On Thu, 10 Sep 2009, Gabriel Rossetti wrote: > Hello everyone, > > From my thread on the libusb mailing list > (http://www.nabble.com/Does-libusb-1.0-get-it%27s-devices-stolen-by-the-OS%27s-native-driver--td25367199.html), > I came to understand that the linux version of the HID driver is unable > to send data to the device. This is of course completely untrue. If the HID driver were unable to send data to devices, how could it turn the NumLock and CapsLock LEDs in USB keyboards on and off? > From what I read on the > HID spec (http://www.usb.org/developers/devclass_docs/HID1_11.pdf) it > says that a report can define input and output (section 6.2.2.5, page > 29). Is this correct, that the linux HID driver only partially > implements the HID spec and that the driver won't let me send arbitrary > (non-keyboard led on/off) data to a HID device? It is true that devices can have both input and output reports. It is not true that the Linux HID driver fails to implement output reports or that it prevents you from sending data to HID devices. > If so, from what I > understand, I have two choices : > > 1) write a kernel driver for my device (which is why I made it to be HID > compliant in the first place, to not have to write a driver for linux, > mac & windows) > 2) use a userland lib like libusb/libhid and have to deal with the > kernel's HID driver claiming the device > > Are there any other choices? Did I miss something? You should ask the HID developers rather than the USB developers. See the entries for HID CORE LAYER and USB HID/HIDBP DRIVERS in the kernel source file MAINTAINERS. Alan Stern -- 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
