[WBEL-devel] amd64 (and ia64 news)
Pasi Pirhonen
upi@iki.fi
Mon, 22 Dec 2003 01:23:33 +0200
Hi,
There is always a catch. Friday i thought that 'i have already almoust
everything compiled so it'd be pretty aesy to pack ia64 together'.
Everything is fine a i do it. I tend to grab the boot.iso from emery
release and do NFS-install. I have been trying to generate a full
ISO-image disk1 for my rx2600 to boot, but the damn thing just doesn't
boot from the CD-RW. I actually striped it down to very parameters that
anaconda uses to make boot.iso, but still nothing.
So i am not sure if i have ISO-images ready by morning (local time)
before i have to go to work again.
If i use the boot.iso for NFS-install, it's working IMO allright, so
there is hope. I only have that rx2600, so my reference platforms are
pretty much limited to that tho.
amd64.
I see two paths just now. Doing it strictly by following what Redhat
gets out errdata or making it better.
Last night i took again Jeff Garzik's 2.4.23-libata2 patch. Merged it
into latest errdata kernel. Tweaked the minimum of anaconda to get
libata included and made NFS-install (my manually fixing the initrd
before booting after installer as it was just a quick-n-dirty hack).
I've been hammering my SATA-rooted box and it seems to be stable so far.
The other path i see, is to enhance the release beyond what RedHat
thinks i suitable for customers (maybe they'll release the libata for
RHEL3 too. i don't know).
Anaconda would need more tweaking as would at least mkinitrd (to make it
aware of those sata_* modules needing libata.o).
I see it more desktop enhancement as server-side althought even my
Tyan mobo does have Promise SATA controller integrated.
This could easily enhance the i386 supporting SATA-rooted installs too.
Any thoughts?
PS. latest for ia64.
I took the very boot.img out of the boot.iso and used that one for
generating the CD1. Now it's working, so there is hope :)
--
Pasi Pirhonen - upi@iki.fi - http://iki.fi/upi/