Closed Bug 1491537 Opened 6 years ago Closed 3 years ago

segfault at 0 ip af20d0a7 sp ae9b9080 error 6 in libxul.so[aed61000+6bb0000]

Categories

(Core :: Graphics, defect, P5)

60 Branch
x86
Linux
defect

Tracking

()

RESOLVED WORKSFORME

People

(Reporter: traumschuleriebau, Unassigned)

Details

Firefox 60.2.0esr segfaulted, syslog shows:

Sep 15 10:37:45 t43 kernel: [1075156.616000] Chrome_~dThread[19315]: segfault at 0 ip af20d0a7 sp ae9b9080 error 6 in libxul.so[aed61000+6bb0000]
Sep 15 10:37:46 t43 kernel: [1075157.839491] Chrome_~dThread[19578]: segfault at 0 ip af1940a7 sp ae940080 error 6 in libxul.so[aece8000+6bb0000]
Sep 15 10:37:46 t43 kernel: [1075159.792345] Chrome_~dThread[19538]: segfault at 0 ip af24a0a7 sp ae9f6080 error 6 in libxul.so[aed9e000+6bb0000]
Sep 15 10:37:52 t43 kernel: [1075166.289269] Chrome_~dThread[19559]: segfault at 0 ip af20e0a7 sp ae9ba080 error 6 in libxul.so[aed62000+6bb0000]

Users reporting the same:
https://bugs.launchpad.net/ubuntu/+source/firefox/+bug/1656065
https://bugzilla.redhat.com/show_bug.cgi?id=1560156
https://forums.linuxmint.com/viewtopic.php?t=266558
https://trac.torproject.org/projects/tor/ticket/27110
This report is not actionable because we need either exact steps how to reproduce the crash or we need a stack trace with debug symbols.
You can use a Mozilla.org build and use the integrated crash reporter and send a crash report. 

see https://developer.mozilla.org/en-US/docs/Mozilla/How_to_get_a_stacktrace_for_a_bug_report
Flags: needinfo?(traumschuleriebau)
Thanks for your fast answer! This is at least triggered by (simulating) I/O errors, will generate you a trace next week.
Flags: needinfo?(traumschuleriebau)
need to give it another week. only got Bus Error for lay tries removing the medium and could not reproduce a proper i/o error with fiu so far with:

fiu-run ./firefox-60.1.0esr/firefox &
export ffpid="$(pidof firefox)"
fiu-ctrl -c 'enable name=posix/io/*' $ffpid
Flags: needinfo?(traumschuleriebau)
I'm looking at bug reports in the File Handling component to determine if an action is needed. It seems that here we're waiting for more information in order to determine if this is an actionable bug. If the information is not available, I'd recommend closing the bug.
Thanks for the reminder! This is still waiting for info (by me or someone else).

STR
1. set up a partition with unfriendly fs
2. move firefox there and start it (or with fiu)
3. deny it file access and see it crash after some time

So far i was only able to reproduce Bus errors: https://trac.torproject.org/projects/tor/ticket/27110#comment:2
Priority: -- → P5
I can confirm this bug, but I can't say an exact step to reproduce, it simply occurs, while watching videos on youtube, while using web.whatsapp, it even occurred while having a freedesktop.org page open while an youtube video was being played.

Here is the segfault in dmesg log (Firefox Quantum 60.3.0esr (64-bit) from debian testing): 
[ 1533.633423] do_general_protection: 9 callbacks suppressed
[ 1533.633425] traps: Web Content[1928] general protection ip:7fd9c4f9b41d sp:7fff3ba0fd90 error:0 in libxul.so[7fd9c4f2a000+57e2000]
[ 3899.332815] Web Content[2536]: segfault at 800000 ip 00007f63b0c7e6d8 sp 00007ffc61ed6768 error 4 in libxul.so[7f63ad5d8000+57e2000]
[ 3899.332824] Code: 00 00 8b 47 0c 83 f8 01 76 20 83 f8 02 74 13 48 83 ef 08 e9 ca b2 00 00 66 2e 0f 1f 84 00 00 00 00 00 f3 c3 66 0f 1f 44 00 00 <48> 8b 36 e9 60 ff ff ff 8b 47 0c 83 f8 01 76 20 83 f8 02 74 13 48 


What I can say is that I'm using Ryzen and it's being a complete hell to use it under linux. Even my PCI-Ex gpus are failing because of the problems.
(In reply to allan4229 from comment #6)

> What I can say is that I'm using Ryzen and it's being a complete hell to use
> it under linux. Even my PCI-Ex gpus are failing because of the problems.

Possibly the same error message for multiple problems on the other links but I see something that may be similar to this.

For me with Ryzen (B450 chipset) + NVIDIA (GeForce 390.87) these type of segfault messages occur without any obvious pattern when using Firefox 63.0.1 compiled from source using GCC 8.2.0 - no such problems with the binary from https://archive.mozilla.org/pub/firefox/releases/. (this setup also has independent PCIe and MCE errors in dmesg).

After swapping out the graphics card for an AMD one (amdgpu driver) the compiled from source Firefox runs fine without segfaults (none of the other dmesg errors either).

(In reply to Chris Mayo from comment #7)

For me with Ryzen (B450 chipset) + NVIDIA (GeForce 390.87) these type of
segfault messages occur without any obvious pattern when using Firefox
63.0.1 compiled from source using GCC 8.2.0 - no such problems with the
binary from https://archive.mozilla.org/pub/firefox/releases/. (this setup
also has independent PCIe and MCE errors in dmesg).

After swapping out the graphics card for an AMD one (amdgpu driver) the
compiled from source Firefox runs fine without segfaults (none of the other
dmesg errors either).

Moving to gfx based on this, though it's unclear if this bug is still current...

Component: File Handling → Graphics
Product: Firefox → Core
Flags: needinfo?(traumschuleriebau)

Allan, traums,
Can you still reproduce this issue using newer versions?

Flags: needinfo?(traumschuleriebau)
Flags: needinfo?(allan4229)

Hello, I need to test it. But probably not.

Regarding the Ryzen problems, I had a defective chip that had to be replaced by AMD. Also, after the replacement I had to replace the RAM chips (Corsair) since they weren't following JEDEC specifications properly. This note is just for reference, since I mentioned Ryzen at the time.

I'll do further checkings and I'll report here later.

Assuming this is no longer a problem => WFM

Status: UNCONFIRMED → RESOLVED
Closed: 3 years ago
Flags: needinfo?(traumschuleriebau)
Flags: needinfo?(allan4229)
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.