diff options
Diffstat (limited to 'arch')
40 files changed, 897 insertions, 388 deletions
diff --git a/arch/arm/include/asm/mmu.h b/arch/arm/include/asm/mmu.h index 53099d4ee421..b561584d04a1 100644 --- a/arch/arm/include/asm/mmu.h +++ b/arch/arm/include/asm/mmu.h @@ -24,7 +24,6 @@ typedef struct { * modified for 2.6 by Hyok S. Choi <hyok.choi@samsung.com> */ typedef struct { - struct vm_list_struct *vmlist; unsigned long end_brk; } mm_context_t; diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c index 67960017dc8f..310e479309ef 100644 --- a/arch/arm/mm/dma-mapping.c +++ b/arch/arm/mm/dma-mapping.c @@ -71,7 +71,7 @@ static DEFINE_SPINLOCK(consistent_lock); * the amount of RAM found at boot time.) I would imagine that get_vm_area() * would have to initialise this each time prior to calling vm_region_alloc(). */ -struct vm_region { +struct arm_vm_region { struct list_head vm_list; unsigned long vm_start; unsigned long vm_end; @@ -79,20 +79,20 @@ struct vm_region { int vm_active; }; -static struct vm_region consistent_head = { +static struct arm_vm_region consistent_head = { .vm_list = LIST_HEAD_INIT(consistent_head.vm_list), .vm_start = CONSISTENT_BASE, .vm_end = CONSISTENT_END, }; -static struct vm_region * -vm_region_alloc(struct vm_region *head, size_t size, gfp_t gfp) +static struct arm_vm_region * +arm_vm_region_alloc(struct arm_vm_region *head, size_t size, gfp_t gfp) { unsigned long addr = head->vm_start, end = head->vm_end - size; unsigned long flags; - struct vm_region *c, *new; + struct arm_vm_region *c, *new; - new = kmalloc(sizeof(struct vm_region), gfp); + new = kmalloc(sizeof(struct arm_vm_region), gfp); if (!new) goto out; @@ -127,9 +127,9 @@ vm_region_alloc(struct vm_region *head, size_t size, gfp_t gfp) return NULL; } -static struct vm_region *vm_region_find(struct vm_region *head, unsigned long addr) +static struct arm_vm_region *arm_vm_region_find(struct arm_vm_region *head, unsigned long addr) { - struct vm_region *c; + struct arm_vm_region *c; list_for_each_entry(c, &head->vm_list, vm_list) { if (c->vm_active && c->vm_start == addr) @@ -149,7 +149,7 @@ __dma_alloc(struct device *dev, size_t size, dma_addr_t *handle, gfp_t gfp, pgprot_t prot) { struct page *page; - struct vm_region *c; + struct arm_vm_region *c; unsigned long order; u64 mask = ISA_DMA_THRESHOLD, limit; @@ -214,7 +214,7 @@ __dma_alloc(struct device *dev, size_t size, dma_addr_t *handle, gfp_t gfp, /* * Allocate a virtual address in the consistent mapping region. */ - c = vm_region_alloc(&consistent_head, size, + c = arm_vm_region_alloc(&consistent_head, size, gfp & ~(__GFP_DMA | __GFP_HIGHMEM)); if (c) { pte_t *pte; @@ -311,13 +311,13 @@ static int dma_mmap(struct device *dev, struct vm_area_struct *vma, void *cpu_addr, dma_addr_t dma_addr, size_t size) { unsigned long flags, user_size, kern_size; - struct vm_region *c; + struct arm_vm_region *c; int ret = -ENXIO; user_size = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; spin_lock_irqsave(&consistent_lock, flags); - c = vm_region_find(&consistent_head, (unsigned long)cpu_addr); + c = arm_vm_region_find(&consistent_head, (unsigned long)cpu_addr); spin_unlock_irqrestore(&consistent_lock, flags); if (c) { @@ -359,7 +359,7 @@ EXPORT_SYMBOL(dma_mmap_writecombine); */ void dma_free_coherent(struct device *dev, size_t size, void *cpu_addr, dma_addr_t handle) { - struct vm_region *c; + struct arm_vm_region *c; unsigned long flags, addr; pte_t *ptep; int idx; @@ -378,7 +378,7 @@ void dma_free_coherent(struct device *dev, size_t size, void *cpu_addr, dma_addr size = PAGE_ALIGN(size); spin_lock_irqsave(&consistent_lock, flags); - c = vm_region_find(&consistent_head, (unsigned long)cpu_addr); + c = arm_vm_region_find(&consistent_head, (unsigned long)cpu_addr); if (!c) goto no_area; diff --git a/arch/blackfin/include/asm/mmu.h b/arch/blackfin/include/asm/mmu.h index 757e43906ed4..dbfd686360e6 100644 --- a/arch/blackfin/include/asm/mmu.h +++ b/arch/blackfin/include/asm/mmu.h @@ -10,7 +10,6 @@ struct sram_list_struct { }; typedef struct { - struct vm_list_struct *vmlist; unsigned long end_brk; unsigned long stack_start; diff --git a/arch/blackfin/kernel/ptrace.c b/arch/blackfin/kernel/ptrace.c index d2d388536630..594e325b40e4 100644 --- a/arch/blackfin/kernel/ptrace.c +++ b/arch/blackfin/kernel/ptrace.c @@ -160,15 +160,15 @@ put_reg(struct task_struct *task, int regno, unsigned long data) static inline int is_user_addr_valid(struct task_struct *child, unsigned long start, unsigned long len) { - struct vm_list_struct *vml; + struct vm_area_struct *vma; struct sram_list_struct *sraml; /* overflow */ if (start + len < start) return -EIO; - for (vml = child->mm->context.vmlist; vml; vml = vml->next) - if (start >= vml->vma->vm_start && start + len < vml->vma->vm_end) + vma = find_vma(child->mm, start); + if (vma && start >= vma->vm_start && start + len <= vma->vm_end) return 0; for (sraml = child->mm->context.sram_list; sraml; sraml = sraml->next) diff --git a/arch/blackfin/kernel/traps.c b/arch/blackfin/kernel/traps.c index 17d8e4172896..5b0667da8d05 100644 --- a/arch/blackfin/kernel/traps.c +++ b/arch/blackfin/kernel/traps.c @@ -32,6 +32,7 @@ #include <linux/module.h> #include <linux/kallsyms.h> #include <linux/fs.h> +#include <linux/rbtree.h> #include <asm/traps.h> #include <asm/cacheflush.h> #include <asm/cplb.h> @@ -83,6 +84,7 @@ static void decode_address(char *buf, unsigned long address) struct mm_struct *mm; unsigned long flags, offset; unsigned char in_atomic = (bfin_read_IPEND() & 0x10) || in_atomic(); + struct rb_node *n; #ifdef CONFIG_KALLSYMS unsigned long symsize; @@ -128,9 +130,10 @@ static void decode_address(char *buf, unsigned long address) if (!mm) continue; - vml = mm->context.vmlist; - while (vml) { - struct vm_area_struct *vma = vml->vma; + for (n = rb_first(&mm->mm_rb); n; n = rb_next(n)) { + struct vm_area_struct *vma; + + vma = rb_entry(n, struct vm_area_struct, vm_rb); if (address >= vma->vm_start && address < vma->vm_end) { char _tmpbuf[256]; @@ -176,8 +179,6 @@ static void decode_address(char *buf, unsigned long address) goto done; } - - vml = vml->next; } if (!in_atomic) mmput(mm); diff --git a/arch/frv/kernel/ptrace.c b/arch/frv/kernel/ptrace.c index 709e9bdc6126..5e7d401d21e7 100644 --- a/arch/frv/kernel/ptrace.c +++ b/arch/frv/kernel/ptrace.c @@ -69,7 +69,8 @@ static inline int put_reg(struct task_struct *task, int regno, } /* - * check that an