On 30.04.2012 17:58, Jerome Glisse wrote: > On Mon, Apr 30, 2012 at 10:50 AM, Christian K?nig > <deathsimple at vodafone.de> wrote: >> Make the suballocator self containing to locking. >> >> v2: split the bugfix into a seperate patch. >> >> Signed-off-by: Christian K?nig<deathsimple at vodafone.de> > I would say NAK but i don't have better solution yet to the issue. > Idea is that cs mutex protect the SA, i am trying hard to avoid adding > new lock, in fact i would like to remove more lock and have some idea > for that. I looked at the whole patch set and the issue is with radeon > semaphore allocation in ttm move blit case, all other case have the cs > mutex taken already. So if no better solution than yeah introduce a > new lock.
Well, my idea was to remove the cs mutex in the long term. But just reading your other mail, I think I better describe the picture as a whole there. Just let me mention that I intended to be this the lock for the "transaction memory allocator", so it pretty much matches your idea also. Christian.
