Re: [linux-usb-devel] [PATCH] ipaq.c bugfixes

[Frank Gevaerts <frank.gevaerts@fks.be>]

On Fri, Jun 02, 2006 at 09:59:35AM -0300, Luiz Fernando N. Capitulino wrote:
> On Thu, 1 Jun 2006 21:16:54 +0200
> Frank Gevaerts <frank.gevaerts@fks.be> wrote:
> 
> | When I changed te ipaq_open code to only submit the read urb after the
> | control message succeeds, these disappear.
> | 
> | Whenever the usb_control_msg returns with ETIMEDOUT (-110), in both code
> | variants, when the device is disconnected we get:
> 
>  Did you mean that it happens even if you send the read urb after the
> control message?

Yes.

> | This seems to be 100% reproducible. I noticed that serial_open (in
> | usb-serial.c) sets port->tty = tty and tty->driver_data = port, but
> | doesn't set them back to NULL if the type->open() fails. Is that correct
> | ?
> 
>  I don't think it would cause the problem you're getting. 'port' is
> valid even if the driver's open() function fails.

We are currently trying a version where these are both set to NULL on
failure, and it does seem to help.

> | Also, we have discovered that the slow response of the ipaq on connect
> | is actually largely caused by the control message retries, so we have
> | added a sleep at the start of ipaq_open.
> | 
> | The current patch we are testing (not yet ready for inclusion, since
> | it doesn't work correctly yet and it produces some output that is not
> | useful in general) is :
> 
>  Well, I'm afraid that every new patch leads to a new problem..
> Maybe it's something else..

I'm not sure it's a new problem every time. Some of the earlier bugs
could have been the same as this one (the stack trace was the same)

Frank

> -- 
> Luiz Fernando N. Capitulino

-- 
Frank Gevaerts                                 frank.gevaerts@fks.be
fks bvba - Formal and Knowledge Systems        http://www.fks.be/
Stationsstraat 108                             Tel:  ++32-(0)11-21 49 11
B-3570 ALKEN                                   Fax:  ++32-(0)11-22 04 19