Severity of this is minor for an experienced person. It's a true blocker for a neophyte. On initial boot of the cdrom, the system hangs when it encounters certian perexisting partitions. This is uncommon. I've had it happen with five out of 30 or so drives. Example output: >> Activating udev >> Making tmpfs for /newroot >> Attempting to mount CD:- /dev/hda >> Attempting to mount CD:- /dev/hda1 >> Attempting to mount CD:- /dev/hdb In this case /dev/hda has a single old Linux (type 83) partition and /dev/hdb has a single old Windows 98se partition (type c). /dev/hdc happens to be the cdrom, but the installation hung at /dev/hdb and never reached it. I've had this happen with multiple disks from multiple vendors ( Western Digital, Maxtor, Quantum ). These machines, and disks, come from businesses that have discarded them as obsolete. They are all "older" disks, IDE or EIDE, less than 30 gig. ( <30 gig may not be significant as I've only installed Gentoo on two systems with ATA drives larger than 30 gig. ) I've also seen it on P2, P3, and Duron systems. It has occurred on several different types of Windows partitions as well as on Linux partitions. Sometimes the hang is on the device itself; other times it is on a specific partition. The hang, when it occures, occures in the same place on the given disk on every boot. Rearrainging the disks is only significant it the CD device is placed in front of the offending disk so that it is found first. My best guess is that something is wrong in the partition itself. Rebuilding the partition table will not help if the first block of the partition remains the same. Running mke2fs always cures it. The problem is getting to a shell prompt to run the command. WORKAROUND #1: Place place the "bad" disk after the cdrom in the search sequence. It works well enough, but involves opening the case and monkeying around with the hardware. If Average Joe is trying to install Gentoo on the old computer that Aunt Sally gave him to monkey with, it may be a show-stopper. WORKAROUND #2: I've been booting a Trustix installation disk in rescue mode, repartitioning as needed and running mke2fs. This works quite well for me ... a minor inconvience at most. SUGGESTED FIX: I doubt if this is a huge problem, but it would be nice if there was a per-device timeout on your search for the CD. That would make it possible to reach the CD and the shell where normal installation is likely to fix the problem. I normally would consider this to be merely a problem with the drive partitions and not a true bug. The means to fix it, however, lies just on the other side of the hang, and I'm not experiencing this hang with Fedora or Trustix ( same systems in the same state ). Paul
This is a LiveCD problem, not an installer problem.
I'm assigning this to genkernel, as that is the place where we are doing the searching. I'm not sure if we can resolve this with the way that we boot, but if we can, we'll resolve it there, then reassign it to release@ until a new release is built with the fixed version of genkernel.
Ok, so it looks like what's happening is when it tries to mount the partition it locks up... maybe this will fix it by forcing ISO mounting (otherwise the kernel would try mounting it as vfat and what-not): diff -u -r1.47 initrd.scripts --- generic/initrd.scripts 17 Apr 2006 15:07:57 -0000 1.47 +++ generic/initrd.scripts 7 May 2006 17:57:16 -0000 @@ -71,7 +71,7 @@ if [ -b "${x}" ] then good_msg "Attempting to mount CD:- ${x}" - mount -r ${x} ${NEW_ROOT}/mnt/cdrom > /dev/null 2>&1 + mount -r -t iso9660 ${x} ${NEW_ROOT}/mnt/cdrom > /dev/null 2>&1 if [ "$?" = '0' ] then # Check for a LiveCD
Fixed in 3.4.0_pre1, thanks.
