Minki/linux
1
0
Fork 1
mirror of https://github.com/torvalds/linux.git synced 2024-11-10 22:21:40 +00:00
Code Issues Packages Projects Releases Wiki Activity
38e11df134
linux/kernel/trace/trace_benchmark.c

199 lines
4.3 KiB
C
Raw Normal View History

tracing: Add tracepoint benchmark tracepoint In order to help benchmark the time tracepoints take, a new config option is added called CONFIG_TRACEPOINT_BENCHMARK. When this option is set a tracepoint is created called "benchmark:benchmark_event". When the tracepoint is enabled, it kicks off a kernel thread that goes into an infinite loop (calling cond_sched() to let other tasks run), and calls the tracepoint. Each iteration will record the time it took to write to the tracepoint and the next iteration that data will be passed to the tracepoint itself. That is, the tracepoint will report the time it took to do the previous tracepoint. The string written to the tracepoint is a static string of 128 bytes to keep the time the same. The initial string is simply a write of "START". The second string records the cold cache time of the first write which is not added to the rest of the calculations. As it is a tight loop, it benchmarks as hot cache. That's fine because we care most about hot paths that are probably in cache already. An example of the output: START first=3672 [COLD CACHED] last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666 Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-05-30 02:49:07 +00:00
#include <linux/delay.h>
#include <linux/module.h>
#include <linux/kthread.h>
#include <linux/trace_clock.h>
#define CREATE_TRACE_POINTS
#include "trace_benchmark.h"
static struct task_struct *bm_event_thread;
static char bm_str[BENCHMARK_EVENT_STRLEN] = "START";
static u64 bm_total;
static u64 bm_totalsq;
static u64 bm_last;
static u64 bm_max;
static u64 bm_min;
static u64 bm_first;
tracing: Only calculate stats of tracepoint benchmarks for 2^32 times When calculating the average and standard deviation, it is required that the count be less than UINT_MAX, otherwise the do_div() will get undefined results. After 2^32 counts of data, the average and standard deviation should pretty much be set anyway. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-06-06 03:34:02 +00:00
static u64 bm_cnt;
static u64 bm_stddev;
static unsigned int bm_avg;
static unsigned int bm_std;
tracing: Add tracepoint benchmark tracepoint In order to help benchmark the time tracepoints take, a new config option is added called CONFIG_TRACEPOINT_BENCHMARK. When this option is set a tracepoint is created called "benchmark:benchmark_event". When the tracepoint is enabled, it kicks off a kernel thread that goes into an infinite loop (calling cond_sched() to let other tasks run), and calls the tracepoint. Each iteration will record the time it took to write to the tracepoint and the next iteration that data will be passed to the tracepoint itself. That is, the tracepoint will report the time it took to do the previous tracepoint. The string written to the tracepoint is a static string of 128 bytes to keep the time the same. The initial string is simply a write of "START". The second string records the cold cache time of the first write which is not added to the rest of the calculations. As it is a tight loop, it benchmarks as hot cache. That's fine because we care most about hot paths that are probably in cache already. An example of the output: START first=3672 [COLD CACHED] last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666 Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-05-30 02:49:07 +00:00
/*
* This gets called in a loop recording the time it took to write
* the tracepoint. What it writes is the time statistics of the last
* tracepoint write. As there is nothing to write the first time
* it simply writes "START". As the first write is cold cache and
* the rest is hot, we save off that time in bm_first and it is
* reported as "first", which is shown in the second write to the
* tracepoint. The "first" field is writen within the statics from
* then on but never changes.
