[PATCH v3 3/3] usb: dwc3: imx8mp: disable auto suspend for host role

[Xu Yang]

On Tue, May 19, 2026 at 10:54:58AM +0200, Franz Schnyder wrote:
> Hi Xu
> 
> On Tue, May 12, 2026 at 05:53:57PM +0800, Xu Yang wrote:
> > 
> > OK. I mean, does dwc3_imx8mp_probe() still succeed after the kernel dumps
> > at the end?
> Ah yes, afterwards it still succeeds.
> 
> > 
> > OK. More debug information will be helpful.
> > 
> 
> I've attached the logs below with and without the commit. looking at 
> what happens before the warning I think the problem is that after probe
> deferral the depopulate path races with the device link state changes 
> from the fw_devlink cycle fixup.
> 
> One important thing to mention is, that we build dwc3 and typec as modules.
> If I build them as built-in I can't reproduce the kernel warning so far. 

Thank your for the information.

It's indeed an important finding. I can reproduce the issue when I build the
driver as modules.

But I guess it's a case that the driver core doesn't handle well. I will
seek help from the driver-core community to see how to improve it.

https://lore.kernel.org/driver-core/DU2PR04MB8822AF9D63FDC335D80DABC08C082@DU2PR04MB8822.eurprd04.prod.outlook.com/T/#u

Thanks,
Xu Yang