address falls within the bounds of the target process's memory mappings + * check that an address falls within the bounds of the target process's memory + * mappings */ static inline int is_user_addr_valid(struct task_struct *child, unsigned long start, unsigned long len) @@ -79,11 +80,11 @@ static inline int is_user_addr_valid(struct task_struct *child, return -EIO; return 0; #else - struct vm_list_struct *vml; + struct vm_area_struct *vma; - for (vml = child->mm->context.vmlist; vml; vml = vml->next) - if (start >= vml->vma->vm_start && start + len <= vml->vma->vm_end) - return 0; + vma = find_vma(child->mm, start); + if (vma && start >= vma->vm_start && start + len <= vma->vm_end) + return 0; return -EIO; #endif diff --git a/arch/h8300/include/asm/mmu.h b/arch/h8300/include/asm/mmu.h index 2ce06ea46104..31309969df70 100644 --- a/arch/h8300/include/asm/mmu.h +++ b/arch/h8300/include/asm/mmu.h @@ -4,7 +4,6 @@ /* Copyright (C) 2002, David McCullough <davidm@snapgear.com> */ typedef struct { - struct vm_list_struct *vmlist; unsigned long end_brk; } mm_context_t; diff --git a/arch/m68knommu/include/asm/mmu.h b/arch/m68knommu/include/asm/mmu.h index 5fa6b68353ba..e2da1e6f09fe 100644 --- a/arch/m68knommu/include/asm/mmu.h +++ b/arch/m68knommu/include/asm/mmu.h @@ -4,7 +4,6 @@ /* Copyright (C) 2002, David McCullough <davidm@snapgear.com> */ typedef struct { - struct vm_list_struct *vmlist; unsigned long end_brk; } mm_context_t; diff --git a/arch/parisc/include/asm/Kbuild b/arch/parisc/include/asm/Kbuild index f88b252e419c..2121d99f8364 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 83095c5bb379..da66029c4cb2 100644 --- a/arch/parisc/include/asm/byteorder.h +++ b/arch/parisc/include/asm/byteorder.h @@ -1,69 +1,7 @@ #ifndef _PARISC_BYTEORDER_H #define _PARISC_BYTEORDER_H -#include <asm/types.h> -#include <linux/compiler.h> - -#define __BIG_ENDIAN -#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 */ - -#include <linux/byteorder.h> +#include <asm/swab.h> +#include <linux/byteorder/big_endian.h> #endif /* _PARISC_BYTEORDER_H */ diff --git a/arch/parisc/include/asm/swab.h b/arch/parisc/include/asm/swab.h new file mode 100644 index 000000000000..3ff16c5a3358 --- /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/powerpc/include/asm/cell-pmu.h b/arch/powerpc/include/asm/cell-pmu.h index 8066eede3a0c..b4b7338ad79e 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 95035c602ba6..639dc96077ab 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 dfdbffa06818..964b93974d89 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 83faa958b9d4..9305ddaac512 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 17807acb05d9..21f16edf6c8d 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 25a4ec2514a3..ae06c6236d9c 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/s390/include/asm/chpid.h b/arch/s390/include/asm/chpid.h index dfe3c7f3439a..fc71d8a6709b 100644 --- a/arch/s390/include/asm/chpid.h +++ b/arch/s390/include/asm/chpid.h @@ -9,7 +9,7 @@ #define _ASM_S390_CHPID_H _ASM_S390_CHPID_H #include <linux/string.h> -#include <asm/types.h> +#include <linux/types.h> #define __MAX_CHPID 255 diff --git a/arch/s390/include/asm/chsc.h b/arch/s390/include/asm/chsc.h index d38d0cf62d4b..807997f7414b 100644 --- a/arch/s390/include/asm/chsc.h +++ b/arch/s390/include/asm/chsc.h @@ -8,6 +8,7 @@ #ifndef _ASM_CHSC_H #define _ASM_CHSC_H +#include <linux/types.h> #include <asm/chpid.h> #include <asm/schid.h> diff --git a/arch/s390/include/asm/cmb.h b/arch/s390/include/asm/cmb.h index 50196857d27a..39ae03294794 100644 --- a/arch/s390/include/asm/cmb.h +++ b/arch/s390/include/asm/cmb.h @@ -1,5 +1,8 @@ #ifndef S390_CMB_H #define S390_CMB_H + +#include <linux/types.h> + /** * struct cmbdata - channel measurement block data for user space * @size: size of the stored data diff --git a/arch/s390/include/asm/dasd.h b/arch/s390/include/asm/dasd.h index 55b2b80cdf6e..e2db6f16d9c8 100644 --- a/arch/s390/include/asm/dasd.h +++ b/arch/s390/include/asm/dasd.h @@ -14,6 +14,7 @@ #ifndef DASD_H #define DASD_H +#include <linux/types.h> #include <linux/ioctl.h> #define DASD_IOCTL_LETTER 'D' @@ -78,6 +79,7 @@ typedef struct dasd_information2_t { #define DASD_FEATURE_USEDIAG 0x02 #define DASD_FEATURE_INITIAL_ONLINE 0x04 #define DASD_FEATURE_ERPLOG 0x08 +#define DASD_FEATURE_FAILFAST 0x10 #define DASD_PARTN_BITS 2 diff --git a/arch/s390/include/asm/kvm.h b/arch/s390/include/asm/kvm.h index d74002f95794..e1f54654e3ae 100644 --- a/arch/s390/include/asm/kvm.h +++ b/arch/s390/include/asm/kvm.h @@ -13,7 +13,7 @@ * Author(s): Carsten Otte <cotte@de.ibm.com> * Christian Borntraeger <borntraeger@de.ibm.com> */ -#include <asm/types.h> +#include <linux/types.h> /* for KVM_GET_IRQCHIP and KVM_SET_IRQCHIP */ struct kvm_pic_state { diff --git a/arch/s390/include/asm/posix_types.h b/arch/s390/include/asm/posix_types.h index 397d93fba3a7..8cc113f92523 100644 --- a/arch/s390/include/asm/posix_types.h +++ b/arch/s390/include/asm/posix_types.h @@ -68,11 +68,7 @@ typedef unsigned short __kernel_old_dev_t; #endif /* __s390x__ */ typedef struct { -#if defined(__KERNEL__) || defined(__USE_ALL) int val[2]; -#else /* !defined(__KERNEL__) && !defined(__USE_ALL)*/ - int __val[2]; -#endif /* !defined(__KERNEL__) && !defined(__USE_ALL)*/ } __kernel_fsid_t; diff --git a/arch/s390/include/asm/ptrace.h b/arch/s390/include/asm/ptrace.h index 5396f9f12263..8920025c3c02 100644 --- a/arch/s390/include/asm/ptrace.h +++ b/arch/s390/include/asm/ptrace.h @@ -272,12 +272,15 @@ typedef struct #define PSW_ASC_SECONDARY 0x0000800000000000UL #define PSW_ASC_HOME 0x0000C00000000000UL -extern long psw_user32_bits; - #endif /* __s390x__ */ +#ifdef __KERNEL__ extern long psw_kernel_bits; extern long psw_user_bits; +#ifdef CONFIG_64BIT +extern long psw_user32_bits; +#endif +#endif /* This macro merges a NEW PSW mask specified by the user into the currently active PSW mask CURRENT, modifying only those diff --git a/arch/s390/include/asm/qeth.h b/arch/s390/include/asm/qeth.h index 930d378ef75a..06cbd1e8c943 100644 --- a/arch/s390/include/asm/qeth.h +++ b/arch/s390/include/asm/qeth.h @@ -10,6 +10,7 @@ */ #ifndef __ASM_S390_QETH_IOCTL_H__ #define __ASM_S390_QETH_IOCTL_H__ +#include <linux/types.h> #include <linux/ioctl.h> #define SIOC_QETH_ARP_SET_NO_ENTRIES (SIOCDEVPRIVATE) diff --git a/arch/s390/include/asm/schid.h b/arch/s390/include/asm/schid.h index 825503cf3dc2..3e4d401b4e45 100644 --- a/arch/s390/include/asm/schid.h +++ b/arch/s390/include/asm/schid.h @@ -1,6 +1,8 @@ #ifndef ASM_SCHID_H #define ASM_SCHID_H +#include <linux/types.h> + struct subchannel_id { __u32 cssid : 8; __u32 : 4; diff --git a/arch/s390/include/asm/swab.h b/arch/s390/include/asm/swab.h index bd9321aa55a9..eb18dc1f327b 100644 --- a/arch/s390/include/asm/swab.h +++ b/arch/s390/include/asm/swab.h @@ -9,7 +9,7 @@ * Author(s): Martin Schwidefsky (schwidefsky@de.ibm.com) */ -#include <asm/types.h> +#include <linux/types.h> #ifndef __s390x__ # define __SWAB_64_THRU_32__ diff --git a/arch/s390/include/asm/types.h b/arch/s390/include/asm/types.h index 41c547656130..3dc3fc228812 100644 --- a/arch/s390/include/asm/types.h +++ b/arch/s390/include/asm/types.h @@ -9,11 +9,7 @@ #ifndef _S390_TYPES_H #define _S390_TYPES_H -#ifndef __s390x__ -# include <asm-generic/int-ll64.h> -#else -# include <asm-generic/int-l64.h> -#endif +#include <asm-generic/int-ll64.h> #ifndef __ASSEMBLY__ diff --git a/arch/s390/kernel/entry.h b/arch/s390/kernel/entry.h index 6b1896345eda..a65afc91e8aa 100644 --- a/arch/s390/kernel/entry.h +++ b/arch/s390/kernel/entry.h @@ -54,7 +54,5 @@ long sys_sigreturn(void); long sys_rt_sigreturn(void); long sys32_sigreturn(void); long sys32_rt_sigreturn(void); -long old_select(struct sel_arg_struct __user *arg); -long sys_ptrace(long request, long pid, long addr, long data); #endif /* _ENTRY_H */ diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c index 9c0ccb532a45..2d337cbb9329 100644 --- a/arch/s390/kernel/smp.c +++ b/arch/s390/kernel/smp.c @@ -685,7 +685,8 @@ void __init smp_prepare_cpus(unsigned int max_cpus) if (MACHINE_HAS_IEEE) lowcore->extended_save_area_addr = (u32) save_area; #else - BUG_ON(vdso_alloc_per_cpu(smp_processor_id(), lowcore)); + if (vdso_alloc_per_cpu(smp_processor_id(), lowcore)) + BUG(); #endif set_prefix((u32)(unsigned long) lowcore); local_mcck_enable(); diff --git a/arch/s390/kernel/sys_s390.c b/arch/s390/kernel/sys_s390.c index 4fe952e557ac..c34be4568b80 100644 --- a/arch/s390/kernel/sys_s390.c +++ b/arch/s390/kernel/sys_s390.c @@ -103,25 +103,6 @@ out: return error; } -#ifndef CONFIG_64BIT -struct sel_arg_struct { - unsigned long n; - fd_set __user *inp, *outp, *exp; - struct timeval __user *tvp; -}; - -asmlinkage long old_select(struct sel_arg_struct __user *arg) -{ - struct sel_arg_struct a; - - if (copy_from_user(&a, arg, sizeof(a))) - return -EFAULT; - /* sys_select() does the appropriate kernel locking */ - return sys_select(a.n, a.inp, a.outp, a.exp, a.tvp); - -} -#endif /* CONFIG_64BIT */ - /* * sys_ipc() is the de-multiplexer for the SysV IPC calls.. * diff --git a/arch/s390/kernel/vdso.c b/arch/s390/kernel/vdso.c index 25a6a82f1c02..690e17819686 100644 --- a/arch/s390/kernel/vdso.c +++ b/arch/s390/kernel/vdso.c @@ -322,7 +322,8 @@ static int __init vdso_init(void) vdso64_pagelist[vdso64_pages - 1] = virt_to_page(vdso_data); vdso64_pagelist[vdso64_pages] = NULL; #ifndef CONFIG_SMP - BUG_ON(vdso_alloc_per_cpu(0, S390_lowcore)); + if (vdso_alloc_per_cpu(0, &S390_lowcore)) + BUG(); #endif vdso_init_cr5(); #endif /* CONFIG_64BIT */ diff --git a/arch/s390/kernel/vdso32/gettimeofday.S b/arch/s390/kernel/vdso32/gettimeofday.S index