linux/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c, branch v4.9

drm/amdgpu: fix fence slab teardown

2016-10-24T21:01:22Z

To free fences, call_rcu() is used, which calls amdgpu_fence_free() after a grace period. During teardown, there is no guarantee all callbacks have finished, so amdgpu_fence_slab may be destroyed before all fences have been freed. If we are lucky, this results in some slab warnings, if not, we get a crash in one of rcu threads because callback is called after amdgpu has already been unloaded. Fix it with a rcu_barrier(). Fixes: b44135351a3a ("drm/amdgpu: RCU protected amdgpu_fence_release") Acked-by: Chunming Zhou Reviewed-by: Christian König Signed-off-by: Grazvydas Ignotas Signed-off-by: Alex Deucher

drm/amdgpu: don't leave dangling pointers around

2016-09-27T17:00:51Z

Right now it's possible to trigger fence_drv.fences[] dereference after the array has been freed. While the real problem is elsewhere, this still results in confusing errors that depend on how the freed memory was reused (I've seen "kernel tried to execute NX-protected page"), it's better to clear them and get NULL dereference so that it's obvious what's going wrong. Signed-off-by: Grazvydas Ignotas Signed-off-by: Alex Deucher

drm/amdgpu: always signal all fences

2016-07-14T20:46:07Z

A little fallout from "drm/amdgpu: sanitize fence numbers", we sometimes need to signal all fences in the ring. Signed-off-by: Christian König Reviewed-by: Alex Deucher Reviewed-by: Chunming Zhou Signed-off-by: Alex Deucher

drm/amdgpu: sanitize fence numbers

2016-07-14T20:46:03Z

Looks like the VCE block sometimes still sends nonsense fence numbers on startup. Signed-off-by: Christian König Reviewed-by: Alex Deucher Signed-off-by: Alex Deucher

drm/amdgpu: fix bug in fence driver fini

2016-05-25T13:42:53Z

Using wrong counter for walking fences. Fixes a crash when unloading the driver. Signed-off-by: Monk Liu Reviewed-by: Christian König Signed-off-by: Alex Deucher

drm/amdgpu: create fence slab once when amdgpu module init.

2016-05-13T18:30:22Z

This avoids problems with multiple GPUs. For example, if the first GPU failed before amdgpu_fence_init() was called, amdgpu_fence_slab_ref is still 0 and it will get decremented in amdgpu_fence_driver_fini(). This will lead to a crash during init of the second GPU since amdgpu_fence_slab_ref is not 0. v2: add functions for init/exit instead of moving the variables into the driver. Signed-off-by: Rex Zhu Reviewed-by: Christian König Signed-off-by: Alex Deucher

drm/amdgpu: Replace rcu_assign_pointer() with RCU_INIT_POINTER()

2016-05-05T00:31:06Z

The rcu_assign_pointer() ensures that the initialization of a structure is carried out before storing a pointer to that structre. It is always safe to use RCU_INIT_POINTER() to NULL a pointer, instead of rcu_assign_pointer(). This results in slightly smaller/faster code. The following semantic patch was used: @@ @@ - rcu_assign_pointer + RCU_INIT_POINTER (..., NULL) Reviewed-by: Christian König Signed-off-by: Muhammad Falak R Wani Signed-off-by: Alex Deucher

drm/amdgpu: double fence slot

2016-05-05T00:22:48Z

we introduced vmid fence, so one hw submission could produce two fences. Signed-off-by: Chunming Zhou Reviewed-by: Christian König Signed-off-by: Alex Deucher

drm/amdgpu: Mark all instances of struct drm_info_list as const

2016-05-05T00:20:10Z

All these are compile time constand and the drm_debugfs_create/remove_files functions take a const pointer argument. Reviewed-by: Christian König Signed-off-by: Nils Wallménius Signed-off-by: Alex Deucher

drm/amdgpu: fence wait old rcu slot

2016-04-01T14:08:39Z

since the rcu slot was initialized to be num_hw_submission, if command submission doesn't use scheduler, this limitation will be invalid like uvd test. Signed-off-by: Chunming Zhou Reviewed-by: Christian König Signed-off-by: Alex Deucher