On 07/26/2011 11:32 PM, Michael Prokop wrote:
01_fix_output_file.patch
applied, thanks.
07_support_findiso.patch
misses manpage entry, and please do the silencing of the umount call in a seperate commit/patch, so that it can (if needed) reverted independently from the fromiso thing.
10_validateroot.patch
intrigeri commented about that already.
16_nodhcp.patch
who uses NODHCPHOSTNAME? i couldn't find any reference in live-boot nor initramfs-tools that makes use of this variable.
19_revert_toram_bootoption_to_expected_behaviour.patch
since we have d-i with l-i by default on the images these days, we should by default copy the whole image to ram, and, if too few space available, either automatically fall back to copying the filesystem only (plus giving a message about it), or, failing with the advice to use a parameter to specify that the filesystem should be copied only.
however, just 'reverting' the command without above doesn't make sense.
25_support_lvm_for_live-media.patch
reformated to match the documented codying style, otherwise applied, thanks.
27_support_static_ip.patch
looks rather intrusive, in particular, has broken formating, and i wonder if there's no code in intramfs-tools already that can be sourced for that.
30_support_multiarch_dns.patch
i'll wait for that on comments from Michal.
* Daniel Baumann wrote [04.08.11 21:17]: Hi,
16_nodhcp.patch
who uses NODHCPHOSTNAME? i couldn't find any reference in live-boot nor initramfs-tools that makes use of this variable.
Hm? There is no NODHCPHOSTNAME in http://git.grml.org/?p=live-boot-grml.git;a=blob;f=debian/patches/16_nodhcp....
27_support_static_ip.patch
looks rather intrusive, in particular, has broken formating, and i wonder if there's no code in intramfs-tools already that can be sourced for that.
There is code in initramfs-tools which supports some features. I can create a patch which tries to reuse the existing code in initramfs-tools and talk with maks about necessary changes.
Ulrich
On 08/04/2011 09:54 PM, Ulrich Dangel wrote:
Hm? There is no NODHCPHOSTNAME in http://git.grml.org/?p=live-boot-grml.git;a=blob;f=debian/patches/16_nodhcp....
indeed, my fault.. i accidentally looked at 11_dhcphostname.patch instead of 16_nodhcp.patch in the checkout of the sources. sorry.
however.. NODHCP is not used either.
There is code in initramfs-tools which supports some features. I can create a patch which tries to reuse the existing code in initramfs-tools and talk with maks about necessary changes.
that would be most appreciated. the whole network config stuff in live-boot is really messy, but so far, noone had the time to look into cleaning it up properly.
* Daniel Baumann wrote [04.08.11 22:14]:
On 08/04/2011 09:54 PM, Ulrich Dangel wrote:
Hm? There is no NODHCPHOSTNAME in http://git.grml.org/?p=live-boot-grml.git;a=blob;f=debian/patches/16_nodhcp....
indeed, my fault.. i accidentally looked at 11_dhcphostname.patch instead of 16_nodhcp.patch in the checkout of the sources. sorry.
however.. NODHCP is not used either.
It is used in 27_support_static_ip. It seems though that it should be enough to unset the variable.
On 4 August 2011 22:59, Ulrich Dangel mru@spamt.net wrote:
- Daniel Baumann wrote [04.08.11 22:14]:
On 08/04/2011 09:54 PM, Ulrich Dangel wrote:
Hm? There is no NODHCPHOSTNAME in http://git.grml.org/?p=live-boot-grml.git;a=blob;f=debian/patches/16_nodhcp....
indeed, my fault.. i accidentally looked at 11_dhcphostname.patch instead of 16_nodhcp.patch in the checkout of the sources. sorry.
however.. NODHCP is not used either.
It is used in 27_support_static_ip. It seems though that it should be enough to unset the variable.
There is already code to disable dhcp when netbooting and code to use network configuration from the live media, and code to set particular a ip address. What is missing that you need added by this patch?
Thanks
Michal
Hi,
Michal Suchanek wrote (04 Aug 2011 22:40:10 GMT) :
On 4 August 2011 22:59, Ulrich Dangel mru@spamt.net wrote:
- Daniel Baumann wrote [04.08.11 22:14]:
however.. NODHCP is not used either.
It is used in 27_support_static_ip. It seems though that it should be enough to unset the variable.
There is already code to disable dhcp when netbooting and code to use network configuration from the live media, and code to set particular a ip address. What is missing that you need added by this patch?
Grml folks: I can't do anything of these two patches of yours unless you answer Michal's question:
16_nodhcp.patch 27_support_static_ip.patch
Cheers,
* intrigeri intrigeri+debian-live@boum.org [110909 09:53]:
Michal Suchanek wrote (04 Aug 2011 22:40:10 GMT) :
On 4 August 2011 22:59, Ulrich Dangel mru@spamt.net wrote:
- Daniel Baumann wrote [04.08.11 22:14]:
however.. NODHCP is not used either.
It is used in 27_support_static_ip. It seems though that it should be enough to unset the variable.
There is already code to disable dhcp when netbooting and code to use network configuration from the live media, and code to set particular a ip address. What is missing that you need added by this patch?
Grml folks: I can't do anything of these two patches of yours unless you answer Michal's question:
16_nodhcp.patch 27_support_static_ip.patch
This patch is required to support standard ip= style IP configuration, as supported by the kernel and klibc. Unfortunately the code in initramfs-tools isn't flexible enough so we couldn't reuse it.
Also the patch simplifies the code quite a bit.
Hi,
Christian Hofstaedtler wrote (11 Sep 2011 22:01:03 GMT) :
16_nodhcp.patch 27_support_static_ip.patch
This patch is required to support standard ip= style IP configuration, as supported by the kernel and klibc. Unfortunately the code in initramfs-tools isn't flexible enough so we couldn't reuse it.
Also the patch simplifies the code quite a bit.
Thanks for the explanation. Christian and other grml folks, there are a few things that need to be improved before I'm happy to merge these two patches in:
1. The rationale for get_ipconfig_para, and the way this function works, should be minimally documented. Two lines of inline comments should be enough.
2. The fact standard ip= parameters can now be used must be documented in the manpage.
3. Your patch seems to remove support for ip=[frommedia]. If this is intended, please explain why, and update the manpage accordingly. If this is not intended, how about getting the deleted code back?
Other than that, I've reviewed these patches quite carefully and they seem right to me. Not only do they add support for standard ip= arguments (and allow using it for multiple devices, btw), but they generally improve the robustness and maintainability of the do_netsetup code. However, I'm far from having a shard global view of the live-boot networking -related code, so I'd welcome other reviews before we merge these patches.
Bye,
Hello
On 4 August 2011 21:17, Daniel Baumann daniel.baumann@progress-technologies.net wrote:
On 07/26/2011 11:32 PM, Michael Prokop wrote:
30_support_multiarch_dns.patch
i'll wait for that on comments from Michal.
I propose to add this code to to initramfs-hooks to handle nss modules and similar ( http://paste.debian.net/plain/125089 ):
# include a module dynamically loaded by a library # $1 - directory to search for the library (may be / to search all of initramfs) # $2 - library to search for # $3 - module to include relative to library found # example: lib_module /lib 'libc.so.*' 'libnss_dns.so.*' # lib_module /usr/lib 'libpango-*.so.*' 'pango/*/modules/pango-basic-fc.so' # Does not handle spaces in directory or module names and .. in module names. lib_module() { local dir lib mod lib_dir i j dir="$1" lib="$2" mod="$3" { find "${DESTDIR}${dir}" -name "${lib}" -type l find "${DESTDIR}${dir}" -name "${lib}" -type f ; } | { while read i ; do lib_dir="$(dirname "$i" | sed -e "s ^${DESTDIR} " )" ls "${lib_dir}"/${mod} | { while read j ; do copy_exec "$j" done ; } done ; } }
Note that this installs the module exactly for the libraries that are included in the initramfs, and does not install 32bit modules from ia32-libs when the 32bit library is not used by any binary in initramfs. This requires that the module is located in the same place relative to the library which seems to be the case so far wherever the library is installed.
