Bug#493863: firmware-bnx2: doesn't rebuild all initrds
Peter Palfrader
Version: 0.4+etchnhalf.1
Severity: important
I installed the etchnhalf kernel, rebooted only to find out that I had
not network, booted the old kernel again and installed firmware-bnx2:
| thelma:~# apt-get install firmware-bnx2
| Reading package lists... Done
| Building dependency tree... Done
| The following NEW packages will be installed:
| firmware-bnx2
| 0 upgraded, 1 newly installed, 0 to remove and 0 not upgraded.
| Need to get 104kB of archives.
| After unpacking 279kB of additional disk space will be used.
| Get:1 http://mirror.came.sbg.ac.at etch/non-free firmware-bnx2 0.4+etchnhalf.1 [104kB]
| Fetched 104kB in 0s (1818kB/s)
| Selecting previously deselected package firmware-bnx2.
| (Reading database ... 62627 files and directories currently installed.)
| Unpacking firmware-bnx2 (from .../firmware-bnx2_0.4+etchnhalf.1_all.deb) ...
| Setting up firmware-bnx2 (0.4+etchnhalf.1) ...
| update-initramfs: Generating /boot/initrd.img-2.6.18-5-amd64
Note how it doesn't rebuild the 2.6.24 etchnhalf initrd. Needless to say
the system still didn't properly boot, only purging and reinstalling the
linux-image-2.6.24-etchnhalf.1-amd64 did the trick.
Maybe the firmware-bnx2 package should do "update-initramfs -u -k all"
instead of just "update-initramfs -u".
--
To UNSUBSCRIBE, email to debian-bugs-...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listm...@lists.debian.org
Debian Bug Tracking System
Your message dated Sun, 29 Mar 2009 18:35:34 +0200
with message-id <2009032916...@stro.at>
and subject line Re: firmware-bnx2: doesn't rebuild all initrds
has caused the Debian Bug report #493863,
regarding firmware-bnx2: doesn't rebuild all initrds
to be marked as done.
This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.
(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact ow...@bugs.debian.org
immediately.)
--
493863: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=493863
Debian Bug Tracking System
Contact ow...@bugs.debian.org with problems
Peter Palfrader
thanks
On Sun, 29 Mar 2009, Debian Bug Tracking System wrote:
> > Maybe the firmware-bnx2 package should do "update-initramfs -u -k all"
> > instead of just "update-initramfs -u".
>
> no.
> it was explicitly asked that packages in postinst only update
> the newest initramfs (which can be the one which has the symlink).
> yes conservative people want to have least possible "damage" policy,
> but it is easily overridable.
It didn't update the newest initramfs - for 2.6.24. It updated the one
it was running at the time - 2.6.18.
--
| .''`. ** Debian GNU/Linux **
Peter Palfrader | : :' : The universal
http://www.palfrader.org/ | `. `' Operating System
| `- http://www.debian.org/
maximilian attems
> reopen 493863
> thanks
no fun those games, thanks.
>
> It didn't update the newest initramfs - for 2.6.24. It updated the one
> it was running at the time - 2.6.18.
sure you had the initrd symlink pointing to that one.
so it got attributed as the newest.
maximilian attems
> maximilian attems schrieb am Sonntag, dem 29. März 2009:
>
> > > It didn't update the newest initramfs - for 2.6.24. It updated the one
> > > it was running at the time - 2.6.18.
> >
> > sure you had the initrd symlink pointing to that one.
> > so it got attributed as the newest.
>
please post on that box output of
sh -x update-initramfs -u
Peter Palfrader
> > It didn't update the newest initramfs - for 2.6.24. It updated the one
> > it was running at the time - 2.6.18.
>
> sure you had the initrd symlink pointing to that one.
> so it got attributed as the newest.
I don't have symlinks.
--
| .''`. ** Debian GNU/Linux **
Peter Palfrader | : :' : The universal
http://www.palfrader.org/ | `. `' Operating System
| `- http://www.debian.org/
--
Peter Palfrader
> On Sun, Mar 29, 2009 at 08:25:46PM +0200, Peter Palfrader wrote:
> > maximilian attems schrieb am Sonntag, dem 29. März 2009:
> >
> > > > It didn't update the newest initramfs - for 2.6.24. It updated the one
> > > > it was running at the time - 2.6.18.
> > >
> > > sure you had the initrd symlink pointing to that one.
> > > so it got attributed as the newest.
> >
> > I don't have symlinks.
>
> please post on that box output of
> sh -x update-initramfs -u
Ah. there were symlinks in / - tho they were never ever used on this
system.
Seems pretty broken to rely on symlinks that might (and are) years out
of date.
--
| .''`. ** Debian GNU/Linux **
Peter Palfrader | : :' : The universal
http://www.palfrader.org/ | `. `' Operating System
| `- http://www.debian.org/
--
maximilian attems
retilte 493863 update-initramfs follows symlink before searching for newest kernel
stop
On Sun, Mar 29, 2009 at 08:34:26PM +0200, Peter Palfrader wrote:
> maximilian attems schrieb am Sonntag, dem 29. März 2009:
>
> > On Sun, Mar 29, 2009 at 08:25:46PM +0200, Peter Palfrader wrote:
> > > maximilian attems schrieb am Sonntag, dem 29. März 2009:
> > >
> > > > > It didn't update the newest initramfs - for 2.6.24. It updated the one
> > > > > it was running at the time - 2.6.18.
> > > >
> > > > sure you had the initrd symlink pointing to that one.
> > > > so it got attributed as the newest.
> > >
> > > I don't have symlinks.
> >
> > please post on that box output of
> > sh -x update-initramfs -u
>
> Ah. there were symlinks in / - tho they were never ever used on this
> system.
>
> Seems pretty broken to rely on symlinks that might (and are) years out
> of date.
has been such since version 0.65, where i changed to first use the
link then a heuristic for the newest image and afterwards only
the current kernel (which was first before).
as this symlink usage should be phased out i agree with you
that an update-initramfs -u call should try for highest version
first before falling back to the linked one.
maximilian attems
>
> Ah. there were symlinks in / - tho they were never ever used on this
> system.
>
> Seems pretty broken to rely on symlinks that might (and are) years out
> of date.
dpkg --compare-versions 2.6.29-rc8-amd64 '>' 2.6.29-1-amd64 && echo true
true
grrr
so now it assumes that 2.6.29-rc8 is newer then 2.6.29,
guess that is just an issue for dev boxes having such $(uname -r) versions
lying around but still enerving.
Debian Bug Tracking System
Your message dated Thu, 02 Apr 2009 11:02:04 +0000
with message-id <E1LpKgO-...@ries.debian.org>
and subject line Bug#493863: fixed in initramfs-tools 0.93.2
has caused the Debian Bug report #493863,