From bc284e5d9d6da48934a177db92bf8e09b96a9cb8 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Mon, 21 Sep 2009 16:56:10 +0000 Subject: [PATCH 01/32] powerpc: perf_event: Log invalid data addresses as all 1s When we take an exception and the SDAR isn't synchronised we currently log 0 as the address. Unfortunately this is a pretty common value, so use ~0UL instead. Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- arch/powerpc/kernel/perf_event.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/powerpc/kernel/perf_event.c b/arch/powerpc/kernel/perf_event.c index bbcbae183e92..73c3b73bf009 100644 --- a/arch/powerpc/kernel/perf_event.c +++ b/arch/powerpc/kernel/perf_event.c @@ -1162,7 +1162,7 @@ static void record_and_restart(struct perf_event *event, unsigned long val, */ if (record) { struct perf_sample_data data = { - .addr = 0, + .addr = ~0ULL, .period = event->hw.last_period, }; From 81cd5ae303e88a1e9d3a3e0f1fe8abd100edde16 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Tue, 27 Oct 2009 18:31:29 +0000 Subject: [PATCH 02/32] powerpc: perf_event: Enable SDAR in continous sample mode In continuous sampling mode we want the SDAR to update. While we can select between dcache misses and ERAT (L1-TLB) misses, a decent default is to enable both. Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- arch/powerpc/include/asm/reg.h | 2 ++ arch/powerpc/kernel/power5+-pmu.c | 4 ---- arch/powerpc/kernel/power5-pmu.c | 6 +----- arch/powerpc/kernel/power6-pmu.c | 2 +- arch/powerpc/kernel/power7-pmu.c | 6 +----- arch/powerpc/kernel/ppc970-pmu.c | 4 ---- 6 files changed, 5 insertions(+), 19 deletions(-) diff --git a/arch/powerpc/include/asm/reg.h b/arch/powerpc/include/asm/reg.h index 6315edc205d8..bc8dd53f718a 100644 --- a/arch/powerpc/include/asm/reg.h +++ b/arch/powerpc/include/asm/reg.h @@ -489,6 +489,8 @@ #define SPRN_MMCR1 798 #define SPRN_MMCRA 0x312 #define MMCRA_SDSYNC 0x80000000UL /* SDAR synced with SIAR */ +#define MMCRA_SDAR_DCACHE_MISS 0x40000000UL +#define MMCRA_SDAR_ERAT_MISS 0x20000000UL #define MMCRA_SIHV 0x10000000UL /* state of MSR HV when SIAR set */ #define MMCRA_SIPR 0x08000000UL /* state of MSR PR when SIAR set */ #define MMCRA_SLOT 0x07000000UL /* SLOT bits (37-39) */ diff --git a/arch/powerpc/kernel/power5+-pmu.c b/arch/powerpc/kernel/power5+-pmu.c index 0f4c1c73a6ad..199de527d411 100644 --- a/arch/powerpc/kernel/power5+-pmu.c +++ b/arch/powerpc/kernel/power5+-pmu.c @@ -72,10 +72,6 @@ #define MMCR1_PMCSEL_SH(n) (MMCR1_PMC1SEL_SH - (n) * 8) #define MMCR1_PMCSEL_MSK 0x7f -/* - * Bits in MMCRA - */ - /* * Layout of constraint bits: * 6666555555555544444444443333333333222222222211111111110000000000 diff --git a/arch/powerpc/kernel/power5-pmu.c b/arch/powerpc/kernel/power5-pmu.c index c351b3a57fbb..98b6a729a9dd 100644 --- a/arch/powerpc/kernel/power5-pmu.c +++ b/arch/powerpc/kernel/power5-pmu.c @@ -72,10 +72,6 @@ #define MMCR1_PMCSEL_SH(n) (MMCR1_PMC1SEL_SH - (n) * 8) #define MMCR1_PMCSEL_MSK 0x7f -/* - * Bits in MMCRA - */ - /* * Layout of constraint bits: * 6666555555555544444444443333333333222222222211111111110000000000 @@ -390,7 +386,7 @@ static int power5_compute_mmcr(u64 event[], int n_ev, unsigned int hwc[], unsigned long mmcr[]) { unsigned long mmcr1 = 0; - unsigned long mmcra = 0; + unsigned long mmcra = MMCRA_SDAR_DCACHE_MISS | MMCRA_SDAR_ERAT_MISS; unsigned int pmc, unit, byte, psel; unsigned int ttm, grp; int i, isbus, bit, grsel; diff --git a/arch/powerpc/kernel/power6-pmu.c b/arch/powerpc/kernel/power6-pmu.c index ca399ba5034c..84a607bda8fb 100644 --- a/arch/powerpc/kernel/power6-pmu.c +++ b/arch/powerpc/kernel/power6-pmu.c @@ -178,7 +178,7 @@ static int p6_compute_mmcr(u64 event[], int n_ev, unsigned int hwc[], unsigned long mmcr[]) { unsigned long mmcr1 = 0; - unsigned long mmcra = 0; + unsigned long mmcra = MMCRA_SDAR_DCACHE_MISS | MMCRA_SDAR_ERAT_MISS; int i; unsigned int pmc, ev, b, u, s, psel; unsigned int ttmset = 0; diff --git a/arch/powerpc/kernel/power7-pmu.c b/arch/powerpc/kernel/power7-pmu.c index 28a4daacdc02..852f7b7f6b40 100644 --- a/arch/powerpc/kernel/power7-pmu.c +++ b/arch/powerpc/kernel/power7-pmu.c @@ -50,10 +50,6 @@ #define MMCR1_PMCSEL_SH(n) (MMCR1_PMC1SEL_SH - (n) * 8) #define MMCR1_PMCSEL_MSK 0xff -/* - * Bits in MMCRA - */ - /* * Layout of constraint bits: * 6666555555555544444444443333333333222222222211111111110000000000 @@ -230,7 +226,7 @@ static int power7_compute_mmcr(u64 event[], int n_ev, unsigned int hwc[], unsigned long mmcr[]) { unsigned long mmcr1 = 0; - unsigned long mmcra = 0; + unsigned long mmcra = MMCRA_SDAR_DCACHE_MISS | MMCRA_SDAR_ERAT_MISS; unsigned int pmc, unit, combine, l2sel, psel; unsigned int pmc_inuse = 0; int i; diff --git a/arch/powerpc/kernel/ppc970-pmu.c b/arch/powerpc/kernel/ppc970-pmu.c index 479574413a93..8eff48e20dba 100644 --- a/arch/powerpc/kernel/ppc970-pmu.c +++ b/arch/powerpc/kernel/ppc970-pmu.c @@ -83,10 +83,6 @@ static short mmcr1_adder_bits[8] = { MMCR1_PMC8_ADDER_SEL_SH }; -/* - * Bits in MMCRA - */ - /* * Layout of constraint bits: * 6666555555555544444444443333333333222222222211111111110000000000 From f7d7986060b2890fc26db6ab5203efbd33aa2497 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Sun, 18 Oct 2009 01:09:29 +0000 Subject: [PATCH 03/32] perf_event: Add alignment-faults and emulation-faults software events Add two more software events that are common to many cpus. Alignment faults: When a load or store is not aligned properly. Emulation faults: When an instruction is emulated in software. Both cause a very significant slowdown (100x or worse), so identifying and fixing them is very important. Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- include/linux/perf_counter.h | 2 ++ include/linux/perf_event.h | 2 ++ kernel/perf_event.c | 2 ++ tools/perf/design.txt | 2 ++ tools/perf/util/parse-events.c | 4 ++++ 5 files changed, 12 insertions(+) diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h index 7b7fbf433cff..d6b95d1e79f0 100644 --- a/include/linux/perf_counter.h +++ b/include/linux/perf_counter.h @@ -106,6 +106,8 @@ enum perf_sw_ids { PERF_COUNT_SW_CPU_MIGRATIONS = 4, PERF_COUNT_SW_PAGE_FAULTS_MIN = 5, PERF_COUNT_SW_PAGE_FAULTS_MAJ = 6, + PERF_COUNT_SW_ALIGNMENT_FAULTS = 7, + PERF_COUNT_SW_EMULATION_FAULTS = 8, PERF_COUNT_SW_MAX, /* non-ABI */ }; diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 2e6d95f97419..a33707a3a788 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -102,6 +102,8 @@ enum perf_sw_ids { PERF_COUNT_SW_CPU_MIGRATIONS = 4, PERF_COUNT_SW_PAGE_FAULTS_MIN = 5, PERF_COUNT_SW_PAGE_FAULTS_MAJ = 6, + PERF_COUNT_SW_ALIGNMENT_FAULTS = 7, + PERF_COUNT_SW_EMULATION_FAULTS = 8, PERF_COUNT_SW_MAX, /* non-ABI */ }; diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 9d0b5c665883..0683b33cbb28 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -4186,6 +4186,8 @@ static const struct pmu *sw_perf_event_init(struct perf_event *event) case PERF_COUNT_SW_PAGE_FAULTS_MAJ: case PERF_COUNT_SW_CONTEXT_SWITCHES: case PERF_COUNT_SW_CPU_MIGRATIONS: + case PERF_COUNT_SW_ALIGNMENT_FAULTS: + case PERF_COUNT_SW_EMULATION_FAULTS: if (!event->parent) { atomic_inc(&perf_swevent_enabled[event_id]); event->destroy = sw_perf_event_destroy; diff --git a/tools/perf/design.txt b/tools/perf/design.txt index fdd42a824c98..f000c30877ac 100644 --- a/tools/perf/design.txt +++ b/tools/perf/design.txt @@ -137,6 +137,8 @@ enum sw_event_ids { PERF_COUNT_SW_CPU_MIGRATIONS = 4, PERF_COUNT_SW_PAGE_FAULTS_MIN = 5, PERF_COUNT_SW_PAGE_FAULTS_MAJ = 6, + PERF_COUNT_SW_ALIGNMENT_FAULTS = 7, + PERF_COUNT_SW_EMULATION_FAULTS = 8, }; Counters of the type PERF_TYPE_TRACEPOINT are available when the ftrace event diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 8cfb48cbbea0..34bd84423933 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -46,6 +46,8 @@ static struct event_symbol event_symbols[] = { { CSW(PAGE_FAULTS_MAJ), "major-faults", "" }, { CSW(CONTEXT_SWITCHES), "context-switches", "cs" }, { CSW(CPU_MIGRATIONS), "cpu-migrations", "migrations" }, + { CSW(ALIGNMENT_FAULTS), "alignment-faults", "" }, + { CSW(EMULATION_FAULTS), "emulation-faults", "" }, }; #define __PERF_EVENT_FIELD(config, name) \ @@ -74,6 +76,8 @@ static const char *sw_event_names[] = { "CPU-migrations", "minor-faults", "major-faults", + "alignment-faults", + "emulation-faults", }; #define MAX_ALIASES 8 From eecff81d1fcda22cd0029d11fe2a71dceed11dad Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Tue, 27 Oct 2009 18:46:55 +0000 Subject: [PATCH 04/32] powerpc: Create PPC_WARN_ALIGNMENT to match PPC_WARN_EMULATED perf_event wants a separate event for alignment and emulation faults, so create another emulation event. This will make it easy to hook in perf_event at one spot. We pass in regs which will be required for these events. Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- arch/powerpc/include/asm/emulated_ops.h | 7 +++++-- arch/powerpc/kernel/align.c | 12 ++++++------ arch/powerpc/kernel/traps.c | 18 +++++++++--------- 3 files changed, 20 insertions(+), 17 deletions(-) diff --git a/arch/powerpc/include/asm/emulated_ops.h b/arch/powerpc/include/asm/emulated_ops.h index 9154e8526732..640c4e456aa5 100644 --- a/arch/powerpc/include/asm/emulated_ops.h +++ b/arch/powerpc/include/asm/emulated_ops.h @@ -57,7 +57,7 @@ extern u32 ppc_warn_emulated; extern void ppc_warn_emulated_print(const char *type); -#define PPC_WARN_EMULATED(type) \ +#define __PPC_WARN_EMULATED(type) \ do { \ atomic_inc(&ppc_emulated.type.val); \ if (ppc_warn_emulated) \ @@ -66,8 +66,11 @@ extern void ppc_warn_emulated_print(const char *type); #else /* !CONFIG_PPC_EMULATED_STATS */ -#define PPC_WARN_EMULATED(type) do { } while (0) +#define __PPC_WARN_EMULATED(type) do { } while (0) #endif /* !CONFIG_PPC_EMULATED_STATS */ +#define PPC_WARN_EMULATED(type, regs) __PPC_WARN_EMULATED(type) +#define PPC_WARN_ALIGNMENT(type, regs) __PPC_WARN_EMULATED(type) + #endif /* _ASM_POWERPC_EMULATED_OPS_H */ diff --git a/arch/powerpc/kernel/align.c b/arch/powerpc/kernel/align.c index a5b632e52fae..3839839f83c7 100644 --- a/arch/powerpc/kernel/align.c +++ b/arch/powerpc/kernel/align.c @@ -732,7 +732,7 @@ int fix_alignment(struct pt_regs *regs) #ifdef CONFIG_SPE if ((instr >> 26) == 0x4) { - PPC_WARN_EMULATED(spe); + PPC_WARN_ALIGNMENT(spe, regs); return emulate_spe(regs, reg, instr); } #endif @@ -786,7 +786,7 @@ int fix_alignment(struct pt_regs *regs) flags |= SPLT; nb = 8; } - PPC_WARN_EMULATED(vsx); + PPC_WARN_ALIGNMENT(vsx, regs); return