More problems with Kernel 2.4.10-pre12 and mounting

More problems with Kernel 2.4.10-pre12 and mounting

[patrick]

To continue with 2.4.10-pre12 problems..

(This is a 2.4.9 with the 2.4.10-pre12 patch applied, everything
downloaded from ftp.kernel.org)

I've currently got two SCSI drives on my system.  /dev/sda is my dedicated
Linux drive, and /dev/sdb is my shared drive between Linux and
MacOS.  /dev/sda runs off of an Adaptec 2940UW card and /dev/sdb off of
the internal MESH of the PowerMac 8500.

If I try to do a

 mount -t hfs /dev/sdb3 /mnt/macos
 or
 mount /dev/sdb3 /mnt/macos

I get a message:

mount: /dev/sdb3 has wrong major or minor number

In addtion, if I issue:

pdisk -l /dev/sdb

I get a message:

pdisk: Can't read block 0 from '/dev/sdb'

Any ideas?

(BTW: This is not a hardware problem.  If I reboot with 2.2.18 everything
works fine).

--
patrick m pritchard
disposable hero.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: More problems with Kernel 2.4.10-pre12 and mounting

[Timothy A. Seufert]

At 12:21 AM -0400 9/23/01, patrick wrote:

>Any ideas?

You could always give up and connect sdb to your 2940UW's internal
50-pin connector instead of mesh.  (only half joking -- I'm under the
impression that the mesh driver is not well maintained these days.)
--
Tim Seufert

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: More problems with Kernel 2.4.10-pre12 and mounting

[patrick]

On Sat, 22 Sep 2001, Timothy A. Seufert wrote:

> You could always give up and connect sdb to your 2940UW's internal
> 50-pin connector instead of mesh.  (only half joking -- I'm under the
> impression that the mesh driver is not well maintained these days.)

I wish it were that easy.

I've just noticed that it not only MESH, but the Adaptec card as
well.  Which means the problem isn't limited to the driver (MESH or
Adaptec), and it isn't my own hardware (again, 2.2.x works fine).

Me thinks it is something to do with the Apple partition map support in
general.

--
patrick m pritchard
disposable hero.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: More problems with Kernel 2.4.10-pre12 and mounting

[Ethan Benson]

[-- Attachment #1: Type: text/plain, Size: 932 bytes --]

On Sun, Sep 23, 2001 at 03:14:39AM -0400, patrick wrote:
> 
> On Sat, 22 Sep 2001, Timothy A. Seufert wrote:
> 
> > You could always give up and connect sdb to your 2940UW's internal
> > 50-pin connector instead of mesh.  (only half joking -- I'm under the
> > impression that the mesh driver is not well maintained these days.)
> 
> I wish it were that easy.
> 
> I've just noticed that it not only MESH, but the Adaptec card as
> well.  Which means the problem isn't limited to the driver (MESH or
> Adaptec), and it isn't my own hardware (again, 2.2.x works fine).
> 
> Me thinks it is something to do with the Apple partition map support in
> general.

what happens if you run the following:

dd if=/dev/sdb of=/dev/null bs=512 count=1

i am betting you will get a `No such device' error which means the
kernel thinks there is no second scsi device.

-- 
Ethan Benson
http://www.alaska.net/~erbenson/

[-- Attachment #2: Type: application/pgp-signature, Size: 0 bytes --]

Re: More problems with Kernel 2.4.10-pre12 and mounting

[patrick]

On Sat, 22 Sep 2001, Ethan Benson wrote:

> what happens if you run the following:
>
> dd if=/dev/sdb of=/dev/null bs=512 count=1
>
> i am betting you will get a `No such device' error which means the
> kernel thinks there is no second scsi device.

Nope, I get:

1+0 records in
1+0 records out

Related, my partition map for /dev/sdb is:

Partition   Dec
0           Apple driver
1           ??
2           ??
3           HFS
4           e2fs

I can mount /dev/sdb4.

This all points to a problem with Apple Partition Support..?

--
patrick m pritchard
disposable hero.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: More problems with Kernel 2.4.10-pre12 and mounting

[Ethan Benson]

[-- Attachment #1: Type: text/plain, Size: 1126 bytes --]

On Sun, Sep 23, 2001 at 01:33:54PM -0400, patrick wrote:
> On Sat, 22 Sep 2001, Ethan Benson wrote:
> 
> > what happens if you run the following:
> > 
> > dd if=/dev/sdb of=/dev/null bs=512 count=1
> > 
> > i am betting you will get a `No such device' error which means the
> > kernel thinks there is no second scsi device.
> 
> Nope, I get:
> 
> 1+0 records in
> 1+0 records out
> 
> Related, my partition map for /dev/sdb is:
> 
> Partition   Dec
> 0           Apple driver
> 1           ??
> 2           ??
> 3           HFS
> 4           e2fs
> 
> I can mount /dev/sdb4.
> 
> This all points to a problem with Apple Partition Support..?

is it possible you have an x86 BIOS partition table on that disk as
well?  if mac-fdisk -l doesn't work then your mac partition table is
corrupt.  but if that does work but the kernel is getting other ideas
about your partitions you probably have both an x86 and mac partition
table (its actually possible for them to coexist, but its a gross and
uneeded hack).  what does /proc/partitions say?

-- 
Ethan Benson
http://www.alaska.net/~erbenson/

[-- Attachment #2: Type: application/pgp-signature, Size: 0 bytes --]

Re: More problems with Kernel 2.4.10-pre12 and mounting

[patrick]

On Sun, 23 Sep 2001, Ethan Benson wrote:

> is it possible you have an x86 BIOS partition table on that disk as
> well?  if mac-fdisk -l doesn't work then your mac partition table is
> corrupt.  but if that does work but the kernel is getting other ideas
> about your partitions you probably have both an x86 and mac partition
> table (its actually possible for them to coexist, but its a gross and
> uneeded hack).  what does /proc/partitions say?

But if the Mac partition table were corrupt, wouldn't it also be
skewed under my 2.2.18 Kernel as well?

Can't make heads or tails of it (I don't know what the columsn mean)
but here it is:

major minor  #blocks  name

   8     0    8971292 sda
   8     1         31 sda1
   8     2         27 sda2
   8     3         37 sda3
   8     4        100 sda4
   8     5        256 sda5
   8     6        256 sda6
   8     7    2661427 sda7
   8     8    5714420 sda8
   8     9     594722 sda9
   8    10          9 sda10
   8    16    4194157 sdb
   8    17         31 sdb1
   8    18         64 sdb2
   8    19    2000000 sdb3
   8    20    2194040 sdb4
   8    21         21 sdb5


--
patrick m pritchard
disposable hero.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: More problems with Kernel 2.4.10-pre12 and mounting

[Ethan Benson]

[-- Attachment #1: Type: text/plain, Size: 1436 bytes --]

On Sun, Sep 23, 2001 at 11:09:24PM -0400, patrick wrote:
> On Sun, 23 Sep 2001, Ethan Benson wrote:
> 
> > is it possible you have an x86 BIOS partition table on that disk as
> > well?  if mac-fdisk -l doesn't work then your mac partition table is
> > corrupt.  but if that does work but the kernel is getting other ideas
> > about your partitions you probably have both an x86 and mac partition
> > table (its actually possible for them to coexist, but its a gross and
> > uneeded hack).  what does /proc/partitions say?
> 
> But if the Mac partition table were corrupt, wouldn't it also be
> skewed under my 2.2.18 Kernel as well?  

yes most likely.

>    8    16    4194157 sdb
>    8    17         31 sdb1
>    8    18         64 sdb2
>    8    19    2000000 sdb3
>    8    20    2194040 sdb4
>    8    21         21 sdb5

the kernel is detecting your partition table just fine it seems.  the
8 is the major number of the device, 16, 17... are the minor numbers,
they correlate with the major/minors of the /dev/ nodes.  the third is
the number of 1k blocks that device has, sdb1 is clearly the mac
partition table as they are almost always 31.5k

the dd trick shows you can read the device without problems, are you
sure the filesystem is not corrupt? or that the disk is not failing?

try:

badblocks -sv /dev/sdb 

and see what happens.

-- 
Ethan Benson
http://www.alaska.net/~erbenson/

[-- Attachment #2: Type: application/pgp-signature, Size: 0 bytes --]

Re: More problems with Kernel 2.4.10-pre12 and mounting

[patrick]

On Sun, 23 Sep 2001, Ethan Benson wrote:

> the dd trick shows you can read the device without problems, are you
> sure the filesystem is not corrupt? or that the disk is not failing?

But the problem isn't limited to /dev/sdb, since /dev/sda also has
problems.  (I can't do any of the tricks, i.e., dd or pdisk -l /dev/sda
with it either.  Again, this is all under 2.4.10-pre12, 2.2.18 works
fine.)

> badblocks -sv /dev/sdb

Nothing bad happened, and no error messages came out (I checked
the entire /dev/sdb drive), so I'm assuming my drive is ok.

--
patrick m pritchard
disposable hero.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: More problems with Kernel 2.4.10-pre12 and mounting

[Ethan Benson]

[-- Attachment #1: Type: text/plain, Size: 921 bytes --]

On Mon, Sep 24, 2001 at 12:22:22AM -0400, patrick wrote:
> 
> On Sun, 23 Sep 2001, Ethan Benson wrote:
> 
> > the dd trick shows you can read the device without problems, are you
> > sure the filesystem is not corrupt? or that the disk is not failing?
> 
> But the problem isn't limited to /dev/sdb, since /dev/sda also has
> problems.  (I can't do any of the tricks, i.e., dd or pdisk -l /dev/sda
> with it either.  Again, this is all under 2.4.10-pre12, 2.2.18 works
> fine.)

well sounds to me like your scsi driver is hosed then.  
 
