From: Robin Gareus <robin(a)gareus.org>
Date: Sun, 13 May 2012 20:15:49 +0200
On 05/13/2012 07:45 PM, plutek wrote:
From: Robin Gareus <robin(a)gareus.org>
Date: Sun, 13 May 2012 08:57:26 +0200
On 05/13/2012 03:13 AM, plutek wrote:
From: Karsten Wiese
<fzuuzf(a)googlemail.com>
Date: Sat, 12 May 2012 23:18:22 +0200
> 2012/5/12 Clemens Ladisch <clemens(a)ladisch.de>de>:
>> plutek wrote:
>>> i've been running AVLinux5.0.3 on a ThinkPad T61p for a while now....
>>> with an ART USB Dual Tube Pre, i can run Jack at 64 frames, 2 periods
>>>
>>> now i've got a new ThinkPad T520, and if i start jack on USB at 64*2
>>> i get hundreds of xruns in seconds.
>>
>> Both machines run (I assume) the same software? Then the problem is
>> with the hardware, or its drivers.
>>
>> Please show the output of lspci and the contents of /proc/interrupts.
>>
>
> Thinkwiki says its got a QM67. That does USB 1.1 via the "Rate
> matching hub" and EHCI.
> I guess linux's hub driver is the culprit.
hmmmm.... shame.... lenovo specs said, of course, "USB 2.0".
mrph.
any way around this currently?
yes,
http://lists.linuxaudio.org/pipermail/linux-audio-dev/2010-October/029511.h…
after i do:
# cd /sys/bus/pci/drivers/ehci_hcd/
# echo -n 0000:00:1d.0 > unbind
...the USB audio device no longer shows up in qjackctl.
did i misunderstand?
No, not that I can tell. "0000:00:1d.0" is the USB hub/controller, right?
yes.
The idea is to use the uhci_hcd (USB1) driver instead
of ehci_hcd
(USB2). -- Maybe the module is not loaded?
`lsmod | grep uhci_hcd`
yeah... nothing.
If they're modules: `rmmod ehci_hcd; modprobe
uhci_hcd`; that'll force
all USB ports into 1.X mode; may be worth a try before playing with
dedicated per port bindings.
well, when i do that, this happens:
rmmod ehci_hcd
(lsmod shows neither ehci nor uhci)
modprobe uhci_hcd
(lsmod shows BOTH ehci and uhci)
rmmod uhci_hcd
(lsmod shows only ehci)
(USB devices don't appear in qjackctl)
i.e., it seems like USB devices show ONLY when ehci is loaded.
Anyway, there might be something else going on: as
both Karsten and
Clemens noted, these "Intel Corporation Cougar Point [..] controllers
use a different driver" and "There have been fixes recently. Try
updating the kernel."
yeah, i'm starting to play with more recent kernels.... so far nothing does any
better! :-(
cheers!
.pltk.