On 2021-03-17 9:29 a.m., Simon Ser wrote: > On Tuesday, March 16th, 2021 at 10:50 PM, Sefa Eyeoglu <contact@xxxxxxxxxxxxx> wrote: > >> Sometimes the primary plane might not be initialized (yet), which >> causes dm_check_crtc_cursor to divide by zero. >> Apparently a weird state before a S3-suspend causes the aforementioned >> divide-by-zero error when resuming from S3. This was explained in >> bug 212293 on Bugzilla. >> >> To avoid this divide-by-zero error we check if the primary plane's fb >> isn't NULL. If it's NULL the src_w and src_h attributes will be 0, >> which would cause a divide-by-zero. >> >> This fixes Bugzilla report 212293 >> https://bugzilla.kernel.org/show_bug.cgi?id=212293 >> >> Fixes: 12f4849a1cfd69f3 ("drm/amd/display: check cursor scaling") >> Signed-off-by: Sefa Eyeoglu <contact@xxxxxxxxxxxxx> > > Thanks for the fix! In theory we should return -EINVAL here, because we can't > enable the cursor plane without the primary plane. But that would break the > legacy API translation layer in DRM core, which expects that planes can always > be disabled individually. The core DRM code can deal with being unable to enable the CRTC while the primary plane is disabled. If you have evidence to the contrary, I'd like to see it. -- Earthling Michel Dänzer | https://redhat.com Libre software enthusiast | Mesa and X developer _______________________________________________ amd-gfx mailing list amd-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/amd-gfx
