[SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

nash · 2023-11-06 18:27:31

Since a few weeks ago chromium is giving me this errors in some pages (joining any Zoom meet and an entry in MyAnimeList).

I get the Aw, Snap! error page with Error code: SIGILL. This only crashes the tab and not the entire browser.

My system is fully updated. Chromium version is 119.0.6045.105-1, but the problem was present in previous versions since this started a few weeks ago, don't remember exactly when.

Already tried clearing cache, changing security settings to "No protection" and using incognito windows.

dmesg gives me the following output when I try going to the MAL link:

 [12671.439407] traps: chromium[8548] trap invalid opcode ip:55abc33f8074 sp:7ffd97b137b0 error:0 in chromium[55abb9e90000+c895000]

Has this ever happened to anyone?

Apparently this happens in other chromium based browsers in other distros as well and sometimes the cause appears to be faulty hardware. I really hope it's not that. I've yet to run CPU and RAM checks. I have an Intel i5-7400.

Anyways any help would be appreciated.

Last edited by nash (2023-12-08 00:49:41)

seth · 2023-11-06 21:40:02

the cause appears to be faulty hardware

https://bbs.archlinux.org/viewtopic.php?id=287571

https://wiki.archlinux.org/title/Stress … MemTest86+
Run it at least over night (16h) or until an error shows up.
If it's a RAM error, make sure the system isn't overheating and try to downclock the RAM/choose the most conservative timings in the BIOS/UEFI.

nash · 2023-11-08 22:58:11

I did 6 passes and got no errors.

It could be due to this chromium bug, but I don't really understand.

The page is received correctly I think (I get a 200 response) and it kinda starts rendering but breaks midway, so it could be an issue with that.

kittykat · 2023-11-09 02:11:02

I can confirm I am having this problem recently (maybe the last month).

linux 6.5.9, chrome 119.0.6045.123-1

For me steps to reproduce involve joining a zoom meeting that has video going.

I have reproduced this on both my desktop and laptop.
3700x/5700xt and 1185g7

In both cases I get

kernel: traps: ThreadPoolForeg[484722] trap invalid opcode ip:5600e4567253 sp:7faaed3fcd90 error:0 in chromium[5600e2fc2000+c895000]

seth · 2023-11-09 07:38:40

Do you get a backtrace that results in FT_List_Iterate ?
https://bugs.chromium.org/p/chromium/is … id=1433651 seems private?

nash · 2023-11-09 15:39:19

That issue does seem private.

How do I get a stack trace?

seth · 2023-11-09 15:42:35

https://wiki.archlinux.org/title/Core_dump
https://wiki.archlinux.org/title/Debugg … ing_traces

nash · 2023-11-10 15:59:32

Maybe I'm doing something wrong but the stack trace from the coredump doesn't look similar to the one in the chromium issue. I think this is the top of the stack:

Core was generated by `/usr/lib/chromium/chromium --type=renderer --crashpad-handler-pid=1160 --enable'.
Program terminated with signal SIGILL, Illegal instruction.
#0  0x0000555f4096cb44 in blink::NGLineBreaker::BreakText(blink::NGInlineItemResult*, blink::NGInlineItem const&, blink::ShapeResult const&, blink::LayoutUnit, blink::LayoutUnit, blink::NGLineInfo*) ()

The rest is pretty much just calls within this blink namespace or whatever. Sorry I'm not very knowledgeable of this stuff.

seth · 2023-11-10 17:29:55

Can you please post the entire backtrace?

nash · 2023-11-11 21:38:15

I ran

coredumpctl debug

and enabled debuginfod for the session.

This is the stack trace: https://pastebin.com/NJ9dBeYi

seth · 2023-11-11 22:02:54

Not in the FT code at all, happens when breaking text in a block layout.
The pattern doesn't show up in google but for one occasion of "blink::NGLineBreaker::HandleText" in relation to an OOM… do you have a swap partition/file?

nash · 2023-11-11 22:10:11

I do have a swap file but running the stat command on it, it hasn't been modified since creation.

seth · 2023-11-11 22:20:44

swapon

nash · 2023-11-11 22:44:17

Yeah it's on I think. It's specified to mount as swap in the fstab and running

free -h

shows it properly.

I think I've just never ran OOM so it was never used.

Last edited by nash (2023-11-11 22:45:29)

seth · 2023-11-11 22:59:49

What are the ouputs of "swapon" and "free"?
The swapfile timestamps won't change because it's (usually) not accessed through the filesystem.
Unless you've a humongous amount of RAM it's rather unlikely that you've never ever swapped out anything ever.

nash · 2023-11-11 23:06:33

swapon output:

NAME      TYPE SIZE USED PRIO
/swapfile file   2G   0B   -2

free output:

               total        used        free      shared  buff/cache   available
Mem:        16325560     4701976     8028384      136404     4068436    11623584
Swap:        2097148           0     2097148

seth · 2023-11-12 09:05:03

There's a 2G swapfile and it has, so far, not been used.
Also the anime link you posted isn't overly complex, nor cjk-ridden.

Do you get the same behavior w/ a completely fresh user account?
Do the backtrace for all the crashes look the same or are they more all over the place?

nash · 2023-11-12 20:12:04

I get the same behaviour for incognito window and clean cache.

