Chromium-browser armhf crashes with "InitializeSandbox() called with multiple threads in process gpu-process"

Bug #1563184 reported by David Burke on 2016-03-29
280
This bug affects 51 people
Affects Status Importance Assigned to Milestone
chromium-browser (Arch Linux)
Confirmed
Undecided
Unassigned
chromium-browser (Ubuntu)
High
Unassigned

Bug Description

crashes on startup with

[2457:2457:0322/182419:ERROR:sandbox_linux.cc(334)] InitializeSandbox() called with multiple threads in process gpu-process
Segmentation fault

System ODROID XU3 armhf

Nonny Moose (moosenonny10) wrote :

I have had almost the exact same problem on a Raspberry Pi 2 running Ubuntu Mate 15.10.
uname -a:
Linux RaspberryPi 3.18.0-20-rpi2 #21-Ubuntu SMP PREEMPT Sun Apr 5 01:56:02 UTC 2015 armv7l armv7l armv7l GNU/Linux

$ chromium-browser
[2815:2815:0403/092234:ERROR:sandbox_linux.cc(334)] InitializeSandbox() called with multiple threads in process gpu-process
Segmentation fault (core dumped)

Changed in chromium-browser (Ubuntu):
status: New → Confirmed
Aworan (aworan) wrote :

Same problem here with raspberry pi 2 or raspberry pi 3 under ubuntu mate and lubuntu 15.10.
chromium-browser works before version 49.
If I go back to version 48 it works.
The bug is with armhf architecture package !
The bug is confirmed by ubuntu mate for raspberry pi community users, see :
https://ubuntu-mate.community/t/chromium-crashes-when-starting-segfaults/4578/14

Ross Holder (rossholder) wrote :

I too am having the same issue as described above.

William Woods (shatnerps) wrote :

Confirming that I am having the same issue on both an Orange Pi and a ODroid running Ubuntu Mate 15.10.

Chazall1 (lishdigg) wrote :

Confirming that I am having the same issue on R-Pi 3 . Running Ubuntu Mate 16.04.

Joao Driessen (joaodriessen) wrote :

Problem also exists in archlinux arm.

Nonny Moose (moosenonny10) wrote :

I upgraded my Raspberry Pi 2 to Ubuntu Mate 16.10 LTS and Chromium continues to segfault when launched from the menu, but when I launched it from the terminal it opened a completely transparent window that proceeded to do absolutely nothing. When I finally got impatient and ^Ced it, <i>then</i> it had a segfault and printed out a stack trace, which was essentially garbage.

Nonny Moose (moosenonny10) wrote :

I have had several other bugs have this error:
``

I think that the name should be changed to be more descriptive:

`Chromium-browser armhf crashes with "InitializeSandbox() called with multiple threads in process gpu-process"`

summary: - InitializeSandbox() called with multiple threads in process gpu-process
+ Chromium-browser armhf crashes with "InitializeSandbox() called with
+ multiple threads in process gpu-process"

Same issue. RPi3

Remis (plremis) wrote :

Same issue, RPi3

Tianon Gravi (tianon) wrote :

Hit the same issue on my own new Raspberry Pi 3 running Ubuntu Xenial (installed via http://www.finnie.org/software/raspberrypi/ubuntu-rpi3/ubuntu-16.04-preinstalled-server-armhf+raspi3.img.xz linked from https://wiki.ubuntu.com/ARM/RaspberryPi).

I managed to get a little bit further by installing the "chromium-browser" package from "trusty" (which appears to be the same underlying Chromium version backported; more exactly, package version "49.0.2623.108-0ubuntu0.14.04.1.1113").

That at least lets the browser come up, but all the pages crash (as opposed to the entire browser crashing, like the xenial version does).

When I add "--no-sandbox" (which is inherently dangerous and adds an appropriate warning to the UI every time you launch chrome), I can get much further and actually have a functional Chromium instance (but using "--no-sandbox" is obviously pretty far from ideal).

Same here: Debian 8, Mate 1.8.1

Tianon Gravi (tianon) wrote :

The debdiff from xenial to trusty (attached) doesn't appear terribly interesting. :(

I try Google Chrome OS for SBC ...

From this link

http://www.chromiumosforsbc.org/cros_rpi3-v05-notes

They able to fix html5 video driver and can play youtube without any issues
at all.

Maybe you need to use that OS to debug and fix the issues in ubuntu mate...

