diff options
Diffstat (limited to 'arch/sparc/kernel')
-rw-r--r-- | arch/sparc/kernel/Makefile | 5 | ||||
-rw-r--r-- | arch/sparc/kernel/cpu.c | 5 | ||||
-rw-r--r-- | arch/sparc/kernel/head_32.S | 5 | ||||
-rw-r--r-- | arch/sparc/kernel/idprom.c | 2 | ||||
-rw-r--r-- | arch/sparc/kernel/ioport.c | 32 | ||||
-rw-r--r-- | arch/sparc/kernel/irq_32.c | 5 | ||||
-rw-r--r-- | arch/sparc/kernel/leon_kernel.c | 203 | ||||
-rw-r--r-- | arch/sparc/kernel/nmi.c | 72 | ||||
-rw-r--r-- | arch/sparc/kernel/of_device_32.c | 40 | ||||
-rw-r--r-- | arch/sparc/kernel/pcr.c | 14 | ||||
-rw-r--r-- | arch/sparc/kernel/perf_counter.c | 557 | ||||
-rw-r--r-- | arch/sparc/kernel/prom_32.c | 33 | ||||
-rw-r--r-- | arch/sparc/kernel/prom_common.c | 10 | ||||
-rw-r--r-- | arch/sparc/kernel/setup_32.c | 5 | ||||
-rw-r--r-- | arch/sparc/kernel/sysfs.c | 1 | ||||
-rw-r--r-- | arch/sparc/kernel/systbls_32.S | 2 | ||||
-rw-r--r-- | arch/sparc/kernel/systbls_64.S | 4 |
17 files changed, 958 insertions, 37 deletions
diff --git a/arch/sparc/kernel/Makefile b/arch/sparc/kernel/Makefile index 29b88a58066..247cc620cee 100644 --- a/arch/sparc/kernel/Makefile +++ b/arch/sparc/kernel/Makefile @@ -41,6 +41,8 @@ obj-y += of_device_common.o obj-y += of_device_$(BITS).o obj-$(CONFIG_SPARC64) += prom_irqtrans.o +obj-$(CONFIG_SPARC_LEON)+= leon_kernel.o + obj-$(CONFIG_SPARC64) += reboot.o obj-$(CONFIG_SPARC64) += sysfs.o obj-$(CONFIG_SPARC64) += iommu.o @@ -101,3 +103,6 @@ obj-$(CONFIG_SUN_LDOMS) += ldc.o vio.o viohs.o ds.o obj-$(CONFIG_AUDIT) += audit.o audit--$(CONFIG_AUDIT) := compat_audit.o obj-$(CONFIG_COMPAT) += $(audit--y) + +pc--$(CONFIG_PERF_COUNTERS) := perf_counter.o +obj-$(CONFIG_SPARC64) += $(pc--y) diff --git a/arch/sparc/kernel/cpu.c b/arch/sparc/kernel/cpu.c index d85c3dc4953..1446df90ef8 100644 --- a/arch/sparc/kernel/cpu.c +++ b/arch/sparc/kernel/cpu.c @@ -312,7 +312,12 @@ void __cpuinit cpu_probe(void) psr = get_psr(); put_psr(psr | PSR_EF); +#ifdef CONFIG_SPARC_LEON + fpu_vers = 7; +#else fpu_vers = ((get_fsr() >> 17) & 0x7); +#endif + put_psr(psr); set_cpu_and_fpu(psr_impl, psr_vers, fpu_vers); diff --git a/arch/sparc/kernel/head_32.S b/arch/sparc/kernel/head_32.S index 6b4d8acc4c8..439d82a95ac 100644 --- a/arch/sparc/kernel/head_32.S +++ b/arch/sparc/kernel/head_32.S @@ -809,6 +809,11 @@ found_version: nop got_prop: +#ifdef CONFIG_SPARC_LEON + /* no cpu-type check is needed, it is a SPARC-LEON */ + ba sun4c_continue_boot + nop +#endif set cputypval, %o2 ldub [%o2 + 0x4], %l1 diff --git a/arch/sparc/kernel/idprom.c b/arch/sparc/kernel/idprom.c index 57922f69c3f..52a15fe2db1 100644 --- a/arch/sparc/kernel/idprom.c +++ b/arch/sparc/kernel/idprom.c @@ -31,6 +31,8 @@ static struct Sun_Machine_Models Sun_Machines[NUM_SUN_MACHINES] = { { .name = "Sun 4/200 Series", .id_machtype = (SM_SUN4 | SM_4_260) }, { .name = "Sun 4/300 Series", .id_machtype = (SM_SUN4 | SM_4_330) }, { .name = "Sun 4/400 Series", .id_machtype = (SM_SUN4 | SM_4_470) }, +/* Now Leon */ +{ .name = "Leon3 System-on-a-Chip", .id_machtype = (M_LEON | M_LEON3_SOC) }, /* Now, Sun4c's */ { .name = "Sun4c SparcStation 1", .id_machtype = (SM_SUN4C | SM_4C_SS1) }, { .name = "Sun4c SparcStation IPC", .id_machtype = (SM_SUN4C | SM_4C_IPC) }, diff --git a/arch/sparc/kernel/ioport.c b/arch/sparc/kernel/ioport.c index edbea232c61..9f61fd8cbb7 100644 --- a/arch/sparc/kernel/ioport.c +++ b/arch/sparc/kernel/ioport.c @@ -35,6 +35,7 @@ #include <linux/slab.h> #include <linux/pci.h> /* struct pci_dev */ #include <linux/proc_fs.h> +#include <linux/seq_file.h> #include <linux/scatterlist.h> #include <linux/of_device.h> @@ -683,26 +684,33 @@ EXPORT_SYMBOL(dma_set_mask); #ifdef CONFIG_PROC_FS -static int -_sparc_io_get_info(char *buf, char **start, off_t fpos, int length, int *eof, - void *data) +static int sparc_io_proc_show(struct seq_file *m, void *v) { - char *p = buf, *e = buf + length; - struct resource *r; + struct resource *root = m->private, *r; const char *nm; - for (r = ((struct resource *)data)->child; r != NULL; r = r->sibling) { - if (p + 32 >= e) /* Better than nothing */ - break; + for (r = root->child; r != NULL; r = r->sibling) { if ((nm = r->name) == 0) nm = "???"; - p += sprintf(p, "%016llx-%016llx: %s\n", + seq_printf(m, "%016llx-%016llx: %s\n", (unsigned long long)r->start, (unsigned long long)r->end, nm); } - return p-buf; + return 0; } +static int sparc_io_proc_open(struct inode *inode, struct file *file) +{ + return single_open(file, sparc_io_proc_show, PDE(inode)->data); +} + +static const struct file_operations sparc_io_proc_fops = { + .owner = THIS_MODULE, + .open = sparc_io_proc_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; #endif /* CONFIG_PROC_FS */ /* @@ -727,7 +735,7 @@ static struct resource *_sparc_find_resource(struct resource *root, static void register_proc_sparc_ioport(void) { #ifdef CONFIG_PROC_FS - create_proc_read_entry("io_map",0,NULL,_sparc_io_get_info,&sparc_iomap); - create_proc_read_entry("dvma_map",0,NULL,_sparc_io_get_info,&_sparc_dvma); + proc_create_data("io_map", 0, NULL, &sparc_io_proc_fops, &sparc_iomap); + proc_create_data("dvma_map", 0, NULL, &sparc_io_proc_fops, &_sparc_dvma); #endif } diff --git a/arch/sparc/kernel/irq_32.c b/arch/sparc/kernel/irq_32.c index ad800b80c71..e1af4372832 100644 --- a/arch/sparc/kernel/irq_32.c +++ b/arch/sparc/kernel/irq_32.c @@ -45,6 +45,7 @@ #include <asm/pcic.h> #include <asm/cacheflush.h> #include <asm/irq_regs.h> +#include <asm/leon.h> #include "kernel.h" #include "irq.h" @@ -661,6 +662,10 @@ void __init init_IRQ(void) sun4d_init_IRQ(); break; + case sparc_leon: + leon_init_IRQ(); + break; + default: prom_printf("Cannot initialize IRQs on this Sun machine..."); break; diff --git a/arch/sparc/kernel/leon_kernel.c b/arch/sparc/kernel/leon_kernel.c new file mode 100644 index 00000000000..54d8a5bd482 --- /dev/null +++ b/arch/sparc/kernel/leon_kernel.c @@ -0,0 +1,203 @@ +/* + * Copyright (C) 2009 Daniel Hellstrom (daniel@gaisler.com) Aeroflex Gaisler AB + * Copyright (C) 2009 Konrad Eisele (konrad@gaisler.com) Aeroflex Gaisler AB + */ + +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/errno.h> +#include <linux/mutex.h> +#include <linux/slab.h> +#include <linux/of.h> +#include <linux/of_platform.h> +#include <linux/interrupt.h> +#include <linux/of_device.h> +#include <asm/oplib.h> +#include <asm/timer.h> +#include <asm/prom.h> +#include <asm/leon.h> +#include <asm/leon_amba.h> + +#include "prom.h" +#include "irq.h" + +struct leon3_irqctrl_regs_map *leon3_irqctrl_regs; /* interrupt controller base address, initialized by amba_init() */ +struct leon3_gptimer_regs_map *leon3_gptimer_regs; /* timer controller base address, initialized by amba_init() */ +struct amba_apb_device leon_percpu_timer_dev[16]; + +int leondebug_irq_disable; +int leon_debug_irqout; +static int dummy_master_l10_counter; + +unsigned long leon3_gptimer_irq; /* interrupt controller irq number, initialized by amba_init() */ +unsigned int sparc_leon_eirq; +#define LEON_IMASK ((&leon3_irqctrl_regs->mask[0])) + +/* Return the IRQ of the pending IRQ on the extended IRQ controller */ +int sparc_leon_eirq_get(int eirq, int cpu) +{ + return LEON3_BYPASS_LOAD_PA(&leon3_irqctrl_regs->intid[cpu]) & 0x1f; +} + +irqreturn_t sparc_leon_eirq_isr(int dummy, void *dev_id) +{ + printk(KERN_ERR "sparc_leon_eirq_isr: ERROR EXTENDED IRQ\n"); + return IRQ_HANDLED; +} + +/* The extended IRQ controller has been found, this function registers it */ +void sparc_leon_eirq_register(int eirq) +{ + int irq; + + /* Register a "BAD" handler for this interrupt, it should never happen */ + irq = request_irq(eirq, sparc_leon_eirq_isr, + (IRQF_DISABLED | SA_STATIC_ALLOC), "extirq", NULL); + + if (irq) { + printk(KERN_ERR + "sparc_leon_eirq_register: unable to attach IRQ%d\n", + eirq); + } else { + sparc_leon_eirq = eirq; + } + +} + +static inline unsigned long get_irqmask(unsigned int irq) +{ + unsigned long mask; + + if (!irq || ((irq > 0xf) && !sparc_leon_eirq) + || ((irq > 0x1f) && sparc_leon_eirq)) { + printk(KERN_ERR + "leon_get_irqmask: false irq number: %d\n", irq); + mask = 0; + } else { + mask = LEON_HARD_INT(irq); + } + return mask; +} + +static void leon_enable_irq(unsigned int irq_nr) +{ + unsigned long mask, flags; + mask = get_irqmask(irq_nr); + local_irq_save(flags); + LEON3_BYPASS_STORE_PA(LEON_IMASK, + (LEON3_BYPASS_LOAD_PA(LEON_IMASK) | (mask))); + local_irq_restore(flags); +} + +static void leon_disable_irq(unsigned int irq_nr) +{ + unsigned long mask, flags; + mask = get_irqmask(irq_nr); + local_irq_save(flags); + LEON3_BYPASS_STORE_PA(LEON_IMASK, + (LEON3_BYPASS_LOAD_PA(LEON_IMASK) & ~(mask))); + local_irq_restore(flags); + +} + +void __init leon_init_timers(irq_handler_t counter_fn) +{ + int irq; + + leondebug_irq_disable = 0; + leon_debug_irqout = 0; + master_l10_counter = (unsigned int *)&dummy_master_l10_counter; + dummy_master_l10_counter = 0; + + if (leon3_gptimer_regs && leon3_irqctrl_regs) { + LEON3_BYPASS_STORE_PA(&leon3_gptimer_regs->e[0].val, 0); + LEON3_BYPASS_STORE_PA(&leon3_gptimer_regs->e[0].rld, + (((1000000 / 100) - 1))); + LEON3_BYPASS_STORE_PA(&leon3_gptimer_regs->e[0].ctrl, 0); + + } else { + printk(KERN_ERR "No Timer/irqctrl found\n"); + BUG(); + } + + irq = request_irq(leon3_gptimer_irq, + counter_fn, + (IRQF_DISABLED | SA_STATIC_ALLOC), "timer", NULL); + + if (irq) { + printk(KERN_ERR "leon_time_init: unable to attach