Hi Dinh:
Do you think it might be a problem of a un-aligned DMA buffer
address? We know that the usbnet driver submits lots of URBs with a
2-byte aligned buffer address. But in Buffer DMA mode, the starting DMA
address must be DWORD-aligned, only if turn on Descriptor DMA mode this
4-byte aligned limit disappear. We have met this kind of problem in our
platform which powered by a arm based SoC, RK3288, using the upstream
dwc2 driver.
发自我的 iPad
> on 7-15-2014，10:39，Dinh Nguyen <dinh.linux@xxxxxxxxx> wrote：
>
> Hi Paul,
>
> On 6/25/14, 1:24 PM, Paul Zimmerman wrote:
>>> From: Dinh Nguyen [mailto:dinh.linux@xxxxxxxxx]
>>> Sent: Wednesday, June 25, 2014 8:52 AM
>>>
>>> I was wondering if you have ever tested this driver with a USB/ethernet
>>> dongle? I'm using the apple usb/ethernet dongle, which is basically just
>>> the ASIX AX88xxx Based Ethernet Adapter. From brief debugging, it
>>> appears that Bulk and Interrupt endpoint data are getting corrupted
>>> during the data transfer.
>>>
>>> Will continue to debug, but was just curious if you have tested this
>>> kind of device?
>> Hi Dinh,
>>
>> Yes, I have tested the driver with a couple of different Ethernet
>> dongles. That was a while ago though, so I don't know if anything has
>> changed lately to break that. I don't remember what brand of Ethernet
>> dongle I used, unfortunately.
> I managed to debug this a bit further. The ethernet dongles all work
> fine on my
> raspberry pi. But on the SOCFPGA platform, which has version 2.93a of
> the USB
> IP, the ethernet dongle seems to only work if I turn on Descriptor DMA and
> uframe_sched=0.
>
> Dinh
> --
> To unsubscribe from this list: send the line "unsubscribe linux-usb" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
