We've seen a couple of bugs now involving mistakes in the BEGIN_RING/ADVANCE_RING counts or mistakes of missing them entirely. I'd like to add code to check that what happens inside of BEGIN/ADVANCE pairs matches expectations. Attached is a patch which has verified a couple of bugs. I don't expect the extra conditionals are expensive enough to warrant conditionalizing them at compile time.
Created attachment 713 [details] [review] radeon-dma-sanity.diff
Committed.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.