[Qemu-devel] usb_linux_update_endp_table: Broken pipe

[Qemu-devel] usb_linux_update_endp_table: Broken pipe

[Paul Wise]

[-- Attachment #1: Type: text/plain, Size: 1636 bytes --]

Hi all,

[Please CC me on any replies, I'm not subscribed]

I'm attempting to get a nicer setup for reverse engineering my scanner's
USB protocol. By using libpcap 1.0, wireshark and Linux's usbmon I can
capture USB packets. Add qemu's USB pass-through to that mix and I can
capture the communication between the Windows driver and my scanner.

Unfortunately qemu fails to add my USB scanner to the guest:

$ qemu -M pc -m 256 -smp 1 -name winders -boot c -drive file=winders.img,if=ide,index=0 -usb -usbdevice tablet -usbdevice host:043d:00bb
husb: open device 4.6
husb: config #1 need -1
husb: 3 interfaces claimed for configuration 1
husb: grabbed usb device 4.6
usb_linux_update_endp_table: Broken pipe
Warning: could not add USB device host:043d:00bb

I note these messages on the list about this issue:

http://lists.gnu.org/archive/html/qemu-devel/2008-03/msg00121.html
http://lists.gnu.org/archive/html/qemu-devel/2008-09/msg00277.html
http://lists.gnu.org/archive/html/qemu-devel/2008-09/msg00304.html

I note a workaround in this post and I confirm it works (FSVO):

http://ubuntuforums.org/showpost.php?p=5744172&postcount=5

Any thoughts as to what a proper fix might be? It would be good to not
have to rebuild qemu with the workaround every time I update my system.

The above workaround works in that it allows Windows to detect the
scanner but there seems to be some other issue because the Windows
scanner app can initiate scans but comes up with a generic error dialog
half way through. Unsure if this has anything to do with qemu.

-- 
bye,
pabs

http://bonedaddy.net/pabs3/

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: [Qemu-devel] usb_linux_update_endp_table: Broken pipe

[Pantelis Koukousoulas]

> Unfortunately qemu fails to add my USB scanner to the guest:
>
> $ qemu -M pc -m 256 -smp 1 -name winders -boot c -drive file=winders.img,if=ide,index=0 -usb -usbdevice tablet -usbdevice host:043d:00bb
> husb: open device 4.6
> husb: config #1 need -1
> husb: 3 interfaces claimed for configuration 1
> husb: grabbed usb device 4.6
> usb_linux_update_endp_table: Broken pipe
> Warning: could not add USB device host:043d:00bb

Yes, this is the typical symptom

> I note a workaround in this post and I confirm it works (FSVO):
>
> http://ubuntuforums.org/showpost.php?p=5744172&postcount=5
>
> Any thoughts as to what a proper fix might be? It would be good to not
> have to rebuild qemu with the workaround every time I update my system.

I think it would be worth to just merge this patch. I asked myself a few days
ago, but it seems to be a period where qemu moves quite fast (with the move
to git and all) and there is no time to review usb patches.

Perhaps we should have a usb-specific git tree so that we can test/fix all
usb-related patches that are in limbo right now and occasionally propose
merges.

>
> The above workaround works in that it allows Windows to detect the
> scanner but there seems to be some other issue because the Windows
> scanner app can initiate scans but comes up with a generic error dialog
> half way through. Unsure if this has anything to do with qemu.

There could be several reasons for that. Note that your device is also a bit
on the peculiar side, in that it actually has 3 interfaces (and thus 3 drivers).
It could be also a speed/stall problem but this is unlikely since max's work
has made the usb backend reasonably fast these days.

Cheers,
Pantelis

Re: [Qemu-devel] usb_linux_update_endp_table: Broken pipe

[Paul Wise]

[-- Attachment #1: Type: text/plain, Size: 1477 bytes --]

[Please CC me on any replies, I'm not subscribed]

On Tue, 2009-05-12 at 11:33 +0300, Pantelis Koukousoulas wrote:

> I think it would be worth to just merge this patch. I asked myself a few days
> ago, but it seems to be a period where qemu moves quite fast (with the move
> to git and all) and there is no time to review usb patches.

Seems like it would enable pass-through for a few devices, from Google,
USB-serial devices are the most common thing people run into.

> Perhaps we should have a usb-specific git tree so that we can test/fix all
> usb-related patches that are in limbo right now and occasionally propose
> merges.

Not being a qemu or Linux developer I probably shouldn't comment, but
the -staging tree for dumping stuff to be merged seems to work well for
Linux so maybe qemu could do the same.

> There could be several reasons for that. Note that your device is also a bit
> on the peculiar side, in that it actually has 3 interfaces (and thus 3 drivers).
> It could be also a speed/stall problem but this is unlikely since max's work
> has made the usb backend reasonably fast these days.

Yeah, it is an all-in-one scanner-printer-fax. One of the interfaces is
the printer, one the scanner and not sure about the other one. I was
using qemu 0.10.3 when I had these issues, how recent is max's work? It
actually feels like a speed/stall problem but it really is hard to say.

-- 
bye,
pabs

http://bonedaddy.net/pabs3/

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: [Qemu-devel] usb_linux_update_endp_table: Broken pipe

[Pantelis Koukousoulas]

> Not being a qemu or Linux developer I probably shouldn't comment, but
> the -staging tree for dumping stuff to be merged seems to work well for
> Linux so maybe qemu could do the same.

I 'm not a qemu developer either. I 'm just interested in seeing these problems
fixed eventually.

> Yeah, it is an all-in-one scanner-printer-fax. One of the interfaces is
> the printer, one the scanner and not sure about the other one.

Fwiw, I 'd guess the 3rd interface is a HID one for the scanner button.
lsusb -v can tell you for sure.

> I was using qemu 0.10.3 when I had these issues, how recent is max's work? It
> actually feels like a speed/stall problem but it really is hard to say.

Max's work is in 0.10.x so the problem is not there.

Cheers,
Pantelis