[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gentoo-dev
Subject:    [gentoo-dev] Results of testing the build-ix86-20010629.tbz2 snapshot build
From:       Jonas Berlin <jonas () berlin ! vg>
Date:       2001-06-30 1:34:02
[Download RAW message or body]

I tested the latest snapshot build tbz2, using:

19a768ea676e307c936787a0fbd5f11e  build-ix86-20010629.tbz2
acc9a79bddde60c5c8aa7dc1d650e89b  INSTALL-BUILD

  (If someone needs the version of INSTALL-BUILD I was using, drop me a
   line..)

Summary: No major problems were encountered during the process, except for
         the linux kernel image being too big.

-----------------------------------------------------------------------

In step 1, we should make a note that some tars (older ?) use the "-I"
option instead of "-j".

-----------------------------------------------------------------------

In step 2, we should instead of

  chroot /tmp/build /bin/bash

have

  env -i TERM=$TERM `which chroot` /tmp/build /bin/bash

This, to zero out the environment, copying only the TERM setting. I
don't think we should make or have to make any assumptions about the
environment when installing. Additionally, `which chroot` is needed
because the PATH is empty by the time env runs the command.

-----------------------------------------------------------------------

Because of the change in step 2, the PATH is empty and thus in step 3
we need to change

  env-update

to

  /usr/sbin/env-update

-----------------------------------------------------------------------

Step 5:

Number of files transferred: 2974

-----------------------------------------------------------------------

In step 7, I noticed this build included the /etc/make.profile symlink
already, or was created by env-update, dunno...

The "-march=pentiumpro" bug (imake compatibilty problems w/x11-4.1.0)
hadn't been fixed in this release. I used the following settings:

CHOST="i686-pc-linux-gnu"
CFLAGS="-march=i686 -O3 -pipe"
CXXFLAGS="-march=i686 -O3 -pipe"

(non-MAINTAINER mode)

-----------------------------------------------------------------------

In step 8, I chose to use bootstrap-20010617.packages.

Bootstrap failed, because:

-rw-r--r--    1 root     root       154124 Jun 29 17:12 /usr/bin/wget

After fixing it I checked the dir and established that all other files
had their executable bits set.

-----------------------------------------------------------------------

In step 9, I chose not to modify the USE variables in make.conf to
make sure the default install works.

-----------------------------------------------------------------------

In step 11, I chose to do:

  emerge /usr/portage/sys-kernel/linux/linux-2.4.4.9.ebuild

-----------------------------------------------------------------------

In step 12, as I already have grub installed, I chose not to do it
again, instead I'll boot from the grub commandline.

-----------------------------------------------------------------------

Jumping on to http://gentoo.org/doc/install.html / 3. Configuration..

Add "fi-latin1" to the keymap list ;-) (yes, it's a dash, not a
underscore).

The example /etc/fstab had <dump/pass>, but I made it 1 1 instead
(since I don't have a separate /boot partition, otherwise I would have
made it 1 2, or maybe 0 2, hmm..). Also, in the web the root was
mounted with "defaults" and in the example it was "notail,noatime". I
chose "defaults".

Hmm, why is the bzImage installed to /boot/boot/bzImage and not
/boot/bzImage or something.. some fhs2.2 issue ?

-----------------------------------------------------------------------

Ok, booting of the 2.4.4.9 kernel failed in a message something like
"Ran out of input data", and my guess that the kernel image would
maybe be too big was confirmed:

-rw-r--r--    1 root      1217940 Jun 30 04:00 bzImage

Don't know about you guys, but for me, the kernel stops booting when
it gets somewhere over 1 meg.. Dunno where the exact limit goes, it's
not 1048576 bytes but something a bit bigger.. Maybe it's 1 meg + the
code to purge the compressed image?

At this point, I didn't have time to do any more testing on this, so I
had to leave it here for now.

-----------------------------------------------------------------------

- xkr47


[prev in list] [next in list] [prev in thread] [next in thread]