On Thu, Mar 06, 2014 at 08:17:51AM -0800, Jesse Barnes wrote: > On Thu, 6 Mar 2014 09:12:40 +0000 > Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> wrote: > > > On Wed, Mar 05, 2014 at 02:48:27PM -0800, Jesse Barnes wrote: > > > This gets us out of our init code and out to userspace quite a bit > > > faster, but does open us up to some bugs given the state of our init > > > time locking. > > > > Why are we hand rolling an async task for this? See > > http://lists.freedesktop.org/archives/intel-gfx/2010-August/007642.html > > > > And the locking issue was the main reason why we haven't been able to > > proceed so far... > > In looking at the async domains it didn't appear that they would > actually save me much if any code in most of these cases. I agree that it may not save actual lines of code. I am more thinking of the need to try and integrate with existing work along the same paths. I presume that the async init debug code is more refined (or at least could be) etc. > The locking is worrisome, but I added some extra WARNs and things are > solid across multiple boots, reloads, and suspend/resume cycles. I > haven't tried lockdep though... Things have changed wrt locking, but I still expect that to make lockdep happy will require some ugly reworking. -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
