On 11/04/2010, at 12:36 PM, Rob Landley wrote: > On Tuesday 06 April 2010 16:59:22 Steve Bennett wrote: >> On 02/04/2010, at 2:50 PM, Denys Vlasenko wrote: >>> On Friday 02 April 2010 01:14, Steve Bennett wrote: >>>>> This treats creation actions as deletions too - it will >>>>> delete the device node if it exist. This is confusing. >>>> >>>> Possibly. What does delete even mean when the "create" action >>>> is to not create the device node? >>> >>> Delete will (try to) delete the node. >>> Users which created the node by some other mechanism >>> won't like to find their nodes nuked by mdev. > > If you tell mdev to manage the /dev directory, and the kernel says that the > node went away, mdev will delete it. If you're saying the kernel will say a > node went away when it didn't actually go away, that would be a kernel bug. > > You don't seem to understand how mdev is for. Maybe you want to be using > something other than mdev, rather than complicating mdev? And you haven't > given a real-world example of where this is actually causing you a problem... > > I'm confused...
ls -l /dev/mtd* crw-rw---- 1 root root 90, 0 Apr 5 13:29 /dev/mtd0 crw-rw---- 1 root root 90, 2 Apr 5 13:29 /dev/mtd1 ... crw-rw---- 1 root root 90, 1 Apr 5 13:29 /dev/mtdro0 crw-rw---- 1 root root 90, 3 Apr 5 13:29 /dev/mtdro1 ... But I don't need those those read-only device nodes. I don't want them to be created. So in mdev.conf I add: mtd[0-9]ro 0:0 660 ! Any they are gone. That's it. Cheers, Steve > > Rob > -- > Latency is more important than throughput. It's that simple. - Linus Torvalds > -- µWeb: Embedded Web Framework - http://uweb.workware.net.au/ WorkWare Systems Pty Ltd W: www.workware.net.au P: 0434 921 300 E: [email protected] F: 07 3102 9221 _______________________________________________ busybox mailing list [email protected] http://lists.busybox.net/mailman/listinfo/busybox
