System: Any x86 system, but for the record my test machine is: - ASUS K8V SE Deluxe - 521MB DDR ECC RAM on a single module; - AMD ATHLON 64 3200+ What I tried (and failed): - Follow the procedure in http://www.gentoo.org/doc/en/liveusb.xml to prepare a bootable USB pendrive containing the equivalent of the CD install-x86-minimal-2006.1.iso; What happened: - boot aborted when init tried to find the install media, with the following error: >> No bootable medium found. Waiting for new devices... !! Could not find CD to boot, something else needed! What the problem is: - I determined that the problem was in the file /etc/initrd.scripts, in the function "findcdmount" to be more precise: it invokes mount with "-t iso9660" option, which is completelly unnecessary on a CD boot (Linux "mount" is completelly capable of determining what file system to use by himself), and WILL FAIL IF THE MEDIA IS NOT A CD, like the aforementioned USB pendrive. How I fixed it: - opened up "gentoo.igz", modified /etc/initrd.scripts (see diff below), generated new gentoo.igz with the modified file: worked perfectly after that. =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-[ --- etc/initrd.scripts.orig 2006-09-11 08:04:01.000000000 -0300 +++ etc/initrd.scripts 2006-09-11 10:08:07.000000000 -0300 @@ -71,7 +71,7 @@ if [ -b "${x}" ] then good_msg "Attempting to mount CD:- ${x}" - mount -r -t iso9660 ${x} ${NEW_ROOT}/mnt/cdrom > /dev/null 2>&1 + mount -r ${x} ${NEW_ROOT}/mnt/cdrom > /dev/null 2>&1 if [ "$?" = '0' ] then # Check for a LiveCD =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-] Hope this helps. Also: - The documentation on the above URL is wrong when it mentions a "dobladecenter" boot option: init doesn't check for anything with that name and will silently ignore it. From examination of the /init script inside gentoo.igz, I've determined that the correct option is "scandelay". - I've also found a "subdir=" option when examining /init; very useful, as it enables one to put all the gentoo files inside a single dir, reducing clutter (and making wrong-deletion errors more difficult for unwary users); On the other hand, the test code in the aforementioned "findcdmount" function will always look for the "livecd" file on "/", which is irritating and pointless. To fix it, here's an additional diff: =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-[ --- etc/initrd.scripts.orig 2006-09-11 08:04:01.000000000 -0300 +++ etc/initrd.scripts 2006-09-11 10:16:02.000000000 -0300 @@ -75,7 +75,7 @@ if [ "$?" = '0' ] then # Check for a LiveCD - if [ -e ${NEW_ROOT}/mnt/cdrom/livecd ] + if [ -e ${NEW_ROOT}/mnt/cdrom/${SUBDIR}/livecd ] then REAL_ROOT="${x}" break =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-] Hope to have been of help. Best Regards, -- Durval Menezes <jm165@tmp.com.br>
I just ran into this myself, with some hardware that absolutely needed 2006.1 (the 2006.0 kernel doesn't boot on it). I changed: -t iso9660 to: -t auto Because mount will not always default to auto, depending on the circumstances.
*** Bug 147265 has been marked as a duplicate of this bug. ***
(In reply to comment #0) > - The documentation on the above URL is wrong when it mentions a > "dobladecenter" boot option: init doesn't check for anything with > that name and will silently ignore it. From examination of the /init > script inside gentoo.igz, I've determined that the correct option is > "scandelay". Actually, the correct option is "slowusb" rather than scandelay. Anyway, the documentation isn't maintained by us, and I actually never knew it became an "official" document, as Release Engineering does not support LiveUSB and the developer who did the work on it has left the project. That being said, I've fixed the initrd.scripts for both the iso9660 and SUBDIR bugs. These will be fixed in the next genkernel, but a new Gentoo release using it will not be out for quite some time.
*** Bug 145932 has been marked as a duplicate of this bug. ***
Fixed in 3.4.1, thanks!
*** Bug 158530 has been marked as a duplicate of this bug. ***
Actually... this should probably stay open until 2007.0 is released.
Well, that was fixed but it appears that there's a different error, which is handled in a different bug (173117).
Actually, 143774 is the bug for the exact fix.