Bug 7903 – small eclass task -> bootloader.eclass

• Status: RESOLVED FIXED
• Product: Portage Development

[Description Donny Davies (RETIRED) 2002-09-13 23:41:06 UTC]

Just so I dont forget..

Somebody looking to familiarize themselves with eclasses or such might want to 
look at making grub and lilo use a consistent pkg_setup() method.

Seems it keeps getting tweaked in one pkg, then the other, would be nice to 
have them share a function or something :/

Not volunteering immediately.  :)

[Comment 1 Maurizio Disimino 2002-09-15 09:37:47 UTC]

Created attachment 3946 [details]
boot.eclass

Well, here's an initial import a bit reworked from the lilo pkg_setup().
Works fine here with lilo-22.3.3.ebuild. Maybe you can start CC'ing
people to test and improve this eclass :) azarah, spanky, seemant and others.

Maurizio

[Comment 2 Markus Krainer 2002-09-16 08:34:39 UTC]

Please have also a look at bug #6820 where I submitted an ebuild for 
grub-0.92-r1. This ebuild contains a rewrite of pkg_setup that also takes 
into account the possibility of /boot beeing mounted read-only and remounts
it rw.

Markus

[Comment 3 Martin Schlemmer (RETIRED) 2002-09-17 14:00:25 UTC]

Havent looked yet at Markus Krainer's stuff (didnt feel like going back to
galeon1 and USE=-gtk2 moz to get it working properly), but that in lilo is 
tested, and works well.  We might just add the "ro" stuff to it.

[Comment 4 Donny Davies (RETIRED) 2002-09-17 14:58:32 UTC]

j2ee in #gentoo asked me to comment on this.

not sure why, but here my comment:

yes, i do want this to get completed, that why i reported it as a bug :p
no, i dont want to code it, thats why i hoped somebody else would :p
im happy to see somebody jumped up to it :p

yes, people, please *do* make sure the eclass gets the best parts from lilo and 
grub and make it as good as can be.  i got tired of seeing one package get a 
nice fix, then the other get a fix, then they get out of sync.. you know how 
the rest goes.

so go ahead and commit the sucker, and change the .ebuilds to use it :p

[Comment 5 Maurizio Disimino 2002-09-18 08:05:37 UTC]

Created attachment 3993 [details]
boot.eclass 

boot.eclass with support for /boot mounted as ready-only.
thanks Markus.
PLEASE test.

Maurizio aka j2ee

[Comment 6 Maurizio Disimino 2002-09-19 09:12:46 UTC]

Created attachment 4008 [details]
boot.eclass updated. PLEASE test and report

[Comment 7 Maurizio Disimino 2002-09-20 07:38:34 UTC]

Created attachment 4026 [details]
final (I hope) boot.eclass

this eclass works has been tested with lilo and grub and works fine.
Someone should update the lilo & grub ebuild to include this.

Maurizio

[Comment 8 Markus Krainer 2002-09-20 10:23:19 UTC]

Maurizio,

consider /boot not mounted and the following fstab entry:
/dev/hda1   /boot       ext2        ro,noatime

then your pkg_setup will mount /boot read-only!

Markus

[Comment 9 Donny Davies (RETIRED) 2002-09-20 13:43:32 UTC]

alright i added the last attachment as

  mount-boot.eclass

more testing needed; if you have another fix for this in
the meantime, attach as a patch against the .eclass please...

thank you very much to everybody involved for your continuing
contributions/participation!  i really appreciate it :)

[Comment 10 SpanKY 2002-09-20 22:38:20 UTC]

sys-apps/quik <- you think you want to incorporate this eclass into it too ?

[Comment 11 Maurizio Disimino 2002-09-21 10:02:34 UTC]

Created attachment 4052 [details]
mount-boot.eclass

markus: tnx, fixed :)
spanky: think so
woodchip: It's your turn now :) Works fine here with both lilo and grub.

Maurizio

[Comment 12 SpanKY 2002-09-21 23:13:33 UTC]

sys-apps/memtest86 <- might want to throw into this sucker too :D

[Comment 13 Donny Davies (RETIRED) 2002-09-22 19:01:39 UTC]

if people could try lilo-22.3.3-r1 and comment please.

please, again, if you have a fix, attach it as a unified
diff against the eclass in portage, not the whole eclass,
ive already spent time cleaning whitespace issues, so... :)

markus, did you have an update for grub-0.92, aside from
the eclass addition?  i think you posted an -r1 attachment,
but again i cant easily see atm what your changes were
if any (or just related to the eclass).

hopefully a bunch of these lilo/grub related bugs can be
closed up fairly soon (*yay*).  im aware memtest86 should
use the eclass too ;).

[Comment 14 Maurizio Disimino 2002-09-23 08:55:59 UTC]

Created attachment 4100 [details]
oops :) I'm sorry, here's the diff.

[Comment 15 Maurizio Disimino 2002-09-23 09:02:36 UTC]

Created attachment 4101 [details]
revised mount-boot.diff

auff .. I was still naming the eclass "boot" instead of "mount-boot"
fixed.

Maurizio

[Comment 16 SpanKY 2002-09-23 11:41:36 UTC]

*** Bug 8256 has been marked as a duplicate of this bug. ***

[Comment 17 Markus Krainer 2002-09-23 13:47:37 UTC]

Created attachment 4106 [details]
grub.diff.tgz

contains grub.ebuild.diff (0.92 to 0.92-r2!) and ChangeLog.diff

[Comment 18 Donny Davies (RETIRED) 2002-09-29 20:33:44 UTC]

okay i've committed a bunch of work on this into CVS.

as for the last attachment re: grub, i'll look at that asap.

any known problems with either grub-0.92 or lilo-22.3.3-r1?
im getting awefully itchy on unmasking both soon.

[Comment 19 Donny Davies (RETIRED) 2002-10-02 07:06:40 UTC]

i merged the last of markus' changes into grub-0.92-r1

that pretty much completes this item.

thanks again to everyone.