Bug#773868: live-boot: Does not include overlayfs support in initramfs, due to wrong kernel module naming
Source: live-boot Version: 4.0.2-1 Followup-For: Bug #773868 Hello, is there any plan to fix this in jessie? Thanks Michal -- System Information: Debian Release: 8.0 APT prefers testing APT policy: (910, 'testing'), (900, 'stable'), (410, 'unstable'), (400, 'experimental') Architecture: amd64 (x86_64) Foreign Architectures: i386 Kernel: Linux 3.18.0-trunk-amd64 (SMP w/4 CPU cores) Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) (ignored: LC_ALL set to en_US.UTF-8) Shell: /bin/sh linked to /bin/bash Init: systemd (via /run/systemd/system) -- To UNSUBSCRIBE, email to debian-live-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20150213135206.2095.1829.report...@optiplex960.ruk.cuni.cz
Bug#704571: support for multiple kernel versions in binary
Package: live-build
Version: 4.0.5-1
Followup-For: Bug #704571
Control: retitle -1 build with multiple kernels of the same flavour may fail
Hello,
the build still fails when multiple kernel flavours are installed and
multiple kernels of the same flavour are installed. Live-build contains
this code:
mv binary/${_INITRAMFS}/vmlinuz-*-${_FLAVOUR}
binary/${_INITRAMFS}/vmlinuz${_NUMBER}
Obvoiusly, in case vmlinuz-*-${_FLAVOUR} expands to multiple files
this command fails.
Since none of the patches fixing the problem was accepted I will leave
it at this.
Thanks
Michal
-- Package-specific info:
-- System Information:
Debian Release: 8.0
APT prefers testing
APT policy: (910, 'testing'), (900, 'stable'), (410, 'unstable'), (400,
'experimental')
Architecture: amd64 (x86_64)
Foreign Architectures: i386
Kernel: Linux 3.18.0-trunk-amd64 (SMP w/4 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) (ignored: LC_ALL
set to en_US.UTF-8)
Shell: /bin/sh linked to /bin/bash
Init: systemd (via /run/systemd/system)
Versions of packages live-build depends on:
ii debootstrap 1.0.66
Versions of packages live-build recommends:
ii cpio2.11+dfsg-4
ii live-boot-doc 4.0.1-1
ii live-config-doc 4.0.2-1
pn live-manual-html | live-manual
live-build suggests no packages.
-- no debconf information
--
To UNSUBSCRIBE, email to debian-live-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive:
https://lists.debian.org/20150213140814.2716.74898.report...@optiplex960.ruk.cuni.cz
Processed: Re: support for multiple kernel versions in binary
Processing control commands: > retitle -1 build with multiple kernels of the same flavour may fail Bug #704571 [live-build] support for multiple kernel versions in binary Bug #573173 [live-build] support for multiple kernel versions in binary Changed Bug title to 'build with multiple kernels of the same flavour may fail' from 'support for multiple kernel versions in binary' Changed Bug title to 'build with multiple kernels of the same flavour may fail' from 'support for multiple kernel versions in binary' -- 573173: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=573173 704571: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=704571 Debian Bug Tracking System Contact ow...@bugs.debian.org with problems -- To UNSUBSCRIBE, email to debian-live-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/handler.s.b704571.14238370733664.transcr...@bugs.debian.org
Bug#773868: Patch to support overlay
I modified live-boot to support overlay from Linux kernel >=3.18 a few
days ago, by referring to kiwi from OpenSuSE.
Attached please find the patch to support overlayfs for live-boot 4.
My 2 cents.
Steven.
--
Steven Shiau
Public Key Server PGP Key ID: 4096R/47CF935C
Fingerprint: 0240 1FEB 695D 7112 62F0 8796 11C1 12DA 47CF 935C
--- a/components/9990-misc-helpers.sh
+++ b/components/9990-misc-helpers.sh
@@ -1322,7 +1322,7 @@
pidof unionfs-fuse >> /run/sendsigs.omit.d/unionfs-fuse || true )
;;
- overlayfs)
+ overlayfs|overlay)
# XXX: can multiple unionro be used? (overlayfs only handles two dirs, but perhaps they can be chained?)
# XXX: and can unionro be optional? i.e. can overlayfs skip lowerdir?
if echo ${unionro} | grep -q " "
@@ -1333,7 +1333,18 @@
panic "Overlayfs needs at least one lower filesystem (read-only branch)."
