1318504 - Firefox crashes with 'BadAccess (attempt to access private resource denied)' with HWA enabled on Arch Linux

Status: RESOLVED DUPLICATE of bug 1271100

(Core :: Graphics, defect)

Description

[David Santiago]

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:50.0) Gecko/20100101 Firefox/50.0
Build ID: 20161115075801

Steps to reproduce:

This is happening in Arch Linux distro with firefox 50.0. No issues with chromium.

1. Go to https://www.sportisimo.cz/prodejny/ceska-republika/hlavni-mesto-praha/
2. Click on the map button, to show the google map
3. Move click in the the map to move it around




Actual results:

This will cause firefox to crash with the following error:
(firefox:10228): Gdk-ERROR **: The program 'firefox' received an X Window System error.
This probably reflects a bug in the program.
The error was 'BadAccess (attempt to access private resource denied)'.
  (Details: serial 8043 error_code 10 request_code 130 (MIT-SHM) minor_code 1)
  (Note to programmers: normally, X errors are reported asynchronously;
   that is, you will receive the error a while after causing it.
   To debug your program, run it with the GDK_SYNCHRONIZE environment
   variable to change this behavior. You can then get a meaningful
   backtrace from your debugger if you break on the gdk_x_error() function.)
fish: 'firefox' terminated by signal SIGTRAP (Trace or breakpoint trap)


Expected results:

No crash

Comment 1

[Emil Pasca [:emilpasca]]

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:53.0) Gecko/20100101 Firefox/53.0

I have tested this issue on Ubuntu 16.04 x64 with the latest Firefox release (50.0) and the latest Nightly (53.0a1-20161121030224) and could not to reproduce it.
After navigating to the link provided in the description and clicking on the map, when I move click in the map to move it around, the map is moved as expected, I have also managed to zoom in or zoom out without any issues.

Can you please try re-testing this with clean Firefox profile, maybe even safe mode, to eliminate custom settings as a possible cause (https://goo.gl/PNe90E).

Comment 2

[David Santiago]

With the safe mode (either by using the --safe-mode or by choosing it after the browser crashed) didn't solve. The browser kept crashing.

However if i delete the .mozilla folder, the problem doesn't happen anymore.

I'm checking the preferences and the settings to see what's different....

Comment 3

[Tim Ruffing]

Same here, Mozilla/5.0 (X11; Linux x86_64; rv:50.0) Gecko/20100101 Firefox/50.0.
It crashes in safe mode but not with a fresh profile.

Quoting myself from downstream (Arch Linux, https://bugs.archlinux.org/task/51881):
Go to https://openstreetmap.us/iD/release/ , click "edit" and zoom/move the map around. Crashes after less than one second for me.
I've tried to run it with GDK_SYNCHRONIZE but then (unfortunately?) the bug does not appear...

Comment 4

[Tim Ruffing]

It does NOT crash when I set layers.acceleration.force-enabled to true.

Then about:support shows for HW_COMPOSITING:
blocked by default: Acceleration blocked by platform
force_enabled by user: Force-enabled by pref

Comment 5

[David Santiago]

Hi Tim,

Thanks for your debugging. I can confirm that setting the option

layers.acceleration.force-enabled to true will stop the crash.

Thanks for the excelent debug. I was trying the options that were changed by the user and not finding anything...

Comment 6

[Daniel M. Capella]

I get the same error and crash (unrelated to maps) with a Stylish theme. More details available in the Arch Linux bug shared earlier.
https://github.com/stylish-userstyles/stylish/issues/302

layers.acceleration.force-enabled doesn't help there.

Comment 7

[Daniel M. Capella]

The setting is effective after restarting Firefox.

Comment 8

[David Santiago]

Unfortunately if i set that option a true, i'm unable to see youtube videos.

It keeps showing the first frame, mixed with the current frame.

So i have to set the option as false.

Comment 9

[Daniel M. Capella]

Something like this could help in the meantime:
https://addons.mozilla.org/en-us/firefox/addon/watch-with-mpv/

Comment 10

[David Santiago]

Thanks. It will work for now. There's also youtube-viewer app in the arch linux communit repo.

