Hi On Sat, Apr 5, 2014 at 11:44 AM, Daniel Vetter <daniel.vetter@xxxxxxxx> wrote: > Hi all, > > Rainy w/e here and I got a bit bored, so looked at coverity again. I've closed > all outstanding issues in drivers/gpu now either as false positives or fixed in > this series expect for vmwgfx/ttm stuff (not enough clue) and one insane savage > init issue (no desire to wake dragons today). All patches besides "drm/ast: Remove dead code from cbr_scan2" are: Reviewed-by: David Herrmann <dh.herrmann@xxxxxxxxx> However, a few notes: - ttm_bo_unref() sets "bo = NULL;", so the following condition is always true. Your commit-message argues with an unconditional kref_unref(). I think that's misleading (but still true) as that does not change the fact that bo is always NULL afterwards. - I am a big fan of "if (!obj) return;" in destructors. It simplifies error-paths considerably and we can call them despite objects being already cleared. But that's just personal style, and I don't deal much with TTM anyway. But generally, I think we shouldn't remove these checks blindly. - The AST patch should be reviewed by someone who knows that hw. The code is obviously wrong, but that doesn't mean we cannot fix it properly. Thanks for the cleanups, the coverity reports have been pending here for months.. David _______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/dri-devel
