Re: UAS gadget driver & UAS host driver [was: Re: [RFC/PATCH 0/4] UASP device driver]

[Sarah Sharp <sarah.a.sharp@linux.intel.com>]

On Mon, Feb 21, 2011 at 05:35:47PM +0200, Tanya Brokhman wrote:
> Hi Sarah
> 
> > That particular error is from the host side.  It means the xHCI driver
> > tried to move the ring dequeue pointer past the last part of your
> > command (probably the status phase), and the host refused to do so,
> > because that was the "active" stream ID.  Can you please send me the
> > full dmesg for the stalled request with CONFIG_USB_XHCI_HCD_DEBUGGING
> > turned on?
> 
> Np. I've already overcame this issue but I'll roll back and send you the
> dmesg output. When I tried enabling this flag the dmesg was swamped with
> xhci messages and the messages I was looking for weren't there already. 
> I read that in order to increase the dmesg buffer size one needs to update
> the #define of LOG_BUF_LEN in linux/kernel/printk.c. I haven't tried that
> yet. Is this correct?

There will be a lot of messages, but you can capture them by running
this command line:

tail -f /var/log/kern.log | tee logfile.txt

Then send me logfile.txt.

I've been meaning to cut back on the less-than-useful debugging in the
xHCI driver and add an event trace infrastructure, but other features
have taken priority.

> Do you need the LECroy recording as well?

No, that's not necessary.

> > I think it's worth rolling back your firmware for this because I really
> > need to test this corner case.  I think it may be an unhandled corner
> > case that I've been warning Intel's xHCI spec architect about.  There
> > seems to be no way to modify the active stream's rings for cases like
> > these where the transfer stalls.
> > 
> > Which host controller are you running under?  If it's an NEC host
> > controller, please send me the line that looks something like this:
> > 
> > NEC firmware version xx.xx
> > 
> 
> We're using NEC but I'm not sure where to retrieve the version from. The
> line you mentioned isn't present in the dmesg output and I found nothing
> similar written on the chip either.

You'll see that line in the dmesg once you enable
CONFIG_USB_XHCI_HCD_DEBUGGING.

Sarah Sharp