At the moment, it looks like something is really broke with layers.acceleration.force-enabled and probably with HW acceleration.

Comment 11

[Loic]

1) Do you have crash reports (bp-...) in the page about:crashes? If yes, provide some crash IDs.

2) Could you type about:support and paste the section "graphics".

3) As it could be a regression in FF50, could you install the tool Mozregresion (you need python 2.7) to narrow down a regression range. Run the command "mozregression --good=49" then copy here the final pushlog from the repository mozilla-inbound.
See http://mozilla.github.io/mozregression/ fo details.

Comment 12

[pingplug]

1) Arch build has an option "--disable-crashreporter", and there is no about:crashes page. see https://git.archlinux.org/svntogit/packages.git/tree/trunk/mozconfig?h=packages/firefox
2) 
{
  "graphics": {
    "numTotalWindows": 1,
    "numAcceleratedWindows": 0,
    "windowLayerManagerType": "Basic",
    "windowLayerManagerRemote": true,
    "supportsHardwareH264": "No",
    "currentAudioBackend": "pulse",
    "numAcceleratedWindowsMessage": [
      ""
    ],
    "adapterDescription": "Intel Open Source Technology Center -- Mesa DRI Intel(R) Ivybridge Mobile ",
    "adapterVendorID": "Intel Open Source Technology Center",
    "adapterDeviceID": "Mesa DRI Intel(R) Ivybridge Mobile ",
    "adapterRAM": "",
    "adapterDrivers": "",
    "driverVersion": "3.0 Mesa 13.0.1",
    "driverDate": "",
    "webglRenderer": "Intel Open Source Technology Center -- Mesa DRI Intel(R) Ivybridge Mobile ",
    "webgl2Renderer": "(no info)",
    "info": {
      "AzureCanvasBackend": "skia",
      "AzureCanvasAccelerated": 0,
      "AzureFallbackCanvasBackend": "none",
      "AzureContentBackend": "cairo",
      "CairoUseXRender": 0
    },
    "featureLog": {
      "features": [
        {
          "name": "HW_COMPOSITING",
          "description": "Compositing",
          "status": "blocked",
          "log": [
            {
              "type": "default",
              "status": "blocked",
              "message": "Acceleration blocked by platform"
            }
          ]
        },
        {
          "name": "OPENGL_COMPOSITING",
          "description": "OpenGL Compositing",
          "status": "unavailable",
          "log": [
            {
              "type": "default",
              "status": "unavailable",
              "message": "Hardware compositing is disabled"
            }
          ]
        }
      ],
      "fallbacks": []
    },
    "crashGuards": []
  }
}
3) I'll do it later.

Comment 13

[David Santiago]

I re-compiled firefox without the --disable-crashreporter however the crashes aren't getting reported...

On the about:crashes i get always "No crash reports have been submitted."

These are the configuration options:
--enable-application=browser --enable-update-channel=release --enable-rust --enable-system-hunspell --enable-gold --enable-official-branding --enable-pie --enable-release --enable-startup-notification --enable-system-ffi --enable-system-pixman --enable-system-sqlite --disable-updater --prefix=/usr --with-google-api-keyfile=/home/demanuel/PKGBUILD/firefox/src/firefox-50.0/google-api-key --with-mozilla-api-keyfile=/home/demanuel/PKGBUILD/firefox/src/firefox-50.0/mozilla-api-key --with-system-bz2 --with-system-icu --with-system-jpeg --with-system-libevent --with-system-libvpx --with-system-nspr --with-system-nss --with-system-zlib

What options should i use so the crash is reported?

Comment 14

[David Santiago]

These are the graphics options:

Graphics
--------

Features
Compositing: Basic
Asynchronous Pan/Zoom: none
WebGL Renderer: Intel Open Source Technology Center -- Mesa DRI Intel(R) HD Graphics 530 (Skylake GT2)
WebGL2 Renderer: (no info)
Hardware H264 Decoding: No
Audio Backend: pulse
GPU #1
Active: Yes
Description: Intel Open Source Technology Center -- Mesa DRI Intel(R) HD Graphics 530 (Skylake GT2)
Vendor ID: Intel Open Source Technology Center
Device ID: Mesa DRI Intel(R) HD Graphics 530 (Skylake GT2)
Driver Version: 3.0 Mesa 13.0.1

Diagnostics
AzureCanvasAccelerated: 0
AzureCanvasBackend: skia
AzureContentBackend: cairo
AzureFallbackCanvasBackend: none
CairoUseXRender: 0
Decision Log
HW_COMPOSITING:
blocked by default: Acceleration blocked by platform
OPENGL_COMPOSITING:
unavailable by default: Hardware compositing is disabled

Comment 15

[Daniel M. Capella]

Evangelos mentioned here the upstream builds have crash reporting:
https://bugs.archlinux.org/task/41994#comment127670

1) Like David, the crashes didn't generate any reports..

2)
Features
Compositing	OpenGL
Asynchronous Pan/Zoom	wheel input enabled; touch input enabled
WebGL Renderer	X.Org -- Gallium 0.4 on AMD PITCAIRN (DRM 2.46.0 / 4.8.10-1-zen, LLVM 3.9.0)
WebGL2 Renderer	(no info)
Hardware H264 Decoding	No
Audio Backend	pulse
GPU #1
Active	Yes
Description	X.Org -- Gallium 0.4 on AMD PITCAIRN (DRM 2.46.0 / 4.8.10-1-zen, LLVM 3.9.0)
Vendor ID	X.Org
Device ID	Gallium 0.4 on AMD PITCAIRN (DRM 2.46.0 / 4.8.10-1-zen, LLVM 3.9.0)
Driver Version	3.0 Mesa 13.0.1
Diagnostics
AzureCanvasAccelerated	0
AzureCanvasBackend	skia
AzureContentBackend	cairo
AzureFallbackCanvasBackend	none
CairoUseXRender	0
Decision Log
HW_COMPOSITING	
blocked by default: Acceleration blocked by platform
force_enabled by user: Force-enabled by pref

3) Couldn't reproduce with `mozregression --good=49 --bad=50` o.O.. Will try to give it another shot.

Comment 16

[David Santiago]

3) I'm unable to reproduce it with `mozregression --good=49` or even with `mozregression --good=48`


Was this nightly build good, bad, or broken? (type 'good', 'bad', 'skip', 'retry' or 'exit' and press Enter): good
 3:05.86 ERROR: Build was expected to be bad! The initial good/bad range seems incorrect.



This also seem to be related with https://bugzilla.mozilla.org/show_bug.cgi?id=1317994 because since when i set to true the option layers.acceleration.force-enabled i'm unable to watch youtube videos.

Comment 17

[Tim Ruffing]

This is getting annoying, because the crash is somewhat random (I guess related to concurrency).
Some observations when trying mozregression:
 - I've managed to reproduce it with a somewhat fresh profile too but it's much more unlikely.
 - In mozregression I've managed to reproduce it only when force-enabled is ON, which seems to confirm that it's rather non-deterministic. It seems to depend on the interplay of all the options.
 - I've never seen a crash with e10s enabled.
 - The first revision I could crash was the following.


