On Wed, Mar 09, 2011 at 07:44:01AM -0800, Greg KH wrote:
> On Wed, Mar 09, 2011 at 10:33:01AM -0500, Elly Jones wrote:
> > On Tue, Mar 08, 2011 at 03:10:46PM -0800, Greg KH wrote:
> > > On Tue, Mar 08, 2011 at 05:53:08PM -0500, Elly Jones wrote:
> > > > As I understand it, gregkh rewrote Qualcomm's GPLed QCSerial2k driver to produce
> > > > qcserial.c. Since Qualcomm just released Gobi 3000 drivers under the GPL, I've
> > > > been working on hacking qcserial.c to support Gobi 3000 as well as 2000.
> > > > However, I don't understand the reasoning for one of the changes in the
> > > > rewrite.
> > >
> > > <snip>
> > >
> > > > I don't understand why we need the extra test against bNumEndpoints and
> > > > usb_endpoint_is_bulk_{in,out} here. Can anyone shed any light?
> > >
> > > Did you look at the git history for this driver? It will show you
> > > exactly why this change was made, and who made it.
> >
> > I looked, but was not enlightened. The driver was added by you on Feb 17 2009,
> > as part of commit a78b42824dd7c2b40d72fb01f1b1842f7e845f3a; the message for that
> > commit says "Cleaned up the probe logic to make a bit more sense". Since then,
> > that code has been refactored (by Anssi Hannula). The code I'm asking about was
> > present in the very first commit made to this driver.
>
> Ah, sorry, I thought it was added by Anssi's commit, you are correct.
>
> > > After looking at that, if you have further questions, please let us
> > > know.
> >
> > I still do not understand why the test for usb_endpoint_is_bulk_{in,out} is
> > necessary, when it wasn't in the original qualcomm driver (and isn't in their
> > new gobi3k serial driver). I feel like I must have missed something obvious, but
> > I cannot see what it is.
>
> I think I had some devices where this check was required to be made,
> otherwise it was picking up the wrong interface. It was needed to get
> something to work properly, sorry if I can't recall the exact model it
> was done for, that was a long time ago that I rewrote it.
S'ok.
> Does this code somehow cause you problems with your device? The check
> should be safe for you, right?
Yeah, it's perfectly safe; I was just wondering if it was important, since I'm
about to send up a patch to add Gobi 3000 support to qcserial.c and the qualcomm
gobi 3000 driver doesn't have that check either.
> thanks,
>
> greg k-h
-- Elly
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