IRQ%d\n", + LEON_INTERRUPT_TIMER1); + prom_halt(); + } + + if (leon3_gptimer_regs) { + LEON3_BYPASS_STORE_PA(&leon3_gptimer_regs->e[0].ctrl, + LEON3_GPTIMER_EN | + LEON3_GPTIMER_RL | + LEON3_GPTIMER_LD | LEON3_GPTIMER_IRQEN); + } +} + +void leon_clear_clock_irq(void) +{ +} + +void leon_load_profile_irq(int cpu, unsigned int limit) +{ + BUG(); +} + + + + +void __init leon_trans_init(struct device_node *dp) +{ + if (strcmp(dp->type, "cpu") == 0 && strcmp(dp->name, "<NULL>") == 0) { + struct property *p; + p = of_find_property(dp, "mid", (void *)0); + if (p) { + int mid; + dp->name = prom_early_alloc(5 + 1); + memcpy(&mid, p->value, p->length); + sprintf((char *)dp->name, "cpu%.2d", mid); + } + } +} + +void __initdata (*prom_amba_init)(struct device_node *dp, struct device_node ***nextp) = 0; + +void __init leon_node_init(struct device_node *dp, struct device_node ***nextp) +{ + if (prom_amba_init && + strcmp(dp->type, "ambapp") == 0 && + strcmp(dp->name, "ambapp0") == 0) { + prom_amba_init(dp, nextp); + } +} + +void __init leon_init_IRQ(void) +{ + sparc_init_timers = leon_init_timers; + + BTFIXUPSET_CALL(enable_irq, leon_enable_irq, BTFIXUPCALL_NORM); + BTFIXUPSET_CALL(disable_irq, leon_disable_irq, BTFIXUPCALL_NORM); + BTFIXUPSET_CALL(enable_pil_irq, leon_enable_irq, BTFIXUPCALL_NORM); + BTFIXUPSET_CALL(disable_pil_irq, leon_disable_irq, BTFIXUPCALL_NORM); + + BTFIXUPSET_CALL(clear_clock_irq, leon_clear_clock_irq, + BTFIXUPCALL_NORM); + BTFIXUPSET_CALL(load_profile_irq, leon_load_profile_irq, + BTFIXUPCALL_NOP); + +#ifdef CONFIG_SMP + BTFIXUPSET_CALL(set_cpu_int, leon_set_cpu_int, BTFIXUPCALL_NORM); + BTFIXUPSET_CALL(clear_cpu_int, leon_clear_ipi, BTFIXUPCALL_NORM); + BTFIXUPSET_CALL(set_irq_udt, leon_set_udt, BTFIXUPCALL_NORM); +#endif + +} + +void __init leon_init(void) +{ + prom_build_more = &leon_node_init; +} diff --git a/arch/sparc/kernel/nmi.c b/arch/sparc/kernel/nmi.c index b75bf502cd4..378eb53e077 100644 --- a/arch/sparc/kernel/nmi.c +++ b/arch/sparc/kernel/nmi.c @@ -19,6 +19,7 @@ #include <linux/delay.h> #include <linux/smp.h> +#include <asm/perf_counter.h> #include <asm/ptrace.h> #include <asm/local.h> #include <asm/pcr.h> @@ -31,13 +32,19 @@ * level 14 as our IRQ off level. */ -static int nmi_watchdog_active; static int panic_on_timeout; -int nmi_usable; -EXPORT_SYMBOL_GPL(nmi_usable); +/* nmi_active: + * >0: the NMI watchdog is active, but can be disabled + * <0: the NMI watchdog has not been set up, and cannot be enabled + * 0: the NMI watchdog is disabled, but can be enabled + */ +atomic_t nmi_active = ATOMIC_INIT(0); /* oprofile uses this */ +EXPORT_SYMBOL(nmi_active); static unsigned int nmi_hz = HZ; +static DEFINE_PER_CPU(short, wd_enabled); +static int endflag __initdata; static DEFINE_PER_CPU(unsigned int, last_irq_sum); static DEFINE_PER_CPU(local_t, alert_counter); @@ -45,7 +52,7 @@ static DEFINE_PER_CPU(int, nmi_touch); void touch_nmi_watchdog(void) { - if (nmi_watchdog_active) { + if (atomic_read(&nmi_active)) { int cpu; for_each_present_cpu(cpu) { @@ -78,6 +85,7 @@ static void die_nmi(const char *str, struct pt_regs *regs, int do_panic) if (do_panic || panic_on_oops) panic("Non maskable interrupt"); + nmi_exit(); local_irq_enable(); do_exit(SIGBUS); } @@ -92,6 +100,8 @@ notrace __kprobes void perfctr_irq(int irq, struct pt_regs *regs) local_cpu_data().__nmi_count++; + nmi_enter(); + if (notify_die(DIE_NMI, "nmi", regs, 0, pt_regs_trap_type(regs), SIGINT) == NOTIFY_STOP) touched = 1; @@ -110,10 +120,12 @@ notrace __kprobes void perfctr_irq(int irq, struct pt_regs *regs) __get_cpu_var(last_irq_sum) = sum; local_set(&__get_cpu_var(alert_counter), 0); } - if (nmi_usable) { + if (__get_cpu_var(wd_enabled)) { write_pic(picl_value(nmi_hz)); pcr_ops->write(pcr_enable); } + + nmi_exit(); } static inline unsigned int get_nmi_count(int cpu) @@ -121,8 +133,6 @@ static inline unsigned int get_nmi_count(int cpu) return cpu_data(cpu).__nmi_count; } -static int endflag __initdata; - static __init void nmi_cpu_busy(void *data) { local_irq_enable_in_hardirq(); @@ -143,12 +153,15 @@ static void report_broken_nmi(int cpu, int *prev_nmi_count) printk(KERN_WARNING "and attach the output of the 'dmesg' command.