Re: livecd-creator unmounting temp image, running daemons.
by Jerry Vonau
Martin Langhoff wrote:
> On Mon, Jul 28, 2008 at 4:52 AM, Jerry Vonau <jvonau(a)shaw.ca> wrote:
>> Your cc to the livecd list are not appearing there, are you
subscribed?
>
> No - I thought it was an open list. I'll sub...
>
So did I....
> ...
>> Think the issue is going to be how the kernel modules differ from
>> version 3.2 and 3.3 of squashfs-tools.
>>
>> So lets get this clear, F7 is using the 2.6.23.?? and does not boot,
>> F8 is now booting correctly? with which kernel?
>
> Correct. This is with 2.6.23.1-21.fc7
>
>> Think the only way you'll get this to fly is to get the source for
>> squashfs-tools3.3 http://www.squashfs-lzma.org/dl/squashfs3.3.tar.gz
>> and patch your 2.6.23 kernel and recompile the module and replace the
>> module in the initrd.
>
> Thanks for pointing me to lzma. So the squashfs 3.3 on F9 is patched
> to have lzma support? Good to know... after a bit of investigation, it
> turns out that between 3.2 and 3.3 the squashfs project took the lzma
> pactches, and forgot to mention it in the changelog. How misterious
> and cute :-/
>
Hold it, lzma is not part of Fedora's squash support, but could be
patched it. The link just has all the steps the are needed to patch a
kernel, then recompile the kernel's modules. Just happens to be squashfs
as the example. ;-) There is something "cute" between 3.2 and 3.3
thou...
> After a bit of testing, squashfs-3.2 on F9 produces a squashfs I can
> mount on F7.
>
> But the image still does not boot. Something else in the initrd must
> be messing up.
What is the message now?
Jerry
15 years, 8 months
great news for Asus eee users
by Valent Turkovic
I have posted a question on ath5k wireless devel list on support for
AR2425 (AR5007EG) chipset (one that Asus eee 701 has) and got an
answer that the patches have been sent upstream. So when they got
included Asus eee's wireless will work in Fedora out of the box,
right?
This is a great news for me as a Asus eee user because wireless is the
only part of Asus eee that isn't working out of the box on Fedora 9.
For me as a future Fedora 10 user this is a great because Fedora 10
will work on Asus eee out of the box, including working wireless.
Cheers,
Valent.
--
http://kernelreloaded.blog385.com/
linux, blog, anime, spirituality, windsurf, wireless
registered as user #367004 with the Linux Counter, http://counter.li.org.
ICQ: 2125241, Skype: valent.turkovic
15 years, 8 months
Application for GSoC Project - Package WebUI
by Pavel Khardikov
Hi, Thorsten!
I'm interested in the GSoC Package WebUI project idea.
I am well acquainted with the basic principles and methods of working
with repoview and pkgdb.
I have a desire to make a site that provides information on all packages
and has convenient tools for searching, grouping, resolving
dependencies, etc...
I consider that the best variant for development of the new application
is the use or improvement repoview or pkgdb.
I think that http://packages.ubuntu.com/ and
http://www.debian.org/distrib/packages are very convenient and
functional tools.
I have a good experience in user interface design and development of web
applications:
-- scripting languages: perl, python, php and the popular frameworks for
them, like Catalyst, Django and ZendFramework
-- XML, HTML / XHTML, CSS, semantic and cross-browser making-up
-- graphic tools
What is the main criteria for application?
http://fedoraproject.org/wiki/SummerCoding/2008/Ideas
--
Best regards. Pavel Khardikov
15 years, 8 months
Re: livecd-creator unmounting temp image, running daemons.
by Jerry Vonau
Martin Langhoff wrote:
> On Sun, Jul 27, 2008 at 3:25 PM, Martin Langhoff
> <martin.langhoff(a)gmail.com> wrote:
>> After a bit more investigation. the livecd-tools package in F9
>> (017.1-1.fc9) can only build F9 correctly, and the problem boils down
>> to incorrect placement of the .ko files in the initrd. Here is how to
>> repro with F9 vs F7. In my testing, F8 shows the same problems.
