On Fri, Nov 25, 2011 at 12:35:06AM +0530, Aruna Balakrishaiah wrote:
>
> Hi Greg,
>
> I had not received this mail, got it from mailing list archive. Please
> find my answers below.
>
> On Wed, Nov 23, 2011 at 10:32:33PM +0530, Aruna Balakrishnaiah wrote:
> > Signed-off-by: Aruna Balakrishnaiah <aruna@...>
> >
> > lsusb throws unusual error message "unable to initialize libusb: -99"
>
> > with exit status of 1 when usb devices are not listed. Patch handles it
> > without printing anything and returns exit status of 0.
>
> >Why would we want to return "good" if there was an error?
>
> Since there are no usb devices in the system , it should exit saying no
> USB devices or something similar to it. Throwing message about failure of
> initialization of libusb may lead to confusion among users and it might not
> be their concern. for example: ls on an empty directory will return with an
> exit status of 0 itself. It is not treated as an error.
Have you ever seen this error before? As package-maintainer in a number
of distros for this package, no one has ever complained about it before.
And I do think that the program should return an error if there are no
USB devices present, that means something is wrong and is easier to
check for, right?
> > +static int check_for_usbdevice(void)
> > +{
> > + struct dirent *d;
> > + int n = 0;
> > + DIR *dir = opendir(sysbususb);
> > +
> > + while ((d = readdir(dir)) != NULL) {
> > + if (++n > 2)
> > + break;
> > + }
> > +
> > + closedir(dir);
> > +
> > + if (n <= 2)
> > + return 1;
> > + else
> > + return 0;
> > +}
>
> >No, we don't want to be doing this, this is why we use libusb, it
> >handles this type of thing for us.
>
> Since libusb_init returns error codes , we would need a seperate function
> to check for usb devices.
Ok, why not change libusb to properly handle this if it's really a big
deal. Wouldn't all of the packages that use also have this same issue?
> > - err = libusb_init(&ctx);
> > + if (check_for_usbdevice())
> > + return 0;
>
> >Again, returning 0 for when there is an error is not a good idea.
>
> >Why is the existing code somehow not working properly for you? When
> >USB devices can't be found, it's good to return an error, right?
>
> As mentioned issuing ls on an empty directory cant be treated as an error
> right?User might not be concerned about libusb library. Since the command
> is not failing its just there are no USB devices to list.
But are you sure that your check_for_usbdevice() will actually work
properly for all distros and kernel versions (hint, I don't think
so...)
libusb has this logic already in it, I do not want to duplicate it
again, that's why we use the library.
And yes, I do consider the fact that if you run lsusb on a system with
no usb devices, it should return an error, just like I also think lspci
does the same thing, right?
thanks,
greg k-h
--
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
