On 2019-11-11 6:25 p.m., Yong Zhao wrote:
Given control stack is now in the userspace context save restore area
on GFX10, the same as GFX8, it is not needed to copy it back to userspace.

Change-Id: I063ddc3026eefa57713ec47b466a90f9bf9d49b8
Signed-off-by: Yong Zhao <yong.z...@amd.com>
Patches 1 and 2 are

Reviewed-by: Felix Kuehling <felix.kuehl...@amd.com>


Patch 3 should arguably not be part of this series, because it does not affect GFXv10.

Regards,
  Felix



---
  drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c | 14 +++++++++-----
  1 file changed, 9 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c 
b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c
index 4884cd6c65ce..954dc8ac4ff1 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c
@@ -251,18 +251,22 @@ static int get_wave_state(struct mqd_manager *mm, void 
*mqd,
  {
        struct v10_compute_mqd *m;
- /* Control stack is located one page after MQD. */
-       void *mqd_ctl_stack = (void *)((uintptr_t)mqd + PAGE_SIZE);
-
        m = get_mqd(mqd);
+ /* Control stack is written backwards, while workgroup context data
+        * is written forwards. Both starts from m->cp_hqd_cntl_stack_size.
+        * Current position is at m->cp_hqd_cntl_stack_offset and
+        * m->cp_hqd_wg_state_offset, respectively.
+        */
        *ctl_stack_used_size = m->cp_hqd_cntl_stack_size -
                m->cp_hqd_cntl_stack_offset;
        *save_area_used_size = m->cp_hqd_wg_state_offset -
                m->cp_hqd_cntl_stack_size;
- if (copy_to_user(ctl_stack, mqd_ctl_stack, m->cp_hqd_cntl_stack_size))
-               return -EFAULT;
+       /* Control stack is not copied to user mode for GFXv10 because
+        * it's part of the context save area that is already
+        * accessible to user mode
+        */
return 0;
  }
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Reply via email to