Package Details: linux-drm-tip-headers-git 6.3.r1170814.08c3a5cb8ecc-1

Git Clone URL: (read-only, click to copy)
Package Base: linux-drm-tip-git
Description: Headers and scripts for building modules for the Linux kernel with bleeding-edge GPU drivers kernel
Upstream URL:
Keywords: amdgpu drm git intel iris kernel linux
Licenses: GPL2
Submitter: cdown
Maintainer: yurikoles
Last Packager: yurikoles
Votes: 4
Popularity: 0.000000
First Submitted: 2017-03-08 14:47 (UTC)
Last Updated: 2023-03-24 02:59 (UTC)

Pinned Comments

yurikoles commented on 2019-06-07 08:08 (UTC) (edited on 2019-06-07 08:09 (UTC) by yurikoles)

PRs are welcome:

Latest Comments

1 2 3 4 Next › Last »

yurikoles commented on 2023-03-24 10:44 (UTC)

Joel, behrad,

Thank you for the report, and sorry for not so welcome self-service. The fix for prepare() was straight-forward, but I didn't want to push update until I figured out how to fix a BTF-related error during linkage of kernel. I decided that a faster fail is better than one in the middle of the build.

joelishness commented on 2023-03-17 19:39 (UTC) (edited on 2023-03-17 20:41 (UTC) by joelishness)

prepare() error is because upstream removed --save-scmversion option

Actually, maybe there are more commits involved

behrad commented on 2023-03-16 07:05 (UTC)

Hi, thank you for maintaining this package, but I get this error:

==> Making package: linux-drm-tip-git 6.2.r1157488.648a70b879da-1 (Thu 16 Mar 2023 10:33:24 AM +0330)
==> Checking runtime dependencies...
==> Checking buildtime dependencies...
==> Retrieving sources...
  -> Updating linux-drm-tip-git git repo...
  -> Found config
==> Validating source files with sha256sums...
    linux-drm-tip-git ... Skipped
    config ... Passed
==> Extracting sources...
  -> Creating working copy of linux-drm-tip-git git repo...
Reset branch 'makepkg'
==> Starting prepare()...
Setting version...
Usage: scripts/setlocalversion [srctree]
==> ERROR: A failure occurred in prepare().

rayzorben commented on 2022-07-30 23:29 (UTC)

It's 2022 so I didn't think the previous issue in 2020 would still be around, although it seems related. Build fails for me, and the closest error output is this. Any idea what to do to get past this?

drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c:806:20: error: passing argument 1 of ‘mutex_lock’ from incompatible pointer type [-Werror=incompatible-pointer-types]
  806 |         mutex_lock(&mgr->lock);
      |                    ^~~~~~~~~~
      |                    |
      |                    spinlock_t * {aka struct spinlock *}
./include/linux/mutex.h:199:38: note: expected ‘struct mutex *’ but argument is of type ‘spinlock_t *’ {aka ‘struct spinlock *’}
  199 | extern void mutex_lock(struct mutex *lock);
      |                        ~~~~~~~~~~~~~~^~~~
drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c:814:9: error: implicit declaration of function ‘drm_buddy_fini’ [-Werror=implicit-function-declaration]
  814 |         drm_buddy_fini(&mgr->mm);
      |         ^~~~~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c:815:22: error: passing argument 1 of ‘mutex_unlock’ from incompatible pointer type [-Werror=incompatible-pointer-types]
  815 |         mutex_unlock(&mgr->lock);
      |                      ^~~~~~~~~~
      |                      |
      |                      spinlock_t * {aka struct spinlock *}
./include/linux/mutex.h:218:40: note: expected ‘struct mutex *’ but argument is of type ‘spinlock_t *’ {aka ‘struct spinlock *’}
  218 | extern void mutex_unlock(struct mutex *lock);
      |                          ~~~~~~~~~~~~~~^~~~
drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c: In function ‘amdgpu_vram_mgr_first_block’:
drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c:57:1: error: control reaches end of non-void function [-Werror=return-type]
   57 | }
      | ^
cc1: some warnings being treated as errors

jghodd commented on 2020-05-30 19:42 (UTC) (edited on 2020-05-30 20:24 (UTC) by jghodd)

@yurikoles the issue is still there. reported at the link you provided - issue 1156 -

Edit: here's a patch to fix it.


--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c  2020-05-30 10:34:11.738183991 -0400
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c  2020-05-30 16:01:12.296946979 -0400
@@ -1354,7 +1354,7 @@

        /* Free the BO*/
-       drm_gem_object_put_unlocked(&mem->bo->tbo.base);
+       drm_gem_object_put(&mem->bo->tbo.base);

jghodd commented on 2020-05-30 14:31 (UTC) (edited on 2020-05-30 14:36 (UTC) by jghodd)

@yurikoles thanks for the link. i worked around the issue by commenting out the AMDGPU option in the config file since what i really needed were the Intel modules anyway. i noticed that torvalds signed off on 5.7-rc7 a few days ago, so i presume that the build error has been fixed. i'll be doing a new build today with AMDGPU restored to config, so i guess i'll see if the problem is gone. cheers.

yurikoles commented on 2020-05-24 20:09 (UTC)

@jghodd sorry for outdated info. The new bug tracker is here: