search

dvdhrm / kmscon

Linux KMS/DRM based virtual Console Emulator
http://www.freedesktop.org/wiki/Software/kmscon
Other
436 stars 81 forks source link

kmscon seg faults in EGL #102

Closed kalrish closed 11 years ago

kalrish commented 11 years ago

The latest commits make kmscon seg fault when switching to another VT. It still works fine if I stay in the same VT, but, if I try to change to another one, no kmscon will start (and the current process will also die, when I try to return back). Here is the relevant message:

Nov 04 18:40:58 [HOST] kernel: kmscon[548]: segfault at 24 ip 00007f0cb986ac71 sp 00007fffaa9ccb70 error 6 in libEGL.so.1.0.0[7f0cb9854000+1f000]

I'm sorry there's nothing more. I have built kmscon-8 (which worked fine) and am working with it again. I've reviewed the commits, but I'm not a developer, so I have no idea what might be the problem.

Using an HD 3200 Radeon card with the open-source driver. Kernel 3.12. Mesa 9.2.2.

dvdhrm commented 11 years ago

Can you run kmscon with "--debug" and show me the output? You can pipe it to a file via kmscon --debug 2>log_file. If you run systemd, the backtrace would help a lot. You can get it via: systemd-coredumpctl gdb kmscon and then run "bt" in gdb.

I cannot reproduce it locally. Can you try running with --no-hwaccel?

dvdhrm commented 11 years ago

btw., I pushed a small fix, but I'm very uncertain whether it fixes your issue.

kalrish commented 11 years ago

It seems that last commit has fixed it. Building it with the same options as usual results in a normal-behaving kmscon. I've tried enabling and disabling hardware acceleration, using bbulk and gltex, and the problem has just disappeared.

I'm sorry I've wasted your time - thank you for the assistance. I'm stunned!

dvdhrm commented 11 years ago

Hehe, you cannot waste my time. Don't worry. Thanks for the report, seems radeon-DRM has some weird inter-process dependencies. I will ping the developers on IRC.