Closed jjbarr closed 5 years ago
UPDATE: I have just tested more Goldsrc games. This behavior appears to affect all Goldsrc games save for Half-Life. At least, all the ones I own. This includes:
-Counter-Strike -Day of Defeat -Half-Life: Blue Shift -Half-Life: Opposing Force -Deathmatch Classic -Ricochet -Team Fortress Classic
I have not tested any mods. I also have not tested CSC0, as I do not own it. Anyone else suffering from this issue is welcome to test that game.
...D'oh. I am... really stupid.
Okay, it turns out having __GL_THREADED_OPTIMIZATIONS=1
breaks most goldsrc games. I forgot about this! Just set the envvar to 0 to fix the problem.
I'm not sure if I have the same issue as you, but I have the same symptoms. When I launch counter strike 1.6 or half life 1, steam says it is running, but no window appears. I'm using an ATI card instead of an NVIDIA one, so the __GL_THREADED_OPTIMIZATIONS
has no effect.
Interestingly enough, the game launches fine when I start it directly from the command line.
I rebooted my machine and it works fine now. I have no idea what changed.
I suspect that the issue resulted from a version mismatch between some aspect of your driver in RAM and your driver on disk.
This is common when you update your drivers or kernel and don't reboot.
Your system information
Please describe your issue in as much detail as possible:
For the first time after the 8/21 update, I attempted to launch Counter-Strike 1.6. The game was marked as "Running" in Steam, but nothing was actually happening. Further investigation (
ps xu
) indicated that the game was, in fact, happily running. However, I was still unable to get to a menu (or even a black screen!), and the game wouldn't exit without akill -9
from me.When I straced the process, (my logs are available here), it indicated that what was occurring was repeated calls to
sched_yield(2)
. GDB reported (I think—my GDB-fu is weak) that these calls were taking place inside of pthreads.To me, at least, that indicates the bug is some kind of race condition leading to a deadlock.
Once, while attempting to run the the game, it actually did launch fine (it was under
strace
, and&|
ed to tee, but I'd some how done it wrong so the output wasn't captured). However, this was not reproducible ever again, and I've launched the game like 20 times at this point.I've spoken to other Linux users. Some of them have the same issue. Others do not. I don't know what the difference is that triggers this.
One more note: Half-Life, which runs from the same executable, runs fine, and does not have this issue. However, Deathmatch Classic locks up in much the same way. I have not yet tested the other goldsrc games in my library, but it seems the issue does not affect all of them, nor does it only affect CS.
Steps for reproducing this issue: