From 2af4f9b8596afbbd7667a18fa71d117bac227dea Mon Sep 17 00:00:00 2001 From: Len Brown Date: Sat, 30 Jan 2021 23:43:10 -0500 Subject: tools/power turbostat: add built-in-counter for IPC -- Instructions per Cycle Use linux-perf to access the hardware instructions-retired counter. This is necessary because the counter is not enabled by default, and also the counter is prone to roll-over -- both of which perf manages. It is not necessary to use perf for the cycle counter, because turbostat already needs to collect delta-aperf to calcuate frequency. Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 84 +++++++++++++++++++++++++++++++++++ 1 file changed, 84 insertions(+) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index a7c4f0772e53..b82295eaa744 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -33,10 +33,13 @@ #include #include #include +#include +#include char *proc_stat = "/proc/stat"; FILE *outf; int *fd_percpu; +int *fd_instr_count_percpu; struct timeval interval_tv = {5, 0}; struct timespec interval_ts = {5, 0}; unsigned int num_iterations; @@ -75,6 +78,7 @@ char *output_buffer, *outp; unsigned int do_rapl; unsigned int do_dts; unsigned int do_ptm; +unsigned int do_ipc; unsigned long long gfx_cur_rc6_ms; unsigned long long cpuidle_cur_cpu_lpi_us; unsigned long long cpuidle_cur_sys_lpi_us; @@ -173,6 +177,7 @@ struct thread_data { unsigned long long aperf; unsigned long long mperf; unsigned long long c1; + unsigned long long instr_count; unsigned long long irq_count; unsigned int smi_count; unsigned int cpu_id; @@ -490,6 +495,39 @@ int get_msr_fd(int cpu) return fd; } +static long perf_event_open(struct perf_event_attr *hw_event, pid_t pid, int cpu, int group_fd, unsigned long flags) +{ + return syscall(__NR_perf_event_open, hw_event, pid, cpu, group_fd, flags); +} + +static int perf_instr_count_open(int cpu_num) +{ + struct perf_event_attr pea; + int fd; + + memset(&pea, 0, sizeof(struct perf_event_attr)); + pea.type = PERF_TYPE_HARDWARE; + pea.size = sizeof(struct perf_event_attr); + pea.config = PERF_COUNT_HW_INSTRUCTIONS; + + /* counter for cpu_num, including user + kernel and all processes */ + fd = perf_event_open(&pea, -1, cpu_num, -1, 0); + if (fd == -1) + err(-1, "cpu%d: perf instruction counter\n", cpu_num); + + return fd; +} + +int get_instr_count_fd(int cpu) +{ + if (fd_instr_count_percpu[cpu]) + return fd_instr_count_percpu[cpu]; + + fd_instr_count_percpu[cpu] = perf_instr_count_open(cpu); + + return fd_instr_count_percpu[cpu]; +} + int get_msr(int cpu, off_t offset, unsigned long long *msr) { ssize_t retval; @@ -561,6 +599,7 @@ struct msr_counter bic[] = { { 0x0, "X2APIC" }, { 0x0, "Die" }, { 0x0, "GFXAMHz" }, + { 0x0, "IPC" }, }; #define MAX_BIC (sizeof(bic) / sizeof(struct msr_counter)) @@ -616,6 +655,7 @@ struct msr_counter bic[] = { #define BIC_X2APIC (1ULL << 49) #define BIC_Die (1ULL << 50) #define BIC_GFXACTMHz (1ULL << 51) +#define BIC_IPC (1ULL << 52) #define BIC_DISABLED_BY_DEFAULT (BIC_USEC | BIC_TOD | BIC_APIC | BIC_X2APIC) @@ -627,6 +667,7 @@ unsigned long long bic_present = BIC_USEC | BIC_TOD | BIC_sysfs | BIC_APIC | BIC #define ENABLE_BIC(COUNTER_NAME) (bic_enabled |= COUNTER_NAME) #define BIC_PRESENT(COUNTER_BIT) (bic_present |= COUNTER_BIT) #define BIC_NOT_PRESENT(COUNTER_BIT) (bic_present &= ~COUNTER_BIT) +#define BIC_IS_ENABLED(COUNTER_BIT) (bic_enabled & COUNTER_BIT) #define MAX_DEFERRED 16 @@ -764,6 +805,9 @@ void print_header(char *delim) if (DO_BIC(BIC_TSC_MHz)) outp += sprintf(outp, "%sTSC_MHz", (printed++ ? delim : "")); + if (DO_BIC(BIC_IPC)) + outp += sprintf(outp, "%sIPC", (printed++ ? delim : "")); + if (DO_BIC(BIC_IRQ)) { if (sums_need_wide_columns) outp += sprintf(outp, "%s IRQ", (printed++ ? delim : "")); @@ -926,6 +970,9 @@ int dump_counters(struct thread_data *t, struct core_data *c, outp += sprintf(outp, "mperf: %016llX\n", t->mperf); outp += sprintf(outp, "c1: %016llX\n", t->c1); + if (DO_BIC(BIC_IPC)) + outp += sprintf(outp, "IPC: %lld\n", t->instr_count); + if (DO_BIC(BIC_IRQ)) outp += sprintf(outp, "IRQ: %lld\n", t->irq_count); if (DO_BIC(BIC_SMI)) @@ -1105,6 +1152,9 @@ int format_counters(struct thread_data *t, struct core_data *c, if (DO_BIC(BIC_TSC_MHz)) outp += sprintf(outp, "%s%.0f", (printed++ ? delim : ""), 1.0 * t->tsc/units/interval_float); + if (DO_BIC(BIC_IPC)) + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 1.0 * t->instr_count / t->aperf); + /* IRQ */ if (DO_BIC(BIC_IRQ)) { if (sums_need_wide_columns) @@ -1482,6 +1532,9 @@ delta_thread(struct thread_data *new, struct thread_data *old, old->mperf = 1; /* divide by 0 protection */ } + if (DO_BIC(BIC_IPC)) + old->instr_count = new->instr_count - old->instr_count; + if (DO_BIC(BIC_IRQ)) old->irq_count = new->irq_count - old->irq_count; @@ -1536,6 +1589,8 @@ void clear_counters(struct thread_data *t, struct core_data *c, struct pkg_data t->mperf = 0; t->c1 = 0; + t->instr_count = 0; + t->irq_count = 0; t->smi_count = 0; @@ -1611,6 +1666,8 @@ int sum_counters(struct thread_data *t, struct core_data *c, average.threads.mperf += t->mperf; average.threads.c1 += t->c1; + average.threads.instr_count += t->instr_count; + average.threads.irq_count += t->irq_count; average.threads.smi_count += t->smi_count; @@ -1707,6 +1764,7 @@ void compute_average(struct thread_data *t, struct core_data *c, average.threads.tsc /= topo.num_cpus; average.threads.aperf /= topo.num_cpus; average.threads.mperf /= topo.num_cpus; + average.threads.instr_count /= topo.num_cpus; average.threads.c1 /= topo.num_cpus; if (average.threads.irq_count > 9999999) @@ -1989,6 +2047,10 @@ retry: t->mperf = t->mperf * aperf_mperf_multiplier; } + if (DO_BIC(BIC_IPC)) + if (read(get_instr_count_fd(cpu), &t->instr_count, sizeof(long long)) != sizeof(long long)) + return -4; + if (DO_BIC(BIC_IRQ)) t->irq_count = irqs_per_cpu[cpu]; if (DO_BIC(BIC_SMI)) { @@ -5031,6 +5093,26 @@ void print_dev_latency(void) close(fd); } + +/* + * Linux-perf manages the the HW instructions-retired counter + * by enabling when requested, and hiding rollover + */ +void linux_perf_init(void) +{ + if (!BIC_IS_ENABLED(BIC_IPC)) + return; + + if (access("/proc/sys/kernel/perf_event_paranoid", F_OK)) + return; + + fd_instr_count_percpu = calloc(topo.max_cpu_num + 1, sizeof(int)); + if (fd_instr_count_percpu == NULL) + err(-1, "calloc fd_instr_count_percpu"); + + BIC_PRESENT(BIC_IPC); +} + void process_cpuid() { unsigned int eax, ebx, ecx, edx; @@ -5642,6 +5724,7 @@ void turbostat_init() check_dev_msr(); check_permissions(); process_cpuid(); + linux_perf_init(); if (!quiet) @@ -6087,6 +6170,7 @@ void cmdline(int argc, char **argv) {"debug", no_argument, 0, 'd'}, /* internal, not documented */ {"enable", required_argument, 0, 'e'}, {"interval", required_argument, 0, 'i'}, + {"IPC", no_argument, 0, 'I'}, {"num_iterations", required_argument, 0, 'n'}, {"help", no_argument, 0, 'h'}, {"hide", required_argument, 0, 'H'}, // meh, -h taken by --help -- cgit v1.2.3-59-g8ed1b From ed0757b83a00d1799c249073d688b018b82d0093 Mon Sep 17 00:00:00 2001 From: Len Brown Date: Thu, 4 Feb 2021 14:44:12 -0500 Subject: tools/power turbostat: print microcode patch level (also available via "grep microcode /proc/cpuinfo") Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index b82295eaa744..e1bc7937b1ec 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -5118,6 +5118,7 @@ void process_cpuid() unsigned int eax, ebx, ecx, edx; unsigned int fms, family, model, stepping, ecx_flags, edx_flags; unsigned int has_turbo; + unsigned long long ucode_patch = 0; eax = ebx = ecx = edx = 0; @@ -5131,8 +5132,8 @@ void process_cpuid() hygon_genuine = 1; if (!quiet) - fprintf(outf, "CPUID(0): %.4s%.4s%.4s ", - (char *)&ebx, (char *)&edx, (char *)&ecx); + fprintf(outf, "CPUID(0): %.4s%.4s%.4s 0x%x CPUID levels\n", + (char *)&ebx, (char *)&edx, (char *)&ecx, max_level); __cpuid(1, fms, ebx, ecx, edx); family = (fms >> 8) & 0xf; @@ -5145,6 +5146,9 @@ void process_cpuid() ecx_flags = ecx; edx_flags = edx; + if (get_msr(sched_getcpu(), MSR_IA32_UCODE_REV, &ucode_patch)) + warnx("get_msr(UCODE)\n"); + /* * check max extended function levels of CPUID. * This is needed to check for invariant TSC. @@ -5154,8 +5158,9 @@ void process_cpuid() __cpuid(0x80000000, max_extended_level, ebx, ecx, edx); if (!quiet) { - fprintf(outf, "0x%x CPUID levels; 0x%x xlevels; family:model:stepping 0x%x:%x:%x (%d:%d:%d)\n", - max_level, max_extended_level, family, model, stepping, family, model, stepping); + fprintf(outf, "CPUID(1): family:model:stepping 0x%x:%x:%x (%d:%d:%d) microcode 0x%x\n", + family, model, stepping, family, model, stepping, (unsigned int)((ucode_patch >> 32) & 0xFFFFFFFF)); + fprintf(outf, "CPUID(0x80000000): max_extended_levels: 0x%x\n", max_extended_level); fprintf(outf, "CPUID(1): %s %s %s %s %s %s %s %s %s %s\n", ecx_flags & (1 << 0) ? "SSE3" : "-", ecx_flags & (1 << 3) ? "MONITOR" : "-", -- cgit v1.2.3-59-g8ed1b From 5683460b85a8a14c5eec10e363635ad4660eb961 Mon Sep 17 00:00:00 2001 From: Chen Yu Date: Wed, 3 Feb 2021 16:19:59 +0800 Subject: tools/power turbostat: Support Alder Lake Mobile Share the code between Alder Lake Mobile and Alder Lake Desktop. Signed-off-by: Chen Yu Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index e1bc7937b1ec..a4745825047f 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -5056,6 +5056,7 @@ unsigned int intel_model_duplicates(unsigned int model) case INTEL_FAM6_ROCKETLAKE: case INTEL_FAM6_LAKEFIELD: case INTEL_FAM6_ALDERLAKE: + case INTEL_FAM6_ALDERLAKE_L: return INTEL_FAM6_CANNONLAKE_L; case INTEL_FAM6_ATOM_TREMONT_L: -- cgit v1.2.3-59-g8ed1b From 6c5c656006cf314196faea7bd76eebbfa0941cd1 Mon Sep 17 00:00:00 2001 From: Chen Yu Date: Wed, 3 Feb 2021 16:26:32 +0800 Subject: tools/power turbostat: Support Ice Lake D Ice Lake D is low-end server version of Ice Lake X, reuse the code accordingly. Tested-by: Wendy Wang Signed-off-by: Chen Yu Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index a4745825047f..d27e899328f9 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -5063,6 +5063,7 @@ unsigned int intel_model_duplicates(unsigned int model) return INTEL_FAM6_ATOM_TREMONT; case INTEL_FAM6_ICELAKE_X: + case INTEL_FAM6_ICELAKE_D: case INTEL_FAM6_SAPPHIRERAPIDS_X: return INTEL_FAM6_SKYLAKE_X; } -- cgit v1.2.3-59-g8ed1b From b2b94be787bf47eedd5890a249f3318bf9f1f1d5 Mon Sep 17 00:00:00 2001 From: Len Brown Date: Thu, 11 Mar 2021 18:36:35 -0500 Subject: Revert "tools/power turbostat: adjust for temperature offset" This reverts commit 6ff7cb371c4bea3dba03a56d774da925e78a5087. Apparently the TCC offset should not be used to adjust what temperature we show the user after all. (on most systems, TCC offset is 0, FWIW) Fixes: 6ff7cb371c4b Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 62 +++++++++++++++++++---------------- 1 file changed, 33 insertions(+), 29 deletions(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index d27e899328f9..98a0a731da8a 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -4879,33 +4879,12 @@ double discover_bclk(unsigned int family, unsigned int model) * below this value, including the Digital Thermal Sensor (DTS), * Package Thermal Management Sensor (PTM), and thermal event thresholds. */ -int read_tcc_activation_temp() +int set_temperature_target(struct thread_data *t, struct core_data *c, struct pkg_data *p) { unsigned long long msr; - unsigned int tcc, target_c, offset_c; - - /* Temperature Target MSR is Nehalem and newer only */ - if (!do_nhm_platform_info) - return 0; - - if (get_msr(base_cpu, MSR_IA32_TEMPERATURE_TARGET, &msr)) - return 0; - - target_c = (msr >> 16) & 0xFF; - - offset_c = (msr >> 24) & 0xF; - - tcc = target_c - offset_c; - - if (!quiet) - fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C) (%d default - %d offset)\n", - base_cpu, msr, tcc, target_c, offset_c); - - return tcc; -} + unsigned int target_c_local; + int cpu; -int set_temperature_target(struct thread_data *t, struct core_data *c, struct pkg_data *p) -{ /* tcc_activation_temp is used only for dts or ptm */ if (!(do_dts || do_ptm)) return 0; @@ -4914,18 +4893,43 @@ int set_temperature_target(struct thread_data *t, struct core_data *c, struct pk if (!(t->flags & CPU_IS_FIRST_THREAD_IN_CORE) || !(t->flags & CPU_IS_FIRST_CORE_IN_PACKAGE)) return 0; + cpu = t->cpu_id; + if (cpu_migrate(cpu)) { + fprintf(outf, "Could not migrate to CPU %d\n", cpu); + return -1; + } + if (tcc_activation_temp_override != 0) { tcc_activation_temp = tcc_activation_temp_override; - fprintf(outf, "Using cmdline TCC Target (%d C)\n", tcc_activation_temp); + fprintf(outf, "cpu%d: Using cmdline TCC Target (%d C)\n", + cpu, tcc_activation_temp); return 0; } - tcc_activation_temp = read_tcc_activation_temp(); - if (tcc_activation_temp) - return 0; + /* Temperature Target MSR is Nehalem and newer only */ + if (!do_nhm_platform_info) + goto guess; + + if (get_msr(base_cpu, MSR_IA32_TEMPERATURE_TARGET, &msr)) + goto guess; + + target_c_local = (msr >> 16) & 0xFF; + + if (!quiet) + fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C)\n", + cpu, msr, target_c_local); + + if (!target_c_local) + goto guess; + + tcc_activation_temp = target_c_local; + + return 0; +guess: tcc_activation_temp = TJMAX_DEFAULT; - fprintf(outf, "Guessing tjMax %d C, Please use -T to specify\n", tcc_activation_temp); + fprintf(outf, "cpu%d: Guessing tjMax %d C, Please use -T to specify\n", + cpu, tcc_activation_temp); return 0; } -- cgit v1.2.3-59-g8ed1b From abdc75ab53b7fd2ef42c79e88cf0caf2d007c4f2 Mon Sep 17 00:00:00 2001 From: Zhang Rui Date: Thu, 11 Mar 2021 10:05:13 +0800 Subject: tools/power turbostat: Fix DRAM Energy Unit on SKX SKX uses fixed DRAM Energy Unit, just like HSX and BDX. Signed-off-by: Zhang Rui Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 98a0a731da8a..5afe85efca5c 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -4272,6 +4272,7 @@ rapl_dram_energy_units_probe(int model, double rapl_energy_units) switch (model) { case INTEL_FAM6_HASWELL_X: /* HSX */ case INTEL_FAM6_BROADWELL_X: /* BDX */ + case INTEL_FAM6_SKYLAKE_X: /* SKX */ case INTEL_FAM6_XEON_PHI_KNL: /* KNL */ return (rapl_dram_energy_units = 15.3 / 1000000); default: -- cgit v1.2.3-59-g8ed1b From ba58ecde5eec898f647bba7cb07e6ec6ea1b875c Mon Sep 17 00:00:00 2001 From: Len Brown Date: Fri, 12 Mar 2021 17:30:30 -0500 Subject: tools/power turbostat: update version number --- tools/power/x86/turbostat/turbostat.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 5afe85efca5c..ace100dd5a83 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -5834,7 +5834,7 @@ int get_and_dump_counters(void) } void print_version() { - fprintf(outf, "turbostat version 20.09.30" + fprintf(outf, "turbostat version 21.03.12" " - Len Brown \n"); } -- cgit v1.2.3-59-g8ed1b From 301b1d3a9104f4f3a8ab4171cf88d0f55d632b41 Mon Sep 17 00:00:00 2001 From: Bas Nieuwenhuizen Date: Wed, 28 Apr 2021 17:09:03 +0800 Subject: tools/power/turbostat: Fix turbostat for AMD Zen CPUs It was reported that on Zen+ system turbostat started exiting, which was tracked down to the MSR_PKG_ENERGY_STAT read failing because offset_to_idx wasn't returning a non-negative index. This patch combined the modification from Bingsong Si and Bas Nieuwenhuizen and addd the MSR to the index system as alternative for MSR_PKG_ENERGY_STATUS. Fixes: 9972d5d84d76 ("tools/power turbostat: Enable accumulate RAPL display") Reported-by: youling257 Tested-by: youling257 Tested-by: Kurt Garloff Tested-by: Bingsong Si Tested-by: Artem S. Tashkinov Co-developed-by: Bingsong Si Co-developed-by: Terry Bowman Signed-off-by: Bas Nieuwenhuizen Reviewed-by: Chen Yu Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index ace100dd5a83..b5f4ec24fea9 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -302,7 +302,10 @@ int idx_to_offset(int idx) switch (idx) { case IDX_PKG_ENERGY: - offset = MSR_PKG_ENERGY_STATUS; + if (do_rapl & RAPL_AMD_F17H) + offset = MSR_PKG_ENERGY_STAT; + else + offset = MSR_PKG_ENERGY_STATUS; break; case IDX_DRAM_ENERGY: offset = MSR_DRAM_ENERGY_STATUS; @@ -331,6 +334,7 @@ int offset_to_idx(int offset) switch (offset) { case MSR_PKG_ENERGY_STATUS: + case MSR_PKG_ENERGY_STAT: idx = IDX_PKG_ENERGY; break; case MSR_DRAM_ENERGY_STATUS: @@ -358,7 +362,7 @@ int idx_valid(int idx) { switch (idx) { case IDX_PKG_ENERGY: - return do_rapl & RAPL_PKG; + return do_rapl & (RAPL_PKG | RAPL_AMD_F17H); case IDX_DRAM_ENERGY: return do_rapl & RAPL_DRAM; case IDX_PP0_ENERGY: -- cgit v1.2.3-59-g8ed1b From 13a779de4175df602366d129e41782ad7168cef0 Mon Sep 17 00:00:00 2001 From: Calvin Walton Date: Wed, 28 Apr 2021 17:09:16 +0800 Subject: tools/power turbostat: Fix offset overflow issue in index converting The idx_to_offset() function returns type int (32-bit signed), but MSR_PKG_ENERGY_STAT is u32 and would be interpreted as a negative number. The end result is that it hits the if (offset < 0) check in update_msr_sum() which prevents the timer callback from updating the stat in the background when long durations are used. The similar issue exists in offset_to_idx() and update_msr_sum(). Fix this issue by converting the 'int' to 'off_t' accordingly. Fixes: 9972d5d84d76 ("tools/power turbostat: Enable accumulate RAPL display") Signed-off-by: Calvin Walton Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index b5f4ec24fea9..f3cb06a115b5 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -296,9 +296,9 @@ struct msr_sum_array { /* The percpu MSR sum array.