turbostat for 4.10 merge

PATCH New Len Brown 1 participant 18 messages Dec 24, 2016 Turbostat utility

lore

Share & tools

lore

Len Brown

author maintainer

2016-12-24 20:56 UTC #0 lore raw

cover turbostat for 4.10 merge

These patches bring the upstream tree up to date with the version of turbostat that I am currently using.

Let me know if you have any troubles, or suggestions.

thanks,
-Len Brown, Intel Open Source Technology Center

Len Brown

author maintainer

2016-12-24 20:56 UTC #1 lore raw

PATCH 1/17 tools/power turbostat: tidy up output on Joule counter overflow

From: Len Brown <[email protected]>

The RAPL Joules counter is limited in capacity.
Turbostat estimates how soon it can roll-over
based on the max TDP of the processor --
which tells us the maximum increment rate.

eg.
RAPL: 2759 sec. Joule Counter Range, at 95 Watts

So if a sample duration is longer than 2759 seconds on this system, '**' replace the decimal place in the display to indicate that the results may be suspect.

But the display had an extra ' ' in this case, throwing off the columns.

Also, the -J "Joules" option appended an extra "time" column to the display. While this may be useful, it printed the interval time, which may not be the accurate time per processor. Remove this column, which appeared only when using '-J',
as we plan to add accurate per-cpu interval times in a future commit.

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

tools/power/x86/turbostat/turbostat.c +1 -5

index 3e199b508a96..a302a333c084 100644
@@ -423,8 +423,6 @@ void print_header(void)
 			outp += sprintf(outp, "   PKG_%%");
 		if (do_rapl & RAPL_DRAM_PERF_STATUS)
 			outp += sprintf(outp, "   RAM_%%");
-		outp += sprintf(outp, "   time");
-
 	}
     done:
 	outp += sprintf(outp, "\n");

@@ -665,7 +663,7 @@ int format_counters(struct thread_data *t, struct core_data *c,
 	if (interval_float < rapl_joule_counter_range)
 		fmt8 = "%8.2f";
 	else
-		fmt8 = " %6.0f**";
+		fmt8 = "%6.0f**";
 
 	if (do_rapl && !rapl_joules) {
 		if (do_rapl & RAPL_PKG)

@@ -697,8 +695,6 @@ 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);
 	}
 done:
 	outp += sprintf(outp, "\n");

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #2 lore raw

PATCH 2/17 tools/power turbostat: Gracefully handle ACPI S3

In reply to Len Brown

From: Len Brown <[email protected]>

turbostat gives valid results across suspend to idle, aka freeze, whether invoked in interval mode, or in command mode. Indeed, this can be used to measure suspend to idle:

turbostat echo freeze > /sys/power/state

But this does not work across suspend to ACPI S3, because the processor counters, including the TSC, are reset on resume. Further, when turbostat detects a problem, it does't forgive the hardware, and interval mode will print *'s from there on out.

Instead, upon detecting counters going backwards, simply reset and start over.

Interval mode across ACPI S3: (observe TSC going backwards)

root@sharkbay:/home/lenb/turbostat-src# ./turbostat -M 0x10
     CPU Avg_MHz   Busy% Bzy_MHz TSC_MHz           MSR 0x010
       -       1    0.06     858    2294  0x0000000000000000
       0       0    0.06     847    2294  0x0000002a254b98ac
       1       1    0.06     878    2294  0x0000002a254efa3a
       2       1    0.07     843    2294  0x0000002a2551df65
       3       0    0.05     863    2294  0x0000002a2553fea2
turbostat: re-initialized with num_cpus 4
     CPU Avg_MHz   Busy% Bzy_MHz TSC_MHz           MSR 0x010
       -       2    0.20     849    2294  0x0000000000000000
       0       2    0.26     856    2294  0x0000000449abb60d
       1       2    0.20     844    2294  0x0000000449b087ec
       2       2    0.21     850    2294  0x0000000449b35d5d
       3       1    0.12     839    2294  0x0000000449b5fd5a
^C

Command mode across ACPI S3:

root@sharkbay:/home/lenb/turbostat-src# ./turbostat -M 0x10 sleep 10
./turbostat: Counter reset detected
14.196299 sec

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 77 ++++++++++++++++-------------------
 1 file changed, 35 insertions(+), 42 deletions(-)

tools/power/x86/turbostat/turbostat.c +35 ~~-42~~

index a302a333c084..c0dade0c1921 100644
@@ -51,8 +51,6 @@ unsigned int debug;
 unsigned int rapl_joules;
 unsigned int summary_only;
 unsigned int dump_only;
-unsigned int skip_c0;
-unsigned int skip_c1;
 unsigned int do_nhm_cstates;
 unsigned int do_snb_cstates;
 unsigned int do_knl_cstates;

@@ -145,7 +143,6 @@ unsigned int has_hwp_pkg;		/* IA32_HWP_REQUEST_PKG */
 
 #define MAX(a, b) ((a) > (b) ? (a) : (b))
 
-int aperf_mperf_unstable;
 int backwards_count;
 char *progname;

@@ -546,12 +543,8 @@ int format_counters(struct thread_data *t, struct core_data *c,
 			1.0 / units * t->aperf / interval_float);
 
 	/* Busy% */
-	if (has_aperf) {
-		if (!skip_c0)
-			outp += sprintf(outp, "%8.2f", 100.0 * t->mperf/t->tsc/tsc_tweak);
-		else
-			outp += sprintf(outp, "********");
-	}
+	if (has_aperf)
+		outp += sprintf(outp, "%8.2f", 100.0 * t->mperf/t->tsc/tsc_tweak);
 
 	/* Bzy_MHz */
 	if (has_aperf) {

@@ -591,12 +584,8 @@ int format_counters(struct thread_data *t, struct core_data *c,
 	if (do_smi)
 		outp += sprintf(outp, "%8d", t->smi_count);
 
-	if (do_nhm_cstates) {
-		if (!skip_c1)
-			outp += sprintf(outp, "%8.2f", 100.0 * t->c1/t->tsc);
-		else
-			outp += sprintf(outp, "********");
-	}
+	if (do_nhm_cstates)
+		outp += sprintf(outp, "%8.2f", 100.0 * t->c1/t->tsc);
 
 	/* print per-core data only for 1st thread in core */
 	if (!(t->flags & CPU_IS_FIRST_THREAD_IN_CORE))

@@ -622,8 +611,8 @@ int format_counters(struct thread_data *t, struct core_data *c,
 
 	/* GFXrc6 */
 	if (do_gfx_rc6_ms) {
-		if (p->gfx_rc6_ms == -1) {	/* detect counter reset */
-			outp += sprintf(outp, "  ***.**");
+		if (p->gfx_rc6_ms == -1) {	/* detect GFX counter reset */
+			outp += sprintf(outp, "   **.**");
 		} else {
 			outp += sprintf(outp, "%8.2f",
 				p->gfx_rc6_ms / 10.0 / interval_float);