c32f29c3d70c..ad8acfc949fb 100644 --- a/arch/s390/kernel/vdso32/gettimeofday.S +++ b/arch/s390/kernel/vdso32/gettimeofday.S @@ -13,10 +13,6 @@ #include <asm/asm-offsets.h> #include <asm/unistd.h> -#include <asm/vdso.h> -#include <asm/asm-offsets.h> -#include <asm/unistd.h> - .text .align 4 .globl __kernel_gettimeofday diff --git a/arch/s390/kvm/diag.c b/arch/s390/kvm/diag.c index a0775e1f08df..8300309698fa 100644 --- a/arch/s390/kvm/diag.c +++ b/arch/s390/kvm/diag.c @@ -47,7 +47,7 @@ static int __diag_ipl_functions(struct kvm_vcpu *vcpu) vcpu->run->s390_reset_flags |= KVM_S390_RESET_IPL; vcpu->run->s390_reset_flags |= KVM_S390_RESET_CPU_INIT; vcpu->run->exit_reason = KVM_EXIT_S390_RESET; - VCPU_EVENT(vcpu, 3, "requesting userspace resets %lx", + VCPU_EVENT(vcpu, 3, "requesting userspace resets %llx", vcpu->run->s390_reset_flags); return -EREMOTE; } diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c index 2960702b4824..f4fe28a2521a 100644 --- a/arch/s390/kvm/interrupt.c +++ b/arch/s390/kvm/interrupt.c @@ -160,7 +160,7 @@ static void __do_deliver_interrupt(struct kvm_vcpu *vcpu, break; case KVM_S390_INT_VIRTIO: - VCPU_EVENT(vcpu, 4, "interrupt: virtio parm:%x,parm64:%lx", + VCPU_EVENT(vcpu, 4, "interrupt: virtio parm:%x,parm64:%llx", inti->ext.ext_params, inti->ext.ext_params2); vcpu->stat.deliver_virtio_interrupt++; rc = put_guest_u16(vcpu, __LC_EXT_INT_CODE, 0x2603); @@ -360,7 +360,7 @@ int kvm_s390_handle_wait(struct kvm_vcpu *vcpu) vcpu->arch.ckc_timer.expires = jiffies + sltime; add_timer(&vcpu->arch.ckc_timer); - VCPU_EVENT(vcpu, 5, "enabled wait timer:%lx jiffies", sltime); + VCPU_EVENT(vcpu, 5, "enabled wait timer:%llx jiffies", sltime); no_timer: spin_lock_bh(&vcpu->arch.local_int.float_int->lock); spin_lock_bh(&vcpu->arch.local_int.lock); @@ -491,7 +491,7 @@ int kvm_s390_inject_vm(struct kvm *kvm, switch (s390int->type) { case KVM_S390_INT_VIRTIO: - VM_EVENT(kvm, 5, "inject: virtio parm:%x,parm64:%lx", + VM_EVENT(kvm, 5, "inject: virtio parm:%x,parm64:%llx", s390int->parm, s390int->parm64); inti->type = s390int->type; inti->ext.ext_params = s390int->parm; diff --git a/arch/s390/kvm/priv.c b/arch/s390/kvm/priv.c index cce40ff2913b..3605df45dd41 100644 --- a/arch/s390/kvm/priv.c +++ b/arch/s390/kvm/priv.c @@ -118,7 +118,7 @@ static int handle_store_cpu_address(struct kvm_vcpu *vcpu) goto out; } - VCPU_EVENT(vcpu, 5, "storing cpu address to %lx", useraddr); + VCPU_EVENT(vcpu, 5, "storing cpu address to %llx", useraddr); out: return 0; } diff --git a/arch/sh/include/asm/mmu.h b/arch/sh/include/asm/mmu.h index fdcb93bc6d11..6c43625bb1a5 100644 --- a/arch/sh/include/asm/mmu.h +++ b/arch/sh/include/asm/mmu.h @@ -9,7 +9,6 @@ typedef struct { mm_context_id_t id; void *vdso; #else - struct vm_list_struct *vmlist; unsigned long end_brk; #endif #ifdef CONFIG_BINFMT_ELF_FDPIC diff --git a/arch/x86/include/asm/bitops.h b/arch/x86/include/asm/bitops.h index 9fa9dcdf344b..e02a359d2aa5 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/early-quirks.c b/arch/x86/kernel/early-quirks.c index 744aa7fc49d5..76b8cd953dee 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 98658f25f542..8fdf06e4edf9 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 */ |