Re: [RFC 1/1] SWDEV476969 - dm: Fail dm_atomic_check if cursor overlay is required at MAX_SURFACES

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 






On 25/10/2024 16:37, Zaeem Mohamed wrote:
[why]
Prevent index-out-of-bounds due to requiring cursor overlay when
plane_count is MAX_SURFACES.
Hi Zaeem,

Thanks for working on this fix.

[how]
Bounds check on plane_count when requiring overlay cursor.
I agree. Atomic check makes sense.

1) Since the native cursor mode was previously the unique mode avaliable, I
wonder if the driver should fall to native cursor mode in favor of the overlay
planes advertised. I.e. if driver says it supports two overlay planes and
the userspace requested both, cursor overlay mode should not be available or
should switch to native cursor mode, as before the introduction of cursor
overlay mode.

2) Then my second question: can we increase the number of surfaces to 4 first to accommodate more than one active overlay plane with cursor overly mode enabled.
If four is still possible, this increase can reduce the number of commit
failure scenarios and mitigate current userspace issues first. After addressing
current array-out-of-bounds, follow-up patches can do the proper changes and
checks.

3) IMHO, the incoherence between MAX_SURFACE_NUM and MAX_SURFACE should be
addressed before adding debugging points. For example, there are part of the
DC code using MAX_SURFACE_NUM == MAX_PLANE == 6 to allocate dc_surface_update arrays, instead of using MAX_SURFACE value. You can find one of this case here:
https://gitlab.freedesktop.org/agd5f/linux/-/blob/amd-staging-drm-next/drivers/gpu/drm/amd/display/dc/core/dc.c#L4507
It doesn't make sense to me and it can contribute to an incomplete solution.

Also, please add the references of bugs reported in the amd tracker, so far:

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/3693
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/3594
Co-developed-by: Melissa Wen <mwen@xxxxxxxxxx>
I don't think I contributed enough to your code to get any credits.
Thanks, but you can remove my co-dev-by :)

Best Regards,

Melissa
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@xxxxxxx>
---
  amdgpu_dm/amdgpu_dm.c | 16 +++++++++++++++-
  1 file changed, 15 insertions(+), 1 deletion(-)

diff --git a/amdgpu_dm/amdgpu_dm.c b/amdgpu_dm/amdgpu_dm.c
index df83e7b42b..c2595efb74 100644
--- a/amdgpu_dm/amdgpu_dm.c
+++ b/amdgpu_dm/amdgpu_dm.c
@@ -11676,6 +11676,12 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev,
  		 * need to be added for DC to not disable a plane by mistake
  		 */
  		if (dm_new_crtc_state->cursor_mode == DM_CURSOR_OVERLAY_MODE) {
+			if(dc->current_state->stream_status->plane_count >= MAX_SURFACES){
+				drm_dbg_driver(crtc->dev,
+				       "Can't enable cursor plane with %d planes\n", MAX_SURFACES);
+				ret = -EINVAL;
+				goto fail;
+			}
  			ret = drm_atomic_add_affected_planes(state, crtc);
  			if (ret)
  				goto fail;
@@ -11769,8 +11775,16 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev,
/* Overlay cusor not subject to native cursor restrictions */
  		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
-		if (dm_new_crtc_state->cursor_mode == DM_CURSOR_OVERLAY_MODE)
+		if (dm_new_crtc_state->cursor_mode == DM_CURSOR_OVERLAY_MODE){
+			if(dc->current_state->stream_status->plane_count > MAX_SURFACES){
+				drm_dbg_driver(crtc->dev,
+				       "Can't enable cursor plane with %d planes\n", MAX_SURFACES);
+				ret = -EINVAL;
+				goto fail;
+			}
+			
  			continue;
+		}
/* Check if rotation or scaling is enabled on DCN401 */
  		if ((drm_plane_mask(crtc->cursor) & new_crtc_state->plane_mask) &&




[Index of Archives]     [Linux DRI Users]     [Linux Intel Graphics]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux