On 2018-12-06 10:36 a.m., Grodzovsky, Andrey wrote: > Not an expert on Freesync so maybe stupid question but from he comment > looks like this pipe locking is only for the sake of Freesync mode there > - why is it then called unconditionally w/o checking if you even run in > Freesync mode ? > > Andrey I don't think there's any sort of state tracking done in DC for FreeSync state changes on a stream so it was probably easier to do this. Given the age of the pipe lock addition I would be worried about other behavior relying on it now if it was just conditionally done on that.
In practice I think the worst case effect of guarding these calls with a mutex is <100us of delay. That's for two separate threads issuing two commits at the same time with the planes changed. Nicholas Kazlauskas > > > On 12/06/2018 08:42 AM, Kazlauskas, Nicholas wrote: >> A reader/writer lock wouldn't help here, actually. >> >> Fast updates (and page flips in particular) need to lock the pipe as >> well, see: commit_planes_for_stream. >> >> Nicholas Kazlauskas > _______________________________________________ amd-gfx mailing list amd-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/amd-gfx