@@ -748,7 +737,7 @@ void format_all_counters(struct thread_data *t, struct core_data *c, struct pkg_
 		old = 0x100000000 + new - old;	\
 	}
 
-void
+int
 delta_package(struct pkg_data *new, struct pkg_data *old)
 {

@@ -784,6 +773,8 @@ delta_package(struct pkg_data *new, struct pkg_data *old)
 	DELTA_WRAP32(new->energy_dram, old->energy_dram);
 	DELTA_WRAP32(new->rapl_pkg_perf_status, old->rapl_pkg_perf_status);
 	DELTA_WRAP32(new->rapl_dram_perf_status, old->rapl_dram_perf_status);
+
+	return 0;
 }
 
 void

@@ -798,7 +789,7 @@ delta_core(struct core_data *new, struct core_data *old)
 /*
  * old = new - old
  */
-void
+int
 delta_thread(struct thread_data *new, struct thread_data *old,
 	struct core_data *core_delta)
 {

@@ -817,20 +808,7 @@ delta_thread(struct thread_data *new, struct thread_data *old,
 			old->aperf = new->aperf - old->aperf;
 			old->mperf = new->mperf - old->mperf;
 		} else {
-
-			if (!aperf_mperf_unstable) {
-				fprintf(outf, "%s: APERF or MPERF went backwards *\n", progname);
-				fprintf(outf, "* Frequency results do not cover entire interval *\n");
-				fprintf(outf, "* fix this by running Linux-2.6.30 or later *\n");
-
-				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;
+			return -1;
 		}
 	}

@@ -877,24 +855,30 @@ delta_thread(struct thread_data *new, struct thread_data *old,
 
 	if (do_smi)
 		old->smi_count = new->smi_count - old->smi_count;
+
+	return 0;
 }
 
 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)
 {
+	int retval = 0;
+
 	/* calculate core delta only for 1st thread in core */
 	if (t->flags & CPU_IS_FIRST_THREAD_IN_CORE)
 		delta_core(c, c2);
 
 	/* always calculate thread delta */
-	delta_thread(t, t2, c2);	/* c2 is core delta */
+	retval = delta_thread(t, t2, c2);	/* c2 is core delta */
+	if (retval)
+		return retval;
 
 	/* calculate package delta only for 1st core in package */
 	if (t->flags & CPU_IS_FIRST_CORE_IN_PACKAGE)
-		delta_package(p, p2);
+		retval = delta_package(p, p2);
 
-	return 0;
+	return retval;
 }
 
 void clear_counters(struct thread_data *t, struct core_data *c, struct pkg_data *p)

@@ -2067,7 +2051,10 @@ void turbostat_loop()
 		}
 		gettimeofday(&tv_odd, (struct timezone *)NULL);
 		timersub(&tv_odd, &tv_even, &tv_delta);
-		for_all_cpus_2(delta_cpu, ODD_COUNTERS, EVEN_COUNTERS);
+		if (for_all_cpus_2(delta_cpu, ODD_COUNTERS, EVEN_COUNTERS)) {
+			re_initialize();
+			goto restart;
+		}
 		compute_average(EVEN_COUNTERS);
 		format_all_counters(EVEN_COUNTERS);
 		flush_output_stdout();

@@ -2083,7 +2070,10 @@ void turbostat_loop()
 		}
 		gettimeofday(&tv_even, (struct timezone *)NULL);
 		timersub(&tv_even, &tv_odd, &tv_delta);
-		for_all_cpus_2(delta_cpu, EVEN_COUNTERS, ODD_COUNTERS);
+		if (for_all_cpus_2(delta_cpu, EVEN_COUNTERS, ODD_COUNTERS)) {
+			re_initialize();
+			goto restart;
+		}
 		compute_average(ODD_COUNTERS);
 		format_all_counters(ODD_COUNTERS);
 		flush_output_stdout();

@@ -3693,9 +3683,12 @@ int fork_it(char **argv)
 	for_all_cpus(get_counters, ODD_COUNTERS);
 	gettimeofday(&tv_odd, (struct timezone *)NULL);
 	timersub(&tv_odd, &tv_even, &tv_delta);
-	for_all_cpus_2(delta_cpu, ODD_COUNTERS, EVEN_COUNTERS);
-	compute_average(EVEN_COUNTERS);
-	format_all_counters(EVEN_COUNTERS);
+	if (for_all_cpus_2(delta_cpu, ODD_COUNTERS, EVEN_COUNTERS))
+		fprintf(outf, "%s: Counter reset detected\n", progname);
+	else {
+		compute_average(EVEN_COUNTERS);
+		format_all_counters(EVEN_COUNTERS);
+	}
 
 	fprintf(outf, "%.6f sec\n", tv_delta.tv_sec + tv_delta.tv_usec/1000000.0);

@@ -3722,7 +3715,7 @@ int get_and_dump_counters(void)
 }
 
 void print_version() {
-	fprintf(outf, "turbostat version 4.12 5 Apr 2016"
+	fprintf(outf, "turbostat version 4.13 22 Apr 2016"
 		" - Len Brown <[email protected]>\n");
 }

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #3 lore raw

PATCH 3/17 tools/power turbostat: switch to tab delimited output

In reply to Len Brown

From: Len Brown <[email protected]>

Switch to tab-delimited output from fixed-width columns to make it simpler to import into spreadsheets.

As the fixed width columnns were 8-spaces wide,
the output on the screen should not change.

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 170 +++++++++++++++++-----------------
 1 file changed, 85 insertions(+), 85 deletions(-)

tools/power/x86/turbostat/turbostat.c +85 ~~-85~~

index c0dade0c1921..be50892f27aa 100644
@@ -323,103 +323,103 @@ int get_msr(int cpu, off_t offset, unsigned long long *msr)
 void print_header(void)
 {
 	if (show_pkg)
-		outp += sprintf(outp, " Package");
+		outp += sprintf(outp, "\tPackage");
 	if (show_core)
-		outp += sprintf(outp, "    Core");
+		outp += sprintf(outp, "\tCore");
 	if (show_cpu)
-		outp += sprintf(outp, "     CPU");
+		outp += sprintf(outp, "\tCPU");
 	if (has_aperf)
-		outp += sprintf(outp, " Avg_MHz");
+		outp += sprintf(outp, "\tAvg_MHz");
 	if (has_aperf)
-		outp += sprintf(outp, "   Busy%%");
+		outp += sprintf(outp, "\tBusy%%");
 	if (has_aperf)
-		outp += sprintf(outp, " Bzy_MHz");
-	outp += sprintf(outp, " TSC_MHz");
+		outp += sprintf(outp, "\tBzy_MHz");
+	outp += sprintf(outp, "\tTSC_MHz");
 
 	if (extra_delta_offset32)
-		outp += sprintf(outp, "  count 0x%03X", extra_delta_offset32);
+		outp += sprintf(outp, "\tcount 0x%03X", extra_delta_offset32);
 	if (extra_delta_offset64)
-		outp += sprintf(outp, "  COUNT 0x%03X", extra_delta_offset64);
+		outp += sprintf(outp, "\tCOUNT 0x%03X", extra_delta_offset64);
 	if (extra_msr_offset32)
-		outp += sprintf(outp, "   MSR 0x%03X", extra_msr_offset32);
+		outp += sprintf(outp, "\tMSR 0x%03X", extra_msr_offset32);
 	if (extra_msr_offset64)
-		outp += sprintf(outp, "           MSR 0x%03X", extra_msr_offset64);
+		outp += sprintf(outp, "\tMSR 0x%03X", extra_msr_offset64);
 
 	if (!debug)
 		goto done;
 
 	if (do_irq)
-		outp += sprintf(outp, "     IRQ");
+		outp += sprintf(outp, "\tIRQ");
 	if (do_smi)
-		outp += sprintf(outp, "     SMI");
+		outp += sprintf(outp, "\tSMI");
 
 	if (do_nhm_cstates)
-		outp += sprintf(outp, "  CPU%%c1");
+		outp += sprintf(outp, "\tCPU%%c1");
 	if (do_nhm_cstates && !do_slm_cstates && !do_knl_cstates)
-		outp += sprintf(outp, "  CPU%%c3");
+		outp += sprintf(outp, "\tCPU%%c3");
 	if (do_nhm_cstates)
-		outp += sprintf(outp, "  CPU%%c6");
+		outp += sprintf(outp, "\tCPU%%c6");
 	if (do_snb_cstates)
-		outp += sprintf(outp, "  CPU%%c7");
+		outp += sprintf(outp, "\tCPU%%c7");
 
 	if (do_dts)
-		outp += sprintf(outp, " CoreTmp");
+		outp += sprintf(outp, "\tCoreTmp");
 	if (do_ptm)
-		outp += sprintf(outp, "  PkgTmp");
+		outp += sprintf(outp, "\tPkgTmp");
 
 	if (do_gfx_rc6_ms)
-		outp += sprintf(outp, " GFX%%rc6");
+		outp += sprintf(outp, "\tGFX%%rc6");
 
 	if (do_gfx_mhz)
-		outp += sprintf(outp, "  GFXMHz");
+		outp += sprintf(outp, "\tGFXMHz");
 
 	if (do_skl_residency) {
-		outp += sprintf(outp, " Totl%%C0");
-		outp += sprintf(outp, "  Any%%C0");
-		outp += sprintf(outp, "  GFX%%C0");
-		outp += sprintf(outp, " CPUGFX%%");
+		outp += sprintf(outp, "\tTotl%%C0");
+		outp += sprintf(outp, "\tAny%%C0");
+		outp += sprintf(outp, "\tGFX%%C0");
+		outp += sprintf(outp, "\tCPUGFX%%");
 	}
 
 	if (do_pc2)
-		outp += sprintf(outp, " Pkg%%pc2");
+		outp += sprintf(outp, "\tPkg%%pc2");
 	if (do_pc3)
-		outp += sprintf(outp, " Pkg%%pc3");
+		outp += sprintf(outp, "\tPkg%%pc3");
 	if (do_pc6)
-		outp += sprintf(outp, " Pkg%%pc6");
+		outp += sprintf(outp, "\tPkg%%pc6");
 	if (do_pc7)
-		outp += sprintf(outp, " Pkg%%pc7");
+		outp += sprintf(outp, "\tPkg%%pc7");
 	if (do_c8_c9_c10) {
-		outp += sprintf(outp, " Pkg%%pc8");
-		outp += sprintf(outp, " Pkg%%pc9");
-		outp += sprintf(outp, " Pk%%pc10");
+		outp += sprintf(outp, "\tPkg%%pc8");
+		outp += sprintf(outp, "\tPkg%%pc9");
+		outp += sprintf(outp, "\tPk%%pc10");
 	}
 
 	if (do_rapl && !rapl_joules) {
 		if (do_rapl & RAPL_PKG)
-			outp += sprintf(outp, " PkgWatt");
+			outp += sprintf(outp, "\tPkgWatt");
 		if (do_rapl & RAPL_CORES)
-			outp += sprintf(outp, " CorWatt");
+			outp += sprintf(outp, "\tCorWatt");
 		if (do_rapl & RAPL_GFX)
-			outp += sprintf(outp, " GFXWatt");
+			outp += sprintf(outp, "\tGFXWatt");
 		if (do_rapl & RAPL_DRAM)
-			outp += sprintf(outp, " RAMWatt");
+			outp += sprintf(outp, "\tRAMWatt");
 		if (do_rapl & RAPL_PKG_PERF_STATUS)
-			outp += sprintf(outp, "   PKG_%%");
+			outp += sprintf(outp, "\tPKG_%%");
 		if (do_rapl & RAPL_DRAM_PERF_STATUS)
-			outp += sprintf(outp, "   RAM_%%");
+			outp += sprintf(outp, "\tRAM_%%");
 	} else if (do_rapl && rapl_joules) {
 		if (do_rapl & RAPL_PKG)
-			outp += sprintf(outp, "   Pkg_J");
+			outp += sprintf(outp, "\tPkg_J");
 		if (do_rapl & RAPL_CORES)
-			outp += sprintf(outp, "   Cor_J");
+			outp += sprintf(outp, "\tCor_J");
 		if (do_rapl & RAPL_GFX)
-			outp += sprintf(outp, "   GFX_J");
+			outp += sprintf(outp, "\tGFX_J");
 		if (do_rapl & RAPL_DRAM)
-			outp += sprintf(outp, "   RAM_J");
+			outp += sprintf(outp, "\tRAM_J");
 		if (do_rapl & RAPL_PKG_PERF_STATUS)
-			outp += sprintf(outp, "   PKG_%%");
+			outp += sprintf(outp, "\tPKG_%%");
 		if (do_rapl & RAPL_DRAM_PERF_STATUS)
-			outp += sprintf(outp, "   RAM_%%");
+			outp += sprintf(outp, "\tRAM_%%");
 	}
     done:
 	outp += sprintf(outp, "\n");

