> On Mon, Nov 17, 2003 at 09:51:29PM +0100, Olivier Blin wrote: > >By the way, what should we do about devfs ? > >It is deprecated, but is there any working alternative yet ? > >udev needs userspace tools IIRC. > > uhm, good question, we still ship it enabled by default in 9.2, but > there is nothing (i belive) that depends on having it enabled > i would keep things as they are, supported but not compulsory, even if > that becomes an hassle to maintain. >
IMHO there is nothing available that could replace it udev is pretty far from finished but do wee need it i don't know > >> I will try to add cramfs/tmpfs support if someone kindly notices the > >> fact i am doing it, if this stuff is not scheduled to be merged > >> sometimes then i will not bother. > > > >Okay, I let you do it :) > ok :) > > >It would be a good start to have a look at Debian's linuxrc and > >mkinitrd, since they use exactly what you want to do. > >http://http.us.debian.org/debian/pool/main/i/initrd-tools/initrd-tools_0.1.55.tar.gz > It is a much cleaner implementation that the rh based mkinitrd we are > using, but i don't know how much space does it waste wrt ours. > Their mkinitrd is completely written in shell and the guy knows about > shell programming. It will be complex to master. and it's HUGE at least in the default config [EMAIL PROTECTED] disk2]# ll /mnt/initrd.cramfs -rw-r--r-- 1 root root 3461120 Nov 17 16:27 /mnt/initrd.cramfs [EMAIL PROTECTED] disk2]# du 242K ./bin 4.5K ./dev 1.0K ./etc 1.2M ./lib/i686 147K ./lib/modules/2.6.0-t9.ruby.7mdk/kernel/drivers/block 22K ./lib/modules/2.6.0-t9.ruby.7mdk/kernel/drivers/ide/legacy 8.5K ./lib/modules/2.6.0-t9.ruby.7mdk/kernel/drivers/ide/pci 62K ./lib/modules/2.6.0-t9.ruby.7mdk/kernel/drivers/ide 90K ./lib/modules/2.6.0-t9.ruby.7mdk/kernel/drivers/md 44K ./lib/modules/2.6.0-t9.ruby.7mdk/kernel/drivers/message/fusion 44K ./lib/modules/2.6.0-t9.ruby.7mdk/kernel/drivers/message 20K ./lib/modules/2.6.0-t9.ruby.7mdk/kernel/drivers/scsi/aacraid 189K ./lib/modules/2.6.0-t9.ruby.7mdk/kernel/drivers/scsi/aic7xxx 36K ./lib/modules/2.6.0-t9.ruby.7mdk/kernel/drivers/scsi/sym53c8xx_2 1.1M ./lib/modules/2.6.0-t9.ruby.7mdk/kernel/drivers/scsi 1.5M ./lib/modules/2.6.0-t9.ruby.7mdk/kernel/drivers <snip -- all filesystems> 3.2M ./lib/modules 4.4M ./lib 76K ./sbin 12K ./usr/sbin 13K ./usr 512 ./dev2 512 ./devfs 512 ./mnt 512 ./proc 512 ./scripts 512 ./sys 512 ./tmp 512 ./var 4.8M . may be switching busybox could help (staticly linked with diet libs) there are good sides too it should support any thinkable sitution but on that price ? and is it really nessecary to use dynamicly linked binaries + libc > I would really like to rewrite initrd since the current implementation > sucks a lot, but: > 1) there are a some things our mkinitrd does that i do not understand > or have no means to test (i.e. root on a loopback file) > 2) before embarking in such a project i would like some consensus on > whether it is needed or not > 3) i do not have that much spare time, so i'would need some help > > at the moment i was thinking on going on hacking the current mkinitrd. it's probably the better solution, it would be _damn_hard_ to switch to what debian use in a good way may be you could also check geninitrd from PLD linux (a polish distribution) it seems to support romfs|cramfs|ext2 but it doesn't seem to support 2.6, loop ... (debian do it really nice :( ... evms, lvm2,....) > >It was about to patch our linuxrc, but if you wanna do it, you're > >welcome :) > I do not particularly like it, but i do not want my previous work for > integrating lvm2 to get lost. > > >Can you have a look at the mkinitrd/make-initrd patches I've begun ? > i have already integrated them in my tree, > i am now working on mounting tmpfs on /dev (see my other mail) svetljo -- GMX Weihnachts-Special: Seychellen-Traumreise zu gewinnen! Rentier entlaufen. Finden Sie Rudolph! Als Belohnung winken tolle Preise. http://www.gmx.net/de/cgi/special/ +++ GMX - die erste Adresse für Mail, Message, More! +++