Re: Help with debugging no-callback problem with IrDA USB dongle

Alan Stern escribió:
> On Tue, 1 May 2007 [email protected] wrote:
>
>   
>> After a big delay, I am now trying to write a driver for my USB-to-Irda
>> dongle.
>> This device reports the following in lsusb:
>>
>> Bus 002 Device 002: ID 07c0:4200 Code Mercenaries Hard- und Software GmbH
>> Device Descriptor:
>>   bLength                18
>>   bDescriptorType         1
>>   bcdUSB               1.10
>>     
>
> What version of the Linux kernel are you using?  If you turn on
> CONFIG_USB_DEBUG, what shows up in the dmesg log?
>
>   
All of this was tested with vanilla 2.6.21 from kernel.org. I will try 
enabling CONFIG_USB_DEBUG when I get home.
>> submit the interrupt urbs for either endpoint in the driver code, my callback
>> code never gets called at all - not even to signal an error while
>> submitting the
>> urb.
>>     
>
> That never happens.  The callback routine is invoked when the URB
> completes, and if the submission failed then the URB wasn't accepted so of
> course it can never complete.
>
>   
Hmm... I was under the impression that if usb_submit_urb() returned 0, 
then the callback routine would eventually get called, either because it 
completed, or because an error occurred in submission - such as a timeout.
>> However, when the irda0 network interface is brought down manually
>> (ifconfig irda0 down), I do get calls to my callbacks, with errors
>> corresponding
>> to the device being shut down. I have a suspicion that this is caused by the
>> required urb teardown in kingsun_net_close.
>>
>> Here is a dump of usbmon when my driver takes control of the dongle:
>>
>> Plugin of the dongle:
>>
>> ddf58140 410742248 C Ii:001:01 0 1 = 02
>> ddf58140 410742276 S Ii:001:01 -115 2 <
>> cb8ebb40 410793172 S Ci:001:00 s a3 00 0000 0001 0004 4 <
>> cb8ebb40 410793182 C Ci:001:00 0 4 = 01030100
>>     
>
> So this is a low-speed device, right?
>
>   
Yes.
> ...
>   
>> da870b40 411151178 S Ci:007:00 s 80 06 0302 0409 00ff 255 <
>> da870b40 416154270 C Ci:007:00 -2 24 = 18035500 53004200 20007400 6f002000
>> 49007200 64006100
>>     
>
> This shouldn't happen!  Reception of a short message should never fail 
> with a timeout.  Something must be wrong with your host controller or its 
> driver.  The dmesg log might provide a clue.
>
> ...
>   
>> da87f140 421243731 S Co:007:00 s 00 09 0001 0000 0000 0
>> da87f140 421246338 C Co:007:00 0 0
>>     
>
> Apparently your driver doesn't send anything to the device when it is 
> probed.
>
>   
The stir4200 driver also doesn't do anything to the device, other than 
usb_reset_configuration(). What kind of operation might be necessary to 
do at the probe function?
>> ifconfig irda0 up
>>
>> da87fec0 474705925 S Co:007:00 s 02 01 0000 0081 0000 0
>> da87fec0 474708443 C Co:007:00 0 0
>> da87fec0 474708548 S Co:007:00 s 02 01 0000 0002 0000 0
>> da87fec0 474711422 C Co:007:00 0 0
>>     
>
> Why does your driver bother to do a Clear-Halt?  It should not be needed.
>
>   
Again, because stir4200 does it and I was taking it as one of the models 
for writing my driver. I thought this was a required part of the setup. 
I will try removing it at home and see if it makes any difference.
>> da87fec0 474711537 S Ii:007:01 -115 8 <
>> da870b40 475671204 S Io:007:02 -115 30 = ffffffff ffffffff ffffffff
>> c0ff3f01 c5257160 ffffffff 01000064 d2c1
>>
>> ifconfig irda0 down
>>
>> da870b40 533710546 C Io:007:02 -2 0
>> da87fec0 533711539 C Ii:007:01 -2 0
>>
>> Please notice that usbmon does log the IrDA discovery packet (the one that
>> starts with ffffffff), only my code does not get called for any response
>> on the
>> send.
>>     
>
> It gets logged because your driver submitted the URB to be sent.  The 
> completion handler doesn't get called because the device never accepted 
> the data.
>
> Like Oliver said, you probably need to do some sort of initialization of 
> the device before it will start working.
>
> Alan Stern
>
>   
Maybe, but I could not find any evidence of such an initialization 
sequence in the SnoopyPro dump done under Windows XP (attached). I 
deduced that no initialization sequence was needed. So, I am not looking 
hard enough, or SnoopyPro is buggy / hides information from me.

-- 
perl -e '$x = 2.4; print sprintf("%.0f + %.0f = %.0f\n", $x, $x, $x + $x);'

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/

_______________________________________________
[email protected]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel