On Wed, Dec 26, 2007 at 11:37:11AM -0700, jimc wrote: > Luciano Rocha wrote: > > On Mon, Dec 24, 2007 at 09:21:05PM +0530, Shourya Sarcar wrote: > > > >> I had a few related questions on initramfs/initrd. If someone could > >> provide me some help and/or point to places where I can go and look, that > >> would be very nice. > >> > >> Is there a difference between initramfs and initrd ? If not, what's the > >> difference between doing a "mkinitrd" and "update-initramfs -cv" > >> > > > > Initrd is usually an image of a filesystem (cramfs, ext2fs, etc.). > > Initramfs is a series of cpio archives (compressed or not). > > > > mkinitrd on current RedHat, Fedora, etc., really create initramfs > > archives, not the old, filesystem image version. > > > > > > pardon me for freeloading on this thread, but it gets close to a notion I > had a while back. > > Enhance kconfig system to accept CONFIG_FOO='e' as a near synonym for ='m' > > 'e' would carry current meaning, but additionally would indicate that the > module > (and its dependencies) be added into the initrd file. > > It doesnt really add any fundamentally new capabilities, but it would > simplify the > creation of customized initrds. > > Anyone here think this would be in bad taste ? or perhaps a good idea ? Hm, the current mkinitrd scripts (at least in RedHat) look for what is needed to boot, then add it to the image. That is, IMHO, a better approach: run-time detection Vs. compile-time option. Besides, if you know the module is needed to boot, why not compile it in? There are only a few modules I'd rather have it as modules instead of built-in. Anyway, the mkinitrd, and the initrd created, does need to improve. Here's for klibc improvement and standardization. -- Luciano Rocha <luciano@xxxxxxxxxxx> Eurotux Informática, S.A. <http://www.eurotux.com/>
Attachment:
pgp0P7RjlHbfG.pgp
Description: PGP signature