>
> I tracked this down to the switch from mayflower to mkinitrd, which
> lead me to commit
> 11dbd0bb5ba4b845e80109e990e4e780ca402218 where mayflower gets
> installed and called during %post.
>
> So I updated my ks file as below. This still fails to build a bootable
> F8 or F7, both drop to an emergency shell after failing to find root
> (see below for more details). I am using git's master for these
> builds.
>
> Current kickstart file
>
> --8<----8<----8<--
> lang en_US.UTF-8
> keyboard us
> timezone US/Eastern
> auth --useshadow --enablemd5
> selinux --disabled
> firewall --disabled
> part / --size 1024
>
> repo --name=released
>
--mirrorlist=http://mirrors.fedoraproject.org/mirrorlist?repo=fedora-8&arch=$basearch
>
> %packages
> @core
> bash
> kernel
> passwd
> policycoreutils
> chkconfig
> authconfig
> rootfiles
>
> # for live initrd
> livecd-tools
> # livecd bits to set up the livecd and be able to install
> anaconda
> #isomd5sum
>
> %end
>
> %post
> # make the initrd we care about
> rm -f /boot/initrd*.img
> cp /etc/sysconfig/mkinitrd /etc/mayflower.conf
> ver=`ls /boot/vmlinuz* |head -n 1 |sed -e 's;/boot/vmlinuz-;;'`
> /usr/lib/livecd-creator/mayflower -f /boot/initrd-$ver.img $ver
> rm -f /etc/mayflower.conf
> %end
>
> %post --nochroot
> # move the initrd we created to be the booted one
> mv $INSTALL_ROOT/boot/initrd-*.img $LIVE_ROOT/isolinux/initrd0.img
> %end
> --8<----8<----8<--
>
> With this ks file, the initrd is now built correctly. But during boot
> with F8 I see all sorts of odd errors:
>
> WARNING: Bogus /etc/fstab file - cannot have /dev/root as the device
for /
> ...
> starting udevd
> creating devices
> waiting for system to settle
> ...
> SQUASHFS error: Major/Minor mismatch, trying to mount newer 3.1
filesystem
> SQUASHFS error: Please update your kernel
> mount: wrong fstype ...
>
The squsashfs-tools/kernel module on the fc7/8 live disks is to old to
mount the squashfs.img created on the newer f9 box.
> Once on the shell, the dmesg output looks normal except for the
> Squashfs errors, and ls /dev/ does not contain anything that looks
> like a usable block device. Trying to mount /dev/loop0 gives me the
> same squashfs error as before.
>
> hmmmm? hints?
>
>
>
> m
Think you would need to include, maybe recompile, the squashfs-tools rpm
from the f9. Then I think you'll need to recompile the kernel module and
include it in the initrd. I can't recall when this update to squashfs
occurred, but I believe it was tied to a kernel update between fedora
releases 8-9. so I'm not sure if the above will even work.
Jerry
15 years, 8 months
Mailing list support [Fwd: New support lists?]
by Mike Chambers
I sent the original email below to fedora-list, after seeing a certain
topic or two on that list as well as devel-list and wanted to forward it
here as well. Just something along the lines of the other topic about
cleaning up fedora-list.
Note, fedora-testers-list below should be fedora-test-list as I
accidentally called it by different name and didn't mean to create a new
list.
Anyway, just something that might make things easier.
BTW, we all know Red Hat is main primary sponsor/contributor and what
not to Fedora, but shouldn't fedora related material be behind
fedoraproject.org domain or something instead? No biggie or complain,
just asking/putting it out there.
Peace out muh bruthahs LOL
-
Mike Chambers
Fedora Project - Ambassador, Bug Zapper, Tester, User, etc..
mikec302(a)fedoraproject.org
-------- Forwarded Message --------
> From: Mike Chambers <mike(a)miketc.net>
> Reply-to: For users of Fedora <fedora-list(a)redhat.com>
> To: Fedora <fedora-list(a)redhat.com>
> Subject: New support lists?
