I have found this: https://bugzilla.kernel.org/show_bug.cgi?id=42162
That bug report is very similar to this one: - The same GPU softresets with the error "[drm:r600_ring_test] *ERROR* radeon: ring test failed (scratch(0x8504)=0xCAFEDEAD)" - The same errors at the end of the type: "[drm:radeon_cs_ioctl] *ERROR* Failed to schedule IB ! [drm:radeon_ib_schedule] *ERROR* radeon: couldn't schedule IB(n)" Being "n" a number that changes. According to the former link, the patch to solve that bug was merged in Linux v3.1-rc10a; therefore, I suppose that aforementioned patch should be already in the 3.2 kernel. I hope this helps.
