Re: perf script doesn't dump a normal call trace

[Jiri Olsa <jolsa@redhat.com>]

On Sat, Nov 03, 2018 at 06:36:21PM +0900, Xin Long wrote:
> On Fri, Nov 2, 2018 at 7:26 PM Jiri Olsa <jolsa@redhat.com> wrote:
> >
> > On Fri, Nov 02, 2018 at 03:36:13PM +0900, Xin Long wrote:
> > > On upstream kernel(4.19) or RHEL-8 kernel(4.18.0):
> > >
> > > # perf record -e 'skb:consume_skb' -ag
> > > ^C[ perf record: Woken up 1 times to write data ]
> > > [ perf record: Captured and wrote 0.612 MB perf.data (634 samples) ]
> > >
> > > # perf script
> > > swapper     0 [009] 274370.117711: skb:consume_skb: skbaddr=0xffff962c591d5b00
> > >         ffffffffa4abe534 consume_skb+0x64 ([kernel.kallsyms])
> > >
> > > kworker/9:1-eve   926 [009] 274370.117729: skb:consume_skb:
> > > skbaddr=0xffff962c591d5b00
> > >         ffffffffa4abe534 consume_skb+0x64 ([kernel.kallsyms])
> > >
> > > kworker/9:1-eve   926 [009] 274370.117732: skb:consume_skb:
> > > skbaddr=0xffff962c591d4900
> > >         ffffffffa4abe534 consume_skb+0x64 ([kernel.kallsyms])
> > >
> > > swapper     0 [009] 274370.145528: skb:consume_skb: skbaddr=0xffff962c591d4900
> > >         ffffffffa4abe534 consume_skb+0x64 ([kernel.kallsyms])
> > >
> > > kworker/9:1-eve   926 [009] 274370.145545: skb:consume_skb:
> > > skbaddr=0xffff962c591d4900
> > >         ffffffffa4abe534 consume_skb+0x64 ([kernel.kallsyms])
> > >
> > > kworker/9:1-eve   926 [009] 274370.145547: skb:consume_skb:
> > > skbaddr=0xffff962c591d5b00
> > >         ffffffffa4abe534 consume_skb+0x64 ([kernel.kallsyms])
> > >
> > > swapper     0 [009] 274370.173443: skb:consume_skb: skbaddr=0xffff962c591d5b00
> > >         ffffffffa4abe534 consume_skb+0x64 ([kernel.kallsyms])
> >
> > I can see that on upstream as well, probably something within
> > tracepoint entry code, because I have it working nicely when
> > from kprobe, like:
> >
> > [root@ibm-x3650m4-02 perf]# ./perf probe 'consume_skb'
> > [root@ibm-x3650m4-02 perf]# ./perf record -g -e probe:consume_skb* -aR ^C
> >
> Thanks Jiri,
> 
> My debugging script is using tracepoint with some filters which I
> don't think probe can support.
> Any one have fixes for this tracepoint issue?

trying to bisect that.. looks like orc code issue,
it works for me when you switch to fp unwind:
  CONFIG_UNWINDER_FRAME_POINTER

jirka