From 5c1de006e8e66b0be05be422416629e344c71652 Mon Sep 17 00:00:00 2001 From: "sriram@marirs.net.in" Date: Mon, 22 Dec 2014 22:17:37 +0530 Subject: [PATCH 1/5] cpupower Makefile change to help run the tool without 'make install' The cpupower tool, when compiled against libcpupower.so fail's to run as the linker file path's are missing during compilation. So added changes in the Makefile to run cpupower tool, which helps us run the tool without doing a 'make install'. Signed-off-by: Sriram Raghunathan Acked-by: Thomas Renninger Signed-off-by: Rafael J. Wysocki --- tools/power/cpupower/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tools/power/cpupower/Makefile b/tools/power/cpupower/Makefile index 2e2ba2efa0d9..3ed7c0476d48 100644 --- a/tools/power/cpupower/Makefile +++ b/tools/power/cpupower/Makefile @@ -209,7 +209,7 @@ $(OUTPUT)%.o: %.c $(OUTPUT)cpupower: $(UTIL_OBJS) $(OUTPUT)libcpupower.so.$(LIB_MAJ) $(ECHO) " CC " $@ - $(QUIET) $(CC) $(CFLAGS) $(LDFLAGS) $(UTIL_OBJS) -lcpupower -lrt -lpci -L$(OUTPUT) -o $@ + $(QUIET) $(CC) $(CFLAGS) $(LDFLAGS) $(UTIL_OBJS) -lcpupower -Wl,-rpath=./ -lrt -lpci -L$(OUTPUT) -o $@ $(QUIET) $(STRIPCMD) $@ $(OUTPUT)po/$(PACKAGE).pot: $(UTIL_SRC) From 98481e79b60a50d699b79292ff1b7e56e7fa8425 Mon Sep 17 00:00:00 2001 From: Len Brown Date: Fri, 15 Aug 2014 00:36:50 -0400 Subject: [PATCH 2/5] tools/power turbostat: relax dependency on root permission For turbostat to run as non-root, it needs to permissions: 1. read access to /dev/cpu/*/msr via standard user/group/world file permissions 2. CAP_SYS_RAWIO eg. # setcap cap_sys_rawio=ep turbostat Yes, running as root still works. Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 53 +++++++++++++++++++++------ 1 file changed, 41 insertions(+), 12 deletions(-) diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 5b1b807265a1..6f29fc11fde6 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -38,6 +38,8 @@ #include #include #include +#include +#include char *proc_stat = "/proc/stat"; unsigned int interval_sec = 5; /* set with -i interval_sec */ @@ -251,15 +253,13 @@ int get_msr(int cpu, off_t offset, unsigned long long *msr) sprintf(pathname, "/dev/cpu/%d/msr", cpu); fd = open(pathname, O_RDONLY); if (fd < 0) - return -1; + err(-1, "%s open failed, try chown or chmod +r /dev/cpu/*/msr, or run as root", pathname); retval = pread(fd, msr, sizeof *msr, offset); close(fd); - if (retval != sizeof *msr) { - fprintf(stderr, "%s offset 0x%llx read failed\n", pathname, (unsigned long long)offset); - return -1; - } + if (retval != sizeof *msr) + err(-1, "%s offset 0x%llx read failed", pathname, (unsigned long long)offset); return 0; } @@ -1462,10 +1462,40 @@ void check_dev_msr() "Try \"# modprobe msr\""); } -void check_super_user() +void check_permissions() { - if (getuid() != 0) - errx(-6, "must be root"); + struct __user_cap_header_struct cap_header_data; + cap_user_header_t cap_header = &cap_header_data; + struct __user_cap_data_struct cap_data_data; + cap_user_data_t cap_data = &cap_data_data; + extern int capget(cap_user_header_t hdrp, cap_user_data_t datap); + int do_exit = 0; + + /* check for CAP_SYS_RAWIO */ + cap_header->pid = getpid(); + cap_header->version = _LINUX_CAPABILITY_VERSION; + if (capget(cap_header, cap_data) < 0) + err(-6, "capget(2) failed"); + + if ((cap_data->effective & (1 << CAP_SYS_RAWIO)) == 0) { + do_exit++; + warnx("capget(CAP_SYS_RAWIO) failed," + " try \"# setcap cap_sys_rawio=ep %s\"", progname); + } + + /* test file permissions */ + if (euidaccess("/dev/cpu/0/msr", R_OK)) { + do_exit++; + warn("/dev/cpu/0/msr open failed, try chown or chmod +r /dev/cpu/*/msr"); + } + + /* if all else fails, thell them to be root */ + if (do_exit) + if (getuid() != 0) + warnx("Or simply run as root"); + + if (do_exit) + exit(-6); } int has_nehalem_turbo_ratio_limit(unsigned int family, unsigned int model) @@ -2299,10 +2329,9 @@ void setup_all_buffers(void) void turbostat_init() { - check_cpuid(); - check_dev_msr(); - check_super_user(); + check_permissions(); + check_cpuid(); setup_all_buffers(); @@ -2441,7 +2470,7 @@ int main(int argc, char **argv) cmdline(argc, argv); if (verbose) - fprintf(stderr, "turbostat v3.7 Feb 6, 2014" + fprintf(stderr, "turbostat v3.8 14-Aug 2014" " - Len Brown \n"); turbostat_init(); From 3a9a941d0b9361eac81fb763a89fb465f70b1c28 Mon Sep 17 00:00:00 2001 From: Len Brown Date: Fri, 15 Aug 2014 02:39:52 -0400 Subject: [PATCH 3/5] tools/power turbostat: decode MSR_*_PERF_LIMIT_REASONS The Processor generation code-named Haswell added MSR_{CORE | GFX | RING}_PERF_LIMIT_REASONS to explain when and how the processor limits frequency. turbostat -v will now decode these bits. Each MSR has an "Active" set of bits which describe current conditions, and a "Logged" set of bits, which describe what has happened since last cleared. Turbostat currently doesn't clear the log bits. Signed-off-by: Len Brown --- arch/x86/include/uapi/asm/msr-index.h | 4 + tools/power/x86/turbostat/turbostat.c | 126 ++++++++++++++++++++++++++ 2 files changed, 130 insertions(+) diff --git a/arch/x86/include/uapi/asm/msr-index.h b/arch/x86/include/uapi/asm/msr-index.h index c8aa65d56027..f721330541cb 100644 --- a/arch/x86/include/uapi/asm/msr-index.h +++ b/arch/x86/include/uapi/asm/msr-index.h @@ -152,6 +152,10 @@ #define MSR_CC6_DEMOTION_POLICY_CONFIG 0x00000668 #define MSR_MC6_DEMOTION_POLICY_CONFIG 0x00000669 +#define MSR_CORE_PERF_LIMIT_REASONS 0x00000690 +#define MSR_GFX_PERF_LIMIT_REASONS 0x000006B0 +#define MSR_RING_PERF_LIMIT_REASONS 0x000006B1 + /* Hardware P state interface */ #define MSR_PPERF 0x0000064e #define MSR_PERF_LIMIT_REASONS 0x0000064f diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 6f29fc11fde6..58913096d7b2 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -83,6 +83,9 @@ unsigned int tcc_activation_temp; unsigned int tcc_activation_temp_override; double rapl_power_units, rapl_energy_units, rapl_time_units; double rapl_joule_counter_range; +unsigned int do_core_perf_limit_reasons; +unsigned int do_gfx_perf_limit_reasons; +unsigned int do_ring_perf_limit_reasons; #define RAPL_PKG (1 << 0) /* 0x610 MSR_PKG_POWER_LIMIT */ @@ -1178,6 +1181,7 @@ void print_verbose_header(void) if (ratio) fprintf(stderr, "%d * %.0f = %.0f MHz max turbo 1 active cores\n", ratio, bclk, ratio * bclk); + } void free_all_buffers(void) @@ -1594,6 +1598,103 @@ int print_epb(struct thread_data *t, struct core_data *c, struct pkg_data *p) return 0; } +/* + * print_perf_limit() + */ +int print_perf_limit(struct thread_data *t, struct core_data *c, struct pkg_data *p) +{ + unsigned long long msr; + int cpu; + + cpu = t->cpu_id; + + /* per-package */ + if (!(t->flags & CPU_IS_FIRST_THREAD_IN_CORE) || !