A curse on LABEL=

[Timothy Murphy]

Alexander Dalloz wrote:

>> I was just bitten again by the LABEL= bug.
> 
> Please be careful with calling a behaviour a bug. Especially if you do
> not fully understand whats happening. There is not bug with the LABEL
> feature.

If I compile a kernel, and then say "make install",
and the installed kernel does not work, I call that a bug.
It's easy to get round - by changing root=LABEL=/ to root=/dev/hda?
in grub.conf - so I would call it a cosmetic bug.

>> For some reason, "make install" 2.6 kernels
>> causes a new entry in grub.conf using root=LABEL=/ ,
>> even though the entry being copied says /dev/hda?
> 
> I do not fully understand that sentence. You get a new entry in
> grub.conf for the new kernel with use of LABEL and with /dev/hda at same
> time?

The previous entry in grub.conf for 2.6.4 had root=/dev/hda6 .
Admittedly there was an old entry with root=LABEL=/
but it is not clear to me why the install process chose to copy this.

>> I don't know why - and don't really want to know why -
>> but this causes a panic on my machine,
>> with the root partition not found,
>> even though it has the correct label according to e2label.
> 
> Well, you don't want to know the reason? Then do not make usage of the
> LABEL feature.

That was just my point.
I don't _want_ to use the "LABEL feature" -
and neither, in my estimation, do the vast majority of Linux users -
but it is forced on me unless I take positive steps to remove it.

> It is easy to avoid LABEL usage. 

Actually, it is not that easy, since it is used by default.
It is clear from postings on the Linux newsgroups
that this is a significant cause of kernel panic,
with the root partition not being found.
And it is clear that a significant number of Linux users
do _not_ immediately diagnose the cause.

> If you want to use the LABEL feature,
> you need to have a correct labeled root partition, an initial ramdisk
> image and BSD label support compiled into the kernel.

If it requires all these preconditions to work properly,
that seems to me a very good reason not to make it the default.


-- 
Timothy Murphy  
e-mail (<80k only): tim /at/ birdsnest.maths.tcd.ie
tel: +353-86-2336090, +353-1-2842366
s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland