Hello Yongxing, Dmitry,
> If a downstream consumer (e.g. drm_bridge_connector attached by the
> msm/dp driver) registers its HPD callback after an upstream driver
> has already reported a HPD event through drm_aux_hpd_bridge_notify(),
> the notification is dropped because bridge->hpd_cb is still NULL.
> This can affect any user of drm_aux_hpd_bridge_notify() whose
> downstream consumer arms HPD only after upstream events have started.
>
> The race has been observed on Qualcomm X1E-based laptops during boot,
> when pmic_glink_altmode reports the initial USB-C DP connection state
> before the DP driver has finished probing and enabled HPD handling on
> the bridge. The consumer then never observes the initial connected
> state and the external display remains dark.
>
> Cache the last HPD status reported through drm_aux_hpd_bridge_notify()
> and replay it when HPD is enabled by the downstream consumer.
>
> The replay is deferred to a work item so that the replayed HPD
> notification is delivered outside drm_bridge_hpd_enable()'s call
> context.
>
> This follows the same pattern as display-connector, which also defers
> an initial HPD notification from .hpd_enable(), but reuses the cached
> status since aux-hpd-bridge cannot re-detect sink presence on its own.
>
> Fixes: e560518a6c2e ("drm/bridge: implement generic DP HPD bridge")
> Signed-off-by: Yongxing Mou <[email protected]>
Is this patch still current?
It looks mostly OK at quick glance, but there are issues reported by
sashiko which look valid to me. If you could reply to those remarks, either
confirming the issue or explaining why it isn't, it would help reviewing.
@Dmitry: I'm not familiar with this code, do you have any comment on this
patch?
Luca
--
Luca Ceresoli, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
