On Sun, 22 Apr 2007, Julien BLACHE wrote: > Brad Sawatzky <[EMAIL PROTECTED]> wrote:
[ ... ] > > Is it possible for the autoprobed scanner device to be indeterminate? The > > I'm not sure what you mean here, could you elaborate ? See below. > > CX6600 is a multifunction printer/scanner unit with built-in card reader so > > many different devices are created when it's plugged in -- could there be > > some weird race condition? Arrgh. > > There's only one node created under /dev/bus/usb unfortunately, that's > a real problem with multifunction devices and the current permissions > scheme - it's hard to share the printer between CUPS and SANE. This is what was in the back of my mind. I'm not sure how libsane identifies the correct scanner device in /dev/. If the scanner can be accessed through more than one device ( ie. /dev/bus/usb/00X/00Y, and /dev/usbdevX.Y_epZZ) then the device libsane picks would depend on which device it tries first and was able to open. If CUPS grabbed the /dev/bus/usb/ device first, would libsane keep searching and eventually try the /dev/usbdevX.Y* devices? Of course, this assumes that a) libsane walks /dev looking for scanner devices, and b) the usbdevX.Y* devices can actually be used to control the scanner despite having different major/minor numbers than the nodes in /dev/bus/usb/... > Exposing the endpoints individually as it's done now could help solve > that, I'd need to check why this was introduced and what the plans are > on the libusb side. -- Brad -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
