From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Brownell Date: Fri, 09 Feb 2001 17:59:08 +0000 Subject: Re: definition of terms Message-Id: List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: linux-hotplug@vger.kernel.org There's "function" too. "Physical device" is something I can pick up, but when I plug it in it may do several things. PCI cards can integrate several functions (up to 8?) and USB devices can have several "interfaces" (up to 255) or be parts of compound devices (most software will ignore that last). Examples (real): - Device: Cardbus * Functions 1,2: USB OHCI host controller * Function 3: USB EHCI host controller - Device: USB * Function 1: audio speaker * Function 2: three buttons - Compound USB device (integrates a hub): * Device 1: USB Hub (single function) * Device 2: USB Keyboard (single function) Right now, /proc/bus/usb names devices, while /proc/bus/pci names functions. In an ideal world, we'd be consistent in those two, and /proc/bus/pcmcia would also be consistent. (Also in an ideal world, /proc/bus/usb would expose stable names based on topology, not unstable ones based on current addresses.) We report hotplug events not in terms of "physical device" as you said, or (device) function as I might prefer ... but in the granularity used by those /proc/bus names. Even in kernels not configured with usbdevfs, and even though multifunction USB devices should arguably get multiple hotplug events. (It'd work better that way on systems that don't configure usbdevfs, too.) Your definition needs to be less precise, if it's to be accurate. More general ... unless we change the code, which should be a separate thread. - Dave _______________________________________________ Linux-hotplug-devel mailing list http://linux-hotplug.sourceforge.net Linux-hotplug-devel@lists.sourceforge.net http://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel