Thank you for your contribution! Sashiko AI review found 1 potential
issue to consider:
- [High] drm/atomic: lost driver state in
  drm_atomic_commit_fill_with_defaults()
--

commit da151559e1bf404876508e3465b551e3f9d62af2
Author: Maxime Ripard <[email protected]>

drm/atomic: Allow filling a commit with pristine object states

This commit adds drm_atomic_commit_fill_with_defaults() to populate an
atomic commit with pristine object states created via atomic_create_state()
while preserving adjusted_mode and best_encoder.

> diff --git a/drivers/gpu/drm/drm_atomic.c b/drivers/gpu/drm/drm_atomic.c
> index 7a77051096c91..43df13aaef1eb 100644
> --- a/drivers/gpu/drm/drm_atomic.c
> +++ b/drivers/gpu/drm/drm_atomic.c
> @@ -1649,6 +1649,327 @@ bool drm_atomic_implements_create_state(struct 
> drm_device *dev)
>  }
>  EXPORT_SYMBOL(drm_atomic_implements_create_state);
>  
> +int drm_atomic_commit_fill_with_defaults(struct drm_file *file_priv,
> +                                      struct drm_atomic_commit *commit)
> +{

[ ... ]

> +             crtc_state = crtc->funcs->atomic_create_state(crtc);
> +             if (IS_ERR(crtc_state))
> +                     return PTR_ERR(crtc_state);

[ ... ]

> +             drm_mode_copy(&crtc_state->adjusted_mode,
> +                           &crtc->state->adjusted_mode);
> +
> +             /*
> +              * The planes come back with the default zpos, which the
> +              * atomic helpers only normalize when the planes bound to a
> +              * CRTC or one of their zpos changed. Re-applying the same
> +              * configuration on top of the reset wouldn't trigger
> +              * either, so ask for a normalization explicitly.
> +              */
> +             crtc_state->zpos_changed = true;
> +
> +             drm_atomic_commit_set_crtc_state(commit, crtc, crtc_state);
> +     }
> +
> +     drm_for_each_plane(plane, dev) {

[ ... ]

> +             plane_state = plane->funcs->atomic_create_state(plane);
> +             if (IS_ERR(plane_state))
> +                     return PTR_ERR(plane_state);

[ ... ]

> +             connector_state = 
> connector->funcs->atomic_create_state(connector);
> +             if (IS_ERR(connector_state)) {
> +                     drm_connector_list_iter_end(&conn_iter);
> +                     ret = PTR_ERR(connector_state);
> +                     return ret;
> +             }
> +
> +             /*
> +              * best_encoder is a routing information maintained by the
> +              * atomic helpers rather than a userspace property, and it
> +              * is only carried over to a new state when that state is a
> +              * duplicate of the current one. Since
> +              * update_connector_routing() uses it to detect the commits
> +              * that keep the connector routing unchanged, carry the
> +              * encoder currently in use over so that a reset on which
> +              * userspace re-applies the exact same configuration doesn't
> +              * require a modeset.
> +              */
> +             connector_state->best_encoder = connector->state->best_encoder;

[Severity: High]
Does creating fresh states via atomic_create_state() while carrying over
adjusted_mode and best_encoder to avoid a modeset drop required
driver-private state in drivers that satisfy
drm_atomic_implements_create_state()?

In msm/mdp5, mdp5_plane_create_state() zeroes mdp5_state->hwpipe and
r_hwpipe, so mdp5_plane_atomic_check_with_state() passes NULL to
mdp5_pipe_release() and permanently leaks the previously assigned hardware
pipes in mdp5_global_state; if a CRTC stays active with no visible planes,
mdp5_crtc_atomic_check() also skips mdp5_crtc_setup_pipeline(), leaving
mdp5_cstate->pipeline.mixer NULL and causing a NULL pointer dereference in
blend_setup() during mdp5_crtc_atomic_flush().

In nouveau dispnv50, skipping a modeset when the active configuration is
re-applied across a reset causes nv50_wndw_atomic_check_acquire() to skip
populating asyw->image when the framebuffer is unchanged (making
curs507a_acquire() reject cursor commits with -EINVAL and zeroing
asyw->ntfy, sema, xlut, and csc handles on primary/overlay planes), leaves
asyh->mode, view, or, dither, procamp, and wndw.owned zeroed in
nv50_head_atomic_check(), and leaves asyc->set.mask zeroed when connector
properties are reset to their defaults.

-- 
Sashiko AI review ยท 
https://sashiko.dev/#/patchset/[email protected]?part=3

Reply via email to