emulate_vsx(addr, reg, areg, regs, flags, nb); } #endif @@ -794,7 +794,7 @@ int fix_alignment(struct pt_regs *regs) * the exception of DCBZ which is handled as a special case here */ if (instr == DCBZ) { - PPC_WARN_EMULATED(dcbz); + PPC_WARN_ALIGNMENT(dcbz, regs); return emulate_dcbz(regs, addr); } if (unlikely(nb == 0)) @@ -804,7 +804,7 @@ int fix_alignment(struct pt_regs *regs) * function */ if (flags & M) { - PPC_WARN_EMULATED(multiple); + PPC_WARN_ALIGNMENT(multiple, regs); return emulate_multiple(regs, addr, reg, nb, flags, instr, swiz); } @@ -825,11 +825,11 @@ int fix_alignment(struct pt_regs *regs) /* Special case for 16-byte FP loads and stores */ if (nb == 16) { - PPC_WARN_EMULATED(fp_pair); + PPC_WARN_ALIGNMENT(fp_pair, regs); return emulate_fp_pair(addr, reg, flags); } - PPC_WARN_EMULATED(unaligned); + PPC_WARN_ALIGNMENT(unaligned, regs); /* If we are loading, get the data from user space, else * get it from register values diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c index 6f0ae1a9bfae..9d1f9354d6ca 100644 --- a/arch/powerpc/kernel/traps.c +++ b/arch/powerpc/kernel/traps.c @@ -759,7 +759,7 @@ static int emulate_instruction(struct pt_regs *regs) /* Emulate the mfspr rD, PVR. */ if ((instword & PPC_INST_MFSPR_PVR_MASK) == PPC_INST_MFSPR_PVR) { - PPC_WARN_EMULATED(mfpvr); + PPC_WARN_EMULATED(mfpvr, regs); rd = (instword >> 21) & 0x1f; regs->gpr[rd] = mfspr(SPRN_PVR); return 0; @@ -767,7 +767,7 @@ static int emulate_instruction(struct pt_regs *regs) /* Emulating the dcba insn is just a no-op. */ if ((instword & PPC_INST_DCBA_MASK) == PPC_INST_DCBA) { - PPC_WARN_EMULATED(dcba); + PPC_WARN_EMULATED(dcba, regs); return 0; } @@ -776,7 +776,7 @@ static int emulate_instruction(struct pt_regs *regs) int shift = (instword >> 21) & 0x1c; unsigned long msk = 0xf0000000UL >> shift; - PPC_WARN_EMULATED(mcrxr); + PPC_WARN_EMULATED(mcrxr, regs); regs->ccr = (regs->ccr & ~msk) | ((regs->xer >> shift) & msk); regs->xer &= ~0xf0000000UL; return 0; @@ -784,19 +784,19 @@ static int emulate_instruction(struct pt_regs *regs) /* Emulate load/store string insn. */ if ((instword & PPC_INST_STRING_GEN_MASK) == PPC_INST_STRING) { - PPC_WARN_EMULATED(string); + PPC_WARN_EMULATED(string, regs); return emulate_string_inst(regs, instword); } /* Emulate the popcntb (Population Count Bytes) instruction. */ if ((instword & PPC_INST_POPCNTB_MASK) == PPC_INST_POPCNTB) { - PPC_WARN_EMULATED(popcntb); + PPC_WARN_EMULATED(popcntb, regs); return emulate_popcntb_inst(regs, instword); } /* Emulate isel (Integer Select) instruction */ if ((instword & PPC_INST_ISEL_MASK) == PPC_INST_ISEL) { - PPC_WARN_EMULATED(isel); + PPC_WARN_EMULATED(isel, regs); return emulate_isel(regs, instword); } @@ -995,7 +995,7 @@ void SoftwareEmulation(struct pt_regs *regs) #ifdef CONFIG_MATH_EMULATION errcode = do_mathemu(regs); if (errcode >= 0) - PPC_WARN_EMULATED(math); + PPC_WARN_EMULATED(math, regs); switch (errcode) { case 0: @@ -1018,7 +1018,7 @@ void SoftwareEmulation(struct pt_regs *regs) #elif defined(CONFIG_8XX_MINIMAL_FPEMU) errcode = Soft_emulate_8xx(regs); if (errcode >= 0) - PPC_WARN_EMULATED(8xx); + PPC_WARN_EMULATED(8xx, regs); switch (errcode) { case 0: @@ -1129,7 +1129,7 @@ void altivec_assist_exception(struct pt_regs *regs) flush_altivec_to_thread(current); - PPC_WARN_EMULATED(altivec); + PPC_WARN_EMULATED(altivec, regs); err = emulate_altivec(regs); if (err == 0) { regs->nip += 4; /* skip emulated instruction */ From 196f02bf900c5eb6f85d889c4f70e7cc11fda7e8 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Sun, 18 Oct 2009 01:13:00 +0000 Subject: [PATCH 05/32] powerpc: perf_event: Add alignment-faults and emulation-faults software events Hook up the alignment-faults and emulation-faults events for powerpc. Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- arch/powerpc/include/asm/emulated_ops.h | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/arch/powerpc/include/asm/emulated_ops.h b/arch/powerpc/include/asm/emulated_ops.h index 640c4e456aa5..f0fb4fc1f6e6 100644 --- a/arch/powerpc/include/asm/emulated_ops.h +++ b/arch/powerpc/include/asm/emulated_ops.h @@ -19,6 +19,7 @@ #define _ASM_POWERPC_EMULATED_OPS_H #include +#include #ifdef CONFIG_PPC_EMULATED_STATS @@ -70,7 +71,18 @@ extern void ppc_warn_emulated_print(const char *type); #endif /* !CONFIG_PPC_EMULATED_STATS */ -#define PPC_WARN_EMULATED(type, regs) __PPC_WARN_EMULATED(type) -#define PPC_WARN_ALIGNMENT(type, regs) __PPC_WARN_EMULATED(type) +#define PPC_WARN_EMULATED(type, regs) \ + do { \ + perf_sw_event(PERF_COUNT_SW_EMULATION_FAULTS, \ + 1, 0, regs, 0); \ + __PPC_WARN_EMULATED(type); \ + } while (0) + +#define PPC_WARN_ALIGNMENT(type, regs) \ + do { \ + perf_sw_event(PERF_COUNT_SW_ALIGNMENT_FAULTS, \ + 1, 0, regs, regs->dar); \ + __PPC_WARN_EMULATED(type); \ + } while (0) #endif /* _ASM_POWERPC_EMULATED_OPS_H */ From 1bf4af165050d90ea6659ffb2536ec8ca783aab5 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Mon, 26 Oct 2009 18:47:42 +0000 Subject: [PATCH 06/32] powerpc: tracing: Add powerpc tracepoints for interrupt entry and exit This adds powerpc-specific tracepoints for interrupt entry and exit. While we already have generic irq_handler_entry and irq_handler_exit tracepoints there are cases on our virtualised powerpc machines where an interrupt is presented to the OS, but subsequently handled by the hypervisor. This means no OS interrupt handler is invoked. Here is an example on a POWER6 machine with the patch below applied: -0 [006] 3243.949840744: irq_entry: pt_regs=c0000000ce31fb10 -0 [006] 3243.949850520: irq_exit: pt_regs=c0000000ce31fb10 -0 [007] 3243.950218208: irq_entry: pt_regs=c0000000ce323b10 -0 [007] 3243.950224080: irq_exit: pt_regs=c0000000ce323b10 -0 [000] 3244.021879320: irq_entry: pt_regs=c000000000a63aa0 -0 [000] 3244.021883616: irq_handler_entry: irq=87 handler=eth0 -0 [000] 3244.021887328: irq_handler_exit: irq=87 return=handled -0 [000] 3244.021897408: irq_exit: pt_regs=c000000000a63aa0 Here we see two phantom interrupts (no handler was invoked), followed by a real interrupt for eth0. Without the tracepoints in this patch we would have missed the phantom interrupts. Signed-off-by: Anton Blanchard Acked-by: Steven Rostedt Signed-off-by: Paul Mackerras --- arch/powerpc/include/asm/trace.h | 53 ++++++++++++++++++++++++++++++++ arch/powerpc/kernel/irq.c | 6 ++++ 2 files changed, 59 insertions(+) create mode 100644 arch/powerpc/include/asm/trace.h diff --git a/arch/powerpc/include/asm/trace.h b/arch/powerpc/include/asm/trace.h new file mode 100644 index 000000000000..187696da5ae7 --- /dev/null +++ b/arch/powerpc/include/asm/trace.h @@ -0,0 +1,53 @@ +#undef TRACE_SYSTEM +#define TRACE_SYSTEM powerpc + +#if !defined(_TRACE_POWERPC_H) || defined(TRACE_HEADER_MULTI_READ) +#define _TRACE_POWERPC_H + +#include + +struct pt_regs; + +TRACE_EVENT(irq_entry, + + TP_PROTO(struct pt_regs *regs), + + TP_ARGS(regs), + + TP_STRUCT__entry( + __field(struct pt_regs *, regs) + ), + + TP_fast_assign( + __entry->regs = regs; + ), + + TP_printk("pt_regs=%p", __entry->regs) +); + +TRACE_EVENT(irq_exit, + + TP_PROTO(struct pt_regs *regs), + + TP_ARGS(regs), + + TP_STRUCT__entry( + __field(struct pt_regs *, regs) + ), + + TP_fast_assign( + __entry->regs = regs; + ), + + TP_printk("pt_regs=%p", __entry->regs) +); + +#endif /* _TRACE_POWERPC_H */ + +#undef TRACE_INCLUDE_PATH +#undef TRACE_INCLUDE_FILE + +#define TRACE_INCLUDE_PATH asm +#define TRACE_INCLUDE_FILE trace + +#include diff --git a/arch/powerpc/kernel/irq.c b/arch/powerpc/kernel/irq.c index e5d121177984..02a334662cc0 100644 --- a/arch/powerpc/kernel/irq.c +++ b/arch/powerpc/kernel/irq.c @@ -70,6 +70,8 @@ #include #include #endif +#define CREATE_TRACE_POINTS +#include int __irq_offset_value; static int ppc_spurious_interrupts; @@ -325,6 +327,8 @@ void do_IRQ(struct pt_regs *regs) struct pt_regs *old_regs = set_irq_regs(regs); unsigned int irq; + trace_irq_entry(regs); + irq_enter(); check_stack_overflow(); @@ -348,6 +352,8 @@ void do_IRQ(struct pt_regs *regs) timer_interrupt(regs); } #endif + + trace_irq_exit(regs); } void __init init_IRQ(void) From 6795b85c6a4f690e61e7be31aa150d945c723fb5 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Mon, 26 Oct 2009 18:49:14 +0000 Subject: [PATCH 07/32] powerpc: tracing: Add powerpc tracepoints for timer entry and exit We can monitor the effectiveness of our power management of both the kernel and hypervisor by probing the timer interrupt. For example, on this box we see 10.37s timer interrupts on an idle core: -0 [010] 3900.671297: timer_interrupt_entry: pt_regs=c0000000ce1e7b10 -0 [010] 3900.671302: timer_interrupt_exit: pt_regs=c0000000ce1e7b10 -0 [010] 3911.042963: timer_interrupt_entry: pt_regs=c0000000ce1e7b10 -0 [010] 3911.042968: timer_interrupt_exit: pt_regs=c0000000ce1e7b10 -0 [010] 3921.414630: timer_interrupt_entry: pt_regs=c0000000ce1e7b10 -0 [010] 3921.414635: timer_interrupt_exit: pt_regs=c0000000ce1e7b10 Since we have a 207MHz decrementer it will go negative and fire every 10.37s even if Linux is completely idle. Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- arch/powerpc/include/asm/trace.h | 34 ++++++++++++++++++++++++++++++++ arch/powerpc/kernel/time.c | 6 ++++++ 2 files changed, 40 insertions(+) diff --git a/arch/powerpc/include/asm/trace.h b/arch/powerpc/include/asm/trace.h index 187696da5ae7..b558c31d409e 100644 --- a/arch/powerpc/include/asm/trace.h +++ b/arch/powerpc/include/asm/trace.h @@ -42,6 +42,40 @@ TRACE_EVENT(irq_exit, TP_printk("pt_regs=%p", __entry->regs) ); +TRACE_EVENT(timer_interrupt_entry, + + TP_PROTO(struct pt_regs *regs), + + TP_ARGS(regs), + + TP_STRUCT__entry( + __field(struct pt_regs *, regs) + ), + + TP_fast_assign( + __entry->regs = regs; + ), + + TP_printk("pt_regs=%p", __entry->regs) +); + +TRACE_EVENT(timer_interrupt_exit, + + TP_PROTO(struct pt_regs *regs), + + TP_ARGS(regs), + + TP_STRUCT__entry( + __field(struct pt_regs *, regs) + ), + + TP_fast_assign( + __entry->regs = regs; + ), + + TP_printk("pt_regs=%p", __entry->regs) +); + #endif /* _TRACE_POWERPC_H */ #undef TRACE_INCLUDE_PATH diff --git a/arch/powerpc/kernel/time.c b/arch/powerpc/kernel/time.c index 92dc844299b6..d6e88df4630c 100644 --- a/arch/powerpc/kernel/time.c +++ b/arch/powerpc/kernel/time.c @@ -54,6 +54,7 @@ #include #include #include +#include #include #include @@ -571,6 +572,8 @@ void timer_interrupt(struct pt_regs * regs) struct clock_event_device *evt = &decrementer->event; u64 now; + trace_timer_interrupt_entry(regs); + /* Ensure a positive value is written to the decrementer, or else * some CPUs will continuue to take decrementer exceptions */ set_dec(DECREMENTER_MAX); @@ -590,6 +593,7 @@ void timer_interrupt(struct pt_regs * regs) now = decrementer->next_tb - now; if (now <= DECREMENTER_MAX) set_dec((int)now); + trace_timer_interrupt_exit(regs); return; } old_regs = set_irq_regs(regs); @@ -620,6 +624,8 @@ void timer_interrupt(struct pt_regs * regs) irq_exit(); set_irq_regs(old_regs); + + trace_timer_interrupt_exit(regs); } void wakeup_decrementer(void) From c8cd093a6e9f96ea6b871576fd4e46d7c818bb89 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Mon, 26 Oct 2009 18:50:29 +0000 Subject: [PATCH 08/32] powerpc: tracing: Add hypervisor call tracepoints Add hcall_entry and hcall_exit tracepoints. This replaces the inline assembly HCALL_STATS code and converts it to use the new tracepoints. To keep the disabled case as quick as possible, we embed a status word in the TOC so we can get at it with a single load. By doing so we keep the overhead at a minimum. Time taken for a null hcall: No tracepoint code: 135.79 cycles Disabled tracepoints: 137.95 cycles For reference, before this patch enabling HCALL_STATS resulted in a null hcall of 201.44 cycles! Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- arch/powerpc/Kconfig.debug | 2 +- arch/powerpc/include/asm/hvcall.h | 2 + arch/powerpc/include/asm/trace.h | 45 +++++++++ arch/powerpc/platforms/pseries/hvCall.S | 101 +++++++++++-------- arch/powerpc/platforms/pseries/hvCall_inst.c | 37 +++++++ arch/powerpc/platforms/pseries/lpar.c | 32 ++++++ 6 files changed, 175 insertions(+), 44 deletions(-) diff --git a/arch/powerpc/Kconfig.debug b/arch/powerpc/Kconfig.debug index 3b1005185390..bf3382f1904d 100644 --- a/arch/powerpc/Kconfig.debug +++ b/arch/powerpc/Kconfig.debug @@ -46,7 +46,7 @@ config DEBUG_STACK_USAGE config HCALL_STATS bool "Hypervisor call instrumentation" - depends on PPC_PSERIES && DEBUG_FS + depends on PPC_PSERIES && DEBUG_FS && TRACEPOINTS help Adds code to keep track of the number of hypervisor calls made and the amount of time spent in hypervisor calls. Wall time spent in diff --git a/arch/powerpc/include/asm/hvcall.h b/arch/powerpc/include/asm/hvcall.h index 6251a4b10be7..c27caac47ad1 100644 --- a/arch/powerpc/include/asm/hvcall.h +++ b/arch/powerpc/include/asm/hvcall.h @@ -274,6 +274,8 @@ struct hcall_stats { unsigned long num_calls; /* number of calls (on this CPU) */ unsigned long tb_total; /* total wall time (mftb) of calls. */ unsigned long purr_total; /* total cpu time (PURR) of calls. */ + unsigned long tb_start; + unsigned long purr_start; }; #define HCALL_STAT_ARRAY_SIZE ((MAX_HCALL_OPCODE >> 2) + 1) diff --git a/arch/powerpc/include/asm/trace.h b/arch/powerpc/include/asm/trace.h index b558c31d409e..9b01c0e43b55 100644 --- a/arch/powerpc/include/asm/trace.h +++ b/arch/powerpc/include/asm/trace.h @@ -76,6 +76,51 @@ TRACE_EVENT(timer_interrupt_exit, TP_printk("pt_regs=%p", __entry->regs) ); +#ifdef CONFIG_PPC_PSERIES +extern void hcall_tracepoint_regfunc(void); +extern void hcall_tracepoint_unregfunc(void); + +TRACE_EVENT_FN(hcall_entry, + + TP_PROTO(unsigned long opcode), + + TP_ARGS(opcode), + + TP_STRUCT__entry( + __field(unsigned long, opcode) + ), + + TP_fast_assign( + __entry->opcode = opcode; + ), + + TP_printk("opcode=%lu", __entry->opcode), + + hcall_tracepoint_regfunc, hcall_tracepoint_unregfunc +); + +TRACE_EVENT_FN(hcall_exit, + + TP_PROTO(unsigned long opcode, unsigned long retval), + + TP_ARGS(opcode, retval), + + TP_STRUCT__entry( + __field(unsigned long, opcode) + __field(unsigned long, retval) + ), + + TP_fast_assign( + __entry->opcode = opcode; + __entry->retval = retval; + ), + + TP_printk("opcode=%lu retval=%lu", __entry->opcode, __entry->retval), + + hcall_tracepoint_regfunc, hcall_tracepoint_unregfunc +); +#endif + #endif /* _TRACE_POWERPC_H */ #undef TRACE_INCLUDE_PATH diff --git a/arch/powerpc/platforms/pseries/hvCall.S b/arch/powerpc/platforms/pseries/hvCall.S index c1427b3634ec..01e95ab18d35 100644 --- a/arch/powerpc/platforms/pseries/hvCall.S +++ b/arch/powerpc/platforms/pseries/hvCall.S @@ -14,20 +14,54 @@ #define STK_PARM(i) (48 + ((i)-3)*8) -#ifdef CONFIG_HCALL_STATS +#ifdef CONFIG_TRACEPOINTS + + .section ".toc","aw" + + .globl hcall_tracepoint_refcount +hcall_tracepoint_refcount: + .llong 0 + + .section ".text" + /* * precall must preserve all registers. use unused STK_PARM() - * areas to save snapshots and opcode. + * areas to save snapshots and opcode. We branch around this + * in early init (eg when populating the MMU hashtable) by using an + * unconditional cpu feature. */ #define HCALL_INST_PRECALL \ - std r3,STK_PARM(r3)(r1); /* save opcode */ \ - mftb r0; /* get timebase and */ \ - std r0,STK_PARM(r5)(r1); /* save for later */ \ BEGIN_FTR_SECTION; \ - mfspr r0,SPRN_PURR; /* get PURR and */ \ - std r0,STK_PARM(r6)(r1); /* save for later */ \ -END_FTR_SECTION_IFSET(CPU_FTR_PURR); - + b 1f; \ +END_FTR_SECTION(0, 1); \ + ld r12,hcall_tracepoint_refcount@toc(r2); \ + cmpdi r12,0; \ + beq+ 1f; \ + mflr r0; \ + std r3,STK_PARM(r3)(r1); \ + std r4,STK_PARM(r4)(r1); \ + std r5,STK_PARM(r5)(r1); \ + std r6,STK_PARM(r6)(r1); \ + std r7,STK_PARM(r7)(r1); \ + std r8,STK_PARM(r8)(r1); \ + std r9,STK_PARM(r9)(r1); \ + std r10,STK_PARM(r10)(r1); \ + std r0,16(r1); \ + stdu r1,-STACK_FRAME_OVERHEAD(r1); \ + bl .__trace_hcall_entry; \ + addi r1,r1,STACK_FRAME_OVERHEAD; \ + ld r0,16(r1); \ + ld r3,STK_PARM(r3)(r1); \ + ld r4,STK_PARM(r4)(r1); \ + ld r5,STK_PARM(r5)(r1); \ + ld r6,STK_PARM(r6)(r1); \ + ld r7,STK_PARM(r7)(r1); \ + ld r8,STK_PARM(r8)(r1); \ + ld r9,STK_PARM(r9)(r1); \ + ld r10,STK_PARM(r10)(r1); \ + mtlr r0; \ +1: + /* * postcall is performed immediately before function return which * allows liberal use of volatile registers. We branch around this @@ -38,40 +72,21 @@ END_FTR_SECTION_IFSET(CPU_FTR_PURR); BEGIN_FTR_SECTION; \ b 1f; \ END_FTR_SECTION(0, 1); \ - ld r4,STK_PARM(r3)(r1); /* validate opcode */ \ - cmpldi cr7,r4,MAX_HCALL_OPCODE; \ - bgt- cr7,1f; \ - \ - /* get time and PURR snapshots after hcall */ \ - mftb r7; /* timebase after */ \ -BEGIN_FTR_SECTION; \ - mfspr r8,SPRN_PURR; /* PURR after */ \ - ld r6,STK_PARM(r6)(r1); /* PURR before */ \ - subf r6,r6,r8; /* delta */ \ -END_FTR_SECTION_IFSET(CPU_FTR_PURR); \ - ld r5,STK_PARM(r5)(r1); /* timebase before */ \ - subf r5,r5,r7; /* time delta */ \ - \ - /* calculate address of stat structure r4 = opcode */ \ - srdi r4,r4,2; /* index into array */ \ - mulli r4,r4,HCALL_STAT_SIZE; \ - LOAD_REG_ADDR(r7, per_cpu__hcall_stats); \ - add r4,r4,r7; \ - ld r7,PACA_DATA_OFFSET(r13); /* per cpu offset */ \ - add r4,r4,r7; \ - \ - /* update stats */ \ - ld r7,HCALL_STAT_CALLS(r4); /* count */ \ - addi r7,r7,1; \ - std r7,HCALL_STAT_CALLS(r4); \ - ld r7,HCALL_STAT_TB(r4); /* timebase */ \ - add r7,r7,r5; \ - std r7,HCALL_STAT_TB(r4); \ -BEGIN_FTR_SECTION; \ - ld r7,HCALL_STAT_PURR(r4); /* PURR */ \ - add r7,r7,r6; \ - std r7,HCALL_STAT_PURR(r4); \ -END_FTR_SECTION_IFSET(CPU_FTR_PURR); \ + ld r12,hcall_tracepoint_refcount@toc(r2); \ + cmpdi r12,0; \ + beq+ 1f; \ + mflr r0; \ + ld r6,STK_PARM(r3)(r1); \ + std r3,STK_PARM(r3)(r1); \ + mr r4,r3; \ + mr r3,r6; \ + std r0,16(r1); \ + stdu r1,-STACK_FRAME_OVERHEAD(r1); \ + bl .__trace_hcall_exit; \ + addi r1,r1,STACK_FRAME_OVERHEAD; \ + ld r0,16(r1); \ + ld r3,STK_PARM(r3)(r1); \ + mtlr r0; \ 1: #else #define HCALL_INST_PRECALL diff --git a/arch/powerpc/platforms/pseries/hvCall_inst.c b/arch/powerpc/platforms/pseries/hvCall_inst.c index 3631a4f277eb..e44e1035f133 100644 --- a/arch/powerpc/platforms/pseries/hvCall_inst.c +++ b/arch/powerpc/platforms/pseries/hvCall_inst.c @@ -26,6 +26,7 @@ #include #include #include +#include DEFINE_PER_CPU(struct hcall_stats[HCALL_STAT_ARRAY_SIZE], hcall_stats); @@ -100,6 +101,34 @@ static const struct file_operations hcall_inst_seq_fops = { #define HCALL_ROOT_DIR "hcall_inst" #define CPU_NAME_BUF_SIZE 32 + +static void probe_hcall_entry(unsigned long opcode) +{ + struct hcall_stats *h; + + if (opcode > MAX_HCALL_OPCODE) + return; + + h = &get_cpu_var(hcall_stats)[opcode / 4]; + h->tb_start = mftb(); + h->purr_start = mfspr(SPRN_PURR); +} + +static void probe_hcall_exit(unsigned long opcode, unsigned long retval) +{ + struct hcall_stats *h; + + if (opcode > MAX_HCALL_OPCODE) + return; + + h = &__get_cpu_var(hcall_stats)[opcode / 4]; + h->num_calls++; + h->tb_total = mftb() - h->tb_start; + h->purr_total = mfspr(SPRN_PURR) - h->purr_start; + + put_cpu_var(hcall_stats); +} + static int __init hcall_inst_init(void) { struct dentry *hcall_root; @@ -110,6 +139,14 @@ static int __init hcall_inst_init(void) if (!firmware_has_feature(FW_FEATURE_LPAR)) return 0; + if (register_trace_hcall_entry(probe_hcall_entry)) + return -EINVAL; + + if (register_trace_hcall_exit(probe_hcall_exit)) { + unregister_trace_hcall_entry(probe_hcall_entry); + return -EINVAL; + } + hcall_root = debugfs_create_dir(HCALL_ROOT_DIR, NULL); if (!hcall_root) return -ENOMEM; diff --git a/arch/powerpc/platforms/pseries/lpar.c b/arch/powerpc/platforms/pseries/lpar.c index 903eb9eec687..4b7b6e8e32de 100644 --- a/arch/powerpc/platforms/pseries/lpar.c +++ b/arch/powerpc/platforms/pseries/lpar.c @@ -39,6 +39,7 @@ #include #include #include +#include #include "plpar_wrappers.h" #include "pseries.h" @@ -661,3 +662,34 @@ void arch_free_page(struct page *page, int order) EXPORT_SYMBOL(arch_free_page); #endif + +#ifdef CONFIG_TRACEPOINTS +/* + * We optimise our hcall path by placing hcall_tracepoint_refcount + * directly in the TOC so we can check if the hcall tracepoints are + * enabled via a single load. + */ + +/* NB: reg/unreg are called while guarded with the tracepoints_mutex */ +extern long hcall_tracepoint_refcount; + +void hcall_tracepoint_regfunc(void) +{ + hcall_tracepoint_refcount++; +} + +void hcall_tracepoint_unregfunc(void) +{ + hcall_tracepoint_refcount--; +} + +void __trace_hcall_entry(unsigned long opcode) +{ + trace_hcall_entry(opcode); +} + +void __trace_hcall_exit(long opcode, unsigned long retval) +{ + trace_hcall_exit(opcode, retval); +} +#endif From 6f26353ca29e96475208bce673efb6a2c58b73f2 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Mon, 26 Oct 2009 18:51:09 +0000 Subject: [PATCH 09/32] powerpc: tracing: Give hypervisor call tracepoints access to arguments While most users of the hcall tracepoints will only want the opcode and return code, some will want all the arguments. To avoid the complexity of using varargs we pass a pointer to the register save area, which contains all the arguments. Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- arch/powerpc/include/asm/trace.h | 9 +++--- arch/powerpc/platforms/pseries/hvCall.S | 31 +++++++++++++------- arch/powerpc/platforms/pseries/hvCall_inst.c | 5 ++-- arch/powerpc/platforms/pseries/lpar.c | 9 +++--- 4 files changed, 34 insertions(+), 20 deletions(-) diff --git a/arch/powerpc/include/asm/trace.h b/arch/powerpc/include/asm/trace.h index 9b01c0e43b55..cbe2297d68b6 100644 --- a/arch/powerpc/include/asm/trace.h +++ b/arch/powerpc/include/asm/trace.h @@ -82,9 +82,9 @@ extern void hcall_tracepoint_unregfunc(void); TRACE_EVENT_FN(hcall_entry, - TP_PROTO(unsigned long opcode), + TP_PROTO(unsigned long opcode, unsigned long *args), - TP_ARGS(opcode), + TP_ARGS(opcode, args), TP_STRUCT__entry( __field(unsigned long, opcode) @@ -101,9 +101,10 @@ TRACE_EVENT_FN(hcall_entry, TRACE_EVENT_FN(hcall_exit, - TP_PROTO(unsigned long opcode, unsigned long retval), + TP_PROTO(unsigned long opcode, unsigned long retval, + unsigned long *retbuf), - TP_ARGS(opcode, retval), + TP_ARGS(opcode, retval, retbuf), TP_STRUCT__entry( __field(unsigned long, opcode) diff --git a/arch/powerpc/platforms/pseries/hvCall.S b/arch/powerpc/platforms/pseries/hvCall.S index 01e95ab18d35..383a5d0e9818 100644 --- a/arch/powerpc/platforms/pseries/hvCall.S +++ b/arch/powerpc/platforms/pseries/hvCall.S @@ -30,7 +30,7 @@ hcall_tracepoint_refcount: * in early init (eg when populating the MMU hashtable) by using an * unconditional cpu feature. */ -#define HCALL_INST_PRECALL \ +#define HCALL_INST_PRECALL(FIRST_REG) \ BEGIN_FTR_SECTION; \ b 1f; \ END_FTR_SECTION(0, 1); \ @@ -47,6 +47,7 @@ END_FTR_SECTION(0, 1); \ std r9,STK_PARM(r9)(r1); \ std r10,STK_PARM(r10)(r1); \ std r0,16(r1); \ + addi r4,r1,STK_PARM(FIRST_REG); \ stdu r1,-STACK_FRAME_OVERHEAD(r1); \ bl .__trace_hcall_entry; \ addi r1,r1,STACK_FRAME_OVERHEAD; \ @@ -68,7 +69,7 @@ END_FTR_SECTION(0, 1); \ * in early init (eg when populating the MMU hashtable) by using an * unconditional cpu feature. */ -#define HCALL_INST_POSTCALL \ +#define __HCALL_INST_POSTCALL \ BEGIN_FTR_SECTION; \ b 1f; \ END_FTR_SECTION(0, 1); \ @@ -88,9 +89,19 @@ END_FTR_SECTION(0, 1); \ ld r3,STK_PARM(r3)(r1); \ mtlr r0; \ 1: + +#define HCALL_INST_POSTCALL_NORETS \ + li r5,0; \ + __HCALL_INST_POSTCALL + +#define HCALL_INST_POSTCALL(BUFREG) \ + mr r5,BUFREG; \ + __HCALL_INST_POSTCALL + #else -#define HCALL_INST_PRECALL -#define HCALL_INST_POSTCALL +#define HCALL_INST_PRECALL(FIRST_ARG) +#define HCALL_INST_POSTCALL_NORETS +#define HCALL_INST_POSTCALL(BUFREG) #endif .text @@ -101,11 +112,11 @@ _GLOBAL(plpar_hcall_norets) mfcr r0 stw r0,8(r1) - HCALL_INST_PRECALL + HCALL_INST_PRECALL(r4) HVSC /* invoke the hypervisor */ - HCALL_INST_POSTCALL + HCALL_INST_POSTCALL_NORETS lwz r0,8(r1) mtcrf 0xff,r0 @@ -117,7 +128,7 @@ _GLOBAL(plpar_hcall) mfcr r0 stw r0,8(r1) - HCALL_INST_PRECALL + HCALL_INST_PRECALL(r5) std r4,STK_PARM(r4)(r1) /* Save ret buffer */ @@ -136,7 +147,7 @@ _GLOBAL(plpar_hcall) std r6, 16(r12) std r7, 24(r12) - HCALL_INST_POSTCALL + HCALL_INST_POSTCALL(r12) lwz r0,8(r1) mtcrf 0xff,r0 @@ -183,7 +194,7 @@ _GLOBAL(plpar_hcall9) mfcr r0 stw r0,8(r1) - HCALL_INST_PRECALL + HCALL_INST_PRECALL(r5) std r4,STK_PARM(r4)(r1) /* Save ret buffer */ @@ -211,7 +222,7 @@ _GLOBAL(plpar_hcall9) std r11,56(r12) std r0, 64(r12) - HCALL_INST_POSTCALL + HCALL_INST_POSTCALL(r12) lwz r0,8(r1) mtcrf 0xff,r0 diff --git a/arch/powerpc/platforms/pseries/hvCall_inst.c b/arch/powerpc/platforms/pseries/hvCall_inst.c index e44e1035f133..2f58c71b7259 100644 --- a/arch/powerpc/platforms/pseries/hvCall_inst.c +++ b/arch/powerpc/platforms/pseries/hvCall_inst.c @@ -102,7 +102,7 @@ static const struct file_operations hcall_inst_seq_fops = { #define CPU_NAME_BUF_SIZE 32 -static void probe_hcall_entry(unsigned long opcode) +static void probe_hcall_entry(unsigned long opcode, unsigned long *args) { struct hcall_stats *h; @@ -114,7 +114,8 @@ static void probe_hcall_entry(unsigned long opcode) h->purr_start = mfspr(SPRN_PURR); } -static void probe_hcall_exit(unsigned long opcode, unsigned long retval) +static void probe_hcall_exit(unsigned long opcode, unsigned long retval, + unsigned long *retbuf) { struct hcall_stats *h; diff --git a/arch/powerpc/platforms/pseries/lpar.c b/arch/powerpc/platforms/pseries/lpar.c index 4b7b6e8e32de..0707653612ba 100644 --- a/arch/powerpc/platforms/pseries/lpar.c +++ b/arch/powerpc/platforms/pseries/lpar.c @@ -683,13 +683,14 @@ void hcall_tracepoint_unregfunc(void) hcall_tracepoint_refcount--; } -void __trace_hcall_entry(unsigned long opcode) +void __trace_hcall_entry(unsigned long opcode, unsigned long *args) { - trace_hcall_entry(opcode); + trace_hcall_entry(opcode, args); } -void __trace_hcall_exit(long opcode, unsigned long retval) +void __trace_hcall_exit(long opcode, unsigned long retval, + unsigned long *retbuf) { - trace_hcall_exit(opcode, retval); + trace_hcall_exit(opcode, retval, retbuf); } #endif From b3c86ee6d128dea7c671380090488887e73fa774 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Mon, 26 Oct 2009 18:51:57 +0000 Subject: [PATCH 10/32] powerpc: Disable HCALL_STATS by default The overhead of HCALL_STATS is quite high and the functionality is very rarely used. Key statistics are also missing (eg min/max). With the new hcall tracepoints much more powerful tracing can be done in a kernel module. Lets disable this by default. Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- arch/powerpc/configs/pseries_defconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/powerpc/configs/pseries_defconfig b/arch/powerpc/configs/pseries_defconfig index f1889abb89b1..c568329723b8 100644 --- a/arch/powerpc/configs/pseries_defconfig +++ b/arch/powerpc/configs/pseries_defconfig @@ -1683,7 +1683,7 @@ CONFIG_HAVE_ARCH_KGDB=y CONFIG_DEBUG_STACKOVERFLOW=y # CONFIG_DEBUG_STACK_USAGE is not set # CONFIG_DEBUG_PAGEALLOC is not set -CONFIG_HCALL_STATS=y +# CONFIG_HCALL_STATS is not set # CONFIG_CODE_PATCHING_SELFTEST is not set # CONFIG_FTR_FIXUP_SELFTEST is not set # CONFIG_MSI_BITMAP_SELFTEST is not set From 907b1f45d901c956e4bcd3f27c4f1f25d6fb36b2 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Mon, 26 Oct 2009 18:52:24 +0000 Subject: [PATCH 11/32] powerpc: Export powerpc_debugfs_root Kernel modules should be able to place their debug output inside our powerpc debugfs directory. Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- arch/powerpc/kernel/setup-common.c | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/powerpc/kernel/setup-common.c b/arch/powerpc/kernel/setup-common.c index 4271f7a655a3..845c72ab7357 100644 --- a/arch/powerpc/kernel/setup-common.c +++ b/arch/powerpc/kernel/setup-common.c @@ -660,6 +660,7 @@ late_initcall(check_cache_coherency); #ifdef CONFIG_DEBUG_FS struct dentry *powerpc_debugfs_root; +EXPORT_SYMBOL(powerpc_debugfs_root); static int powerpc_debugfs_init(void) { From 3cd980dbc1050889acca7306cbcedf79a4ba2f81 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Sun, 18 Oct 2009 01:23:28 +0000 Subject: [PATCH 12/32] powerpc: perf_event: Cleanup copy_page output by hiding setup symbol A lot of hits in "setup" doesn't make much sense, so hide this symbol and allow all the hits to end up in copy_4k_page. Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- arch/powerpc/lib/copypage_64.S | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/powerpc/lib/copypage_64.S b/arch/powerpc/lib/copypage_64.S index 75f3267fdc30..e68beac0a171 100644 --- a/arch/powerpc/lib/copypage_64.S +++ b/arch/powerpc/lib/copypage_64.S @@ -26,11 +26,11 @@ BEGIN_FTR_SECTION srd r8,r5,r11 mtctr r8 -setup: +.Lsetup: dcbt r9,r4 dcbz r9,r3 add r9,r9,r12 - bdnz setup + bdnz .Lsetup END_FTR_SECTION_IFSET(CPU_FTR_CP_USE_DCBTZ) addi r3,r3,-8 srdi r8,r5,7 /* page is copied in 128 byte strides */ From 917e407c762ba6d91d1a4bc1c804d518585082a3 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Sun, 18 Oct 2009 01:24:29 +0000 Subject: [PATCH 13/32] powerpc: perf_event: Hide iseries_check_pending_irqs If CONFIG_PPC_ISERIES isn't defined we end up with iseries_check_pending_irqs and do_work at the same address. perf ends up picking iseries_check_pending_irqs which creates confusing backtraces. Hide it. Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- arch/powerpc/kernel/entry_64.S | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/powerpc/kernel/entry_64.S b/arch/powerpc/kernel/entry_64.S index 900e0eea0099..e722226cabc4 100644 --- a/arch/powerpc/kernel/entry_64.S +++ b/arch/powerpc/kernel/entry_64.S @@ -551,7 +551,7 @@ restore: BEGIN_FW_FTR_SECTION ld r5,SOFTE(r1) FW_FTR_SECTION_ELSE - b iseries_check_pending_irqs + b .Liseries_check_pending_irqs ALT_FW_FTR_SECTION_END_IFCLR(FW_FEATURE_ISERIES) 2: TRACE_AND_RESTORE_IRQ(r5); @@ -623,7 +623,7 @@ ALT_FW_FTR_SECTION_END_IFCLR(FW_FEATURE_ISERIES) #endif /* CONFIG_PPC_BOOK3E */ -iseries_check_pending_irqs: +.Liseries_check_pending_irqs: #ifdef CONFIG_PPC_ISERIES ld r5,SOFTE(r1) cmpdi 0,r5,0 From c86e2eaded39843e1bf4f07d1adfab4494f20894 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Sun, 18 Oct 2009 01:24:06 +0000 Subject: [PATCH 14/32] powerpc: perf_event: Cleanup output by adding symbols Add some dummy symbols for the branches at 0xf00, 0xf20 and 0xf40, otherwise hits end up in trap_0e which is confusing to the user. Signed-off-by: Anton Blanchard Signed-off-by: Paul Mackerras --- arch/powerpc/kernel/exceptions-64s.S | 3 +++ 1 file changed, 3 insertions(+) diff --git a/arch/powerpc/kernel/exceptions-64s.S b/arch/powerpc/kernel/exceptions-64s.S index 1808876edcc9..c7eb4e0eb86c 100644 --- a/arch/powerpc/kernel/exceptions-64s.S +++ b/arch/powerpc/kernel/exceptions-64s.S @@ -185,12 +185,15 @@ END_FTR_SECTION_IFSET(CPU_FTR_REAL_LE) * prolog code of the PerformanceMonitor one. A little * trickery is thus necessary */ +performance_monitor_pSeries_1: . = 0xf00 b performance_monitor_pSeries +altivec_unavailable_pSeries_1: . = 0xf20 b altivec_unavailable_pSeries +vsx_unavailable_pSeries_1: . = 0xf40 b vsx_unavailable_pSeries From c426bba069e65ea438880a04aa4e7c5b880e1728 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Thu, 5 Nov 2009 09:31:31 +0900 Subject: [PATCH 15/32] perf bench: Add new directory and header for new subcommand 'bench' This patch adds bench/ directory and bench/bench.h. bench/ directory will