New issue
Advanced search Search tips

Issue 744622 link

Starred by 1 user

Issue metadata

Status: Untriaged
Owner: ----
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Linux
Pri: 3
Type: Bug



Sign in to add a comment

Stack trace with no useful information

Project Member Reported by mlamouri@chromium.org, Jul 17 2017

Issue description

I've got a stacktrace that looks like this:
```
Received signal 11 SEGV_ACCERR 2f39816ab998
#0 0x7f4a75bb670b base::debug::StackTrace::StackTrace()
#1 0x7f4a75bb540c base::debug::StackTrace::StackTrace()
#2 0x7f4a75bb621f base::debug::(anonymous namespace)::StackDumpSignalHandler()
#3 0x7f4a76053330 <unknown>
#4 0x2f39816ab998 <unknown>
  r8: fffffffffffffed8  r9: fffffffffffffec8 r10: fffffffffffffeb8 r11: 00007f4a5b092110
 r12: 0000564a42761b9b r13: 00007ffd0f954f30 r14: 0000000000000000 r15: 0000000000000000
  di: 00007ffd0f950070  si: 00007ffd0f94ff08  bp: 00007ffd0f94ff40  bx: 0000000000000000
  dx: 0000000000000018  ax: 00002f39816ab998  cx: 00002f39816ab998  sp: 00007ffd0f94fed8
  ip: 00002f39816ab998 efl: 0000000000010202 cgf: 0000000000000033 erf: 0000000000000015
 trp: 000000000000000e msk: 0000000000000000 cr2: 00002f39816ab998
[end of stack trace]
```

It is a crash in the renderer process on a debug build with symbols. It usually works fine but for some reasons, this crash doesn't play well.

This are my gn args:
```
is_component_build = true
enable_nacl = false
is_debug = true
use_goma = true
is_clang = true
use_gold = true
symbol_level = 1

media_use_ffmpeg = true
media_use_libvpx = true
proprietary_codecs = true
ffmpeg_branding = "Chrome"
```

Following primiano@'s advices, I've forced `use_custom_libcxx` to false in the gni file where it is set but it has no effect. Though, he believed that it might be an issue in our stack trace code.

FWIW, gdb provides a readable stack.
 
Cc: thestig@chromium.org jamescook@chromium.org
given that use_custom_libcxx didn't solve the issue, I'd -thomasanderson@chromium.org as my speculation was clearly wrong. :)

Adding some folks who might have any inputs on the in-process stack walking.
Cc: -thomasanderson@chromium.org
Components: -Build Internals>CrashReporting
Sorry, I don't know.

Aside: You probably don't need is_clang and use_gold in your GN args. I doubt it will make a difference, but it might be a worth a try without them.

On Linux, the only GN configs I set are:

is_debug = true
use_goma = true
is_component_build = true

And when I visit about:crash, I get:

Received signal 11 SEGV_MAPERR 000000000000
#0 0x7f69122a378d base::debug::StackTrace::StackTrace()
#1 0x7f69122a1b5c base::debug::StackTrace::StackTrace()
#2 0x7f69122a3145 base::debug::(anonymous namespace)::StackDumpSignalHandler()
#3 0x7f6912902330 <unknown>
#4 0x7f690d2379c3 content::CrashIntentionally()
#5 0x7f690d237ca0 content::MaybeHandleDebugURL()
#6 0x7f690d26fcc1 content::RenderFrameImpl::PrepareRenderViewForNavigation()
#7 0x7f690d24fbb6 content::RenderFrameImpl::NavigateInternal()
#8 0x7f690d247051 content::RenderFrameImpl::OnNavigate()
#9 0x7f690d28dda1 _ZN4base20DispatchToMethodImplIPN7content15RenderFrameImplEMS2_FvRKNS1_22CommonNavigationParamsERKNS1_21StartNavigationParamsERKNS1_23RequestNavigationParamsEERKNSt3__15tupleIJS4_S7_SA_EEEJLm0ELm1ELm2EEEEvRKT_T0_OT1_NS_13IndexSequenceIJXspT2_EEEE
...
#47 0x55ffbdc4c632 main
#48 0x7f68fc44bf45 __libc_start_main
#49 0x55ffbdc4c514 <unknown>
  r8: fffffffffffffed8  r9: fffffffffffffec8 r10: fffffffffffffeb8 r11: 00007f68fc5b4110
...

Which seems good enough for me.

Sign in to add a comment