Red Hat Bugzilla – Bug 34356
Lilo error: Hole found in map file
Last modified: 2007-04-18 12:32:30 EDT
I'm getting a "Hole found in map file (alloc_pages)" from lilo. The README for lilo indicates that this needs to be reported. This is Lilo version 21. Details: I have a drive 255/63/526. It has three partitions. hda1: 1 12 Ms-dos hda2: 13 372 Linux with ReiserFS hda3: 373 526 Linux with ext2 FS We have the reiser as the default boot. This system will run headless, with the power switch as the "turn-off", so we need a self repairing file system. The Ext2 system is there so that incase something does happen, a monitor/keyboard can be hooked up for diagnostics (it is not mounted under normal situations). We have the dos system marked as bootable. Changing this does not affect the problem. Diagnostics: 1. If Lilo is run with NO compact option, we get the error message "Hole found in map file (alloc_pages)" as soon as one of the linux partitions is added (or tried to). 2. If compact is specified, no error message is reported, but the system will not boot properly: A: At boot up, no message is displayed. B: At boot up, the loader prints LILboot: Loading linuxUncompressing Linux... OK, booting the kernel. all on one line. Note capitalization -- Loading _l_inuxUncompressing _L_inux. The lowercase "l"inux is from the default boot -- changing this in lilo.conf does change the output. The message file specified in lilo.conf is NOT displayed. No pause occurs to enter a new kernel. The only reason the root is properly found is that we used rdev to specify the default root on hda2. (We got kernel panics before that). This is going out into the field today or tomorrow. We need this solved. Lilo.conf: boot=/dev/hda map=/boot/map install=/boot/boot.b prompt timeout=50 linear default=reiser # default=linux message=/boot/message compact other = /dev/hda1 label = dos table = /dev/hda image=/boot/vmlinuz-2.2.14-5.0 label=linux read-only root=/dev/hda3 image=/boot/vmlinuz-2.2.14-5.0 root=/dev/hda2 read-only label=reiser Partition: Disk /dev/hda: 255 heads, 63 sectors, 526 cylinders Units = cylinders of 16065 * 512 bytes Device Boot Start End Blocks Id System /dev/hda1 * 1 12 96358+ 6 FAT16 /dev/hda2 13 372 2891700 83 Linux /dev/hda3 373 526 1237005 83 Linux Command (m for help): q
Additional information: Lilo has no problems if it uses the kernels/ message files on the ext2 partition, when booted on the ext2 root. It did have a problem when the reiser partition was root, using the reisfer files. In fact, testing cross setups (booting on one, using files from the other) shows that the problem only occurs when the map file is on the reiser file system. This is confirmed -- the /boot/map file can be on a dos partition, or a ext2 partition, just not a reiser partition. This gives us a work-around -- it no longer has to be considered high priority.
The work-around was premature. /sbin/lilo will not complain if the map file is on dos or ext2. However, the bootup behavior is the same. Putting the entire /boot directory onto the dos partition... Yes. If all of /boot is on the dos partition, it works fine. This gives us a usable work-around. (Still would prefer to have it all on reiser :-).
Does this work better with newer releases? note that we don't really support reiserfs and its possible that tail-merging there may cause problems, so it's really only interesting if you hvae problems on ext2 or ext3
Closing due to inactivity. If you have further information, please reopen this bug.
Changed to 'CLOSED' state since 'RESOLVED' has been deprecated.