Hi Stefan,

> >> https://lists.berlios.de/pipermail/linux-uvc-devel/2007-March/001476.htm
> >>l
> >>
> >> .. the reset functionality is a part of a patch that is applied against
> >> the UVC Module that get compiled and included in the kernel.  From
> >> my reading of the code, it automatically resets the USB device (without
> >> user intervention).  Thus no command is required.
> >>
> >> But, to my understanding, this patch is not included by default in UVC.
> >
> > That's right.  The patch helped a few people, but failed to fix the bug
> > for others.  As this is a pretty intrusive workaround, I'm not really
> > keen on including it by default if it can't fix the problem for most
> > users.
>
> Hmm... I'm not sure what you mean by "intrusive" here (I don't understand
> the UVC code, really, but when I applied the patch by hand, it seemed
> reasonably conceptually clean and unintrusive in that it just hooks into
> a few places that are not performance sensitive;  Its behavior is obviously
> an ugly hack).
>
> I wouldn't want to have the patch included as-is, and especially the hack
> shouldn't be enabled by default, but I'm wondering: what changes would you
> want to see before such a patch could be considered as acceptable.
> I'd obviously start with adding a module argument named something like
> "use_reset_hack", but what else?
> Would you want the whole hack to be carefully wrapped in #ifdefs?

Before including such hacks in the driver, I want to make sure they are 
actually helpful for more than a few users. If the patch can solve timeout 
issues for more than 50% of the user base (which does not mean preventing 
timeouts once every other time, but really preventing all timeouts to happen 
for most users), I would considering including it. A friend of mine has a 
laptop which suffers a lot from the hardware bug. Until now no patch has been 
able to help him.

The other issue is that such a fix would better be made in the USB core, not 
in the UVC driver. Basically, I'd like a "this works because..." patch better 
than a "this seems to work but I don't know why" patch.

Best regards,

Laurent Pinchart
_______________________________________________
Linux-uvc-devel mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/linux-uvc-devel

Reply via email to