 0:16.11 INFO: Running mozilla-central build built on 2016-01-02 00:20:50.591000, revision c9ed86a8
 0:27.14 INFO: Launching /tmp/tmpIZXgqS/firefox/firefox
 0:27.14 INFO: Application command: /tmp/tmpIZXgqS/firefox/firefox -profile /tmp/tmpxVxiD6
 0:27.14 INFO: application_buildid: 20160101150640
 0:27.14 INFO: application_changeset: c9ed86a8b757e783fa14c2afcfff4a61e0d7b574
 0:27.14 INFO: application_name: Firefox
 0:27.14 INFO: application_repository: https://hg.mozilla.org/mozilla-central
 0:27.14 INFO: application_version: 46.0a1
Was this inbound build good, bad, or broken? (type 'good', 'bad', 'skip', 'retry' or 'exit' and press Enter):  0:29.11 INFO: 1479991866175	addons.update-checker	WARN	Update manifest for [e-mail] did not contain an updates property
 1:52.46 INFO: ATTENTION: default value of option force_s3tc_enable overridden by environment.
 1:57.39 INFO: ATTENTION: default value of option force_s3tc_enable overridden by environment.
 2:14.36 INFO: 
 2:14.36 INFO: (firefox:23080): Gdk-ERROR **: The program 'firefox' received an X Window System error.
 2:14.36 INFO: This probably reflects a bug in the program.
 2:14.36 INFO: The error was 'BadAccess (attempt to access private resource denied)'.
 2:14.36 INFO:   (Details: serial 77652 error_code 10 request_code 130 (MIT-SHM) minor_code 1)
 2:14.36 INFO:   (Note to programmers: normally, X errors are reported asynchronously;
 2:14.36 INFO:    that is, you will receive the error a while after causing it.
 2:14.36 INFO:    To debug your program, run it with the GDK_SYNCHRONIZE environment
 2:14.36 INFO:    variable to change this behavior. You can then get a meaningful
 2:14.36 INFO:    backtrace from your debugger if you break on the gdk_x_error() function.)

 2:14.89 WARNING: Process exited with code -133

Comment 18

[pingplug]

3) I can't reproduce this bug so far.

I can reproduce it with firefox 50.0, but I can't reproduce it with 50.0b1 so far.
I'm building firefox 50.0b6...

Comment 19

[pingplug]

Firefox 50.0b6 crashed and give more message:

(firefox:13496): Gdk-ERROR **: The program 'firefox' received an X Window System error.
This probably reflects a bug in the program.
The error was 'BadAccess (attempt to access private resource denied)'.
  (Details: serial 4249 error_code 10 request_code 130 (MIT-SHM) minor_code 1)
  (Note to programmers: normally, X errors are reported asynchronously;
   that is, you will receive the error a while after causing it.
   To debug your program, run it with the GDK_SYNCHRONIZE environment
   variable to change this behavior. You can then get a meaningful
   backtrace from your debugger if you break on the gdk_x_error() function.)
[NPAPI 13589] ###!!! ABORT: Aborting on channel error.: file /home/pingplug/firefox-50b6/src/firefox-50.0b6/ipc/glue/MessageChannel.cpp, line 2056
[NPAPI 13589] ###!!! ABORT: Aborting on channel error.: file /home/pingplug/firefox-50b6/src/firefox-50.0b6/ipc/glue/MessageChannel.cpp, line 2056

I'll try 50.0b3

Comment 20

[pingplug]

50.0b3 seems OK
50.0b4 seems OK
50.0b5 seems OK
50.0b6 crash
I think it's between firefox 50.0b5 and firefox 50.0b6

Comment 21

[Loic]

Thanks for the regression range. Can you open about:buildconfig for 50.0b5 and 50.0b6 and copy here the link from the line "Built from".

Comment 22

[pingplug]

(In reply to Loic from comment #21)
> Thanks for the regression range. Can you open about:buildconfig for 50.0b5
> and 50.0b6 and copy here the link from the line "Built from".

There is no "Built from" line with Arch's build config. I use the source from:

Firefox 50.0b5:
https://ftp.mozilla.org/pub/firefox/releases/50.0b5/source/firefox-50.0b5.source.tar.xz

Firefox 50.0b6:
https://ftp.mozilla.org/pub/firefox/releases/50.0b6/source/firefox-50.0b6.source.tar.xz

Comment 23

[David Santiago]

(In reply to Loic from comment #21)
> Thanks for the regression range. Can you open about:buildconfig for 50.0b5
> and 50.0b6 and copy here the link from the line "Built from".

Besides that is there any other info you require?

