RE: usb_submit_urb errors

Hi,

On Sun, 4 Apr 2004, Christopher Piggott HOME wrote:

> This is really disturbing.  I went through and determined that the -ENOMEM
> was not coming from net/core/dev.c as I thought it was.  So, I went through
> at76c503.c and looked for all possible places that there could be an -ENOMEM
> returned and I added this:
>
> 	printk("chrisp: kmalloc failed in " __FILE__ ", line %d\n", __LINE);
>
> recompiled it, and it started WORKING.  I never saw any of the debug
> messages fire.  When I say it works, I mean it completely works - it
> associates with the AP and dhcp gets an address, WEP is working, it's full
> connectable.  It's even happy talking to my dhcp server.


Hm, in at76c503.c (both version 0.11 and the current cvs - you haven't
told which version you saw that error with) I see only one line (3092 -
v0.11, 3614 - cvs) producing the error you've posted before:
"rx, usb_submit_urb failed ..."
where in both cases the ret val comes directly from the USB subsystem.

Have you tried to run the original driver (0.11 / cvs) again? Maybe
something inside your kernel (modules) changed meanwhile.

/Jörg