> > badblocks -sv /dev/sdb
> 
> Nothing bad happened, and no error messages came out (I checked
> the entire /dev/sdb drive), so I'm assuming my drive is ok.

well if badblocks has no problem reading the entire device why should
anything else.  this clearly has nothing to do with partition table
support.

-- 
Ethan Benson
http://www.alaska.net/~erbenson/

[-- Attachment #2: Type: application/pgp-signature, Size: 0 bytes --]

Re: More problems with Kernel 2.4.10-pre12 and mounting

[patrick]

On Sun, 23 Sep 2001, Ethan Benson wrote:

> well sounds to me like your scsi driver is hosed then.

Well if the SCSI driver is hosed, then shouldn't I *not* be able to do
anything else then?  My setup is as follows:

SCSI0 - Adaptec 2940 UW Card   /dev/sda
SCSI1 - Mesh Internal SCSI     /dev/sdb

And all of the drivers have been compiled into the kernel (no modules).

> well if badblocks has no problem reading the entire device why should
> anything else.  this clearly has nothing to do with partition table
> support.

Then that leaves it to the program I'm using to read the partition map.
I've seen you reference "mac-fdisk".  Is this the same as pdisk?  Or
is an entirely different beast?

Perhaps it is pdisk that doesn't like the new kernel.

patrick@tendim.cjb.net
http://www.tendim.cjb.net/
Like anime?   Got hotline?
hotline://twoaday.cjb.net/


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: More problems with Kernel 2.4.10-pre12 and mounting

[Ethan Benson]

[-- Attachment #1: Type: text/plain, Size: 1148 bytes --]

On Mon, Sep 24, 2001 at 05:39:59PM -0400, patrick wrote:
> On Sun, 23 Sep 2001, Ethan Benson wrote:
> 
> > well sounds to me like your scsi driver is hosed then.
> 
> Well if the SCSI driver is hosed, then shouldn't I *not* be able to do
> anything else then?  My setup is as follows:
> 
> SCSI0 - Adaptec 2940 UW Card   /dev/sda
> SCSI1 - Mesh Internal SCSI     /dev/sdb
> 
> And all of the drivers have been compiled into the kernel (no modules).
> 
> > well if badblocks has no problem reading the entire device why should
> > anything else.  this clearly has nothing to do with partition table
> > support.
> 
> Then that leaves it to the program I'm using to read the partition map.
> I've seen you reference "mac-fdisk".  Is this the same as pdisk?  Or
> is an entirely different beast?

its a debian modified version of pdisk thats more reliable.  get the
one from debian woody.

> Perhaps it is pdisk that doesn't like the new kernel.

i don't really see how.. its not doing much more then reading the
first 32k of of the disk to parse the partition table.

-- 
Ethan Benson
http://www.alaska.net/~erbenson/

[-- Attachment #2: Type: application/pgp-signature, Size: 0 bytes --]

Re: More problems with Kernel 2.4.10-pre12 and mounting

[Timothy A. Seufert]

At 3:04 PM -0800 9/24/01, Ethan Benson wrote:

>>  Perhaps it is pdisk that doesn't like the new kernel.
>
>i don't really see how.. its not doing much more then reading the
>first 32k of of the disk to parse the partition table.

Actually, I think the LinuxPPC 2000 Q4 version of pdisk does have a
problem with reading block 0 on 2.4.x kernels.  I remember having
that problem.  Doesn't happen with the pdisk from YDL 2.0.
--
Tim Seufert

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: More problems with Kernel 2.4.10-pre12 and mounting (SOLVED!)

[patrick]

On Mon, 24 Sep 2001, Ethan Benson wrote:

> its a debian modified version of pdisk thats more reliable.  get the
> one from debian woody.

Couldn't find it.

> > Perhaps it is pdisk that doesn't like the new kernel.
>
> i don't really see how.. its not doing much more then reading the
> first 32k of of the disk to parse the partition table.

Well, I feel stupid.

As Timothy verified, LinuxPPCQ4 pdisk doesn't like 2.4.x kernels, so I
upgraded to a new pdisk found at YellowdogLinux.com.  That let me see my
drives at least.

But I still couldn't mount.  I figured if pdisk was broken, mount might
be, so I upgraded to a newer version of mount from linuxppc.org.

*This* version gave me error messages I could read!  More importantly:

mount: filesystem hfs not supported by kernel

It would seem I turned off the experimental prompts in make menuconfig,
forgot about Apple HFS File System support, and blindly assumed that Apple
Partition Map support implied Apple HFS support.

D'oh.

Thanks for everyone's patience!


--
patrick m pritchard
disposable hero.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/