On Wed, Nov 02, 2011 at 02:09:20AM +0000, Ken Moffat wrote: > On Tue, Nov 01, 2011 at 11:46:47PM +0000, Ken Moffat wrote: > > On Tue, Nov 01, 2011 at 10:44:32PM +0000, Andrew Benton wrote: > > > > > > I've never had this problem, so I can't think what could be the cause. > > > What I would suggest is that you compile your kernel with > > > > > > CONFIG_DEVTMPFS=y > > > CONFIG_DEVTMPFS_MOUNT=y > > > > Thanks, that is working, both for 3.0.4 and for 2.6.32.43. > Interestingly, udev claims to start on both kernels. I'll see if I > agree after I've tried to use an external usb drive for backups! > Surprisingly, on 2.6.32.43 udev *is* working enough to create nodes when I plug in an external disk, and remove them when I unplug it.
Meanwhile, I think I've found the root of my problem - I said this would be embarrassing :-( The server system was *well* past its best-before date. I'd hoped to upgrade the system about a year ago, but at that time I was on the previous machine with PCI cards for the two mirror drives. Those were using ide=reverse on the commandline to relegate them after the system disk (i.e. sdb, sdc). That worked on 2.6.24.7, but I never managed to boot 2.6.32 which had lost that option. So, a year ago I was aware that the system used old udev, but in the move to a new machine I had forgotten - I blame the other changes I had to make at that time for making me forget this: move to grub2 (lilo refuses to accept root= for a partition it cannot see when it is updated, so it wouldn't let me use root=/dev/sdc7 or whatever to setup the new kernel parms from the old ones), change the framebuffer, and fiddle with grub to use the 'graphics' display for the framebuffer (not needed on my desktops!). <sigh!/> The config entries I think are causing the problem are CONFIG_SYSFS_DEPRECATED=y CONFIG_SYSFS_DEPRECATED_V2=y Those are certainly not wanted for a recent system. ISTR someone had a similar problem in the distant past. OTOH, the system is now working ok, so I'm in no rush to rebuild the kernel and reboot, and anyway the results won't be conclusive if I leave in the DEVTMPFS options. So, tentatively, the cause is now understood. ĸen -- das eine Mal als Tragödie, das andere Mal als Farce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
