cpuinit init_per_cpu(int cpunum) { int ret; struct pdc_coproc_cfg coproc_cfg; @@ -307,8 +308,8 @@ int __init init_per_cpu(int cpunum) /* FWIW, FP rev/model is a more accurate way to determine ** CPU type. CPU rev/model has some ambiguous cases. */ - cpu_data[cpunum].fp_rev = coproc_cfg.revision; - cpu_data[cpunum].fp_model = coproc_cfg.model; + per_cpu(cpu_data, cpunum).fp_rev = coproc_cfg.revision; + per_cpu(cpu_data, cpunum).fp_model = coproc_cfg.model; printk(KERN_INFO "FP[%d] enabled: Rev %ld Model %ld\n", cpunum, coproc_cfg.revision, coproc_cfg.model); @@ -344,16 +345,17 @@ int __init init_per_cpu(int cpunum) int show_cpuinfo (struct seq_file *m, void *v) { - int n; + unsigned long cpu; - for(n=0; n<boot_cpu_data.cpu_count; n++) { + for_each_online_cpu(cpu) { + const struct cpuinfo_parisc *cpuinfo = &per_cpu(cpu_data, cpu); #ifdef CONFIG_SMP - if (0 == cpu_data[n].hpa) + if (0 == cpuinfo->hpa) continue; #endif - seq_printf(m, "processor\t: %d\n" + seq_printf(m, "processor\t: %lu\n" "cpu family\t: PA-RISC %s\n", - n, boot_cpu_data.family_name); + cpu, boot_cpu_data.family_name); seq_printf(m, "cpu\t\t: %s\n", boot_cpu_data.cpu_name ); @@ -365,8 +367,8 @@ show_cpuinfo (struct seq_file *m, void *v) seq_printf(m, "model\t\t: %s\n" "model name\t: %s\n", boot_cpu_data.pdc.sys_model_name, - cpu_data[n].dev ? - cpu_data[n].dev->name : "Unknown" ); + cpuinfo->dev ? + cpuinfo->dev->name : "Unknown"); seq_printf(m, "hversion\t: 0x%08x\n" "sversion\t: 0x%08x\n", @@ -377,8 +379,8 @@ show_cpuinfo (struct seq_file *m, void *v) show_cache_info(m); seq_printf(m, "bogomips\t: %lu.%02lu\n", - cpu_data[n].loops_per_jiffy / (500000 / HZ), - (cpu_data[n].loops_per_jiffy / (5000 / HZ)) % 100); + cpuinfo->loops_per_jiffy / (500000 / HZ), + (cpuinfo->loops_per_jiffy / (5000 / HZ)) % 100); seq_printf(m, "software id\t: %ld\n\n", boot_cpu_data.pdc.model.sw_id); diff --git a/arch/parisc/kernel/setup.c b/arch/parisc/kernel/setup.c index 7d27853ff8c..82131ca8e05 100644 --- a/arch/parisc/kernel/setup.c +++ b/arch/parisc/kernel/setup.c @@ -58,11 +58,6 @@ int parisc_bus_is_phys __read_mostly = 1; /* Assume no IOMMU is present */ EXPORT_SYMBOL(parisc_bus_is_phys); #endif -/* This sets the vmerge boundary and size, it's here because it has to - * be available on all platforms (zero means no-virtual merging) */ -unsigned long parisc_vmerge_boundary = 0; -unsigned long parisc_vmerge_max_size = 0; - void __init setup_cmdline(char **cmdline_p) { extern unsigned int boot_args[]; @@ -321,7 +316,7 @@ static int __init parisc_init(void) processor_init(); printk(KERN_INFO "CPU(s): %d x %s at %d.%06d MHz\n", - boot_cpu_data.cpu_count, + num_present_cpus(), boot_cpu_data.cpu_name, boot_cpu_data.cpu_hz / 1000000, boot_cpu_data.cpu_hz % 1000000 ); @@ -387,8 +382,8 @@ void start_parisc(void) if (ret >= 0 && coproc_cfg.ccr_functional) { mtctl(coproc_cfg.ccr_functional, 10); - cpu_data[cpunum].fp_rev = coproc_cfg.revision; - cpu_data[cpunum].fp_model = coproc_cfg.model; + per_cpu(cpu_data, cpunum).fp_rev = coproc_cfg.revision; + per_cpu(cpu_data, cpunum).fp_model = coproc_cfg.model; asm volatile ("fstd %fr0,8(%sp)"); } else { diff --git a/arch/parisc/kernel/smp.c b/arch/parisc/kernel/smp.c index 80bc000523f..9995d7ed581 100644 --- a/arch/parisc/kernel/smp.c +++ b/arch/parisc/kernel/smp.c @@ -56,16 +56,17 @@ static int smp_debug_lvl = 0; if (lvl >= smp_debug_lvl) \ printk(printargs); #else -#define smp_debug(lvl, ...) +#define smp_debug(lvl, ...) do { } while(0) #endif /* DEBUG_SMP */ DEFINE_SPINLOCK(smp_lock); volatile struct task_struct *smp_init_current_idle_task; -static volatile int cpu_now_booting __read_mostly = 0; /* track which CPU is booting */ +/* track which CPU is booting */ +static volatile int cpu_now_booting __cpuinitdata; -static int parisc_max_cpus __read_mostly = 1; +static int parisc_max_cpus __cpuinitdata = 1; DEFINE_PER_CPU(spinlock_t, ipi_lock) = SPIN_LOCK_UNLOCKED; @@ -123,7 +124,7 @@ irqreturn_t ipi_interrupt(int irq, void *dev_id) { int this_cpu = smp_processor_id(); - struct cpuinfo_parisc *p = &cpu_data[this_cpu]; + struct cpuinfo_parisc *p = &per_cpu(cpu_data, this_cpu); unsigned long ops; unsigned long flags; @@ -202,13 +203,13 @@ ipi_interrupt(int irq, void *dev_id) static inline void ipi_send(int cpu, enum ipi_message_type op) { - struct cpuinfo_parisc *p = &cpu_data[cpu]; + struct cpuinfo_parisc *p = &per_cpu(cpu_data, cpu); spinlock_t *lock = &per_cpu(ipi_lock, cpu); unsigned long flags; spin_lock_irqsave(lock, flags); p->pending_ipi |= 1 << op; - gsc_writel(IPI_IRQ - CPU_IRQ_BASE, cpu_data[cpu].hpa); + gsc_writel(IPI_IRQ - CPU_IRQ_BASE, p->hpa); spin_unlock_irqrestore(lock, flags); } @@ -224,10 +225,7 @@ send_IPI_mask(cpumask_t mask, enum ipi_message_type op) static inline void send_IPI_single(int dest_cpu, enum ipi_message_type op) { - if (dest_cpu == NO_PROC_ID) { - BUG(); - return; - } + BUG_ON(dest_cpu == NO_PROC_ID); ipi_send(dest_cpu, op); } @@ -309,8 +307,7 @@ smp_cpu_init(int cpunum) /* Initialise the idle task for this CPU */ atomic_inc(&init_mm.mm_count); current->active_mm = &init_mm; - if(current->mm) - BUG(); + BUG_ON(current->mm); enter_lazy_tlb(&init_mm, current); init_IRQ(); /* make sure no IRQs are enabled or pending */ @@ -345,6 +342,7 @@ void __init smp_callin(void) */ int __cpuinit smp_boot_one_cpu(int cpuid) { + const struct cpuinfo_parisc *p = &per_cpu(cpu_data, cpuid); struct task_struct *idle; long timeout; @@ -376,7 +374,7 @@ int __cpuinit smp_boot_one_cpu(int cpuid) smp_init_current_idle_task = idle ; mb(); - printk("Releasing cpu %d now, hpa=%lx\n", cpuid, cpu_data[cpuid].hpa); + printk(KERN_INFO "Releasing cpu %d now, hpa=%lx\n", cpuid, p->hpa); /* ** This gets PDC to release the CPU from a very tight loop. @@ -387,7 +385,7 @@ int __cpuinit smp_boot_one_cpu(int cpuid) ** EIR{0}). MEM_RENDEZ is valid only when it is nonzero and the ** contents of memory are valid." */ - gsc_writel(TIMER_IRQ - CPU_IRQ_BASE, cpu_data[cpuid].hpa); + gsc_writel(TIMER_IRQ - CPU_IRQ_BASE, p->hpa); mb(); /* @@ -419,12 +417,12 @@ alive: return 0; } -void __devinit smp_prepare_boot_cpu(void) +void __init smp_prepare_boot_cpu(void) { - int bootstrap_processor=cpu_data[0].cpuid; /* CPU ID of BSP */ + int bootstrap_processor = per_cpu(cpu_data, 0).cpuid; /* Setup BSP mappings */ - printk("SMP: bootstrap CPU ID is %d\n",bootstrap_processor); + printk(KERN_INFO "SMP: bootstrap CPU ID is %d\n", bootstrap_processor); cpu_set(bootstrap_processor, cpu_online_map); cpu_set(bootstrap_processor, cpu_present_map); diff --git a/arch/parisc/kernel/time.c b/arch/parisc/kernel/time.c index 4d09203bc69..9d46c43a415 100644 --- a/arch/parisc/kernel/time.c +++ b/arch/parisc/kernel/time.c @@ -60,7 +60,7 @@ irqreturn_t timer_interrupt(int irq, void *dev_id) unsigned long cycles_elapsed, ticks_elapsed; unsigned long cycles_remainder; unsigned int cpu = smp_processor_id(); - struct cpuinfo_parisc *cpuinfo = &cpu_data[cpu]; + struct cpuinfo_parisc *cpuinfo = &per_cpu(cpu_data, cpu); /* gcc can optimize for "read-only" case with a local clocktick */ unsigned long cpt = clocktick; @@ -213,7 +213,7 @@ void __init start_cpu_itimer(void) mtctl(next_tick, 16); /* kick off Interval Timer (CR16) */ - cpu_data[cpu].it_value = next_tick; + per_cpu(cpu_data, cpu).it_value = next_tick; } struct platform_device rtc_parisc_dev = { diff --git a/arch/parisc/kernel/topology.c b/arch/parisc/kernel/topology.c index d71cb018a21..f5159381fdd 100644 --- a/arch/parisc/kernel/topology.c +++ b/arch/parisc/kernel/topology.c @@ -22,14 +22,14 @@ #include <linux/cpu.h> #include <linux/cache.h> -static struct cpu cpu_devices[NR_CPUS] __read_mostly; +static DEFINE_PER_CPU(struct cpu, cpu_devices); static int __init topology_init(void) { int num; for_each_present_cpu(num) { - register_cpu(&cpu_devices[num], num); + register_cpu(&per_cpu(cpu_devices, num), num); } return 