Closed Bug 1728476 Opened 3 years ago Closed 3 years ago

Crash in [@ @0x0 | iHD_drv_video.so@0x42d3ba]

Categories

(Core :: Audio/Video, defect, P5)

Firefox 88
x86_64
Linux
defect

Tracking

()

RESOLVED DUPLICATE of bug 1619585

People

(Reporter: ihor.chelnokov, Unassigned)

Details

Crash Data

Crash report: https://crash-stats.mozilla.org/report/index/f2731267-aaf4-474d-a6db-4da9a0210831

Reason: SIGSEGV /SEGV_MAPERR

Top 7 frames of crashing thread:

0  @0x0 
1 iHD_drv_video.so iHD_drv_video.so@0x42d3ba 
2 iHD_drv_video.so iHD_drv_video.so@0x423399 
3 libigdgmm.so.11 libigdgmm.so.11@0x6637f 
4 libigdgmm.so.11 libigdgmm.so.11@0x66a1f 
5 libigdgmm.so.11 libigdgmm.so.11@0x7349f 
6 libigdgmm.so.11 libigdgmm.so.11@0x734df 

This is on ThinkPad E14 Gen2 (i5-1135G7 w/ Iris Xe Graphics).
Running on Debian 11/Bullseye, GNOME DE, with firefox, libigdgmm11 and intel-media-va-driver from Unstable.

This appears to be one single occurrence of this crash signature whioch happened on Firefox v88.0.1.

  1. Do you know what might have caused it to happen? What were you doing right before it happened?
  2. Has it ever occurred before or after this report? It is recurrent or a one-time occurance?
  3. Can you update the browser to the latest current live version (v92.0.2) and retest?

Thank you for your contribution, ihor.chelnokov!

Flags: needinfo?(ihor.chelnokov)
  1. I believe it happens due to hardware acceleration with Intel's Iris Xe Graphics. I have a laptop with a much older CPU (7th or 8th gen, don't remember) and it works fine in a very similar setup.
  2. I can reliably reproduce this on Reddit (guaranteed to crash withing seconds of opening a thread on new design), sometimes this affects Google search and even Jira.
  3. I can't update, since I'm using the version from Debian Unstable, but I can back up settings and try running the one from official tarball, but I only see 91.0.2.
Flags: needinfo?(ihor.chelnokov)

I have tested version 91.0.2 and it seems to work. Now gotta wait for it to arrive to Unstable repo...

Good evening,

I just installed 91.0.2 from Debian Experimental and I can still reproduce this crash. I ran it from a terminal, so here's the output:

libva info: VA-API version 1.12.0
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/iHD_drv_video.so
libva info: Found init function __vaDriverInit_1_12
Sandbox: seccomp sandbox violation: pid 26147, tid 26261, syscall 64, args 1140872792 1 1974 139969145340120 0 0.
Sandbox: seccomp sandbox violation: pid 26147, tid 26261, syscall 64, args 1140872792 1 438 139969145340120 0 0.

Here's the crash report and here's link to Reddit post that caused the crash.

Confirming based on the latest crash reports. This occurs on Release v91.0.1. Don't even have a flag to set as affected so I won't set any of them.
Furthermore, I will set this component since it's related to video drivers, but it may also be related to memory.
Please set a more appropriate component if this is incorrect. Thank you!

Status: UNCONFIRMED → NEW
Component: General → Audio/Video
Ever confirmed: true
Product: Firefox → Core

The crash report shows it's on the decoder thread so it looks like VAAPI related.

Severity: -- → S3
Priority: -- → P5

(In reply to ihor.chelnokov)

Running on Debian 11/Bullseye
intel-media-va-driver
syscall 64

That's bug 1619585.
Temporary solution:
apt-get remove intel-media-va-driver intel-media-va-driver-non-free
apt-get install i965-va-driver

No longer blocks: egl-linux-vaapi
Status: NEW → RESOLVED
Closed: 3 years ago
Resolution: --- → DUPLICATE

(Darkspirit from bug 1683808 comment 26)

The Arch wiki should unambiguously state that disabling the content process sandbox means disabling the protection against attackers. Don't ever disable the content process sandbox. If iHD (intel-media-driver) causes a crash (sandbox violation), then uninstall it and install i965 (intel-vaapi-driver == i965-va-driver). It's already a risk to use VAAPI in the content process, please don't do the dirtiest workaround.
One needs the sandbox more than one needs VAAPI: If your hardware forces you to use the crashing iHD driver, then better wait until VAAPI has been moved into the GPU process - after a GPU process has been implemented for Wayland.

(In reply to ihor.chelnokov from comment #0)

Iris Xe Graphics

According to another user your hardware might only be supported by the iHD driver.
Then you can only decide between

  • waiting for the implementation of a Wayland GPU process (the right thing to do) or
  • entirely disabling the sandbox, your protection against attackers (don't do this).

(Esokrarkose from bug 1683808 comment #23)

To me it looks like https://bugzilla.mozilla.org/show_bug.cgi?id=1619585#c26 indeed describes to problem. Since I have the new Intel Iris Xe, I can't make use of the older i965 driver, thus I'm forced to the new iHD driver.

(In reply to Darkspirit from comment #9)

(In reply to ihor.chelnokov from comment #0)
According to another user your hardware might only be supported by the iHD driver.
Then you can only decide between

  • waiting for the implementation of a Wayland GPU process (the right thing to do) or
  • entirely disabling the sandbox, your protection against attackers (don't do this).

Alright, guess I'm waiting...

You need to log in before you can comment on or make changes to this bug.