Page MenuHomePhabricator
Create Task
Maniphest T8840

E crashes after a while of unactivity
Closed, InvalidPublic
Actions

Description

Watching a movie for a while (which means, no activity on mouse/keyboard), E crashes without reason, there's a dump:

http://paste.openstack.org/show/799040/

  • EFL version: branch-1.25.1
  • E version: recent build from master at: d084520660

OT Question: since I see so many lines in the crashdump file, it is created in incremental mode?

Related Objects

raster added a comment.Oct 14 2020, 9:48 AM

that backtrace implies its segfaulting inside select... which smells wrong.

fyi the crashdump log file is appended to so you might want to delete the file then get a clean backtrace set with just that crash.

ProhtMeyhet triaged this task as High priority.Oct 14 2020, 9:49 AM
ProhtMeyhet added a subscriber: ProhtMeyhet.

OT Question: since I see so many lines in the crashdump file, it is created in incremental mode?

AFAIK it is, and that is a problem.

Could you please delete your crashdump and crash e like that again? (this is something that should be adressed)

Thanatermesis added a comment.Oct 14 2020, 10:01 AM

Thanks for the details, I wonder if the backtrace files should be re-created as new ones instead, or adding something like a bunch of spaces or date stamps to differentiate between the reports

Sure! I already deleted it, I will add the new one when it will happen

raster added a comment.Oct 14 2020, 5:02 PM

they are logged to so if you get multiple crashes... you have them all there. the latest (one at end of log) is .. the latest crash.

ProhtMeyhet added a comment.Oct 16 2020, 4:55 AM
In T8840#154321, @raster wrote:

they are logged to so if you get multiple crashes... you have them all there. the latest (one at end of log) is .. the latest crash.

It would still be a good idea to have a crashlog start with a datetime.

Thanatermesis added a comment.Oct 30 2020, 7:26 PM

I think I was finally able to reproduce the issue, it was not in an "inactivity moment" but while watching a documentary in the external TV plugged on hdmi (like in the other moments that happened)

http://paste.openstack.org/show/799575/

raster added a comment.Nov 11 2020, 2:51 AM

that bt doesn't show e actually crashed. if it crashds inside select... i am guessing something did a "kill -SEGV pidof enlightenment" as that is the only way i can see this happening there.

ProhtMeyhet added a comment.Nov 15 2020, 4:18 AM
In T8840#154460, @raster wrote:

that bt doesn't show e actually crashed. if it crashds inside select... i am guessing something did a "kill -SEGV pidof enlightenment" as that is the only way i can see this happening there.

Just to clarify: A SEGV is not catchable, therefore a backtrace is not possible.

So this is very likely a driver issue. Please open a bug report with your operating system vendor.

Thanatermesis closed this task as Invalid.Nov 15 2020, 10:31 AM

driver issue should be not an E issue, so let's close it...

Thanatermesis added a comment.Nov 18 2020, 6:14 PM

Just for the record, I attach another crashdump: http://paste.openstack.org/show/800183

It happened while watching a movie with external screen on hdmi, movie was paused and later the screen suspended, when I was back and playing the video again, one minute later, E segfaulted

raster added a comment.Nov 19 2020, 1:13 AM

that new bt is different to the other one - completely different. this says "heap is corrupted". the bug happened before this bt - so the bt itself is useless. you need to use ASAN or valgrind to find the real source. can't do anything with that backtrace