0; } diff --git a/arch/parisc/kernel/traps.c b/arch/parisc/kernel/traps.c index 4c771cd580e..ba658d2086f 100644 --- a/arch/parisc/kernel/traps.c +++ b/arch/parisc/kernel/traps.c @@ -745,6 +745,10 @@ void handle_interruption(int code, struct pt_regs *regs) /* Fall Through */ case 27: /* Data memory protection ID trap */ + if (code == 27 && !user_mode(regs) && + fixup_exception(regs)) + return; + die_if_kernel("Protection id trap", regs, code); si.si_code = SEGV_MAPERR; si.si_signo = SIGSEGV; @@ -821,8 +825,8 @@ void handle_interruption(int code, struct pt_regs *regs) int __init check_ivt(void *iva) { + extern u32 os_hpmc_size; extern const u32 os_hpmc[]; - extern const u32 os_hpmc_end[]; int i; u32 check = 0; @@ -839,8 +843,7 @@ int __init check_ivt(void *iva) *ivap++ = 0; /* Compute Checksum for HPMC handler */ - - length = os_hpmc_end - os_hpmc; + length = os_hpmc_size; ivap[7] = length; hpmcp = (u32 *)os_hpmc; diff --git a/arch/parisc/kernel/unwind.c b/arch/parisc/kernel/unwind.c index 6773c582e45..69dad5a850a 100644 --- a/arch/parisc/kernel/unwind.c +++ b/arch/parisc/kernel/unwind.c @@ -372,7 +372,7 @@ void unwind_frame_init_from_blocked_task(struct unwind_frame_info *info, struct struct pt_regs *r = &t->thread.regs; struct pt_regs *r2; - r2 = kmalloc(sizeof(struct pt_regs), GFP_KERNEL); + r2 = kmalloc(sizeof(struct pt_regs), GFP_ATOMIC); if (!r2) return; *r2 = *r; diff --git a/arch/parisc/lib/iomap.c b/arch/parisc/lib/iomap.c index 9abed07db7f..5069e8b2ca7 100644 --- a/arch/parisc/lib/iomap.c +++ b/arch/parisc/lib/iomap.c @@ -261,7 +261,7 @@ static const struct iomap_ops iomem_ops = { iomem_write32r, }; -const struct iomap_ops *iomap_ops[8] = { +static const struct iomap_ops *iomap_ops[8] = { [0] = &ioport_ops, [7] = &iomem_ops }; diff --git a/arch/parisc/lib/memcpy.c b/arch/parisc/lib/memcpy.c index 2d68431fc22..bbda909c866 100644 --- a/arch/parisc/lib/memcpy.c +++ b/arch/parisc/lib/memcpy.c @@ -275,7 +275,7 @@ handle_store_error: /* Returns 0 for success, otherwise, returns number of bytes not transferred. */ -unsigned long pa_memcpy(void *dstp, const void *srcp, unsigned long len) +static unsigned long pa_memcpy(void *dstp, const void *srcp, unsigned long len) { register unsigned long src, dst, t1, t2, t3; register unsigned char *pcs, *pcd; diff --git a/arch/parisc/mm/fault.c b/arch/parisc/mm/fault.c index b2e3e9a8cec..92c7fa4ecc3 100644 --- a/arch/parisc/mm/fault.c +++ b/arch/parisc/mm/fault.c @@ -139,13 +139,41 @@ parisc_acctyp(unsigned long code, unsigned int inst) } #endif +int fixup_exception(struct pt_regs *regs) +{ + const struct exception_table_entry *fix; + + fix = search_exception_tables(regs->iaoq[0]); + if (fix) { + struct exception_data *d; + d = &__get_cpu_var(exception_data); + d->fault_ip = regs->iaoq[0]; + d->fault_space = regs->isr; + d->fault_addr = regs->ior; + + regs->iaoq[0] = ((fix->fixup) & ~3); + /* + * NOTE: In some cases the faulting instruction + * may be in the delay slot of a branch. We + * don't want to take the branch, so we don't + * increment iaoq[1], instead we set it to be + * iaoq[0]+4, and clear the B bit in the PSW + */ + regs->iaoq[1] = regs->iaoq[0] + 4; + regs->gr[0] &= ~PSW_B; /* IPSW in gr[0] */ + + return 1; + } + + return 0; +} + void do_page_fault(struct pt_regs *regs, unsigned long code, unsigned long address) { struct vm_area_struct *vma, *prev_vma; struct task_struct *tsk = current; struct mm_struct *mm = tsk->mm; - const struct exception_table_entry *fix; unsigned long acc_type; int fault; @@ -229,32 +257,8 @@ bad_area: no_context: - if (!user_mode(regs)) { - fix = search_exception_tables(regs->iaoq[0]); - - if (fix) { - struct exception_data *d; - - d = &__get_cpu_var(exception_data); - d->fault_ip = regs->iaoq[0]; - d->fault_space = regs->isr; - d->fault_addr = regs->ior; - - regs->iaoq[0] = ((fix->fixup) & ~3); - - /* - * NOTE: In some cases the faulting instruction - * may be in the delay slot of a branch. We - * don't want to take the branch, so we don't - * increment iaoq[1], instead we set it to be - * iaoq[0]+4, and clear the B bit in the PSW - */ - - regs->iaoq[1] = regs->iaoq[0] + 4; - regs->gr[0] &= ~PSW_B; /* IPSW in gr[0] */ - - return; - } + if (!user_mode(regs) && fixup_exception(regs)) { + return; } parisc_terminate("Bad Address (null pointer deref?)", regs, code, address); diff --git a/arch/powerpc/include/asm/cell-pmu.h b/arch/powerpc/include/asm/cell-pmu.h index 8066eede3a0..b4b7338ad79 100644 --- a/arch/powerpc/include/asm/cell-pmu.h +++ b/arch/powerpc/include/asm/cell-pmu.h @@ -37,9 +37,11 @@ #define CBE_PM_STOP_AT_MAX 0x40000000 #define CBE_PM_TRACE_MODE_GET(pm_control) (((pm_control) >> 28) & 0x3) #define CBE_PM_TRACE_MODE_SET(mode) (((mode) & 0x3) << 28) +#define CBE_PM_TRACE_BUF_OVFLW(bit) (((bit) & 0x1) << 17) #define CBE_PM_COUNT_MODE_SET(count) (((count) & 0x3) << 18) #define CBE_PM_FREEZE_ALL_CTRS 0x00100000 #define CBE_PM_ENABLE_EXT_TRACE 0x00008000 +#define CBE_PM_SPU_ADDR_TRACE_SET(msk) (((msk) & 0x3) << 9) /* Macros for the trace_address register. */ #define CBE_PM_TRACE_BUF_FULL 0x00000800 diff --git a/arch/powerpc/include/asm/oprofile_impl.h b/arch/powerpc/include/asm/oprofile_impl.h index 95035c602ba..639dc96077a 100644 --- a/arch/powerpc/include/asm/oprofile_impl.h +++ b/arch/powerpc/include/asm/oprofile_impl.h @@ -32,6 +32,12 @@ struct op_system_config { unsigned long mmcr0; unsigned long mmcr1; unsigned long mmcra; +#ifdef CONFIG_OPROFILE_CELL + /* Register for oprofile user tool to check cell kernel profiling + * suport. + */ + unsigned long cell_support; +#endif #endif unsigned long enable_kernel; unsigned long enable_user; diff --git a/arch/powerpc/oprofile/cell/pr_util.h b/arch/powerpc/oprofile/cell/pr_util.h index dfdbffa0681..964b93974d8 100644 --- a/arch/powerpc/oprofile/cell/pr_util.h +++ b/arch/powerpc/oprofile/cell/pr_util.h @@ -30,6 +30,10 @@ extern struct delayed_work spu_work; extern int spu_prof_running; +#define TRACE_ARRAY_SIZE 1024 + +extern spinlock_t oprof_spu_smpl_arry_lck; + struct spu_overlay_info { /* map of sections within an SPU overlay */ unsigned int vma; /* SPU virtual memory address from elf */ unsigned int size; /* size of section from elf */ @@ -89,10 +93,11 @@ void vma_map_free(struct vma_to_fileoffset_map *map); * Entry point for SPU profiling. * cycles_reset is the SPU_CYCLES count value specified by the user. */ -int start_spu_profiling(unsigned int cycles_reset); - -void stop_spu_profiling(void); +int start_spu_profiling_cycles(unsigned int cycles_reset); +void start_spu_profiling_events(void); +void stop_spu_profiling_cycles(void); +void stop_spu_profiling_events(void); /* add the necessary profiling hooks */ int spu_sync_start(void); diff --git a/arch/powerpc/oprofile/cell/spu_profiler.c b/arch/powerpc/oprofile/cell/spu_profiler.c index 83faa958b9d..9305ddaac51 100644 --- a/arch/powerpc/oprofile/cell/spu_profiler.c +++ b/arch/powerpc/oprofile/cell/spu_profiler.c @@ -18,11 +18,21 @@ #include <asm/cell-pmu.h> #include "pr_util.h" -#define TRACE_ARRAY_SIZE 1024 #define SCALE_SHIFT 14 static u32 *samples; +/* spu_prof_running is a flag used to indicate if spu profiling is enabled + * or not. It is set by the routines start_spu_profiling_cycles() and + * start_spu_profiling_events(). The flag is cleared by the routines + * stop_spu_profiling_cycles() and stop_spu_profiling_events(). These + * routines are called via global_start() and global_stop() which are called in + * op_powerpc_start() and op_powerpc_stop(). These routines are called once + * per system as a result of the user starting/stopping oprofile. Hence, only + * one CPU per user at a time will be changing the value of spu_prof_running. + * In general, OProfile does not protect against multiple users trying to run + * OProfile at a time. + */ int spu_prof_running; static unsigned int profiling_interval; @@ -31,8 +41,8 @@ static unsigned int profiling_interval; #define SPU_PC_MASK 0xFFFF -static DEFINE_SPINLOCK(sample_array_lock); -unsigned long sample_array_lock_flags; +DEFINE_SPINLOCK(oprof_spu_smpl_arry_lck); +unsigned long oprof_spu_smpl_arry_lck_flags; void set_spu_profiling_frequency(unsigned int freq_khz, unsigned int cycles_reset) { @@ -145,13 +155,13 @@ static enum hrtimer_restart profile_spus(struct hrtimer *timer) * sample array must be loaded and then processed for a given * cpu. The sample array is not per cpu. */ - spin_lock_irqsave(&sample_array_lock, - sample_array_lock_flags); + spin_lock_irqsave(&oprof_spu_smpl_arry_lck, + oprof_spu_smpl_arry_lck_flags); num_samples = cell_spu_pc_collection(cpu); if (num_samples == 0) { - spin_unlock_irqrestore(&sample_array_lock, - sample_array_lock_flags); + spin_unlock_irqrestore(&oprof_spu_smpl_arry_lck, + oprof_spu_smpl_arry_lck_flags); continue; } @@ -162,8 +172,8 @@ static enum hrtimer_restart profile_spus(struct hrtimer *timer) num_samples); } - spin_unlock_irqrestore(&sample_array_lock, - sample_array_lock_flags); + spin_unlock_irqrestore(&oprof_spu_smpl_arry_lck, + oprof_spu_smpl_arry_lck_flags); } smp_wmb(); /* insure spu event buffer updates are written */ @@ -182,13 +192,13 @@ static enum hrtimer_restart profile_spus(struct hrtimer *timer) static struct hrtimer timer; /* - * Entry point for SPU profiling. + * Entry point for SPU cycle profiling. * NOTE: SPU profiling is done system-wide, not per-CPU. * * cycles_reset is the count value specified by the user when * setting up OProfile to count SPU_CYCLES. */ -int start_spu_profiling(unsigned int cycles_reset) +int start_spu_profiling_cycles(unsigned int cycles_reset) { ktime_t kt; @@ -212,10 +222,30 @@ int start_spu_profiling(unsigned int cycles_reset) return 0; } -void stop_spu_profiling(void) +/* + * Entry point for SPU event profiling. + * NOTE: SPU profiling is done system-wide, not per-CPU. + * + * cycles_reset is the count value specified by the user when + * setting up OProfile to count SPU_CYCLES. + */ +void start_spu_profiling_events(void) +{ + spu_prof_running = 1; + schedule_delayed_work(&spu_work, DEFAULT_TIMER_EXPIRE); + + return; +} + +void stop_spu_profiling_cycles(void) { spu_prof_running = 0; hrtimer_cancel(&timer); kfree(samples); - pr_debug("SPU_PROF: stop_spu_profiling issued\n"); + pr_debug("SPU_PROF: stop_spu_profiling_cycles issued\n"); +} + +void stop_spu_profiling_events(void) +{ + spu_prof_running = 0; } diff --git a/arch/powerpc/oprofile/common.c b/arch/powerpc/oprofile/common.c index 17807acb05d..21f16edf6c8 100644 --- a/arch/powerpc/oprofile/common.c +++ b/arch/powerpc/oprofile/common.c @@ -132,6 +132,28 @@ static int op_powerpc_create_files(struct super_block *sb, struct dentry *root) oprofilefs_create_ulong(sb, root, "mmcr0", &sys.mmcr0); oprofilefs_create_ulong(sb, root, "mmcr1", &sys.mmcr1); oprofilefs_create_ulong(sb, root, "mmcra", &sys.mmcra); +#ifdef CONFIG_OPROFILE_CELL + /* create a file the user tool can check to see what level of profiling + * support exits with this kernel. Initialize bit mask to indicate + * what support the kernel has: + * bit 0 - Supports SPU event profiling in addition to PPU + * event and cycles; and SPU cycle profiling + * bits 1-31 - Currently unused. + * + * If the file does not exist, then the kernel only supports SPU + * cycle profiling, PPU event and cycle profiling. + */ + oprofilefs_create_ulong(sb, root, "cell_support", &sys.cell_support); + sys.cell_support = 0x1; /* Note, the user OProfile tool must check + * that this bit is set before attempting to + * user SPU event profiling. Older kernels + * will not have this file, hence the user + * tool is not allowed to do SPU event + * profiling on older kernels. Older kernels + * will accept SPU events but collected data + * is garbage. + */ +#endif #endif for (i = 0; i < model->num_counters; ++i) { diff --git a/arch/powerpc/oprofile/op_model_cell.c b/arch/powerpc/oprofile/op_model_cell.c index 25a4ec2514a..ae06c6236d9 100644 --- a/arch/powerpc/oprofile/op_model_cell.c +++ b/arch/powerpc/oprofile/op_model_cell.c @@ -40,14 +40,15 @@ #include "../platforms/cell/interrupt.h" #include "cell/pr_util.h" -static void cell_global_stop_spu(void); +#define PPU_PROFILING 0 +#define SPU_PROFILING_CYCLES 1 +#define SPU_PROFILING_EVENTS 2 -/* - * spu_cycle_reset is the number of cycles between samples. - * This variable is used for SPU profiling and should ONLY be set - * at the beginning of cell_reg_setup; otherwise, it's read-only. - */ -static unsigned int spu_cycle_reset; +#define SPU_EVENT_NUM_START 4100 +#define SPU_EVENT_NUM_STOP 4399 +#define SPU_PROFILE_EVENT_ADDR 4363 /* spu, address trace, decimal */ +#define SPU_PROFILE_EVENT_ADDR_MASK_A 0x146 /* sub unit set to zero */ +#define SPU_PROFILE_EVENT_ADDR_MASK_B 0x186 /* sub unit set to zero */ #define NUM_SPUS_PER_NODE 8 #define SPU_CYCLES_EVENT_NUM 2 /* event number for SPU_CYCLES */ @@ -66,6 +67,21 @@ static unsigned int spu_cycle_reset; #define MAX_SPU_COUNT 0xFFFFFF /* maximum 24 bit LFSR value */ +/* Minumum HW interval timer setting to send value to trace buffer is 10 cycle. + * To configure counter to send value every N cycles set counter to + * 2^32 - 1 - N. + */ +#define NUM_INTERVAL_CYC 0xFFFFFFFF - 10 + +/* + * spu_cycle_reset is the number of cycles between samples. + * This variable is used for SPU profiling and should ONLY be set + * at the beginning of cell_reg_setup; otherwise, it's read-only. + */ +static unsigned int spu_cycle_reset; +static unsigned int profiling_mode; +static int spu_evnt_phys_spu_indx; + struct pmc_cntrl_data { unsigned long vcntr; unsigned long evnts; @@ -105,6 +121,8 @@ struct pm_cntrl { u16 trace_mode; u16 freeze; u16 count_mode; + u16 spu_addr_trace; + u8 trace_buf_ovflw; }; static struct { @@ -122,7 +140,7 @@ static struct { #define GET_INPUT_CONTROL(x) ((x & 0x00000004) >> 2) static DEFINE_PER_CPU(unsigned long[NR_PHYS_CTRS], pmc_values); - +static unsigned long spu_pm_cnt[MAX_NUMNODES * NUM_SPUS_PER_NODE]; static struct pmc_cntrl_data pmc_cntrl[NUM_THREADS][NR_PHYS_CTRS]; /* @@ -152,6 +170,7 @@ static u32 hdw_thread; static u32 virt_cntr_inter_mask; static struct timer_list timer_virt_cntr; +static struct timer_list timer_spu_event_swap; /* * pm_signal needs to be global since it is initialized in @@ -165,7 +184,7 @@ static int spu_rtas_token; /* token for SPU cycle profiling */ static u32 reset_value[NR_PHYS_CTRS]; static int num_counters; static int oprofile_running; -static DEFINE_SPINLOCK(virt_cntr_lock); +static DEFINE_SPINLOCK(cntr_lock); static u32 ctr_enabled; @@ -336,13 +355,13 @@ static void set_pm_event(u32 ctr, int event, u32 unit_mask) for (i = 0; i < NUM_DEBUG_BUS_WORDS; i++) { if (bus_word & (1 << i)) { pm_regs.debug_bus_control |= - (bus_type << (30 - (2 * i))); + (bus_type << (30 - (2 * i))); for (j = 0; j < NUM_INPUT_BUS_WORDS; j++) { if (input_bus[j] == 0xff) { input_bus[j] = i; pm_regs.group_control |= - (i << (30 - (2 * j))); + (i << (30 - (2 * j))); break; } @@ -367,12 +386,16 @@ static void write_pm_cntrl(int cpu) if (pm_regs.pm_cntrl.stop_at_max == 1) val |= CBE_PM_STOP_AT_MAX; - if (pm_regs.pm_cntrl.trace_mode == 1) + if (pm_regs.pm_cntrl.trace_mode != 0) val |= CBE_PM_TRACE_MODE_SET(pm_regs.pm_cntrl.trace_mode); + if (pm_regs.pm_cntrl.trace_buf_ovflw == 1) + val |= CBE_PM_TRACE_BUF_OVFLW(pm_regs.pm_cntrl.trace_buf_ovflw); if (pm_regs.pm_cntrl.freeze == 1) val |= CBE_PM_FREEZE_ALL_CTRS; + val |= CBE_PM_SPU_ADDR_TRACE_SET(pm_regs.pm_cntrl.spu_addr_trace); + /* * Routine set_count_mode must be called previously to set * the count mode based on the user selection of user and kernel. @@ -441,7 +464,7 @@ static void cell_virtual_cntr(unsigned long data) * not both playing with the