*/
static void trace_do_benchmark(void)
{
u64 start;
u64 stop;
u64 delta;
tracing: Convert stddev into u64 in tracepoint benchmark I've been told that do_div() expects an unsigned 64 bit number, and is undefined if a signed is used. This gave a warning on the MIPS build. I'm not sure if a signed 64 bit dividend is really an issue or not, but the calculation this is used for is standard deviation, and that isn't going to be negative. We can just convert it to unsigned and be safe. Reported-by: David Daney <ddaney.cavm@gmail.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-06-06 00:35:30 +00:00
u64 stddev;
tracing: Add tracepoint benchmark tracepoint In order to help benchmark the time tracepoints take, a new config option is added called CONFIG_TRACEPOINT_BENCHMARK. When this option is set a tracepoint is created called "benchmark:benchmark_event". When the tracepoint is enabled, it kicks off a kernel thread that goes into an infinite loop (calling cond_sched() to let other tasks run), and calls the tracepoint. Each iteration will record the time it took to write to the tracepoint and the next iteration that data will be passed to the tracepoint itself. That is, the tracepoint will report the time it took to do the previous tracepoint. The string written to the tracepoint is a static string of 128 bytes to keep the time the same. The initial string is simply a write of "START". The second string records the cold cache time of the first write which is not added to the rest of the calculations. As it is a tight loop, it benchmarks as hot cache. That's fine because we care most about hot paths that are probably in cache already. An example of the output: START first=3672 [COLD CACHED] last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666 Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-05-30 02:49:07 +00:00
u64 seed;
u64 last_seed;
unsigned int avg;
unsigned int std = 0;
/* Only run if the tracepoint is actually active */
tracing: Only benchmark the time tracepoints take if tracing is on There's no need to record the time tracepoints take when tracing is off. This is because: 1) We cannot see these records since ring_buffer record is off at that moment. 2) If tracing is off and benchmark tracepoint is enabled, the time tracepoint takes is fewer than the same situation when tracing is on, since the tracepoints need to be wrote into ring_buffer, it would take more time. If turn on tracing at this moment, the average and standard deviation cannot exactly present the time that tracepoints take to write data into ring_buffer. Link: http://lkml.kernel.org/r/1445947933-27955-1-git-send-email-zhang.chunyan@linaro.org Signed-off-by: Chunyan Zhang <zhang.chunyan@linaro.org> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2015-10-27 12:12:13 +00:00
if (!trace_benchmark_event_enabled() || !tracing_is_on())
tracing: Add tracepoint benchmark tracepoint In order to help benchmark the time tracepoints take, a new config option is added called CONFIG_TRACEPOINT_BENCHMARK. When this option is set a tracepoint is created called "benchmark:benchmark_event". When the tracepoint is enabled, it kicks off a kernel thread that goes into an infinite loop (calling cond_sched() to let other tasks run), and calls the tracepoint. Each iteration will record the time it took to write to the tracepoint and the next iteration that data will be passed to the tracepoint itself. That is, the tracepoint will report the time it took to do the previous tracepoint. The string written to the tracepoint is a static string of 128 bytes to keep the time the same. The initial string is simply a write of "START". The second string records the cold cache time of the first write which is not added to the rest of the calculations. As it is a tight loop, it benchmarks as hot cache. That's fine because we care most about hot paths that are probably in cache already. An example of the output: START first=3672 [COLD CACHED] last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666 Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-05-30 02:49:07 +00:00
return;
local_irq_disable();
start = trace_clock_local();
trace_benchmark_event(bm_str);
stop = trace_clock_local();
local_irq_enable();
bm_cnt++;
delta = stop - start;
/*
* The first read is cold cached, keep it separate from the
* other calculations.
*/
if (bm_cnt == 1) {
bm_first = delta;
scnprintf(bm_str, BENCHMARK_EVENT_STRLEN,
"first=%llu [COLD CACHED]", bm_first);
return;
}
bm_last = delta;
if (delta > bm_max)
bm_max = delta;
if (!bm_min || delta < bm_min)
bm_min = delta;
tracing: Only calculate stats of tracepoint benchmarks for 2^32 times When calculating the average and standard deviation, it is required that the count be less than UINT_MAX, otherwise the do_div() will get undefined results. After 2^32 counts of data, the average and standard deviation should pretty much be set anyway. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-06-06 03:34:02 +00:00
/*
* When bm_cnt is greater than UINT_MAX, it breaks the statistics
* accounting. Freeze the statistics when that happens.
* We should have enough data for the avg and stddev anyway.
*/
if (bm_cnt > UINT_MAX) {
scnprintf(bm_str, BENCHMARK_EVENT_STRLEN,
"last=%llu first=%llu max=%llu min=%llu ** avg=%u std=%d std^2=%lld",
bm_last, bm_first, bm_max, bm_min, bm_avg, bm_std, bm_stddev);
return;
}
bm_total += delta;
bm_totalsq += delta * delta;
tracing: Add tracepoint benchmark tracepoint In order to help benchmark the time tracepoints take, a new config option is added called CONFIG_TRACEPOINT_BENCHMARK. When this option is set a tracepoint is created called "benchmark:benchmark_event". When the tracepoint is enabled, it kicks off a kernel thread that goes into an infinite loop (calling cond_sched() to let other tasks run), and calls the tracepoint. Each iteration will record the time it took to write to the tracepoint and the next iteration that data will be passed to the tracepoint itself. That is, the tracepoint will report the time it took to do the previous tracepoint. The string written to the tracepoint is a static string of 128 bytes to keep the time the same. The initial string is simply a write of "START". The second string records the cold cache time of the first write which is not added to the rest of the calculations. As it is a tight loop, it benchmarks as hot cache. That's fine because we care most about hot paths that are probably in cache already. An example of the output: START first=3672 [COLD CACHED] last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666 Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-05-30 02:49:07 +00:00
if (bm_cnt > 1) {
/*
* Apply Welford's method to calculate standard deviation:
* s^2 = 1 / (n * (n-1)) * (n * \Sum (x_i)^2 - (\Sum x_i)^2)
*/
stddev = (u64)bm_cnt * bm_totalsq - bm_total * bm_total;
tracing: Only calculate stats of tracepoint benchmarks for 2^32 times When calculating the average and standard deviation, it is required that the count be less than UINT_MAX, otherwise the do_div() will get undefined results. After 2^32 counts of data, the average and standard deviation should pretty much be set anyway. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-06-06 03:34:02 +00:00
do_div(stddev, (u32)bm_cnt);
do_div(stddev, (u32)bm_cnt - 1);
tracing: Add tracepoint benchmark tracepoint In order to help benchmark the time tracepoints take, a new config option is added called CONFIG_TRACEPOINT_BENCHMARK. When this option is set a tracepoint is created called "benchmark:benchmark_event". When the tracepoint is enabled, it kicks off a kernel thread that goes into an infinite loop (calling cond_sched() to let other tasks run), and calls the tracepoint. Each iteration will record the time it took to write to the tracepoint and the next iteration that data will be passed to the tracepoint itself. That is, the tracepoint will report the time it took to do the previous tracepoint. The string written to the tracepoint is a static string of 128 bytes to keep the time the same. The initial string is simply a write of "START". The second string records the cold cache time of the first write which is not added to the rest of the calculations. As it is a tight loop, it benchmarks as hot cache. That's fine because we care most about hot paths that are probably in cache already. An example of the output: START first=3672 [COLD CACHED] last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666 Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-05-30 02:49:07 +00:00
} else
stddev = 0;
delta = bm_total;
do_div(delta, bm_cnt);
avg = delta;
if (stddev > 0) {
int i = 0;
/*
* stddev is the square of standard deviation but
* we want the actualy number. Use the average
* as our seed to find the std.
*
* The next try is:
* x = (x + N/x) / 2
*
* Where N is the squared number to find the square
* root of.
*/
seed = avg;
do {
last_seed = seed;
seed = stddev;
if (!last_seed)
break;
do_div(seed, last_seed);
seed += last_seed;
do_div(seed, 2);
} while (i++ < 10 && last_seed != seed);
std = seed;
}
scnprintf(bm_str, BENCHMARK_EVENT_STRLEN,
"last=%llu first=%llu max=%llu min=%llu avg=%u std=%d std^2=%lld",
bm_last, bm_first, bm_max, bm_min, avg, std, stddev);
tracing: Only calculate stats of tracepoint benchmarks for 2^32 times When calculating the average and standard deviation, it is required that the count be less than UINT_MAX, otherwise the do_div() will get undefined results. After 2^32 counts of data, the average and standard deviation should pretty much be set anyway. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-06-06 03:34:02 +00:00
bm_std = std;
bm_avg = avg;
bm_stddev = stddev;
tracing: Add tracepoint benchmark tracepoint In order to help benchmark the time tracepoints take, a new config option is added called CONFIG_TRACEPOINT_BENCHMARK. When this option is set a tracepoint is created called "benchmark:benchmark_event". When the tracepoint is enabled, it kicks off a kernel thread that goes into an infinite loop (calling cond_sched() to let other tasks run), and calls the tracepoint. Each iteration will record the time it took to write to the tracepoint and the next iteration that data will be passed to the tracepoint itself. That is, the tracepoint will report the time it took to do the previous tracepoint. The string written to the tracepoint is a static string of 128 bytes to keep the time the same. The initial string is simply a write of "START". The second string records the cold cache time of the first write which is not added to the rest of the calculations. As it is a tight loop, it benchmarks as hot cache. That's fine because we care most about hot paths that are probably in cache already. An example of the output: START first=3672 [COLD CACHED] last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666 Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-05-30 02:49:07 +00:00
}
static int benchmark_event_kthread(void *arg)
{
/* sleep a bit to make sure the tracepoint gets activated */
msleep(100);
while (!kthread_should_stop()) {
trace_do_benchmark();
/*
* We don't go to sleep, but let others
* run as well.