contain modules for bench subcommand. bench/bench.h is for listing prototypes of module functions. Signed-off-by: Hitoshi Mitake Cc: Rusty Russell Cc: Peter Zijlstra Cc: Mike Galbraith Cc: Arnaldo Carvalho de Melo Cc: fweisbec@gmail.com Cc: Jiri Kosina LKML-Reference: <1257381097-4743-2-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/bench/bench.h | 9 +++++++++ 1 file changed, 9 insertions(+) create mode 100644 tools/perf/bench/bench.h diff --git a/tools/perf/bench/bench.h b/tools/perf/bench/bench.h new file mode 100644 index 000000000000..59adb279cd7a --- /dev/null +++ b/tools/perf/bench/bench.h @@ -0,0 +1,9 @@ +#ifndef BENCH_H +#define BENCH_H + +extern int bench_sched_messaging(int argc, const char **argv, + const char *prefix); +extern int bench_sched_pipe(int argc, const char **argv, + const char *prefix); + +#endif From e27454cc6352c4226ddc76f5e3a5dedd7dff456a Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Thu, 5 Nov 2009 09:31:32 +0900 Subject: [PATCH 16/32] perf bench: Add sched-messaging.c: Benchmark for scheduler and IPC mechanisms based on hackbench This patch adds bench/sched-messaging.c. This benchmark measures performance of scheduler and IPC mechanisms, and is based on hackbench by Rusty Russell. Example of usage: % perf bench sched messaging -g 20 -l 1000 -s 5.432 # in sec % perf bench sched messaging # run with default options (20 sender and receiver processes per group) (10 groups == 400 processes run) Total time:0.308 sec % perf bench sched messaging -t -g 20 # # be multi-thread, with 20 groups (20 sender and receiver threads per group) (20 groups == 800 threads run) Total time:0.582 sec ( Rusty is the original author of hackbench.c and he said the code is and was under the GPLv2 so fine to be merged. ) Signed-off-by: Hitoshi Mitake Acked-by: Rusty Russell Cc: Peter Zijlstra Cc: Mike Galbraith Cc: Arnaldo Carvalho de Melo Cc: fweisbec@gmail.com Cc: Jiri Kosina LKML-Reference: <1257381097-4743-3-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/bench/sched-messaging.c | 332 +++++++++++++++++++++++++++++ 1 file changed, 332 insertions(+) create mode 100644 tools/perf/bench/sched-messaging.c diff --git a/tools/perf/bench/sched-messaging.c b/tools/perf/bench/sched-messaging.c new file mode 100644 index 000000000000..36b62c549e28 --- /dev/null +++ b/tools/perf/bench/sched-messaging.c @@ -0,0 +1,332 @@ +/* + * + * builtin-bench-messaging.c + * + * messaging: Benchmark for scheduler and IPC mechanisms + * + * Based on hackbench by Rusty Russell + * Ported to perf by Hitoshi Mitake + * + */ + +#include "../perf.h" +#include "../util/util.h" +#include "../util/parse-options.h" +#include "../builtin.h" +#include "bench.h" + +/* Test groups of 20 processes spraying to 20 receivers */ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#define DATASIZE 100 + +static int use_pipes = 0; +static unsigned int loops = 100; +static unsigned int thread_mode = 0; +static unsigned int num_groups = 10; +static int simple = 0; + +struct sender_context { + unsigned int num_fds; + int ready_out; + int wakefd; + int out_fds[0]; +}; + +struct receiver_context { + unsigned int num_packets; + int in_fds[2]; + int ready_out; + int wakefd; +}; + +static void barf(const char *msg) +{ + fprintf(stderr, "%s (error: %s)\n", msg, strerror(errno)); + exit(1); +} + +static void fdpair(int fds[2]) +{ + if (use_pipes) { + if (pipe(fds) == 0) + return; + } else { + if (socketpair(AF_UNIX, SOCK_STREAM, 0, fds) == 0) + return; + } + + barf(use_pipes ? "pipe()" : "socketpair()"); +} + +/* Block until we're ready to go */ +static void ready(int ready_out, int wakefd) +{ + char dummy; + struct pollfd pollfd = { .fd = wakefd, .events = POLLIN }; + + /* Tell them we're ready. */ + if (write(ready_out, &dummy, 1) != 1) + barf("CLIENT: ready write"); + + /* Wait for "GO" signal */ + if (poll(&pollfd, 1, -1) != 1) + barf("poll"); +} + +/* Sender sprays loops messages down each file descriptor */ +static void *sender(struct sender_context *ctx) +{ + char data[DATASIZE]; + unsigned int i, j; + + ready(ctx->ready_out, ctx->wakefd); + + /* Now pump to every receiver. */ + for (i = 0; i < loops; i++) { + for (j = 0; j < ctx->num_fds; j++) { + int ret, done = 0; + +again: + ret = write(ctx->out_fds[j], data + done, + sizeof(data)-done); + if (ret < 0) + barf("SENDER: write"); + done += ret; + if (done < DATASIZE) + goto again; + } + } + + return NULL; +} + + +/* One receiver per fd */ +static void *receiver(struct receiver_context* ctx) +{ + unsigned int i; + + if (!thread_mode) + close(ctx->in_fds[1]); + + /* Wait for start... */ + ready(ctx->ready_out, ctx->wakefd); + + /* Receive them all */ + for (i = 0; i < ctx->num_packets; i++) { + char data[DATASIZE]; + int ret, done = 0; + +again: + ret = read(ctx->in_fds[0], data + done, DATASIZE - done); + if (ret < 0) + barf("SERVER: read"); + done += ret; + if (done < DATASIZE) + goto again; + } + + return NULL; +} + +static pthread_t create_worker(void *ctx, void *(*func)(void *)) +{ + pthread_attr_t attr; + pthread_t childid; + int err; + + if (!thread_mode) { + /* process mode */ + /* Fork the receiver. */ + switch (fork()) { + case -1: + barf("fork()"); + break; + case 0: + (*func) (ctx); + exit(0); + break; + default: + break; + } + + return (pthread_t)0; + } + + if (pthread_attr_init(&attr) != 0) + barf("pthread_attr_init:"); + +#ifndef __ia64__ + if (pthread_attr_setstacksize(&attr, PTHREAD_STACK_MIN) != 0) + barf("pthread_attr_setstacksize"); +#endif + + err = pthread_create(&childid, &attr, func, ctx); + if (err != 0) { + fprintf(stderr, "pthread_create failed: %s (%d)\n", + strerror(err), err); + exit(-1); + } + return childid; +} + +static void reap_worker(pthread_t id) +{ + int proc_status; + void *thread_status; + + if (!thread_mode) { + /* process mode */ + wait(&proc_status); + if (!WIFEXITED(proc_status)) + exit(1); + } else { + pthread_join(id, &thread_status); + } +} + +/* One group of senders and receivers */ +static unsigned int group(pthread_t *pth, + unsigned int num_fds, + int ready_out, + int wakefd) +{ + unsigned int i; + struct sender_context *snd_ctx = malloc(sizeof(struct sender_context) + + num_fds * sizeof(int)); + + if (!snd_ctx) + barf("malloc()"); + + for (i = 0; i < num_fds; i++) { + int fds[2]; + struct receiver_context *ctx = malloc(sizeof(*ctx)); + + if (!ctx) + barf("malloc()"); + + + /* Create the pipe between client and server */ + fdpair(fds); + + ctx->num_packets = num_fds * loops; + ctx->in_fds[0] = fds[0]; + ctx->in_fds[1] = fds[1]; + ctx->ready_out = ready_out; + ctx->wakefd = wakefd; + + pth[i] = create_worker(ctx, (void *)receiver); + + snd_ctx->out_fds[i] = fds[1]; + if (!thread_mode) + close(fds[0]); + } + + /* Now we have all the fds, fork the senders */ + for (i = 0; i < num_fds; i++) { + snd_ctx->ready_out = ready_out; + snd_ctx->wakefd = wakefd; + snd_ctx->num_fds = num_fds; + + pth[num_fds+i] = create_worker(snd_ctx, (void *)sender); + } + + /* Close the fds we have left */ + if (!thread_mode) + for (i = 0; i < num_fds; i++) + close(snd_ctx->out_fds[i]); + + /* Return number of children to reap */ + return num_fds * 2; +} + +static const struct option options[] = { + OPT_BOOLEAN('p', "pipe", &use_pipes, + "Use pipe() instead of socketpair()"), + OPT_BOOLEAN('t', "thread", &thread_mode, + "Be multi thread instead of multi process"), + OPT_INTEGER('g', "group", &num_groups, + "Specify number of groups"), + OPT_INTEGER('l', "loop", &loops, + "Specify number of loops"), + OPT_BOOLEAN('s', "simple-output", &simple, + "Do simple output (this maybe useful for" + "processing by scripts or graph tools like gnuplot)"), + OPT_END() +}; + +static const char * const bench_sched_message_usage[] = { + "perf bench sched messaging ", + NULL +}; + +int bench_sched_messaging(int argc, const char **argv, + const char *prefix __used) +{ + unsigned int i, total_children; + struct timeval start, stop, diff; + unsigned int num_fds = 20; + int readyfds[2], wakefds[2]; + char dummy; + pthread_t *pth_tab; + + argc = parse_options(argc, argv, options, + bench_sched_message_usage, 0); + + pth_tab = malloc(num_fds * 2 * num_groups * sizeof(pthread_t)); + if (!pth_tab) + barf("main:malloc()"); + + fdpair(readyfds); + fdpair(wakefds); + + total_children = 0; + for (i = 0; i < num_groups; i++) + total_children += group(pth_tab+total_children, num_fds, + readyfds[1], wakefds[0]); + + /* Wait for everyone to be ready */ + for (i = 0; i < total_children; i++) + if (read(readyfds[0], &dummy, 1) != 1) + barf("Reading for readyfds"); + + gettimeofday(&start, NULL); + + /* Kick them off */ + if (write(wakefds[1], &dummy, 1) != 1) + barf("Writing to start them"); + + /* Reap them all */ + for (i = 0; i < total_children; i++) + reap_worker(pth_tab[i]); + + gettimeofday(&stop, NULL); + + timersub(&stop, &start, &diff); + + if (simple) + printf("%lu.%03lu\n", diff.tv_sec, diff.tv_usec/1000); + else { + printf("(%d sender and receiver %s per group)\n", + num_fds, thread_mode ? "threads" : "processes"); + printf("(%d groups == %d %s run)\n\n", + num_groups, num_groups * 2 * num_fds, + thread_mode ? "threads" : "processes"); + printf("\tTotal time:%lu.