> Date: Mon, 28 Jul 2008 14:15:54 -0500
>
> Hi all,
>
> Reading the last couple emails from another thread on maybe new list or
> two, and to segregate the content to them was brought up and have couple
> new list names that might help make sense or to get the ball rolling?
>
> fedora-list(this one) - General Fedora talk, opinions, suggestions,
> comments, distro wars, all that type stuff not support related.
>
> fedora-user-list or fedora-support-list - Bug references, problems with
> software, computer won't boot, email don't work, firefox plugins not
> working, etc.. where problems are asked about and responded to on
> *official releases*.
>
> fedora-testers-list - as is, for testing alpha, beta, preview, rc
> releases when preparing a new release.
>
> fedora-devel-list - as is, for development of fedora, and the rest that
> is currently used on that list.
>
> See the url below for currently created lists already being used..
>
> https://listman.redhat.com/mailman/listinfo/
>
> Would this help make things better?
15 years, 8 months
livecd-creator unmounting temp image, running daemons.
by Martin Langhoff
I am using a F9 host to build a F7 liveCD -- the School Server image
-- and I am finding that livecd-creator never manages to unmount the
temporary partition, and the resulting image fails to boot (tested
with qemu and on real hw).
When livecd-creator tries to unmount the CD, two processes are still
running - httpd, and a custom network daemon written in python. lsof
shows them as the only processes keeping files open. Once I kill those
processes, I can unmount cleanly.
Is this expected? Are any incompatibilities with F7 known? Should
livecd-creator try to run all the relevant init scripts with stop, and
perhaps run lsof to see if any programs need to be killed or
kill-9'ed?
The boot failures (fails to load the kernel modules, and kills init)
probably have to do with the image being incomplete.
This is the kickstart file I am using...
http://dev.laptop.org/git?p=projects/xs-livecd;a=blob;f=kickstarts/livecd...
cheers,
martin
--
martin.langhoff(a)gmail.com
martin(a)laptop.org -- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working code first
- http://wiki.laptop.org/go/User:Martinlanghoff
15 years, 8 months
trouble building OLPC's evince fork
by Daniel Drake
Hi,
I'm new to RPM building, so apologies if I am missing anything obvious.
For OLPC we need to build our forked version of evince. Our forked
version is a little out of date, and requires a poppler version older
than the one in F9.
We'll update our evince fork and hopefully get things upstreamed soon,
but for now we need things working for our efforts to stabilise our new
software builds.
So, I first built our old poppler version in the OLPC-3 disttag, as
poppler-0.6.2-5.olpc3:
http://koji.fedoraproject.org/koji/buildinfo?buildID=57583
That seems to have worked.
Next, locally, I have modified our sugar-evince spec file to do this:
%define poppler_version 0.6.2-5
BuildRequires: poppler-devel = %{poppler_version}
I tried to scratch build it, but it failed:
http://koji.fedoraproject.org/koji/taskinfo?taskID=743472
The error seems to be:
DEBUG util.py:250: No Package Found for poppler-devel = 0.6.2-5
Why is it unable to find that build?
What am I missing?
Thanks,
Daniel
15 years, 8 months
A PackageKit browser plugin
by Owen Taylor
We've been working on a website for browsing (rating, commenting,
etc) available applications for Fedora. As part of that we wanted
to be able to install/run software from the website.
I spent the last day hacking up a quick browser plugin:
If the package is not installed but is available in the package
repository, the plugin will show:
+------------------------------------+
| _Install GNU Backgammon Now_ |
| Version: 20061119-14.fc9 |
+------------------------------------+
Click on the plugin, and it will fire off gpk-install-package to install
the package; the display changes to:
+------------------------------------+
| GNU Backgammon |
| Installing... |
+------------------------------------+
once that is done (or if the package was already installed), the plugin
will show:
+------------------------------------+
| _Run GNU Backgammon_ |
| Installed version: 20061119-14.fc9 |
+------------------------------------+
Given wide usage of PackageKit and availability of the plugin, this
could also be pretty neat to put on third-party project pages. The
information you provide as parameters to the plugin is:
The name of the application for display
A list of possible package names
A list of possible desktop file names for the application
So it should be pretty robust against inter-distro differences in
package names.
