A contributor chasing garbled video on a 2007 iMac used OpenAI’s Codex to trace the fault to how the radeon driver feeds the UVD decoder, and his fix for RV6xx GPUs is now queued for Linux 7.5.
Some of the most satisfying kernel patches have nothing to do with new silicon. The latest AMD graphics pull request headed for Linux 7.5 carries the usual pile of work for current and upcoming GPUs, but tucked into it is a fix for hardware that first shipped in 2007: the RV6xx chips behind the Radeon HD 2000 and HD 3000 series.
The patch comes from Ted Bullock, a new contributor who ran into video corruption on an older iMac7,1. That machine uses an RV610-based mobile GPU (the M74, PCI ID 1002:94c8), and hardware-accelerated video playback through the Unified Video Decoder kept producing broken frames.
What was actually going wrong #
Bullock’s write-up on the kernel mailing list is refreshingly concrete. When the decoder read compressed video out of system memory while the GPU was busy drawing, the frames came out corrupted. Move that input into VRAM and the frames decoded cleanly. Stop the graphics draws and the corruption also went away.
That points to contention between the 3D engine and the decoder when UVD fetches its buffers over the system-memory path, something these early chips evidently do not handle well. The radeon driver already forces UVD buffers into VRAM for some other GPUs, so the fix simply adds the RV6xx family to that list. It covers RV610, RV620, RV630, RV635 and RV670, which share the same decoder path and legacy R600 firmware.
Bullock was upfront that he only had the RV610 to test with, on Debian, and that someone with the other chips would need to confirm the behaviour there. He was equally upfront about the tooling: he wrote that he used Codex to help with the investigation, the testing and the patch itself.
Why an AI-assisted patch matters here #
AMD stopped actively maintaining these R600-era GPUs years ago. Bugs on hardware this old tend to sit unfixed because the people who understand the driver internals have moved on, and the people who still own the cards rarely know where to start. An assistant that can read through an unfamiliar driver, suggest experiments and narrow the problem down lowers that barrier considerably.
The part that should reassure skeptics is that nothing about the process skipped review. The change went through the normal mailing-list route and was picked up by AMD’s kernel graphics maintainers. As Phoronix noted, it landed in the pull request sent on November 12, listed simply as an “R6xx UVD fix” under the radeon driver.
The rest of that pull request is mostly fixes and enablement for current hardware: display, power management, GPU reset and SR-IOV work in amdgpu, plus EDID parsing support for HDMI 2.2 FRL rates in the core DRM code. The new GFX13 (RDNA 5) driver code is not part of this batch.
The Linux 7.5 merge window opens around the end of the year, so the fix should reach mainline in early 2027. For the handful of people still running a Radeon HD 2400, HD 2600 or HD 3870 as a media box, or reviving a mid-2007 iMac with a modern distribution, it means accelerated video playback should finally behave.