On Wed, Nov 18, 2026 at 04:02:40PM +0000, [email protected] wrote:
> Thank you for your contribution! Sashiko AI review found 3 potential
> issues to consider:
> - [Medium] drm/imx/dcif: unclocked MMIO access when runtime PM resume
>   fails
> - [Medium] drm/imx: dcif: unbalanced clock disable if
>   dcif_runtime_resume() fails
> - [Low] drm/imx/dcif: missing shadow load and trigger clear in
>   dcif_crc_source_disable()
> --
> 
> commit 373caa0eccbfd7fa4334093ea8869b540640e7ab
> Author: Sandor Yu <[email protected]>
> 
> drm/imx: Add support for i.MX94 DCIF
> 
> This commit adds a DRM/KMS driver for the NXP i.MX94 Display Control
> Interface (DCIF), supporting primary and overlay planes, YCbCr-to-RGB CSC
> on layer 0, and frame/ROI CRC generation.
> 
> > diff --git a/drivers/gpu/drm/imx/dcif/dcif-crc.c 
> > b/drivers/gpu/drm/imx/dcif/dcif-crc.c
> > new file mode 100644
> > index 0000000000000..ad94b458a6c59
> > --- /dev/null
> > +++ b/drivers/gpu/drm/imx/dcif/dcif-crc.c
> [ ... ]
> > +void dcif_crc_source_enable(struct dcif_dev *dcif, enum dcif_crc_source 
> > source,
> > +                       struct drm_rect *roi, int ncrc)
> > +{
> > +   int ret;
> > +
> > +   if (ncrc >= MAX_DCIF_CRC_NUM)
> > +           return;
> > +
> > +   if (source == DCIF_CRC_SRC_NONE)
> > +           return;
> > +
> > +   if (dcif->crc_is_enabled)
> > +           return;
> > +
> > +   ret = drm_crtc_vblank_get(&dcif->crtc);
> > +   if (ret) {
> > +           drm_err(&dcif->drm, "Could not get vblank reference: %d\n", 
> > ret);
> > +           return;
> > +   }
> > +
> > +   dcif_crc_config(dcif, roi, ncrc);
> > +
> > +   regmap_set_bits(dcif->regmap, DCIF_CRC_CTRL,
> > +                   DCIF_CRC_CTRL_CRC_MODE | 
> > DCIF_CRC_CTRL_CRC_SHADOW_LOAD_EN |
> > +                   DCIF_CRC_CTRL_CRC_TRIG);
> > +
> > +   scoped_guard(spinlock_irqsave, &dcif->drm.event_lock)
> > +           dcif->crc_is_enabled = true;
> > +}
> > +
> > +void dcif_crc_source_disable(struct dcif_dev *dcif, int ncrc)
> > +{
> > +   if (!dcif->crc_is_enabled)
> > +           return;
> > +
> > +   if (ncrc >= MAX_DCIF_CRC_NUM)
> > +           return;
> > +
> > +   regmap_clear_bits(dcif->regmap, DCIF_CRC_CTRL, 
> > DCIF_CRC_CTRL_CRC_EN(ncrc));
> 
> [Severity: Low]
> Should dcif_crc_source_disable() also clear DCIF_CRC_CTRL_CRC_TRIG and set
> DCIF_CRC_CTRL_CRC_SHADOW_LOAD_EN when disabling CRC on an active CRTC?
> 
> Clearing only DCIF_CRC_CTRL_CRC_EN(ncrc) leaves the disabled state
> unlatched, and if DCIF_CRC_CTRL_CRC_MODE, DCIF_CRC_CTRL_CRC_SHADOW_LOAD_EN
> and DCIF_CRC_CTRL_CRC_TRIG stay set in DCIF_CRC_CTRL, a later
> dcif_crc_source_enable() call via regmap_set_bits() will see no bit changes
> and skip writing the register.

It appears this one is correct. The CRC_EN(n) is shadowed and CRC_TRIG
is not self-clearing in continuous mode. Will fix in v13.

> 
> [ ... ]
> 
> > diff --git a/drivers/gpu/drm/imx/dcif/dcif-crtc.c 
> > b/drivers/gpu/drm/imx/dcif/dcif-crtc.c
> > new file mode 100644
> > index 0000000000000..233e149929964
> > --- /dev/null
> > +++ b/drivers/gpu/drm/imx/dcif/dcif-crtc.c
> [ ... ]
> > +static void dcif_crtc_atomic_enable(struct drm_crtc *crtc,
> > +                               struct drm_atomic_commit *state)
> > +{
> > +   struct drm_crtc_state *crtc_state = 
> > drm_atomic_get_new_crtc_state(state, crtc);
> > +   struct dcif_crtc_state *dcif_crtc_state = 
> > to_dcif_crtc_state(crtc_state);
> > +   struct drm_display_mode *adj = &crtc_state->adjusted_mode;
> > +   struct dcif_dev *dcif = crtc_to_dcif_dev(crtc);
> > +   struct drm_device *drm = crtc->dev;
> > +   int ret;
> > +
> > +   dev_dbg(drm->dev, "mode " DRM_MODE_FMT "\n", DRM_MODE_ARG(adj));
> > +
> > +   /* enable power when we start to set mode for CRTC */
> > +   ret = pm_runtime_resume_and_get(drm->dev);
> > +   if (ret < 0) {
> > +           drm_err(drm, "failed to resume DCIF, ret = %d\n", ret);
> > +           return;
> > +   }
> > +   dcif->crtc_pm_enabled = true;
> [ ... ]
> > +static void dcif_crtc_atomic_disable(struct drm_crtc *crtc,
> > +                                struct drm_atomic_commit *state)
> > +{
> > +   struct dcif_dev *dcif = crtc_to_dcif_dev(crtc);
> > +   struct drm_device *drm = crtc->dev;
> > +
> > +   if (dcif->has_crc)
> > +           dcif_crc_source_disable(dcif, 0);
> > +
> > +   dcif_disable_controller(dcif);
> > +   dcif_disable_plane_panic(dcif);
> > +
> > +   drm_crtc_vblank_off(crtc);
> > +
> > +   if (dcif->crtc_pm_enabled) {
> > +           dcif->crtc_pm_enabled = false;
> > +           pm_runtime_put_sync(drm->dev);
> > +   }
> 
> [Severity: Medium]
> Can a failure in pm_runtime_resume_and_get() lead to MMIO register accesses
> while the DCIF clocks are disabled?
> 
> If pm_runtime_resume_and_get() fails in dcif_crtc_atomic_enable(),
> dcif->crtc_pm_enabled remains false, but crtc_state->active is still true.
> In the same commit tail, drm_atomic_helper_commit_planes() still calls
> dcif_plane_atomic_update() and dcif_crtc_atomic_flush(), and a later
> dcif_crtc_atomic_disable() calls dcif_crc_source_disable(),
> dcif_disable_controller(), and dcif_disable_plane_panic() before checking
> dcif->crtc_pm_enabled.  Accessing DCIF registers while the block is
> unclocked can cause an external abort or bus hang on i.MX94.

This one is not worth fixing. First, the only thing that can fail in the
runtime resume is clk_bulk_prepare_enable(). If that fails, the platform
is already broken since the clocks are managed by the SCMI firmware. Second,
adding checks dor crtc_pm_enabled in every commit path (flush, plane
update, etc) would add a lot of unnecessary noise for something that
will probably never happen in practice.

> 
> [ ... ]
> 
> > diff --git a/drivers/gpu/drm/imx/dcif/dcif-drv.c 
> > b/drivers/gpu/drm/imx/dcif/dcif-drv.c
> > new file mode 100644
> > index 0000000000000..80b554f1b6843
> > --- /dev/null
> > +++ b/drivers/gpu/drm/imx/dcif/dcif-drv.c
> [ ... ]
> > +static int dcif_suspend(struct device *dev)
> > +{
> > +   struct dcif_dev *dcif = dev_get_drvdata(dev);
> > +   int ret;
> > +
> > +   ret = drm_mode_config_helper_suspend(&dcif->drm);
> > +   if (ret < 0)
> > +           return ret;
> > +
> > +   if (pm_runtime_suspended(dev))
> > +           return 0;
> > +
> > +   return dcif_runtime_suspend(dev);
> > +}
> > +
> > +static int dcif_resume(struct device *dev)
> > +{
> > +   struct dcif_dev *dcif = dev_get_drvdata(dev);
> > +   int ret;
> > +
> > +   if (!pm_runtime_suspended(dev)) {
> > +           ret = dcif_runtime_resume(dev);
> > +           if (ret < 0)
> > +                   return ret;
> > +   }
> > +
> > +   return drm_mode_config_helper_resume(&dcif->drm);
> > +}
> 
> [Severity: Medium]
> Can a failure in dcif_runtime_resume() during system resume cause an
> unbalanced clk_bulk_disable_unprepare()?
> 
> When the CRTC is active entering system suspend, the PM core's
> device_prepare() holds a runtime PM reference, so the pm_runtime_put_sync()
> in dcif_crtc_atomic_disable() leaves the device's runtime PM status as
> RPM_ACTIVE.  dcif_suspend() then calls dcif_runtime_suspend() directly,
> disabling clocks without updating runtime_status.  If dcif_runtime_resume()
> fails in dcif_resume(), clocks remain disabled while runtime_status is
> still RPM_ACTIVE, and device_complete() -> pm_runtime_put() will invoke
> dcif_runtime_suspend() a second time.  Using pm_runtime_force_suspend() and
> pm_runtime_force_resume() would keep runtime_status aligned with the clock
> state.

This is related to the previous issue and also not worth fixing since
the resume is unlikely to fail, as I explained above.

-- 
Thanks,
Laurentiu

Reply via email to