The stack trace is the same for that MAL page but it's different for Zoom meetings. It looks like this then:

[Thread debugging using libthread_db enabled]                                                
Using host libthread_db library "/usr/lib/libthread_db.so.1".
Core was generated by `/usr/lib/chromium/chromium --type=renderer --crashpad-handler-pid=1172 --enable'.
Program terminated with signal SIGILL, Illegal instruction.
#0  0x00005588a4477253 in media::FFmpegGlue::OpenContext(bool) ()
[Current thread is 1 (Thread 0x7f5051dfe6c0 (LWP 3))]
(gdb) bt
#0  0x00005588a4477253 in media::FFmpegGlue::OpenContext(bool) ()
#1  0x00005588a3430bc4 in void base::internal::ReturnAsParamAdapter<bool>(base::OnceCallback<bool ()>, std::unique_ptr<bool, std::default_delete<bool> >*) ()
#2  0x00005588a845d667 in base::internal::PostTaskAndReplyRelay::RunTaskAndPostReply(base::internal::PostTaskAndReplyRelay) ()
#3  0x00005588a845d7db in base::internal::Invoker<base::internal::BindState<void (*)(base::internal::PostTaskAndReplyRelay), base::internal::PostTaskAndReplyRelay>, void ()>::RunOnce(base::internal::BindStateBase*) [clone .cfi] ()
#4  0x00005588a8438a7c in base::TaskAnnotator::RunTaskImpl(base::PendingTask&) ()
#5  0x00005588a8460aa5 in base::internal::TaskTracker::RunSkipOnShutdown(base::internal::Task&, base::TaskTraits const&, base::internal::TaskSource*, base::SequenceToken const&) ()
#6  0x00005588a8460622 in base::internal::TaskTracker::RunTask(base::internal::Task, base::internal::TaskSource*, base::TaskTraits const&) ()
#7  0x00005588a84601eb in base::internal::TaskTracker::RunAndPopNextTask(base::internal::RegisteredTaskSource) ()
#8  0x00005588a8472c0d in base::internal::WorkerThread::RunWorker() ()
#9  0x00005588a847274a in base::internal::WorkerThread::RunPooledWorker() ()
#10 0x00005588a8472593 in base::internal::WorkerThread::ThreadMain() ()
#11 0x00005588a848f71b in base::(anonymous namespace)::ThreadFunc(void*) [clone .a67435033112019129ad5c28ddc47327] [clone .cfi] ()
#12 0x00007f50670859eb in start_thread (arg=<optimized out>) at pthread_create.c:444
#13 0x00007f5067109654 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:100

seth · 2023-11-12 20:44:22

I get the same behaviour for incognito window and clean cache.

That's not what I suggested - we don't want chromium to look at anything in your $HOME.

The stack trace is the same for that MAL page

Depending on how many traces you checked, that's a bit too consistent for RAM issues.

Do all zoom crashes look the same as well?

nash · 2023-11-12 20:56:18

Oh ok. I'll check with another user later. Would it be the same if I just delete the .config/chromium and .cache/chromium? Is there any other chromium stuff in $HOME?

The stack trace for the Zoom is identical as well. Even the memory addresses.

seth · 2023-11-12 21:03:16

There's gonna be IPC to other processes and eg. your GTK theme (and some settings?) will probably be loaded etcetc.
We just want to know whether this is specific to your user or affects the system globally and the easiest way is a fresh user account (before you start moving stuff from your home, and then miss stuff in /var/run and some other locations.

nash · 2023-11-14 17:33:24

It also breaks with a different user. Same problem.

seth · 2023-11-14 20:52:18

Do you get the same issue with other electron-based clients (brave, chrome, (vs)code, discord)?
Do any other clients exhibit frequent instability? (Check https://wiki.archlinux.org/title/Core_d … _core_dump )

nash · 2023-11-14 21:16:56

I don't use any other electron-based apps I think.

Are the IntelliJ IDEs electron-based? They don't show any weird behaviour anyway.

Discord web works fine as well.

Also consider the fact that so far I've only had this problem with those 2 sites from the op (and for MAL it's just that one link), so it might be hard to find similar problems in other software that I don't use as much as the web browser.

seth · 2023-11-14 21:23:14

You could probably open those pages in brave (or vivaldi) and see what happens.

Arch Linux

#1 2023-11-06 18:27:31

[SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#2 2023-11-06 21:40:02

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#3 2023-11-08 22:58:11

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#4 2023-11-09 02:11:02

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#5 2023-11-09 07:38:40

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#6 2023-11-09 15:39:19

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#7 2023-11-09 15:42:35

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#8 2023-11-10 15:59:32

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#9 2023-11-10 17:29:55

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#10 2023-11-11 21:38:15

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#11 2023-11-11 22:02:54

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#12 2023-11-11 22:10:11

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#13 2023-11-11 22:20:44

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#14 2023-11-11 22:44:17

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#15 2023-11-11 22:59:49

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#16 2023-11-11 23:06:33

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#17 2023-11-12 09:05:03

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#18 2023-11-12 20:12:04

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#19 2023-11-12 20:44:22

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#20 2023-11-12 20:56:18

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#21 2023-11-12 21:03:16

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#22 2023-11-14 17:33:24

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#23 2023-11-14 20:52:18

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#24 2023-11-14 21:16:56

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

#25 2023-11-14 21:23:14

Re: [SOLVED] "Aw, Snap!" error in Chromium. Trap invalid opcode

Board footer