Re: Fedora fun 517438: Bulk crowds out ISO?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, 2009-12-04 at 16:02 -0500, Alan Stern wrote:
> On Fri, 4 Dec 2009, Pete Zaitcev wrote:
> 
> > On Fri, 4 Dec 2009 11:59:25 -0800, Greg KH <greg@xxxxxxxxx> wrote:
> > > On Fri, Dec 04, 2009 at 12:46:57PM -0700, Pete Zaitcev wrote:
> > 
> > > > A user found that copying to a USB disk crowds out ISO transfers
> > > > to audio. How is this even possible?

FWIW, "a user" being me.

> > > > https://bugzilla.redhat.com/show_bug.cgi?id=517438
> > > > 
> > > > > Whilst starting a Torrent with Transmission on a VFAT USB hard drive, audio
> > > > > (through a USB 5.1 sound card) started skipping like nobody's business.
> > > > 
> > > > > ALSA sound/usb/usbaudio.c:882: cannot submit datapipe for urb 0, error -28: not enough bandwidth
> > > > > ALSA sound/usb/usbaudio.c:882: cannot submit datapipe for urb 0, error -28: not enough bandwidth
> 
> I would guess that the problem is the extra computational load caused 
> by running Torrent.

I'd say uber-bogus. There's no computation yet, it's Transmission
creating the "how many gigs" files for the torrent. If you ran it on
ext4 (as I have the exact same drive setup for a couple of days now),
there wouldn't be any traffic to speak of, because ext4 can do sparse
reservation for files.

dd'ing /dev/null to the USB disk would probably have the same effect.

The audio device is a "cheap as chips" Aureon MKII 5.1 sound card, with
all 6 channels filled up with audio through PulseAudio.

>   It could prevent the usbaudio driver from running 
> in a timely fashion (because of the added interrupt latency), thereby 
> losing its bandwidth allocation.  A usbmon trace would tell whether 
> this is really what's happening.
<snip>
> > The whole reason for ISO to exist is to receive the guaranteed bandwidth,
> > isn't it (see clause 3.3 of 2.0 spec)?
> > 
> > I understand that until the ISO URB is submitted, there's no way for the
> > HCD to know that it should allocate the bandwidth. What is our solution,
> > then? Should we create some sort of "channels" that allocate the bandwidth
> > (what spec calls "prenegotiated")?
> 
> That is more or less our plan (preallocate periodic bandwidth as soon
> as a config or an altsetting is installed).  It's going to take a while
> to implement, however.

/me crosses fingers!

--
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

[Index of Archives]     [Linux Media]     [Linux Input]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Old Linux USB Devel Archive]

  Powered by Linux