They even use a new kernel for it...
On 18 May 2016 11:01 pm, "Tianon Gravi" <email address hidden> wrote:

The debdiff from xenial to trusty (attached) doesn't appear terribly
interesting. :(

** Patch added: "xenial-to-trusty.debdiff"

https://bugs.launchpad.net/ubuntu/+source/chromium-browser/+bug/1563184/+attachment/4665562/+files/xenial-to-trusty.debdiff

--
You received this bug notification because you are subscribed to a
duplicate bug report (1576079).
https://bugs.launchpad.net/bugs/1563184

Title:
  Chromium-browser armhf crashes with "InitializeSandbox() called with
  multiple threads in process gpu-process"

Status in chromium-browser package in Ubuntu:
  Confirmed

Bug description:
  crashes on startup with

  [2457:2457:0322/182419:ERROR:sandbox_linux.cc(334)] InitializeSandbox()
called with multiple threads in process gpu-process
  Segmentation fault

  System ODROID XU3 armhf

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/chromium-browser/+bug/1563184/+subscriptions

Will Glynn (willglynn) wrote :
Download full text (7.9 KiB)

I encountered this on a Raspberry Pi 2 running a minimal Ubuntu 16.04. The initial install was based on ubuntu-16.04-preinstalled-server-armhf+raspi2.img.xz from https://wiki.ubuntu.com/ARM/RaspberryPi.

Starting chromium-browser 50.0.2661.102-0ubuntu0.16.04.1 (armhf) with no args, I get:

    $ chromium-browser

    (chromium-browser:9976): GLib-GIO-CRITICAL **: g_settings_schema_source_lookup: assertion 'source != NULL' failed
    [10031:10031:0526/000302:ERROR:sandbox_linux.cc(334)] InitializeSandbox() called with multiple threads in process gpu-process

    (chromium-browser:9976): GConf-WARNING **: Client failed to connect to the D-BUS daemon:
    /usr/bin/dbus-launch terminated abnormally without any error message
    chro: pthread_getattr_np.c:71: pthread_getattr_np: Assertion `abs (thread->pid) == thread->tid' failed.
    Received signal 4 <unknown> 0000721733e6
    #0 0x000076d9fe1a base::debug::StackTrace::StackTrace()
    #1 0x000076da0104 <unknown>
    #2 0x000072173260 <unknown>
    #3 0x0000721733e6 abort
    [end of stack trace]
    Segmentation fault

If I allow it to dump core, I get a backtrace:

    $ ulimit -c unlimited
    $ chromium-browser

    (chromium-browser:13589): GLib-GIO-CRITICAL **: g_settings_schema_source_lookup: assertion 'source != NULL' failed

    (chromium-browser:13589): GConf-WARNING **: Client failed to connect to the D-BUS daemon:
    /usr/bin/dbus-launch terminated abnormally without any error message
    chro: pthread_getattr_np.c:71: pthread_getattr_np: Assertion `abs (thread->pid) == thread->tid' failed.
    Received signal 4 <unknown> 0000721bc3e6
    #0 0x000076de8e1a base::debug::StackTrace::StackTrace()
    #1 0x000076de9104 <unknown>
    #2 0x0000721bc260 <unknown>
    #3 0x0000721bc3e6 abort
    [end of stack trace]
    [13645:13645:0526/002513:ERROR:sandbox_linux.cc(334)] InitializeSandbox() called with multiple threads in process gpu-process
    Segmentation fault (core dumped)

    $ gdb /usr/lib/chromium-browser/chromium-browser core
    GNU gdb (Ubuntu 7.11-0ubuntu1) 7.11
    …
    Core was generated by `chromium-browser --enable-pinch '.
    Program terminated with signal SIGSEGV, Segmentation fault.
    #0 0x72426bc6 in ?? () from /usr/lib/arm-linux-gnueabihf/libstdc++.so.6
    [Current thread is 1 (Thread 0x673c8230 (LWP 13616))]
    (gdb) bt
    #0 0x72426bc6 in ?? () from /usr/lib/arm-linux-gnueabihf/libstdc++.so.6
    #1 0x5509389c in std::_Rb_tree_iterator<std::pair<unsigned long long const, int> > std::_Rb_tree<unsigned long long, std::pair<unsigned long long const, int>, std::_Select1st<std::pair<unsigned long long const, int> >, std::less<unsigned long long>, std::allocator<std::pair<unsigned long long const, int> > >::_M_emplace_hint_unique<std::piecewise_construct_t const&, std::tuple<unsigned long long const&>, std::tuple<> >(std::_Rb_tree_const_iterator<std::pair<unsigned long long const, int> >, std::piecewise_construct_t const&, std::tuple<unsigned long long const&>&&, std::tuple<>&&) ()
    #2 0x553ea784 in ?? ()
    #3 0x553eec6c in ?? ()
    #4 0x54f04afa in ?? ()
    #5 0x741be894 in net::URLRequestHttpJob::StartTransactionInter...

Read more...

Liam Proven (lproven) wrote :

Replicated on Lubuntu 16.04 on RasPi 3. Fully updated, Chromium 50 segfaults.

--no-sandbox does not work; still segfaults.

As per this page:
https://ubuntu-mate.community/t/chromium-crashes-when-starting-segfaults/4578

I also tried --disable-pinch. Doesn't work; still segfaults.

Nonny Moose (moosenonny10) wrote :

I was aware that the InitializeSandbox error was not related to the issue because I get that from like half of my applications. XD
The chromium-browser <i>script</i> that starts chromium-browser supports the flag --no-touch-pinch, which I find a little odd because it essentially turns it into --disable-pinch when it sends the flags to the binary itself.

mordant (mordant23) wrote :

To provide more information, the specific stacktrace that breaks the sandbox is:
[2781:2861:0529/230241:ERROR:zygote_host_impl_linux.cc(162)] Failed to adjust OOM score of renderer with pid 2956: Permission denied
[2781:2861:0529/230241:ERROR:zygote_host_impl_linux.cc(162)] Failed to adjust OOM score of renderer with pid 2997: Permission denied
[2781:2861:0529/230241:ERROR:zygote_host_impl_linux.cc(162)] Failed to adjust OOM score of renderer with pid 2998: Permission denied
[2781:2861:0529/230241:ERROR:zygote_host_impl_linux.cc(162)] Failed to adjust OOM score of renderer with pid 3000: Permission denied
[2781:2861:0529/230241:ERROR:zygote_host_impl_linux.cc(162)] Failed to adjust OOM score of renderer with pid 3004: Permission denied
Segmentation fault (core dumped)
bbliveservice@bbliveservice-desktop:~$ [2948:2948:0529/230244:ERROR:sandbox_linux.cc(334)] InitializeSandbox() called with multiple threads in process gpu-process

This indicates it may have to do with https://bugs.chromium.org/p/chromium/issues/detail?id=576409

Davide Alberelli (dadexix86) wrote :

Same here with 16.04 on RaspberryPi 3 and Chromium 51.0.2704.79-0ubuntu0.16.04.1.1242.

Chad Miller (cmiller) wrote :

Reproduced. Thanks, all. I don't need more me-toos for now.

Changed in chromium-browser (Ubuntu):
importance: Undecided → High
assignee: nobody → Chad Miller (cmiller)
Nonny Moose (moosenonny10) wrote :

@cmiller If you need any impersonal information about the crash on my system, I would be happy to porvide it.

bandyt (dantololoi) wrote :

I also reached to this ticket and I will be happy of any progress. Thanks!

Francesco Strappini (fstraps) wrote :

Same here with Xubuntu 16.04 on RPi 3

I'm also subscribing. Just bought my first Raspberry Pi 3b, and installed Lubuntu 16.04 on it, and I have the exact same error message when trying to start Chrome.

Tony U. (tugando) wrote :

Man, this bug is a big show stopper for me. I use Chromium on my RaspPi-3 to turn my local printer into a cloud ready printer, and I have not been able to use this feature since this bug came about.

Russell Faull (rfaull) wrote :

@Tony U Workaround using older version of Chromium for RPI is at https://ubuntu-mate.community/t/trying-to-fix-chromium-issues/9235/10

XJDHDR (xjdhdr) wrote :

I was experiencing this bug as well on my Odroid XU3.
Eventually, I switched to Firefox and that works fine.

Henry Wertz (hwertz10) wrote :

I wonder if this is a compiler error? Can anyone tell if Chromium for Ubuntu 14.04 and Chromium for 16.04 have different build flags, or if their configure scripts detect something different? To me, it seems like if they have the same or very similar build flags, 16.04 build breaks but 14.04 build doesn't, it implies gcc-5.4 is misbuilding something that gcc-4.8 isn't (or, much less likely, gcc-5.4 is building something correctly but chromium relies on some misbehavior of older gcc.)

I have an Acer Chromebook 13 (Tegra K1) which had 14.04 on it; Chromium 52 or 53 worked fine. Went to 16.04 (managing to hold back X so I could keep the nvidia driver...) and chromium blew up as others have reported.

I also found the 14.04 build of (at this point, chromium 53) worked fine after I set "--disable-namespace-sandbox"... I had the "Aw, snap!" but found (per google) that "--disable-namespace-sandbox" fixed it without the ill effects of "--no-sandbox".

information type: Public → Public Security
information type: Public Security → Public

For the record, it works on an Ubuntu 17.04

Adam Smith (adamsmith) wrote :

I can confirm armhf and arm64 chromium works in 17.04

This is on a raspberry pi 3. On 16.04 chromium displays "aw, snap" unless I start with --no-sandbox

These https://github.com/raspberrypi-ui/chromium_patches are the patches used in raspbian. I don't know if there is anything useful in there to do with this bug.

Adam Smith (adamsmith) wrote :

Just to clarify, I've only tested 16.04 on armhf

Olivier Tilloy (osomon) on 2018-03-09
Changed in chromium-browser (Ubuntu):
assignee: Chad Miller (cmiller) → nobody
Kurt Vogt (pacco) wrote :

crashes on startup with

[ATTENTION: default value of option force_s3tc_enable overridden by environment.
[4005:4005:0326/150642.816056:ERROR:sandbox_linux.cc(375)] InitializeSandbox() called with multiple threads in process gpu-process.

System XUbuntu 16.04

Olivier Tilloy (osomon) wrote :

I haven't had a chance to get my hands on an armhf board yet.
To everyone affected, have things improved with the latest version of chromium, by any chance?

herrtimson (herrtimson) wrote :

With trusty 14.04 and chromium 65, it gives a few warning about 'Failed to launch GPU process' and 'Lost UI shared context', but these are no show stoppers, browser works fine. For the newer ubuntu versions I don't know, the newer kernels don't work with this device.

Xenial and above got 66, are there any plans to push it for trusty as well?

Anthony Martin (telamont) wrote :
Download full text (3.9 KiB)

I am also getting this error on X86 platform running default Ubuntu Desktop install of 18.04.1. Hardware is:

system**
    description: Desktop Computer
    product: MS-7721 (To be filled by O.E.M.)
    vendor: MSI
    version: 1.0
    serial: To be filled by O.E.M.
    width: 64 bits
    capabilities: smbios-2.7 dmi-2.7 smp vsyscall32
    configuration: boot=normal chassis=desktop family=To be filled by O.E.M. sku=To be filled by O.E.M. uuid=00000000-0000-0000-0000-D43D7E9BC232
  *-core
       description: Motherboard
       product: FM2-A75MA-E35 (MS-7721)
       vendor: MSI
       physical id: 0
       version: 1.0
       serial: To be filled by O.E.M.
       slot: To be filled by O.E.M.
     *-firmware
          description: BIOS
          vendor: American Megatrends Inc.
          physical id: 0
          version: V1.7
          date: 01/29/2013
          size: 64KiB
          capacity: 8128KiB
          capabilities: pci upgrade shadowing cdboot bootselect socketedrom edd int13floppy1200 int13floppy720 int13floppy2880 int5printscreen int9keyboard int14serial int17printer acpi usb biosbootspecification uefi
     *-memory
          description: System Memory
          physical id: 22
          slot: System board or motherboard
          size: 4GiB
        *-bank:0
             description: DIMM DDR3 Synchronous Unbuffered (Unregistered) 667 MHz (1.5 ns)
             product: CMX4GX3M1A1333C9
             vendor: Corsair
             physical id: 0
             serial: 41896574
             slot: A1_DIMM0
             size: 4GiB
             width: 64 bits
             clock: 667MHz (1.5ns)
        *-bank:1
             description: DIMM Synchronous [empty]
             product: Array1_PartNumber1
             vendor: A1_Manufacturer1
             physical id: 1
             serial: A1_SerNum1
             slot: A1_DIMM1
        *-bank:2
             description: DIMM Synchronous [empty]
             product: Array1_PartNumber2
             vendor: A1_Manufacturer2
             physical id: 2
             serial: A1_SerNum2
             slot: A1_DIMM2
        *-bank:3
             description: DIMM Synchronous [empty]
             product: Array1_PartNumber3
             vendor: A1_Manufacturer3
             physical id: 3
             serial: A1_SerNum3
             slot: A1_DIMM3
     *-cache:0
          description: L1 cache
          physical id: 30
          slot: L1 CACHE
          size: 96KiB
          capacity: 96KiB
          clock: 1GHz (1.0ns)
          capabilities: pipeline-burst internal write-back unified
          configuration: level=1
     *-cache:1
          description: L2 cache
          physical id: 31
          slot: L2 CACHE
          size: 1MiB
          capacity: 1MiB
          clock: 1GHz (1.0ns)
          capabilities: pipeline-burst internal write-back unified
          configuration: level=2
     *-cpu
          description: CPU
          product: AMD A4-5300 APU with Radeon(tm) HD Graphics
          vendor: Advanced Micro Devices [AMD]
          physical id: 35
          bus info: cpu@0
          version: AMD A4-5300 APU with Radeon(tm) HD Graphics
          slot: P0
          size: 1861MH...

Read more...

Well.. it happened again.. this nasty bug is back in Chromium 70.0.3538.102-1 on Arch linux.

Changed in chromium-browser (Arch Linux):
status: New → Confirmed
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Duplicates of this bug

Other bug subscribers