*/ struct msr_sum_array *per_cpu_msr_sum; -int idx_to_offset(int idx) +off_t idx_to_offset(int idx) { - int offset; + off_t offset; switch (idx) { case IDX_PKG_ENERGY: @@ -328,7 +328,7 @@ int idx_to_offset(int idx) return offset; } -int offset_to_idx(int offset) +int offset_to_idx(off_t offset) { int idx; @@ -3338,7 +3338,7 @@ static int update_msr_sum(struct thread_data *t, struct core_data *c, struct pkg for (i = IDX_PKG_ENERGY; i < IDX_COUNT; i++) { unsigned long long msr_cur, msr_last; - int offset; + off_t offset; if (!idx_valid(i)) continue; @@ -3347,7 +3347,8 @@ static int update_msr_sum(struct thread_data *t, struct core_data *c, struct pkg continue; ret = get_msr(cpu, offset, &msr_cur); if (ret) { - fprintf(outf, "Can not update msr(0x%x)\n", offset); + fprintf(outf, "Can not update msr(0x%llx)\n", + (unsigned long long)offset); continue; } -- cgit v1.2.3-59-g8ed1b From 25368d7cefcd87a94ccabcc6f9f31796607bbe4e Mon Sep 17 00:00:00 2001 From: Chen Yu Date: Tue, 4 May 2021 17:52:34 +0300 Subject: tools/power/turbostat: Remove Package C6 Retention on Ice Lake Server Currently the turbostat treats ICX the same way as SKX and shares the code among them. But one difference is that ICX does not support Package C6 Retention, unlike SKX and CLX. So this patch: 1. Splitting SKX and ICX in turbostat. 2. Removing Package C6 Rentention for ICX. And after this split, it would be easier to cutomize Ice Lake Server in turbostat in the future. Suggested-by: Artem Bityutskiy Signed-off-by: Chen Yu Reviewed-by: Artem Bityutskiy Tested-by: Artem Bityutskiy Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 36 +++++++++++++++++++++++++++++------ 1 file changed, 30 insertions(+), 6 deletions(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index f3cb06a115b5..b43816e4c1ff 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -2263,7 +2263,7 @@ int amt_pkg_cstate_limits[16] = {PCLUNL, PCL__1, PCL__2, PCLRSV, PCLRSV, PCLRSV, int phi_pkg_cstate_limits[16] = {PCL__0, PCL__2, PCL_6N, PCL_6R, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV}; int glm_pkg_cstate_limits[16] = {PCLUNL, PCL__1, PCL__3, PCL__6, PCL__7, PCL_7S, PCL__8, PCL__9, PCL_10, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV}; int skx_pkg_cstate_limits[16] = {PCL__0, PCL__2, PCL_6N, PCL_6R, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV}; - +int icx_pkg_cstate_limits[16] = {PCL__0, PCL__2, PCL__6, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV}; static void calculate_tsc_tweak() @@ -2378,6 +2378,7 @@ int has_turbo_ratio_group_limits(int family, int model) switch (model) { case INTEL_FAM6_ATOM_GOLDMONT: case INTEL_FAM6_SKYLAKE_X: + case INTEL_FAM6_ICELAKE_X: case INTEL_FAM6_ATOM_GOLDMONT_D: case INTEL_FAM6_ATOM_TREMONT_D: return 1; @@ -3618,6 +3619,10 @@ int probe_nhm_msrs(unsigned int family, unsigned int model) pkg_cstate_limits = skx_pkg_cstate_limits; has_misc_feature_control = 1; break; + case INTEL_FAM6_ICELAKE_X: /* ICX */ + pkg_cstate_limits = icx_pkg_cstate_limits; + has_misc_feature_control = 1; + break; case INTEL_FAM6_ATOM_SILVERMONT: /* BYT */ no_MSR_MISC_PWR_MGMT = 1; case INTEL_FAM6_ATOM_SILVERMONT_D: /* AVN */ @@ -3706,6 +3711,20 @@ int is_skx(unsigned int family, unsigned int model) } return 0; } + +int is_icx(unsigned int family, unsigned int model) +{ + + if (!genuine_intel) + return 0; + + switch (model) { + case INTEL_FAM6_ICELAKE_X: + return 1; + } + return 0; +} + int is_ehl(unsigned int family, unsigned int model) { if (!genuine_intel) @@ -3808,6 +3827,7 @@ int has_glm_turbo_ratio_limit(unsigned int family, unsigned int model) switch (model) { case INTEL_FAM6_ATOM_GOLDMONT: case INTEL_FAM6_SKYLAKE_X: + case INTEL_FAM6_ICELAKE_X: return 1; default: return 0; @@ -3833,6 +3853,7 @@ int has_config_tdp(unsigned int family, unsigned int model) case INTEL_FAM6_SKYLAKE_L: /* SKL */ case INTEL_FAM6_CANNONLAKE_L: /* CNL */ case INTEL_FAM6_SKYLAKE_X: /* SKX */ + case INTEL_FAM6_ICELAKE_X: /* ICX */ case INTEL_FAM6_XEON_PHI_KNL: /* Knights Landing */ return 1; @@ -4363,6 +4384,7 @@ void rapl_probe_intel(unsigned int family, unsigned int model) case INTEL_FAM6_HASWELL_X: /* HSX */ case INTEL_FAM6_BROADWELL_X: /* BDX */ case INTEL_FAM6_SKYLAKE_X: /* SKX */ + case INTEL_FAM6_ICELAKE_X: /* ICX */ case INTEL_FAM6_XEON_PHI_KNL: /* KNL */ do_rapl = RAPL_PKG | RAPL_DRAM | RAPL_DRAM_POWER_INFO | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS | RAPL_PKG_POWER_INFO; BIC_PRESENT(BIC_PKG__); @@ -4519,7 +4541,8 @@ void perf_limit_reasons_probe(unsigned int family, unsigned int model) void automatic_cstate_conversion_probe(unsigned int family, unsigned int model) { - if (is_skx(family, model) || is_bdx(family, model)) + if (is_skx(family, model) || is_bdx(family, model) || + is_icx(family, model)) has_automatic_cstate_conversion = 1; } @@ -4734,6 +4757,7 @@ int has_snb_msrs(unsigned int family, unsigned int model) case INTEL_FAM6_SKYLAKE_L: /* SKL */ case INTEL_FAM6_CANNONLAKE_L: /* CNL */ case INTEL_FAM6_SKYLAKE_X: /* SKX */ + case INTEL_FAM6_ICELAKE_X: /* ICX */ case INTEL_FAM6_ATOM_GOLDMONT: /* BXT */ case INTEL_FAM6_ATOM_GOLDMONT_PLUS: case INTEL_FAM6_ATOM_GOLDMONT_D: /* DNV */ @@ -5072,10 +5096,9 @@ unsigned int intel_model_duplicates(unsigned int model) case INTEL_FAM6_ATOM_TREMONT_L: return INTEL_FAM6_ATOM_TREMONT; - case INTEL_FAM6_ICELAKE_X: case INTEL_FAM6_ICELAKE_D: case INTEL_FAM6_SAPPHIRERAPIDS_X: - return INTEL_FAM6_SKYLAKE_X; + return INTEL_FAM6_ICELAKE_X; } return model; } @@ -5185,8 +5208,9 @@ void process_cpuid() edx_flags & (1 << 28) ? "HT" : "-", edx_flags & (1 << 29) ? "TM" : "-"); } - if (genuine_intel) + if (genuine_intel) { model = intel_model_duplicates(model); + } if (!(edx_flags & (1 << 5))) errx(1, "CPUID: no MSR"); @@ -5364,7 +5388,7 @@ void process_cpuid() BIC_NOT_PRESENT(BIC_Pkgpc7); use_c1_residency_msr = 1; } - if (is_skx(family, model)) { + if (is_skx(family, model) || is_icx(family, model)) { BIC_NOT_PRESENT(BIC_CPU_c3); BIC_NOT_PRESENT(BIC_Pkgpc3); BIC_NOT_PRESENT(BIC_CPU_c7); -- cgit v1.2.3-59-g8ed1b From 1e3ec5cdfb63bc2a1ff06145faa2be08d6ec9594 Mon Sep 17 00:00:00 2001 From: Randy Dunlap Date: Thu, 25 Mar 2021 13:13:33 -0700 Subject: tools/power turbostat: unmark non-kernel-doc comment Do not mark a comment as kernel-doc notation when it is not meant to be in kernel-doc notation. Signed-off-by: Randy Dunlap Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index b43816e4c1ff..407e80e72546 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -2516,7 +2516,7 @@ dump_knl_turbo_ratio_limits(void) fprintf(outf, "cpu%d: MSR_TURBO_RATIO_LIMIT: 0x%08llx\n", base_cpu, msr); - /** + /* * Turbo encoding in KNL is as follows: * [0] -- Reserved * [7:1] -- Base value of number of active cores of bucket 1. -- cgit v1.2.3-59-g8ed1b From 8c69da293041352d15a2b6e8010c141822a416c5 Mon Sep 17 00:00:00 2001 From: Chen Yu Date: Wed, 28 Apr 2021 10:51:57 +0800 Subject: tools/power turbostat: Enable tsc_tweak for Elkhart Lake and Jasper Lake It was found that on Elkhart Lake the TSC frequency is driven by a separate crystal-clock domain, which is different from the BCLK domain which includes mperf. This has result in small different speed thus inconsistence between TSC and the mperf, which caused the Busy% to be higher than 100%. On this platform it seems that the mperf runs faster than tsc when the CPU is 100% utilized: delta tsc(18815473183) < delta mperf(18958403680) for 10 seconds. To align TSC with mperf, leverage the tsc_tweak mechanism introduced for cores newer than Skylake, so that TSC and mperf would be calculated in the same domain. Reported-by: Zhang Rui Signed-off-by: Chen Yu Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 407e80e72546..9ec13f06c0f3 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -5440,7 +5440,7 @@ void process_cpuid() if (!quiet) dump_sysfs_pstate_config(); - if (has_skl_msrs(family, model)) + if (has_skl_msrs(family, model) || is_ehl(family, model)) calculate_tsc_tweak(); if (!access("/sys/class/drm/card0/power/rc6_residency_ms", R_OK)) -- cgit v1.2.3-59-g8ed1b From aeb01e6d71ffaf3011ac755c3083cc200ed57cb4 Mon Sep 17 00:00:00 2001 From: Chen Yu Date: Wed, 28 Apr 2021 12:18:12 +0800 Subject: tools/power turbostat: Print the C-state Pre-wake settings C-state pre-wake setting[1] is an optimization for some Intel CPUs to be woken up from deep C-states in order to reduce latency. According to the spec, the BIT30 is the C-state Pre-wake Disable. Expose this setting accordingly. Sample output from turbostat: ... cpu51: MSR_IA32_POWER_CTL: 0x1a00a40059 (C1E auto-promotion: DISabled) C-state Pre-wake: ENabled cpu51: MSR_TURBO_RATIO_LIMIT: 0x2021212121212224 ... [1] https://intel.github.io/wult/#c-state-pre-wake Signed-off-by: Chen Yu Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 9ec13f06c0f3..e1ed14c666db 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -91,6 +91,7 @@ double rapl_dram_energy_units, rapl_energy_units; double rapl_joule_counter_range; unsigned int do_core_perf_limit_reasons; unsigned int has_automatic_cstate_conversion; +unsigned int dis_cstate_prewake; unsigned int do_gfx_perf_limit_reasons; unsigned int do_ring_perf_limit_reasons; unsigned int crystal_hz; @@ -2271,6 +2272,8 @@ calculate_tsc_tweak() tsc_tweak = base_hz / tsc_hz; } +void prewake_cstate_probe(unsigned int family, unsigned int model); + static void dump_nhm_platform_info(void) { @@ -2293,6 +2296,11 @@ dump_nhm_platform_info(void) fprintf(outf, "cpu%d: MSR_IA32_POWER_CTL: 0x%08llx (C1E auto-promotion: %sabled)\n", base_cpu, msr, msr & 0x2 ? "EN" : "DIS"); + /* C-state Pre-wake Disable (CSTATE_PREWAKE_DISABLE) */ + if (dis_cstate_prewake) + fprintf(outf, "C-state Pre-wake: %sabled\n", + msr & 0x40000000 ? "DIS" : "EN"); + return; } @@ -4546,6 +4554,12 @@ void automatic_cstate_conversion_probe(unsigned int family, unsigned int model) has_automatic_cstate_conversion = 1; } +void prewake_cstate_probe(unsigned int family, unsigned int model) +{ + if (is_icx(family, model)) + dis_cstate_prewake = 1; +} + int print_thermal(struct thread_data *t, struct core_data *c, struct pkg_data *p) { unsigned long long msr; -- cgit v1.2.3-59-g8ed1b From 7ab5ff4937a338783d147ec2d8c8714f48a5de79 Mon Sep 17 00:00:00 2001 From: Zhang Rui Date: Wed, 21 Apr 2021 22:22:47 +0800 Subject: tools/power turbostat: Fix Core C6 residency on Atom CPUs For Atom CPUs that have core cstate deeper than C6, MSR_CORE_C6_RESIDENCY actually returns the residency for both CC6 and deeper Core cstates. Thus, the real Core C6 residency should be the subtraction of MSR_CORE_C6_RESIDENCY return value and MSR_CORE_C6_RESIDENCY return value. Signed-off-by: Zhang Rui Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 39 ++++++++++++++++++++++++++++++++++- 1 file changed, 38 insertions(+), 1 deletion(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index e1ed14c666db..ee18966f65a4 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -35,6 +35,7 @@ #include #include #include +#include char *proc_stat = "/proc/stat"; FILE *outf; @@ -185,6 +186,7 @@ struct thread_data { unsigned int apic_id; unsigned int x2apic_id; unsigned int flags; + bool is_atom; #define CPU_IS_FIRST_THREAD_IN_CORE 0x2 #define CPU_IS_FIRST_CORE_IN_PACKAGE 0x4 unsigned long long counter[MAX_ADDED_THREAD_COUNTERS]; @@ -2090,9 +2092,19 @@ retry: return -7; } - if (DO_BIC(BIC_CPU_c7) || soft_c1_residency_display(BIC_CPU_c7)) + if (DO_BIC(BIC_CPU_c7) || soft_c1_residency_display(BIC_CPU_c7)) { if (get_msr(cpu, MSR_CORE_C7_RESIDENCY, &c->c7)) return -8; + else if (t->is_atom) { + /* + * For Atom CPUs that has core cstate deeper than c6, + * MSR_CORE_C6_RESIDENCY returns residency of cc6 and deeper. + * Minus CC7 (and deeper cstates) residency to get + * accturate cc6 residency. + */ + c->c6 -= c->c7; + } + } if (DO_BIC(BIC_Mod_c6)) if (get_msr(cpu, MSR_MODULE_C6_RES_MS, &c->mc6_us)) @@ -4911,6 +4923,28 @@ double discover_bclk(unsigned int family, unsigned int model) return 133.33; } +int get_cpu_type(struct thread_data *t, struct core_data *c, struct pkg_data *p) +{ + unsigned int eax, ebx, ecx, edx; + + if (!genuine_intel) + return 0; + + if (cpu_migrate(t->cpu_id)) { + fprintf(outf, "Could not migrate to CPU %d\n", t->cpu_id); + return -1; + } + + if (max_level < 0x1a) + return 0; + + __cpuid(0x1a, eax, ebx, ecx, edx); + eax = (eax >> 24) & 0xFF; + if (eax == 0x20 ) + t->is_atom = true; + return 0; +} + /* * MSR_IA32_TEMPERATURE_TARGET indicates the temperature where * the Thermal Control Circuit (TCC) activates. @@ -5796,6 +5830,9 @@ void turbostat_init() for_all_cpus(set_temperature_target, ODD_COUNTERS); + for_all_cpus(get_cpu_type, ODD_COUNTERS); + for_all_cpus(get_cpu_type, EVEN_COUNTERS); + if (!quiet) for_all_cpus(print_thermal, ODD_COUNTERS); -- cgit v1.2.3-59-g8ed1b From e9d3092f6d7c21031c8ac10ba2016ae0482a39fe Mon Sep 17 00:00:00 2001 From: Zhang Rui Date: Mon, 26 Apr 2021 10:05:27 +0800 Subject: tools/power turbostat: save original CPU model CPU model may get changed in intel_model_duplicates() for code reuse. But there are still some cases we need the original CPU model to handle minor differences between generations. Thus save the original CPU model. Signed-off-by: Zhang Rui Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index ee18966f65a4..d1ae6b248377 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -43,6 +43,10 @@ int *fd_percpu; int *fd_instr_count_percpu; struct timeval interval_tv = {5, 0}; struct timespec interval_ts = {5, 0}; + +/* Save original CPU model */ +unsigned int model_orig; + unsigned int num_iterations; unsigned int debug; unsigned int quiet; @@ -5257,6 +5261,7 @@ void process_cpuid() edx_flags & (1 << 29) ? "TM" : "-"); } if (genuine_intel) { + model_orig = model; model = intel_model_duplicates(model); } -- cgit v1.2.3-59-g8ed1b From 0b9a0b9be991656f125b58a240065cdf72077244 Mon Sep 17 00:00:00 2001 From: Zhang Rui Date: Wed, 21 Apr 2021 23:22:14 +0800 Subject: tools/power turbostat: add TCC Offset support The length of TCC Offset bits varies on different platforms. Decode TCC Offset bits only for the platforms that we have verified. For the others, only show default TCC activation temperature. Signed-off-by: Zhang Rui Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 58 +++++++++++++++++++++++++++++++++-- 1 file changed, 55 insertions(+), 3 deletions(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index d1ae6b248377..6326bee97c0b 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -91,6 +91,7 @@ unsigned int gfx_cur_mhz; unsigned int gfx_act_mhz; unsigned int tcc_activation_temp; unsigned int tcc_activation_temp_override; +int tcc_offset_bits; double rapl_power_units, rapl_time_units; double rapl_dram_energy_units, rapl_energy_units; double rapl_joule_counter_range; @@ -3886,6 +3887,40 @@ int has_config_tdp(unsigned int family, unsigned int model) } } +/* + * tcc_offset_bits: + * 0: Tcc Offset not supported (Default) + * 6: Bit 29:24 of MSR_PLATFORM_INFO + * 4: Bit 27:24 of MSR_PLATFORM_INFO + */ +void check_tcc_offset(int model) +{ + unsigned long long msr; + + if (!genuine_intel) + return; + + switch (model) { + case INTEL_FAM6_SKYLAKE_L: + case INTEL_FAM6_SKYLAKE: + case INTEL_FAM6_KABYLAKE_L: + case INTEL_FAM6_KABYLAKE: + case INTEL_FAM6_ICELAKE_L: + case INTEL_FAM6_ICELAKE: + case INTEL_FAM6_TIGERLAKE_L: + case INTEL_FAM6_TIGERLAKE: + case INTEL_FAM6_COMETLAKE: + if (!get_msr(base_cpu, MSR_PLATFORM_INFO, &msr)) { + msr = (msr >> 30) & 1; + if (msr) + tcc_offset_bits = 6; + } + return; + default: + return; + } +} + static void remove_underbar(char *s) { @@ -4964,7 +4999,7 @@ int get_cpu_type(struct thread_data *t, struct core_data *c, struct pkg_data *p) int set_temperature_target(struct thread_data *t, struct core_data *c, struct pkg_data *p) { unsigned long long msr; - unsigned int target_c_local; + unsigned int target_c_local, tcc_offset; int cpu; /* tcc_activation_temp is used only for dts or ptm */ @@ -4997,9 +5032,24 @@ int set_temperature_target(struct thread_data *t, struct core_data *c, struct pk