Lars Volker has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/8612 )

Change subject: IMPALA-4132: Use -fno-omit-frame-pointer
......................................................................


Patch Set 2:

> Patch Set 2:
>
> I spoke to Gabor about this a few days ago. I agree it would be good to 
> validate it but we we didn't plan for that - the JIRA didn't have a 
> reproduction of a bad stacktrace scenario and the comments on the JIRA were 
> confident that it would be beneficial.
>
> We could try building a release binary, stripping it with strip --strip-debug 
> and inducing a crash. @Lars do you have any examples of crashes from 
> minidumps where we got a "bad" stacktrace?

I couldn't think of a particular case that I could share. However, Breakpad 
first tries to use call frame information that it had extracted from the DWARF 
info during symbol extraction.

 2  impalad!impala::HandleSignal [minidump.cc : 101 + 0x1e]
    rbx = 0x0000000000000000   rbp = 0x00007ffff99871f0
    rsp = 0x00007ffff99871e0   r12 = 0x00007ffff9987860
    r13 = 0x000000000dc50040   r14 = 0x0000000000000000
    r15 = 0x0000000000000000   rip = 0x0000000001a74959
    Found by: call frame info

If that fails, it tries to use the frame pointer:

 1  impalad!impala::SleepForMs(long) [time.cc : 30 + 0x1f]
    rbp = 0x00007f365e6b3aa0   rsp = 0x00007f365e6b3a80
    rip = 0x0000000001ac61e3
    Found by: previous frame's frame pointer

If that fails, it falls back to scanning the stack for addresses that look like 
return addresses:

 8  impalad!boost::detail::function::void_function_invoker0<void (*)(), 
void>::invoke(boost::detail::function::function_buffer&) [function_template.hpp 
: 112 + 0x6]
    rsp = 0x00007f365eeb4b00   rip = 0x000000000163f151
    Found by: stack scanning

Resolving a minidump written by a stripped release binary should not show 
"stack scanning" for Impala's frames.


--
To view, visit http://gerrit.cloudera.org:8080/8612
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Ib7d0d88ba015a847356ed0274fe91017b98cb941
Gerrit-Change-Number: 8612
Gerrit-PatchSet: 2
Gerrit-Owner: Gabor Kaszab <[email protected]>
Gerrit-Reviewer: Bharath Vissapragada <[email protected]>
Gerrit-Reviewer: Dan Hecht <[email protected]>
Gerrit-Reviewer: Gabor Kaszab <[email protected]>
Gerrit-Reviewer: Lars Volker <[email protected]>
Gerrit-Reviewer: Mostafa Mokhtar <[email protected]>
Gerrit-Reviewer: Philip Zeyliger <[email protected]>
Gerrit-Reviewer: Tim Armstrong <[email protected]>
Gerrit-Comment-Date: Thu, 07 Dec 2017 19:12:15 +0000
Gerrit-HasComments: No

Reply via email to