lkml.org 
[lkml]   [1999]   [Apr]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: PROBLEM: MO mounting not working w/ 2.2.3+
    From alan@lxorguk.ukuu.org.uk Thu Apr  8 04:23:32 1999

> The present state of affairs is that you have to rewrite your 2.2.2
> partition table for 2.2.3. (By taking all entries and dividing them
> by four.) Remind me to mail you a program that does this - write to
> aeb@cwi.nl.

[I have now made this offer on two fora and nobody took me up on it.
That is good, since it means that people have no problems handling
the situation themselves.]

Learning enough about partition tables to make the code auto switch
is still on my TODO list.

Hmm. Probably you know all there is to know.

But auto switch?
If the partition table says that /dev/sda1 extends from
sector 1 to sector 200, then how are you going to decide
what sector size was meant?

I do not think auto switch is possible in the kernel.
User space might try heuristics and examine the raw disk
for things that look like the start of an ext2 or FAT filesystem.
Extremely ugly.



A vaguely related matter:

Long ago someone had boot problems because the kernel does
autodetect the type of the root filesystem, incorrectly in
his case. Indeed, it is quite possible that a partition by
coincidence has the magic numbers for several different
filesystem types.

Several people, including me, have had problems with a disk
containing random garbage. It happens (with 4.6% probability)
that the kernel at boot time interprets the garbage as a
partition table with an extended partition somewhere, next
tries to read the first sector of that extended partition,
which almost surely is past the end of the disk, gets a read
error and dies (it is long ago, I forgot the details of the
death, probably an infinite `resetting SCSI bus').

This means that we need two boot-time flags:
1) rootfstype=
2) skip_pt_read

where the 2nd one would tell the kernel only to look at
the root device and not worry about the partition tables
elsewhere.
It might even mean skipping the partition table reading
entirely if we had a third parameter telling where on the
root device the root partition lives
3) rootpt=START,LENGTH

These days we are getting more and more cruft in genhd.c
to read all kind of partition schemes. Such schemes should
be handled like filesystems are in VFS: invoked, perhaps
from a module, when we mention their name, not autodetected.


[You see, you want to autodetect. I hate autodetection and
heuristics, schemes that are right almost all the time.
Because they will fail. This partition reading is one place
where it should be possible to stop kernel autodetection.]

Andries

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:51    [W:0.092 / U:0.800 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site