When run "perf record -e", the number of samples showed up is wrong on some 32 bit systems, i.e. powerpc and arm.
For example, run the below commands on 32 bit powerpc:
perf probe -x /lib/libc.so.6 malloc perf record -e probe_libc:malloc -a ls perf.data [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.036 MB perf.data (13829241621624967218 samples) ]
Actually, "perf script" just shows 21 samples. The number of samples is also absurd since samples is long type, but it is printed as PRIu64.
Build test is run on x86-64, x86, aarch64, arm, mips, ppc and ppc64.
Signed-off-by: Yang Shi yang.shi@linaro.org --- tools/perf/builtin-record.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 142eeb3..e54aa4c 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -49,7 +49,7 @@ struct record { int realtime_prio; bool no_buildid; bool no_buildid_cache; - long samples; + unsigned long long samples; };
static int record__write(struct record *rec, void *bf, size_t size)
Em Tue, Sep 29, 2015 at 02:49:43PM -0700, Yang Shi escreveu:
When run "perf record -e", the number of samples showed up is wrong on some 32 bit systems, i.e. powerpc and arm.
For example, run the below commands on 32 bit powerpc:
perf probe -x /lib/libc.so.6 malloc perf record -e probe_libc:malloc -a ls perf.data [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.036 MB perf.data (13829241621624967218 samples) ]
Actually, "perf script" just shows 21 samples. The number of samples is also absurd since samples is long type, but it is printed as PRIu64.
Build test is run on x86-64, x86, aarch64, arm, mips, ppc and ppc64.
Sure?
AR /tmp/build/perf/libperf.a builtin-record.c: In function ‘__cmd_record’: builtin-record.c:689:12: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare]
- Arnaldo
Signed-off-by: Yang Shi yang.shi@linaro.org
tools/perf/builtin-record.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 142eeb3..e54aa4c 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -49,7 +49,7 @@ struct record { int realtime_prio; bool no_buildid; bool no_buildid_cache;
- long samples;
- unsigned long long samples;
}; static int record__write(struct record *rec, void *bf, size_t size) -- 2.0.2
Em Fri, Oct 02, 2015 at 04:08:38PM -0300, Arnaldo Carvalho de Melo escreveu:
Em Tue, Sep 29, 2015 at 02:49:43PM -0700, Yang Shi escreveu:
When run "perf record -e", the number of samples showed up is wrong on some 32 bit systems, i.e. powerpc and arm.
For example, run the below commands on 32 bit powerpc:
perf probe -x /lib/libc.so.6 malloc perf record -e probe_libc:malloc -a ls perf.data [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.036 MB perf.data (13829241621624967218 samples) ]
Actually, "perf script" just shows 21 samples. The number of samples is also absurd since samples is long type, but it is printed as PRIu64.
Build test is run on x86-64, x86, aarch64, arm, mips, ppc and ppc64.
Sure?
AR /tmp/build/perf/libperf.a builtin-record.c: In function ‘__cmd_record’: builtin-record.c:689:12: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare]
You missed something important, I am fixing this and applying your patch:
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 974065f8ce80..24ace2f318c1 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -667,7 +667,7 @@ static int __cmd_record(struct record *rec, int argc, const char **argv)
auxtrace_snapshot_enabled = 1; for (;;) { - int hits = rec->samples; + unsigned long long hits = rec->samples;
if (record__mmap_read_all(rec) < 0) { auxtrace_snapshot_enabled = 0;
On 10/2/2015 12:10 PM, Arnaldo Carvalho de Melo wrote:
Em Fri, Oct 02, 2015 at 04:08:38PM -0300, Arnaldo Carvalho de Melo escreveu:
Em Tue, Sep 29, 2015 at 02:49:43PM -0700, Yang Shi escreveu:
When run "perf record -e", the number of samples showed up is wrong on some 32 bit systems, i.e. powerpc and arm.
For example, run the below commands on 32 bit powerpc:
perf probe -x /lib/libc.so.6 malloc perf record -e probe_libc:malloc -a ls perf.data [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.036 MB perf.data (13829241621624967218 samples) ]
Actually, "perf script" just shows 21 samples. The number of samples is also absurd since samples is long type, but it is printed as PRIu64.
Build test is run on x86-64, x86, aarch64, arm, mips, ppc and ppc64.
Sure?
AR /tmp/build/perf/libperf.a builtin-record.c: In function ‘__cmd_record’: builtin-record.c:689:12: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare]
You missed something important, I am fixing this and applying your patch:
Thanks a lot. Yes, it was wrong, however, my compiler just throws a warning instead of an error. And, the warning was hidden by my OE build environment.
Regards, Yang
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 974065f8ce80..24ace2f318c1 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -667,7 +667,7 @@ static int __cmd_record(struct record *rec, int argc, const char **argv)
auxtrace_snapshot_enabled = 1; for (;;) {
int hits = rec->samples;
unsigned long long hits = rec->samples;if (record__mmap_read_all(rec) < 0) { auxtrace_snapshot_enabled = 0;
linaro-kernel@lists.linaro.org
-
Arnaldo Carvalho de Melo -
Arnaldo Carvalho de Melo -
Shi, Yang -
Yang Shi