Re: usb HC busted?

[Mathias Nyman <mathias.nyman-VuQAYsv1563Yd54FQh9/CA@public.gmane.org>]

On 27.06.2018 14:59, Sudip Mukherjee wrote:
>>> Can you share a bit more details on the platform you are using, and what types of test you are running.
>>
>> It is a board based on "Intel(R) Atom(TM) CPU  E3840  @ 1.91GHz".
>> The usb device in question is a bluetooth device:
>>
>> Bus 001 Device 012: ID 8087:07dc Intel Corp.
> <snip>
>>
>> And the problem that we are seeing is with phone calls via bluetooth.
>>
>>> Does my test above trigger the case? (show "MATTU dmatest match!")
>>
>> I have kept it for tonight, will see the results tomorrow morning.
>> And I am using that same device in the usb script to change "authrized".
> 
> No, your test did not trigger the error. :(
> 
> But, my last night's test (with an added debug to get some extra trace for addresses) showed the same error of -
> "Looking for event-dma", but looking at the ftrace, I could not see it getting same address from dma_pool_zalloc().
> 
> Can you please have a look at the dmesg and ftrace at:
> https://drive.google.com/open?id=1nMy_qVxOQzcZNYa9bw7az9WiS2MZzdKo
> 

There is however freeing of the same dma address:

<...>-28448 [003] ....   492.025808: xhci_ring_free: ISOC f1ffb700: enq 0x000000002d31bcc0(0x000000002d31b000) deq 0x000000002d31b000(0x000000002d31b000) segs 2 stream 0 free_trbs 305 bounce 17 cycle 0
<...>-28448 [003] ....   492.025818: xhci_ring_mem_detail: MATTU xhci segment free seg->dma @ 0x000000002d31b000
<...>-28448 [003] ....   492.025823: xhci_ring_mem_detail: MATTU xhci segment free seg->dma @ 0x000000002d31b000
<...>-28448 [003] ....   492.025826: xhci_ring_free: ISOC f1f9b380: enq 0x000000002d31b140(0x000000002d31b000) deq 0x000000002d31b000(0x000000002d31b000) segs 2 stream 0 free_trbs 489 bounce 17 cycle 1
<...>-28448 [003] ....   492.025828: xhci_ring_mem_detail: MATTU xhci segment free seg->dma @ 0x000000002d31b000
<...>-28448 [003] ....   492.025830: xhci_ring_mem_detail: MATTU xhci segment free seg->dma @ 0x000000002d31b000

I'd guess it's still the same cause, maybe trace is not complete?

-Mathias