Comment 24

[Loic]

(In reply to David Santiago from comment #23)
> (In reply to Loic from comment #21)
> > Thanks for the regression range. Can you open about:buildconfig for 50.0b5
> > and 50.0b6 and copy here the link from the line "Built from".
> 
> Besides that is there any other info you require?

In the page about:support, there is a line about build configuration.

Comment 25

[David Santiago]

This is all the information that shows in about:buildconfig:

about:buildconfig

Build platform
target
x86_64-pc-linux-gnu

Build tools
Compiler 	Version 	Compiler flags

/usr/bin/gcc -std=gnu99 	6.2.1 	-Wall -Wempty-body -Wignored-qualifiers -Wpointer-arith -Wsign-compare -Wtype-limits -Wunreachable-code -Wno-error=maybe-uninitialized -Wno-error=deprecated-declarations -Wno-error=array-bounds -march=x86-64 -mtune=generic -O2 -pipe -fstack-protector-strong -fno-strict-aliasing -ffunction-sections -fdata-sections -fno-math-errno -pthread -pipe

/usr/bin/g++ -std=gnu++11 	6.2.1 	-D_FORTIFY_SOURCE=2 -O2 -Wall -Wc++11-compat -Wempty-body -Wignored-qualifiers -Woverloaded-virtual -Wpointer-arith -Wsign-compare -Wtype-limits -Wunreachable-code -Wwrite-strings -Wc++14-compat -Wno-invalid-offsetof -Wno-error=maybe-uninitialized -Wno-error=deprecated-declarations -Wno-error=array-bounds -march=x86-64 -mtune=generic -O2 -pipe -fstack-protector-strong -fno-delete-null-pointer-checks -fno-schedule-insns2 -fno-exceptions -fno-strict-aliasing -fno-rtti -ffunction-sections -fdata-sections -fno-exceptions -fno-math-errno -pthread -pipe -g -freorder-blocks -Os -fomit-frame-pointer -D_FORTIFY_SOURCE=2 -O2


Configure options

--enable-application=browser --enable-update-channel=release --enable-rust --enable-system-hunspell --disable-crashreporter --enable-gold --enable-official-branding --enable-pie --enable-release --enable-startup-notification --enable-system-ffi --enable-system-pixman --enable-system-sqlite --disable-updater --prefix=/usr --with-google-api-keyfile=/build/firefox/src/firefox-50.0/google-api-key --with-mozilla-api-keyfile=/build/firefox/src/firefox-50.0/mozilla-api-key --with-system-bz2 --with-system-icu --with-system-jpeg --with-system-libevent --with-system-libvpx --with-system-nspr --with-system-nss --with-system-zlib

Comment 27

[Peter]

(In reply to Tim Ruffing from comment #4)
> It does NOT crash when I set layers.acceleration.force-enabled to true.
> 
> Then about:support shows for HW_COMPOSITING:
> blocked by default: Acceleration blocked by platform
> force_enabled by user: Force-enabled by pref

With FF 5.0.2-1 on Arch Linux this fix works for embedded maps but NOT with maps/streetview in their own tab. If I revert the setting to 'false' then the conditions are reversed - crashes with embedded maps but not with maps in their own tab.

Comment 28

[Loic]

I think it's a duplicate of bug 1271100 which if fixed in 50.1.0 and later.

Peter, could you download Beta 51 and confirm it's fixed in all the cases you hit the crash, please.
https://www.mozilla.org/en-US/firefox/beta/all/

Comment 29

[Daniel M. Capella]

51.0b6 fixes my issue (crash with Stylish theme). Thank you.

Comment 30

[Peter]

51.0b6 fixes the issues with Street View and embedded maps regardless of the setting o layers.acceleraion.force-enabled.

Thanks

Comment 31

[Peter]

Forgot to clear the needinfo flag...

Comment 32

[Loic]

Ok, 50.1.0 is the next release, so it should be fixed tomorrow.

Comment 33

[David Santiago]

Since i opened this, i just want to inform that 50.1.0 fixed this issue.

Thank you all.