Re: [PATCH] drm/bridge: sn65dsi83: perform host_attach on attach instead of probe

[Maxime Ripard]

Hi,x

On Tue, Jul 07, 2026 at 10:46:55AM +0200, Paul Geurts wrote:
> When connecting the sn65dsi83 device to the Samsung MIPI DSI host
> controller found in i.MX8M Mini, probing the host and device
> infinitely defers, waiting for each other to appear. on probe of the
> host, it tries to attach to the device, which has not been probed yet.
> It therefore does not register the host, and defers. When the sn65dsi83
> probes, it tries to attach to the host, which is not registered as it is
> deferred, and thereofore it defers as well, without registering. This
> means the two drivers are infinetely waiting for each other to appear.
> This is the case on the Samsung MIPI DSI host controller, but might be
> the case with other host controllers as well.
> 
> Only register the sn65dsi83 on probe, and attach to the host later. This
> means the host has a bridge to attache to in probe. When the host
> attaches to the bridge, the bridge can attach to the host, breaking the
> infinite defer.
> 
> Signed-off-by: Paul Geurts <paul.geurts@xxxxxxxxxxxxxxxxxxxxxxxxx>
> Fixes: 6ef7ee48765f ("drm/bridge: sn65dsi83: Register and attach our DSI device at probe")

I think the problem is actually that DSIM doesn't attach the bridge in
its MIPI-DSI host attach hook, as documented here:

https://www.kernel.org/doc/html/latest/gpu/drm-kms-helpers.html#special-care-with-mipi-dsi-bridges

If it doesn't fix the issue either, we should revise the doc further.

Maxime

Attachment: signature.asc
Description: PGP signature