On Sat, 20 Feb 2016, Rodrigo Vivi rodrigo.vivi@gmail.com wrote:
We have a duplicated patch on drm-intel-nightly
commits d7006964d and cc1de6e80
[I snipped the extra trailing 'q' off the 2nd commit id.]
causing:
drivers/gpu/drm//amd/amdgpu/amdgpu_ttm.c:818:6: error: redefinition of ‘amdgpu_ttm_tt_affect_userptr’ bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start, ^ drivers/gpu/drm//amd/amdgpu/amdgpu_ttm.c:799:6: note: previous definition of ‘amdgpu_ttm_tt_affect_userptr’ was here bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
One cc'ed stable and other didn't. So I'm not sure how this duplication happened or in which tree, so I'm pinging you all that know better how to identify and solve this issue.
There's nothing unusual here. The patch was apparently first applied to drm-next, and then cherry-picked to drm-fixes and this one is now in Linus' master. We do this all the time in i915.
The build fail is, or rather was, in our integration tree. I thought I had screwed up the conflict resolution [1], but turns out the fail was in a silent conflict. The function got added twice. It's now fixed in our tree.
Heads up Dave, I presume this will happen when drm-next gets merged to Linus' tree in the next merge window. (Or perhaps in linux-next already.)
BR, Jani.
[1] http://mid.gmane.org/87ziuwvj0o.fsf@intel.com