On Sun, Mar 01, 2026 at 08:43:43AM -0800, Barry K. Nathan wrote:
On 3/1/26 00:49, Barry K. Nathan wrote:
Unfortunately, 6.19.6-rc1 won't even build for me:
Warning: drivers/gpu/drm/i915/intel_wakeref.h:156 expecting prototype for __intel_wakeref_put(). Prototype was for INTEL_WAKEREF_PUT_ASYNC() instead 1 warnings as errors make[9]: *** [drivers/gpu/drm/i915/Makefile:449: drivers/gpu/drm/i915/intel_wakeref.hdrtest] Error 3 make[8]: *** [scripts/Makefile.build:546: drivers/gpu/drm/i915] Error 2 make[8]: *** Waiting for unfinished jobs....
This only happens with 6.19.6-rc1, not any of this weekend's other stable rc's. (I'm still testing 6.12.75-rc1 and 6.18.16-rc1, but they're doing well so far. I have successfully built 5.15.202-rc1 and 6.1.165-rc1 but I won't have a chance to do any further testing of them before they're released.)
As soon as I can (in the next hour or two) I'll minimize my config a little to shorten the compile time, then I'll start bisecting.
Result of bisecting: first bad commit: [0ef5d235ab57bc90831ddf38eb1742ff68f345e1] docs: kdoc: fix logic to handle unissued warnings
This commit breaks the i915 DRM build if (and only if) CONFIG_DRM_I915_WERROR=y, whether CONFIG_WERROR is enabled or disabled. However, the "bad" commit is definitely fixing a real bug, and this build failure doesn't happen on current mainline as of this writing (commit eb71ab2bf722), so I don't think dropping the patch is the correct way forward.
Rather, adding commit 524696a19e34598c9173fdd5b32fb7e5d16a91d3 drm/i915/wakeref: clean up INTEL_WAKEREF_PUT_* flag macros (it applies cleanly) fixes the warning, thereby fixing the build.
The resulting kernel works fine in my testing, too. I'm using 6.19.6-rc1 + 524696a19e34598c9173fdd5b32fb7e5d16a91d3 to write and send this email from my ThinkPad T14 Gen 1, which uses the i915 DRM driver for its Intel integrated graphics. (I also tested it on my 2017 MacBook Air, which also uses i915 DRM for its Intel integrated graphics.)
I'll queue 524696a19e345 up, thanks for the report!