\n"); - nmi_usable = 0; + per_cpu(wd_enabled, cpu) = 0; + atomic_dec(&nmi_active); } -static void stop_watchdog(void *unused) +void stop_nmi_watchdog(void *unused) { pcr_ops->write(PCR_PIC_PRIV); + __get_cpu_var(wd_enabled) = 0; + atomic_dec(&nmi_active); } static int __init check_nmi_watchdog(void) @@ -156,6 +169,9 @@ static int __init check_nmi_watchdog(void) unsigned int *prev_nmi_count; int cpu, err; + if (!atomic_read(&nmi_active)) + return 0; + prev_nmi_count = kmalloc(nr_cpu_ids * sizeof(unsigned int), GFP_KERNEL); if (!prev_nmi_count) { err = -ENOMEM; @@ -172,12 +188,15 @@ static int __init check_nmi_watchdog(void) mdelay((20 * 1000) / nmi_hz); /* wait 20 ticks */ for_each_online_cpu(cpu) { + if (!per_cpu(wd_enabled, cpu)) + continue; if (get_nmi_count(cpu) - prev_nmi_count[cpu] <= 5) report_broken_nmi(cpu, prev_nmi_count); } endflag = 1; - if (!nmi_usable) { + if (!atomic_read(&nmi_active)) { kfree(prev_nmi_count); + atomic_set(&nmi_active, -1); err = -ENODEV; goto error; } @@ -188,12 +207,26 @@ static int __init check_nmi_watchdog(void) kfree(prev_nmi_count); return 0; error: - on_each_cpu(stop_watchdog, NULL, 1); + on_each_cpu(stop_nmi_watchdog, NULL, 1); return err; } -static void start_watchdog(void *unused) +void start_nmi_watchdog(void *unused) { + __get_cpu_var(wd_enabled) = 1; + atomic_inc(&nmi_active); + + pcr_ops->write(PCR_PIC_PRIV); + write_pic(picl_value(nmi_hz)); + + pcr_ops->write(pcr_enable); +} + +static void nmi_adjust_hz_one(void *unused) +{ + if (!__get_cpu_var(wd_enabled)) + return; + pcr_ops->write(PCR_PIC_PRIV); write_pic(picl_value(nmi_hz)); @@ -203,13 +236,13 @@ static void start_watchdog(void *unused) void nmi_adjust_hz(unsigned int new_hz) { nmi_hz = new_hz; - on_each_cpu(start_watchdog, NULL, 1); + on_each_cpu(nmi_adjust_hz_one, NULL, 1); } EXPORT_SYMBOL_GPL(nmi_adjust_hz); static int nmi_shutdown(struct notifier_block *nb, unsigned long cmd, void *p) { - on_each_cpu(stop_watchdog, NULL, 1); + on_each_cpu(stop_nmi_watchdog, NULL, 1); return 0; } @@ -221,18 +254,19 @@ int __init nmi_init(void) { int err; - nmi_usable = 1; - - on_each_cpu(start_watchdog, NULL, 1); + on_each_cpu(start_nmi_watchdog, NULL, 1); err = check_nmi_watchdog(); if (!err) { err = register_reboot_notifier(&nmi_reboot_notifier); if (err) { - nmi_usable = 0; - on_each_cpu(stop_watchdog, NULL, 1); + on_each_cpu(stop_nmi_watchdog, NULL, 1); + atomic_set(&nmi_active, -1); } } + if (!err) + init_hw_perf_counters(); + return err; } diff --git a/arch/sparc/kernel/of_device_32.c b/arch/sparc/kernel/of_device_32.c index 90396702ea2..4c26eb59e74 100644 --- a/arch/sparc/kernel/of_device_32.c +++ b/arch/sparc/kernel/of_device_32.c @@ -9,6 +9,8 @@ #include <linux/irq.h> #include <linux/of_device.h> #include <linux/of_platform.h> +#include <asm/leon.h> +#include <asm/leon_amba.h> #include "of_device_common.h" @@ -97,6 +99,35 @@ static unsigned long of_bus_sbus_get_flags(const u32 *addr, unsigned long flags) return IORESOURCE_MEM; } + /* + * AMBAPP bus specific translator + */ + +static int of_bus_ambapp_match(struct device_node *np) +{ + return !strcmp(np->name, "ambapp"); +} + +static void of_bus_ambapp_count_cells(struct device_node *child, + int *addrc, int *sizec) +{ + if (addrc) + *addrc = 1; + if (sizec) + *sizec = 1; +} + +static int of_bus_ambapp_map(u32 *addr, const u32 *range, + int na, int ns, int pna) +{ + return of_bus_default_map(addr, range, na, ns, pna); +} + +static unsigned long of_bus_ambapp_get_flags(const u32 *addr, + unsigned long flags) +{ + return IORESOURCE_MEM; +} /* * Array of bus specific translators @@ -121,6 +152,15 @@ static struct of_bus of_busses[] = { .map = of_bus_default_map, .get_flags = of_bus_sbus_get_flags, }, + /* AMBA */ + { + .name = "ambapp", + .addr_prop_name = "reg", + .match = of_bus_ambapp_match, + .count_cells = of_bus_ambapp_count_cells, + .map = of_bus_ambapp_map, + .get_flags = of_bus_ambapp_get_flags, + }, /* Default */ { .name = "default", diff --git a/arch/sparc/kernel/pcr.c b/arch/sparc/kernel/pcr.c index 1ae8cdd7e70..68ff0010707 100644 --- a/arch/sparc/kernel/pcr.c +++ b/arch/sparc/kernel/pcr.c @@ -7,6 +7,8 @@ #include <linux/init.h> #include <linux/irq.h> +#include <linux/perf_counter.h> + #include <asm/pil.h> #include <asm/pcr.h> #include <asm/nmi.h> @@ -34,10 +36,20 @@ unsigned int picl_shift; */ void deferred_pcr_work_irq(int irq, struct pt_regs *regs) { + struct pt_regs *old_regs; + clear_softint(1 << PIL_DEFERRED_PCR_WORK); + + old_regs = set_irq_regs(regs); + irq_enter(); +#ifdef CONFIG_PERF_COUNTERS + perf_counter_do_pending(); +#endif + irq_exit(); + set_irq_regs(old_regs); } -void schedule_deferred_pcr_work(void) +void set_perf_counter_pending(void) { set_softint(1 << PIL_DEFERRED_PCR_WORK); } diff --git a/arch/sparc/kernel/perf_counter.c b/arch/sparc/kernel/perf_counter.c new file mode 100644 index 00000000000..09de4035eaa --- /dev/null +++ b/arch/sparc/kernel/perf_counter.c @@ -0,0 +1,557 @@ +/* Performance counter support for sparc64. + * + * Copyright (C) 2009 David S. Miller <davem@davemloft.net> + * + * This code is based almost entirely upon the x86 perf counter + * code, which is: + * + * Copyright (C) 2008 Thomas Gleixner <tglx@linutronix.de> + * Copyright (C) 2008-2009 Red