counters on the same node. */ - spin_lock_irqsave(&virt_cntr_lock, flags); + spin_lock_irqsave(&cntr_lock, flags); prev_hdw_thread = hdw_thread; @@ -480,7 +503,7 @@ static void cell_virtual_cntr(unsigned long data) cbe_disable_pm_interrupts(cpu); for (i = 0; i < num_counters; i++) { per_cpu(pmc_values, cpu + prev_hdw_thread)[i] - = cbe_read_ctr(cpu, i); + = cbe_read_ctr(cpu, i); if (per_cpu(pmc_values, cpu + next_hdw_thread)[i] == 0xFFFFFFFF) @@ -527,7 +550,7 @@ static void cell_virtual_cntr(unsigned long data) cbe_enable_pm(cpu); } - spin_unlock_irqrestore(&virt_cntr_lock, flags); + spin_unlock_irqrestore(&cntr_lock, flags); mod_timer(&timer_virt_cntr, jiffies + HZ / 10); } @@ -541,38 +564,146 @@ static void start_virt_cntrs(void) add_timer(&timer_virt_cntr); } -/* This function is called once for all cpus combined */ -static int cell_reg_setup(struct op_counter_config *ctr, +static int cell_reg_setup_spu_cycles(struct op_counter_config *ctr, struct op_system_config *sys, int num_ctrs) { - int i, j, cpu; - spu_cycle_reset = 0; + spu_cycle_reset = ctr[0].count; - if (ctr[0].event == SPU_CYCLES_EVENT_NUM) { - spu_cycle_reset = ctr[0].count; + /* + * Each node will need to make the rtas call to start + * and stop SPU profiling. Get the token once and store it. + */ + spu_rtas_token = rtas_token("ibm,cbe-spu-perftools"); + + if (unlikely(spu_rtas_token == RTAS_UNKNOWN_SERVICE)) { + printk(KERN_ERR + "%s: rtas token ibm,cbe-spu-perftools unknown\n", + __func__); + return -EIO; + } + return 0; +} + +/* Unfortunately, the hardware will only support event profiling + * on one SPU per node at a time. Therefore, we must time slice + * the profiling across all SPUs in the node. Note, we do this + * in parallel for each node. The following routine is called + * periodically based on kernel timer to switch which SPU is + * being monitored in a round robbin fashion. + */ +static void spu_evnt_swap(unsigned long data) +{ + int node; + int cur_phys_spu, nxt_phys_spu, cur_spu_evnt_phys_spu_indx; + unsigned long flags; + int cpu; + int ret; + u32 interrupt_mask; + + + /* enable interrupts on cntr 0 */ + interrupt_mask = CBE_PM_CTR_OVERFLOW_INTR(0); + + hdw_thread = 0; + + /* Make sure spu event interrupt handler and spu event swap + * don't access the counters simultaneously. + */ + spin_lock_irqsave(&cntr_lock, flags); + + cur_spu_evnt_phys_spu_indx = spu_evnt_phys_spu_indx; + + if (++(spu_evnt_phys_spu_indx) == NUM_SPUS_PER_NODE) + spu_evnt_phys_spu_indx = 0; + + pm_signal[0].sub_unit = spu_evnt_phys_spu_indx; + pm_signal[1].sub_unit = spu_evnt_phys_spu_indx; + pm_signal[2].sub_unit = spu_evnt_phys_spu_indx; + + /* switch the SPU being profiled on each node */ + for_each_online_cpu(cpu) { + if (cbe_get_hw_thread_id(cpu)) + continue; + + node = cbe_cpu_to_node(cpu); + cur_phys_spu = (node * NUM_SPUS_PER_NODE) + + cur_spu_evnt_phys_spu_indx; + nxt_phys_spu = (node * NUM_SPUS_PER_NODE) + + spu_evnt_phys_spu_indx; /* - * Each node will need to make the rtas call to start - * and stop SPU profiling. Get the token once and store it. + * stop counters, save counter values, restore counts + * for previous physical SPU */ - spu_rtas_token = rtas_token("ibm,cbe-spu-perftools"); + cbe_disable_pm(cpu); + cbe_disable_pm_interrupts(cpu); - if (unlikely(spu_rtas_token == RTAS_UNKNOWN_SERVICE)) { - printk(KERN_ERR - "%s: rtas token ibm,cbe-spu-perftools unknown\n", - __func__); - return -EIO; - } + spu_pm_cnt[cur_phys_spu] + = cbe_read_ctr(cpu, 0); + + /* restore previous count for the next spu to sample */ + /* NOTE, hardware issue, counter will not start if the + * counter value is at max (0xFFFFFFFF). + */ + if (spu_pm_cnt[nxt_phys_spu] >= 0xFFFFFFFF) + cbe_write_ctr(cpu, 0, 0xFFFFFFF0); + else + cbe_write_ctr(cpu, 0, spu_pm_cnt[nxt_phys_spu]); + + pm_rtas_reset_signals(cbe_cpu_to_node(cpu)); + + /* setup the debug bus measure the one event and + * the two events to route the next SPU's PC on + * the debug bus + */ + ret = pm_rtas_activate_signals(cbe_cpu_to_node(cpu), 3); + if (ret) + printk(KERN_ERR "%s: pm_rtas_activate_signals failed, " + "SPU event swap\n", __func__); + + /* clear the trace buffer, don't want to take PC for + * previous SPU*/ + cbe_write_pm(cpu, trace_address, 0); + + enable_ctr(cpu, 0, pm_regs.pm07_cntrl); + + /* Enable interrupts on the CPU thread that is starting */ + cbe_enable_pm_interrupts(cpu, hdw_thread, + interrupt_mask); + cbe_enable_pm(cpu); } - pm_rtas_token = rtas_token("ibm,cbe-perftools"); + spin_unlock_irqrestore(&cntr_lock, flags); + /* swap approximately every 0.1 seconds */ + mod_timer(&timer_spu_event_swap, jiffies + HZ / 25); +} + +static void start_spu_event_swap(void) +{ + init_timer(&timer_spu_event_swap); + timer_spu_event_swap.function = spu_evnt_swap; + timer_spu_event_swap.data = 0UL; + timer_spu_event_swap.expires = jiffies + HZ / 25; + add_timer(&timer_spu_event_swap); +} + +static int cell_reg_setup_spu_events(struct op_counter_config *ctr, + struct op_system_config *sys, int num_ctrs) +{ + int i; + + /* routine is called once for all nodes */ + + spu_evnt_phys_spu_indx = 0; /* - * For all events excetp PPU CYCLEs, each node will need to make + * For all events except PPU CYCLEs, each node will need to make * the rtas cbe-perftools call to setup and reset the debug bus. * Make the token lookup call once and store it in the global * variable pm_rtas_token. */ + pm_rtas_token = rtas_token("ibm,cbe-perftools"); + if (unlikely(pm_rtas_token == RTAS_UNKNOWN_SERVICE)) { printk(KERN_ERR "%s: rtas token ibm,cbe-perftools unknown\n", @@ -580,6 +711,58 @@ static int cell_reg_setup(struct op_counter_config *ctr, return -EIO; } + /* setup the pm_control register settings, + * settings will be written per node by the + * cell_cpu_setup() function. + */ + pm_regs.pm_cntrl.trace_buf_ovflw = 1; + + /* Use the occurrence trace mode to have SPU PC saved + * to the trace buffer. Occurrence data in trace buffer + * is not used. Bit 2 must be set to store SPU addresses. + */ + pm_regs.pm_cntrl.trace_mode = 2; + + pm_regs.pm_cntrl.spu_addr_trace = 0x1; /* using debug bus + event 2 & 3 */ + + /* setup the debug bus event array with the SPU PC routing events. + * Note, pm_signal[0] will be filled in by set_pm_event() call below. + */ + pm_signal[1].signal_group = SPU_PROFILE_EVENT_ADDR / 100; + pm_signal[1].bus_word = GET_BUS_WORD(SPU_PROFILE_EVENT_ADDR_MASK_A); + pm_signal[1].bit = SPU_PROFILE_EVENT_ADDR % 100; + pm_signal[1].sub_unit = spu_evnt_phys_spu_indx; + + pm_signal[2].signal_group = SPU_PROFILE_EVENT_ADDR / 100; + pm_signal[2].bus_word = GET_BUS_WORD(SPU_PROFILE_EVENT_ADDR_MASK_B); + pm_signal[2].bit = SPU_PROFILE_EVENT_ADDR % 100; + pm_signal[2].sub_unit = spu_evnt_phys_spu_indx; + + /* Set the user selected spu event to profile on, + * note, only one SPU profiling event is supported + */ + num_counters = 1; /* Only support one SPU event at a time */ + set_pm_event(0, ctr[0].event, ctr[0].unit_mask); + + reset_value[0] = 0xFFFFFFFF - ctr[0].count; + + /* global, used by cell_cpu_setup */ + ctr_enabled |= 1; + + /* Initialize the count for each SPU to the reset value */ + for (i=0; i < MAX_NUMNODES * NUM_SPUS_PER_NODE; i++) + spu_pm_cnt[i] = reset_value[0]; + + return 0; +} + +static int cell_reg_setup_ppu(struct op_counter_config *ctr, + struct op_system_config *sys, int num_ctrs) +{ + /* routine is called once for all nodes */ + int i, j, cpu; + num_counters = num_ctrs; if (unlikely(num_ctrs > NR_PHYS_CTRS)) { @@ -589,14 +772,6 @@ static int cell_reg_setup(struct op_counter_config *ctr, __func__); return -EIO; } - pm_regs.group_control = 0; - pm_regs.debug_bus_control = 0; - - /* setup the pm_control register */ - memset(&pm_regs.pm_cntrl, 0, sizeof(struct pm_cntrl)); - pm_regs.pm_cntrl.stop_at_max = 1; - pm_regs.pm_cntrl.trace_mode = 0; - pm_regs.pm_cntrl.freeze = 1; set_count_mode(sys->enable_kernel, sys->enable_user); @@ -665,6 +840,63 @@ static int cell_reg_setup(struct op_counter_config *ctr, } +/* This function is called once for all cpus combined */ +static int cell_reg_setup(struct op_counter_config *ctr, + struct op_system_config *sys, int num_ctrs) +{ + int ret=0; + spu_cycle_reset = 0; + + /* initialize the spu_arr_trace value, will be reset if + * doing spu event profiling. + */ + pm_regs.group_control = 0; + pm_regs.debug_bus_control = 0; + pm_regs.pm_cntrl.stop_at_max = 1; + pm_regs.pm_cntrl.trace_mode = 0; + pm_regs.pm_cntrl.freeze = 1; + pm_regs.pm_cntrl.trace_buf_ovflw = 0; + pm_regs.pm_cntrl.spu_addr_trace = 0; + + /* + * For all events except PPU CYCLEs, each node will need to make + * the rtas cbe-perftools call to setup and reset the debug bus. + * Make the token lookup call once and store it in the global + * variable pm_rtas_token. + */ + pm_rtas_token = rtas_token("ibm,cbe-perftools"); + + if (unlikely(pm_rtas_token == RTAS_UNKNOWN_SERVICE)) { + printk(KERN_ERR + "%s: rtas token ibm,cbe-perftools unknown\n", + __func__); + return -EIO; + } + + if (ctr[0].event == SPU_CYCLES_EVENT_NUM) { + profiling_mode = SPU_PROFILING_CYCLES; + ret = cell_reg_setup_spu_cycles(ctr, sys, num_ctrs); + } else if ((ctr[0].event >= SPU_EVENT_NUM_START) && + (ctr[0].event <= SPU_EVENT_NUM_STOP)) { + profiling_mode = SPU_PROFILING_EVENTS; + spu_cycle_reset = ctr[0].count; + + /* for SPU event profiling, need to setup the + * pm_signal array with the events to route the + * SPU PC before making the FW call. Note, only + * one SPU event for profiling can be specified + * at a time. + */ + cell_reg_setup_spu_events(ctr, sys, num_ctrs); + } else { + profiling_mode = PPU_PROFILING; + ret = cell_reg_setup_ppu(ctr, sys, num_ctrs); + } + + return ret; +} + + /* This function is called once for each cpu */ static int cell_cpu_setup(struct op_counter_config *cntr) @@ -672,8 +904,13 @@ static int cell_cpu_setup(struct op_counter_config *cntr) u32 cpu = smp_processor_id(); u32 num_enabled = 0; int i; + int ret; - if (spu_cycle_reset) + /* Cycle based SPU profiling does not use the performance + * counters. The trace array is configured to collect + * the data. + */ + if (profiling_mode == SPU_PROFILING_CYCLES) return 0; /* There is one performance monitor per processor chip (i.e. node), @@ -686,7 +923,6 @@ static int cell_cpu_setup(struct op_counter_config *cntr) cbe_disable_pm(cpu); cbe_disable_pm_interrupts(cpu); - cbe_write_pm(cpu, pm_interval, 0); cbe_write_pm(cpu, pm_start_stop, 0); cbe_write_pm(cpu, group_control, pm_regs.group_control); cbe_write_pm(cpu, debug_bus_control, pm_regs.debug_bus_control); @@ -703,7 +939,20 @@ static int cell_cpu_setup(struct op_counter_config *cntr) * The pm_rtas_activate_signals will return -EIO if the FW * call failed. */ - return pm_rtas_activate_signals(cbe_cpu_to_node(cpu), num_enabled); + if (profiling_mode == SPU_PROFILING_EVENTS) { + /* For SPU event profiling also need to setup the + * pm interval timer + */ + ret = pm_rtas_activate_signals(cbe_cpu_to_node(cpu), + num_enabled+2); + /* store PC from debug bus to Trace buffer as often + * as possible (every 10 cycles) + */ + cbe_write_pm(cpu, pm_interval, NUM_INTERVAL_CYC); + return ret; + } else + return pm_rtas_activate_signals(cbe_cpu_to_node(cpu), + num_enabled); } #define ENTRIES 303 @@ -885,7 +1134,122 @@ static struct notifier_block cpu_freq_notifier_block = { }; #endif -static int cell_global_start_spu(struct op_counter_config *ctr) +/* + * Note the generic OProfile stop calls do not support returning + * an error on stop. Hence, will not return an error if the FW + * calls fail on stop. Failure to reset the debug bus is not an issue. + * Failure to disable the SPU profiling is not an issue. The FW calls + * to enable the performance counters and debug bus will work even if + * the hardware was not cleanly reset. + */ +static void cell_global_stop_spu_cycles(void) +{ + int subfunc, rtn_value; + unsigned int lfsr_value; + int cpu; + + oprofile_running = 0; + smp_wmb(); + +#ifdef CONFIG_CPU_FREQ + cpufreq_unregister_notifier(&cpu_freq_notifier_block, + CPUFREQ_TRANSITION_NOTIFIER); +#endif + + for_each_online_cpu(cpu) { + if (cbe_get_hw_thread_id(cpu)) + continue; + + subfunc = 3; /* + * 2 - activate SPU tracing, + * 3 - deactivate + */ + lfsr_value = 0x8f100000; + + rtn_value = rtas_call(spu_rtas_token, 3, 1, NULL, + subfunc, cbe_cpu_to_node(cpu), + lfsr_value); + + if (unlikely(rtn_value != 0)) { + printk(KERN_ERR + "%s: rtas call ibm,cbe-spu-perftools " \ + "failed, return = %d\n", + __func__, rtn_value); + } + + /* Deactivate the signals */ + pm_rtas_reset_signals(cbe_cpu_to_node(cpu)); + } + + stop_spu_profiling_cycles(); +} + +static void cell_global_stop_spu_events(void) +{ + int cpu; + oprofile_running = 0; + + stop_spu_profiling_events(); + smp_wmb(); + + for_each_online_cpu(cpu) { + if (cbe_get_hw_thread_id(cpu)) + continue; + + cbe_sync_irq(cbe_cpu_to_node(cpu)); + /* Stop the counters */ + cbe_disable_pm(cpu); + cbe_write_pm07_control(cpu, 0, 0); + + /* Deactivate the signals */ + pm_rtas_reset_signals(cbe_cpu_to_node(cpu)); + + /* Deactivate interrupts */ + cbe_disable_pm_interrupts(cpu); + } + del_timer_sync(&timer_spu_event_swap); +} + +static void cell_global_stop_ppu(void) +{ + int cpu; + + /* + * This routine will be called once for the system. + * There is one performance monitor per node, so we + * only need to perform this function once per node. + */ + del_timer_sync(&timer_virt_cntr); + oprofile_running = 0; + smp_wmb(); + + for_each_online_cpu(cpu) { + if (cbe_get_hw_thread_id(cpu)) + continue; + + cbe_sync_irq(cbe_cpu_to_node(cpu)); + /* Stop the counters */ + cbe_disable_pm(cpu); + + /* Deactivate the signals */ + pm_rtas_reset_signals(cbe_cpu_to_node(cpu)); + + /* Deactivate interrupts */ + cbe_disable_pm_interrupts(cpu); + } +} + +static void cell_global_stop(void) +{ + if (profiling_mode == PPU_PROFILING) + cell_global_stop_ppu(); + else if (profiling_mode == SPU_PROFILING_EVENTS) + cell_global_stop_spu_events(); + else + cell_global_stop_spu_cycles(); +} + +static int cell_global_start_spu_cycles(struct op_counter_config *ctr) { int subfunc; unsigned int lfsr_value; @@ -951,18 +1315,18 @@ static int cell_global_start_spu(struct op_counter_config *ctr) /* start profiling */ ret = rtas_call(spu_rtas_token, 3, 1, NULL, subfunc, - cbe_cpu_to_node(cpu), lfsr_value); + cbe_cpu_to_node(cpu), lfsr_value); if (unlikely(ret != 0)) { printk(KERN_ERR - "%s: rtas call ibm,cbe-spu-perftools failed, return = %d\n", - __func__, ret); + "%s: rtas call ibm,cbe-spu-perftools failed, " \ + "return = %d\n", __func__, ret); rtas_error = -EIO; goto out; } } - rtas_error = start_spu_profiling(spu_cycle_reset); + rtas_error = start_spu_profiling_cycles(spu_cycle_reset); if (rtas_error) goto out_stop; @@ -970,11 +1334,74 @@ static int cell_global_start_spu(struct op_counter_config *ctr) return 0; out_stop: - cell_global_stop_spu(); /* clean up the PMU/debug bus */ + cell_global_stop_spu_cycles(); /* clean up the PMU/debug bus */ out: return rtas_error; } +static int cell_global_start_spu_events(struct op_counter_config *ctr) +{ + int cpu; + u32 interrupt_mask = 0; + int rtn = 0; + + hdw_thread = 0; + + /* spu event profiling, uses the performance counters to generate + * an interrupt. The hardware is setup to store the SPU program + * counter into the trace array. The occurrence mode is used to + * enable storing data to the trace buffer. The bits are set + * to send/store the SPU address in the trace buffer. The debug + * bus must be setup to route the SPU program counter onto the + * debug bus. The occurrence data in the trace buffer is not used. + */ + + /* This routine gets called once for the system. + * There is one performance monitor