Messages in this thread | | | Date | Wed, 18 Nov 2015 15:59:16 +0800 | From | "Wangnan (F)" <> | Subject | Re: [PATCH 2/3] perf tools: Add callchain order support for libunwind DWARF unwinder |
| |
On 2015/11/18 15:54, Jiri Olsa wrote: > On Wed, Nov 18, 2015 at 12:13:08PM +0800, Wangnan (F) wrote: > > SNIP > >> 3.38% a.out a.out [.] funcc >> | >> ---funcc >> | >> |--2.70%-- funcb >> | funca >> | main >> | __libc_start_main >> | _start >> | >> --0.68%-- 0 >> >> >> I'm not sure whether we can regard this behavior changing as a bugfix? I >> think >> there may be some reason the original code explicitly avoid creating an '0' >> entry. >> >> Then I tried to find why perf can't get call frame on my case, and >> I guess there's something wrong whe dealing with 'call' command, because >> the instruction on it I can't get callchain from libunwind is a 'callq': >> >> ... >> 4005bf: be 00 00 00 00 mov $0x0,%esi >> 4005c4: 48 89 c7 mov %rax,%rdi >> 4005c7: e8 74 fe ff ff callq 400440 <gettimeofday@plt> >> us2 = tv2.tv_sec * 1000000 + tv2.tv_usec; >> 4005cc: 48 8b 04 24 mov (%rsp),%rax >> ... >> >> But this is another problem, we can discuss it in a new thread. > so the problem is you dont see the gettimeofday call at the end?
No. The problem is when sample is taken at 'callq' perf is unable to unwind correctly, even with dwarf and user stack.
> could you share the test code?
I have posted a detail analysis in [1] (rechecked, you are in cc-list). Test code is here:
#include <stdio.h> #include <unistd.h> #include <sys/time.h>
static volatile int x = 0;
int funcc(void) { struct timeval tv1, tv2; unsigned long us1, us2;
gettimeofday(&tv1, NULL);
us1 = tv1.tv_sec * 1000000 + tv1.tv_usec;
while(1) { x = x + 100; gettimeofday(&tv2, NULL); us2 = tv2.tv_sec * 1000000 + tv2.tv_usec; if (us2 - us1 >= 3000000) break; } return x; } int funcb(void) { return funcc();} int funca(void) { return funcb();} int main() { funca(); return 0;}
Thank you.
[1] http://lkml.kernel.org/r/564C26C4.2040603@huawei.com
> > thanks, > jirka
| |