From Bugzilla Helper: User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a1) Gecko/20060210 Firefox/1.6a1 Description of problem: This change in initscripts-8.26: - don't mount /sys and /proc in rc.sysinit - the initrd already does (<pjones>) causes breakage on non RH kernels. Reverting to 8.25 and the problem goes away. Here's the output of a failed boot with 8.26 using a vanilla kernel: VFS: Mounted root (reiserfs filesystem) readonly. Freeing unused kernel memory: 196k freed Write protecting the kernel read-only data: 358k INIT: version 2.86 booting awk: cmd. line:1: fatal: cannot open file `/proc/mounts' for reading (No such file or directory) Welcome to Fedora Core Press 'I' to enter interactive startup. Setting clock (utc): Sun Feb 12 00:40:46 NZDT 2006 [ OK ] cat: /proc/cmdline: No such file or directory Setting hostname tornado.reub.net: [ OK ] mdadm: failed to create /dev/md1 mdadm: failed to create /dev/md4 ERROR: finding sysfs mount point ERROR: No arguments allowed with -i ERROR: finding sysfs mount point ERROR: No arguments allowed with -i ERROR: finding sysfs mount point ERROR: No arguments allowed with -i ERROR: finding sysfs mount point ERROR: No arguments allowed with -i Checking filesystems Checking all file systems. [/sbin/fsck.ext3 (1) -- /boot] fsck.ext3 -a /dev/sda1 /dev/sda1: clean, 36/6024 files, 8360/24064 blocks [ OK ] fgrep: /proc/mounts: No such file or directory awk: cmd. line:1: fatal: cannot open file `/proc/mounts' for reading (No such file or directory) Remounting root filesystem in read-write mode: [ OK ] Mounting local filesystems: mount: special device /dev/md1 does not exist mount: special device /dev/md4 does not exist mount: special device /dev/sdc1 does not exist [FAILED] touch: setting times of `/var/log/wtmp': Function not implemented Enabling swap space: [ OK ] Version-Release number of selected component (if applicable): initscripts-8.26-1 How reproducible: Always Steps to Reproduce: 1. Install initscripts-8.26 2. Boot a non-FC non-initrd kernel 3. Reboot Actual Results: Failed to boot Additional info: I don't think that blindly assuming that they are always mounted via initrd is a good idea. In my case I often build kernels which do not require any modules be loaded until well after the box has started booting. A good fix IMHO would be to test to see if /sysfs and /proc had already been mouted in the initrd. If they had then continue on, else if not, mount them. The impact on testing this before assuming would be small. If this is not going to change then it should at least be documented in the release notes for FC5.
Will put something back in 8.29-1, *and* document in the relnotes that booting without initrd is deprecated, and may be removed in a later release.