per node, so we + * only need to perform this function once per node. + */ + + for_each_online_cpu(cpu) { + if (cbe_get_hw_thread_id(cpu)) + continue; + + /* + * Setup SPU event-based profiling. + * Set perf_mon_control bit 0 to a zero before + * enabling spu collection hardware. + * + * Only support one SPU event on one SPU per node. + */ + if (ctr_enabled & 1) { + cbe_write_ctr(cpu, 0, reset_value[0]); + enable_ctr(cpu, 0, pm_regs.pm07_cntrl); + interrupt_mask |= + CBE_PM_CTR_OVERFLOW_INTR(0); + } else { + /* Disable counter */ + cbe_write_pm07_control(cpu, 0, 0); + } + + cbe_get_and_clear_pm_interrupts(cpu); + cbe_enable_pm_interrupts(cpu, hdw_thread, interrupt_mask); + cbe_enable_pm(cpu); + + /* clear the trace buffer */ + cbe_write_pm(cpu, trace_address, 0); + } + + /* Start the timer to time slice collecting the event profile + * on each of the SPUs. Note, can collect profile on one SPU + * per node at a time. + */ + start_spu_event_swap(); + start_spu_profiling_events(); + oprofile_running = 1; + smp_wmb(); + + return rtn; +} + static int cell_global_start_ppu(struct op_counter_config *ctr) { u32 cpu, i; @@ -994,8 +1421,7 @@ static int cell_global_start_ppu(struct op_counter_config *ctr) if (ctr_enabled & (1 << i)) { cbe_write_ctr(cpu, i, reset_value[i]); enable_ctr(cpu, i, pm_regs.pm07_cntrl); - interrupt_mask |= - CBE_PM_CTR_OVERFLOW_INTR(i); + interrupt_mask |= CBE_PM_CTR_OVERFLOW_INTR(i); } else { /* Disable counter */ cbe_write_pm07_control(cpu, i, 0); @@ -1024,99 +1450,162 @@ static int cell_global_start_ppu(struct op_counter_config *ctr) static int cell_global_start(struct op_counter_config *ctr) { - if (spu_cycle_reset) - return cell_global_start_spu(ctr); + if (profiling_mode == SPU_PROFILING_CYCLES) + return cell_global_start_spu_cycles(ctr); + else if (profiling_mode == SPU_PROFILING_EVENTS) + return cell_global_start_spu_events(ctr); else return cell_global_start_ppu(ctr); } -/* - * Note the generic OProfile stop calls do not support returning - * an error on stop. Hence, will not return an error if the FW - * calls fail on stop. Failure to reset the debug bus is not an issue. - * Failure to disable the SPU profiling is not an issue. The FW calls - * to enable the performance counters and debug bus will work even if - * the hardware was not cleanly reset. + +/* The SPU interrupt handler + * + * SPU event profiling works as follows: + * The pm_signal[0] holds the one SPU event to be measured. It is routed on + * the debug bus using word 0 or 1. The value of pm_signal[1] and + * pm_signal[2] contain the necessary events to route the SPU program + * counter for the selected SPU onto the debug bus using words 2 and 3. + * The pm_interval register is setup to write the SPU PC value into the + * trace buffer at the maximum rate possible. The trace buffer is configured + * to store the PCs, wrapping when it is full. The performance counter is + * intialized to the max hardware count minus the number of events, N, between + * samples. Once the N events have occured, a HW counter overflow occurs + * causing the generation of a HW counter interrupt which also stops the + * writing of the SPU PC values to the trace buffer. Hence the last PC + * written to the trace buffer is the SPU PC that we want. Unfortunately, + * we have to read from the beginning of the trace buffer to get to the + * last value written. We just hope the PPU has nothing better to do then + * service this interrupt. The PC for the specific SPU being profiled is + * extracted from the trace buffer processed and stored. The trace buffer + * is cleared, interrupts are cleared, the counter is reset to max - N. + * A kernel timer is used to periodically call the routine spu_evnt_swap() + * to switch to the next physical SPU in the node to profile in round robbin + * order. This way data is collected for all SPUs on the node. It does mean + * that we need to use a relatively small value of N to ensure enough samples + * on each SPU are collected each SPU is being profiled 1/8 of the time. + * It may also be necessary to use a longer sample collection period. */ -static void cell_global_stop_spu(void) +static void cell_handle_interrupt_spu(struct pt_regs *regs, + struct op_counter_config *ctr) { - int subfunc, rtn_value; - unsigned int lfsr_value; - int cpu; + u32 cpu, cpu_tmp; + u64 trace_entry; + u32 interrupt_mask; + u64 trace_buffer[2]; + u64 last_trace_buffer; + u32 sample; + u32 trace_addr; + unsigned long sample_array_lock_flags; + int spu_num; + unsigned long flags; - oprofile_running = 0; + /* Make sure spu event interrupt handler and spu event swap + * don't access the counters simultaneously. + */ + cpu = smp_processor_id(); + spin_lock_irqsave(&cntr_lock, flags); -#ifdef CONFIG_CPU_FREQ - cpufreq_unregister_notifier(&cpu_freq_notifier_block, - CPUFREQ_TRANSITION_NOTIFIER); -#endif + cpu_tmp = cpu; + cbe_disable_pm(cpu); - for_each_online_cpu(cpu) { - if (cbe_get_hw_thread_id(cpu)) - continue; + interrupt_mask = cbe_get_and_clear_pm_interrupts(cpu); - subfunc = 3; /* - * 2 - activate SPU tracing, - * 3 - deactivate - */ - lfsr_value = 0x8f100000; + sample = 0xABCDEF; + trace_entry = 0xfedcba; + last_trace_buffer = 0xdeadbeaf; - rtn_value = rtas_call(spu_rtas_token, 3, 1, NULL, - subfunc, cbe_cpu_to_node(cpu), - lfsr_value); + if ((oprofile_running == 1) && (interrupt_mask != 0)) { + /* disable writes to trace buff */ + cbe_write_pm(cpu, pm_interval, 0); - if (unlikely(rtn_value != 0)) { - printk(KERN_ERR - "%s: rtas call ibm,cbe-spu-perftools failed, return = %d\n", - __func__, rtn_value); + /* only have one perf cntr being used, cntr 0 */ + if ((interrupt_mask & CBE_PM_CTR_OVERFLOW_INTR(0)) + && ctr[0].enabled) + /* The SPU PC values will be read + * from the trace buffer, reset counter + */ + + cbe_write_ctr(cpu, 0, reset_value[0]); + + trace_addr = cbe_read_pm(cpu, trace_address); + + while (!(trace_addr & CBE_PM_TRACE_BUF_EMPTY)) { + /* There is data in the trace buffer to process + * Read the buffer until you get to the last + * entry. This is the value we want. + */ + + cbe_read_trace_buffer(cpu, trace_buffer); + trace_addr = cbe_read_pm(cpu, trace_address); } - /* Deactivate the signals */ - pm_rtas_reset_signals(cbe_cpu_to_node(cpu)); - } + /* SPU Address 16 bit count format for 128 bit + * HW trace buffer is used for the SPU PC storage + * HDR bits 0:15 + * SPU Addr 0 bits 16:31 + * SPU Addr 1 bits 32:47 + * unused bits 48:127 + * + * HDR: bit4 = 1 SPU Address 0 valid + * HDR: bit5 = 1 SPU Address 1 valid + * - unfortunately, the valid bits don't seem to work + * + * Note trace_buffer[0] holds bits 0:63 of the HW + * trace buffer, trace_buffer[1] holds bits 64:127 + */ - stop_spu_profiling(); -} + trace_entry = trace_buffer[0] + & 0x00000000FFFF0000; -static void cell_global_stop_ppu(void) -{ - int cpu; + /* only top 16 of the 18 bit SPU PC address + * is stored in trace buffer, hence shift right + * by 16 -2 bits */ + sample = trace_entry >> 14; + last_trace_buffer = trace_buffer[0]; - /* - * This routine will be called once for the system. - * There is one performance monitor per node, so we - * only need to perform this function once per node. - */ - del_timer_sync(&timer_virt_cntr); - oprofile_running = 0; - smp_wmb(); + spu_num = spu_evnt_phys_spu_indx + + (cbe_cpu_to_node(cpu) * NUM_SPUS_PER_NODE); - for_each_online_cpu(cpu) { - if (cbe_get_hw_thread_id(cpu)) - continue; + /* make sure only one process at a time is calling + * spu_sync_buffer() + */ + spin_lock_irqsave(&oprof_spu_smpl_arry_lck, + sample_array_lock_flags); + spu_sync_buffer(spu_num, &sample, 1); + spin_unlock_irqrestore(&oprof_spu_smpl_arry_lck, + sample_array_lock_flags); - cbe_sync_irq(cbe_cpu_to_node(cpu)); - /* Stop the counters */ - cbe_disable_pm(cpu); + smp_wmb(); /* insure spu event buffer updates are