On Mon, Mar 06, 2017 at 06:59:26PM +0100, Wargreen wrote:
> Here a new try, with a process name more explicit, but without X crash
> this time.
> So it's the graphic's ressources demanding process.
>
> [ 497.053562] 2 locks held by Xorg/2154:
> [ 497.053563] #0: (&rdev->exclusive_lock){++++.+}, at: [<ffffffffa90f3710>] rt_down_read+0x10/0x20
> [ 497.053571] #1: (&rdev->pm.mclk_lock){++++++}, at: [<ffffffffa90f3710>] rt_down_read+0x10/0x20
> [ 497.053589] 2 locks held by gnome-shell/2344:
> [ 497.053590] #0: (&rdev->pm.mclk_lock){++++++}, at: [<ffffffffa90f3710>] rt_down_read+0x10/0x20
> [ 497.053597] #1: (reservation_ww_class_mutex){+.+.+.}, at: [<ffffffffc0af47b0>] ttm_bo_vm_fault+0x60/0x530 [ttm]
Unfortunately, this doesn't get us any more data than we already had.
For some reason, a thread (G.Main in the prior trace, gnome-shell here)
in this state fails to make progress.
We need to better understand what this thread is doing. Is it
Runnable/spinning? Is it itself waiting on a lock? Dumping the stack
from it in this state might be revealing. The contents of
/proc/<pid>/stack might be of interest.
Julia
Attachment:
signature.asc
Description: PGP signature
