Hi,
Sekhar Nori <nsek...@ti.com> writes:
>> Roger Quadros <rog...@ti.com> writes:
>>>> From: Janusz Dziedzic <januszx.dzied...@intel.com>
>>>>
>>>> In the case of bounced ep0 requests, we must delay DMA operation until
>>>> after ->complete() otherwise we might overwrite contents of req->buf.
>>>>
>>>> This caused problems with RNDIS gadget.
>>>>
>>>> Signed-off-by: Janusz Dziedzic <januszx.dzied...@intel.com>
>>>> Signed-off-by: Felipe Balbi <felipe.ba...@linux.intel.com>
>>>> ---
>>>
>>> Sekhar found out that this fixes RNDIS for us on v4.9.
>>> Could you please tag this for stable? Thanks.
>>
>> Was this a regression on v4.9 or was it already broken? Was it broken
>> forever or which commit regressed it? If it falls into
>> "has-never-worked-before" then I'm not sure we should tag stable. If we
>> can blame a commit, then I can send the patch to stable, no issues.
>
> Mainline v4.9 works. v4.9.20 is broken because commit d62145929992
> ("usb: dwc3: gadget: always unmap EP0 requests") was backported it.
>
> I think the $SUBJECT patch needs to be propagated to all the stable
> kernels to which commit d62145929992 was back-ported to.okay, thanks for letting me know :-) -- balbi
signature.asc
Description: PGP signature