Hat, Inc., Ingo Molnar + * Copyright (C) 2009 Jaswinder Singh Rajput + * Copyright (C) 2009 Advanced Micro Devices, Inc., Robert Richter + * Copyright (C) 2008-2009 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com> + */ + +#include <linux/perf_counter.h> +#include <linux/kprobes.h> +#include <linux/kernel.h> +#include <linux/kdebug.h> +#include <linux/mutex.h> + +#include <asm/cpudata.h> +#include <asm/atomic.h> +#include <asm/nmi.h> +#include <asm/pcr.h> + +/* Sparc64 chips have two performance counters, 32-bits each, with + * overflow interrupts generated on transition from 0xffffffff to 0. + * The counters are accessed in one go using a 64-bit register. + * + * Both counters are controlled using a single control register. The + * only way to stop all sampling is to clear all of the context (user, + * supervisor, hypervisor) sampling enable bits. But these bits apply + * to both counters, thus the two counters can't be enabled/disabled + * individually. + * + * The control register has two event fields, one for each of the two + * counters. It's thus nearly impossible to have one counter going + * while keeping the other one stopped. Therefore it is possible to + * get overflow interrupts for counters not currently "in use" and + * that condition must be checked in the overflow interrupt handler. + * + * So we use a hack, in that we program inactive counters with the + * "sw_count0" and "sw_count1" events. These count how many times + * the instruction "sethi %hi(0xfc000), %g0" is executed. It's an + * unusual way to encode a NOP and therefore will not trigger in + * normal code. + */ + +#define MAX_HWCOUNTERS 2 +#define MAX_PERIOD ((1UL << 32) - 1) + +#define PIC_UPPER_INDEX 0 +#define PIC_LOWER_INDEX 1 + +struct cpu_hw_counters { + struct perf_counter *counters[MAX_HWCOUNTERS]; + unsigned long used_mask[BITS_TO_LONGS(MAX_HWCOUNTERS)]; + unsigned long active_mask[BITS_TO_LONGS(MAX_HWCOUNTERS)]; + int enabled; +}; +DEFINE_PER_CPU(struct cpu_hw_counters, cpu_hw_counters) = { .enabled = 1, }; + +struct perf_event_map { + u16 encoding; + u8 pic_mask; +#define PIC_NONE 0x00 +#define PIC_UPPER 0x01 +#define PIC_LOWER 0x02 +}; + +struct sparc_pmu { + const struct perf_event_map *(*event_map)(int); + int max_events; + int upper_shift; + int lower_shift; + int event_mask; + int hv_bit; + int irq_bit; + int upper_nop; + int lower_nop; +}; + +static const struct perf_event_map ultra3i_perfmon_event_map[] = { + [PERF_COUNT_HW_CPU_CYCLES] = { 0x0000, PIC_UPPER | PIC_LOWER }, + [PERF_COUNT_HW_INSTRUCTIONS] = { 0x0001, PIC_UPPER | PIC_LOWER }, + [PERF_COUNT_HW_CACHE_REFERENCES] = { 0x0009, PIC_LOWER }, + [PERF_COUNT_HW_CACHE_MISSES] = { 0x0009, PIC_UPPER }, +}; + +static const struct perf_event_map *ultra3i_event_map(int event) +{ + return &ultra3i_perfmon_event_map[event]; +} + +static const struct sparc_pmu ultra3i_pmu = { + .event_map = ultra3i_event_map, + .max_events = ARRAY_SIZE(ultra3i_perfmon_event_map), + .upper_shift = 11, + .lower_shift = 4, + .event_mask = 0x3f, + .upper_nop = 0x1c, + .lower_nop = 0x14, +}; + +static const struct perf_event_map niagara2_perfmon_event_map[] = { + [PERF_COUNT_HW_CPU_CYCLES] = { 0x02ff, PIC_UPPER | PIC_LOWER }, + [PERF_COUNT_HW_INSTRUCTIONS] = { 0x02ff, PIC_UPPER | PIC_LOWER }, + [PERF_COUNT_HW_CACHE_REFERENCES] = { 0x0208, PIC_UPPER | PIC_LOWER }, + [PERF_COUNT_HW_CACHE_MISSES] = { 0x0302, PIC_UPPER | PIC_LOWER }, + [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = { 0x0201, PIC_UPPER | PIC_LOWER }, + [PERF_COUNT_HW_BRANCH_MISSES] = { 0x0202, PIC_UPPER | PIC_LOWER }, +}; + +static const struct perf_event_map *niagara2_event_map(int event) +{ + return &niagara2_perfmon_event_map[event]; +} + +static const struct sparc_pmu niagara2_pmu = { + .event_map = niagara2_event_map, + .max_events = ARRAY_SIZE(niagara2_perfmon_event_map), + .upper_shift = 19, + .lower_shift = 6, + .event_mask = 0xfff, + .hv_bit = 0x8, + .irq_bit = 0x03, + .upper_nop = 0x220, + .lower_nop = 0x220, +}; + +static const struct sparc_pmu *sparc_pmu __read_mostly; + +static u64 event_encoding(u64 event, int idx) +{ + if (idx == PIC_UPPER_INDEX) + event <<= sparc_pmu->upper_shift; + else + event <<= sparc_pmu->lower_shift; + return event; +} + +static u64 mask_for_index(int idx) +{ + return event_encoding(sparc_pmu->event_mask, idx); +} + +static u64 nop_for_index(int idx) +{ + return event_encoding(idx == PIC_UPPER_INDEX ? + sparc_pmu->upper_nop : + sparc_pmu->lower_nop, idx); +} + +static inline void sparc_pmu_enable_counter(struct hw_perf_counter *hwc, + int idx) +{ + u64 val, mask = mask_for_index(idx); + + val = pcr_ops->read(); + pcr_ops->write((val & ~mask) | hwc->config); +} + +static inline void