Messages in this thread | | | Date | Sun, 9 Aug 1998 12:02:58 +1000 | From | Richard Gooch <> | Subject | Re: DEVFSv50 and /dev/fb? (or /dev/fb/? ???) |
| |
Alan Cox writes: > > > Dev_fs breaks completely under such circumstances you mean. Firewire and USB > > > need potentially quite a lot of nonvolatile state storing. For that they need > > > a daemon if they have a daemon they don't need devfs > > > > What state are you referring to? The permissions? And how many device > > nodes are you talking about? > > Well for USB you have to remember the state of every device plugged > into the system on a permanent basis. Not only is this needed for > things like hot plugging when a device loses its settings as you > move it between ports but also to ensure that when you plug it back > in it keeps its old configuration from last use on this machine. > > The amount of state per device isnt really bounded, although I would > expect it to be small. The point I was making is you do this by > having a usb daemon sitting around getting "someone plugged this USB > device in, its serial id is blah, its address is xyz and its a > camera" and taking action in user space accordingly > > Policy in user space > > The pcmcia services also use this technique. > > So it doesnt matter if you are using devfs or not - the daemon will > be making the device files in either and firing things up as > needed. It has more to fire up than just devices though - insert a > mouse - perhaps that loads gpm ?
OK, this sounds like it will be handled nicely by the devfsd I mentioned earlier. When a driver registers a new device entry, the daemon is notified. It can then do whatever it requires.
Regards,
Richard....
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html
| |