[RFC] [PATCH] usbatm.[ch]: multiple changes

[matthieu castet]

Hi Roman,

Roman Kagan wrote:
>>Also the bulk mode don't work well with eagle-usb, there are lot's of 
>>errors like usb-atm. That's why iso is the default.
> 
> 
> Hmm, neither the forum nor the mailing list archive at eagle-usb.org
> seems to mention anybody having babble problem in bulk mode.
That easy, try to find people that have bulk mode working ;)
The mode is choosen at compilation time, and iso is the default one, do 
you expect users to recompile their driver and if so to change the 
default configuration ?
For the people who have choosen the bulk mode, how many have a high rate 
connection ?

Also I found (don't remember the link) a people who used bulk mode and 
which have lot's of ATM corruption.
> 
> 
>>>there's any reference to the kind of errors you see in their forum or
>>>docs, which makes me suspect a hardware problem.
>>
>>No this was confirmed by a developper :
>>https://mail.gna.org/public/eagleusb-dev/2005-01/msg00083.html
> 
> 
> Same concern here: the message you quote is the only one mentioning
> -EILSEQ, and that developer only appears to remember someone else
> reporting it...
Of course do you expect user to edit the driver and try to catch all the 
errors ?
If the error was masked, I think it is normal that nobody discover the 
error...


> 
> Please understand that putting a workaround like that is usually frowned
> upon, because it may hide a real problem.  So I think we need to try to
> accumulate some statistics on the matter, and to isolate the source of
> the problem as narrow as possible.  In particular, if that turns out to
> be a firmware bug, it can be reported to ADI and hopefully fixed by
> them.
> 
> Maybe you can get people at eagle-usb.org who have a high enough
> connection speed to try to use your driver in both bulk and iso mode
> without the EILSEQ workaround?  (Naturally I also suggest trying with my
> latest patch if you feel like it :)
> 
So I believe this 2 problem are linked to the modem and they were masked 
by the adi driver. I agree that they shouldn't happen. I will try to do 
some test and ask people to do some. I will also try the windows driver 
and see if it produce errors in bulk mode.


Cheers

Matthieu