@@ -515,91 +515,91 @@ int format_counters(struct thread_data *t, struct core_data *c,
 	/* topo columns, print blanks on 1st (average) line */
 	if (t == &average.threads) {
 		if (show_pkg)
-			outp += sprintf(outp, "       -");
+			outp += sprintf(outp, "\t-");
 		if (show_core)
-			outp += sprintf(outp, "       -");
+			outp += sprintf(outp, "\t-");
 		if (show_cpu)
-			outp += sprintf(outp, "       -");
+			outp += sprintf(outp, "\t-");
 	} else {
 		if (show_pkg) {
 			if (p)
-				outp += sprintf(outp, "%8d", p->package_id);
+				outp += sprintf(outp, "\t%d", p->package_id);
 			else
-				outp += sprintf(outp, "       -");
+				outp += sprintf(outp, "\t-");
 		}
 		if (show_core) {
 			if (c)
-				outp += sprintf(outp, "%8d", c->core_id);
+				outp += sprintf(outp, "\t%d", c->core_id);
 			else
-				outp += sprintf(outp, "       -");
+				outp += sprintf(outp, "\t-");
 		}
 		if (show_cpu)
-			outp += sprintf(outp, "%8d", t->cpu_id);
+			outp += sprintf(outp, "\t%d", t->cpu_id);
 	}
 
 	/* Avg_MHz */
 	if (has_aperf)
-		outp += sprintf(outp, "%8.0f",
+		outp += sprintf(outp, "\t%.0f",
 			1.0 / units * t->aperf / interval_float);
 
 	/* Busy% */
 	if (has_aperf)
-		outp += sprintf(outp, "%8.2f", 100.0 * t->mperf/t->tsc/tsc_tweak);
+		outp += sprintf(outp, "\t%.2f", 100.0 * t->mperf/t->tsc/tsc_tweak);
 
 	/* Bzy_MHz */
 	if (has_aperf) {
 		if (has_base_hz)
-			outp += sprintf(outp, "%8.0f", base_hz / units * t->aperf / t->mperf);
+			outp += sprintf(outp, "\t%.0f", base_hz / units * t->aperf / t->mperf);
 		else
-			outp += sprintf(outp, "%8.0f",
+			outp += sprintf(outp, "\t%.0f",
 				1.0 * t->tsc / units * t->aperf / t->mperf / interval_float);
 	}
 
 	/* TSC_MHz */
-	outp += sprintf(outp, "%8.0f", 1.0 * t->tsc/units/interval_float);
+	outp += sprintf(outp, "\t%.0f", 1.0 * t->tsc/units/interval_float);
 
 	/* delta */
 	if (extra_delta_offset32)
-		outp += sprintf(outp, "  %11llu", t->extra_delta32);
+		outp += sprintf(outp, "\t%11llu", t->extra_delta32);
 
 	/* DELTA */
 	if (extra_delta_offset64)
-		outp += sprintf(outp, "  %11llu", t->extra_delta64);
+		outp += sprintf(outp, "\t%11llu", t->extra_delta64);
 	/* msr */
 	if (extra_msr_offset32)
-		outp += sprintf(outp, "  0x%08llx", t->extra_msr32);
+		outp += sprintf(outp, "\t0x%08llx", t->extra_msr32);
 
 	/* MSR */
 	if (extra_msr_offset64)
-		outp += sprintf(outp, "  0x%016llx", t->extra_msr64);
+		outp += sprintf(outp, "\t0x%016llx", t->extra_msr64);
 
 	if (!debug)
 		goto done;
 
 	/* IRQ */
 	if (do_irq)
-		outp += sprintf(outp, "%8d", t->irq_count);
+		outp += sprintf(outp, "\t%d", t->irq_count);
 
 	/* SMI */
 	if (do_smi)
-		outp += sprintf(outp, "%8d", t->smi_count);
+		outp += sprintf(outp, "\t%d", t->smi_count);
 
 	if (do_nhm_cstates)
-		outp += sprintf(outp, "%8.2f", 100.0 * t->c1/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * t->c1/t->tsc);
 
 	/* print per-core data only for 1st thread in core */
 	if (!(t->flags & CPU_IS_FIRST_THREAD_IN_CORE))
 		goto done;
 
 	if (do_nhm_cstates && !do_slm_cstates && !do_knl_cstates)
-		outp += sprintf(outp, "%8.2f", 100.0 * c->c3/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * c->c3/t->tsc);
 	if (do_nhm_cstates)
-		outp += sprintf(outp, "%8.2f", 100.0 * c->c6/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * c->c6/t->tsc);
 	if (do_snb_cstates)
-		outp += sprintf(outp, "%8.2f", 100.0 * c->c7/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * c->c7/t->tsc);
 
 	if (do_dts)
-		outp += sprintf(outp, "%8d", c->core_temp_c);
+		outp += sprintf(outp, "\t%d", c->core_temp_c);
 
 	/* print per-package data only for 1st core in package */
 	if (!(t->flags & CPU_IS_FIRST_CORE_IN_PACKAGE))

@@ -607,42 +607,42 @@ int format_counters(struct thread_data *t, struct core_data *c,
 
 	/* PkgTmp */
 	if (do_ptm)
-		outp += sprintf(outp, "%8d", p->pkg_temp_c);
+		outp += sprintf(outp, "\t%d", p->pkg_temp_c);
 
 	/* GFXrc6 */
 	if (do_gfx_rc6_ms) {
 		if (p->gfx_rc6_ms == -1) {	/* detect GFX counter reset */
-			outp += sprintf(outp, "   **.**");
+			outp += sprintf(outp, "\t**.**");
 		} else {
-			outp += sprintf(outp, "%8.2f",
+			outp += sprintf(outp, "\t%.2f",
 				p->gfx_rc6_ms / 10.0 / interval_float);
 		}
 	}
 
 	/* GFXMHz */
 	if (do_gfx_mhz)
-		outp += sprintf(outp, "%8d", p->gfx_mhz);
+		outp += sprintf(outp, "\t%d", p->gfx_mhz);
 
 	/* Totl%C0, Any%C0 GFX%C0 CPUGFX% */
 	if (do_skl_residency) {
-		outp += sprintf(outp, "%8.2f", 100.0 * p->pkg_wtd_core_c0/t->tsc);
-		outp += sprintf(outp, "%8.2f", 100.0 * p->pkg_any_core_c0/t->tsc);
-		outp += sprintf(outp, "%8.2f", 100.0 * p->pkg_any_gfxe_c0/t->tsc);
-		outp += sprintf(outp, "%8.2f", 100.0 * p->pkg_both_core_gfxe_c0/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * p->pkg_wtd_core_c0/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * p->pkg_any_core_c0/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * p->pkg_any_gfxe_c0/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * p->pkg_both_core_gfxe_c0/t->tsc);
 	}
 
 	if (do_pc2)
-		outp += sprintf(outp, "%8.2f", 100.0 * p->pc2/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * p->pc2/t->tsc);
 	if (do_pc3)
-		outp += sprintf(outp, "%8.2f", 100.0 * p->pc3/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * p->pc3/t->tsc);
 	if (do_pc6)
-		outp += sprintf(outp, "%8.2f", 100.0 * p->pc6/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * p->pc6/t->tsc);
 	if (do_pc7)
-		outp += sprintf(outp, "%8.2f", 100.0 * p->pc7/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * p->pc7/t->tsc);
 	if (do_c8_c9_c10) {
-		outp += sprintf(outp, "%8.2f", 100.0 * p->pc8/t->tsc);
-		outp += sprintf(outp, "%8.2f", 100.0 * p->pc9/t->tsc);
-		outp += sprintf(outp, "%8.2f", 100.0 * p->pc10/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * p->pc8/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * p->pc9/t->tsc);
+		outp += sprintf(outp, "\t%.2f", 100.0 * p->pc10/t->tsc);
 	}
 