%03lu sec\n", + diff.tv_sec, diff.tv_usec/1000); + } + + return 0; +} From c7d9300f367f480aee4663a0e3695c5b48859a1a Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Thu, 5 Nov 2009 09:31:33 +0900 Subject: [PATCH 17/32] perf bench: Add sched-pipe.c: Benchmark for pipe() system call This patch adds bench/sched-pipe.c. bench/sched-pipe.c is a benchmark program to measure performance of pipe() system call. This benchmark is based on pipe-test-1m.c by Ingo Molnar: http://people.redhat.com/mingo/cfs-scheduler/tools/pipe-test-1m.c Example of use: % perf bench sched pipe (executing 1000000 pipe operations between two tasks) Total time:4.499 sec 4.499179 usecs/op 222262 ops/sec % perf bench sched pipe -s -l 1000 0.015 Signed-off-by: Hitoshi Mitake Cc: Rusty Russell Cc: Peter Zijlstra Cc: Mike Galbraith Cc: Arnaldo Carvalho de Melo Cc: fweisbec@gmail.com Cc: Jiri Kosina LKML-Reference: <1257381097-4743-4-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/bench/sched-pipe.c | 113 ++++++++++++++++++++++++++++++++++ 1 file changed, 113 insertions(+) create mode 100644 tools/perf/bench/sched-pipe.c diff --git a/tools/perf/bench/sched-pipe.c b/tools/perf/bench/sched-pipe.c new file mode 100644 index 000000000000..3214ed20b1aa --- /dev/null +++ b/tools/perf/bench/sched-pipe.c @@ -0,0 +1,113 @@ +/* + * + * builtin-bench-pipe.c + * + * pipe: Benchmark for pipe() + * + * Based on pipe-test-1m.c by Ingo Molnar + * http://people.redhat.com/mingo/cfs-scheduler/tools/pipe-test-1m.c + * Ported to perf by Hitoshi Mitake + * + */ + +#include "../perf.h" +#include "../util/util.h" +#include "../util/parse-options.h" +#include "../builtin.h" +#include "bench.h" + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#define LOOPS_DEFAULT 1000000 +static int loops = LOOPS_DEFAULT; +static int simple = 0; + +static const struct option options[] = { + OPT_INTEGER('l', "loop", &loops, + "Specify number of loops"), + OPT_BOOLEAN('s', "simple-output", &simple, + "Do simple output (this maybe useful for" + "processing by scripts or graph tools like gnuplot)"), + OPT_END() +}; + +static const char * const bench_sched_pipe_usage[] = { + "perf bench sched pipe ", + NULL +}; + +int bench_sched_pipe(int argc, const char **argv, + const char *prefix __used) +{ + int pipe_1[2], pipe_2[2]; + int m = 0, i; + struct timeval start, stop, diff; + unsigned long long result_usec = 0; + + /* + * why does "ret" exist? + * discarding returned value of read(), write() + * causes error in building environment for perf + */ + int ret; + pid_t pid; + + argc = parse_options(argc, argv, options, + bench_sched_pipe_usage, 0); + + assert(!pipe(pipe_1)); + assert(!pipe(pipe_2)); + + pid = fork(); + assert(pid >= 0); + + gettimeofday(&start, NULL); + + if (!pid) { + for (i = 0; i < loops; i++) { + ret = read(pipe_1[0], &m, sizeof(int)); + ret = write(pipe_2[1], &m, sizeof(int)); + } + } else { + for (i = 0; i < loops; i++) { + ret = write(pipe_1[1], &m, sizeof(int)); + ret = read(pipe_2[0], &m, sizeof(int)); + } + } + + gettimeofday(&stop, NULL); + timersub(&stop, &start, &diff); + + if (pid) + return 0; + + if (simple) + printf("%lu.%03lu\n", + diff.tv_sec, diff.tv_usec / 1000); + else { + printf("(executing %d pipe operations between two tasks)\n\n", + loops); + + result_usec = diff.tv_sec * 1000000; + result_usec += diff.tv_usec; + + printf("\tTotal time:%lu.%03lu sec\n", + diff.tv_sec, diff.tv_usec / 1000); + printf("\t\t%lf usecs/op\n", + (double)result_usec / (double)loops); + printf("\t\t%d ops/sec\n", + (int)((double)loops / + ((double)result_usec / (double)1000000))); + } + + return 0; +} From 629cc356653719c206a05f4dee5c5e242edb6546 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Thu, 5 Nov 2009 09:31:34 +0900 Subject: [PATCH 18/32] perf bench: Add builtin-bench.c: General framework for benchmark suites This patch adds builtin-bench.c builtin-bench.c is a general framework for benchmark suites. Signed-off-by: Hitoshi Mitake Cc: Rusty Russell Cc: Peter Zijlstra Cc: Mike Galbraith Cc: Arnaldo Carvalho de Melo Cc: fweisbec@gmail.com Cc: Jiri Kosina LKML-Reference: <1257381097-4743-5-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/builtin-bench.c | 128 +++++++++++++++++++++++++++++++++++++ 1 file changed, 128 insertions(+) create mode 100644 tools/perf/builtin-bench.c diff --git a/tools/perf/builtin-bench.c b/tools/perf/builtin-bench.c new file mode 100644 index 000000000000..31f41643b0cd --- /dev/null +++ b/tools/perf/builtin-bench.c @@ -0,0 +1,128 @@ +/* + * + * builtin-bench.c + * + * General benchmarking subsystem provided by perf + * + * Copyright (C) 2009, Hitoshi Mitake + * + */ + +/* + * + * Available subsystem list: + * sched ... scheduler and IPC mechanism + * + */ + +#include "perf.h" +#include "util/util.h" +#include "util/parse-options.h" +#include "builtin.h" +#include "bench/bench.h" + +#include +#include +#include + +struct bench_suite { + const char *name; + const char *summary; + int (*fn)(int, const char **, const char *); +}; + +static struct bench_suite sched_suites[] = { + { "messaging", + "Benchmark for scheduler and IPC mechanisms", + bench_sched_messaging }, + { "pipe", + "Flood of communication over pipe() between two processes", + bench_sched_pipe }, + { NULL, + NULL, + NULL } +}; + +struct bench_subsys { + const char *name; + const char *summary; + struct bench_suite *suites; +}; + +static struct bench_subsys subsystems[] = { + { "sched", + "scheduler and IPC mechanism", + sched_suites }, + { NULL, + NULL, + NULL } +}; + +static void dump_suites(int subsys_index) +{ + int i; + + printf("List of available suites for %s...\n\n", + subsystems[subsys_index].name); + + for (i = 0; subsystems[subsys_index].suites[i].name; i++) + printf("\t%s: %s\n", + subsystems[subsys_index].suites[i].name, + subsystems[subsys_index].suites[i].summary); + + printf("\n"); + return; +} + +int cmd_bench(int argc, const char **argv, const char *prefix __used) +{ + int i, j, status = 0; + + if (argc < 2) { + /* No subsystem specified. */ + printf("Usage: perf bench []\n\n"); + printf("List of available subsystems...\n\n"); + + for (i = 0; subsystems[i].name; i++) + printf("\t%s: %s\n", + subsystems[i].name, subsystems[i].summary); + printf("\n"); + + goto end; + } + + for (i = 0; subsystems[i].name; i++) { + if (strcmp(subsystems[i].name, argv[1])) + continue; + + if (argc < 3) { + /* No suite specified. */ + dump_suites(i); + goto end; + } + + for (j = 0; subsystems[i].suites[j].name; j++) { + if (strcmp(subsystems[i].suites[j].name, argv[2])) + continue; + + status = subsystems[i].suites[j].fn(argc - 2, + argv + 2, prefix); + goto end; + } + + if (!strcmp(argv[2], "-h") || !strcmp(argv[2], "--help")) { + dump_suites(i); + goto end; + } + + printf("Unknown suite:%s for %s\n", argv[2], argv[1]); + status = 1; + goto end; + } + + printf("Unknown subsystem:%s\n", argv[1]); + status = 1; + +end: + return status; +} From 11bd341c043348ecb7462d3bd8e1ad6d00f6892a Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Thu, 5 Nov 2009 09:31:35 +0900 Subject: [PATCH 19/32] perf bench: Modify builtin.h for new prototype This patch modifies builtin.h to add prototype of cmd_bench(). Signed-off-by: Hitoshi Mitake Cc: Rusty Russell Cc: Peter Zijlstra Cc: Mike Galbraith Cc: Arnaldo Carvalho de Melo Cc: fweisbec@gmail.com Cc: Jiri Kosina LKML-Reference: <1257381097-4743-6-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/builtin.h | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/perf/builtin.h b/tools/perf/builtin.h index e11d8d231c3b..f0cd5b139b7c 100644 --- a/tools/perf/builtin.h +++ b/tools/perf/builtin.h @@ -15,6 +15,7 @@ extern int read_line_with_nul(char *buf, int size, FILE *file); extern int check_pager_config(const char *cmd); extern int cmd_annotate(int argc, const char **argv, const char *prefix); +extern int cmd_bench(int argc, const char **argv, const char *prefix); extern int cmd_help(int argc, const char **argv, const char *prefix); extern int cmd_sched(int argc, const char **argv, const char *prefix); extern int cmd_list(int argc, const char **argv, const char *prefix); From dcba8848d3bc83ec9ee0858b9ae6e4f1c1fa7fa3 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Thu, 5 Nov 2009 09:31:36 +0900 Subject: [PATCH 20/32] perf bench: Add new subcommand 'bench' to perf.c This patch modifies perf.c for invoking 'bench' subcommand. Signed-off-by: Hitoshi Mitake Cc: Rusty Russell Cc: Peter Zijlstra Cc: Mike Galbraith Cc: Arnaldo Carvalho de Melo Cc: fweisbec@gmail.com Cc: Jiri Kosina LKML-Reference: <1257381097-4743-7-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/perf.c | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/perf/perf.c b/tools/perf/perf.c index 624e62d9d1e0..f90ca5ec83f3 100644 --- a/tools/perf/perf.c +++ b/tools/perf/perf.c @@ -288,6 +288,7 @@ static void handle_internal_command(int argc, const char **argv) { "list", cmd_list, 0 }, { "record", cmd_record, 0 }, { "report", cmd_report, 0 }, + { "bench", cmd_bench, 0 }, { "stat", cmd_stat, 0 }, { "timechart", cmd_timechart, 0 }, { "top", cmd_top, 0 }, From bfde82ef51e3ea6ab8634d0fdbf5adcdd1b429cb Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Thu, 5 Nov 2009 09:31:37 +0900 Subject: [PATCH 21/32] perf bench: Add subcommand 'bench' to the Makefile This patch modifies Makefile for new files related to 'bench' subcommand. The new code is active from this point on. Signed-off-by: Hitoshi Mitake Cc: Rusty Russell Cc: Peter Zijlstra Cc: Mike Galbraith Cc: Arnaldo Carvalho de Melo Cc: fweisbec@gmail.com Cc: Jiri Kosina LKML-Reference: <1257381097-4743-8-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/Makefile | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/tools/perf/Makefile b/tools/perf/Makefile index 542b29e2e382..0a2542844a62 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -416,6 +416,13 @@ LIB_OBJS += util/hist.o LIB_OBJS += util/data_map.o BUILTIN_OBJS += builtin-annotate.o + +BUILTIN_OBJS += builtin-bench.o + +# Benchmark modules +BUILTIN_OBJS += bench/sched-messaging.o +BUILTIN_OBJS += bench/sched-pipe.o + BUILTIN_OBJS += builtin-help.o BUILTIN_OBJS += builtin-sched.o BUILTIN_OBJS += builtin-list.o From 5ff0cfc67f00fe0feaa1da0b2359232ea4aa0ee7 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Mon, 9 Nov 2009 12:31:05 +0900 Subject: [PATCH 22/32] perf bench: Fix bench/sched-pipe.c to wait for child process Ingo reported this small 'perf bench sched pipe' output problem: | $ ./perf bench sched pipe | (executing 1000000 pipe operations between two tasks) | | Total time:4.898 sec | $ 4.898586 usecs/op | 204140 ops/sec | | the shell prompt came back before the usecs/op and ops/sec line | was printed. Process teardown race, lack of wait() or so? This caused by lack of calling waitpid() by parent process, so I added it. Signed-off-by: Hitoshi Mitake Cc: Rusty Russell Cc: Peter Zijlstra Cc: Paul Mackerras Cc: Mike Galbraith Cc: Arnaldo Carvalho de Melo Cc: Frederic Weisbecker Cc: Jiri Kosina LKML-Reference: <1257737465-7546-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/bench/sched-pipe.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/tools/perf/bench/sched-pipe.c b/tools/perf/bench/sched-pipe.c index 3214ed20b1aa..6a29100e9282 100644 --- a/tools/perf/bench/sched-pipe.c +++ b/tools/perf/bench/sched-pipe.c @@ -26,6 +26,7 @@ #include #include #include +#include #define LOOPS_DEFAULT 1000000 static int loops = LOOPS_DEFAULT; @@ -58,8 +59,8 @@ int bench_sched_pipe(int argc, const char **argv, * discarding returned value of read(), write() * causes error in building environment for perf */ - int ret; - pid_t pid; + int ret, wait_stat; + pid_t pid, retpid; argc = parse_options(argc, argv, options, bench_sched_pipe_usage, 0); @@ -87,8 +88,11 @@ int bench_sched_pipe(int argc, const char **argv, gettimeofday(&stop, NULL); timersub(&stop, &start, &diff); - if (pid) + if (pid) { + retpid = waitpid(pid, &wait_stat, 0); + assert((retpid == pid) && WIFEXITED(wait_stat)); return 0; + } if (simple) printf("%lu.