Am 31.03.22 um 10:15 schrieb Lucas De Marchi:
On Thu, Mar 31, 2022 at 09:35:50AM +0200, Christian König wrote:
Well the fix is trivial, but somehow rebuilding drm-tip always fails for me while merging drm-intel-next.
I probably have somehow messed up reverting the conflict resolution. Ideas?
It looks like the error is in the wrong conflict resolution in 629041de3a0c M─┐ Merge remote-tracking branch 'drm-misc/drm-misc-next' into drm-tip you'd need to redo the merge (-i) after telling git rerere to forget the previous conflict resolution.
https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdrm.pages....
it's also useful to use dim -i, to make sure we can build on each merge afterwards.
AFAICS the revert would be on 141577c6d0bd ("2022y-03m-29d-09h-01m-04s UTC: drm-tip rerere cache update")
taking a stab on it, should be like below. It even builds, but I have no confidence to commit it.
Please go ahead with that, it looks correct to me.
I somehow messed up my installation so badly that even a "drm rebuild-tip" doesn't work any more and fails to merge drm-intel-next with a conflict which seems to already be resolved.
And when I try to resolve it once more drm-tip says only "Preimage für 'drivers/gpu/drm/i915/intel_pm.c' aufgezeichnet." and doesn't seem to remember the conflict resolution.
Any advise how I can fix this?
Thanks, Christian.
Lucas De Marchi
diff --cc drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c index dd78402e3cb0,68494b959116..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c @@@ -200,17 -187,20 +187,17 @@@ static void amdgpu_gtt_mgr_del(struct t * * Re-init the gart for each known BO in the GTT. */ -int amdgpu_gtt_mgr_recover(struct amdgpu_gtt_mgr *mgr) +void amdgpu_gtt_mgr_recover(struct amdgpu_gtt_mgr *mgr) { - struct amdgpu_gtt_node *node; + struct ttm_range_mgr_node *node; struct drm_mm_node *mm_node; struct amdgpu_device *adev; - int r = 0; adev = container_of(mgr, typeof(*adev), mman.gtt_mgr); spin_lock(&mgr->lock); drm_mm_for_each_node(mm_node, &mgr->mm) { - node = container_of(mm_node, typeof(*node), base.mm_nodes[0]); - amdgpu_ttm_recover_gart(node->tbo); + node = container_of(mm_node, typeof(*node), mm_nodes[0]); - r = amdgpu_ttm_recover_gart(node->base.bo); - if (r) - break; ++ amdgpu_ttm_recover_gart(node->base.bo); } spin_unlock(&mgr->lock); diff --cc drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index fc4563cf2828,28f5e8b21a99..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@@ -375,9 -375,7 +375,9 @@@ static void amdgpu_vm_bo_base_init(stru if (bo->tbo.base.resv != vm->root.bo->tbo.base.resv) return; + dma_resv_assert_held(vm->root.bo->tbo.base.resv); + - vm->bulk_moveable = false; + ttm_bo_set_bulk_move(&bo->tbo, &vm->lru_bulk_move); if (bo->tbo.type == ttm_bo_type_kernel && bo->parent) amdgpu_vm_bo_relocated(base); else @@@ -2660,12 -2597,9 +2606,12 @@@ void amdgpu_vm_bo_del(struct amdgpu_dev struct amdgpu_vm *vm = bo_va->base.vm; struct amdgpu_vm_bo_base **base; + dma_resv_assert_held(vm->root.bo->tbo.base.resv); + if (bo) { + dma_resv_assert_held(bo->tbo.base.resv); if (bo->tbo.base.resv == vm->root.bo->tbo.base.resv) - vm->bulk_moveable = false; + ttm_bo_set_bulk_move(&bo->tbo, NULL); for (base = &bo_va->base.bo->vm_bo; *base; base = &(*base)->next) {
Christian.
Am 31.03.22 um 08:28 schrieb Christian König:
I'm going to take a look, but need to figure out how to find the broken merge
Christian.
Am 31.03.22 um 00:33 schrieb Dixit, Ashutosh:
Is anyone looking into fixing this:
drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c: In function ‘amdgpu_gtt_mgr_recover’: drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c:200:31: error: ‘struct ttm_range_mgr_node’ has no member named ‘tbo’ amdgpu_ttm_recover_gart(node->tbo); ^~ make[4]: *** [scripts/Makefile.build:288: drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.o] Error 1
Thanks.