Re: [PATCH 16/35] perf annotate: Add samples into struct annotation_line

[Jiri Olsa]

On Tue, Nov 14, 2017 at 11:29:54AM +0100, Jiri Olsa wrote:
> On Tue, Nov 14, 2017 at 03:45:27PM +0530, Ravi Bangoria wrote:
> > Hi Jiri,
> > 
> > On 11/14/2017 03:01 PM, Jiri Olsa wrote:
> > > On Mon, Nov 13, 2017 at 09:14:38PM +0100, Jiri Olsa wrote:
> > > > On Mon, Nov 13, 2017 at 09:16:20PM +0530, Ravi Bangoria wrote:
> > > > > Hi Jiri,
> > > > > 
> > > > > This patch seems to be causing segfault with "perf top --stdio".
> > > > > 
> > > > > Steps to reproduce:
> > > > > 1. start "perf top --stdio" in one terminal
> > > > > 2. run some simple workload in another terminal, let it get finished.
> > > > > 3. annotate function from previous workload in perf top (press 'a' followed
> > > > > by 's')
> > > > > 
> > > > > Perf will crash with:
> > > > > 
> > > > >    perf: Segmentation fault
> > > > >    Obtained 8 stack frames.
> > > > >    ./perf(sighandler_dump_stack+0x3e) [0x4f1b6e]
> > > > >    /lib64/libc.so.6(+0x36a7f) [0x7ff3aa7e4a7f]
> > > > >    ./perf() [0x4a27fd]
> > > > >    ./perf(symbol__annotate+0x199) [0x4a4439]
> > > > >    ./perf() [0x44e32d]
> > > > >    ./perf() [0x44f098]
> > > > >    /lib64/libpthread.so.0(+0x736c) [0x7ff3acee836c]
> > > > >    /lib64/libc.so.6(clone+0x3e) [0x7ff3aa8bee1e]
> > > > > 
> > > > > Can you please check.
> > > > hum, I'm getting following crash after resizing the terminal window:
> > > > 
> > > > perf: Floating point exception
> > > > Obtained 8 stack frames.
> > > > ./perf(dump_stack+0x2e) [0x510c89]
> > > > ./perf(sighandler_dump_stack+0x2e) [0x510d69]
> > > > /lib64/libc.so.6(+0x36a80) [0x7f9419588a80]
> > > > ./perf(perf_top__header_snprintf+0x208) [0x4f42c1]
> > > > ./perf() [0x453c09]
> > > > ./perf() [0x454ddb]
> > > > /lib64/libpthread.so.0(+0x736d) [0x7f941bc8c36d]
> > > > /lib64/libc.so.6(clone+0x3f) [0x7f9419662e1f]
> > > > Floating point exception (core dumped)
> > > > 
> > > > working on fix
> > > so my crash is caused by bogus resize code, I have it working with fix for
> > > memory corruption happening in SIGWINCH signal handler (attached)
> > > could you please check if that fixes the code for you?
> > 
> > Yes, this fixes the crash caused by resize.
> > 
> > But original crash I reported is still there. Issue seems to be with evsel
> > being NULL and we are trying to de-reference it somewhere inside
> > annotation_line__new().
> > 
> > Will try to spend more time on it.
> 
> right, I can see it now.. we are passing NULL as evsel in
> the top but does not check on that.. attached patch prevents
> the crash for me, but I'll need to double check if that's
> correct fix

I ended up with few other annotation fixes, could you please
check following branch if it's working for you?

  https://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf.git
  perf/fixes

thanks,
jirka