[Yaird-devel] Bug#457467: Bug#457467: failures due to changed kernel interfaces is a feature, not a bug

[maximilian attems]

tags 457177 - fixed
stop


On Sat, 29 Dec 2007, Jonas Smedegaard wrote:

> package netatalk
> severity 457467 normal
> block 457467 by 457177
> tags 457467 fixed
> thanks
>
> 
> Yaird tries very hard to only only build initial ramdisks that it is 
> confident will work. This causes failure each time the kernel changes 
> its boot-related hints in /sys or other places needed to safely resolve 
> booting needs of the new kernel from a running kernel.
> 
> It is not true that yaird "breaks on every new kernel".

yaird breaks on every new kernel, just experienced as i wanted
to let you know another important failure:
uname -a
Linux dual 2.6.24-rc7-amd64 #1 SMP Wed Jan 16 09:27:04 UTC 2008 x86_64 GNU/Linux
 mkinitrd.yaird -o /tmp/foo
 yaird error: unrecognised device: /sys/devices/LNXSYSTM:00
 yaird error: unrecognised device: /sys/devices/LNXSYSTM:00/device:00
 yaird error: unrecognised device:
 /sys/devices/LNXSYSTM:00/device:00/PNP0A08:00
 yaird error: unrecognised device:
 /sys/devices/LNXSYSTM:00/device:00/PNP0A08:00/LNXVIDEO:00
 yaird error: unrecognised device: /sys/devices/LNXSYSTM:00
 yaird error: unrecognised device: /sys/devices/LNXSYSTM:00/device:00
 yaird error: unrecognised device:
 /sys/devices/LNXSYSTM:00/device:00/PNP0A08:00
 yaird error: unrecognised device:
 /sys/devices/LNXSYSTM:00/device:00/PNP0A08:00/device:05
 yaird error: unrecognised device:
 /sys/devices/LNXSYSTM:00/device:00/PNP0A08:00/device:05/LNXVIDEO:01
 yaird error: there were errors in this run, aborting now without output
 (fatal)


 
> Current behaviour is a careful design decision, not a bug.

your target is Linux and not solaris. if you design only for your
past users you have lost. Linux is a fast moving target.

this bug is of an rc severity as hinted by the debian testing rm.
/sysfs, /proc entries and config variables change, but yaird is
still expected to produce a damm initramfs. aboves output is fine
as warning, but must not be fatal.
 
> Sure each change needs an update to yaird. But the potential for these 
> kinds of bugs are not considered a bug by the maintainer of this 
> package.
> 
> Keeping this bugreport open until bug#457177 is resolved.

as seen aboves you may want to clone this bugreport for newly
breakage for uname > 2.6.23. it is strictly unreasoned that each
linux kernel release has to break the initramfs generator.
and no we are not yet even speaking of subsys switching ide -> pata.

best regards

-- 
maks