%03lu\n", From 242aa14a67f4e19453fc8a51cffc5ac5ee5bcbd1 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Tue, 10 Nov 2009 08:19:59 +0900 Subject: [PATCH 23/32] perf bench: Add format constants to bench.h for unified output formatting This patch adds some constants and extern declaration to bench.h. These are used for unified output formatting of 'perf bench'. Signed-off-by: Hitoshi Mitake Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <1257808802-9420-2-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/bench/bench.h | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/tools/perf/bench/bench.h b/tools/perf/bench/bench.h index 59adb279cd7a..42167ea41944 100644 --- a/tools/perf/bench/bench.h +++ b/tools/perf/bench/bench.h @@ -6,4 +6,13 @@ extern int bench_sched_messaging(int argc, const char **argv, extern int bench_sched_pipe(int argc, const char **argv, const char *prefix); +#define BENCH_FORMAT_DEFAULT_STR "default" +#define BENCH_FORMAT_DEFAULT 0 +#define BENCH_FORMAT_SIMPLE_STR "simple" +#define BENCH_FORMAT_SIMPLE 1 + +#define BENCH_FORMAT_UNKNOWN -1 + +extern int bench_format; + #endif From 386d7e9e542c2115d5d300747e57f503458a1617 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Tue, 10 Nov 2009 08:20:00 +0900 Subject: [PATCH 24/32] perf bench: Modify builtin-bench.c for processing common options This patch modifies builtin-bench.c for processing common options. The first option added is "--format". Users of perf bench will be able to specify output style by --format. Usage example: % ./perf bench sched messaging # with no style specify (20 sender and receiver processes per group) (10 groups == 400 processes run) Total time:1.431 sec % ./perf bench --format=simple sched messaging # specified simple 1.431 Signed-off-by: Hitoshi Mitake Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <1257808802-9420-3-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/builtin-bench.c | 79 +++++++++++++++++++++++++++++++------- 1 file changed, 65 insertions(+), 14 deletions(-) diff --git a/tools/perf/builtin-bench.c b/tools/perf/builtin-bench.c index 31f41643b0cd..c7505eaff84b 100644 --- a/tools/perf/builtin-bench.c +++ b/tools/perf/builtin-bench.c @@ -74,53 +74,104 @@ static void dump_suites(int subsys_index) return; } +static char *bench_format_str; +int bench_format = BENCH_FORMAT_DEFAULT; + +static const struct option bench_options[] = { + OPT_STRING('f', "format", &bench_format_str, "default", + "Specify format style"), + OPT_END() +}; + +static const char * const bench_usage[] = { + "perf bench [] []", + NULL +}; + +static void print_usage(void) +{ + int i; + + printf("Usage: \n"); + for (i = 0; bench_usage[i]; i++) + printf("\t%s\n", bench_usage[i]); + printf("\n"); + + printf("List of available subsystems...\n\n"); + + for (i = 0; subsystems[i].name; i++) + printf("\t%s: %s\n", + subsystems[i].name, subsystems[i].summary); + printf("\n"); +} + +static int bench_str2int(char *str) +{ + if (!str) + return BENCH_FORMAT_DEFAULT; + + if (!strcmp(str, BENCH_FORMAT_DEFAULT_STR)) + return BENCH_FORMAT_DEFAULT; + else if (!strcmp(str, BENCH_FORMAT_SIMPLE_STR)) + return BENCH_FORMAT_SIMPLE; + + return BENCH_FORMAT_UNKNOWN; +} + int cmd_bench(int argc, const char **argv, const char *prefix __used) { int i, j, status = 0; if (argc < 2) { /* No subsystem specified. */ - printf("Usage: perf bench []\n\n"); - printf("List of available subsystems...\n\n"); + print_usage(); + goto end; + } - for (i = 0; subsystems[i].name; i++) - printf("\t%s: %s\n", - subsystems[i].name, subsystems[i].summary); - printf("\n"); + argc = parse_options(argc, argv, bench_options, bench_usage, + PARSE_OPT_STOP_AT_NON_OPTION); + bench_format = bench_str2int(bench_format_str); + if (bench_format == BENCH_FORMAT_UNKNOWN) { + printf("Unknown format descriptor:%s\n", bench_format_str); + goto end; + } + + if (argc < 1) { + print_usage(); goto end; } for (i = 0; subsystems[i].name; i++) { - if (strcmp(subsystems[i].name, argv[1])) + if (strcmp(subsystems[i].name, argv[0])) continue; - if (argc < 3) { + if (argc < 2) { /* No suite specified. */ dump_suites(i); goto end; } for (j = 0; subsystems[i].suites[j].name; j++) { - if (strcmp(subsystems[i].suites[j].name, argv[2])) + if (strcmp(subsystems[i].suites[j].name, argv[1])) continue; - status = subsystems[i].suites[j].fn(argc - 2, - argv + 2, prefix); + status = subsystems[i].suites[j].fn(argc - 1, + argv + 1, prefix); goto end; } - if (!strcmp(argv[2], "-h") || !strcmp(argv[2], "--help")) { + if (!strcmp(argv[1], "-h") || !strcmp(argv[1], "--help")) { dump_suites(i); goto end; } - printf("Unknown suite:%s for %s\n", argv[2], argv[1]); + printf("Unknown suite:%s for %s\n", argv[1], argv[0]); status = 1; goto end; } - printf("Unknown subsystem:%s\n", argv[1]); + printf("Unknown subsystem:%s\n", argv[0]); status = 1; end: From cced06c62a9db6bd6d77e3f0a57dbe47a26d881e Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Tue, 10 Nov 2009 08:20:01 +0900 Subject: [PATCH 25/32] perf bench: Modify bench/bench-messaging.c to adopt unified output formatting This patch modifies bench/bench-messaging.c to adopt unified output formatting: --format option. Usage example: % ./perf bench sched messaging # with no style specify (20 sender and receiver processes per group) (10 groups == 400 processes run) Total time:1.431 sec % ./perf bench --format=simple sched messaging # specified simple 1.431 Signed-off-by: Hitoshi Mitake Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <1257808802-9420-4-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/bench/sched-messaging.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-) diff --git a/tools/perf/bench/sched-messaging.c b/tools/perf/bench/sched-messaging.c index 36b62c549e28..2cc5edcca3d6 100644 --- a/tools/perf/bench/sched-messaging.c +++ b/tools/perf/bench/sched-messaging.c @@ -35,7 +35,6 @@ static int use_pipes = 0; static unsigned int loops = 100; static unsigned int thread_mode = 0; static unsigned int num_groups = 10; -static int simple = 0; struct sender_context { unsigned int num_fds; @@ -261,9 +260,6 @@ static const struct option options[] = { "Specify number of groups"), OPT_INTEGER('l', "loop", &loops, "Specify number of loops"), - OPT_BOOLEAN('s', "simple-output", &simple, - "Do simple output (this maybe useful for" - "processing by scripts or graph tools like gnuplot)"), OPT_END() }; @@ -316,9 +312,8 @@ int bench_sched_messaging(int argc, const char **argv, timersub(&stop, &start, &diff); - if (simple) - printf("%lu.%03lu\n", diff.tv_sec, diff.tv_usec/1000); - else { + switch (bench_format) { + case BENCH_FORMAT_DEFAULT: printf("(%d sender and receiver %s per group)\n", num_fds, thread_mode ? "threads" : "processes"); printf("(%d groups == %d %s run)\n\n", @@ -326,6 +321,15 @@ int bench_sched_messaging(int argc, const char **argv, thread_mode ? "threads" : "processes"); printf("\tTotal time:%lu.%03lu sec\n", diff.tv_sec, diff.tv_usec/1000); + break; + case BENCH_FORMAT_SIMPLE: + printf("%lu.%03lu\n", diff.tv_sec, diff.tv_usec/1000); + break; + default: + /* reaching here is something disaster */ + fprintf(stderr, "Unknown format:%d\n", bench_format); + exit(1); + break; } return 0; From 158ba827f6deef4102c5247ed4b6a587f0bd6a07 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Tue, 10 Nov 2009 08:20:02 +0900 Subject: [PATCH 26/32] perf bench: Modify builtin-pipe.c for processing common options This patch modifies builtin-pipe.c for processing common options. The first option added is "--format". Users of perf bench will be able to specify output style by --format. Usage example: % ./perf bench sched pipe # with no style specify (executing 1000000 pipe operations between two tasks) Total time:5.855 sec 5.855061 usecs/op 170792 ops/sec % ./perf bench --format=simple sched pipe # specified simple 5.988 Signed-off-by: Hitoshi Mitake Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <1257808802-9420-5-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar Cc: Peter Zijlstra Cc: Paul Mackerras --- tools/perf/bench/sched-pipe.c | 22 ++++++++++++++-------- 1 file changed, 14 insertions(+), 8 deletions(-) diff --git a/tools/perf/bench/sched-pipe.c b/tools/perf/bench/sched-pipe.c index 6a29100e9282..a9ac186714b4 100644 --- a/tools/perf/bench/sched-pipe.c +++ b/tools/perf/bench/sched-pipe.c @@ -30,14 +30,10 @@ #define LOOPS_DEFAULT 1000000 static int loops = LOOPS_DEFAULT; -static int simple = 0; static const struct option options[] = { OPT_INTEGER('l', "loop", &loops, "Specify number of loops"), - OPT_BOOLEAN('s', "simple-output", &simple, - "Do simple output (this maybe useful for" - "processing by scripts or graph tools like gnuplot)"), OPT_END() }; @@ -94,10 +90,8 @@ int bench_sched_pipe(int argc, const char **argv, return 0; } - if (simple) - printf("%lu.%03lu\n", - diff.tv_sec, diff.tv_usec / 1000); - else { + switch (bench_format) { + case BENCH_FORMAT_DEFAULT: printf("(executing %d pipe operations between two tasks)\n\n", loops); @@ -111,6 +105,18 @@ int bench_sched_pipe(int argc, const char **argv, printf("\t\t%d ops/sec\n", (int)((double)loops / ((double)result_usec / (double)1000000))); + break; + + case BENCH_FORMAT_SIMPLE: + printf("%lu.