(t->flags & CPU_IS_FIRST_CORE_IN_PACKAGE)) + return 0; + + if (cpu_migrate(cpu)) { + fprintf(stderr, "Could not migrate to CPU %d\n", cpu); + return -1; + } + + if (do_core_perf_limit_reasons) { + get_msr(cpu, MSR_CORE_PERF_LIMIT_REASONS, &msr); + fprintf(stderr, "cpu%d: MSR_CORE_PERF_LIMIT_REASONS, 0x%08llx", cpu, msr); + fprintf(stderr, " (Active: %s%s%s%s%s%s%s%s%s%s%s%s%s%s)", + (msr & 1 << 0) ? "PROCHOT, " : "", + (msr & 1 << 1) ? "ThermStatus, " : "", + (msr & 1 << 2) ? "bit2, " : "", + (msr & 1 << 4) ? "Graphics, " : "", + (msr & 1 << 5) ? "Auto-HWP, " : "", + (msr & 1 << 6) ? "VR-Therm, " : "", + (msr & 1 << 8) ? "Amps, " : "", + (msr & 1 << 9) ? "CorePwr, " : "", + (msr & 1 << 10) ? "PkgPwrL1, " : "", + (msr & 1 << 11) ? "PkgPwrL2, " : "", + (msr & 1 << 12) ? "MultiCoreTurbo, " : "", + (msr & 1 << 13) ? "Transitions, " : "", + (msr & 1 << 14) ? "bit14, " : "", + (msr & 1 << 15) ? "bit15, " : ""); + fprintf(stderr, " (Logged: %s%s%s%s%s%s%s%s%s%s%s%s%s%s)\n", + (msr & 1 << 16) ? "PROCHOT, " : "", + (msr & 1 << 17) ? "ThermStatus, " : "", + (msr & 1 << 18) ? "bit18, " : "", + (msr & 1 << 20) ? "Graphics, " : "", + (msr & 1 << 21) ? "Auto-HWP, " : "", + (msr & 1 << 22) ? "VR-Therm, " : "", + (msr & 1 << 24) ? "Amps, " : "", + (msr & 1 << 25) ? "CorePwr, " : "", + (msr & 1 << 26) ? "PkgPwrL1, " : "", + (msr & 1 << 27) ? "PkgPwrL2, " : "", + (msr & 1 << 28) ? "MultiCoreTurbo, " : "", + (msr & 1 << 29) ? "Transitions, " : "", + (msr & 1 << 30) ? "bit30, " : "", + (msr & 1 << 31) ? "bit31, " : ""); + + } + if (do_gfx_perf_limit_reasons) { + get_msr(cpu, MSR_GFX_PERF_LIMIT_REASONS, &msr); + fprintf(stderr, "cpu%d: MSR_GFX_PERF_LIMIT_REASONS, 0x%08llx", cpu, msr); + fprintf(stderr, " (Active: %s%s%s%s%s%s%s%s)", + (msr & 1 << 0) ? "PROCHOT, " : "", + (msr & 1 << 1) ? "ThermStatus, " : "", + (msr & 1 << 4) ? "Graphics, " : "", + (msr & 1 << 6) ? "VR-Therm, " : "", + (msr & 1 << 8) ? "Amps, " : "", + (msr & 1 << 9) ? "GFXPwr, " : "", + (msr & 1 << 10) ? "PkgPwrL1, " : "", + (msr & 1 << 11) ? "PkgPwrL2, " : ""); + fprintf(stderr, " (Logged: %s%s%s%s%s%s%s%s)\n", + (msr & 1 << 16) ? "PROCHOT, " : "", + (msr & 1 << 17) ? "ThermStatus, " : "", + (msr & 1 << 20) ? "Graphics, " : "", + (msr & 1 << 22) ? "VR-Therm, " : "", + (msr & 1 << 24) ? "Amps, " : "", + (msr & 1 << 25) ? "GFXPwr, " : "", + (msr & 1 << 26) ? "PkgPwrL1, " : "", + (msr & 1 << 27) ? "PkgPwrL2, " : ""); + } + if (do_ring_perf_limit_reasons) { + get_msr(cpu, MSR_RING_PERF_LIMIT_REASONS, &msr); + fprintf(stderr, "cpu%d: MSR_RING_PERF_LIMIT_REASONS, 0x%08llx", cpu, msr); + fprintf(stderr, " (Active: %s%s%s%s%s%s)", + (msr & 1 << 0) ? "PROCHOT, " : "", + (msr & 1 << 1) ? "ThermStatus, " : "", + (msr & 1 << 6) ? "VR-Therm, " : "", + (msr & 1 << 8) ? "Amps, " : "", + (msr & 1 << 10) ? "PkgPwrL1, " : "", + (msr & 1 << 11) ? "PkgPwrL2, " : ""); + fprintf(stderr, " (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, " : ""); + } + return 0; +} + #define RAPL_POWER_GRANULARITY 0x7FFF /* 15 bit power granularity */ #define RAPL_TIME_GRANULARITY 0x3F /* 6 bit time granularity */ @@ -1683,6 +1784,27 @@ void rapl_probe(unsigned int family, unsigned int model) return; } +void perf_limit_reasons_probe(family, model) +{ + if (!genuine_intel) + return; + + if (family != 6) + return; + + switch (model) { + case 0x3C: /* HSW */ + case 0x45: /* HSW */ + case 0x46: /* HSW */ + do_gfx_perf_limit_reasons = 1; + case 0x3F: /* HSX */ + do_core_perf_limit_reasons = 1; + do_ring_perf_limit_reasons = 1; + default: + return; + } +} + int print_thermal(struct thread_data *t, struct core_data *c, struct pkg_data *p) { unsigned long long msr; @@ -2104,6 +2226,7 @@ void check_cpuid() do_nehalem_turbo_ratio_limit = has_nehalem_turbo_ratio_limit(family, model); do_ivt_turbo_ratio_limit = has_ivt_turbo_ratio_limit(family, model); rapl_probe(family, model); + perf_limit_reasons_probe(family, model); return; } @@ -2341,6 +2464,9 @@ void turbostat_init() if (verbose) for_all_cpus(print_epb, ODD_COUNTERS); + if (verbose) + for_all_cpus(print_perf_limit, ODD_COUNTERS); + if (verbose) for_all_cpus(print_rapl, ODD_COUNTERS); From d7899447535929b3672442b7b42a09ae4e48fa91 Mon Sep 17 00:00:00 2001 From: Len Brown Date: Fri, 23 Jan 2015 00:12:33 -0500 Subject: [PATCH 4/5] tools/power turbostat: relax dependency on invariant TSC Turbostat can be useful on systems that do not support invariant TSC, so allow it to run on those systgems. All arithmetic in turbostat using the TSC value is per-processsor, so it does not depend on the TSC values being in sync acrosss processors. Turbostat uses gettimeofday() for the measurement interval rather than using the TSC directly, so that key metric is also immune from variable TSC. Turbostat prints a TSC sanity check column: TSC_MHz = TSC_delta/interval If this column is constant and is close to the processor base frequency, then the TSC is behaving properly. The other key turbostat columns are calculated this way: Avg_Mhz = APERF_delta/interval %Busy = MPERF_delta/TSC_delta Bzy_MHz = TSC_delta/APERF_delta/MPERF_delta/interval Tested on Core2 and Core2-Xeon, and so this patch includes a few other changes to remove the assumption that target systems are Nehalem and newer. Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 114 +++++++++++++++++--------- 1 file changed, 75 insertions(+), 39 deletions(-) diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 58913096d7b2..b654b641d4c3 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -61,8 +61,8 @@ unsigned int has_epb; unsigned int units = 1000000; /* MHz etc */ unsigned int genuine_intel; unsigned int has_invariant_tsc; -unsigned int do_nehalem_platform_info; -unsigned int do_nehalem_turbo_ratio_limit; +unsigned int do_nhm_platform_info; +unsigned int do_nhm_turbo_ratio_limit; unsigned int do_ivt_turbo_ratio_limit; unsigned int extra_msr_offset32; unsigned int extra_msr_offset64; @@ -284,7 +284,7 @@ void print_header(void) outp += sprintf(outp, " CPU"); if (has_aperf) outp += sprintf(outp, " Avg_MHz"); - if (do_nhm_cstates) + if (has_aperf) outp += sprintf(outp, " %%Busy"); if (has_aperf) outp += sprintf(outp, " Bzy_MHz"); @@ -340,7 +340,7 @@ void print_header(void) outp += sprintf(outp, " PKG_%%"); if (do_rapl & RAPL_DRAM_PERF_STATUS) outp += sprintf(outp, " RAM_%%"); - } else { + } else if (do_rapl && rapl_joules) { if (do_rapl & RAPL_PKG) outp += sprintf(outp, " Pkg_J"); if (do_rapl & RAPL_CORES) @@ -460,25 +460,25 @@ int format_counters(struct thread_data *t, struct core_data *c, outp += sprintf(outp, "%8d", t->cpu_id); } - /* AvgMHz */ + /* Avg_MHz */ if (has_aperf) outp += sprintf(outp, "%8.0f", 1.0 / units * t->aperf / interval_float); - /* %c0 */ - if (do_nhm_cstates) { + /* %Busy */ + if (has_aperf) { if (!skip_c0) outp += sprintf(outp, "%8.2f", 100.0 * t->mperf/t->tsc); else outp += sprintf(outp, "********"); } - /* BzyMHz */ + /* Bzy_MHz */ if (has_aperf) outp += sprintf(outp, "%8.0f", 1.0 * t->tsc / units * t->aperf / t->mperf / interval_float); - /* TSC */ + /* TSC_MHz */ outp += sprintf(outp, "%8.0f", 1.0 * t->tsc/units/interval_float); /* SMI */ @@ -564,7 +564,7 @@ int format_counters(struct thread_data *t, struct core_data *c, outp += sprintf(outp, fmt8, 100.0 * p->rapl_pkg_perf_status * rapl_time_units / interval_float); if (do_rapl & RAPL_DRAM_PERF_STATUS) outp += sprintf(outp, fmt8, 100.0 * p->rapl_dram_perf_status * rapl_time_units / interval_float); - } else { + } else if (do_rapl && rapl_joules) { if (do_rapl & RAPL_PKG) outp += sprintf(outp, fmt8, p->energy_pkg * rapl_energy_units); @@ -581,8 +581,8 @@ int format_counters(struct thread_data *t, struct core_data *c, outp += sprintf(outp, fmt8, 100.0 * p->rapl_pkg_perf_status * rapl_time_units / interval_float); if (do_rapl & RAPL_DRAM_PERF_STATUS) outp += sprintf(outp, fmt8, 100.0 * p->rapl_dram_perf_status * rapl_time_units / interval_float); - outp += sprintf(outp, fmt8, interval_float); + outp += sprintf(outp, fmt8, interval_float); } done: outp += sprintf(outp, "\n"); @@ -1022,7 +1022,7 @@ void print_verbose_header(void) unsigned long long msr; unsigned int ratio; - if (!do_nehalem_platform_info) + if (!do_nhm_platform_info) return; get_msr(0, MSR_NHM_PLATFORM_INFO, &msr); @@ -1135,7 +1135,7 @@ void print_verbose_header(void) } fprintf(stderr, ")\n"); - if (!do_nehalem_turbo_ratio_limit) + if (!do_nhm_turbo_ratio_limit) return; get_msr(0, MSR_NHM_TURBO_RATIO_LIMIT, &msr); @@ -1462,8 +1462,7 @@ void check_dev_msr() struct stat sb; if (stat("/dev/cpu/0/msr", &sb)) - err(-5, "no /dev/cpu/0/msr\n" - "Try \"# modprobe msr\""); + err(-5, "no /dev/cpu/0/msr, Try \"# modprobe msr\" "); } void check_permissions() @@ -1496,13 +1495,27 @@ void check_permissions() /* if all else fails, thell them to be root */ if (do_exit) if (getuid() != 0) - warnx("Or simply run as root"); + warnx("... or simply run as root"); if (do_exit) exit(-6); } -int has_nehalem_turbo_ratio_limit(unsigned int family, unsigned int model) +/* + * NHM adds support for additional MSRs: + * + * MSR_SMI_COUNT 0x00000034 + * + * MSR_NHM_PLATFORM_INFO 0x000000ce + * MSR_NHM_SNB_PKG_CST_CFG_CTL 0x000000e2 + * + * MSR_PKG_C3_RESIDENCY 0x000003f8 + * MSR_PKG_C6_RESIDENCY 0x000003f9 + * MSR_CORE_C3_RESIDENCY 0x000003fc + * MSR_CORE_C6_RESIDENCY 0x000003fd + * + */ +int has_nhm_msrs(unsigned int family, unsigned int model) { if (!genuine_intel) return 0; @@ -1529,13 +1542,27 @@ int has_nehalem_turbo_ratio_limit(unsigned int family, unsigned int model) case 0x3D: /* BDW */ case 0x4F: /* BDX */ case 0x56: /* BDX-DE */ - return 1; case 0x2E: /* Nehalem-EX Xeon - Beckton */ case 0x2F: /* Westmere-EX Xeon - Eagleton */ + return 1; default: return 0; } } +int has_nhm_turbo_ratio_limit(unsigned int family, unsigned int model) +{ + if (!has_nhm_msrs(family, model)) + return 0; + + switch (model) { + /* Nehalem compatible, but do not include turbo-ratio limit support */ + case 0x2E: /* Nehalem-EX Xeon - Beckton */ + case 0x2F: /* Westmere-EX Xeon - Eagleton */ + return 0; + default: + return 1; + } +} int has_ivt_turbo_ratio_limit(unsigned int family, unsigned int model) { if (!genuine_intel) @@ -1994,8 +2021,15 @@ int print_rapl(struct thread_data *t, struct core_data *c, struct pkg_data *p) return 0; } +/* + * SNB