fi
unionmountopts="-o noatime,lowerdir=${unionro},upperdir=${unionrw}"
- mount -t ${UNIONTYPE} ${unionmountopts} ${UNIONTYPE} "${unionmountpoint}"
+ if ! mount -t ${UNIONTYPE} ${unionmountopts} ${UNIONTYPE} "${unionmountpoint}" 2>/dev/null; then
+# Ref: kiwi from OpenSuse kiwi-7.02.18-1.1
+# overlayfs in version >= v22 behaves differently
+ # + renamed from overlayfs to overlay
+ # + requires a workdir to become mounted
+ # + requires workdir and upperdir to reside under the same mount
+ # + requires workdir and upperdir to be in separate subdirs
+mkdir ${unionrw}/rw
+mkdir ${unionrw}/work
+unionmountopts="-o noatime,lowerdir=${unionro},upperdir=${unionrw}/rw,workdir=${unionrw}/work"
+mount -t ${UNIONTYPE} ${unionmountopts} ${UNIONTYPE} "${unionmountpoint}"
+ fi
;;
*)
--- a/components/9990-overlay.sh
+++ b/components/9990-overlay.sh
@@ -9,7 +9,7 @@
addimage_directory="${3}"
case ${UNIONTYPE} in
- aufs|unionfs|overlayfs)
+ aufs|unionfs|overlayfs|overlay)
modprobe -q -b ${UNIONTYPE}
if ! cut -f2 /proc/filesystems | grep -q "^${UNIONTYPE}\$" && [ -x /bin/unionfs-fuse ]
Bug#778317: live-build: bootstrap cache should be per-architecture
Package: live-build Version: 4.0.5-1mk1 Severity: normal Hello, When you change the live architecture and don't delete the cache directory bootstrap of the old architecture is restored from cache and image of the old architecture is built. This is trivially solved by including the architecture name in the bootstrap cache folder name. Package cache may be shared and some packages may be even reused on multiple architectures once the bootstrap is correct. Thanks Michal -- System Information: Debian Release: 8.0 APT prefers testing APT policy: (910, 'testing'), (900, 'stable'), (410, 'unstable'), (400, 'experimental') Architecture: amd64 (x86_64) Foreign Architectures: i386 Kernel: Linux 3.18.0-trunk-amd64 (SMP w/4 CPU cores) Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) (ignored: LC_ALL set to en_US.UTF-8) Shell: /bin/sh linked to /bin/bash Init: systemd (via /run/systemd/system) Versions of packages live-build depends on: ii debootstrap 1.0.66 Versions of packages live-build recommends: ii cpio2.11+dfsg-4 ii live-boot-doc 4.0.1-1 ii live-config-doc 4.0.2-1 pn live-manual-html | live-manual live-build suggests no packages. -- no debconf information -- To UNSUBSCRIBE, email to debian-live-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20150213142146.13221.11802.report...@optiplex960.ruk.cuni.cz
Bug#778327: live-build: build fails in empty directory
Package: live-build Version: 4.0.5-1mk1 Severity: normal Hello, I tried this: mkdir dl-test cd dl-test lb build No image was produced. Man page says: lb_build(1) alias for all stages lb build says: [2015-02-13 15:12:21] lb build W: No config/ directory; using defaults for all options P: live-build 4.0.5 This is clearly not true. lb config && lb build produces an image but lb build does not. IMHO this will be fixed by adding a note to run lb config in the man page and changing the W: to E: so the build fails early rather than running for an extended time producing nothing. Thanks Michal -- Package-specific info: -- System Information: Debian Release: 8.0 APT prefers testing APT policy: (910, 'testing'), (900, 'stable'), (410, 'unstable'), (400, 'experimental') Architecture: amd64 (x86_64) Foreign Architectures: i386 Kernel: Linux 3.18.0-trunk-amd64 (SMP w/4 CPU cores) Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) (ignored: LC_ALL set to en_US.UTF-8) Shell: /bin/sh linked to /bin/bash Init: systemd (via /run/systemd/system) Versions of packages live-build depends on: ii debootstrap 1.0.66 Versions of packages live-build recommends: ii cpio2.11+dfsg-4 ii live-boot-doc 4.0.1-1 ii live-config-doc 4.0.2-1 pn live-manual-html | live-manual live-build suggests no packages. -- no debconf information -- To UNSUBSCRIBE, email to debian-live-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20150213155315.2854.52744.report...@optiplex960.ruk.cuni.cz
Bug#704571: support for multiple kernel versions in binary
On 13/02/2015 14:08, Michal Suchanek wrote:
> the build still fails when multiple kernel flavours are installed and
> multiple kernels of the same flavour are installed. Live-build contains
> this code:
> mv binary/${_INITRAMFS}/vmlinuz-*-${_FLAVOUR}
> binary/${_INITRAMFS}/vmlinuz${_NUMBER}
> Obvoiusly, in case vmlinuz-*-${_FLAVOUR} expands to multiple files
> this command fails.
