On 06/10/25 04:42, Dmitry Baryshkov wrote:
On Mon, Sep 29, 2025 at 02:09:33PM +0530, Harikrishna Shenoy wrote:
From: Jayesh Choudhary <[email protected]>
In case if we get errors in cdns_mhdp_link_up() or cdns_mhdp_reg_read()
in atomic_enable, we will go to cdns_mhdp_modeset_retry_fn() and will hit
NULL pointer while trying to access the mutex. We need the connector to
be set before that. Unlike in legacy !(DBANC) cases, we do not have
connector initialised in bridge_attach(). Now that we have the connector
pointer in mhdp bridge structure, so set the mhdp->connector in
atomic_enable() earlier to avoid possible NULL pointer dereference
in recovery paths like modeset_retry_fn() with the DBANC flag set.
Fixes: c932ced6b585 ("drm/tidss: Update encoder/bridge chain connect model")
This Fixes tag means that this patch can be attempted to be backported
back to v6.5 (even w/o cc:stable, etc). I know that it is a pain, but
please move all Fixes to the top of the series. Yes, you want to drop
non-DBANC case first and then fix everything. It doesn't look like it is
a correct approach for the sake of maintaing the -stable branches.
Hi Dmitry,
The patch which drops non-DBANC case can be moved after fixes, but for
making fixes i need the pointer structure so adding fixes tag to PATCH
2/6 "drm/bridge: cadence: cdns-mhdp8546*: Change
drm_connector from structure to pointer"
Let me know your thoughts on this, will re-spin accordingly.
Regards.
Signed-off-by: Jayesh Choudhary <[email protected]>
Signed-off-by: Harikrishna Shenoy <[email protected]>
---
.../drm/bridge/cadence/cdns-mhdp8546-core.c | 20 +++++++++----------
1 file changed, 10 insertions(+), 10 deletions(-)
