Re: [PATCH v2 1/1] usb: dwc3: gadget: fix IRQ storm on invalid event buffer count

[Thinh Nguyen <Thinh.Nguyen@synopsys.com>]

On Wed, Sep 02, 2026, Liu Jiazi wrote:
> Hi @Thinh Nguyen
> 
> Tested on our device with the patch you provided.
> Trigger the error by writing a bogus value to GEVNTCOUNT and the
> controller recovers successfully
> USB re-enumrates and ADB comes back. No kernel panic observed.
> 
> uart print:
> [ 1892.781261][    C0] dwc3 b1800000.usb: invalid count(61436) >
> evt->length(4096)
> [ 1892.834341][ T6793] dwc3 b1800000.usb: request ffffff81475856c0 was
> not queued to ep0out
> [ 1895.042752][ T9114] ffs_data_put(): freeing
> [ 1896.206669][ T9142] read descriptors
> [ 1896.207465][ T9142] bcdVersion must be 0x0100, stored in Little
> Endian order. Userspace driver should be
> fixed, accepting 0x0001 for compatibility.
> [ 1896.209638][ T9142] bcdVersion must be 0x0100, stored in Little
> Endian order. Userspace driver should be
> fixed, accepting 0x0001 for compatibility.
> [ 1896.211754][ T9142] read strings
> [ 1897.244130][ T9123] FFS endpoints ready, proceeding with UDC bind
> [ 1897.258493][ T9123] bcdVersion must be 0x0100, stored in Little
> Endian order. Userspace driver should be
> fixed, accepting 0x0001 for compatibility.
> [ 1897.260970][ T9123] bcdVersion must be 0x0100, stored in Little
> Endian order. Userspace driver should be
> fixed, accepting 0x0001 for compatibility.
> 

Thanks for testing and for verifying the fix.

If there are no remaining concerns, could you please submit the patch?
Otherwise, let me know if you notice anything else that needs to be
addressed.

Thanks,
Thinh