I have just recently completed a large chunk of work improving the
existing bootloader code in live-build. You will find an archive
containing a set of 73 commits attached to the last post of bug #775322
(which I used for tracking this work).
While I have not specifically tackled support for multiple versions of
the same kernel flavour (perhaps you could actually help me understand
the use case for this), I have improved the piece of code you refer to
here, which now applies the same logic as when a single kernel flavour
is requested (using only the latest version for each flavour).
You will find this change specifically in patch #29 in the above
mentioned set of 73 bootloader improvement patches. I then moved this
logic directly into binary_linux-image in patch #34, thus applying
simple kernel file renaming to all bootloaders, thus helping pave the
way to the introduction of EFI bootloader support (which I have a done a
lot of the work for already on my local system).
--
To UNSUBSCRIBE, email to debian-live-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/54de5101.8060...@gmail.com
Processed: retitle, set version
Processing commands for cont...@bugs.debian.org: > found 778317 4.0.5-1 Bug #778317 [live-build] live-build: bootstrap cache should be per-architecture Marked as found in versions live-build/4.0.5-1. > retitle 778317 live-build: improve cache support for rebuilds with different > arch Bug #778317 [live-build] live-build: bootstrap cache should be per-architecture Changed Bug title to 'live-build: improve cache support for rebuilds with different arch' from 'live-build: bootstrap cache should be per-architecture' > End of message, stopping processing here. Please contact me if you need assistance. -- 778317: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=778317 Debian Bug Tracking System Contact ow...@bugs.debian.org with problems -- To UNSUBSCRIBE, email to debian-live-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/handler.s.c.14238569822266.transcr...@bugs.debian.org
Bug#778317: live-build: bootstrap cache should be per-architecture
On 13/02/2015 14:21, Michal Suchanek wrote: > When you change the live architecture and don't delete the cache > directory bootstrap of the old architecture is restored from cache and > image of the old architecture is built. > > This is trivially solved by including the architecture name in the > bootstrap cache folder name. Package cache may be shared and some > packages may be even reused on multiple architectures once the bootstrap > is correct. This issue is already known about (at least to me, and it's been encountered and discussed in the mailing list), and I have had it in a mental todo list to consider writing such a patch as you describe, but I had not yet gotten around to even writing a bug report, so thanks for making one. I am not totally convinced yet whether it is better for users to simply use a separate build directory for each architecture. There are two issues this would help address: 1) Users can easily switch architecture in their config and then unknowingly run into issues upon rebuilding. 2) For users building multiple images for different architectures, while you could just use a separate build directory for each, this would allow using the same build directory if desired, without having to throw away or rename the existing cached bootstrap, things would just work automatically. In addition to the bootstrap cache however, there are also other cached items to consider: 1) Optional caching of the rootfs image. This would also require the arch being added to the name. 2) Stuff like dist info files (Contents.gz, etc) and d-i. The d-i files are already impervious to issues here due to the way they are named. I'm not certain about the others, though they certainly will be once my bug #718225 (secure wget downloads) patches get adopted, which includes some improvements in the caching area, so no point investigating on the assumption that that work will be adopted soon. -- To UNSUBSCRIBE, email to debian-live-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54de56fb.4050...@gmail.com
Bug#778327: live-build: build fails in empty directory
On 13/02/2015 15:53, Michal Suchanek wrote: > Man page says: > >lb_build(1) > alias for all stages > > lb build says: > > [2015-02-13 15:12:21] lb build > W: No config/ directory; using defaults for all options > P: live-build 4.0.5 > > This is clearly not true. > > lb config && lb build > > produces an image but > > lb build > > does not. Firstly, the man pages are very much out of date. I have gone a long way to fixing this in bug #774730, and in one or two additional misc patches here and there, none of which has been adopted into the git repo yet, never mind an official version. The manual [1] does clearly state to use lb config before lb build, but with that said, my impression from my recent work on the code base was that in this situation build would automatically run config to generate a config, and I can confirm having actually tried it now that this is not what is happening. It is certainly not ideal that build is just pushing on ahead and crashing in the middle of the chroot_hooks script, so this needs looking into. Either a check for config directory existence needs adding to chroot_hooks, or whatever is preventing config from being auto created needs fixing. [1] http://live.debian.net/manual/4.x/html/live-manual.en.html -- To UNSUBSCRIBE, email to debian-live-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54de65b9.7030...@gmail.com