README File:
http://git.fishsoup.net/cgit/packagekit-plugin/tree/README
Getting the source:
git clone git://git.fishsoup.net/packagekit-plugin
What do people think... does this make sense as part of the PackageKit
project?
- Owen
15 years, 8 months
Re: livecd-creator unmounting temp image, running daemons.
by Jerry Vonau
Martin:
Your cc to the livecd list are not appearing there, are you subscribed?
Martin Langhoff wrote:
> On Sun, Jul 27, 2008 at 8:45 PM, Martin Langhoff
> <martin.langhoff(a)gmail.com> wrote:
>> Sounds complex. I will try going in the opposite direction: downgrade
>> the package on F9 or pass flags to disable the new fanciness.
>
> Right - with the following patch to fs.py the F7 kernel won't panic
> anymore, and a basic F8 will boot. The mkfs tweak is probably
> unneeded, the main difference seems to come from -no-sparse:
>
Think the issue is going to be how the kernel modules differ from
version 3.2 and 3.3 of squashfs-tools.
So lets get this clear, F7 is using the 2.6.23.?? and does not boot,
F8 is now booting correctly? with which kernel?
> --- a/imgcreate/fs.py
> +++ b/imgcreate/fs.py
> @@ -38,7 +38,7 @@ def makedirs(dirname):
> raise
>
> def mksquashfs(in_img, out_img):
> - args = ["/sbin/mksquashfs", in_img, out_img]
> + args = ["/sbin/mksquashfs", in_img, out_img, '-no-sparse', '-b',
'131072']
>
> if not sys.stdout.isatty():
> args.append("-no-progress")
> @@ -200,7 +200,7 @@ class SparseExtLoopbackMount(SparseLoopbackMount):
>
> def __format_filesystem(self):
> rc = subprocess.call(["/sbin/mkfs." + self.fstype,
> - "-F", "-L", self.fslabel,
> + '-I', '128', "-F", "-L", self.fslabel,
> "-m", "1", "-b", str(self.blocksize),
> self.lofile,
> str(self.size / self.blocksize)])
>
>
> On F7 - my actual target ATM - the problem is not gone through. Init
> is dropping me to a panic shell, and I cannot figure out how to get
> something I can mount there. When I get dropped in the shell, lsmod
> makes it seem like no interesting modules have been loaded.
> modprobe-ing libata, cdrom and other modules works. I don't know the
> major/minor numbers enough to mknod my way around here.
>
see above
> So I suspect of the initrd, but I am unsure of what to do next. The
> mayflower initrd seems to work though it does throw some (IMO)
> meaningless errors - complaining about old module names that are not
> relevant to the 2.6.23 kernel I have:
>
> Building an initramfs at /boot/initrd-2.6.23.1-21.fc7.img for kernel
> 2.6.23.1-21.fc7
> FATAL: Module ide_cd not found.
> FATAL: Module =ata not found.
> FATAL: Module usbhid not found.
> FATAL: Module ieee1394 not found.
> Done; initramfs is 4.2M.
>
> any more hints on this track?
>
Think the only way you'll get this to fly is to get the source for
squashfs-tools3.3 http://www.squashfs-lzma.org/dl/squashfs3.3.tar.gz
and patch your 2.6.23 kernel and recompile the module and replace the
module in the initrd. There is a good post on the livecd list about
patching squashfs for lsma support and covers all the steps, perhaps you
could adapt the steps(leave out the lzma patches).
https://www.redhat.com/archives/fedora-livecd-list/2008-July/msg00007.html
https://www.redhat.com/archives/fedora-livecd-list/2008-July/msg00005.html
Repacking is initrd and iso are not that hard, but I have some notes
around if you need them.
Jerry
15 years, 8 months