adds support for additional MSRs: + * + * MSR_PKG_C7_RESIDENCY 0x000003fa + * MSR_CORE_C7_RESIDENCY 0x000003fe + * MSR_PKG_C2_RESIDENCY 0x0000060d + */ -int is_snb(unsigned int family, unsigned int model) +int has_snb_msrs(unsigned int family, unsigned int model) { if (!genuine_intel) return 0; @@ -2017,7 +2051,14 @@ int is_snb(unsigned int family, unsigned int model) return 0; } -int has_c8_c9_c10(unsigned int family, unsigned int model) +/* + * HSW adds support for additional MSRs: + * + * MSR_PKG_C8_RESIDENCY 0x00000630 + * MSR_PKG_C9_RESIDENCY 0x00000631 + * MSR_PKG_C10_RESIDENCY 0x00000632 + */ +int has_hsw_msrs(unsigned int family, unsigned int model) { if (!genuine_intel) return 0; @@ -2069,7 +2110,7 @@ double slm_bclk(void) double discover_bclk(unsigned int family, unsigned int model) { - if (is_snb(family, model)) + if (has_snb_msrs(family, model)) return 100.00; else if (is_slm(family, model)) return slm_bclk(); @@ -2117,7 +2158,7 @@ int set_temperature_target(struct thread_data *t, struct core_data *c, struct pk } /* Temperature Target MSR is Nehalem and newer only */ - if (!do_nehalem_platform_info) + if (!do_nhm_platform_info) goto guess; if (get_msr(0, MSR_IA32_TEMPERATURE_TARGET, &msr)) @@ -2181,18 +2222,15 @@ void check_cpuid() ebx = ecx = edx = 0; __get_cpuid(0x80000000, &max_level, &ebx, &ecx, &edx); - if (max_level < 0x80000007) - errx(1, "CPUID: no invariant TSC (max_level 0x%x)", max_level); + if (max_level >= 0x80000007) { - /* - * Non-Stop TSC is advertised by CPUID.EAX=0x80000007: EDX.bit8 - * this check is valid for both Intel and AMD - */ - __get_cpuid(0x80000007, &eax, &ebx, &ecx, &edx); - has_invariant_tsc = edx & (1 << 8); - - if (!has_invariant_tsc) - errx(1, "No invariant TSC"); + /* + * Non-Stop TSC is advertised by CPUID.EAX=0x80000007: EDX.bit8 + * this check is valid for both Intel and AMD + */ + __get_cpuid(0x80000007, &eax, &ebx, &ecx, &edx); + has_invariant_tsc = edx & (1 << 8); + } /* * APERF/MPERF is advertised by CPUID.EAX=0x6: ECX.bit0 @@ -2215,15 +2253,13 @@ void check_cpuid() if (!has_aperf) errx(-1, "No APERF"); - do_nehalem_platform_info = genuine_intel && has_invariant_tsc; - do_nhm_cstates = genuine_intel; /* all Intel w/ non-stop TSC have NHM counters */ - do_smi = do_nhm_cstates; - do_snb_cstates = is_snb(family, model); - do_c8_c9_c10 = has_c8_c9_c10(family, model); + do_nhm_platform_info = do_nhm_cstates = do_smi = has_nhm_msrs(family, model); + do_snb_cstates = has_snb_msrs(family, model); + do_c8_c9_c10 = has_hsw_msrs(family, model); do_slm_cstates = is_slm(family, model); bclk = discover_bclk(family, model); - do_nehalem_turbo_ratio_limit = has_nehalem_turbo_ratio_limit(family, model); + do_nhm_turbo_ratio_limit = has_nhm_turbo_ratio_limit(family, model); do_ivt_turbo_ratio_limit = has_ivt_turbo_ratio_limit(family, model); rapl_probe(family, model); perf_limit_reasons_probe(family, model); From a729617c58529be0be8faa22c5d45748bb0f12e5 Mon Sep 17 00:00:00 2001 From: Len Brown Date: Fri, 23 Jan 2015 01:33:58 -0500 Subject: [PATCH 5/5] tools/power turbostat: relax dependency on APERF_MSR While turbostat is significantly less useful on systems with no APERF_MSR, it seems more friendly to run on such systems and report what we can, rather than refusing to run. Update man page to reflect recent changes. Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.8 | 64 ++++++++++++++++++--------- tools/power/x86/turbostat/turbostat.c | 47 ++++++++++---------- 2 files changed, 67 insertions(+), 44 deletions(-) diff --git a/tools/power/x86/turbostat/turbostat.8 b/tools/power/x86/turbostat/turbostat.8 index 56bfb523c5bb..9b950699e63d 100644 --- a/tools/power/x86/turbostat/turbostat.8 +++ b/tools/power/x86/turbostat/turbostat.8 @@ -12,16 +12,16 @@ turbostat \- Report processor frequency and idle statistics .RB [ "\-i interval_sec" ] .SH DESCRIPTION \fBturbostat \fP reports processor topology, frequency, -idle power-state statistics, temperature and power on modern X86 processors. -Either \fBcommand\fP is forked and statistics are printed -upon its completion, or statistics are printed periodically. - -\fBturbostat \fP -must be run on root, and -minimally requires that the processor -supports an "invariant" TSC, plus the APERF and MPERF MSRs. -Additional information is reported depending on hardware counter support. +idle power-state statistics, temperature and power on X86 processors. +There are two ways to invoke turbostat. +The first method is to supply a +\fBcommand\fP, which is forked and statistics are printed +upon its completion. +The second method is to omit the command, +and turbodstat will print statistics every 5 seconds. +The 5-second interval can changed using the -i option. +Some information is not availalbe on older processors. .SS Options The \fB-p\fP option limits output to the 1st thread in 1st core of each package. .PP @@ -130,12 +130,13 @@ cpu3: MSR_IA32_THERM_STATUS: 0x884e0000 (27 C +/- 1) ... .fi The \fBmax efficiency\fP frequency, a.k.a. Low Frequency Mode, is the frequency -available at the minimum package voltage. The \fBTSC frequency\fP is the nominal -maximum frequency of the processor if turbo-mode were not available. This frequency +available at the minimum package voltage. The \fBTSC frequency\fP is the base +frequency of the processor -- this should match the brand string +in /proc/cpuinfo. This base frequency should be sustainable on all CPUs indefinitely, given nominal power and cooling. The remaining rows show what maximum turbo frequency is possible -depending on the number of idle cores. Note that this information is -not available on all processors. +depending on the number of idle cores. Note that not all information is +available on all processors. .SH FORK EXAMPLE If turbostat is invoked with a command, it will fork that command and output the statistics gathered when the command exits. @@ -176,6 +177,11 @@ not including any non-busy idle time. .B "turbostat " must be run as root. +Alternatively, non-root users can be enabled to run turbostat this way: + +# setcap cap_sys_rawio=ep ./turbostat + +# chmod +r /dev/cpu/*/msr .B "turbostat " reads hardware counters, but doesn't write them. @@ -184,15 +190,33 @@ multiple invocations of itself. \fBturbostat \fP may work poorly on Linux-2.6.20 through 2.6.29, -as \fBacpi-cpufreq \fPperiodically cleared the APERF and MPERF +as \fBacpi-cpufreq \fPperiodically cleared the APERF and MPERF MSRs in those kernels. -If the TSC column does not make sense, then -the other numbers will also make no sense. -Turbostat is lightweight, and its data collection is not atomic. -These issues are usually caused by an extremely short measurement -interval (much less than 1 second), or system activity that prevents -turbostat from being able to run