Re: uhci_hcd: Possible corruption of DMA pool uhci->td_pool

[Eugene Shatokhin <eugene.shatokhin@rosalab.ru>]

On 12/11/2013 08:41 PM, Alan Stern wrote:
> On Wed, 11 Dec 2013, Eugene Shatokhin wrote:
>
>> Hi,
>>
>> On ROSA Linux with kernel 3.10.21 with DMA debug options enabled, the
>> kernel sometimes issues a warning about DMA pool corruption (see the log
>> below).
>>
>> That happens sometimes, when the system boots or resumes from
>> hibernation with Samson C01U USB microphone attached.
>>
>> The affected DMA pool is 'uhci->td_pool', uhci_alloc_td() from
>> drivers/usb/host/uhci-hcd.c makes the relevant dma_pool_alloc() calls.
>>
>> Any ideas about how to find what causes this and how to fix it?
>
> This is not an easy sort of thing to track down...
>
>> Here is the relevant part of the system log:
>> ----------------------------
>> [   22.264332] usb 2-1: new full-speed USB device number 2 using uhci_hcd
>> [   22.450609] usb 2-1: New USB device found, idVendor=17a0, idProduct=0001
>> [   22.450626] usb 2-1: New USB device strings: Mfr=1, Product=2,
>> SerialNumber=0
>> [   22.450639] usb 2-1: Product: Samson C01U
>> [   22.450649] usb 2-1: Manufacturer: Samson Technologies
>> <...>
>> [  280.703483] retire_capture_urb: 4494 callbacks suppressed
>> [  284.961087] uhci_hcd 0000:00:1d.1: dma_pool_alloc uhci_td, efb7b060
>> (corruped)
>> [  284.961087] 00000000: 00 06 00 00 af 00 00 03 a7 a7 a7 a7 a7 a7 a7 a7
>>    ................
>> [  284.961087] 00000010: a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7
>>    ................
>> [  284.961087] 00000020: a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7
>>    ................
>> [  284.961087] retire_capture_urb: 4343 callbacks suppressed
>> [  284.961087] uhci_hcd 0000:00:1d.1: dma_pool_alloc uhci_td, efb7b5d0
>> (corruped)
>> [  284.961087] 00000000: 00 06 00 00 af 00 00 03 a7 a7 a7 a7 a7 a7 a7 a7
>>    ................
>> [  284.961087] 00000010: a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7
>>    ................
>> [  284.961087] 00000020: a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7
>>    ................
>> [  284.961087] cannot submit urb (err = -27)
>> [  284.961087] cannot submit urb (err = -27)
>> [  284.961087] cannot submit urb (err = -27)
>> [  284.961087] cannot submit urb (err = -27)
>> [  284.961087] cannot submit urb (err = -27)
>> [  284.961087] cannot submit urb (err = -27)
>> [  284.961087] cannot submit urb (err = -27)
>> [  284.961087] cannot submit urb (err = -27)
>> ----------------------------
>>
>> 0xa7 is POOL_POISON_FREED. The memory pages to be allocated from the
>> pool should be filled with such bytes.
>>
>> Each time I observed this problem, the first 8 bytes of the listed
>> memory area were overwritten, with different data each time.
>
> It kind of looks like a hardware bug.  Still, it's hard to say.
>
> Can you test the current 3.13-rc kernel?  There have been a few recent
> changes in this area that might have an effect.
>

I tested 3.13-rc3 - the problem has not shown up so far.

Regards,
Eugene

-- 
Eugene Shatokhin, ROSA Laboratory.
www.rosalab.com