I have been digging into this, and I was wrong it occurred at clBuildProgram
(was just the logs printing was deferred - sorry).
Using gdb, I have been able to see that the crash occurs inside the call
pipe_semaphore_signal(&ws->cs_queued);
Here is the calling stack just before the hang:
(gdb) bt
Hi,
I am using kernel 3.12.6-gentoo, Mesa 10.0.1 and once every two calls to
clBuildProgram, the GPU goes to reset after 10 seconds.
This also happens on Debian unstable with Mesa 9.2. First hello_world works,
the next one hangs, third works, and so on.
Despite this hang on this particular Open