I am waiting for feedback from initramfs-tools people.
Thanks
Michal
Hi Michal and everyone involved,
Michal Suchanek wrote (04 Aug 2011 22:48:27 GMT) :
I propose to add this code to to initramfs-hooks to handle nss modules and similar ( http://paste.debian.net/plain/125089 ):
[...]
I am waiting for feedback from initramfs-tools people.
Any news on this?
For the record, I'm awaiting your confirmation before I dig into your lib_module() and Grml's 30_support_multiarch_dns.patch any deeper.
Bye,
On 8 September 2011 16:05, intrigeri intrigeri+debian-live@boum.org wrote:
Hi Michal and everyone involved,
Michal Suchanek wrote (04 Aug 2011 22:48:27 GMT) :
I propose to add this code to to initramfs-hooks to handle nss modules and similar ( http://paste.debian.net/plain/125089 ):
[...]
 I am waiting for feedback from initramfs-tools people.
Any news on this?
For the record, I'm awaiting your confirmation before I dig into your lib_module() and Grml's 30_support_multiarch_dns.patch any deeper.
I am not sure what to do about this. It would be nice to get lib_module in initramfs-tools but initramfs-tools folks did not say anything about the patch.
Clearly, some solution for this is needed, and it would better be one that works for other stuff as well (eg. the gtk libs for plymouth).
Thanks
Michal
Hi,
Michal Suchanek wrote (09 Sep 2011 07:20:43 GMT) :
On 8 September 2011 16:05, intrigeri intrigeri+debian-live@boum.org wrote:
For the record, I'm awaiting your confirmation before I dig into your lib_module() and Grml's 30_support_multiarch_dns.patch any deeper.
I am not sure what to do about this. It would be nice to get lib_module in initramfs-tools but initramfs-tools folks did not say anything about the patch.
Any pointer to the discussion with initramfs-tools folks? Any idea what others, such as me, can do to help this move forward?
Bye,
On 9 September 2011 11:21, intrigeri intrigeri+debian-live@boum.org wrote:
Hi,
Michal Suchanek wrote (09 Sep 2011 07:20:43 GMT) :
On 8 September 2011 16:05, intrigeri intrigeri+debian-live@boum.org wrote:
For the record, I'm awaiting your confirmation before I dig into your lib_module() and Grml's 30_support_multiarch_dns.patch any deeper.
I am not sure what to do about this. It would be nice to get lib_module in initramfs-tools but initramfs-tools folks did not say anything about the patch.
Any pointer to the discussion with initramfs-tools folks? Any idea what others, such as me, can do to help this move forward?
Somebody on IRC said they will discuss it. I did not receive any feedback in a few days so I filed a bug report. I did not hear back from initramfs maintainers. You can look at the hook function and review it I guess but I don't know what can be done to move this forward.
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=636697
Thanks
Michal
Hi,
Daniel Baumann wrote (04 Aug 2011 19:17:32 GMT) :
07_support_findiso.patch
misses manpage entry
Grml folks, would you mind:
- adding needed doc to the manpage - replacing "grep -q /live/findiso /proc/mounts" with mountpoint (+ adding to the initramfs using copy_exec)
... and resubmitting? Seems like these two minor changes are all that's needed to merge this patch into upstream live-boot, which I'd like to do.
and please do the silencing of the umount call in a seperate commit/patch, so that it can (if needed) reverted independently from the fromiso thing.
I extracted and pushed this part (commit 2d4d31b in debian-next).
Bye,
Teilnehmer (5)
-
Christian Hofstaedtler -
Daniel Baumann -
intrigeri -
Michal Suchanek -
Ulrich Dangel