 	/*

@@ -650,7 +650,7 @@ int format_counters(struct thread_data *t, struct core_data *c,
  	 * indicate that results are suspect by printing "**" in fraction place.
  	 */
 	if (interval_float < rapl_joule_counter_range)
-		fmt8 = "%8.2f";
+		fmt8 = "\t%.2f";
 	else
 		fmt8 = "%6.0f**";

@@ -3715,7 +3715,7 @@ int get_and_dump_counters(void)
 }
 
 void print_version() {
-	fprintf(outf, "turbostat version 4.13 22 Apr 2016"
+	fprintf(outf, "turbostat version 4.14 22 Apr 2016"
 		" - Len Brown <[email protected]>\n");
 }

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #4 lore raw

PATCH 4/17 tools/power turbostat: Allocate correct amount of fd and irq entries

In reply to Len Brown

From: Mika Westerberg <[email protected]>

The tool uses topo.max_cpu_num to determine number of entries needed for fd_percpu[] and irqs_per_cpu[]. For example on a system with 4 CPUs topo.max_cpu_num is 3 so we get too small array for holding per-CPU items.

Fix this to use right number of entries, which is topo.max_cpu_num + 1.

Signed-off-by Mika Westerberg <[email protected]>

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

tools/power/x86/turbostat/turbostat.c +4 -4

index be50892f27aa..741fefa2f07e 100644
@@ -1618,7 +1618,7 @@ void free_fd_percpu(void)
 {
 	int i;
 
-	for (i = 0; i < topo.max_cpu_num; ++i) {
+	for (i = 0; i < topo.max_cpu_num + 1; ++i) {
 		if (fd_percpu[i] != 0)
 			close(fd_percpu[i]);
 	}

@@ -3584,7 +3584,7 @@ void allocate_output_buffer()
 }
 void allocate_fd_percpu(void)
 {
-	fd_percpu = calloc(topo.max_cpu_num, sizeof(int));
+	fd_percpu = calloc(topo.max_cpu_num + 1, sizeof(int));
 	if (fd_percpu == NULL)
 		err(-1, "calloc fd_percpu");
 }

@@ -3594,9 +3594,9 @@ void allocate_irq_buffers(void)
 	if (irq_column_2_cpu == NULL)
 		err(-1, "calloc %d", topo.num_cpus);
 
-	irqs_per_cpu = calloc(topo.max_cpu_num, sizeof(int));
+	irqs_per_cpu = calloc(topo.max_cpu_num + 1, sizeof(int));
 	if (irqs_per_cpu == NULL)
-		err(-1, "calloc %d", topo.max_cpu_num);
+		err(-1, "calloc %d", topo.max_cpu_num + 1);
 }
 void setup_all_buffers(void)
 {

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #5 lore raw

PATCH 5/17 tools/power turbostat: fix error case overflow read of slm_freq_table[]

In reply to Len Brown

From: Colin Ian King <[email protected]>

When i >= SLM_BCLK_FREQS, the frequency read from the slm_freq_table is off the end of the array because msr is set to 3 rather than the actual array index i. Set i to 3 rather than msr to fix this.

Signed-off-by Colin Ian King <[email protected]>

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

tools/power/x86/turbostat/turbostat.c +1 -1

index 741fefa2f07e..12a6eea2d6ae 100644
@@ -3036,7 +3036,7 @@ double slm_bclk(void)
 	i = msr & 0xf;
 	if (i >= SLM_BCLK_FREQS) {
 		fprintf(outf, "SLM BCLK[%d] invalid\n", i);
-		msr = 3;
+		i = 3;
 	}
 	freq = slm_freq_table[i];

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #6 lore raw

PATCH 6/17 tools/power/turbostat: split core MSR support into status + limit

In reply to Len Brown

From: Jacob Pan <[email protected]>

Some CPUs may not have PP0/Core domain power limit MSRs. We should still allow its domain energy status to be used. This patch splits PP0/Core RAPL into two separate flags for power limit and energy status such that energy status can continue to be reported without power limit.

Without this patch, turbostat will not be able to use the remaining RAPL features if some PL MSRs are not present.

Signed-off-by Jacob Pan <[email protected]>

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)

tools/power/x86/turbostat/turbostat.c +11 -9

index 12a6eea2d6ae..3a80f949e5ba 100644
@@ -129,9 +129,8 @@ unsigned int has_hwp_pkg;		/* IA32_HWP_REQUEST_PKG */
 #define RAPL_DRAM_POWER_INFO	(1 << 5)
 					/* 0x61c MSR_DRAM_POWER_INFO */
 
-#define RAPL_CORES		(1 << 6)
+#define RAPL_CORES_POWER_LIMIT	(1 << 6)
 					/* 0x638 MSR_PP0_POWER_LIMIT */
-					/* 0x639 MSR_PP0_ENERGY_STATUS */
 #define RAPL_CORE_POLICY	(1 << 7)
 					/* 0x63a MSR_PP0_POLICY */

@@ -139,6 +138,10 @@ unsigned int has_hwp_pkg;		/* IA32_HWP_REQUEST_PKG */
 					/* 0x640 MSR_PP1_POWER_LIMIT */
 					/* 0x641 MSR_PP1_ENERGY_STATUS */
 					/* 0x642 MSR_PP1_POLICY */
+
+#define RAPL_CORES_ENERGY_STATUS	(1 << 9)
+					/* 0x639 MSR_PP0_ENERGY_STATUS */
+#define RAPL_CORES (RAPL_CORES_ENERGY_STATUS | RAPL_CORES_POWER_LIMIT)
 #define	TJMAX_DEFAULT	100
 
 #define MAX(a, b) ((a) > (b) ? (a) : (b))

@@ -397,7 +400,7 @@ void print_header(void)
 	if (do_rapl && !rapl_joules) {
 		if (do_rapl & RAPL_PKG)
 			outp += sprintf(outp, "\tPkgWatt");
-		if (do_rapl & RAPL_CORES)
+		if (do_rapl & RAPL_CORES_ENERGY_STATUS)
 			outp += sprintf(outp, "\tCorWatt");
 		if (do_rapl & RAPL_GFX)
 			outp += sprintf(outp, "\tGFXWatt");

@@ -410,7 +413,7 @@ void print_header(void)
 	} else if (do_rapl && rapl_joules) {
 		if (do_rapl & RAPL_PKG)
 			outp += sprintf(outp, "\tPkg_J");
-		if (do_rapl & RAPL_CORES)
+		if (do_rapl & RAPL_CORES_ENERGY_STATUS)
 			outp += sprintf(outp, "\tCor_J");
 		if (do_rapl & RAPL_GFX)
 			outp += sprintf(outp, "\tGFX_J");

@@ -657,7 +660,7 @@ int format_counters(struct thread_data *t, struct core_data *c,
 	if (do_rapl && !rapl_joules) {
 		if (do_rapl & RAPL_PKG)
 			outp += sprintf(outp, fmt8, p->energy_pkg * rapl_energy_units / interval_float);
-		if (do_rapl & RAPL_CORES)
+		if (do_rapl & RAPL_CORES_ENERGY_STATUS)
 			outp += sprintf(outp, fmt8, p->energy_cores * rapl_energy_units / interval_float);
 		if (do_rapl & RAPL_GFX)
 			outp += sprintf(outp, fmt8, p->energy_gfx * rapl_energy_units / interval_float);

@@ -1217,7 +1220,7 @@ int get_counters(struct thread_data *t, struct core_data *c, struct pkg_data *p)
 			return -13;
 		p->energy_pkg = msr & 0xFFFFFFFF;
 	}
-	if (do_rapl & RAPL_CORES) {
+	if (do_rapl & RAPL_CORES_ENERGY_STATUS) {
 		if (get_msr(cpu, MSR_PP0_ENERGY_STATUS, &msr))
 			return -14;
 		p->energy_cores = msr & 0xFFFFFFFF;

@@ -2657,7 +2660,7 @@ void rapl_probe(unsigned int family, unsigned int model)
 		break;
 	case 0x37:	/* BYT */
 	case 0x4D:	/* AVN */
-		do_rapl = RAPL_PKG | RAPL_CORES ;
+		do_rapl = RAPL_PKG | RAPL_CORES;
 		break;
 	default:
 		return;

@@ -2872,9 +2875,8 @@ int print_rapl(struct thread_data *t, struct core_data *c, struct pkg_data *p)
 			fprintf(outf, "cpu%d: MSR_PP0_POLICY: %lld\n", cpu, msr & 0xF);
 		}
 	}
-	if (do_rapl & RAPL_CORES) {
+	if (do_rapl & RAPL_CORES_POWER_LIMIT) {
 		if (debug) {
-
 			if (get_msr(cpu, MSR_PP0_POWER_LIMIT, &msr))
 				return -9;
 			fprintf(outf, "cpu%d: MSR_PP0_POWER_LIMIT: 0x%08llx (%slocked)\n",

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #7 lore raw

PATCH 7/17 tools/power/turbostat: Add Denverton support

In reply to Len Brown

From: Jacob Pan <[email protected]>

Denverton is an Atom based micro server which shares the same Goldmont architecture as Broxton. The available C-states on Denverton is a subset of Broxton with only C1, C1e, and C6.

Signed-off-by Jacob Pan <[email protected]>

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 2 ++
 1 file changed, 2 insertions(+)

tools/power/x86/turbostat/turbostat.c +2 -0

index 3a80f949e5ba..16ccada7e439 100644
@@ -2204,6 +2204,7 @@ int probe_nhm_msrs(unsigned int family, unsigned int model)
 		pkg_cstate_limits = phi_pkg_cstate_limits;
 		break;
 	case 0x5C:	/* BXT */
+	case 0x5F:	/* DNV */
 		pkg_cstate_limits = bxt_pkg_cstate_limits;
 		break;
 	default:

@@ -3301,6 +3302,7 @@ void process_cpuid()
 					crystal_hz = 25000000;	/* 25.0 MHz */
 					break;
 				case 0x5C:	/* BXT */
+				case 0x5F:	/* DNV */
 					crystal_hz = 19200000;	/* 19.2 MHz */
 					break;
 				default:

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #8 lore raw

PATCH 8/17 tools/power/turbostat: Add Denverton RAPL support

In reply to Len Brown

From: Jacob Pan <[email protected]>

The Denverton CPU RAPL supports package, core, and DRAM domains.

Signed-off-by Jacob Pan <[email protected]>

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 3 +++
 1 file changed, 3 insertions(+)

tools/power/x86/turbostat/turbostat.c +3 -0

index 16ccada7e439..70a6699c528c 100644
@@ -2663,6 +2663,9 @@ void rapl_probe(unsigned int family, unsigned int model)
 	case 0x4D:	/* AVN */
 		do_rapl = RAPL_PKG | RAPL_CORES;
 		break;
+	case 0x5f:	/* 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;
+		break;
 	default:
 		return;
 	}

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #9 lore raw

PATCH 9/17 tools/power turbostat: use intel-family.h model strings

In reply to Len Brown

From: Len Brown <[email protected]>

All except for model 1F, a Nehalem, which is currently incorrectly indentified as a Westmere in that new header.

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/Makefile    |   1 +
 tools/power/x86/turbostat/turbostat.c | 249 +++++++++++++++++-----------------
 2 files changed, 126 insertions(+), 124 deletions(-)

tools/power/x86/turbostat/Makefile +1 -0

index 8561e7ddca59..8792ad8dbf83 100644
@@ -10,6 +10,7 @@ endif
 turbostat : turbostat.c
 CFLAGS +=	-Wall
 CFLAGS +=	-DMSRHEADER='"../../../../arch/x86/include/asm/msr-index.h"'
+CFLAGS +=	-DINTEL_FAMILY_HEADER='"../../../../arch/x86/include/asm/intel-family.h"'
 
 %: %.c
 	@mkdir -p $(BUILD_OUTPUT)

tools/power/x86/turbostat/turbostat.c +125 ~~-124~~

387 lines not on this page · raw message

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #10 lore raw

PATCH 10/17 tools/power turbostat: fix Denverton BCLK

In reply to Len Brown

From: Xiaolong Wang <[email protected]>

Add Denverton to the group of SandyBridge and later processors, to let the bclk be recognized as 100MHz rather than 133MHz, then avoid the wrong value of the frequencies based on it, including Bzy_MHz, max efficiency freuency, base frequency, and turbo mode frequencies.