*/
cond_resched();
}
return 0;
}
/*
* When the benchmark tracepoint is enabled, it calls this
* function and the thread that calls the tracepoint is created.
*/
void trace_benchmark_reg(void)
{
bm_event_thread = kthread_run(benchmark_event_kthread,
NULL, "event_benchmark");
WARN_ON(!bm_event_thread);
}
/*
* When the benchmark tracepoint is disabled, it calls this
* function and the thread that calls the tracepoint is deleted
* and all the numbers are reset.
*/
void trace_benchmark_unreg(void)
{
if (!bm_event_thread)
return;
kthread_stop(bm_event_thread);
strcpy(bm_str, "START");
bm_total = 0;
bm_totalsq = 0;
bm_last = 0;
bm_max = 0;
bm_min = 0;
bm_cnt = 0;
tracing: Only calculate stats of tracepoint benchmarks for 2^32 times When calculating the average and standard deviation, it is required that the count be less than UINT_MAX, otherwise the do_div() will get undefined results. After 2^32 counts of data, the average and standard deviation should pretty much be set anyway. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-06-06 03:34:02 +00:00
/* These don't need to be reset but reset them anyway */
tracing: Add tracepoint benchmark tracepoint In order to help benchmark the time tracepoints take, a new config option is added called CONFIG_TRACEPOINT_BENCHMARK. When this option is set a tracepoint is created called "benchmark:benchmark_event". When the tracepoint is enabled, it kicks off a kernel thread that goes into an infinite loop (calling cond_sched() to let other tasks run), and calls the tracepoint. Each iteration will record the time it took to write to the tracepoint and the next iteration that data will be passed to the tracepoint itself. That is, the tracepoint will report the time it took to do the previous tracepoint. The string written to the tracepoint is a static string of 128 bytes to keep the time the same. The initial string is simply a write of "START". The second string records the cold cache time of the first write which is not added to the rest of the calculations. As it is a tight loop, it benchmarks as hot cache. That's fine because we care most about hot paths that are probably in cache already. An example of the output: START first=3672 [COLD CACHED] last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666 Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-05-30 02:49:07 +00:00
bm_first = 0;
tracing: Only calculate stats of tracepoint benchmarks for 2^32 times When calculating the average and standard deviation, it is required that the count be less than UINT_MAX, otherwise the do_div() will get undefined results. After 2^32 counts of data, the average and standard deviation should pretty much be set anyway. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-06-06 03:34:02 +00:00
bm_std = 0;
bm_avg = 0;
bm_stddev = 0;
tracing: Add tracepoint benchmark tracepoint In order to help benchmark the time tracepoints take, a new config option is added called CONFIG_TRACEPOINT_BENCHMARK. When this option is set a tracepoint is created called "benchmark:benchmark_event". When the tracepoint is enabled, it kicks off a kernel thread that goes into an infinite loop (calling cond_sched() to let other tasks run), and calls the tracepoint. Each iteration will record the time it took to write to the tracepoint and the next iteration that data will be passed to the tracepoint itself. That is, the tracepoint will report the time it took to do the previous tracepoint. The string written to the tracepoint is a static string of 128 bytes to keep the time the same. The initial string is simply a write of "START". The second string records the cold cache time of the first write which is not added to the rest of the calculations. As it is a tight loop, it benchmarks as hot cache. That's fine because we care most about hot paths that are probably in cache already. An example of the output: START first=3672 [COLD CACHED] last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666 Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-05-30 02:49:07 +00:00
}
Reference in New Issue Copy Permalink