on all CPUS to quickly collect data. +AVG_MHz = APERF_delta/measurement_interval. This is the actual +number of elapsed cycles divided by the entire sample interval -- +including idle time. Note that this calculation is resiliant +to systems lacking a non-stop TSC. + +TSC_MHz = TSC_delta/measurement_interval. +On a system with an invariant TSC, this value will be constant +and will closely match the base frequency value shown +in the brand string in /proc/cpuinfo. On a system where +the TSC stops in idle, TSC_MHz will drop +below the processor's base frequency. + +%Busy = MPERF_delta/TSC_delta + +Bzy_MHz = TSC_delta/APERF_delta/MPERF_delta/measurement_interval + +Note that these calculations depend on TSC_delta, so they +are not reliable during intervals when TSC_MHz is not running at the base frequency. + +Turbostat data collection is not atomic. +Extremely short measurement intervals (much less than 1 second), +or system activity that prevents turbostat from being able +to run on all CPUS to quickly collect data, will result in +inconsistent results. The APERF, MPERF MSRs are defined to count non-halted cycles. Although it is not guaranteed by the architecture, turbostat assumes diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index b654b641d4c3..a02c02f25e88 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -673,24 +673,26 @@ delta_thread(struct thread_data *new, struct thread_data *old, old->c1 = new->c1 - old->c1; - if ((new->aperf > old->aperf) && (new->mperf > old->mperf)) { - old->aperf = new->aperf - old->aperf; - old->mperf = new->mperf - old->mperf; - } else { + if (has_aperf) { + if ((new->aperf > old->aperf) && (new->mperf > old->mperf)) { + old->aperf = new->aperf - old->aperf; + old->mperf = new->mperf - old->mperf; + } else { - if (!aperf_mperf_unstable) { - fprintf(stderr, "%s: APERF or MPERF went backwards *\n", progname); - fprintf(stderr, "* Frequency results do not cover entire interval *\n"); - fprintf(stderr, "* fix this by running Linux-2.6.30 or later *\n"); + if (!aperf_mperf_unstable) { + fprintf(stderr, "%s: APERF or MPERF went backwards *\n", progname); + fprintf(stderr, "* Frequency results do not cover entire interval *\n"); + fprintf(stderr, "* fix this by running Linux-2.6.30 or later *\n"); - aperf_mperf_unstable = 1; + aperf_mperf_unstable = 1; + } + /* + * mperf delta is likely a huge "positive" number + * can not use it for calculating c0 time + */ + skip_c0 = 1; + skip_c1 = 1; } - /* - * mperf delta is likely a huge "positive" number - * can not use it for calculating c0 time - */ - skip_c0 = 1; - skip_c1 = 1; } @@ -2244,14 +2246,11 @@ void check_cpuid() has_epb = ecx & (1 << 3); if (verbose) - fprintf(stderr, "CPUID(6): %s%s%s%s\n", - has_aperf ? "APERF" : "No APERF!", - do_dts ? ", DTS" : "", - do_ptm ? ", PTM": "", - has_epb ? ", EPB": ""); - - if (!has_aperf) - errx(-1, "No APERF"); + fprintf(stderr, "CPUID(6): %sAPERF, %sDTS, %sPTM, %sEPB\n", + has_aperf ? "" : "No ", + do_dts ? "" : "No ", + do_ptm ? "" : "No ", + has_epb ? "" : "No "); do_nhm_platform_info = do_nhm_cstates = do_smi = has_nhm_msrs(family, model); do_snb_cstates = has_snb_msrs(family, model); @@ -2632,7 +2631,7 @@ int main(int argc, char **argv) cmdline(argc, argv); if (verbose) - fprintf(stderr, "turbostat v3.8 14-Aug 2014" + fprintf(stderr, "turbostat v3.9 23-Jan, 2015" " - Len Brown \n"); turbostat_init();