[PATCH v2] drm: writeback: Document behaviour for framebuffer size mismatch

[PATCH v2] drm: writeback: Document behaviour for framebuffer size mismatch

[Derek Foreman]

Currently we have not explicitly defined the behaviour when the width
and height of the writeback framebuffer do not match the width and
height of the attached CRTC.

Allow the driver to perform a scale operation to fit the content, so
systems with scaling blocks in the writeback path can easily expose them.
A partially filled buffer should never be returned, so failure remains
an option for all the systems that don't have a scaler, or when scaler
limits are exceeded.

Cropping would make less sense, as the region to crop is underspecified
without some other properties to define an origin.

Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
---
I have a draft merge request against weston for this at:
 https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/2071

Changes in v2:
  - Make it clear that a buffer should not be partially filled

 drivers/gpu/drm/drm_writeback.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/gpu/drm/drm_writeback.c b/drivers/gpu/drm/drm_writeback.c
index 4da5d6094721..b4a002e6d043 100644
--- a/drivers/gpu/drm/drm_writeback.c
+++ b/drivers/gpu/drm/drm_writeback.c
@@ -58,6 +58,12 @@
  *	framebuffer to be written by the writeback connector. This property is
  *	similar to the FB_ID property on planes, but will always read as zero
  *	and is not preserved across commits.
+ *	If the width and height of the framebuffer do not match those of the
+ *	attached CRTC, the driver must either fail or scale (not crop) the
+ *	content to exactly fit the framebuffer.
+ *	If the driver is unable to exactly fill the framebuffer for any reason,
+ *	such as hardware scaler constraints or an odd width for a sub-sampled
+ *	format, the writeback must fail instead of partially filling the buffer.
  *	Userspace must set this property to an output buffer every time it
  *	wishes the buffer to get filled.
  *
-- 
2.53.0

Claude review: drm: writeback: Document behaviour for framebuffer size mismatch

[Claude Code Review Bot]

Patch Review

**Correctness:** The documentation accurately captures a sensible policy. It properly handles the three possible outcomes for size mismatch: scale (allowed), crop (disallowed), fail (always available as fallback). The "must fail instead of partially filling" language is a good addition in v2.

**Wording nit:** The phrase "an odd width for a sub-sampled format" at line 65:
```
+ *	such as hardware scaler constraints or an odd width for a sub-sampled
+ *	format, the writeback must fail instead of partially filling the buffer.
```
This example is reasonable but slightly tangential — odd-width-with-subsampling is more of a general format constraint issue than something specific to the scaling-vs-mismatch scenario being documented. It's fine as-is since it illustrates "unable to exactly fill," but a reviewer could argue it muddies the focus. Not worth blocking on.

**Integration with existing text:** The new text slots in cleanly between the description of what `WRITEBACK_FB_ID` stores and the instruction that "Userspace must set this property to an output buffer every time...". The flow reads naturally.

**Missing consideration:** The patch doesn't mention whether aspect ratio must be preserved during scaling, or whether arbitrary stretching is permitted. This is presumably left to the driver, which seems fine for now — adding aspect ratio constraints could come in a follow-up if needed.

**Reviewed-by: Looks good to merge.**

---
Generated by Claude Code Patch Reviewer

Claude review: drm: writeback: Document behaviour for framebuffer size mismatch

[Claude Code Review Bot]

Overall Series Review

Subject: drm: writeback: Document behaviour for framebuffer size mismatch
Author: Derek Foreman <derek.foreman@collabora.com>
Patches: 1
Reviewed: 2026-06-04T12:17:50.607491

---

This is a single documentation-only patch (v2) to the DRM writeback connector subsystem. It adds six lines of kernel-doc commentary to the `WRITEBACK_FB_ID` property documentation in `drm_writeback.c`, specifying what should happen when the writeback framebuffer dimensions differ from the attached CRTC dimensions.

The patch is well-motivated: the existing documentation left this behavior entirely undefined, which is a problem for both driver implementers and userspace (compositors like Weston). The chosen policy — allow scaling, forbid cropping, and require failure over partial fills — is reasonable and pragmatic:

- **Scaling** is the natural operation for hardware with scaler blocks in the writeback path.
- **Cropping** is correctly excluded since there's no property to specify a crop origin.
- **Failure over partial fill** is the safe default, preventing garbage data in the output buffer.

The v2 change (making it explicit that partial fills must not occur) is a good improvement.

No functional code changes, no risk of regressions. This is straightforward UAPI documentation clarification.

**Verdict: Looks good.** Minor wording nit below, but nothing blocking.

---
Generated by Claude Code Patch Reviewer