Bug 926207 – x11-drivers/nvidia-drivers-550.54.14 fails to compile: nvidia-drm-drv.c:199:18: error: initialization of struct drm_framebuffer ()(struct drm_device , struct drm_file , const struct drm_mode_fb_cmd2 ) from incompatible pointer type struct drm_framebuffer

• Product: Gentoo Linux
• Status: CONFIRMED ---

[Description Agostino Sarubbo 2024-03-05 11:47:12 UTC]

https://blogs.gentoo.org/ago/2020/07/04/gentoo-tinderbox/

Issue: x11-drivers/nvidia-drivers-550.54.14 fails to compile.
Discovered on: arm64 (internal ref: tinderbox_arm64)
System: GCC-14-SYSTEM (https://wiki.gentoo.org/wiki/Project:Tinderbox/Common_Issues_Helper#GCC-14)

Info about the issue:
https://wiki.gentoo.org/wiki/Project:Tinderbox/Common_Issues_Helper#CF0003

[Comment 1 Agostino Sarubbo 2024-03-05 11:47:14 UTC]

Created attachment 886753 [details]
build.log

build log and emerge --info

[Comment 2 Ionen Wolkens 2024-03-05 13:26:14 UTC]

Strange, gives the impression that the test to check if need to pass "dev" failed (drm_helper_mode_fill_fb_struct in kernel/conftest.sh), but I can't get the test to fail even with gcc-14.

Shame these tests output go straight to /dev/null (no logs either) so can't see what really happened here.

Test does include the wrong-in-recent-kernels header though, but fwics that just makes it "always pass" with a warning, which is fine given this only needs to fail for unsupported <4.5 kernels.

Maybe there's something special going on with arm64 that I don't see at the moment -- builds fine with gcc14 on amd64 for me anyhow, albeit only tried with a "make defconfig" 6.7.8 kernel (haven't tried 6.1.80).

[Comment 3 Agostino Sarubbo 2024-03-05 13:46:23 UTC]

(In reply to Ionen Wolkens from comment #2)
> Strange, gives the impression that the test to check if need to pass "dev"
> failed (drm_helper_mode_fill_fb_struct in kernel/conftest.sh), but I can't
> get the test to fail even with gcc-14.
> 
> Shame these tests output go straight to /dev/null (no logs either) so can't
> see what really happened here.
> 
> Test does include the wrong-in-recent-kernels header though, but fwics that
> just makes it "always pass" with a warning, which is fine given this only
> needs to fail for unsupported <4.5 kernels.
> 
> Maybe there's something special going on with arm64 that I don't see at the
> moment -- builds fine with gcc14 on amd64 for me anyhow, albeit only tried
> with a "make defconfig" 6.7.8 kernel (haven't tried 6.1.80).

works for me too on amd64 so this should be something arm64 related

[Comment 4 Ionen Wolkens 2024-03-19 16:46:17 UTC]

May or may not be fixed in 550.67 given they finally fixed a (different) gcc14 issue and perhaps they picked this up too. Not that I see clear signs that it is so not too optimistic.

Not that I actually tried this on arm64 myself, so no idea.