Re: [REGRESSION] usb: core: interrupt-IN wBytesPerInterval fixup (727d045d064b) breaks RTL8153 in Lenovo dock

[Michal Pecio <michal.pecio@gmail.com>]

On Tue, 16 Jun 2026 20:22:19 +0330, Amin Vakil wrote:
> On 6/16/26 8:06 PM, Michal Pecio wrote:
> > RTL8153 was one of the chips specifically targeted by this fix, but
> > it affects the CDC configuration which you don't seem to be using.
> > 
> > I have tried linux-hardened-7.0.12.hardened1-1-x86_64 from Arch repo
> > as well as my own build of 7.1-rc6, and my RTL8153 (0bda:8153) in a
> > different kind of USB-C dock still works with a few xHCI controllers.
> > 
> > So it's some weird problem narrowly specific to your system. Now that
> > you have built your own kernel, please also revert the suspect patch
> > to verify if that's really it.  
> 
> Honestly I've tried patching it myself :)
> 
> by changing
> 
> (le16_to_cpu(desc->wBytesPerInterval) < usb_endpoint_maxp(&ep->desc) &&
> 
> to
> 
> (le16_to_cpu(desc->wBytesPerInterval) == 0 &&
> 
> in line 187 of drivers/usb/core/config.c to see if that fixes the issue 
> and it did not, but I didn't specifically reverted the commit to see if 
> that fixes the issue.

This should be enough to prevent overriding wBytesPerInterval on
RTL8153 (and get rid of the new log message).

If the message goes away but the device still doesn't work, the problem
must be somewhere else.

I would start with userspace - if you haven't already, downgrade to
some kernel package which worked in the past. See if it still works.
Or check if 9ddb9c0deca48d2c2a22ebf4d2f35c925a520328 applies to your
dock and try reverting it if it does.

Otherwise, no idea, besides git-bisect I guess.

Regards,
Michal