Signed-off-by Xiaolong Wang <[email protected]>

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 1 +
 1 file changed, 1 insertion(+)

tools/power/x86/turbostat/turbostat.c +1 -0

index 2056c148c684..89c31e90a85e 100644
@@ -2938,6 +2938,7 @@ int has_snb_msrs(unsigned int family, unsigned int model)
 	case INTEL_FAM6_KABYLAKE_DESKTOP:	/* KBL */
 	case INTEL_FAM6_SKYLAKE_X:	/* SKX */
 	case INTEL_FAM6_ATOM_GOLDMONT:	/* BXT */
+	case INTEL_FAM6_ATOM_DENVERTON:	/* DNV */
 		return 1;
 	}
 	return 0;

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #11 lore raw

PATCH 11/17 tools/power turbostat: Display HWP OOB status

In reply to Len Brown

From: Srinivas Pandruvada <[email protected]>

Display if the HWP is enabled in OOB (Out of band) mode.

Signed-off-by Srinivas Pandruvada <[email protected]>

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

tools/power/x86/turbostat/turbostat.c +3 -2

index 89c31e90a85e..83d1fb9750aa 100644
@@ -3168,10 +3168,11 @@ void decode_misc_pwr_mgmt_msr(void)
 		return;
 
 	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)\n",
+		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 << 1) ? "EN" : "DIS",
+			msr & (1 << 8) ? "EN" : "DIS");
 }
 
 void process_cpuid()

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #12 lore raw

PATCH 12/17 tools/power turbostat: Support Knights Mill (KNM)

In reply to Len Brown

From: Len Brown <[email protected]>

Original-author: Piotr Luc <[email protected]>

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 7 +++++++
 1 file changed, 7 insertions(+)

tools/power/x86/turbostat/turbostat.c +7 -0

index 83d1fb9750aa..a54ca3775521 100644
@@ -2202,6 +2202,7 @@ int probe_nhm_msrs(unsigned int family, unsigned int model)
 		pkg_cstate_limits = amt_pkg_cstate_limits;
 		break;
 	case INTEL_FAM6_XEON_PHI_KNL:	/* PHI */
+	case INTEL_FAM6_XEON_PHI_KNM:
 		pkg_cstate_limits = phi_pkg_cstate_limits;
 		break;
 	case INTEL_FAM6_ATOM_GOLDMONT:	/* BXT */

@@ -2228,6 +2229,7 @@ int has_nhm_turbo_ratio_limit(unsigned int family, unsigned int model)
 	case INTEL_FAM6_NEHALEM_EX:	/* Nehalem-EX Xeon - Beckton */
 	case INTEL_FAM6_WESTMERE_EX:	/* Westmere-EX Xeon - Eagleton */
 	case INTEL_FAM6_XEON_PHI_KNL:	/* PHI - Knights Landing (different MSR definition) */
+	case INTEL_FAM6_XEON_PHI_KNM:
 		return 0;
 	default:
 		return 1;

@@ -2275,6 +2277,7 @@ int has_knl_turbo_ratio_limit(unsigned int family, unsigned int model)
 
 	switch (model) {
 	case INTEL_FAM6_XEON_PHI_KNL:	/* Knights Landing */
+	case INTEL_FAM6_XEON_PHI_KNM:
 		return 1;
 	default:
 		return 0;

@@ -2305,6 +2308,7 @@ int has_config_tdp(unsigned int family, unsigned int model)
 	case INTEL_FAM6_SKYLAKE_X:	/* SKX */
 
 	case INTEL_FAM6_XEON_PHI_KNL:	/* Knights Landing */
+	case INTEL_FAM6_XEON_PHI_KNM:
 		return 1;
 	default:
 		return 0;

@@ -2606,6 +2610,7 @@ rapl_dram_energy_units_probe(int  model, double rapl_energy_units)
 	case INTEL_FAM6_BROADWELL_X:	/* BDX */
 	case INTEL_FAM6_BROADWELL_XEON_D:	/* BDX-DE */
 	case INTEL_FAM6_XEON_PHI_KNL:	/* KNL */
+	case INTEL_FAM6_XEON_PHI_KNM:
 		return (rapl_dram_energy_units = 15.3 / 1000000);
 	default:
 		return (rapl_energy_units);

@@ -2654,6 +2659,7 @@ void rapl_probe(unsigned int family, unsigned int model)
 	case INTEL_FAM6_BROADWELL_XEON_D:	/* BDX-DE */
 	case INTEL_FAM6_SKYLAKE_X:	/* SKX */
 	case INTEL_FAM6_XEON_PHI_KNL:	/* KNL */
+	case INTEL_FAM6_XEON_PHI_KNM:
 		do_rapl = RAPL_PKG | RAPL_DRAM | RAPL_DRAM_POWER_INFO | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS | RAPL_PKG_POWER_INFO;
 		break;
 	case INTEL_FAM6_SANDYBRIDGE_X:

@@ -3017,6 +3023,7 @@ int is_knl(unsigned int family, unsigned int model)
 		return 0;
 	switch (model) {
 	case INTEL_FAM6_XEON_PHI_KNL:	/* KNL */
+	case INTEL_FAM6_XEON_PHI_KNM:
 		return 1;
 	}
 	return 0;

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #13 lore raw

PATCH 13/17 tools/power turbostat: fix SKX PKG_CSTATE_LIMIT decoding

In reply to Len Brown

From: Len Brown <[email protected]>

SKX has fewer package C-states than previous generations, and so the decoding of PKG_CSTATE_LIMIT has changed.

This changes the line ending with pkg-cstate-limit=XXX: pcYYY

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

tools/power/x86/turbostat/turbostat.c +4 -1

index a54ca3775521..10ae4ffe137c 100644
@@ -1294,6 +1294,7 @@ int slv_pkg_cstate_limits[16] = {PCL__0, PCL__1, PCLRSV, PCLRSV, PCL__4, PCLRSV,
 int amt_pkg_cstate_limits[16] = {PCL__0, 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 bxt_pkg_cstate_limits[16] = {PCL__0, PCL__2, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, 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, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV};
 
 
 static void

@@ -2191,9 +2192,11 @@ int probe_nhm_msrs(unsigned int family, unsigned int model)
 	case INTEL_FAM6_SKYLAKE_DESKTOP:	/* SKL */
 	case INTEL_FAM6_KABYLAKE_MOBILE:	/* KBL */
 	case INTEL_FAM6_KABYLAKE_DESKTOP:	/* KBL */
-	case INTEL_FAM6_SKYLAKE_X:	/* SKX */
 		pkg_cstate_limits = hsw_pkg_cstate_limits;
 		break;
+	case INTEL_FAM6_SKYLAKE_X:	/* SKX */
+		pkg_cstate_limits = skx_pkg_cstate_limits;
+		break;
 	case INTEL_FAM6_ATOM_SILVERMONT1:	/* BYT */
 	case INTEL_FAM6_ATOM_SILVERMONT2:	/* AVN */
 		pkg_cstate_limits = slv_pkg_cstate_limits;

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #14 lore raw

PATCH 14/17 tools/power turbostat: line up headers when -M is used

In reply to Len Brown

From: Len Brown <[email protected]>

The -M option adds an 18-column item, and the header needs to be wide enough to keep the header aligned with the columns.