%03lu\n", + diff.tv_sec, diff.tv_usec / 1000); + break; + + default: + /* reaching here is something disaster */ + fprintf(stderr, "Unknown format:%d\n", bench_format); + exit(1); + break; } return 0; From 606bc1e18d346fc7d7fb333909cc95b06b1ca5b1 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 10 Nov 2009 20:50:53 +0900 Subject: [PATCH 27/32] perf bench: Clean up bench/bench.h Clean up initializers in bench.h: - No need to break the line for function prototypes, they are more readable in a single line. (even if checkpatch complains about it - We try to align definitions / structure fields vertically, to make it all a bit more readable. Signed-off-by: Ingo Molnar Signed-off-by: Hitoshi Mitake Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <1257853855-28934-2-git-send-email-mitake@dcl.info.waseda.ac.jp> --- tools/perf/bench/bench.h | 16 +++++++--------- 1 file changed, 7 insertions(+), 9 deletions(-) diff --git a/tools/perf/bench/bench.h b/tools/perf/bench/bench.h index 42167ea41944..9fbd8d745fa1 100644 --- a/tools/perf/bench/bench.h +++ b/tools/perf/bench/bench.h @@ -1,17 +1,15 @@ #ifndef BENCH_H #define BENCH_H -extern int bench_sched_messaging(int argc, const char **argv, - const char *prefix); -extern int bench_sched_pipe(int argc, const char **argv, - const char *prefix); +extern int bench_sched_messaging(int argc, const char **argv, const char *prefix); +extern int bench_sched_pipe(int argc, const char **argv, const char *prefix); -#define BENCH_FORMAT_DEFAULT_STR "default" -#define BENCH_FORMAT_DEFAULT 0 -#define BENCH_FORMAT_SIMPLE_STR "simple" -#define BENCH_FORMAT_SIMPLE 1 +#define BENCH_FORMAT_DEFAULT_STR "default" +#define BENCH_FORMAT_DEFAULT 0 +#define BENCH_FORMAT_SIMPLE_STR "simple" +#define BENCH_FORMAT_SIMPLE 1 -#define BENCH_FORMAT_UNKNOWN -1 +#define BENCH_FORMAT_UNKNOWN -1 extern int bench_format; From 9fbc04f2493929a69fd9e53b5fb53c127d7950d5 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Tue, 10 Nov 2009 20:50:54 +0900 Subject: [PATCH 28/32] perf bench: Add new document about perf-bench This patch adds new document about perf-bench. Man page and html will be provided for user. Signed-off-by: Hitoshi Mitake Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <1257853855-28934-3-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/Documentation/perf-bench.txt | 120 ++++++++++++++++++++++++ 1 file changed, 120 insertions(+) create mode 100644 tools/perf/Documentation/perf-bench.txt diff --git a/tools/perf/Documentation/perf-bench.txt b/tools/perf/Documentation/perf-bench.txt new file mode 100644 index 000000000000..ae525ac5a2ce --- /dev/null +++ b/tools/perf/Documentation/perf-bench.txt @@ -0,0 +1,120 @@ +perf-bench(1) +============ + +NAME +---- +perf-bench - General framework for benchmark suites + +SYNOPSIS +-------- +[verse] +'perf bench' [] [] + +DESCRIPTION +----------- +This 'perf bench' command is general framework for benchmark suites. + +COMMON OPTIONS +-------------- +-f:: +--format=:: +Specify format style. +Current available format styles are, + +'default':: +Default style. This is mainly for human reading. +--------------------- +% perf bench sched pipe # with no style specify +(executing 1000000 pipe operations between two tasks) + Total time:5.855 sec + 5.855061 usecs/op + 170792 ops/sec +--------------------- + +'simple':: +This simple style is friendly for automated +processing by scripts. +--------------------- +% perf bench --format=simple sched pipe # specified simple +5.988 +--------------------- + +SUBSYSTEM +--------- + +'sched':: + Scheduler and IPC mechanisms. + +SUITES FOR 'sched' +~~~~~~~~~~~~~~~~~~ +*messaging*:: +Suite for evaluating performance of scheduler and IPC mechanisms. +Based on hackbench by Rusty Russell. + +Options of *pipe* +^^^^^^^^^^^^^^^^^ +-p:: +--pipe:: +Use pipe() instead of socketpair() + +-t:: +--thread:: +Be multi thread instead of multi process + +-g:: +--group=:: +Specify number of groups + +-l:: +--loop=:: +Specify number of loops + +Example of *messaging* +^^^^^^^^^^^^^^^^^^^^^^ + +--------------------- +% perf bench sched messaging # run with default +options (20 sender and receiver processes per group) +(10 groups == 400 processes run) + + Total time:0.308 sec + +% perf bench sched messaging -t -g 20 # be multi-thread,with 20 groups +(20 sender and receiver threads per group) +(20 groups == 800 threads run) + + Total time:0.582 sec +--------------------- + +*pipe*:: +Suite for pipe() system call. +Based on pipe-test-1m.c by Ingo Molnar. + +Options of *pipe* +^^^^^^^^^^^^^^^^^ +-l:: +--loop=:: +Specify number of loops. + +Example of *pipe* +^^^^^^^^^^^^^^^^^ + +--------------------- +% perf bench sched pipe +(executing 1000000 pipe operations between two tasks) + + Total time:8.091 sec + 8.091833 usecs/op + 123581 ops/sec + +% perf bench sched pipe -l 1000 # loop 1000 +(executing 1000 pipe operations between two tasks) + + Total time:0.016 sec + 16.948000 usecs/op + 59004 ops/sec +--------------------- + +SEE ALSO +-------- +linkperf:perf[1] From 8d8d61aadb9d8cce07f7dcdb77a4c20a25d36d07 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Tue, 10 Nov 2009 20:50:55 +0900 Subject: [PATCH 29/32] perf bench: Modify command-list.txt for the entry of perf-bench This patch modifies command-list.txt for the entry of perf-bench. So perf will show 'bench' in command list. Example: % perf usage: perf [--version] [--help] COMMAND [ARGS] The most commonly used perf commands are: annotate Read perf.data (created by perf record) and display annotated code bench General framework for benchmark suites ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ list List all symbolic event types probe Define new dynamic tracepoints record Run a command and record its profile into perf.data report Read perf.data (created by perf record) and display the profile sched Tool to trace/measure scheduler properties (latencies) stat Run a command and gather performance counter statistics timechart Tool to visualize total system behavior during a workload top System profiling tool. trace Read perf.data (created by perf record) and display trace output See 'perf help COMMAND' for more information on a specific command. Signed-off-by: Hitoshi Mitake Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <1257853855-28934-4-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/command-list.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/perf/command-list.txt b/tools/perf/command-list.txt index 00326e230d87..981c40b9a5e2 100644 --- a/tools/perf/command-list.txt +++ b/tools/perf/command-list.txt @@ -3,6 +3,7 @@ # command name category [deprecated] [common] # perf-annotate mainporcelain common +perf-bench mainporcelain common perf-list mainporcelain common perf-sched mainporcelain common perf-record mainporcelain common From 79e295d4bd0f524257299e7c4e42f643f21abcc2 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Wed, 11 Nov 2009 00:04:00 +0900 Subject: [PATCH 30/32] perf bench: Improve builtin-bench.c for more friendly output This patch makes output of perf bench more friendly. Current style of putput, keeping user wait and printing everything suddenly when we finish, may confuse users. So I improved it: | % perf bench sched messaging | # Running sched/messaging benchmark... <- printed right after invocation | # 20 sender and receiver processes per group | # 10 groups == 400 processes run | | Total time: 1.476 [sec] Signed-off-by: Hitoshi Mitake Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <1257865442-20252-2-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/builtin-bench.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/tools/perf/builtin-bench.c b/tools/perf/builtin-bench.c index c7505eaff84b..90c39baae0de 100644 --- a/tools/perf/builtin-bench.c +++ b/tools/perf/builtin-bench.c @@ -156,6 +156,10 @@ int cmd_bench(int argc, const char **argv, const char *prefix __used) if (strcmp(subsystems[i].suites[j].name, argv[1])) continue; + if (bench_format == BENCH_FORMAT_DEFAULT) + printf("# Running %s/%s benchmark...\n", + subsystems[i].name, + subsystems[i].suites[j].name); status = subsystems[i].suites[j].fn(argc - 1, argv + 1, prefix); goto end; From ff676b193a401b23c84a79a7ec06559f3eaae917 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Wed, 11 Nov 2009 00:04:01 +0900 Subject: [PATCH 31/32] perf bench: Improve sched-pipe.c with more comfortable output This patch improves sched-pipe.c with more comfortable output. Change points are comment style description and formatting numerical values and its units. Example: | % ./perf bench sched pipe | # Running sched/pipe benchmark... | # Extecuted 1000000 pipe operations between two tasks | | Total time:5.822 [sec] | | 5.822553 usecs/op | 171745 ops/sec Signed-off-by: Hitoshi Mitake Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <1257865442-20252-3-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- tools/perf/bench/sched-pipe.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/tools/perf/bench/sched-pipe.c b/tools/perf/bench/sched-pipe.c index a9ac186714b4..238185f97977 100644 --- a/tools/perf/bench/sched-pipe.c +++ b/tools/perf/bench/sched-pipe.c @@ -92,17 +92,18 @@ int bench_sched_pipe(int argc, const char **argv, switch (bench_format) { case BENCH_FORMAT_DEFAULT: - printf("(executing %d pipe operations between two tasks)\n\n", + printf("# Extecuted %d pipe operations between two tasks\n\n", loops); result_usec = diff.tv_sec * 1000000; result_usec += diff.tv_usec; - printf("\tTotal time:%lu.%03lu sec\n", - diff.tv_sec, diff.tv_usec / 1000); - printf("\t\t%lf usecs/op\n", + printf(" %14s: %lu.%03lu [sec]\n\n", "Total time", + diff.tv_sec, diff.tv_usec/1000); + + printf(" %14lf usecs/op\n", (double)result_usec / (double)loops); - printf("\t\t%d ops/sec\n", + printf(" %14d ops/sec\n", (int)((double)loops / ((double)result_usec / (double)1000000))); break; From c5659b74f052150791750234f92dcfb29d27efa5 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Wed, 11 Nov 2009 00:04:02 +0900 Subject: [PATCH 32/32] perf bench: Improve sched-message.c with more comfortable output This patch improves sched-message.c with more comfortable output. Change points are comment style description and formatting numerical values and its units. Example: | % perf bench sched messaging | # Running sched/messaging benchmark... | # 20 sender and receiver processes per group | # 10 groups == 400 processes run | | Total time: 1.490 [sec] Signed-off-by: Hitoshi Mitake Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <1257865442-20252-4-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar Cc: Peter Zijlstra Cc: Paul Mackerras --- tools/perf/bench/sched-messaging.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/tools/perf/bench/sched-messaging.c b/tools/perf/bench/sched-messaging.c index 2cc5edcca3d6..605a2a959aa8 100644 --- a/tools/perf/bench/sched-messaging.c +++ b/tools/perf/bench/sched-messaging.c @@ -314,12 +314,12 @@ int bench_sched_messaging(int argc, const char **argv, switch (bench_format) { case BENCH_FORMAT_DEFAULT: - printf("(%d sender and receiver %s per group)\n", + printf("# %d sender and receiver %s per group\n", num_fds, thread_mode ? "threads" : "processes"); - printf("(%d groups == %d %s run)\n\n", + printf("# %d groups == %d %s run\n\n", num_groups, num_groups * 2 * num_fds, thread_mode ? "threads" : "processes"); - printf("\tTotal time:%lu.%03lu sec\n", + printf(" %14s: %lu.%03lu [sec]\n", "Total time", diff.tv_sec, diff.tv_usec/1000); break; case BENCH_FORMAT_SIMPLE: