OpenAI Codex Helps Uncover & Fix A Nearly 20 Year Old Radeon Linux Driver Bug

Sent out today was the latest batch of AMD kernel graphics/compute driver patches destined for the Linux 7.5 merge window opening up at year's end/ Besides a lot of fixes and continued work around new IP enablement for upcoming GPUs, there is even a fix for old ATI/AMD Radeon R600 era hardware after help from OpenAI's Codex. One of these AMD kernel graphics driver patches catching me by surprise for Linux 7.5 is a patch queued up for fixing the Unified Video Decoding (UVD) IP block on R6xx era GPUs... The nearly twenty year old graphics cards from the Radeon HD 2000/3000 days. AMD hasn't been officially maintaining support for the R600 GPUs in many years, but an open-source contributor took up the task with AI after encountering video decoding issues. Ted Bullock explained with a patch for expanding UVD handling check for more RV6** GPUs: "I investigated video corruption on my older iMac7,1 (RV610/M74, device ID 1002:94c8). When the decoder was reading compressed video from system memory while the GPU was drawing, it would corrupt video frames. Moving the input to VRAM allowed frames to decode correctly. Disabling graphics draws also stopped the corruption. Here is a proposal to keep UVD buffers in VRAM for this family of devices, as the driver already does for some other chips. The patch covers RV610, RV620, RV630, RV635 and RV670. These chips share the RV6xx decoder path and legacy R600 firmware, so I think they are likely to have the same hardware requirements. Someone with the other hardware would need to verify that. I tested this on Debian. I used Codex to help with the investigation, testing and patch." Sure enough, the patch from a new open-source contributor was accepted and via today's pull request is on its way to the Linux 7.5 kernel. Today's AMD pull request also brings a variety of other fixes and changes though not particularly as notable or exciting: "amdgpu: - DCN 6 DML2 fix - VCE fixes - GC 11.0.2 fix - SMU 15 fix - Error handling fixes - Partition fix - IMU firmware clean up fix - TLB fence handling fix - Enable JPEG type4 packet support - UALink updates - RAS updates - GC fixes - UserQ updates - GEM_OP fix - CS IB parsing fixes - SMU 15.0.8 updates - GPU reset fixes - vcn/uvd/vce/jpeg idle work handler cleanup - GPUVM fixes - NBIO updates - DC CRC fixes - MES debug extension support - SR-IOV updates - HDMI 2.x fixes - DSC fixes - DC pstate fixes - DC replay fixes - DC FAMS updates - More display KUNIT tests - SPM rework - GC 12.1 updates - VRR/freesync fixes - HDR fixes - eDP/LVDS fixes - Remove fence ops->release() - Debugfs fixes - Backlight fixes - SMU 7 cleanup - Non-DC audio fix - Documentation fixes - DC PSP footer size fix amdkfd: - SVM fixes - CRIU fixes - Fix WPTR BO eviction fence handling - Trap handler fixes - dma-buf handling fixes - CRAT cleanup - Add support for discontiguous and CPU-less NUMA topologies - Handle perfmon clocking when profiler is active - Profiler locking fixes drm: - EDID parsing support for HDMI 2.2 FRL rates from the SCDS radeon: - ATRM bounds checking - R6xx UVD fix UAPI: - Update KFD profiler version number to note perfmon clocking fix" We'll see what comes in the weeks ahead for more AMD kernel graphics/compute driver material for Linux 7.5. Not included in this week's pull request is the new AMD GFX13 (RDNA5) patches so perhaps that will be next week.