On Sun, Oct 27, 2024 at 06:23:50PM +0200, Laurent Pinchart wrote:
> On Thu, Oct 24, 2024 at 11:55:38AM +0200, Herve Codina wrote:
> > In some cases observed during ESD tests, the TI SN65DSI83 cannot recover
> > from errors by itself. A full restart of the bridge is needed in those
> > cases to have the bridge output LVDS signals again.
> >
> > The TI SN65DSI83 has some error detection capabilities. Introduce an
> > error recovery mechanism based on this detection.
> >
> > The errors detected are signaled through an interrupt. On system where
> > this interrupt is not available, the driver uses a polling monitoring
> > fallback to check for errors. When an error is present, the recovery
> > process is launched.
> >
> > Restarting the bridge needs to redo the initialization sequence. This
> > initialization sequence has to be done with the DSI data lanes driven in
> > LP11 state. In order to do that, the recovery process resets the entire
> > pipeline.
> >
> > Signed-off-by: Herve Codina <herve.codina@xxxxxxxxxxx>
> > ---
> > drivers/gpu/drm/bridge/ti-sn65dsi83.c | 128 ++++++++++++++++++++++++++
> > 1 file changed, 128 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/bridge/ti-sn65dsi83.c b/drivers/gpu/drm/bridge/ti-sn65dsi83.c
> > index 96e829163d87..22975b60e80f 100644
> > --- a/drivers/gpu/drm/bridge/ti-sn65dsi83.c
> > +++ b/drivers/gpu/drm/bridge/ti-sn65dsi83.c
> > @@ -35,9 +35,12 @@
> > #include <linux/of_graph.h>
> > #include <linux/regmap.h>
> > #include <linux/regulator/consumer.h>
> > +#include <linux/timer.h>
> > +#include <linux/workqueue.h>
> >
> > #include <drm/drm_atomic_helper.h>
> > #include <drm/drm_bridge.h>
> > +#include <drm/drm_drv.h> /* DRM_MODESET_LOCK_ALL_BEGIN() need drm_drv_uses_atomic_modeset() */
> > #include <drm/drm_mipi_dsi.h>
> > #include <drm/drm_of.h>
> > #include <drm/drm_panel.h>
> > @@ -147,6 +150,9 @@ struct sn65dsi83 {
> > struct regulator *vcc;
> > bool lvds_dual_link;
> > bool lvds_dual_link_even_odd_swap;
> > + bool use_irq;
> > + struct delayed_work monitor_work;
> > + struct work_struct reset_work;
> > };
> >
> > static const struct regmap_range sn65dsi83_readable_ranges[] = {
> > @@ -321,6 +327,92 @@ static u8 sn65dsi83_get_dsi_div(struct sn65dsi83 *ctx)
> > return dsi_div - 1;
> > }
> >
> > +static int sn65dsi83_reset_pipeline(struct sn65dsi83 *sn65dsi83)
> > +{
> > + struct drm_device *dev = sn65dsi83->bridge.dev;
> > + struct drm_modeset_acquire_ctx ctx;
> > + struct drm_atomic_state *state;
> > + int err;
> > +
> > + /* Use operation done in drm_atomic_helper_suspend() followed by
> > + * operation done in drm_atomic_helper_resume() but without releasing
> > + * the lock between suspend()/resume()
> > + */
> > +
> > + DRM_MODESET_LOCK_ALL_BEGIN(dev, ctx, 0, err);
> > +
> > + state = drm_atomic_helper_duplicate_state(dev, &ctx);
> > + if (IS_ERR(state)) {
> > + err = PTR_ERR(state);
> > + goto unlock;
> > + }
> > +
> > + err = drm_atomic_helper_disable_all(dev, &ctx);
> > + if (err < 0)
> > + goto unlock;
> > +
> > + drm_mode_config_reset(dev);
> > +
> > + err = drm_atomic_helper_commit_duplicated_state(state, &ctx);
>
> Committing a full atomic state from a bridge driver in an asynchronous
> way seems quite uncharted territory, and it worries me. It's also a very
> heavyweight, you disable all outputs here, instead of focussing on the
> output connected to the bridge. Can you either implement something more
> local, resetting the bridge only, or create a core helper to handle this
> kind of situation, on a per-output basis ?
I think you can't just shut down the bridge and restart it, since some
require particular power sequences that will only occur if you also shut
down the upstream controller.
So I think we'd need to tear down the CRTC, connector and everything
in between.
I do agree that it needs to be a generic helper though. In fact, it
looks awfully similar to what vc4 and i915 are doing in reset_pipe and
and intel_modeset_commit_pipes, respectively. It looks like a good
opportunity to make it a helper.
Maxime
Attachment:
signature.asc
Description: PGP signature
