[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-usb-devel
Subject:    RE: [linux-usb-devel] Timeouts on setting up EZ-USB cards
From:       "Dunlap, Randy" <randy.dunlap () intel ! com>
Date:       2000-06-29 16:11:07
[Download RAW message or body]

Hello-

> Hi all,
> 
> I am working on a system which involves connecting large numbers of
> EZ-USB cards with several layers of hubs. However, the kernel seems to
> run into difficulties with even 7 connected at once - namely that some
> cards do not respond to a "set address" request, and hence become
> unusable.
> 
> I am connecting 7 EZ-USB development boards to a 7-port hub, which is
> connected to the root hub. If they are all attached to the 
> hub, which is
> then plugged into the computer, usually only 5 are successfully
> initialised. If they are connected one by one, 6 sometimes 
> manage to be initialised - and very occasionally all 7.
> 
> This problem occurs with both OHCI and UHCI, and both with kernels
> 2.4.0-test2 and 2.4.0-test1-ac22. It also occurs when I connect a
> generic serial converter instead of one of the cards (I don't have any
> other USB devices to test). It makes no difference whether the devices
> are plugged in before or after powering on the computer.

Did you test it with both UHCI host controller drivers?
If not, would you try the other UHCI HCD (of usb-uhci and uhci,
or UHCI and UHCI Alternate)?

Other than -ETIMEDOUT (-110) being reported, there is also
one occurrence of -EPROTO (-71) reported.
Both usb-uhci and uhci report this error for BITSTUFF errors,
and usb-uhci and returns this error for "unknown error" code.
If it's a BITSTUFF error from one of the devices, that's bad.
But maybe I shouldn't be focusing on this one error so much.

For UHCI, we need to see more host controller descriptor info
(logs), so you'll need to enable/generate more debug output.
In usb-uhci.c, find /uhci_show/ and on the lines that begin
with "//", remove the comment "//", rebuild it, and test again.
With uhci.c, it looks like most descriptor debugging in it
is already enabled, so I'm guessing that you didn't test with it.

Or maybe some of the people who deal with EZ-USB devices
know of some problems and/or workarounds with them.....


> I am using usbdevfs; I noticed that the devices file is truncated when
> this number of devices is used (due to the EZ-USB boards 
> having lots of
> endpoints). I tried changing the amount of memory allocated for the
> /proc/bus/usb/devices file (in devices.c), and stopped the file being
> truncated; however, this did not affect the problem. Neither did
> changing the TIMEOUT value used in usb_set_address().

Yes, I had noticed that also and made a patch for it, but you are
right, it wouldn't affect the problem.


> Any ideas? Here is some (maybe) helpful output:
> 
> The correct output when a card is plugged:
> 
> Jun 29 16:10:16 localhost kernel: usb.c: USB new device connect,
> assigned device number 7 
> Jun 29 16:10:16 localhost kernel: usb.c: This device is not recognized
> by any installed USB driver. 

[descriptor dumps cut]

> The output after about 5 cards have been plugged:
> 
> Jun 29 16:10:16 localhost kernel: usb.c: USB new device connect,
> assigned device number 8
> Jun 29 16:10:16 localhost kernel: usb.c: USB device not accepting new
> address (error=-110)
> Jun 29 16:10:16 localhost kernel: usb.c: USB disconnect on device -1
> Jun 29 16:10:17 localhost kernel: usb.c: USB new device connect,
> assigned device number 8 
> Jun 29 16:10:17 localhost kernel: usb.c: USB device not accepting new
> address (error=-71) **********************************************
> Jun 29 16:10:17 localhost kernel: usb.c: USB new device connect,
> assigned device number -1 
> Jun 29 16:10:17 localhost kernel: usb.c: USB device not responding,
> giving up (error=-110) 
> Jun 29 16:10:17 localhost kernel: usb.c: USB disconnect on device -1 
> 
> Occasionally other errors are produced, eg.:
> Jun 29 16:10:16 localhost kernel: usb.c: failed to set default
> configuration
> 
> Thanks,
> -- 
> Ashley Reid-Montanaro

~Randy


_______________________________________________
linux-usb-devel mailing list
linux-usb-devel@lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/linux-usb-devel

[prev in list] [next in list] [prev in thread] [next in thread]