target_c_local = (msr >> 16) & 0xFF; - if (!quiet) - fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C)\n", + if (!quiet) { + switch (tcc_offset_bits) { + case 4: + tcc_offset = (msr >> 24) & 0xF; + fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C) (%d default - %d offset)\n", + cpu, msr, target_c_local - tcc_offset, target_c_local, tcc_offset); + break; + case 6: + tcc_offset = (msr >> 24) & 0x3F; + fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C) (%d default - %d offset)\n", + cpu, msr, target_c_local - tcc_offset, target_c_local, tcc_offset); + break; + default: + fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C)\n", cpu, msr, target_c_local); + break; + } + } if (!target_c_local) goto guess; @@ -5483,6 +5533,8 @@ void process_cpuid() perf_limit_reasons_probe(family, model); automatic_cstate_conversion_probe(family, model); + check_tcc_offset(model_orig); + if (!quiet) dump_cstate_pstate_config_info(family, model); -- cgit v1.2.3-59-g8ed1b From 55279aef754c5eab170077ae4ba4ebd304dea64f Mon Sep 17 00:00:00 2001 From: Zhang Rui Date: Mon, 26 Apr 2021 18:49:26 +0800 Subject: tools/power turbostat: rename tcc variables There are two TCC activation temeprature. One is the default TCC activation temperature, also known as TJ_MAX. Another one is the effective TCC activation temperature, which is the subtraction of default TCC activation temperature and TCC offset. The name of variable tcc_activation_temp might be misleading here. Thus rename tcc_activation_temp to tj_max, and use tcc_default and tcc_offset to calculate the effective TCC activation temperature. No functional change in this patch. Signed-off-by: Zhang Rui Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 44 +++++++++++++++++------------------ 1 file changed, 22 insertions(+), 22 deletions(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 6326bee97c0b..8f0a1d8a0366 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -89,8 +89,8 @@ unsigned long long cpuidle_cur_cpu_lpi_us; unsigned long long cpuidle_cur_sys_lpi_us; unsigned int gfx_cur_mhz; unsigned int gfx_act_mhz; -unsigned int tcc_activation_temp; -unsigned int tcc_activation_temp_override; +unsigned int tj_max; +unsigned int tj_max_override; int tcc_offset_bits; double rapl_power_units, rapl_time_units; double rapl_dram_energy_units, rapl_energy_units; @@ -2118,7 +2118,7 @@ retry: if (DO_BIC(BIC_CoreTmp)) { if (get_msr(cpu, MSR_IA32_THERM_STATUS, &msr)) return -9; - c->core_temp_c = tcc_activation_temp - ((msr >> 16) & 0x7F); + c->core_temp_c = tj_max - ((msr >> 16) & 0x7F); } if (do_rapl & RAPL_AMD_F17H) { @@ -2224,7 +2224,7 @@ retry: if (DO_BIC(BIC_PkgTmp)) { if (get_msr(cpu, MSR_IA32_PACKAGE_THERM_STATUS, &msr)) return -17; - p->pkg_temp_c = tcc_activation_temp - ((msr >> 16) & 0x7F); + p->pkg_temp_c = tj_max - ((msr >> 16) & 0x7F); } if (DO_BIC(BIC_GFX_rc6)) @@ -4637,7 +4637,7 @@ int print_thermal(struct thread_data *t, struct core_data *c, struct pkg_data *p dts = (msr >> 16) & 0x7F; fprintf(outf, "cpu%d: MSR_IA32_PACKAGE_THERM_STATUS: 0x%08llx (%d C)\n", - cpu, msr, tcc_activation_temp - dts); + cpu, msr, tj_max - dts); if (get_msr(cpu, MSR_IA32_PACKAGE_THERM_INTERRUPT, &msr)) return 0; @@ -4645,7 +4645,7 @@ int print_thermal(struct thread_data *t, struct core_data *c, struct pkg_data *p dts = (msr >> 16) & 0x7F; dts2 = (msr >> 8) & 0x7F; fprintf(outf, "cpu%d: MSR_IA32_PACKAGE_THERM_INTERRUPT: 0x%08llx (%d C, %d C)\n", - cpu, msr, tcc_activation_temp - dts, tcc_activation_temp - dts2); + cpu, msr, tj_max - dts, tj_max - dts2); } @@ -4658,7 +4658,7 @@ int print_thermal(struct thread_data *t, struct core_data *c, struct pkg_data *p dts = (msr >> 16) & 0x7F; resolution = (msr >> 27) & 0xF; fprintf(outf, "cpu%d: MSR_IA32_THERM_STATUS: 0x%08llx (%d C +/- %d)\n", - cpu, msr, tcc_activation_temp - dts, resolution); + cpu, msr, tj_max - dts, resolution); if (get_msr(cpu, MSR_IA32_THERM_INTERRUPT, &msr)) return 0; @@ -4666,7 +4666,7 @@ int print_thermal(struct thread_data *t, struct core_data *c, struct pkg_data *p dts = (msr >> 16) & 0x7F; dts2 = (msr >> 8) & 0x7F; fprintf(outf, "cpu%d: MSR_IA32_THERM_INTERRUPT: 0x%08llx (%d C, %d C)\n", - cpu, msr, tcc_activation_temp - dts, tcc_activation_temp - dts2); + cpu, msr, tj_max - dts, tj_max - dts2); } return 0; @@ -4999,10 +4999,10 @@ int get_cpu_type(struct thread_data *t, struct core_data *c, struct pkg_data *p) int set_temperature_target(struct thread_data *t, struct core_data *c, struct pkg_data *p) { unsigned long long msr; - unsigned int target_c_local, tcc_offset; + unsigned int tcc_default, tcc_offset; int cpu; - /* tcc_activation_temp is used only for dts or ptm */ + /* tj_max is used only for dts or ptm */ if (!(do_dts || do_ptm)) return 0; @@ -5016,10 +5016,10 @@ int set_temperature_target(struct thread_data *t, struct core_data *c, struct pk return -1; } - if (tcc_activation_temp_override != 0) { - tcc_activation_temp = tcc_activation_temp_override; + if (tj_max_override != 0) { + tj_max = tj_max_override; fprintf(outf, "cpu%d: Using cmdline TCC Target (%d C)\n", - cpu, tcc_activation_temp); + cpu, tj_max); return 0; } @@ -5030,38 +5030,38 @@ int set_temperature_target(struct thread_data *t, struct core_data *c, struct pk if (get_msr(base_cpu, MSR_IA32_TEMPERATURE_TARGET, &msr)) goto guess; - target_c_local = (msr >> 16) & 0xFF; + tcc_default = (msr >> 16) & 0xFF; if (!quiet) { switch (tcc_offset_bits) { case 4: tcc_offset = (msr >> 24) & 0xF; fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C) (%d default - %d offset)\n", - cpu, msr, target_c_local - tcc_offset, target_c_local, tcc_offset); + cpu, msr, tcc_default - tcc_offset, tcc_default, tcc_offset); break; case 6: tcc_offset = (msr >> 24) & 0x3F; fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C) (%d default - %d offset)\n", - cpu, msr, target_c_local - tcc_offset, target_c_local, tcc_offset); + cpu, msr, tcc_default - tcc_offset, tcc_default, tcc_offset); break; default: fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C)\n", - cpu, msr, target_c_local); + cpu, msr, tcc_default); break; } } - if (!target_c_local) + if (!tcc_default) goto guess; - tcc_activation_temp = target_c_local; + tj_max = tcc_default; return 0; guess: - tcc_activation_temp = TJMAX_DEFAULT; + tj_max = TJMAX_DEFAULT; fprintf(outf, "cpu%d: Guessing tjMax %d C, Please use -T to specify\n", - cpu, tcc_activation_temp); + cpu, tj_max); return 0; } @@ -6421,7 +6421,7 @@ void cmdline(int argc, char **argv) summary_only++; break; case 'T': - tcc_activation_temp_override = atoi(optarg); + tj_max_override = atoi(optarg); break; case 'v': print_version(); -- cgit v1.2.3-59-g8ed1b From 1b439f01b67c77a374adbbd97ad0c745b7abb09b Mon Sep 17 00:00:00 2001 From: Len Brown Date: Tue, 4 May 2021 19:21:34 -0400 Subject: tools/power turbostat: formatting Spring is here... run a long overdue Lendent on turbostat.c no functional change Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 943 ++++++++++++++++------------------ 1 file changed, 433 insertions(+), 510 deletions(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 8f0a1d8a0366..13805e460a4d 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -41,8 +41,8 @@ char *proc_stat = "/proc/stat"; FILE *outf; int *fd_percpu; int *fd_instr_count_percpu; -struct timeval interval_tv = {5, 0}; -struct timespec interval_ts = {5, 0}; +struct timeval interval_tv = { 5, 0 }; +struct timespec interval_ts = { 5, 0 }; /* Save original CPU model */ unsigned int model_orig; @@ -84,7 +84,7 @@ unsigned int do_rapl; unsigned int do_dts; unsigned int do_ptm; unsigned int do_ipc; -unsigned long long gfx_cur_rc6_ms; +unsigned long long gfx_cur_rc6_ms; unsigned long long cpuidle_cur_cpu_lpi_us; unsigned long long cpuidle_cur_sys_lpi_us; unsigned int gfx_cur_mhz; @@ -104,12 +104,12 @@ unsigned int crystal_hz; unsigned long long tsc_hz; int base_cpu; double discover_bclk(unsigned int family, unsigned int model); -unsigned int has_hwp; /* IA32_PM_ENABLE, IA32_HWP_CAPABILITIES */ +unsigned int has_hwp; /* IA32_PM_ENABLE, IA32_HWP_CAPABILITIES */ /* IA32_HWP_REQUEST, IA32_HWP_STATUS */ -unsigned int has_hwp_notify; /* IA32_HWP_INTERRUPT */ +unsigned int has_hwp_notify; /* IA32_HWP_INTERRUPT */ unsigned int has_hwp_activity_window; /* IA32_HWP_REQUEST[bits 41:32] */ -unsigned int has_hwp_epp; /* IA32_HWP_REQUEST[bits 31:24] */ -unsigned int has_hwp_pkg; /* IA32_HWP_REQUEST_PKG */ +unsigned int has_hwp_epp; /* IA32_HWP_REQUEST[bits 31:24] */ +unsigned int has_hwp_pkg; /* IA32_HWP_REQUEST_PKG */ unsigned int has_misc_feature_control; unsigned int first_counter_read = 1; int ignore_stdin; @@ -185,7 +185,7 @@ struct thread_data { unsigned long long mperf; unsigned long long c1; unsigned long long instr_count; - unsigned long long irq_count; + unsigned long long irq_count; unsigned int smi_count; unsigned int cpu_id; unsigned int apic_id; @@ -253,12 +253,11 @@ struct pkg_data { ((node_no) * topo.cores_per_node) + \ (core_no)) - #define GET_PKG(pkg_base, pkg_no) (pkg_base + pkg_no) -enum counter_scope {SCOPE_CPU, SCOPE_CORE, SCOPE_PACKAGE}; -enum counter_type {COUNTER_ITEMS, COUNTER_CYCLES, COUNTER_SECONDS, COUNTER_USEC}; -enum counter_format {FORMAT_RAW, FORMAT_DELTA, FORMAT_PERCENT}; +enum counter_scope { SCOPE_CPU, SCOPE_CORE, SCOPE_PACKAGE }; +enum counter_type { COUNTER_ITEMS, COUNTER_CYCLES, COUNTER_SECONDS, COUNTER_USEC }; +enum counter_format { FORMAT_RAW, FORMAT_DELTA, FORMAT_PERCENT }; struct msr_counter { unsigned int msr_num; @@ -294,9 +293,9 @@ int get_msr_sum(int cpu, off_t offset, unsigned long long *msr); struct msr_sum_array { /* get_msr_sum() = sum + (get_msr() - last) */ struct { - /*The accumulated MSR value is updated by the timer*/ + /*The accumulated MSR value is updated by the timer */ unsigned long long sum; - /*The MSR footprint recorded in last timer*/ + /*The MSR footprint recorded in last timer */ unsigned long long last; } entries[IDX_COUNT]; }; @@ -385,6 +384,7 @@ int idx_valid(int idx) return 0; } } + struct sys_counters { unsigned int added_thread_counters; unsigned int added_core_counters; @@ -408,7 +408,7 @@ struct cpu_topology { int logical_node_id; /* 0-based count within the package */ int physical_core_id; int thread_id; - cpu_set_t *put_ids; /* Processing Unit/Thread IDs */ + cpu_set_t *put_ids; /* Processing Unit/Thread IDs */ } *cpus; struct topo_params { @@ -425,7 +425,7 @@ struct topo_params { struct timeval tv_even, tv_odd, tv_delta; -int *irq_column_2_cpu; /* /proc/interrupts column numbers */ +int *irq_column_2_cpu; /* /proc/interrupts column numbers */ int *irqs_per_cpu; /* indexed by cpu_num */ void setup_all_buffers(void); @@ -438,34 +438,31 @@ int cpu_is_not_present(int cpu) { return !CPU_ISSET_S(cpu, cpu_present_setsize, cpu_present_set); } + /* * run func(thread, core, package) in topology order * skip non-present cpus */ -int for_all_cpus(int (func)(struct thread_data *, struct core_data *, struct pkg_data *), - struct thread_data *thread_base, struct core_data *core_base, struct pkg_data *pkg_base) +int for_all_cpus(int (func) (struct thread_data *, struct core_data *, struct pkg_data *), + struct thread_data *thread_base, struct core_data *core_base, struct pkg_data *pkg_base) { int retval, pkg_no, core_no, thread_no, node_no; for (pkg_no = 0; pkg_no < topo.num_packages; ++pkg_no) { for (node_no = 0; node_no < topo.nodes_per_pkg; node_no++) { for (core_no = 0; core_no < topo.cores_per_node; ++core_no) { - for (thread_no = 0; thread_no < - topo.threads_per_core; ++thread_no) { + for (thread_no = 0; thread_no < topo.threads_per_core; ++thread_no) { struct thread_data *t; struct core_data *c; struct pkg_data *p; - t = GET_THREAD(thread_base, thread_no, - core_no, node_no, - pkg_no); + t = GET_THREAD(thread_base, thread_no, core_no, node_no, pkg_no); if (cpu_is_not_present(t->cpu_id)) continue; - c = GET_CORE(core_base, core_no, - node_no, pkg_no); + c = GET_CORE(core_base, core_no, node_no, pkg_no); p = GET_PKG(pkg_base, pkg_no); retval = func(t, c, p); @@ -487,6 +484,7 @@ int cpu_migrate(int cpu) else return 0; } + int get_msr_fd(int cpu) { char pathname[32]; @@ -524,7 +522,7 @@ static int perf_instr_count_open(int cpu_num) /* counter for cpu_num, including user + kernel and all processes */ fd = perf_event_open(&pea, -1, cpu_num, -1, 0); - if (fd == -1) + if (fd == -1) err(-1, "cpu%d: perf instruction counter\n", cpu_num); return fd; @@ -568,7 +566,7 @@ struct msr_counter bic[] = { { 0x0, "Bzy_MHz" }, { 0x0, "TSC_MHz" }, { 0x0, "IRQ" }, - { 0x0, "SMI", "", 32, 0, FORMAT_DELTA, NULL}, + { 0x0, "SMI", "", 32, 0, FORMAT_DELTA, NULL }, { 0x0, "sysfs" }, { 0x0, "CPU%c1" }, { 0x0, "CPU%c3" }, @@ -681,7 +679,6 @@ unsigned long long bic_present = BIC_USEC | BIC_TOD | BIC_sysfs | BIC_APIC | BIC #define BIC_NOT_PRESENT(COUNTER_BIT) (bic_present &= ~COUNTER_BIT) #define BIC_IS_ENABLED(COUNTER_BIT) (bic_enabled & COUNTER_BIT) - #define MAX_DEFERRED 16 char *deferred_skip_names[MAX_DEFERRED]; int deferred_skip_index; @@ -695,42 +692,40 @@ enum show_hide_mode { SHOW_LIST, HIDE_LIST } global_show_hide_mode = HIDE_LIST; void help(void) { fprintf(outf, - "Usage: turbostat [OPTIONS][(--interval seconds) | COMMAND ...]\n" - "\n" - "Turbostat forks the specified COMMAND and prints statistics\n" - "when COMMAND completes.\n" - "If no COMMAND is specified, turbostat wakes every 5-seconds\n" - "to print statistics, until interrupted.\n" - " -a, --add add a counter\n" - " eg. --add msr0x10,u64,cpu,delta,MY_TSC\n" - " -c, --cpu cpu-set limit output to summary plus cpu-set:\n" - " {core | package | j,k,l..m,n-p }\n" - " -d, --debug displays usec, Time_Of_Day_Seconds and more debugging\n" - " -D, --Dump displays the raw counter values\n" - " -e, --enable [all | column]\n" - " shows all or the specified disabled column\n" - " -H, --hide [column|column,column,...]\n" - " hide the specified column(s)\n" - " -i, --interval sec.subsec\n" - " Override default 5-second measurement interval\n" - " -J, --Joules displays energy in Joules instead of Watts\n" - " -l, --list list column headers only\n" - " -n, --num_iterations num\n" - " number of the measurement iterations\n" - " -o, --out file\n" - " create or truncate \"file\" for all output\n" - " -q, --quiet skip decoding system configuration header\n" - " -s, --show [column|column,column,...]\n" - " show only the specified column(s)\n" - " -S, --Summary\n" - " limits output to 1-line system summary per interval\n" - " -T, --TCC temperature\n" - " sets the Thermal Control Circuit temperature in\n" - " degrees Celsius\n" - " -h, --help print this help message\n" - " -v, --version print version information\n" - "\n" - "For more help, run \"man turbostat\"\n"); + "Usage: turbostat [OPTIONS][(--interval seconds) | COMMAND ...]\n" + "\n" + "Turbostat forks the specified COMMAND and prints statistics\n" + "when COMMAND completes.\n" + "If no COMMAND is specified, turbostat wakes every 5-seconds\n" + "to print statistics, until interrupted.\n" + " -a, --add add a counter\n" + " eg. --add msr0x10,u64,cpu,delta,MY_TSC\n" + " -c, --cpu cpu-set limit output to summary plus cpu-set:\n" + " {core | package | j,k,l..m,n-p }\n" + " -d, --debug displays usec, Time_Of_Day_Seconds and more debugging\n" + " -D, --Dump displays the raw counter values\n" + " -e, --enable [all | column]\n" + " shows all or the specified disabled column\n" + " -H, --hide [column|column,column,...]\n" + " hide the specified column(s)\n" + " -i, --interval sec.subsec\n" + " Override default 5-second measurement interval\n" + " -J, --Joules displays energy in Joules instead of Watts\n" + " -l, --list list column headers only\n" + " -n, --num_iterations num\n" + " number of the measurement iterations\n" + " -o, --out file\n" + " create or truncate \"file\" for all output\n" + " -q, --quiet skip decoding system configuration header\n" + " -s, --show [column|column,column,...]\n" + " show only the specified column(s)\n" + " -S, --Summary\n" + " limits output to 1-line system summary per interval\n" + " -T, --TCC temperature\n" + " sets the Thermal Control Circuit temperature in\n" + " degrees Celsius\n" + " -h, --help print this help message\n" + " -v, --version print version information\n" "\n" "For more help, run \"man turbostat\"\n"); } /* @@ -784,7 +779,6 @@ unsigned long long bic_lookup(char *name_list, enum show_hide_mode mode) return retval; } - void print_header(char *delim) { struct msr_counter *mp; @@ -966,8 +960,7 @@ void print_header(char *delim) outp += sprintf(outp, "\n"); } -int dump_counters(struct thread_data *t, struct core_data *c, - struct pkg_data *p) +int dump_counters(struct thread_data *t, struct core_data *c, struct pkg_data *p) { int i; struct msr_counter *mp; @@ -975,8 +968,7 @@ int dump_counters(struct thread_data *t, struct core_data *c, outp += sprintf(outp, "t %p, c %p, p %p\n", t, c, p); if (t) { - outp += sprintf(outp, "CPU: %d flags 0x%x\n", - t->cpu_id, t->flags); + outp += sprintf(outp, "CPU: %d flags 0x%x\n", t->cpu_id, t->flags); outp += sprintf(outp, "TSC: %016llX\n", t->tsc); outp += sprintf(outp, "aperf: %016llX\n", t->aperf); outp += sprintf(outp, "mperf: %016llX\n", t->mperf); @@ -991,8 +983,7 @@ int dump_counters(struct thread_data *t, struct core_data *c, outp += sprintf(outp, "SMI: %d\n", t->smi_count); for (i = 0, mp = sys.tp; mp; i++, mp = mp->next) { - outp += sprintf(outp, "tADDED [%d] msr0x%x: %08llX\n", - i, mp->msr_num, t->counter[i]); + outp += sprintf(outp, "tADDED [%d] msr0x%x: %08llX\n", i, mp->msr_num, t->counter[i]); } } @@ -1005,8 +996,7 @@ int dump_counters(struct thread_data *t, struct core_data *c, outp += sprintf(outp, "Joules: %0X\n", c->core_energy); for (i = 0, mp = sys.cp; mp; i++, mp = mp->next) { - outp += sprintf(outp, "cADDED [%d] msr0x%x: %08llX\n", - i, mp->msr_num, c->counter[i]); + outp += sprintf(outp, "cADDED [%d] msr0x%x: %08llX\n", i, mp->msr_num, c->counter[i]); } outp += sprintf(outp, "mc6_us: %016llX\n", c->mc6_us); } @@ -1035,15 +1025,12 @@ int dump_counters(struct thread_data *t, struct core_data *c, outp += sprintf(outp, "Joules COR: %0llX\n", p->energy_cores); outp += sprintf(outp, "Joules GFX: %0llX\n", p->energy_gfx); outp += sprintf(outp, "Joules RAM: %0llX\n", p->energy_dram); - outp += sprintf(outp, "Throttle PKG: %0llX\n", - p->rapl_pkg_perf_status); - outp += sprintf(outp, "Throttle RAM: %0llX\n", - p->rapl_dram_perf_status); + outp += sprintf(outp, "Throttle PKG: %0llX\n", p->rapl_pkg_perf_status); + outp += sprintf(outp, "Throttle RAM: %0llX\n", p->rapl_dram_perf_status); outp += sprintf(outp, "PTM: %dC\n", p->pkg_temp_c); for (i = 0, mp = sys.pp; mp; i++, mp = mp->next) { - outp += sprintf(outp, "pADDED [%d] msr0x%x: %08llX\n", - i, mp->msr_num, p->counter[i]); + outp += sprintf(outp, "pADDED [%d] msr0x%x: %08llX\n", i, mp->msr_num, p->counter[i]); } } @@ -1055,8 +1042,7 @@ int dump_counters(struct thread_data *t, struct core_data *c, /* * column formatting convention & formats */ -int format_counters(struct thread_data *t, struct core_data *c, - struct pkg_data *p) +int format_counters(struct thread_data *t, struct core_data *c, struct pkg_data *p) { double interval_float, tsc; char *fmt8; @@ -1065,17 +1051,16 @@ int format_counters(struct thread_data *t, struct core_data *c, char *delim = "\t"; int printed = 0; - /* if showing only 1st thread in core and this isn't one, bail out */ + /* if showing only 1st thread in core and this isn't one, bail out */ if (show_core_only && !(t->flags & CPU_IS_FIRST_THREAD_IN_CORE)) return 0; - /* if showing only 1st thread in pkg and this isn't one, bail out */ + /* if showing only 1st thread in pkg and this isn't one, bail out */ if (show_pkg_only && !(t->flags & CPU_IS_FIRST_CORE_IN_PACKAGE)) return 0; /*if not summary line and --cpu is used */ - if ((t != &average.threads) && - (cpu_subset && !CPU_ISSET_S(t->cpu_id, cpu_subset_size, cpu_subset))) + if ((t != &average.threads) && (cpu_subset && !CPU_ISSET_S(t->cpu_id, cpu_subset_size, cpu_subset))) return 0; if (DO_BIC(BIC_USEC)) { @@ -1090,7 +1075,7 @@ int format_counters(struct thread_data *t, struct core_data *c, if (DO_BIC(BIC_TOD)) outp += sprintf(outp, "%10ld.%06ld\t", t->tv_end.tv_sec, t->tv_end.tv_usec); - interval_float = t->tv_delta.tv_sec + t->tv_delta.tv_usec/1000000.0; + interval_float = t->tv_delta.tv_sec + t->tv_delta.tv_usec / 1000000.0; tsc = t->tsc * tsc_tweak; @@ -1126,11 +1111,9 @@ int format_counters(struct thread_data *t, struct core_data *c, if (DO_BIC(BIC_Node)) { if (t) outp += sprintf(outp, "%s%d", - (printed++ ? delim : ""), - cpus[t->cpu_id].physical_node_id); + (printed++ ? delim : ""), cpus[t->cpu_id].physical_node_id); else - outp += sprintf(outp, "%s-", - (printed++ ? delim : "")); + outp += sprintf(outp, "%s-", (printed++ ? delim : "")); } if (DO_BIC(BIC_Core)) { if (c) @@ -1147,22 +1130,22 @@ int format_counters(struct thread_data *t, struct core_data *c, } if (DO_BIC(BIC_Avg_MHz)) - outp += sprintf(outp, "%s%.0f", (printed++ ? delim : ""), - 1.0 / units * t->aperf / interval_float); + outp += sprintf(outp, "%s%.0f", (printed++ ? delim : ""), 1.0 / units * t->aperf / interval_float); if (DO_BIC(BIC_Busy)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * t->mperf/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * t->mperf / tsc); if (DO_BIC(BIC_Bzy_MHz)) { if (has_base_hz) - outp += sprintf(outp, "%s%.0f", (printed++ ? delim : ""), base_hz / units * t->aperf / t->mperf); + outp += + sprintf(outp, "%s%.0f", (printed++ ? delim : ""), base_hz / units * t->aperf / t->mperf); else outp += sprintf(outp, "%s%.0f", (printed++ ? delim : ""), - tsc / units * t->aperf / t->mperf / interval_float); + tsc / units * t->aperf / t->mperf / interval_float); } if (DO_BIC(BIC_TSC_MHz)) - outp += sprintf(outp, "%s%.0f", (printed++ ? delim : ""), 1.0 * t->tsc/units/interval_float); + outp += sprintf(outp, "%s%.0f", (printed++ ? delim : ""), 1.0 * t->tsc / units / interval_float); if (DO_BIC(BIC_IPC)) outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 1.0 * t->instr_count / t->aperf); @@ -1183,7 +1166,8 @@ int format_counters(struct thread_data *t, struct core_data *c, for (i = 0, mp = sys.tp; mp; i++, mp = mp->next) { if (mp->format == FORMAT_RAW) { if (mp->width == 32) - outp += sprintf(outp, "%s0x%08x", (printed++ ? delim : ""), (unsigned int) t->counter[i]); + outp += + sprintf(outp, "%s0x%08x", (printed++ ? delim : ""), (unsigned int)t->counter[i]); else outp += sprintf(outp, "%s0x%016llx", (printed++ ? delim : ""), t->counter[i]); } else if (mp->format == FORMAT_DELTA) { @@ -1193,27 +1177,28 @@ int format_counters(struct thread_data *t, struct core_data *c, outp += sprintf(outp, "%s%lld", (printed++ ? delim : ""), t->counter[i]); } else if (mp->format == FORMAT_PERCENT) { if (mp->type == COUNTER_USEC) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), t->counter[i]/interval_float/10000); + outp += + sprintf(outp, "%s%.2f", (printed++ ? delim : ""), + t->counter[i] / interval_float / 10000); else - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * t->counter[i]/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * t->counter[i] / tsc); } } /* C1 */ if (DO_BIC(BIC_CPU_c1)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * t->c1/tsc); - + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * t->c1 / tsc); /* print per-core data only for 1st thread in core */ if (!(t->flags & CPU_IS_FIRST_THREAD_IN_CORE)) goto done; if (DO_BIC(BIC_CPU_c3)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * c->c3/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * c->c3 / tsc); if (DO_BIC(BIC_CPU_c6)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * c->c6/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * c->c6 / tsc); if (DO_BIC(BIC_CPU_c7)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * c->c7/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * c->c7 / tsc); /* Mod%c6 */ if (DO_BIC(BIC_Mod_c6)) @@ -1225,7 +1210,8 @@ int format_counters(struct thread_data *t, struct core_data *c, for (i = 0, mp = sys.cp; mp; i++, mp = mp->next) { if (mp->format == FORMAT_RAW) { if (mp->width == 32) - outp += sprintf(outp, "%s0x%08x", (printed++ ? delim : ""), (unsigned int) c->counter[i]); + outp += + sprintf(outp, "%s0x%08x", (printed++ ? delim : ""), (unsigned int)c->counter[i]); else outp += sprintf(outp, "%s0x%016llx", (printed++ ? delim : ""), c->counter[i]); } else if (mp->format == FORMAT_DELTA) { @@ -1234,14 +1220,15 @@ int format_counters(struct thread_data *t, struct core_data *c, else outp += sprintf(outp, "%s%lld", (printed++ ? delim : ""), c->counter[i]); } else if (mp->format == FORMAT_PERCENT) { - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * c->counter[i]/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * c->counter[i] / tsc); } } fmt8 = "%s%.2f"; if (DO_BIC(BIC_CorWatt) && (do_rapl & RAPL_PER_CORE_ENERGY)) - outp += sprintf(outp, fmt8, (printed++ ? delim : ""), c->core_energy * rapl_energy_units / interval_float); + outp += + sprintf(outp, fmt8, (printed++ ? delim : ""), c->core_energy * rapl_energy_units / interval_float); if (DO_BIC(BIC_Cor_J) && (do_rapl & RAPL_PER_CORE_ENERGY)) outp += sprintf(outp, fmt8, (printed++ ? delim : ""), c->core_energy * rapl_energy_units); @@ -1259,7 +1246,7 @@ int format_counters(struct thread_data *t, struct core_data *c, outp += sprintf(outp, "%s**.**", (printed++ ? delim : "")); } else { outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), - p->gfx_rc6_ms / 10.0 / interval_float); + p->gfx_rc6_ms / 10.0 / interval_float); } } @@ -1273,42 +1260,49 @@ int format_counters(struct thread_data *t, struct core_data *c, /* Totl%C0, Any%C0 GFX%C0 CPUGFX% */ if (DO_BIC(BIC_Totl_c0)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pkg_wtd_core_c0/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pkg_wtd_core_c0 / tsc); if (DO_BIC(BIC_Any_c0)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pkg_any_core_c0/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pkg_any_core_c0 / tsc); if (DO_BIC(BIC_GFX_c0)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pkg_any_gfxe_c0/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pkg_any_gfxe_c0 / tsc); if (DO_BIC(BIC_CPUGFX)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pkg_both_core_gfxe_c0/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pkg_both_core_gfxe_c0 / tsc); if (DO_BIC(BIC_Pkgpc2)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc2/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc2 / tsc); if (DO_BIC(BIC_Pkgpc3)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc3/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc3 / tsc); if (DO_BIC(BIC_Pkgpc6)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc6/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc6 / tsc); if (DO_BIC(BIC_Pkgpc7)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc7/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc7 / tsc); if (DO_BIC(BIC_Pkgpc8)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc8/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc8 / tsc); if (DO_BIC(BIC_Pkgpc9)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc9/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc9 / tsc); if (DO_BIC(BIC_Pkgpc10)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc10/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc10 / tsc); if (DO_BIC(BIC_CPU_LPI)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->cpu_lpi / 1000000.0 / interval_float); + outp += + sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->cpu_lpi / 1000000.0 / interval_float); if (DO_BIC(BIC_SYS_LPI)) - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->sys_lpi / 1000000.0 / interval_float); + outp += + sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->sys_lpi / 1000000.0 / interval_float); if (DO_BIC(BIC_PkgWatt)) - outp += sprintf(outp, fmt8, (printed++ ? delim : ""), p->energy_pkg * rapl_energy_units / interval_float); + outp += + sprintf(outp, fmt8, (printed++ ? delim : ""), p->energy_pkg * rapl_energy_units / interval_float); if (DO_BIC(BIC_CorWatt) && !(do_rapl & RAPL_PER_CORE_ENERGY)) - outp += sprintf(outp, fmt8, (printed++ ? delim : ""), p->energy_cores * rapl_energy_units / interval_float); + outp += + sprintf(outp, fmt8, (printed++ ? delim : ""), p->energy_cores * rapl_energy_units / interval_float); if (DO_BIC(BIC_GFXWatt)) - outp += sprintf(outp, fmt8, (printed++ ? delim : ""), p->energy_gfx * rapl_energy_units / interval_float); + outp += + sprintf(outp, fmt8, (printed++ ? delim : ""), p->energy_gfx * rapl_energy_units / interval_float); if (DO_BIC(BIC_RAMWatt)) - outp += sprintf(outp, fmt8, (printed++ ? delim : ""), p->energy_dram * rapl_dram_energy_units / interval_float); + outp += + sprintf(outp, fmt8, (printed++ ? delim : ""), + p->energy_dram * rapl_dram_energy_units / interval_float); if (DO_BIC(BIC_Pkg_J)) outp += sprintf(outp, fmt8, (printed++ ? delim : ""), p->energy_pkg * rapl_energy_units); if (DO_BIC(BIC_Cor_J) && !(do_rapl & RAPL_PER_CORE_ENERGY)) @@ -1318,14 +1312,19 @@ int format_counters(struct thread_data *t, struct core_data *c, if (DO_BIC(BIC_RAM_J)) outp += sprintf(outp, fmt8, (printed++ ? delim : ""), p->energy_dram * rapl_dram_energy_units); if (DO_BIC(BIC_PKG__)) - outp += sprintf(outp, fmt8, (printed++ ? delim : ""), 100.0 * p->rapl_pkg_perf_status * rapl_time_units / interval_float); + outp += + sprintf(outp, fmt8, (printed++ ? delim : ""), + 100.0 * p->rapl_pkg_perf_status * rapl_time_units / interval_float); if (DO_BIC(BIC_RAM__)) - outp += sprintf(outp, fmt8, (printed++ ? delim : ""), 100.0 * p->rapl_dram_perf_status * rapl_time_units / interval_float); + outp += + sprintf(outp, fmt8, (printed++ ? delim : ""), + 100.0 * p->rapl_dram_perf_status * rapl_time_units / interval_float); for (i = 0, mp = sys.pp; mp; i++, mp = mp->next) { if (mp->format == FORMAT_RAW) { if (mp->width == 32) - outp += sprintf(outp, "%s0x%08x", (printed++ ? delim : ""), (unsigned int) p->counter[i]); + outp += + sprintf(outp, "%s0x%08x", (printed++ ? delim : ""), (unsigned int)p->counter[i]); else outp += sprintf(outp, "%s0x%016llx", (printed++ ? delim : ""), p->counter[i]); } else if (mp->format == FORMAT_DELTA) { @@ -1334,7 +1333,7 @@ int format_counters(struct thread_data *t, struct core_data *c, else outp += sprintf(outp, "%s%lld", (printed++ ? delim : ""), p->counter[i]); } else if (mp->format == FORMAT_PERCENT) { - outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->counter[i]/tsc); + outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->counter[i] / tsc); } } @@ -1359,12 +1358,14 @@ void flush_output_stdout(void) outp = output_buffer; } + void flush_output_stderr(void) { fputs(output_buffer, outf); fflush(outf); outp = output_buffer; } + void format_all_counters(struct thread_data *t, struct core_data *c, struct pkg_data *p) { static int printed; @@ -1385,13 +1386,11 @@ void format_all_counters(struct thread_data *t, struct core_data *c, struct pkg_ #define DELTA_WRAP32(new, old) \ old = ((((unsigned long long)new << 32) - ((unsigned long long)old << 32)) >> 32); -int -delta_package(struct pkg_data *new, struct pkg_data *old) +int delta_package(struct pkg_data *new, struct pkg_data *old) { int i; struct msr_counter *mp; - if (DO_BIC(BIC_Totl_c0)) old->pkg_wtd_core_c0 = new->pkg_wtd_core_c0 - old->pkg_wtd_core_c0; if (DO_BIC(BIC_Any_c0)) @@ -1416,7 +1415,7 @@ delta_package(struct pkg_data *new, struct pkg_data *old) old->pkg_temp_c = new->pkg_temp_c; /* flag an error when rc6 counter resets/wraps */ - if (old->gfx_rc6_ms > new->gfx_rc6_ms) + if (old->gfx_rc6_ms > new->gfx_rc6_ms) old->gfx_rc6_ms = -1; else old->gfx_rc6_ms = new->gfx_rc6_ms - old->gfx_rc6_ms; @@ -1441,8 +1440,7 @@ delta_package(struct pkg_data *new, struct pkg_data *old) return 0; } -void -delta_core(struct core_data *new, struct core_data *old) +void delta_core(struct core_data *new, struct core_data *old) { int i; struct msr_counter *mp; @@ -1474,9 +1472,7 @@ int soft_c1_residency_display(int bic) /* * old = new - old */ -int -delta_thread(struct thread_data *new, struct thread_data *old, - struct core_data *core_delta) +int delta_thread(struct thread_data *new, struct thread_data *old, struct core_data *core_delta) { int i; struct msr_counter *mp; @@ -1507,8 +1503,7 @@ delta_thread(struct thread_data *new, struct thread_data *old, old->c1 = new->c1 - old->c1; - if (DO_BIC(BIC_Avg_MHz) || DO_BIC(BIC_Busy) || DO_BIC(BIC_Bzy_MHz) || - soft_c1_residency_display(BIC_Avg_MHz)) { + if (DO_BIC(BIC_Avg_MHz) || DO_BIC(BIC_Busy) || DO_BIC(BIC_Bzy_MHz) || soft_c1_residency_display(BIC_Avg_MHz)) { if ((new->aperf > old->aperf) && (new->mperf > old->mperf)) { old->aperf = new->aperf - old->aperf; old->mperf = new->mperf - old->mperf; @@ -1517,7 +1512,6 @@ delta_thread(struct thread_data *new, struct thread_data *old, } } - if (use_c1_residency_msr) { /* * Some models have a dedicated C1 residency MSR, @@ -1534,7 +1528,7 @@ delta_thread(struct thread_data *new, struct thread_data *old, else { /* normal case, derive c1 */ old->c1 = (old->tsc * tsc_tweak) - old->mperf - core_delta->c3 - - core_delta->c6 - core_delta->c7; + - core_delta->c6 - core_delta->c7; } } @@ -1563,8 +1557,7 @@ delta_thread(struct thread_data *new, struct thread_data *old, } int delta_cpu(struct thread_data *t, struct core_data *c, - struct pkg_data *p, struct thread_data *t2, - struct core_data *c2, struct pkg_data *p2) + struct pkg_data *p, struct thread_data *t2, struct core_data *c2, struct pkg_data *p2) { int retval = 0; @@ -1587,7 +1580,7 @@ int delta_cpu(struct thread_data *t, struct core_data *c, void clear_counters(struct thread_data *t, struct core_data *c, struct pkg_data *p) { int i; - struct msr_counter *mp; + struct msr_counter *mp; t->tv_begin.tv_sec = 0; t->tv_begin.tv_usec = 0; @@ -1654,8 +1647,8 @@ void clear_counters(struct thread_data *t, struct core_data *c, struct pkg_data for (i = 0, mp = sys.pp; mp; i++, mp = mp->next) p->counter[i] = 0; } -int sum_counters(struct thread_data *t, struct core_data *c, - struct pkg_data *p) + +int sum_counters(struct thread_data *t, struct core_data *c, struct pkg_data *p) { int i; struct msr_counter *mp; @@ -1756,12 +1749,12 @@ int sum_counters(struct thread_data *t, struct core_data *c, } return 0; } + /* * sum the counters for all cpus in the system * compute the weighted average */ -void compute_average(struct thread_data *t, struct core_data *c, - struct pkg_data *p) +void compute_average(struct thread_data *t, struct core_data *c, struct pkg_data *p) { int i; struct msr_counter *mp; @@ -1842,7 +1835,7 @@ static unsigned long long rdtsc(void) { unsigned int low, high; - asm volatile("rdtsc" : "=a" (low), "=d" (high)); + asm volatile ("rdtsc":"=a" (low), "=d"(high)); return low | ((unsigned long long)high) << 32; } @@ -1858,6 +1851,7 @@ FILE *fopen_or_die(const char *path, const char *mode) err(1, "%s: open failed", path); return filep; } + /* * snapshot_sysfs_counter() * @@ -1889,8 +1883,7 @@ int get_mp(int cpu, struct msr_counter *mp, unsigned long long *counterp) char path[128 + PATH_BYTES]; if (mp->flags & SYSFS_PERCPU) { - sprintf(path, "/sys/devices/system/cpu/cpu%d/%s", - cpu, mp->path); + sprintf(path, "/sys/devices/system/cpu/cpu%d/%s", cpu, mp->path); *counterp = snapshot_sysfs_counter(path); } else { @@ -1950,7 +1943,7 @@ void get_apic_id(struct thread_data *t) eax = ebx = ecx = edx = 0; __cpuid(0x80000001, eax, ebx, ecx, edx); - topology_extensions = ecx & (1 << 22); + topology_extensions = ecx & (1 << 22); if (topology_extensions == 0) return; @@ -1973,8 +1966,7 @@ void get_apic_id(struct thread_data *t) t->x2apic_id = edx; if (debug && (t->apic_id != (t->x2apic_id & 0xff))) - fprintf(outf, "cpu%d: BIOS BUG: apic 0x%x x2apic 0x%x\n", - t->cpu_id, t->apic_id, t->x2apic_id); + fprintf(outf, "cpu%d: BIOS BUG: apic 0x%x x2apic 0x%x\n", t->cpu_id, t->apic_id, t->x2apic_id); } /* @@ -2002,8 +1994,7 @@ int get_counters(struct thread_data *t, struct core_data *c, struct pkg_data *p) retry: t->tsc = rdtsc(); /* we are running on local CPU of interest */ - if (DO_BIC(BIC_Avg_MHz) || DO_BIC(BIC_Busy) || DO_BIC(BIC_Bzy_MHz) || - soft_c1_residency_display(BIC_Avg_MHz)) { + if (DO_BIC(BIC_Avg_MHz) || DO_BIC(BIC_Busy) || DO_BIC(BIC_Bzy_MHz) || soft_c1_residency_display(BIC_Avg_MHz)) { unsigned long long tsc_before, tsc_between, tsc_after, aperf_time, mperf_time; /* @@ -2050,8 +2041,7 @@ retry: if (aperf_mperf_retry_count < 5) goto retry; else - warnx("cpu%d jitter %lld %lld", - cpu, aperf_time, mperf_time); + warnx("cpu%d jitter %lld %lld", cpu, aperf_time, mperf_time); } aperf_mperf_retry_count = 0; @@ -2252,47 +2242,64 @@ done: * (>= PCL__7) and to index pkg_cstate_limit_strings[]. */ -#define PCLUKN 0 /* Unknown */ -#define PCLRSV 1 /* Reserved */ -#define PCL__0 2 /* PC0 */ -#define PCL__1 3 /* PC1 */ -#define PCL__2 4 /* PC2 */ -#define PCL__3 5 /* PC3 */ -#define PCL__4 6 /* PC4 */ -#define PCL__6 7 /* PC6 */ -#define PCL_6N 8 /* PC6 No Retention */ -#define PCL_6R 9 /* PC6 Retention */ -#define PCL__7 10 /* PC7 */ -#define PCL_7S 11 /* PC7 Shrink */ -#define PCL__8 12 /* PC8 */ -#define PCL__9 13 /* PC9 */ -#define PCL_10 14 /* PC10 */ -#define PCLUNL 15 /* Unlimited */ +#define PCLUKN 0 /* Unknown */ +#define PCLRSV 1 /* Reserved */ +#define PCL__0 2 /* PC0 */ +#define PCL__1 3 /* PC1 */ +#define PCL__2 4 /* PC2 */ +#define PCL__3 5 /* PC3 */ +#define PCL__4 6 /* PC4 */ +#define PCL__6 7 /* PC6 */ +#define PCL_6N 8 /* PC6 No Retention */ +#define PCL_6R 9 /* PC6 Retention */ +#define PCL__7 10 /* PC7 */ +#define PCL_7S 11 /* PC7 Shrink */ +#define PCL__8 12 /* PC8 */ +#define PCL__9 13 /* PC9 */ +#define PCL_10 14 /* PC10 */ +#define PCLUNL 15 /* Unlimited */ int pkg_cstate_limit = PCLUKN; char *pkg_cstate_limit_strings[] = { "reserved", "unknown", "pc0", "pc1", "pc2", - "pc3", "pc4", "pc6", "pc6n", "pc6r", "pc7", "pc7s", "pc8", "pc9", "pc10", "unlimited"}; - -int nhm_pkg_cstate_limits[16] = {PCL__0, PCL__1, PCL__3, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV}; -int snb_pkg_cstate_limits[16] = {PCL__0, PCL__2, PCL_6N, PCL_6R, PCL__7, PCL_7S, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV}; -int hsw_pkg_cstate_limits[16] = {PCL__0, PCL__2, PCL__3, PCL__6, PCL__7, PCL_7S, PCL__8, PCL__9, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV}; -int slv_pkg_cstate_limits[16] = {PCL__0, PCL__1, PCLRSV, PCLRSV, PCL__4, PCLRSV, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCL__6, PCL__7}; -int amt_pkg_cstate_limits[16] = {PCLUNL, PCL__1, PCL__2, PCLRSV, PCLRSV, PCLRSV, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV}; -int phi_pkg_cstate_limits[16] = {PCL__0, PCL__2, PCL_6N, PCL_6R, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV}; -int glm_pkg_cstate_limits[16] = {PCLUNL, PCL__1, PCL__3, PCL__6, PCL__7, PCL_7S, PCL__8, PCL__9, PCL_10, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV}; -int skx_pkg_cstate_limits[16] = {PCL__0, PCL__2, PCL_6N, PCL_6R, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV}; -int icx_pkg_cstate_limits[16] = {PCL__0, PCL__2, PCL__6, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV}; + "pc3", "pc4", "pc6", "pc6n", "pc6r", "pc7", "pc7s", "pc8", "pc9", "pc10", "unlimited" +}; -static void -calculate_tsc_tweak() +int nhm_pkg_cstate_limits[16] = + { PCL__0, PCL__1, PCL__3, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, +PCLRSV, PCLRSV }; +int snb_pkg_cstate_limits[16] = + { PCL__0, PCL__2, PCL_6N, PCL_6R, PCL__7, PCL_7S, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, +PCLRSV, PCLRSV }; +int hsw_pkg_cstate_limits[16] = + { PCL__0, PCL__2, PCL__3, PCL__6, PCL__7, PCL_7S, PCL__8, PCL__9, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, +PCLRSV, PCLRSV }; +int slv_pkg_cstate_limits[16] = + { PCL__0, PCL__1, PCLRSV, PCLRSV, PCL__4, PCLRSV, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, +PCL__6, PCL__7 }; +int amt_pkg_cstate_limits[16] = + { PCLUNL, PCL__1, PCL__2, PCLRSV, PCLRSV, PCLRSV, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, +PCLRSV, PCLRSV }; +int phi_pkg_cstate_limits[16] = + { PCL__0, PCL__2, PCL_6N, PCL_6R, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, +PCLRSV, PCLRSV }; +int glm_pkg_cstate_limits[16] = + { PCLUNL, PCL__1, PCL__3, PCL__6, PCL__7, PCL_7S, PCL__8, PCL__9, PCL_10, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, +PCLRSV, PCLRSV }; +int skx_pkg_cstate_limits[16] = + { PCL__0, PCL__2, PCL_6N, PCL_6R, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, +PCLRSV, PCLRSV }; +int icx_pkg_cstate_limits[16] = + { PCL__0, PCL__2, PCL__6, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, +PCLRSV, PCLRSV }; + +static void calculate_tsc_tweak() { tsc_tweak = base_hz / tsc_hz; } void prewake_cstate_probe(unsigned int family, unsigned int model); -static void -dump_nhm_platform_info(void) +static void dump_nhm_platform_info(void) { unsigned long long msr; unsigned int ratio; @@ -2302,12 +2309,10 @@ dump_nhm_platform_info(void) fprintf(outf, "cpu%d: MSR_PLATFORM_INFO: 0x%08llx\n", base_cpu, msr); ratio = (msr >> 40) & 0xFF; - fprintf(outf, "%d * %.1f = %.1f MHz max efficiency frequency\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max efficiency frequency\n", ratio, bclk, ratio * bclk); ratio = (msr >> 8) & 0xFF; - fprintf(outf, "%d * %.1f = %.1f MHz base frequency\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz base frequency\n", ratio, bclk, ratio * bclk); get_msr(base_cpu, MSR_IA32_POWER_CTL, &msr); fprintf(outf, "cpu%d: MSR_IA32_POWER_CTL: 0x%08llx (C1E auto-promotion: %sabled)\n", @@ -2315,14 +2320,12 @@ dump_nhm_platform_info(void) /* C-state Pre-wake Disable (CSTATE_PREWAKE_DISABLE) */ if (dis_cstate_prewake) - fprintf(outf, "C-state Pre-wake: %sabled\n", - msr & 0x40000000 ? "DIS" : "EN"); + fprintf(outf, "C-state Pre-wake: %sabled\n", msr & 0x40000000 ? "DIS" : "EN"); return; } -static void -dump_hsw_turbo_ratio_limits(void) +static void dump_hsw_turbo_ratio_limits(void) { unsigned long long msr; unsigned int ratio; @@ -2333,18 +2336,15 @@ dump_hsw_turbo_ratio_limits(void) ratio = (msr >> 8) & 0xFF; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 18 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 18 active cores\n", ratio, bclk, ratio * bclk); ratio = (msr >> 0) & 0xFF; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 17 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 17 active cores\n", ratio, bclk, ratio * bclk); return; } -static void -dump_ivt_turbo_ratio_limits(void) +static void dump_ivt_turbo_ratio_limits(void) { unsigned long long msr; unsigned int ratio; @@ -2355,45 +2355,38 @@ dump_ivt_turbo_ratio_limits(void) ratio = (msr >> 56) & 0xFF; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 16 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 16 active cores\n", ratio, bclk, ratio * bclk); ratio = (msr >> 48) & 0xFF; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 15 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 15 active cores\n", ratio, bclk, ratio * bclk); ratio = (msr >> 40) & 0xFF; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 14 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 14 active cores\n", ratio, bclk, ratio * bclk); ratio = (msr >> 32) & 0xFF; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 13 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 13 active cores\n", ratio, bclk, ratio * bclk); ratio = (msr >> 24) & 0xFF; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 12 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 12 active cores\n", ratio, bclk, ratio * bclk); ratio = (msr >> 16) & 0xFF; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 11 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 11 active cores\n", ratio, bclk, ratio * bclk); ratio = (msr >> 8) & 0xFF; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 10 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 10 active cores\n", ratio, bclk, ratio * bclk); ratio = (msr >> 0) & 0xFF; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 9 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 9 active cores\n", ratio, bclk, ratio * bclk); return; } + int has_turbo_ratio_group_limits(int family, int model) { @@ -2411,8 +2404,7 @@ int has_turbo_ratio_group_limits(int family, int model) return 0; } -static void -dump_turbo_ratio_limits(int family, int model) +static void dump_turbo_ratio_limits(int family, int model) { unsigned long long msr, core_counts; unsigned int ratio, group_size; @@ -2477,8 +2469,7 @@ dump_turbo_ratio_limits(int family, int model) return; } -static void -dump_atom_turbo_ratio_limits(void) +static void dump_atom_turbo_ratio_limits(void) { unsigned long long msr; unsigned int ratio; @@ -2488,45 +2479,37 @@ dump_atom_turbo_ratio_limits(void) ratio = (msr >> 0) & 0x3F; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz minimum operating frequency\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz minimum operating frequency\n", ratio, bclk, ratio * bclk); ratio = (msr >> 8) & 0x3F; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz low frequency mode (LFM)\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz low frequency mode (LFM)\n", ratio, bclk, ratio * bclk); ratio = (msr >> 16) & 0x3F; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz base frequency\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz base frequency\n", ratio, bclk, ratio * bclk); get_msr(base_cpu, MSR_ATOM_CORE_TURBO_RATIOS, &msr); fprintf(outf, "cpu%d: MSR_ATOM_CORE_TURBO_RATIOS: 0x%08llx\n", base_cpu, msr & 0xFFFFFFFF); ratio = (msr >> 24) & 0x3F; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 4 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 4 active cores\n", ratio, bclk, ratio * bclk); ratio = (msr >> 16) & 0x3F; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 3 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 3 active cores\n", ratio, bclk, ratio * bclk); ratio = (msr >> 8) & 0x3F; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 2 active cores\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 2 active cores\n", ratio, bclk, ratio * bclk); ratio = (msr >> 0) & 0x3F; if (ratio) - fprintf(outf, "%d * %.1f = %.1f MHz max turbo 1 active core\n", - ratio, bclk, ratio * bclk); + fprintf(outf, "%d * %.1f = %.1f MHz max turbo 1 active core\n", ratio, bclk, ratio * bclk); } -static void -dump_knl_turbo_ratio_limits(void) +static void dump_knl_turbo_ratio_limits(void) { const unsigned int buckets_no = 7; @@ -2538,8 +2521,7 @@ dump_knl_turbo_ratio_limits(void) get_msr(base_cpu, MSR_TURBO_RATIO_LIMIT, &msr); - fprintf(outf, "cpu%d: MSR_TURBO_RATIO_LIMIT: 0x%08llx\n", - base_cpu, msr); + fprintf(outf, "cpu%d: MSR_TURBO_RATIO_LIMIT: 0x%08llx\n", base_cpu, msr); /* * Turbo encoding in KNL is as follows: @@ -2584,8 +2566,7 @@ dump_knl_turbo_ratio_limits(void) ratio[i], bclk, ratio[i] * bclk, cores[i]); } -static void -dump_nhm_cst_cfg(void) +static void dump_nhm_cst_cfg(void) { unsigned long long msr; @@ -2598,14 +2579,11 @@ dump_nhm_cst_cfg(void) (msr & SNB_C1_AUTO_UNDEMOTE) ? "UNdemote-C1, " : "", (msr & NHM_C3_AUTO_DEMOTE) ? "demote-C3, " : "", (msr & NHM_C1_AUTO_DEMOTE) ? "demote-C1, " : "", - (msr & (1 << 15)) ? "" : "UN", - (unsigned int)msr & 0xF, - pkg_cstate_limit_strings[pkg_cstate_limit]); + (msr & (1 << 15)) ? "" : "UN", (unsigned int)msr & 0xF, pkg_cstate_limit_strings[pkg_cstate_limit]); #define AUTOMATIC_CSTATE_CONVERSION (1UL << 16) if (has_automatic_cstate_conversion) { - fprintf(outf, ", automatic c-state conversion=%s", - (msr & AUTOMATIC_CSTATE_CONVERSION) ? "on" : "off"); + fprintf(outf, ", automatic c-state conversion=%s", (msr & AUTOMATIC_CSTATE_CONVERSION) ? "on" : "off"); } fprintf(outf, ")\n"); @@ -2613,8 +2591,7 @@ dump_nhm_cst_cfg(void) return; } -static void -dump_config_tdp(void) +static void dump_config_tdp(void) { unsigned long long msr; @@ -2656,7 +2633,7 @@ dump_config_tdp(void) fprintf(outf, ")\n"); } -unsigned int irtl_time_units[] = {1, 32, 1024, 32768, 1048576, 33554432, 0, 0 }; +unsigned int irtl_time_units[] = { 1, 32, 1024, 32768, 1048576, 33554432, 0, 0 }; void print_irtl(void) { @@ -2696,6 +2673,7 @@ void print_irtl(void) (msr & 0x3FF) * irtl_time_units[(msr >> 10) & 0x3]); } + void free_fd_percpu(void) { int i; @@ -2752,7 +2730,6 @@ void free_all_buffers(void) free(cpus); } - /* * Parse a file containing a single int. * Return 0 if file can not be opened @@ -2827,8 +2804,7 @@ void set_node_data(void) * the logical_node_id */ for (cpux = cpu; cpux <= topo.max_cpu_num; cpux++) { - if ((cpus[cpux].physical_package_id == pkg) && - (cpus[cpux].physical_node_id == node)) { + if ((cpus[cpux].physical_package_id == pkg) && (cpus[cpux].physical_node_id == node)) { cpus[cpux].logical_node_id = lnode; cpu_count++; } @@ -2850,8 +2826,7 @@ int get_physical_node_id(struct cpu_topology *thiscpu) int cpu = thiscpu->logical_cpu_id; for (i = 0; i <= topo.max_cpu_num; i++) { - sprintf(path, "/sys/devices/system/cpu/cpu%d/node%i/cpulist", - cpu, i); + sprintf(path, "/sys/devices/system/cpu/cpu%d/node%i/cpulist", cpu, i); filep = fopen(path, "r"); if (!filep) continue; @@ -2881,8 +2856,7 @@ int get_thread_siblings(struct cpu_topology *thiscpu) size = CPU_ALLOC_SIZE((topo.max_cpu_num + 1)); CPU_ZERO_S(size, thiscpu->put_ids); - sprintf(path, - "/sys/devices/system/cpu/cpu%d/topology/thread_siblings", cpu); + sprintf(path, "/sys/devices/system/cpu/cpu%d/topology/thread_siblings", cpu); filep = fopen(path, "r"); if (!filep) { @@ -2899,10 +2873,8 @@ int get_thread_siblings(struct cpu_topology *thiscpu) sib_core = get_core_id(so); if (sib_core == thiscpu->physical_core_id) { CPU_SET_S(so, size, thiscpu->put_ids); - if ((so != cpu) && - (cpus[so].thread_id < 0)) - cpus[so].thread_id = - thread_id++; + if ((so != cpu) && (cpus[so].thread_id < 0)) + cpus[so].thread_id = thread_id++; } } } @@ -2917,41 +2889,31 @@ int get_thread_siblings(struct cpu_topology *thiscpu) * skip non-present cpus */ -int for_all_cpus_2(int (func)(struct thread_data *, struct core_data *, - struct pkg_data *, struct thread_data *, struct core_data *, - struct pkg_data *), struct thread_data *thread_base, - struct core_data *core_base, struct pkg_data *pkg_base, - struct thread_data *thread_base2, struct core_data *core_base2, - struct pkg_data *pkg_base2) +int for_all_cpus_2(int (func) (struct thread_data *, struct core_data *, + struct pkg_data *, struct thread_data *, struct core_data *, + struct pkg_data *), struct thread_data *thread_base, + struct core_data *core_base, struct pkg_data *pkg_base, + struct thread_data *thread_base2, struct core_data *core_base2, struct pkg_data *pkg_base2) { int retval, pkg_no, node_no, core_no, thread_no; for (pkg_no = 0; pkg_no < topo.num_packages; ++pkg_no) { for (node_no = 0; node_no < topo.nodes_per_pkg; ++node_no) { - for (core_no = 0; core_no < topo.cores_per_node; - ++core_no) { - for (thread_no = 0; thread_no < - topo.threads_per_core; ++thread_no) { + for (core_no = 0; core_no < topo.cores_per_node; ++core_no) { + for (thread_no = 0; thread_no < topo.threads_per_core; ++thread_no) { struct thread_data *t, *t2; struct core_data *c, *c2; struct pkg_data *p, *p2; - t = GET_THREAD(thread_base, thread_no, - core_no, node_no, - pkg_no); + t = GET_THREAD(thread_base, thread_no, core_no, node_no, pkg_no); if (cpu_is_not_present(t->cpu_id)) continue; - t2 = GET_THREAD(thread_base2, thread_no, - core_no, node_no, - pkg_no); + t2 = GET_THREAD(thread_base2, thread_no, core_no, node_no, pkg_no); - c = GET_CORE(core_base, core_no, - node_no, pkg_no); - c2 = GET_CORE(core_base2, core_no, - node_no, - pkg_no); + c = GET_CORE(core_base, core_no, node_no, pkg_no); + c2 = GET_CORE(core_base2, core_no, node_no, pkg_no); p = GET_PKG(pkg_base, pkg_no); p2 = GET_PKG(pkg_base2, pkg_no); @@ -2970,7 +2932,7 @@ int for_all_cpus_2(int (func)(struct thread_data *, struct core_data *, * run func(cpu) on every cpu in /proc/stat * return max_cpu number */ -int for_all_proc_cpus(int (func)(int)) +int for_all_proc_cpus(int (func) (int)) { FILE *fp; int cpu_num; @@ -2990,7 +2952,7 @@ int for_all_proc_cpus(int (func)(int)) retval = func(cpu_num); if (retval) { fclose(fp); - return(retval); + return (retval); } } fclose(fp); @@ -3014,16 +2976,14 @@ void set_max_cpu_num(void) base_cpu = sched_getcpu(); if (base_cpu < 0) err(1, "cannot find calling cpu ID"); - sprintf(pathname, - "/sys/devices/system/cpu/cpu%d/topology/thread_siblings", - base_cpu); + sprintf(pathname, "/sys/devices/system/cpu/cpu%d/topology/thread_siblings", base_cpu); filep = fopen_or_die(pathname, "r"); topo.max_cpu_num = 0; while (fscanf(filep, "%lx,", &dummy) == 1) topo.max_cpu_num += BITMASK_SIZE; fclose(filep); - topo.max_cpu_num--; /* 0 based */ + topo.max_cpu_num--; /* 0 based */ } /* @@ -3035,6 +2995,7 @@ int count_cpus(int cpu) topo.num_cpus++; return 0; } + int mark_cpu_present(int cpu) { CPU_SET_S(cpu, cpu_present_setsize, cpu_present_set); @@ -3104,12 +3065,12 @@ int snapshot_proc_interrupts(void) } - while (getc(fp) != '\n') - ; /* flush interrupt description */ + while (getc(fp) != '\n') ; /* flush interrupt description */ } return 0; } + /* * snapshot_gfx_rc6_ms() * @@ -3133,6 +3094,7 @@ int snapshot_gfx_rc6_ms(void) return 0; } + /* * snapshot_gfx_mhz() * @@ -3212,6 +3174,7 @@ int snapshot_cpu_lpi_us(void) return 0; } + /* * snapshot_sys_lpi() * @@ -3235,6 +3198,7 @@ int snapshot_sys_lpi_us(void) return 0; } + /* * snapshot /proc and /sys files * @@ -3266,7 +3230,7 @@ int snapshot_proc_sysfs_files(void) int exit_requested; -static void signal_handler (int signal) +static void signal_handler(int signal) { switch (signal) { case SIGINT: @@ -3373,8 +3337,7 @@ static int update_msr_sum(struct thread_data *t, struct core_data *c, struct pkg continue; ret = get_msr(cpu, offset, &msr_cur); if (ret) { - fprintf(outf, "Can not update msr(0x%llx)\n", - (unsigned long long)offset); + fprintf(outf, "Can not update msr(0x%llx)\n", (unsigned long long)offset); continue; } @@ -3387,8 +3350,7 @@ static int update_msr_sum(struct thread_data *t, struct core_data *c, struct pkg return 0; } -static void -msr_record_handler(union sigval v) +static void msr_record_handler(union sigval v) { for_all_cpus(update_msr_sum, EVEN_COUNTERS); } @@ -3433,9 +3395,9 @@ void msr_sum_record(void) } return; - release_timer: +release_timer: timer_delete(timerid); - release_msr: +release_msr: free(per_cpu_msr_sum); } @@ -3527,7 +3489,7 @@ void check_dev_msr() sprintf(pathname, "/dev/cpu/%d/msr", base_cpu); if (stat(pathname, &sb)) - if (system("/sbin/modprobe msr > /dev/null 2>&1")) + if (system("/sbin/modprobe msr > /dev/null 2>&1")) err(-5, "no /dev/cpu/0/msr, Try \"# modprobe msr\" "); } @@ -3549,8 +3511,7 @@ int check_for_cap_sys_rawio(void) err(-6, "cap_get\n"); if (cap_flag_value != CAP_SET) { - warnx("capget(CAP_SYS_RAWIO) failed," - " try \"# setcap cap_sys_rawio=ep %s\"", progname); + warnx("capget(CAP_SYS_RAWIO) failed," " try \"# setcap cap_sys_rawio=ep %s\"", progname); return 1; } @@ -3559,6 +3520,7 @@ int check_for_cap_sys_rawio(void) return 0; } + void check_permissions(void) { int do_exit = 0; @@ -3664,7 +3626,7 @@ int probe_nhm_msrs(unsigned int family, unsigned int model) case INTEL_FAM6_ATOM_GOLDMONT_PLUS: case INTEL_FAM6_ATOM_GOLDMONT_D: /* DNV */ case INTEL_FAM6_ATOM_TREMONT: /* EHL */ - case INTEL_FAM6_ATOM_TREMONT_D: /* JVL */ + case INTEL_FAM6_ATOM_TREMONT_D: /* JVL */ pkg_cstate_limits = glm_pkg_cstate_limits; break; default: @@ -3680,6 +3642,7 @@ int probe_nhm_msrs(unsigned int family, unsigned int model) has_base_hz = 1; return 1; } + /* * SLV client has support for unique MSRs: * @@ -3700,6 +3663,7 @@ int has_slv_msrs(unsigned int family, unsigned int model) } return 0; } + int is_dnv(unsigned int family, unsigned int model) { @@ -3712,6 +3676,7 @@ int is_dnv(unsigned int family, unsigned int model) } return 0; } + int is_bdx(unsigned int family, unsigned int model) { @@ -3724,6 +3689,7 @@ int is_bdx(unsigned int family, unsigned int model) } return 0; } + int is_skx(unsigned int family, unsigned int model) { @@ -3761,6 +3727,7 @@ int is_ehl(unsigned int family, unsigned int model) } return 0; } + int is_jvl(unsigned int family, unsigned int model) { if (!genuine_intel) @@ -3779,7 +3746,7 @@ int has_turbo_ratio_limit(unsigned int family, unsigned int model) return 0; switch (model) { - /* Nehalem compatible, but do not include turbo-ratio limit support */ + /* Nehalem compatible, but do not include turbo-ratio limit support */ case INTEL_FAM6_NEHALEM_EX: /* Nehalem-EX Xeon - Beckton */ case INTEL_FAM6_XEON_PHI_KNL: /* PHI - Knights Landing (different MSR definition) */ return 0; @@ -3787,6 +3754,7 @@ int has_turbo_ratio_limit(unsigned int family, unsigned int model) return 1; } } + int has_atom_turbo_ratio_limit(unsigned int family, unsigned int model) { if (has_slv_msrs(family, model)) @@ -3794,6 +3762,7 @@ int has_atom_turbo_ratio_limit(unsigned int family, unsigned int model) return 0; } + int has_ivt_turbo_ratio_limit(unsigned int family, unsigned int model) { if (!genuine_intel) @@ -3810,6 +3779,7 @@ int has_ivt_turbo_ratio_limit(unsigned int family, unsigned int model) return 0; } } + int has_hsw_turbo_ratio_limit(unsigned int family, unsigned int model) { if (!genuine_intel) @@ -3841,6 +3811,7 @@ int has_knl_turbo_ratio_limit(unsigned int family, unsigned int model) return 0; } } + int has_glm_turbo_ratio_limit(unsigned int family, unsigned int model) { if (!genuine_intel) @@ -3858,6 +3829,7 @@ int has_glm_turbo_ratio_limit(unsigned int family, unsigned int model) return 0; } } + int has_config_tdp(unsigned int family, unsigned int model) { if (!genuine_intel) @@ -3921,8 +3893,7 @@ void check_tcc_offset(int model) } } -static void -remove_underbar(char *s) +static void remove_underbar(char *s) { char *to = s; @@ -3935,8 +3906,7 @@ remove_underbar(char *s) *to = 0; } -static void -dump_cstate_pstate_config_info(unsigned int family, unsigned int model) +static void dump_cstate_pstate_config_info(unsigned int family, unsigned int model) { if (!do_nhm_platform_info) return; @@ -3981,8 +3951,8 @@ static void dump_sysfs_file(char *path) fprintf(outf, "%s: %s", strrchr(path, '/') + 1, cpuidle_buf); } -static void -dump_sysfs_cstate_config(void) + +static void dump_sysfs_cstate_config(void) { char path[64]; char name_buf[16]; @@ -4002,15 +3972,14 @@ dump_sysfs_cstate_config(void) for (state = 0; state < 10; ++state) { - sprintf(path, "/sys/devices/system/cpu/cpu%d/cpuidle/state%d/name", - base_cpu, state); + sprintf(path, "/sys/devices/system/cpu/cpu%d/cpuidle/state%d/name", base_cpu, state); input = fopen(path, "r"); if (input == NULL) continue; if (!fgets(name_buf, sizeof(name_buf), input)) err(1, "%s: failed to read file", path); - /* truncate "C1-HSW\n" to "C1", or truncate "C1\n" to "C1" */ + /* truncate "C1-HSW\n" to "C1", or truncate "C1\n" to "C1" */ sp = strchr(name_buf, '-'); if (!sp) sp = strchrnul(name_buf, '\n'); @@ -4019,8 +3988,7 @@ dump_sysfs_cstate_config(void) remove_underbar(name_buf); - sprintf(path, "/sys/devices/system/cpu/cpu%d/cpuidle/state%d/desc", - base_cpu, state); + sprintf(path, "/sys/devices/system/cpu/cpu%d/cpuidle/state%d/desc", base_cpu, state); input = fopen(path, "r"); if (input == NULL) continue; @@ -4031,8 +3999,8 @@ dump_sysfs_cstate_config(void) fclose(input); } } -static void -dump_sysfs_pstate_config(void) + +static void dump_sysfs_pstate_config(void) { char path[64]; char driver_buf[64]; @@ -4040,8 +4008,7 @@ dump_sysfs_pstate_config(void) FILE *input; int turbo; - sprintf(path, "/sys/devices/system/cpu/cpu%d/cpufreq/scaling_driver", - base_cpu); + sprintf(path, "/sys/devices/system/cpu/cpu%d/cpufreq/scaling_driver", base_cpu); input = fopen(path, "r"); if (input == NULL) { fprintf(outf, "NSFOD %s\n", path); @@ -4051,8 +4018,7 @@ dump_sysfs_pstate_config(void) err(1, "%s: failed to read file", path); fclose(input); - sprintf(path, "/sys/devices/system/cpu/cpu%d/cpufreq/scaling_governor", - base_cpu); + sprintf(path, "/sys/devices/system/cpu/cpu%d/cpufreq/scaling_governor", base_cpu); input = fopen(path, "r"); if (input == NULL) { fprintf(outf, "NSFOD %s\n", path); @@ -4084,7 +4050,6 @@ dump_sysfs_pstate_config(void) } } - /* * print_epb() * Decode the ENERGY_PERF_BIAS MSR @@ -4130,6 +4095,7 @@ int print_epb(struct thread_data *t, struct core_data *c, struct pkg_data *p) return 0; } + /* * print_hwp() * Decode the MSR_HWP_CAPABILITIES @@ -4156,8 +4122,7 @@ int print_hwp(struct thread_data *t, struct core_data *c, struct pkg_data *p) if (get_msr(cpu, MSR_PM_ENABLE, &msr)) return 0; - fprintf(outf, "cpu%d: MSR_PM_ENABLE: 0x%08llx (%sHWP)\n", - cpu, msr, (msr & (1 << 0)) ? "" : "No-"); + fprintf(outf, "cpu%d: MSR_PM_ENABLE: 0x%08llx (%sHWP)\n", cpu, msr, (msr & (1 << 0)) ? "" : "No-"); /* MSR_PM_ENABLE[1] == 1 if HWP is enabled and MSRs visible */ if ((msr & (1 << 0)) == 0) @@ -4167,25 +4132,23 @@ int print_hwp(struct thread_data *t, struct core_data *c, struct pkg_data *p) return 0; fprintf(outf, "cpu%d: MSR_HWP_CAPABILITIES: 0x%08llx " - "(high %d guar %d eff %d low %d)\n", - cpu, msr, - (unsigned int)HWP_HIGHEST_PERF(msr), - (unsigned int)HWP_GUARANTEED_PERF(msr), - (unsigned int)HWP_MOSTEFFICIENT_PERF(msr), - (unsigned int)HWP_LOWEST_PERF(msr)); + "(high %d guar %d eff %d low %d)\n", + cpu, msr, + (unsigned int)HWP_HIGHEST_PERF(msr), + (unsigned int)HWP_GUARANTEED_PERF(msr), + (unsigned int)HWP_MOSTEFFICIENT_PERF(msr), (unsigned int)HWP_LOWEST_PERF(msr)); if (get_msr(cpu, MSR_HWP_REQUEST, &msr)) return 0; fprintf(outf, "cpu%d: MSR_HWP_REQUEST: 0x%08llx " - "(min %d max %d des %d epp 0x%x window 0x%x pkg 0x%x)\n", - cpu, msr, - (unsigned int)(((msr) >> 0) & 0xff), - (unsigned int)(((msr) >> 8) & 0xff), - (unsigned int)(((msr) >> 16) & 0xff), - (unsigned int)(((msr) >> 24) & 0xff), - (unsigned int)(((msr) >> 32) & 0xff3), - (unsigned int)(((msr) >> 42) & 0x1)); + "(min %d max %d des %d epp 0x%x window 0x%x pkg 0x%x)\n", + cpu, msr, + (unsigned int)(((msr) >> 0) & 0xff), + (unsigned int)(((msr) >> 8) & 0xff), + (unsigned int)(((msr) >> 16) & 0xff), + (unsigned int)(((msr) >> 24) & 0xff), + (unsigned int)(((msr) >> 32) & 0xff3), (unsigned int)(((msr) >> 42) & 0x1)); if (has_hwp_pkg) { if (get_msr(cpu, MSR_HWP_REQUEST_PKG, &msr)) @@ -4197,8 +4160,7 @@ int print_hwp(struct thread_data *t, struct core_data *c, struct pkg_data *p) (unsigned int)(((msr) >> 0) & 0xff), (unsigned int)(((msr) >> 8) & 0xff), (unsigned int)(((msr) >> 16) & 0xff), - (unsigned int)(((msr) >> 24) & 0xff), - (unsigned int)(((msr) >> 32) & 0xff3)); + (unsigned int)(((msr) >> 24) & 0xff), (unsigned int)(((msr) >> 32) & 0xff3)); } if (has_hwp_notify) { if (get_msr(cpu, MSR_HWP_INTERRUPT, &msr)) @@ -4206,18 +4168,14 @@ int print_hwp(struct thread_data *t, struct core_data *c, struct pkg_data *p) fprintf(outf, "cpu%d: MSR_HWP_INTERRUPT: 0x%08llx " "(%s_Guaranteed_Perf_Change, %s_Excursion_Min)\n", - cpu, msr, - ((msr) & 0x1) ? "EN" : "Dis", - ((msr) & 0x2) ? "EN" : "Dis"); + cpu, msr, ((msr) & 0x1) ? "EN" : "Dis", ((msr) & 0x2) ? "EN" : "Dis"); } if (get_msr(cpu, MSR_HWP_STATUS, &msr)) return 0; fprintf(outf, "cpu%d: MSR_HWP_STATUS: 0x%08llx " - "(%sGuaranteed_Perf_Change, %sExcursion_Min)\n", - cpu, msr, - ((msr) & 0x1) ? "" : "No-", - ((msr) & 0x2) ? "" : "No-"); + "(%sGuaranteed_Perf_Change, %sExcursion_Min)\n", + cpu, msr, ((msr) & 0x1) ? "" : "No-", ((msr) & 0x2) ? "" : "No-"); return 0; } @@ -4257,8 +4215,7 @@ int print_perf_limit(struct thread_data *t, struct core_data *c, struct pkg_data (msr & 1 << 5) ? "Auto-HWP, " : "", (msr & 1 << 4) ? "Graphics, " : "", (msr & 1 << 2) ? "bit2, " : "", - (msr & 1 << 1) ? "ThermStatus, " : "", - (msr & 1 << 0) ? "PROCHOT, " : ""); + (msr & 1 << 1) ? "ThermStatus, " : "", (msr & 1 << 0) ? "PROCHOT, " : ""); fprintf(outf, " (Logged: %s%s%s%s%s%s%s%s%s%s%s%s%s%s)\n", (msr & 1 << 31) ? "bit31, " : "", (msr & 1 << 30) ? "bit30, " : "", @@ -4272,8 +4229,7 @@ int print_perf_limit(struct thread_data *t, struct core_data *c, struct pkg_data (msr & 1 << 21) ? "Auto-HWP, " : "", (msr & 1 << 20) ? "Graphics, " : "", (msr & 1 << 18) ? "bit18, " : "", - (msr & 1 << 17) ? "ThermStatus, " : "", - (msr & 1 << 16) ? "PROCHOT, " : ""); + (msr & 1 << 17) ? "ThermStatus, " : "", (msr & 1 << 16) ? "PROCHOT, " : ""); } if (do_gfx_perf_limit_reasons) { @@ -4286,8 +4242,7 @@ int print_perf_limit(struct thread_data *t, struct core_data *c, struct pkg_data (msr & 1 << 6) ? "VR-Therm, " : "", (msr & 1 << 8) ? "Amps, " : "", (msr & 1 << 9) ? "GFXPwr, " : "", - (msr & 1 << 10) ? "PkgPwrL1, " : "", - (msr & 1 << 11) ? "PkgPwrL2, " : ""); + (msr & 1 << 10) ? "PkgPwrL1, " : "", (msr & 1 << 11) ? "PkgPwrL2, " : ""); fprintf(outf, " (Logged: %s%s%s%s%s%s%s%s)\n", (msr & 1 << 16) ? "PROCHOT, " : "", (msr & 1 << 17) ? "ThermStatus, " : "", @@ -4295,8 +4250,7 @@ int print_perf_limit(struct thread_data *t, struct core_data *c, struct pkg_data (msr & 1 << 22) ? "VR-Therm, " : "", (msr & 1 << 24) ? "Amps, " : "", (msr & 1 << 25) ? "GFXPwr, " : "", - (msr & 1 << 26) ? "PkgPwrL1, " : "", - (msr & 1 << 27) ? "PkgPwrL2, " : ""); + (msr & 1 << 26) ? "PkgPwrL1, " : "", (msr & 1 << 27) ? "PkgPwrL2, " : ""); } if (do_ring_perf_limit_reasons) { get_msr(cpu, MSR_RING_PERF_LIMIT_REASONS, &msr); @@ -4306,21 +4260,19 @@ int print_perf_limit(struct thread_data *t, struct core_data *c, struct pkg_data (msr & 1 << 1) ? "ThermStatus, " : "", (msr & 1 << 6) ? "VR-Therm, " : "", (msr & 1 << 8) ? "Amps, " : "", - (msr & 1 << 10) ? "PkgPwrL1, " : "", - (msr & 1 << 11) ? "PkgPwrL2, " : ""); + (msr & 1 << 10) ? "PkgPwrL1, " : "", (msr & 1 << 11) ? "PkgPwrL2, " : ""); fprintf(outf, " (Logged: %s%s%s%s%s%s)\n", (msr & 1 << 16) ? "PROCHOT, " : "", (msr & 1 << 17) ? "ThermStatus, " : "", (msr & 1 << 22) ? "VR-Therm, " : "", (msr & 1 << 24) ? "Amps, " : "", - (msr & 1 << 26) ? "PkgPwrL1, " : "", - (msr & 1 << 27) ? "PkgPwrL2, " : ""); + (msr & 1 << 26) ? "PkgPwrL1, " : "", (msr & 1 << 27) ? "PkgPwrL2, " : ""); } return 0; } #define RAPL_POWER_GRANULARITY 0x7FFF /* 15 bit power granularity */ -#define RAPL_TIME_GRANULARITY 0x3F /* 6 bit time granularity */ +#define RAPL_TIME_GRANULARITY 0x3F /* 6 bit time granularity */ double get_tdp_intel(unsigned int model) { @@ -4349,8 +4301,7 @@ double get_tdp_amd(unsigned int family) * rapl_dram_energy_units_probe() * Energy units are either hard-coded, or come from RAPL Energy Unit MSR. */ -static double -rapl_dram_energy_units_probe(int model, double rapl_energy_units) +static double rapl_dram_energy_units_probe(int model, double rapl_energy_units) { /* only called for genuine_intel, family 6 */ @@ -4402,7 +4353,9 @@ void rapl_probe_intel(unsigned int family, unsigned int model) BIC_PRESENT(BIC_PkgWatt); break; case INTEL_FAM6_ATOM_TREMONT: /* EHL */ - do_rapl = RAPL_PKG | RAPL_CORES | RAPL_CORE_POLICY | RAPL_DRAM | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS | RAPL_GFX | RAPL_PKG_POWER_INFO; + do_rapl = + RAPL_PKG | RAPL_CORES | RAPL_CORE_POLICY | RAPL_DRAM | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS + | RAPL_GFX | RAPL_PKG_POWER_INFO; if (rapl_joules) { BIC_PRESENT(BIC_Pkg_J); BIC_PRESENT(BIC_Cor_J); @@ -4425,7 +4378,9 @@ void rapl_probe_intel(unsigned int family, unsigned int model) break; case INTEL_FAM6_SKYLAKE_L: /* SKL */ case INTEL_FAM6_CANNONLAKE_L: /* CNL */ - do_rapl = RAPL_PKG | RAPL_CORES | RAPL_CORE_POLICY | RAPL_DRAM | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS | RAPL_GFX | RAPL_PKG_POWER_INFO; + do_rapl = + RAPL_PKG | RAPL_CORES | RAPL_CORE_POLICY | RAPL_DRAM | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS + | RAPL_GFX | RAPL_PKG_POWER_INFO; BIC_PRESENT(BIC_PKG__); BIC_PRESENT(BIC_RAM__); if (rapl_joules) { @@ -4445,7 +4400,9 @@ void rapl_probe_intel(unsigned int family, unsigned int model) case INTEL_FAM6_SKYLAKE_X: /* SKX */ case INTEL_FAM6_ICELAKE_X: /* ICX */ case INTEL_FAM6_XEON_PHI_KNL: /* KNL */ - do_rapl = RAPL_PKG | RAPL_DRAM | RAPL_DRAM_POWER_INFO | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS | RAPL_PKG_POWER_INFO; + do_rapl = + RAPL_PKG | RAPL_DRAM | RAPL_DRAM_POWER_INFO | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS | + RAPL_PKG_POWER_INFO; BIC_PRESENT(BIC_PKG__); BIC_PRESENT(BIC_RAM__); if (rapl_joules) { @@ -4458,7 +4415,9 @@ void rapl_probe_intel(unsigned int family, unsigned int model) break; case INTEL_FAM6_SANDYBRIDGE_X: case INTEL_FAM6_IVYBRIDGE_X: - do_rapl = RAPL_PKG | RAPL_CORES | RAPL_CORE_POLICY | RAPL_DRAM | RAPL_DRAM_POWER_INFO | RAPL_PKG_PERF_STATUS | RAPL_DRAM_PERF_STATUS | RAPL_PKG_POWER_INFO; + do_rapl = + RAPL_PKG | RAPL_CORES | RAPL_CORE_POLICY | RAPL_DRAM | RAPL_DRAM_POWER_INFO | RAPL_PKG_PERF_STATUS | + RAPL_DRAM_PERF_STATUS | RAPL_PKG_POWER_INFO; BIC_PRESENT(BIC_PKG__); BIC_PRESENT(BIC_RAM__); if (rapl_joules) { @@ -4483,7 +4442,9 @@ void rapl_probe_intel(unsigned int family, unsigned int model) } break; case INTEL_FAM6_ATOM_GOLDMONT_D: /* DNV */ - do_rapl = RAPL_PKG | RAPL_DRAM | RAPL_DRAM_POWER_INFO | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS | RAPL_PKG_POWER_INFO | RAPL_CORES_ENERGY_STATUS; + do_rapl = + RAPL_PKG | RAPL_DRAM | RAPL_DRAM_POWER_INFO | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS | + RAPL_PKG_POWER_INFO | RAPL_CORES_ENERGY_STATUS; BIC_PRESENT(BIC_PKG__); BIC_PRESENT(BIC_RAM__); if (rapl_joules) { @@ -4600,8 +4561,7 @@ void perf_limit_reasons_probe(unsigned int family, unsigned int model) void automatic_cstate_conversion_probe(unsigned int family, unsigned int model) { - if (is_skx(family, model) || is_bdx(family, model) || - is_icx(family, model)) + if (is_skx(family, model) || is_bdx(family, model) || is_icx(family, model)) has_automatic_cstate_conversion = 1; } @@ -4636,8 +4596,7 @@ int print_thermal(struct thread_data *t, struct core_data *c, struct pkg_data *p return 0; dts = (msr >> 16) & 0x7F; - fprintf(outf, "cpu%d: MSR_IA32_PACKAGE_THERM_STATUS: 0x%08llx (%d C)\n", - cpu, msr, tj_max - dts); + fprintf(outf, "cpu%d: MSR_IA32_PACKAGE_THERM_STATUS: 0x%08llx (%d C)\n", cpu, msr, tj_max - dts); if (get_msr(cpu, MSR_IA32_PACKAGE_THERM_INTERRUPT, &msr)) return 0; @@ -4648,7 +4607,6 @@ int print_thermal(struct thread_data *t, struct core_data *c, struct pkg_data *p cpu, msr, tj_max - dts, tj_max - dts2); } - if (do_dts && debug) { unsigned int resolution; @@ -4678,7 +4636,7 @@ void print_power_limit_msr(int cpu, unsigned long long msr, char *label) cpu, label, ((msr >> 15) & 1) ? "EN" : "DIS", ((msr >> 0) & 0x7FFF) * rapl_power_units, - (1.0 + (((msr >> 22) & 0x3)/4.0)) * (1 << ((msr >> 17) & 0x1F)) * rapl_time_units, + (1.0 + (((msr >> 22) & 0x3) / 4.0)) * (1 << ((msr >> 17) & 0x1F)) * rapl_time_units, (((msr >> 16) & 1) ? "EN" : "DIS")); return; @@ -4719,12 +4677,11 @@ int print_rapl(struct thread_data *t, struct core_data *c, struct pkg_data *p) if (do_rapl & RAPL_PKG_POWER_INFO) { if (get_msr(cpu, MSR_PKG_POWER_INFO, &msr)) - return -5; - + return -5; fprintf(outf, "cpu%d: MSR_PKG_POWER_INFO: 0x%08llx (%.0f W TDP, RAPL %.0f - %.0f W, %f sec.)\n", cpu, msr, - ((msr >> 0) & RAPL_POWER_GRANULARITY) * rapl_power_units, + ((msr >> 0) & RAPL_POWER_GRANULARITY) * rapl_power_units, ((msr >> 16) & RAPL_POWER_GRANULARITY) * rapl_power_units, ((msr >> 32) & RAPL_POWER_GRANULARITY) * rapl_power_units, ((msr >> 48) & RAPL_TIME_GRANULARITY) * rapl_time_units); @@ -4743,17 +4700,17 @@ int print_rapl(struct thread_data *t, struct core_data *c, struct pkg_data *p) cpu, ((msr >> 47) & 1) ? "EN" : "DIS", ((msr >> 32) & 0x7FFF) * rapl_power_units, - (1.0 + (((msr >> 54) & 0x3)/4.0)) * (1 << ((msr >> 49) & 0x1F)) * rapl_time_units, + (1.0 + (((msr >> 54) & 0x3) / 4.0)) * (1 << ((msr >> 49) & 0x1F)) * rapl_time_units, ((msr >> 48) & 1) ? "EN" : "DIS"); } if (do_rapl & RAPL_DRAM_POWER_INFO) { if (get_msr(cpu, MSR_DRAM_POWER_INFO, &msr)) - return -6; + return -6; fprintf(outf, "cpu%d: MSR_DRAM_POWER_INFO,: 0x%08llx (%.0f W TDP, RAPL %.0f - %.0f W, %f sec.)\n", cpu, msr, - ((msr >> 0) & RAPL_POWER_GRANULARITY) * rapl_power_units, + ((msr >> 0) & RAPL_POWER_GRANULARITY) * rapl_power_units, ((msr >> 16) & RAPL_POWER_GRANULARITY) * rapl_power_units, ((msr >> 32) & RAPL_POWER_GRANULARITY) * rapl_power_units, ((msr >> 48) & RAPL_TIME_GRANULARITY) * rapl_time_units); @@ -4762,7 +4719,7 @@ int print_rapl(struct thread_data *t, struct core_data *c, struct pkg_data *p) if (get_msr(cpu, MSR_DRAM_POWER_LIMIT, &msr)) return -9; fprintf(outf, "cpu%d: MSR_DRAM_POWER_LIMIT: 0x%08llx (%slocked)\n", - cpu, msr, (msr >> 31) & 1 ? "" : "UN"); + cpu, msr, (msr >> 31) & 1 ? "" : "UN"); print_power_limit_msr(cpu, msr, "DRAM Limit"); } @@ -4776,7 +4733,7 @@ int print_rapl(struct thread_data *t, struct core_data *c, struct pkg_data *p) if (get_msr(cpu, MSR_PP0_POWER_LIMIT, &msr)) return -9; fprintf(outf, "cpu%d: MSR_PP0_POWER_LIMIT: 0x%08llx (%slocked)\n", - cpu, msr, (msr >> 31) & 1 ? "" : "UN"); + cpu, msr, (msr >> 31) & 1 ? "" : "UN"); print_power_limit_msr(cpu, msr, "Cores Limit"); } if (do_rapl & RAPL_GFX) { @@ -4788,7 +4745,7 @@ int print_rapl(struct thread_data *t, struct core_data *c, struct pkg_data *p) if (get_msr(cpu, MSR_PP1_POWER_LIMIT, &msr)) return -9; fprintf(outf, "cpu%d: MSR_PP1_POWER_LIMIT: 0x%08llx (%slocked)\n", - cpu, msr, (msr >> 31) & 1 ? "" : "UN"); + cpu, msr, (msr >> 31) & 1 ? "" : "UN"); print_power_limit_msr(cpu, msr, "GFX Limit"); } return 0; @@ -4810,24 +4767,24 @@ int has_snb_msrs(unsigned int family, unsigned int model) switch (model) { case INTEL_FAM6_SANDYBRIDGE: case INTEL_FAM6_SANDYBRIDGE_X: - case INTEL_FAM6_IVYBRIDGE: /* IVB */ - case INTEL_FAM6_IVYBRIDGE_X: /* IVB Xeon */ - case INTEL_FAM6_HASWELL: /* HSW */ - case INTEL_FAM6_HASWELL_X: /* HSW */ - case INTEL_FAM6_HASWELL_L: /* HSW */ - case INTEL_FAM6_HASWELL_G: /* HSW */ - case INTEL_FAM6_BROADWELL: /* BDW */ - case INTEL_FAM6_BROADWELL_G: /* BDW */ - case INTEL_FAM6_BROADWELL_X: /* BDX */ - case INTEL_FAM6_SKYLAKE_L: /* SKL */ - case INTEL_FAM6_CANNONLAKE_L: /* CNL */ - case INTEL_FAM6_SKYLAKE_X: /* SKX */ - case INTEL_FAM6_ICELAKE_X: /* ICX */ - case INTEL_FAM6_ATOM_GOLDMONT: /* BXT */ + case INTEL_FAM6_IVYBRIDGE: /* IVB */ + case INTEL_FAM6_IVYBRIDGE_X: /* IVB Xeon */ + case INTEL_FAM6_HASWELL: /* HSW */ + case INTEL_FAM6_HASWELL_X: /* HSW */ + case INTEL_FAM6_HASWELL_L: /* HSW */ + case INTEL_FAM6_HASWELL_G: /* HSW */ + case INTEL_FAM6_BROADWELL: /* BDW */ + case INTEL_FAM6_BROADWELL_G: /* BDW */ + case INTEL_FAM6_BROADWELL_X: /* BDX */ + case INTEL_FAM6_SKYLAKE_L: /* SKL */ + case INTEL_FAM6_CANNONLAKE_L: /* CNL */ + case INTEL_FAM6_SKYLAKE_X: /* SKX */ + case INTEL_FAM6_ICELAKE_X: /* ICX */ + case INTEL_FAM6_ATOM_GOLDMONT: /* BXT */ case INTEL_FAM6_ATOM_GOLDMONT_PLUS: case INTEL_FAM6_ATOM_GOLDMONT_D: /* DNV */ - case INTEL_FAM6_ATOM_TREMONT: /* EHL */ - case INTEL_FAM6_ATOM_TREMONT_D: /* JVL */ + case INTEL_FAM6_ATOM_TREMONT: /* EHL */ + case INTEL_FAM6_ATOM_TREMONT_D: /* JVL */ return 1; } return 0; @@ -4913,7 +4870,7 @@ int is_cnl(unsigned int family, unsigned int model) return 0; switch (model) { - case INTEL_FAM6_CANNONLAKE_L: /* CNL */ + case INTEL_FAM6_CANNONLAKE_L: /* CNL */ return 1; } @@ -4928,7 +4885,7 @@ unsigned int get_aperf_mperf_multiplier(unsigned int family, unsigned int model) } #define SLM_BCLK_FREQS 5 -double slm_freq_table[SLM_BCLK_FREQS] = { 83.3, 100.0, 133.3, 116.7, 80.0}; +double slm_freq_table[SLM_BCLK_FREQS] = { 83.3, 100.0, 133.3, 116.7, 80.0 }; double slm_bclk(void) { @@ -4979,7 +4936,7 @@ int get_cpu_type(struct thread_data *t, struct core_data *c, struct pkg_data *p) __cpuid(0x1a, eax, ebx, ecx, edx); eax = (eax >> 24) & 0xFF; - if (eax == 0x20 ) + if (eax == 0x20) t->is_atom = true; return 0; } @@ -5018,8 +4975,7 @@ int set_temperature_target(struct thread_data *t, struct core_data *c, struct pk if (tj_max_override != 0) { tj_max = tj_max_override; - fprintf(outf, "cpu%d: Using cmdline TCC Target (%d C)\n", - cpu, tj_max); + fprintf(outf, "cpu%d: Using cmdline TCC Target (%d C)\n", cpu, tj_max); return 0; } @@ -5037,16 +4993,15 @@ int set_temperature_target(struct thread_data *t, struct core_data *c, struct pk case 4: tcc_offset = (msr >> 24) & 0xF; fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C) (%d default - %d offset)\n", - cpu, msr, tcc_default - tcc_offset, tcc_default, tcc_offset); + cpu, msr, tcc_default - tcc_offset, tcc_default, tcc_offset); break; case 6: tcc_offset = (msr >> 24) & 0x3F; fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C) (%d default - %d offset)\n", - cpu, msr, tcc_default - tcc_offset, tcc_default, tcc_offset); + cpu, msr, tcc_default - tcc_offset, tcc_default, tcc_offset); break; default: - fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C)\n", - cpu, msr, tcc_default); + fprintf(outf, "cpu%d: MSR_IA32_TEMPERATURE_TARGET: 0x%08llx (%d C)\n", cpu, msr, tcc_default); break; } } @@ -5060,8 +5015,7 @@ int set_temperature_target(struct thread_data *t, struct core_data *c, struct pk guess: tj_max = TJMAX_DEFAULT; - fprintf(outf, "cpu%d: Guessing tjMax %d C, Please use -T to specify\n", - cpu, tj_max); + fprintf(outf, "cpu%d: Guessing tjMax %d C, Please use -T to specify\n", cpu, tj_max); return 0; } @@ -5072,9 +5026,7 @@ void decode_feature_control_msr(void) if (!get_msr(base_cpu, MSR_IA32_FEAT_CTL, &msr)) fprintf(outf, "cpu%d: MSR_IA32_FEATURE_CONTROL: 0x%08llx (%sLocked %s)\n", - base_cpu, msr, - msr & FEAT_CTL_LOCKED ? "" : "UN-", - msr & (1 << 18) ? "SGX" : ""); + base_cpu, msr, msr & FEAT_CTL_LOCKED ? "" : "UN-", msr & (1 << 18) ? "SGX" : ""); } void decode_misc_enable_msr(void) @@ -5102,13 +5054,12 @@ void decode_misc_feature_control(void) return; if (!get_msr(base_cpu, MSR_MISC_FEATURE_CONTROL, &msr)) - fprintf(outf, "cpu%d: MSR_MISC_FEATURE_CONTROL: 0x%08llx (%sL2-Prefetch %sL2-Prefetch-pair %sL1-Prefetch %sL1-IP-Prefetch)\n", - base_cpu, msr, - msr & (0 << 0) ? "No-" : "", - msr & (1 << 0) ? "No-" : "", - msr & (2 << 0) ? "No-" : "", - msr & (3 << 0) ? "No-" : ""); + fprintf(outf, + "cpu%d: MSR_MISC_FEATURE_CONTROL: 0x%08llx (%sL2-Prefetch %sL2-Prefetch-pair %sL1-Prefetch %sL1-IP-Prefetch)\n", + base_cpu, msr, msr & (0 << 0) ? "No-" : "", msr & (1 << 0) ? "No-" : "", + msr & (2 << 0) ? "No-" : "", msr & (3 << 0) ? "No-" : ""); } + /* * Decode MSR_MISC_PWR_MGMT * @@ -5129,10 +5080,9 @@ void decode_misc_pwr_mgmt_msr(void) if (!get_msr(base_cpu, MSR_MISC_PWR_MGMT, &msr)) fprintf(outf, "cpu%d: MSR_MISC_PWR_MGMT: 0x%08llx (%sable-EIST_Coordination %sable-EPB %sable-OOB)\n", base_cpu, msr, - msr & (1 << 0) ? "DIS" : "EN", - msr & (1 << 1) ? "EN" : "DIS", - msr & (1 << 8) ? "EN" : "DIS"); + msr & (1 << 0) ? "DIS" : "EN", msr & (1 << 1) ? "EN" : "DIS", msr & (1 << 8) ? "EN" : "DIS"); } + /* * Decode MSR_CC6_DEMOTION_POLICY_CONFIG, MSR_MC6_DEMOTION_POLICY_CONFIG * @@ -5158,10 +5108,10 @@ void decode_c6_demotion_policy_msr(void) unsigned int intel_model_duplicates(unsigned int model) { - switch(model) { + switch (model) { case INTEL_FAM6_NEHALEM_EP: /* Core i7, Xeon 5500 series - Bloomfield, Gainstown NHM-EP */ case INTEL_FAM6_NEHALEM: /* Core i7 and i5 Processor - Clarksfield, Lynnfield, Jasper Forest */ - case 0x1F: /* Core i7 and i5 Processor - Nehalem */ + case 0x1F: /* Core i7 and i5 Processor - Nehalem */ case INTEL_FAM6_WESTMERE: /* Westmere Client - Clarkdale, Arrandale */ case INTEL_FAM6_WESTMERE_EP: /* Westmere EP - Gulftown */ return INTEL_FAM6_NEHALEM; @@ -5224,13 +5174,11 @@ void print_dev_latency(void) close(fd); return; } - fprintf(outf, "/dev/cpu_dma_latency: %d usec (%s)\n", - value, value == 2000000000 ? "default" : "constrained"); + fprintf(outf, "/dev/cpu_dma_latency: %d usec (%s)\n", value, value == 2000000000 ? "default" : "constrained"); close(fd); } - /* * Linux-perf manages the the HW instructions-retired counter * by enabling when requested, and hiding rollover @@ -5296,7 +5244,8 @@ void process_cpuid() if (!quiet) { fprintf(outf, "CPUID(1): family:model:stepping 0x%x:%x:%x (%d:%d:%d) microcode 0x%x\n", - family, model, stepping, family, model, stepping, (unsigned int)((ucode_patch >> 32) & 0xFFFFFFFF)); + family, model, stepping, family, model, stepping, + (unsigned int)((ucode_patch >> 32) & 0xFFFFFFFF)); fprintf(outf, "CPUID(0x80000000): max_extended_levels: 0x%x\n", max_extended_level); fprintf(outf, "CPUID(1): %s %s %s %s %s %s %s %s %s %s\n", ecx_flags & (1 << 0) ? "SSE3" : "-", @@ -5307,8 +5256,7 @@ void process_cpuid() edx_flags & (1 << 4) ? "TSC" : "-", edx_flags & (1 << 5) ? "MSR" : "-", edx_flags & (1 << 22) ? "ACPI-TM" : "-", - edx_flags & (1 << 28) ? "HT" : "-", - edx_flags & (1 << 29) ? "TM" : "-"); + edx_flags & (1 << 28) ? "HT" : "-", edx_flags & (1 << 29) ? "TM" : "-"); } if (genuine_intel) { model_orig = model; @@ -5364,14 +5312,11 @@ void process_cpuid() has_hwp ? "" : "No-", has_hwp_notify ? "" : "No-", has_hwp_activity_window ? "" : "No-", - has_hwp_epp ? "" : "No-", - has_hwp_pkg ? "" : "No-", - has_epb ? "" : "No-"); + has_hwp_epp ? "" : "No-", has_hwp_pkg ? "" : "No-", has_epb ? "" : "No-"); if (!quiet) decode_misc_enable_msr(); - if (max_level >= 0x7 && !quiet) { int has_sgx; @@ -5403,7 +5348,7 @@ void process_cpuid() eax_crystal, ebx_tsc, crystal_hz); if (crystal_hz == 0) - switch(model) { + switch (model) { case INTEL_FAM6_SKYLAKE_L: /* SKL */ crystal_hz = 24000000; /* 24.0 MHz */ break; @@ -5416,13 +5361,13 @@ void process_cpuid() break; default: crystal_hz = 0; - } + } if (crystal_hz) { - tsc_hz = (unsigned long long) crystal_hz * ebx_tsc / eax_crystal; + tsc_hz = (unsigned long long)crystal_hz *ebx_tsc / eax_crystal; if (!quiet) fprintf(outf, "TSC: %lld MHz (%d Hz * %d / %d / 1000000)\n", - tsc_hz / 1000000, crystal_hz, ebx_tsc, eax_crystal); + tsc_hz / 1000000, crystal_hz, ebx_tsc, eax_crystal); } } } @@ -5517,10 +5462,9 @@ void process_cpuid() BIC_PRESENT(BIC_CPUGFX); } do_slm_cstates = is_slm(family, model); - do_knl_cstates = is_knl(family, model); + do_knl_cstates = is_knl(family, model); - if (do_slm_cstates || do_knl_cstates || is_cnl(family, model) || - is_ehl(family, model)) + if (do_slm_cstates || do_knl_cstates || is_cnl(family, model) || is_ehl(family, model)) BIC_NOT_PRESENT(BIC_CPU_c3); if (!quiet) @@ -5614,7 +5558,7 @@ void topology_probe() if (debug > 1) fprintf(outf, "num_cpus %d max_cpu_num %d\n", topo.num_cpus, topo.max_cpu_num); - cpus = calloc(1, (topo.max_cpu_num + 1) * sizeof(struct cpu_topology)); + cpus = calloc(1, (topo.max_cpu_num + 1) * sizeof(struct cpu_topology)); if (cpus == NULL) err(1, "calloc cpus"); @@ -5693,22 +5637,19 @@ void topology_probe() topo.cores_per_node = max_core_id + 1; if (debug > 1) - fprintf(outf, "max_core_id %d, sizing for %d cores per package\n", - max_core_id, topo.cores_per_node); + fprintf(outf, "max_core_id %d, sizing for %d cores per package\n", max_core_id, topo.cores_per_node); if (!summary_only && topo.cores_per_node > 1) BIC_PRESENT(BIC_Core); topo.num_die = max_die_id + 1; if (debug > 1) - fprintf(outf, "max_die_id %d, sizing for %d die\n", - max_die_id, topo.num_die); + fprintf(outf, "max_die_id %d, sizing for %d die\n", max_die_id, topo.num_die); if (!summary_only && topo.num_die > 1) BIC_PRESENT(BIC_Die); topo.num_packages = max_package_id + 1; if (debug > 1) - fprintf(outf, "max_package_id %d, sizing for %d packages\n", - max_package_id, topo.num_packages); + fprintf(outf, "max_package_id %d, sizing for %d packages\n", max_package_id, topo.num_packages); if (!summary_only && topo.num_packages > 1) BIC_PRESENT(BIC_Package); @@ -5731,21 +5672,15 @@ void topology_probe() fprintf(outf, "cpu %d pkg %d die %d node %d lnode %d core %d thread %d\n", i, cpus[i].physical_package_id, cpus[i].die_id, - cpus[i].physical_node_id, - cpus[i].logical_node_id, - cpus[i].physical_core_id, - cpus[i].thread_id); + cpus[i].physical_node_id, cpus[i].logical_node_id, cpus[i].physical_core_id, cpus[i].thread_id); } } -void -allocate_counters(struct thread_data **t, struct core_data **c, - struct pkg_data **p) +void allocate_counters(struct thread_data **t, struct core_data **c, struct pkg_data **p) { int i; - int num_cores = topo.cores_per_node * topo.nodes_per_pkg * - topo.num_packages; + int num_cores = topo.cores_per_node * topo.nodes_per_pkg * topo.num_packages; int num_threads = topo.threads_per_core * num_cores; *t = calloc(num_threads, sizeof(struct thread_data)); @@ -5773,13 +5708,13 @@ allocate_counters(struct thread_data **t, struct core_data **c, error: err(1, "calloc counters"); } + /* * init_counter() * * set FIRST_THREAD_IN_CORE and FIRST_CORE_IN_PACKAGE */ -void init_counter(struct thread_data *thread_base, struct core_data *core_base, - struct pkg_data *pkg_base, int cpu_id) +void init_counter(struct thread_data *thread_base, struct core_data *core_base, struct pkg_data *pkg_base, int cpu_id) { int pkg_id = cpus[cpu_id].physical_package_id; int node_id = cpus[cpu_id].logical_node_id; @@ -5789,7 +5724,6 @@ void init_counter(struct thread_data *thread_base, struct core_data *core_base, struct core_data *c; struct pkg_data *p; - /* Workaround for systems where physical_node_id==-1 * and logical_node_id==(-1 - topo.num_cpus) */ @@ -5811,7 +5745,6 @@ void init_counter(struct thread_data *thread_base, struct core_data *core_base, p->package_id = pkg_id; } - int initialize_counters(int cpu_id) { init_counter(EVEN_COUNTERS, cpu_id); @@ -5826,12 +5759,14 @@ void allocate_output_buffer() if (outp == NULL) err(-1, "calloc output buffer"); } + void allocate_fd_percpu(void) { fd_percpu = calloc(topo.max_cpu_num + 1, sizeof(int)); if (fd_percpu == NULL) err(-1, "calloc fd_percpu"); } + void allocate_irq_buffers(void) { irq_column_2_cpu = calloc(topo.num_cpus, sizeof(int)); @@ -5842,6 +5777,7 @@ void allocate_irq_buffers(void) if (irqs_per_cpu == NULL) err(-1, "calloc %d", topo.max_cpu_num + 1); } + void setup_all_buffers(void) { topology_probe(); @@ -5872,7 +5808,6 @@ void turbostat_init() process_cpuid(); linux_perf_init(); - if (!quiet) for_all_cpus(print_hwp, ODD_COUNTERS); @@ -5945,7 +5880,7 @@ int fork_it(char **argv) format_all_counters(EVEN_COUNTERS); } - fprintf(outf, "%.6f sec\n", tv_delta.tv_sec + tv_delta.tv_usec/1000000.0); + fprintf(outf, "%.6f sec\n", tv_delta.tv_sec + tv_delta.tv_usec / 1000000.0); flush_output_stderr(); @@ -5970,14 +5905,14 @@ int get_and_dump_counters(void) return status; } -void print_version() { - fprintf(outf, "turbostat version 21.03.12" - " - Len Brown \n"); +void print_version() +{ + fprintf(outf, "turbostat version 21.03.12" " - Len Brown \n"); } int add_counter(unsigned int msr_num, char *path, char *name, - unsigned int width, enum counter_scope scope, - enum counter_type type, enum counter_format format, int flags) + unsigned int width, enum counter_scope scope, + enum counter_type type, enum counter_format format, int flags) { struct msr_counter *msrp; @@ -6003,8 +5938,7 @@ int add_counter(unsigned int msr_num, char *path, char *name, sys.tp = msrp; sys.added_thread_counters++; if (sys.added_thread_counters > MAX_ADDED_THREAD_COUNTERS) { - fprintf(stderr, "exceeded max %d added thread counters\n", - MAX_ADDED_COUNTERS); + fprintf(stderr, "exceeded max %d added thread counters\n", MAX_ADDED_COUNTERS); exit(-1); } break; @@ -6014,8 +5948,7 @@ int add_counter(unsigned int msr_num, char *path, char *name, sys.cp = msrp; sys.added_core_counters++; if (sys.added_core_counters > MAX_ADDED_COUNTERS) { - fprintf(stderr, "exceeded max %d added core counters\n", - MAX_ADDED_COUNTERS); + fprintf(stderr, "exceeded max %d added core counters\n", MAX_ADDED_COUNTERS); exit(-1); } break; @@ -6025,8 +5958,7 @@ int add_counter(unsigned int msr_num, char *path, char *name, sys.pp = msrp; sys.added_package_counters++; if (sys.added_package_counters > MAX_ADDED_COUNTERS) { - fprintf(stderr, "exceeded max %d added package counters\n", - MAX_ADDED_COUNTERS); + fprintf(stderr, "exceeded max %d added package counters\n", MAX_ADDED_COUNTERS); exit(-1); } break; @@ -6163,15 +6095,14 @@ void probe_sysfs(void) for (state = 10; state >= 0; --state) { - sprintf(path, "/sys/devices/system/cpu/cpu%d/cpuidle/state%d/name", - base_cpu, state); + sprintf(path, "/sys/devices/system/cpu/cpu%d/cpuidle/state%d/name", base_cpu, state); input = fopen(path, "r"); if (input == NULL) continue; if (!fgets(name_buf, sizeof(name_buf), input)) err(1, "%s: failed to read file", path); - /* truncate "C1-HSW\n" to "C1", or truncate "C1\n" to "C1" */ + /* truncate "C1-HSW\n" to "C1", or truncate "C1\n" to "C1" */ sp = strchr(name_buf, '-'); if (!sp) sp = strchrnul(name_buf, '\n'); @@ -6187,20 +6118,18 @@ void probe_sysfs(void) if (is_deferred_skip(name_buf)) continue; - add_counter(0, path, name_buf, 64, SCOPE_CPU, COUNTER_USEC, - FORMAT_PERCENT, SYSFS_PERCPU); + add_counter(0, path, name_buf, 64, SCOPE_CPU, COUNTER_USEC, FORMAT_PERCENT, SYSFS_PERCPU); } for (state = 10; state >= 0; --state) { - sprintf(path, "/sys/devices/system/cpu/cpu%d/cpuidle/state%d/name", - base_cpu, state); + sprintf(path, "/sys/devices/system/cpu/cpu%d/cpuidle/state%d/name", base_cpu, state); input = fopen(path, "r"); if (input == NULL) continue; if (!fgets(name_buf, sizeof(name_buf), input)) err(1, "%s: failed to read file", path); - /* truncate "C1-HSW\n" to "C1", or truncate "C1\n" to "C1" */ + /* truncate "C1-HSW\n" to "C1", or truncate "C1\n" to "C1" */ sp = strchr(name_buf, '-'); if (!sp) sp = strchrnul(name_buf, '\n'); @@ -6214,13 +6143,11 @@ void probe_sysfs(void) if (is_deferred_skip(name_buf)) continue; - add_counter(0, path, name_buf, 64, SCOPE_CPU, COUNTER_ITEMS, - FORMAT_DELTA, SYSFS_PERCPU); + add_counter(0, path, name_buf, 64, SCOPE_CPU, COUNTER_ITEMS, FORMAT_DELTA, SYSFS_PERCPU); } } - /* * parse cpuset with following syntax * 1,2,4..6,8-10 and set bits in cpu_subset @@ -6307,37 +6234,35 @@ error: exit(-1); } - void cmdline(int argc, char **argv) { int opt; int option_index = 0; static struct option long_options[] = { - {"add", required_argument, 0, 'a'}, - {"cpu", required_argument, 0, 'c'}, - {"Dump", no_argument, 0, 'D'}, - {"debug", no_argument, 0, 'd'}, /* internal, not documented */ - {"enable", required_argument, 0, 'e'}, - {"interval", required_argument, 0, 'i'}, - {"IPC", no_argument, 0, 'I'}, - {"num_iterations", required_argument, 0, 'n'}, - {"help", no_argument, 0, 'h'}, - {"hide", required_argument, 0, 'H'}, // meh, -h taken by --help - {"Joules", no_argument, 0, 'J'}, - {"list", no_argument, 0, 'l'}, - {"out", required_argument, 0, 'o'}, - {"quiet", no_argument, 0, 'q'}, - {"show", required_argument, 0, 's'}, - {"Summary", no_argument, 0, 'S'}, - {"TCC", required_argument, 0, 'T'}, - {"version", no_argument, 0, 'v' }, - {0, 0, 0, 0 } + { "add", required_argument, 0, 'a' }, + { "cpu", required_argument, 0, 'c' }, + { "Dump", no_argument, 0, 'D' }, + { "debug", no_argument, 0, 'd' }, /* internal, not documented */ + { "enable", required_argument, 0, 'e' }, + { "interval", required_argument, 0, 'i' }, + { "IPC", no_argument, 0, 'I' }, + { "num_iterations", required_argument, 0, 'n' }, + { "help", no_argument, 0, 'h' }, + { "hide", required_argument, 0, 'H' }, // meh, -h taken by --help + { "Joules", no_argument, 0, 'J' }, + { "list", no_argument, 0, 'l' }, + { "out", required_argument, 0, 'o' }, + { "quiet", no_argument, 0, 'q' }, + { "show", required_argument, 0, 's' }, + { "Summary", no_argument, 0, 'S' }, + { "TCC", required_argument, 0, 'T' }, + { "version", no_argument, 0, 'v' }, + { 0, 0, 0, 0 } }; progname = argv[0]; - while ((opt = getopt_long_only(argc, argv, "+C:c:Dde:hi:Jn:o:qST:v", - long_options, &option_index)) != -1) { + while ((opt = getopt_long_only(argc, argv, "+C:c:Dde:hi:Jn:o:qST:v", long_options, &option_index)) != -1) { switch (opt) { case 'a': parse_add_command(optarg); @@ -6372,8 +6297,7 @@ void cmdline(int argc, char **argv) double interval = strtod(optarg, NULL); if (interval < 0.001) { - fprintf(outf, "interval %f seconds is too small\n", - interval); + fprintf(outf, "interval %f seconds is too small\n", interval); exit(2); } @@ -6400,8 +6324,7 @@ void cmdline(int argc, char **argv) num_iterations = strtod(optarg, NULL); if (num_iterations <= 0) { - fprintf(outf, "iterations %d should be positive number\n", - num_iterations); + fprintf(outf, "iterations %d should be positive number\n", num_iterations); exit(2); } break; -- cgit v1.2.3-59-g8ed1b From 38c6663a68903cf1187003129cd1873551979865 Mon Sep 17 00:00:00 2001 From: Len Brown Date: Tue, 4 May 2021 19:27:34 -0400 Subject: tools/power turbostat: elevate priority of interval mode This makes interval mode less likely to see delayed results on a heavily loaded system. Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 66 ++++++++++++++++++++++++++++++----- 1 file changed, 57 insertions(+), 9 deletions(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 13805e460a4d..a0dc39d682cd 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -2266,31 +2266,48 @@ char *pkg_cstate_limit_strings[] = { "reserved", "unknown", "pc0", "pc1", "pc2", int nhm_pkg_cstate_limits[16] = { PCL__0, PCL__1, PCL__3, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, -PCLRSV, PCLRSV }; + PCLRSV, PCLRSV +}; + int snb_pkg_cstate_limits[16] = { PCL__0, PCL__2, PCL_6N, PCL_6R, PCL__7, PCL_7S, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, -PCLRSV, PCLRSV }; + PCLRSV, PCLRSV +}; + int hsw_pkg_cstate_limits[16] = { PCL__0, PCL__2, PCL__3, PCL__6, PCL__7, PCL_7S, PCL__8, PCL__9, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, -PCLRSV, PCLRSV }; + PCLRSV, PCLRSV +}; + int slv_pkg_cstate_limits[16] = { PCL__0, PCL__1, PCLRSV, PCLRSV, PCL__4, PCLRSV, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, -PCL__6, PCL__7 }; + PCL__6, PCL__7 +}; + int amt_pkg_cstate_limits[16] = { PCLUNL, PCL__1, PCL__2, PCLRSV, PCLRSV, PCLRSV, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, -PCLRSV, PCLRSV }; + PCLRSV, PCLRSV +}; + int phi_pkg_cstate_limits[16] = { PCL__0, PCL__2, PCL_6N, PCL_6R, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, -PCLRSV, PCLRSV }; + PCLRSV, PCLRSV +}; + int glm_pkg_cstate_limits[16] = { PCLUNL, PCL__1, PCL__3, PCL__6, PCL__7, PCL_7S, PCL__8, PCL__9, PCL_10, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, -PCLRSV, PCLRSV }; + PCLRSV, PCLRSV +}; + int skx_pkg_cstate_limits[16] = { PCL__0, PCL__2, PCL_6N, PCL_6R, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, -PCLRSV, PCLRSV }; + PCLRSV, PCLRSV +}; + int icx_pkg_cstate_limits[16] = { PCL__0, PCL__2, PCL__6, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, -PCLRSV, PCLRSV }; + PCLRSV, PCLRSV +}; static void calculate_tsc_tweak() { @@ -3401,6 +3418,32 @@ release_msr: free(per_cpu_msr_sum); } +/* + * set_my_sched_priority(pri) + * return previous + */ +int set_my_sched_priority(int priority) +{ + int retval; + int original_priority; + + errno = 0; + original_priority = getpriority(PRIO_PROCESS, 0); + if (errno && (original_priority == -1)) + err(errno, "getpriority"); + + retval = setpriority(PRIO_PROCESS, 0, priority); + if (retval) + err(retval, "setpriority(%d)", priority); + + errno = 0; + retval = getpriority(PRIO_PROCESS, 0); + if (retval != priority) + err(-1, "getpriority(%d) != setpriority(%d)", retval, priority); + + return original_priority; +} + void turbostat_loop() { int retval; @@ -3409,6 +3452,11 @@ void turbostat_loop() setup_signal_handler(); + /* + * elevate own priority for interval mode + */ + set_my_sched_priority(-20); + restart: restarted++; -- cgit v1.2.3-59-g8ed1b From b60c573dc241ab3a8719e990d86a0011b79eebcb Mon Sep 17 00:00:00 2001 From: Len Brown Date: Tue, 4 May 2021 19:56:17 -0400 Subject: tools/power turbostat: Support "turbostat --hide idle" As idle, in particular, can have many columns on some machines... Make it easy to ignore them all at once. Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.8 | 6 ++++-- tools/power/x86/turbostat/turbostat.c | 18 ++++++++++++++++++ 2 files changed, 22 insertions(+), 2 deletions(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.8 b/tools/power/x86/turbostat/turbostat.8 index f6b7e85b121c..9b17097bc3d7 100644 --- a/tools/power/x86/turbostat/turbostat.8 +++ b/tools/power/x86/turbostat/turbostat.8 @@ -54,12 +54,14 @@ name as necessary to disambiguate it from others is necessary. Note that option .PP \fB--cpu cpu-set\fP limit output to system summary plus the specified cpu-set. If cpu-set is the string "core", then the system summary plus the first CPU in each core are printed -- eg. subsequent HT siblings are not printed. Or if cpu-set is the string "package", then the system summary plus the first CPU in each package is printed. Otherwise, the system summary plus the specified set of CPUs are printed. The cpu-set is ordered from low to high, comma delimited with ".." and "-" permitted to denote a range. eg. 1,2,8,14..17,21-44 .PP -\fB--hide column\fP do not show the specified built-in columns. May be invoked multiple times, or with a comma-separated list of column names. Use "--hide sysfs" to hide the sysfs statistics columns as a group. +\fB--hide column\fP do not show the specified built-in columns. May be invoked multiple times, or with a comma-separated list of column names. .PP \fB--enable column\fP show the specified built-in columns, which are otherwise disabled, by default. Currently the only built-in counters disabled by default are "usec", "Time_Of_Day_Seconds", "APIC" and "X2APIC". The column name "all" can be used to enable all disabled-by-default built-in counters. .PP -\fB--show column\fP show only the specified built-in columns. May be invoked multiple times, or with a comma-separated list of column names. Use "--show sysfs" to show the sysfs statistics columns as a group. +\fB--show column\fP show only the specified built-in columns. May be invoked multiple times, or with a comma-separated list of column names. +.PP +\fB--show CATEGORY --hide CATEGORY\fP Show and hide also accept a single CATEGORY of columns: "all", "topology", "idle", "frequency", "power", "sysfs", "other". .PP \fB--Dump\fP displays the raw counter values. .PP diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index a0dc39d682cd..4bb08de4cb71 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -667,6 +667,12 @@ struct msr_counter bic[] = { #define BIC_GFXACTMHz (1ULL << 51) #define BIC_IPC (1ULL << 52) +#define BIC_TOPOLOGY (BIC_Package | BIC_Node | BIC_CoreCnt | BIC_PkgCnt | BIC_Core | BIC_CPU | BIC_Die ) +#define BIC_THERMAL_PWR ( BIC_CoreTmp | BIC_PkgTmp | BIC_PkgWatt | BIC_CorWatt | BIC_GFXWatt | BIC_RAMWatt | BIC_PKG__ | BIC_RAM__) +#define BIC_FREQUENCY ( BIC_Avg_MHz | BIC_Busy | BIC_Bzy_MHz | BIC_TSC_MHz | BIC_GFXMHz | BIC_GFXACTMHz ) +#define BIC_IDLE ( BIC_sysfs | BIC_CPU_c1 | BIC_CPU_c3 | BIC_CPU_c6 | BIC_CPU_c7 | BIC_GFX_rc6 | BIC_Pkgpc2 | BIC_Pkgpc3 | BIC_Pkgpc6 | BIC_Pkgpc7 | BIC_Pkgpc8 | BIC_Pkgpc9 | BIC_Pkgpc10 | BIC_CPU_LPI | BIC_SYS_LPI | BIC_Mod_c6 | BIC_Totl_c0 | BIC_Any_c0 | BIC_GFX_c0 | BIC_CPUGFX) +#define BIC_OTHER ( BIC_IRQ | BIC_SMI | BIC_ThreadC | BIC_CoreTmp | BIC_IPC) + #define BIC_DISABLED_BY_DEFAULT (BIC_USEC | BIC_TOD | BIC_APIC | BIC_X2APIC) unsigned long long bic_enabled = (0xFFFFFFFFFFFFFFFFULL & ~BIC_DISABLED_BY_DEFAULT); @@ -748,6 +754,18 @@ unsigned long long bic_lookup(char *name_list, enum show_hide_mode mode) if (!strcmp(name_list, "all")) return ~0; + if (!strcmp(name_list, "topology")) + return BIC_TOPOLOGY; + if (!strcmp(name_list, "power")) + return BIC_THERMAL_PWR; + if (!strcmp(name_list, "idle")) + return BIC_IDLE; + if (!strcmp(name_list, "frequency")) + return BIC_FREQUENCY; + if (!strcmp(name_list, "other")) + return BIC_OTHER; + if (!strcmp(name_list, "all")) + return 0; for (i = 0; i < MAX_BIC; ++i) { if (!strcmp(name_list, bic[i].name)) { -- cgit v1.2.3-59-g8ed1b From 3c070b2abf85b92455c2721d0a9edc68893ab6c1 Mon Sep 17 00:00:00 2001 From: Len Brown Date: Tue, 4 May 2021 19:58:08 -0400 Subject: tools/power turbostat: version 2021.05.04 Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'tools/power/x86') diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 4bb08de4cb71..47d3ba895d6d 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -3,7 +3,7 @@ * turbostat -- show CPU frequency and C-state residency * on modern Intel and AMD processors. * - * Copyright (c) 2013 Intel Corporation. + * Copyright (c) 2021 Intel Corporation. * Len Brown */ @@ -5973,7 +5973,7 @@ int get_and_dump_counters(void) void print_version() { - fprintf(outf, "turbostat version 21.03.12" " - Len Brown \n"); + fprintf(outf, "turbostat version 21.05.04" " - Len Brown \n"); } int add_counter(unsigned int msr_num, char *path, char *name, -- cgit v1.2.3-59-g8ed1b