written + * don't want events intermingled... */ - /* Deactivate the signals */ - pm_rtas_reset_signals(cbe_cpu_to_node(cpu)); + /* The counters were frozen by the interrupt. + * Reenable the interrupt and restart the counters. + */ + cbe_write_pm(cpu, pm_interval, NUM_INTERVAL_CYC); + cbe_enable_pm_interrupts(cpu, hdw_thread, + virt_cntr_inter_mask); - /* Deactivate interrupts */ - cbe_disable_pm_interrupts(cpu); - } -} + /* clear the trace buffer, re-enable writes to trace buff */ + cbe_write_pm(cpu, trace_address, 0); + cbe_write_pm(cpu, pm_interval, NUM_INTERVAL_CYC); -static void cell_global_stop(void) -{ - if (spu_cycle_reset) - cell_global_stop_spu(); - else - cell_global_stop_ppu(); + /* The writes to the various performance counters only writes + * to a latch. The new values (interrupt setting bits, reset + * counter value etc.) are not copied to the actual registers + * until the performance monitor is enabled. In order to get + * this to work as desired, the permormance monitor needs to + * be disabled while writing to the latches. This is a + * HW design issue. + */ + write_pm_cntrl(cpu); + cbe_enable_pm(cpu); + } + spin_unlock_irqrestore(&cntr_lock, flags); } -static void cell_handle_interrupt(struct pt_regs *regs, - struct op_counter_config *ctr) +static void cell_handle_interrupt_ppu(struct pt_regs *regs, + struct op_counter_config *ctr) { u32 cpu; u64 pc; @@ -1132,7 +1621,7 @@ static void cell_handle_interrupt(struct pt_regs *regs, * routine are not running at the same time. See the * cell_virtual_cntr() routine for additional comments. */ - spin_lock_irqsave(&virt_cntr_lock, flags); + spin_lock_irqsave(&cntr_lock, flags); /* * Need to disable and reenable the performance counters @@ -1185,7 +1674,16 @@ static void cell_handle_interrupt(struct pt_regs *regs, */ cbe_enable_pm(cpu); } - spin_unlock_irqrestore(&virt_cntr_lock, flags); + spin_unlock_irqrestore(&cntr_lock, flags); +} + +static void cell_handle_interrupt(struct pt_regs *regs, + struct op_counter_config *ctr) +{ + if (profiling_mode == PPU_PROFILING) + cell_handle_interrupt_ppu(regs, ctr); + else + cell_handle_interrupt_spu(regs, ctr); } /* @@ -1195,7 +1693,8 @@ static void cell_handle_interrupt(struct pt_regs *regs, */ static int cell_sync_start(void) { - if (spu_cycle_reset) + if ((profiling_mode == SPU_PROFILING_CYCLES) || + (profiling_mode == SPU_PROFILING_EVENTS)) return spu_sync_start(); else return DO_GENERIC_SYNC; @@ -1203,7 +1702,8 @@ static int cell_sync_start(void) static int cell_sync_stop(void) { - if (spu_cycle_reset) + if ((profiling_mode == SPU_PROFILING_CYCLES) || + (profiling_mode == SPU_PROFILING_EVENTS)) return spu_sync_stop(); else return 1; diff --git a/arch/x86/include/asm/bitops.h b/arch/x86/include/asm/bitops.h index 9fa9dcdf344..e02a359d2aa 100644 --- a/arch/x86/include/asm/bitops.h +++ b/arch/x86/include/asm/bitops.h @@ -300,7 +300,7 @@ static inline int test_and_change_bit(int nr, volatile unsigned long *addr) return oldbit; } -static inline int constant_test_bit(int nr, const volatile unsigned long *addr) +static inline int constant_test_bit(unsigned int nr, const volatile unsigned long *addr) { return ((1UL << (nr % BITS_PER_LONG)) & (((unsigned long *)addr)[nr / BITS_PER_LONG])) != 0; diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index 29dc0c89d4a..d37593c2f43 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -47,7 +47,7 @@ #endif static int __initdata acpi_force = 0; - +u32 acpi_rsdt_forced; #ifdef CONFIG_ACPI int acpi_disabled = 0; #else @@ -1374,6 +1374,17 @@ static void __init acpi_process_madt(void) "Invalid BIOS MADT, disabling ACPI\n"); disable_acpi(); } + } else { + /* + * ACPI found no MADT, and so ACPI wants UP PIC mode. + * In the event an MPS table was found, forget it. + * Boot with "acpi=off" to use MPS on such a system. + */ + if (smp_found_config) { + printk(KERN_WARNING PREFIX + "No APIC-table, disabling MPS\n"); + smp_found_config = 0; + } } /* @@ -1809,6 +1820,10 @@ static int __init parse_acpi(char *arg) disable_acpi(); acpi_ht = 1; } + /* acpi=rsdt use RSDT instead of XSDT */ + else if (strcmp(arg, "rsdt") == 0) { + acpi_rsdt_forced = 1; + } /* "acpi=noirq" disables ACPI interrupt routing */ else if (strcmp(arg, "noirq") == 0) { acpi_noirq_set(); diff --git a/arch/x86/kernel/acpi/cstate.c b/arch/x86/kernel/acpi/cstate.c index c2502eb9aa8..a4805b3b409 100644 --- a/arch/x86/kernel/acpi/cstate.c +++ b/arch/x86/kernel/acpi/cstate.c @@ -56,6 +56,7 @@ static struct cstate_entry *cpu_cstate_entry; /* per CPU ptr */ static short mwait_supported[ACPI_PROCESSOR_MAX_POWER]; #define MWAIT_SUBSTATE_MASK (0xf) +#define MWAIT_CSTATE_MASK (0xf) #define MWAIT_SUBSTATE_SIZE (4) #define CPUID_MWAIT_LEAF (5) @@ -98,7 +99,8 @@ int acpi_processor_ffh_cstate_probe(unsigned int cpu, cpuid(CPUID_MWAIT_LEAF, &eax, &ebx, &ecx, &edx); /* Check whether this particular cx_type (in CST) is supported or not */ - cstate_type = (cx->address >> MWAIT_SUBSTATE_SIZE) + 1; + cstate_type = ((cx->address >> MWAIT_SUBSTATE_SIZE) & + MWAIT_CSTATE_MASK) + 1; edx_part = edx >> (cstate_type * MWAIT_SUBSTATE_SIZE); num_cstate_subtype = edx_part & MWAIT_SUBSTATE_MASK; diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c index 806b4e9051b..707c1f6f95f 100644 --- a/arch/x86/kernel/acpi/sleep.c +++ b/arch/x86/kernel/acpi/sleep.c @@ -159,6 +159,8 @@ static int __init acpi_sleep_setup(char *str) #endif if (strncmp(str, "old_ordering", 12) == 0) acpi_old_suspend_ordering(); + if (strncmp(str, "s4_nonvs", 8) == 0) + acpi_s4_no_nvs(); str = strchr(str, ','); if (str != NULL) str += strspn(str, ", \t"); diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c index 65a13943e09..e85826829cf 100644 --- a/arch/x86/kernel/e820.c +++ b/arch/x86/kernel/e820.c @@ -665,6 +665,27 @@ void __init e820_mark_nosave_regions(unsigned long limit_pfn) } #endif +#ifdef CONFIG_HIBERNATION +/** + * Mark ACPI NVS memory region, so that we can save/restore it during + * hibernation and the subsequent resume. + */ +static int __init e820_mark_nvs_memory(void) +{ + int i; + + for (i = 0; i < e820.nr_map; i++) { + struct e820entry *ei = &e820.map[i]; + + if (ei->type == E820_NVS) + hibernate_nvs_register(ei->addr, ei->size); + } + + return 0; +} +core_initcall(e820_mark_nvs_memory); +#endif + /* * Early reserved memory areas. */ diff --git a/arch/x86/kernel/early-quirks.c b/arch/x86/kernel/early-quirks.c index 744aa7fc49d..76b8cd953de 100644 --- a/arch/x86/kernel/early-quirks.c +++ b/arch/x86/kernel/early-quirks.c @@ -201,6 +201,12 @@ struct chipset { void (*f)(int num, int slot, int func); }; +/* + * Only works for devices on the root bus. If you add any devices + * not on bus 0 readd another loop level in early_quirks(). But + * be careful because at least the Nvidia quirk here relies on + * only matching on bus 0. + */ static struct chipset early_qrk[] __initdata = { { PCI_VENDOR_ID_NVIDIA, PCI_ANY_ID, PCI_CLASS_BRIDGE_PCI, PCI_ANY_ID, QFLAG_APPLY_ONCE, nvidia_bugs }, @@ -267,17 +273,17 @@ static int __init check_dev_quirk(int num, int slot, int func) void __init early_quirks(void) { - int num, slot, func; + int slot, func; if (!early_pci_allowed()) return; /* Poor man's PCI discovery */ - for (num = 0; num < 32; num++) - for (slot = 0; slot < 32; slot++) - for (func = 0; func < 8; func++) { - /* Only probe function 0 on single fn devices */ - if (check_dev_quirk(num, slot, func)) - break; - } + /* Only scan the root bus */ + for (slot = 0; slot < 32; slot++) + for (func = 0; func < 8; func++) { + /* Only probe function 0 on single fn devices */ + if (check_dev_quirk(0, slot, func)) + break; + } } diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c index 98658f25f54..8fdf06e4edf 100644 --- a/arch/x86/oprofile/op_model_amd.c +++ b/arch/x86/oprofile/op_model_amd.c @@ -2,7 +2,7 @@ * @file op_model_amd.c * athlon / K7 / K8 / Family 10h model-specific MSR operations * - * @remark Copyright 2002-2008 OProfile authors + * @remark Copyright 2002-2009 OProfile authors * @remark Read the file COPYING * * @author John Levon @@ -10,7 +10,7 @@ * @author Graydon Hoare * @author Robert Richter <robert.richter@amd.com> * @author Barry Kasindorf -*/ + */ #include <linux/oprofile.h> #include <linux/device.h> @@ -60,53 +60,10 @@ static unsigned long reset_value[NUM_COUNTERS]; #define IBS_OP_LOW_VALID_BIT (1ULL<<18) /* bit 18 */ #define IBS_OP_LOW_ENABLE (1ULL<<17) /* bit 17 */ -/* Codes used in cpu_buffer.c */ -/* This produces duplicate code, need to be fixed */ -#define IBS_FETCH_BEGIN 3 -#define IBS_OP_BEGIN 4 - -/* - * The function interface needs to be fixed, something like add - * data. Should then be added to linux/oprofile.h. - */ -extern void -oprofile_add_ibs_sample(struct pt_regs * const regs, - unsigned int * const ibs_sample, int ibs_code); - -struct ibs_fetch_sample { - /* MSRC001_1031 IBS Fetch Linear Address Register */ - unsigned int ibs_fetch_lin_addr_low; - unsigned int ibs_fetch_lin_addr_high; - /* MSRC001_1030 IBS Fetch Control Register */ - unsigned int ibs_fetch_ctl_low; - unsigned int ibs_fetch_ctl_high; - /* MSRC001_1032 IBS Fetch Physical Address Register */ - unsigned int ibs_fetch_phys_addr_low; - unsigned int ibs_fetch_phys_addr_high; -}; - -struct ibs_op_sample { - /* MSRC001_1034 IBS Op Logical Address Register (IbsRIP) */ - unsigned int ibs_op_rip_low; - unsigned int ibs_op_rip_high; - /* MSRC001_1035 IBS Op Data Register */ - unsigned int ibs_op_data1_low; - unsigned int ibs_op_data1_high; - /* MSRC001_1036 IBS Op Data 2 Register */ - unsigned int ibs_op_data2_low; - unsigned int ibs_op_data2_high; - /* MSRC001_1037 IBS Op Data 3 Register */ - unsigned int ibs_op_data3_low; - unsigned int ibs_op_data3_high; - /* MSRC001_1038 IBS DC Linear Address Register (IbsDcLinAd) */ - unsigned int ibs_dc_linear_low; - unsigned int ibs_dc_linear_high; - /* MSRC001_1039 IBS DC Physical Address Register (IbsDcPhysAd) */ - unsigned int ibs_dc_phys_low; - unsigned int ibs_dc_phys_high; -}; +#define IBS_FETCH_SIZE 6 +#define IBS_OP_SIZE 12 -static int ibs_allowed; /* AMD Family10h and later */ +static int has_ibs; /* AMD Family10h and later */ struct op_ibs_config { unsigned long op_enabled; @@ -197,31 +154,29 @@ static inline int op_amd_handle_ibs(struct pt_regs * const regs, struct op_msrs const * const msrs) { - unsigned int low, high; - struct ibs_fetch_sample ibs_fetch; - struct ibs_op_sample ibs_op; + u32 low, high; + u64 msr; + struct op_entry entry; - if (!ibs_allowed) + if (!has_ibs) return 1; if (ibs_config.fetch_enabled) { rdmsr(MSR_AMD64_IBSFETCHCTL, low, high); if (high & IBS_FETCH_HIGH_VALID_BIT) { - ibs_fetch.ibs_fetch_ctl_high = high; - ibs_fetch.ibs_fetch_ctl_low = low; - rdmsr(MSR_AMD64_IBSFETCHLINAD, low, high); - ibs_fetch.ibs_fetch_lin_addr_high = high; - ibs_fetch.ibs_fetch_lin_addr_low = low; - rdmsr(MSR_AMD64_IBSFETCHPHYSAD, low, high); - ibs_fetch.ibs_fetch_phys_addr_high = high; - ibs_fetch.ibs_fetch_phys_addr_low = low; - - oprofile_add_ibs_sample(regs, - (unsigned int *)&ibs_fetch, - IBS_FETCH_BEGIN); + rdmsrl(MSR_AMD64_IBSFETCHLINAD, msr); + oprofile_write_reserve(&entry, regs, msr, + IBS_FETCH_CODE, IBS_FETCH_SIZE); + oprofile_add_data(&entry, (u32)msr); + oprofile_add_data(&entry, (u32)(msr >> 32)); + oprofile_add_data(&entry, low); + oprofile_add_data(&entry, high); + rdmsrl(MSR_AMD64_IBSFETCHPHYSAD, msr); + oprofile_add_data(&entry, (u32)msr); + oprofile_add_data(&entry, (u32)(msr >> 32)); + oprofile_write_commit(&entry); /* reenable the IRQ */ - rdmsr(MSR_AMD64_IBSFETCHCTL, low, high); high &= ~IBS_FETCH_HIGH_VALID_BIT; high |= IBS_FETCH_HIGH_ENABLE; low &= IBS_FETCH_LOW_MAX_CNT_MASK; @@ -232,30 +187,29 @@ op_amd_handle_ibs(struct pt_regs * const regs, if (ibs_config.op_enabled) { rdmsr(MSR_AMD64_IBSOPCTL, low, high); if (low & IBS_OP_LOW_VALID_BIT) { - rdmsr(MSR_AMD64_IBSOPRIP, low, high); - ibs_op.ibs_op_rip_low = low; - ibs_op.ibs_op_rip_high = high; - rdmsr(MSR_AMD64_IBSOPDATA, low, high); - ibs_op.ibs_op_data1_low = low; - ibs_op.ibs_op_data1_high = high; - rdmsr(MSR_AMD64_IBSOPDATA2, low, high); - ibs_op.ibs_op_data2_low = low; - ibs_op.ibs_op_data2_high = high; - rdmsr(MSR_AMD64_IBSOPDATA3, low, high); - ibs_op.ibs_op_data3_low = low; - ibs_op.ibs_op_data3_high = high; - rdmsr(MSR_AMD64_IBSDCLINAD, low, high); - ibs_op.ibs_dc_linear_low = low; - ibs_op.ibs_dc_linear_high = high; - rdmsr(MSR_AMD64_IBSDCPHYSAD, low, high); - ibs_op.ibs_dc_phys_low = low; - ibs_op.ibs_dc_phys_high = high; + rdmsrl(MSR_AMD64_IBSOPRIP, msr); + oprofile_write_reserve(&entry, regs, msr, + IBS_OP_CODE, IBS_OP_SIZE); + oprofile_add_data(&entry, (u32)msr); + oprofile_add_data(&entry, (u32)(msr >> 32)); + rdmsrl(MSR_AMD64_IBSOPDATA, msr); + oprofile_add_data(&entry, (u32)msr); + oprofile_add_data(&entry, (u32)(msr >> 32)); + rdmsrl(MSR_AMD64_IBSOPDATA2, msr); + oprofile_add_data(&entry, (u32)msr); + oprofile_add_data(&entry, (u32)(msr >> 32)); + rdmsrl(MSR_AMD64_IBSOPDATA3, msr); + oprofile_add_data(&entry, (u32)msr); + oprofile_add_data(&entry, (u32)(msr >> 32)); + rdmsrl(MSR_AMD64_IBSDCLINAD, msr); + oprofile_add_data(&entry, (u32)msr); + oprofile_add_data(&entry, (u32)(msr >> 32)); + rdmsrl(MSR_AMD64_IBSDCPHYSAD, msr); + oprofile_add_data(&entry, (u32)msr); + oprofile_add_data(&entry, (u32)(msr >> 32)); + oprofile_write_commit(&entry); /* reenable the IRQ */ - oprofile_add_ibs_sample(regs, - (unsigned int *)&ibs_op, - IBS_OP_BEGIN); - rdmsr(MSR_AMD64_IBSOPCTL, low, high); high = 0; low &= ~IBS_OP_LOW_VALID_BIT; low |= IBS_OP_LOW_ENABLE; @@ -305,14 +259,14 @@ static void op_amd_start(struct op_msrs const * const msrs) } #ifdef CONFIG_OPROFILE_IBS - if (ibs_allowed && ibs_config.fetch_enabled) { + if (has_ibs && ibs_config.fetch_enabled) { low = (ibs_config.max_cnt_fetch >> 4) & 0xFFFF; high = ((ibs_config.rand_en & 0x1) << 25) /* bit 57 */ + IBS_FETCH_HIGH_ENABLE; wrmsr(MSR_AMD64_IBSFETCHCTL, low, high); } - if (ibs_allowed && ibs_config.op_enabled) { + if (has_ibs && ibs_config.op_enabled) { low = ((ibs_config.max_cnt_op >> 4) & 0xFFFF) + ((ibs_config.dispatched_ops & 0x1) << 19) /* bit 19 */ + IBS_OP_LOW_ENABLE; @@ -341,14 +295,14 @@ static void op_amd_stop(struct op_msrs const * const msrs) } #ifdef CONFIG_OPROFILE_IBS - if (ibs_allowed && ibs_config.fetch_enabled) { + if (has_ibs && ibs_config.fetch_enabled) { /* clear max count and enable */ low = 0; high = 0; wrmsr(MSR_AMD64_IBSFETCHCTL, low, high); } - if (ibs_allowed && ibs_config.op_enabled) { + if (has_ibs && ibs_config.op_enabled) { /* clear max count and enable */ low = 0; high = 0; @@ -409,6 +363,7 @@ static int init_ibs_nmi(void) | IBSCTL_LVTOFFSETVAL); pci_read_config_dword(cpu_cfg, IBSCTL, &value); if (value != (ibs_eilvt_off | IBSCTL_LVTOFFSETVAL)) { + pci_dev_put(cpu_cfg); printk(KERN_DEBUG "Failed to setup IBS LVT offset, " "IBSCTL = 0x%08x", value); return 1; @@ -436,20 +391,20 @@ static int init_ibs_nmi(void) /* uninitialize the APIC for the IBS interrupts if needed */ static void clear_ibs_nmi(void) { - if (ibs_allowed) + if (has_ibs) on_each_cpu(apic_clear_ibs_nmi_per_cpu, NULL, 1); } /* initialize the APIC for the IBS interrupts if available */ static void ibs_init(void) { - ibs_allowed = boot_cpu_has(X86_FEATURE_IBS); + has_ibs = boot_cpu_has(X86_FEATURE_IBS); - if (!ibs_allowed) + if (!has_ibs) return; if (init_ibs_nmi()) { - ibs_allowed = 0; + has_ibs = 0; return; } @@ -458,7 +413,7 @@ static void ibs_init(void) static void ibs_exit(void) { - if (!ibs_allowed) + if (!has_ibs) return; clear_ibs_nmi(); @@ -478,7 +433,7 @@ static int setup_ibs_files(struct super_block *sb, struct dentry *root) if (ret) return ret; - if (!ibs_allowed) + if (!has_ibs) return ret; /* model specific files */ |