Comment 40 for bug 1523088

Revision history for this message
In , Matthew-d-roper (matthew-d-roper) wrote :

I'm pretty sure this happens because we wind up re-computing the same watermark values for each CRTC multiple times in a single atomic transaction (once per CRTC involved in the transaction). I'm putting together some patches that make SKL watermarks play more nicely with atomic (and eliminate this unnecessary looping); I should have those ready to post next week sometime.