sparc_pmu_disable_counter(struct hw_perf_counter *hwc, + int idx) +{ + u64 mask = mask_for_index(idx); + u64 nop = nop_for_index(idx); + u64 val = pcr_ops->read(); + + pcr_ops->write((val & ~mask) | nop); +} + +void hw_perf_enable(void) +{ + struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters); + u64 val; + int i; + + if (cpuc->enabled) + return; + + cpuc->enabled = 1; + barrier(); + + val = pcr_ops->read(); + + for (i = 0; i < MAX_HWCOUNTERS; i++) { + struct perf_counter *cp = cpuc->counters[i]; + struct hw_perf_counter *hwc; + + if (!cp) + continue; + hwc = &cp->hw; + val |= hwc->config_base; + } + + pcr_ops->write(val); +} + +void hw_perf_disable(void) +{ + struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters); + u64 val; + + if (!cpuc->enabled) + return; + + cpuc->enabled = 0; + + val = pcr_ops->read(); + val &= ~(PCR_UTRACE | PCR_STRACE | + sparc_pmu->hv_bit | sparc_pmu->irq_bit); + pcr_ops->write(val); +} + +static u32 read_pmc(int idx) +{ + u64 val; + + read_pic(val); + if (idx == PIC_UPPER_INDEX) + val >>= 32; + + return val & 0xffffffff; +} + +static void write_pmc(int idx, u64 val) +{ + u64 shift, mask, pic; + + shift = 0; + if (idx == PIC_UPPER_INDEX) + shift = 32; + + mask = ((u64) 0xffffffff) << shift; + val <<= shift; + + read_pic(pic); + pic &= ~mask; + pic |= val; + write_pic(pic); +} + +static int sparc_perf_counter_set_period(struct perf_counter *counter, + struct hw_perf_counter *hwc, int idx) +{ + s64 left = atomic64_read(&hwc->period_left); + s64 period = hwc->sample_period; + int ret = 0; + + if (unlikely(left <= -period)) { + left = period; + atomic64_set(&hwc->period_left, left); + hwc->last_period = period; + ret = 1; + } + + if (unlikely(left <= 0)) { + left += period; + atomic64_set(&hwc->period_left, left); + hwc->last_period = period; + ret = 1; + } + if (left > MAX_PERIOD) + left = MAX_PERIOD; + + atomic64_set(&hwc->prev_count, (u64)-left); + + write_pmc(idx, (u64)(-left) & 0xffffffff); + + perf_counter_update_userpage(counter); + + return ret; +} + +static int sparc_pmu_enable(struct perf_counter *counter) +{ + struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters); + struct hw_perf_counter *hwc = &counter->hw; + int idx = hwc->idx; + + if (test_and_set_bit(idx, cpuc->used_mask)) + return -EAGAIN; + + sparc_pmu_disable_counter(hwc, idx); + + cpuc->counters[idx] = counter; + set_bit(idx, cpuc->active_mask); + + sparc_perf_counter_set_period(counter, hwc, idx); + sparc_pmu_enable_counter(hwc, idx); + perf_counter_update_userpage(counter); + return 0; +} + +static u64 sparc_perf_counter_update(struct perf_counter *counter, + struct hw_perf_counter *hwc, int idx) +{ + int shift = 64 - 32; + u64 prev_raw_count, new_raw_count; + s64 delta; + +again: + prev_raw_count = atomic64_read(&hwc->prev_count); + new_raw_count = read_pmc(idx); + + if (atomic64_cmpxchg(&hwc->prev_count, prev_raw_count, + new_raw_count) != prev_raw_count) + goto again; + + delta = (new_raw_count << shift) - (prev_raw_count << shift); + delta >>= shift; + + atomic64_add(delta, &counter->count); + atomic64_sub(delta, &hwc->period_left); + + return new_raw_count; +} + +static void sparc_pmu_disable(struct perf_counter *counter) +{ + struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters); + struct hw_perf_counter *hwc = &counter->hw; + int idx = hwc->idx; + + clear_bit(idx, cpuc->active_mask); + sparc_pmu_disable_counter(hwc, idx); + + barrier(); + + sparc_perf_counter_update(counter, hwc, idx); + cpuc->counters[idx] = NULL; + clear_bit(idx, cpuc->used_mask); + + perf_counter_update_userpage(counter); +} + +static void sparc_pmu_read(struct perf_counter *counter) +{ + struct hw_perf_counter *hwc = &counter->hw; + sparc_perf_counter_update(counter, hwc, hwc->idx); +} + +static void sparc_pmu_unthrottle(struct perf_counter *counter) +{ + struct hw_perf_counter *hwc = &counter->hw; + sparc_pmu_enable_counter(hwc, hwc->idx); +} + +static atomic_t active_counters = ATOMIC_INIT(0); +static DEFINE_MUTEX(pmc_grab_mutex); + +void perf_counter_grab_pmc(void) +{ + if (atomic_inc_not_zero(&active_counters)) + return; + + mutex_lock(&pmc_grab_mutex); + if (atomic_read(&active_counters) == 0) { + if (atomic_read(&nmi_active) > 0) { + on_each_cpu(stop_nmi_watchdog, NULL, 1); + BUG_ON(atomic_read(&nmi_active) != 0); + } + atomic_inc(&active_counters); + } + mutex_unlock(&pmc_grab_mutex); +} + +void perf_counter_release_pmc(void) +{ + if (atomic_dec_and_mutex_lock(&active_counters, &pmc_grab_mutex)) { + if (atomic_read(&nmi_active) == 0) + on_each_cpu(start_nmi_watchdog, NULL, 1); + mutex_unlock(&pmc_grab_mutex); + } +} + +static void hw_perf_counter_destroy(struct perf_counter *counter) +{ + perf_counter_release_pmc(); +} + +static int __hw_perf_counter_init(struct perf_counter *counter) +{ + struct perf_counter_attr *attr = &counter->attr; + struct hw_perf_counter *hwc = &counter->hw; + const struct perf_event_map *pmap; + u64 enc; + + if (atomic_read(&nmi_active) < 0) + return -ENODEV; + + if (attr->type != PERF_TYPE_HARDWARE) + return -EOPNOTSUPP; + + if (attr->config >= sparc_pmu->max_events) + return -EINVAL; + + perf_counter_grab_pmc(); + counter->destroy = hw_perf_counter_destroy; + + /* We save the enable bits in the config_base. So to + * turn off sampling just write 'config', and to enable + * things write 'config | config_base'. + */ + hwc->config_base = sparc_pmu->irq_bit; + if (!attr->exclude_user) + hwc->config_base |= PCR_UTRACE; + if (!attr->exclude_kernel) + hwc->config_base |= PCR_STRACE; + if (!attr->exclude_hv) + hwc->config_base |= sparc_pmu->hv_bit; + + if (!hwc->sample_period) { + hwc->sample_period = MAX_PERIOD; + hwc->last_period = hwc->sample_period; + atomic64_set(&hwc->period_left, hwc->sample_period); + } + + pmap = sparc_pmu->event_map(attr->config); + + enc = pmap->encoding; + if (pmap->pic_mask & PIC_UPPER) { + hwc->idx = PIC_UPPER_INDEX; + enc <<= sparc_pmu->upper_shift; + } else { + hwc->idx = PIC_LOWER_INDEX; + enc <<= sparc_pmu->lower_shift; + } + + hwc->config |= enc; + return 0; +} + +static const struct pmu pmu = { + .enable = sparc_pmu_enable, + .disable = sparc_pmu_disable, + .read = sparc_pmu_read, + .unthrottle = sparc_pmu_unthrottle, +}; + +const struct pmu *hw_perf_counter_init(struct perf_counter *counter) +{ + int err = __hw_perf_counter_init(counter); + + if (err) + return ERR_PTR(err); + return &pmu; +} + +void perf_counter_print_debug(void) +{ + unsigned long flags; + u64 pcr, pic; + int cpu; + + if (!sparc_pmu) + return; + + local_irq_save(flags); + + cpu = smp_processor_id(); + + pcr = pcr_ops->read(); + read_pic(pic); + + pr_info("\n"); + pr_info("CPU#%d: PCR[%016llx] PIC[%016llx]\n", + cpu, pcr, pic); + + local_irq_restore(flags); +} + +static int __kprobes perf_counter_nmi_handler(struct notifier_block *self, + unsigned long cmd, void *__args) +{ + struct die_args *args = __args; + struct perf_sample_data data; + struct cpu_hw_counters *cpuc; + struct pt_regs *regs; + int idx; + + if (!atomic_read(&active_counters)) + return NOTIFY_DONE; + + switch (cmd) { + case DIE_NMI: + break; + + default: + return NOTIFY_DONE; + } + + regs = args->regs; + + data.regs = regs; + data.addr = 0; + + cpuc = &__get_cpu_var(cpu_hw_counters); + for (idx = 0; idx < MAX_HWCOUNTERS; idx++) { + struct perf_counter *counter = cpuc->counters[idx]; + struct hw_perf_counter *hwc; + u64 val; + + if (!test_bit(idx, cpuc->active_mask)) + continue; + hwc = &counter->hw; + val = sparc_perf_counter_update(counter, hwc, idx); + if (val & (1ULL << 31)) + continue; + + data.period = counter->hw.last_period; + if (!sparc_perf_counter_set_period(counter, hwc, idx)) + continue; + + if (perf_counter_overflow(counter, 1, &data)) + sparc_pmu_disable_counter(hwc, idx); + } + + return NOTIFY_STOP; +} + +static __read_mostly struct notifier_block perf_counter_nmi_notifier = { + .notifier_call = perf_counter_nmi_handler, +}; + +static bool __init supported_pmu(void) +{ + if (!strcmp(sparc_pmu_type, "ultra3i")) { + sparc_pmu = &ultra3i_pmu; + return true; + } + if (!strcmp(sparc_pmu_type, "niagara2")) { + sparc_pmu = &niagara2_pmu; + return true; + } + return false; +} + +void __init init_hw_perf_counters(void) +{ + pr_info("Performance counters: "); + + if (!supported_pmu()) { + pr_cont("No support for PMU type '%s'\n", sparc_pmu_type); + return; + } + + pr_cont("Supported PMU type is '%s'\n", sparc_pmu_type); + + /* All sparc64 PMUs currently have 2 counters. But this simple + * driver only supports one active counter at a time. + */ + perf_max_counters = 1; + + register_die_notifier(&perf_counter_nmi_notifier); +} diff --git a/arch/sparc/kernel/prom_32.c b/arch/sparc/kernel/prom_32.c index fe43e80772d..0a37e8cfd16 100644 --- a/arch/sparc/kernel/prom_32.c +++ b/arch/sparc/kernel/prom_32.c @@ -24,6 +24,8 @@ #include <asm/prom.h> #include <asm/oplib.h> +#include <asm/leon.h> +#include <asm/leon_amba.h> #include "prom.h" @@ -131,6 +133,35 @@ static void __init ebus_path_component(struct device_node *dp, char *tmp_buf) regs->which_io, regs->phys_addr); } +/* "name:vendor:device@irq,addrlo" */ +static void __init ambapp_path_component(struct device_node *dp, char *tmp_buf) +{ + struct amba_prom_registers *regs; unsigned int *intr; + unsigned int *device, *vendor; + struct property *prop; + + prop = of_find_property(dp, "reg", NULL); + if (!prop) + return; + regs = prop->value; + prop = of_find_property(dp, "interrupts", NULL); + if (!prop) + return; + intr = prop->value; + prop = of_find_property(dp, "vendor", NULL); + if (!prop) + return; + vendor = prop->value; + prop = of_find_property(dp, "device", NULL); + if (!prop) + return; + device = prop->value; + + sprintf(tmp_buf, "%s:%d:%d@%x,%x", + dp->name, *vendor, *device, + *intr, regs->phys_addr); +} + static void __init __build_path_component(struct device_node *dp, char *tmp_buf) { struct device_node *parent = dp->parent; @@ -143,6 +174,8 @@ static void __init __build_path_component(struct device_node *dp, char *tmp_buf) return sbus_path_component(dp, tmp_buf); if (!strcmp(parent->type, "ebus")) return ebus_path_component(dp, tmp_buf); + if (!strcmp(parent->type, "ambapp")) + return ambapp_path_component(dp, tmp_buf); /* "isa" is handled with platform naming */ } diff --git a/arch/sparc/kernel/prom_common.c b/arch/sparc/kernel/prom_common.c index 0fb5789d43c..138910c6720 100644 --- a/arch/sparc/kernel/prom_common.c +++ b/arch/sparc/kernel/prom_common.c @@ -22,9 +22,12 @@ #include <linux/of.h> #include <asm/prom.h> #include <asm/oplib.h> +#include <asm/leon.h> #include "prom.h" +void (*prom_build_more)(struct device_node *dp, struct device_node ***nextp); + struct device_node *of_console_device; EXPORT_SYMBOL(of_console_device); @@ -161,7 +164,7 @@ static struct property * __init build_one_prop(phandle node, char *prev, name = prom_nextprop(node, prev, p->name); } - if (strlen(name) == 0) { + if (!name || strlen(name) == 0) { tmp = p; return NULL; } @@ -242,7 +245,7 @@ static struct device_node * __init prom_create_node(phandle node, return dp; } -static char * __init build_full_name(struct device_node *dp) +char * __init build_full_name(struct device_node *dp) { int len, ourlen, plen; char *n; @@ -289,6 +292,9 @@ static struct device_node * __init prom_build_tree(struct device_node *parent, dp->child = prom_build_tree(dp, prom_getchild(node), nextp); + if (prom_build_more) + prom_build_more(dp, nextp); + node = prom_getsibling(node); } diff --git a/arch/sparc/kernel/setup_32.c b/arch/sparc/kernel/setup_32.c index 998cadb4e7f..16a47ffe03c 100644 --- a/arch/sparc/kernel/setup_32.c +++ b/arch/sparc/kernel/setup_32.c @@ -235,6 +235,8 @@ void __init setup_arch(char **cmdline_p) sparc_cpu_model = sun4e; if (!strcmp(&cputypval,"sun4u")) sparc_cpu_model = sun4u; + if (!strncmp(&cputypval, "leon" , 4)) + sparc_cpu_model = sparc_leon; printk("ARCH: "); switch(sparc_cpu_model) { @@ -256,6 +258,9 @@ void __init setup_arch(char **cmdline_p) case sun4u: printk("SUN4U\n"); break; + case sparc_leon: + printk("LEON\n"); + break; default: printk("UNKNOWN!\n"); break; diff --git a/arch/sparc/kernel/sysfs.c b/arch/sparc/kernel/sysfs.c index d28f496f466..ca39c606fe8 100644 --- a/arch/sparc/kernel/sysfs.c +++ b/arch/sparc/kernel/sysfs.c @@ -2,6 +2,7 @@ * * Copyright (C) 2007 David S. Miller <davem@davemloft.net> */ +#include <linux/sched.h> #include <linux/sysdev.h> #include <linux/cpu.h> #include <linux/smp.h> diff --git a/arch/sparc/kernel/systbls_32.S b/arch/sparc/kernel/systbls_32.S index 69090165729..04181577cb6 100644 --- a/arch/sparc/kernel/systbls_32.S +++ b/arch/sparc/kernel/systbls_32.S @@ -82,5 +82,5 @@ sys_call_table: /*310*/ .long sys_utimensat, sys_signalfd, sys_timerfd_create, sys_eventfd, sys_fallocate /*315*/ .long sys_timerfd_settime, sys_timerfd_gettime, sys_signalfd4, sys_eventfd2, sys_epoll_create1 /*320*/ .long sys_dup3, sys_pipe2, sys_inotify_init1, sys_accept4, sys_preadv -/*325*/ .long sys_pwritev, sys_rt_tgsigqueueinfo +/*325*/ .long sys_pwritev, sys_rt_tgsigqueueinfo, sys_perf_counter_open diff --git a/arch/sparc/kernel/systbls_64.S b/arch/sparc/kernel/systbls_64.S index 2ee7250ba7a..91b06b7f7ac 100644 --- a/arch/sparc/kernel/systbls_64.S +++ b/arch/sparc/kernel/systbls_64.S @@ -83,7 +83,7 @@ sys_call_table32: /*310*/ .word compat_sys_utimensat, compat_sys_signalfd, sys_timerfd_create, sys_eventfd, compat_sys_fallocate .word compat_sys_timerfd_settime, compat_sys_timerfd_gettime, compat_sys_signalfd4, sys_eventfd2, sys_epoll_create1 /*320*/ .word sys_dup3, sys_pipe2, sys_inotify_init1, sys_accept4, compat_sys_preadv - .word compat_sys_pwritev, compat_sys_rt_tgsigqueueinfo + .word compat_sys_pwritev, compat_sys_rt_tgsigqueueinfo, sys_perf_counter_open #endif /* CONFIG_COMPAT */ @@ -158,4 +158,4 @@ sys_call_table: /*310*/ .word sys_utimensat, sys_signalfd, sys_timerfd_create, sys_eventfd, sys_fallocate .word sys_timerfd_settime, sys_timerfd_gettime, sys_signalfd4, sys_eventfd2, sys_epoll_create1 /*320*/ .word sys_dup3, sys_pipe2, sys_inotify_init1, sys_accept4, sys_preadv - .word sys_pwritev, sys_rt_tgsigqueueinfo + .word sys_pwritev, sys_rt_tgsigqueueinfo, sys_perf_counter_open |