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

tools/power/x86/turbostat/turbostat.c +1 -1

index 10ae4ffe137c..cfdb574ea302 100644
@@ -345,7 +345,7 @@ void print_header(void)
 	if (extra_delta_offset64)
 		outp += sprintf(outp, "\tCOUNT 0x%03X", extra_delta_offset64);
 	if (extra_msr_offset32)
-		outp += sprintf(outp, "\tMSR 0x%03X", extra_msr_offset32);
+		outp += sprintf(outp, "\t         MSR 0x%03X", extra_msr_offset32);
 	if (extra_msr_offset64)
 		outp += sprintf(outp, "\tMSR 0x%03X", extra_msr_offset64);

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #15 lore raw

PATCH 15/17 tools/power turbostat: Denverton uses a 25 MHz crystal, not 19.2 MHz

In reply to Len Brown

From: Len Brown <[email protected]>

This changes only the TSC frequency decoding line seen with --debug

old: TSC: 1382 MHz (19200000 Hz * 216 / 3 / 1000000)
new: TSC: 1800 MHz (25000000 Hz * 216 / 3 / 1000000)

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

tools/power/x86/turbostat/turbostat.c +1 -1

index cfdb574ea302..3708386531f2 100644
@@ -3315,10 +3315,10 @@ void process_cpuid()
 					crystal_hz = 24000000;	/* 24.0 MHz */
 					break;
 				case INTEL_FAM6_SKYLAKE_X:	/* SKX */
+				case INTEL_FAM6_ATOM_DENVERTON:	/* DNV */
 					crystal_hz = 25000000;	/* 25.0 MHz */
 					break;
 				case INTEL_FAM6_ATOM_GOLDMONT:	/* BXT */
-				case INTEL_FAM6_ATOM_DENVERTON:	/* DNV */
 					crystal_hz = 19200000;	/* 19.2 MHz */
 					break;
 				default:

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #16 lore raw

PATCH 16/17 tools/power turbostat: Make extensible via the --add parameter

In reply to Len Brown

From: Len Brown <[email protected]>

Create the "--add" parameter. This can be used to teach an existing turbostat binary about any number of any type of counter.

turbostat(8) details the syntax for --add.

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.8 |  22 ++
 tools/power/x86/turbostat/turbostat.c | 396 +++++++++++++++++++++++++++++++++-
 2 files changed, 409 insertions(+), 9 deletions(-)

tools/power/x86/turbostat/turbostat.8 +22 -0

index 492e84fbebfa..39f7daeb6fbd 100644
@@ -25,6 +25,28 @@ Some information is not available on older processors.
 .SS Options
 Options can be specified with a single or double '-', and only as much of the option
 name as necessary to disambiguate it from others is necessary.  Note that options are case-sensitive.
+.PP
+\fB--add attributes\fP add column with counter having specified 'attributes'.  The 'location' attribute is required, all others are optional.
+.nf
+	location: {\fBmsrDDD\fP | \fBmsr0xXXX\fP}
+		msrDDD is a decimal offset, eg. msr16
+		msr0xXXX is a hex offset, eg. msr0x10
+
+	scope: {\fBcpu\fP | \fBcore\fP | \fBpackage\fP}
+		sample and print the counter for every cpu, core, or package.
+		default: cpu
+
+	size: {\fBu32\fP | \fBu64\fP }
+		MSRs are read as 64-bits, u32 truncates the displayed value to 32-bits.
+		default: u64
+
+	format: {\fBraw\fP | \fBdelta\fP | \fBpercent\fP}
+		'raw' shows the MSR contents in hex.
+		'delta' shows the difference in values during the measurement interval.
+		'percent' shows the delta as a percentage of the cycles elapsed.
+		default: delta
+.fi
+.PP
 \fB--Counter MSR#\fP shows the delta of the specified 64-bit MSR counter.
 .PP
 \fB--counter MSR#\fP shows the delta of the specified 32-bit MSR counter.

tools/power/x86/turbostat/turbostat.c +387 -9

654 lines not on this page · raw message

Signature · 1 line

2.11.0.161.g6610af872

Len Brown

author maintainer

2016-12-24 20:56 UTC #17 lore raw

PATCH 17/17 tools/power turbostat: remove obsolete -M, -m, -C, -c options

In reply to Len Brown

From: Len Brown <[email protected]>

The new --add option has replaced the -M, -m, -C, -c options Eg.

-M 0x10 is now --add msr0x10,raw
-m 0x10 is now --add msr0x10,raw,u32
-C 0x10 is now --add msr0x10,delta
-c 0x10 is now --add msr0x10,delta,u32

The --add option can be repeated to add any number of counters, while the previous options were limited to adding one of each type.

In addition, the --add option can accept a column label, and can also display a counter as a percentage of elapsed cycles.

Eg. --add msr0x3fe,core,percent,MY_CC3

Signed-off-by Len Brown <[email protected]>

---
 tools/power/x86/turbostat/turbostat.8 |   8 ---
 tools/power/x86/turbostat/turbostat.c | 104 +---------------------------------
 2 files changed, 2 insertions(+), 110 deletions(-)

tools/power/x86/turbostat/turbostat.8 +0 -8

index 39f7daeb6fbd..03cb639b292e 100644
@@ -47,10 +47,6 @@ name as necessary to disambiguate it from others is necessary.  Note that option
 		default: delta
 .fi
 .PP
-\fB--Counter MSR#\fP shows the delta of the specified 64-bit MSR counter.
-.PP
-\fB--counter MSR#\fP shows the delta of the specified 32-bit MSR counter.
-.PP
 \fB--Dump\fP displays the raw counter values.
 .PP
 \fB--debug\fP displays additional system configuration information.  Invoking this parameter

@@ -65,10 +61,6 @@ The file is truncated if it already exists, and it is created if it does not exi
 .PP
 \fB--Joules\fP displays energy in Joules, rather than dividing Joules by time to print power in Watts.
 .PP
-\fB--MSR MSR#\fP shows the specified 64-bit MSR value.
-.PP
-\fB--msr MSR#\fP shows the specified 32-bit MSR value.
-.PP
 \fB--Package\fP limits output to the system summary plus the 1st thread in each Package.
 .PP
 \fB--processor\fP limits output to the system summary plus the 1st thread in each processor of each package.  Ie. it skips hyper-threaded siblings.

tools/power/x86/turbostat/turbostat.c +2 ~~-102~~

200 lines not on this page · raw message

Signature · 1 line

2.11.0.161.g6610af872