Closed kaihendry closed 10 years ago
If the initramfs stuff got changed, then the git-fs changes should re-applied. I recently did that already, so you might be able to just re-apply those commits (IIRC there were two) on top of the current tree?
These two,right?
I've got a really complex merge here: http://s.natalian.org/2014-09-11/complex.diff
Yes, those two. I'd squash them together and apply them as a single commit (with a message similar to the first commit). If you just upgraded initramfs-tools, dpkg should have replaced the relevant files with the unmodified versions and, provided that upstream didn't change anything big, those commits should re-apply without problems.
I'm not sure what the diff file is that you linked, though?
The file which looks like it needs a complex merge is /lib/live/boot/9990-misc-helpers.sh
Next I need to compare with this live-boot's version?
I think it's more complicated than that since match_files_in_dir is nowhere to be found on updated version IIUC.
https://github.com/Webconverger/webc/blob/testing/lib/live/boot/9990-misc-helpers.sh and upstream http://live.debian.net/gitweb/?p=live-boot.git;a=blob;f=components/9990-misc-helpers.sh;
does not have it and I can't find it in it's history either.
Hmmm, what is the story behind match_files_in_dir? :)
It seems that mount_images_in_directory
used to contain a full list of possible filesystem types and used match_files_in_dir
to see if any of them was present. Now, it just calls is_live_path
, which also contains this same list. In other words, upstream removed some duplicate code which I think means we just have to change is_live_path
(so only the first change from 00b9d28 needs to be applied, the second is no longer needed).
I made the suggested changes I think @matthijskooijman and I'm getting segfaults whilst trying to boot... I assume whilst trying to mount the gitfs.
http://s.natalian.org/2014-10-24/call-trace.png
Would be super grateful if you lend a hand here. Not sure why I can't examine the boot.log either. :/
Turns out the problem was that the 'fuse' module is no longer autoloaded. Adding a modprobe fuse
fixes this.
I'll clean up the changes a bit and send a PR in a minute.
:+1: Nice work. I've pushed your changes to a jessie branch https://github.com/Webconverger/webc/tree/jessie
I might leave the splash as is, since (luckily) the B&W theme fits the new Webconverger logo well: https://gist.github.com/kaihendry/c13f4716243a15c75ee4
Only issue is that it doesn't seem to nicely progress.
:heart: git-fs fetch-revision=jessie
from webc-26.iso WORKS :grinning:
I fear there is a Adobe flash regression using Intel amongst other things, so there is a lot of testing work cut out for me and I'll probably create separate issues for them as I see them.
I've pushed my "testing" branch where I've did a fairly clean dist-upgrade to jessie (wheezy+1)
https://github.com/Webconverger/webc/tree/testing
However it fails to boot on the 686 version, it seems like it's failing to mount something, hence I'm thinking git-fs needs to be re-integrated.
As for the 486 version, it segfaults.
Sooner or later we need to make this step up from wheezy, but in the meantime, I am thinking of just backporting xserver-xorg-video-intel 2:2.21.15-2+b2 as that's the most pressing reason to upgrade.