grub-probe fails to find PC partition due to Apple disklabel

grub-probe fails to find PC partition due to Apple disklabel

[Chris Knadle]

Greetings.

   I've run into an interesting problem on a PC running Debian Sid (Linux) 
where grub-probe fails to find partitions on the first hard disk because it 
finds an Apple disklabel, causing the 'update-grub' program to fail and thus 
not allow installing a new Linux kernel.  [The drive may at one time been in 
an Apple or in an external drive enclosure used by both Apple and PC's 
running Linux.]  The latest grub on Debian Sid now uses grub2 code for the 
grub-common package, so even though the bug report is for grub 0.97-36, it's 
using grub2 code.

      http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=475718

   As far as I can tell, (and forgive me if my terminology is slightly off), 
this boils down to the question of how to handle the situation where a disk 
has multiple disklabels / partition maps from different architectures but 
only has allocated partitions in one of the partition maps.

   Is there, or can we think of, a good way of handling this?

   -- Chris

-- 

Chris Knadle
Chris.Knadle@coredump.us

Re: grub-probe fails to find PC partition due to Apple disklabel

[Robert Millan]

On Sun, Apr 13, 2008 at 05:34:50PM -0400, Chris Knadle wrote:
> Greetings.
> 
>    I've run into an interesting problem on a PC running Debian Sid (Linux) 
> where grub-probe fails to find partitions on the first hard disk because it 
> finds an Apple disklabel, causing the 'update-grub' program to fail and thus 
> not allow installing a new Linux kernel.  [The drive may at one time been in 
> an Apple or in an external drive enclosure used by both Apple and PC's 
> running Linux.]  The latest grub on Debian Sid now uses grub2 code for the 
> grub-common package, so even though the bug report is for grub 0.97-36, it's 
> using grub2 code.
> 
>       http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=475718
> 
>    As far as I can tell, (and forgive me if my terminology is slightly off), 
> this boils down to the question of how to handle the situation where a disk 
> has multiple disklabels / partition maps from different architectures but 
> only has allocated partitions in one of the partition maps.
> 
>    Is there, or can we think of, a good way of handling this?

Hi Chris,

Initially, all partmaps were arch-specific, so that "pc" arch had msdos
and gpt partmaps, etc.  This avoided the problem;  so maybe what we could do
is probe for "native" partmaps first, and then for all the rest?  I can think
of two ways of doing this:

  - Explicitly probe for native partmaps for each arch (#ifdef MACHINE_BIOS,
    etc), then probe for all.  This is ugly and also inefficient (some partmaps
    are probed twice).
  - Change the partmap order in partmap.lst (can be complicated because of the
    build system), and make the prober follow the order in that file.  But, uhm,
    how would we access the file in the first place, before partmaps are probed?
    How does it work _now_ actually?

another way would be to add sanity checks in the prober, so that a partmap is
never identified as "good" when one of its partitions are out of bounds or so.
But then again:

  - Failing completely for an otherwise perfectly-sane partmap just because one
    partition is out of bounds sounds like a bad idea.
  - It doesn't completely fix the problem anyway.

Comments?

-- 
Robert Millan

<GPLv2> I know my rights; I want my phone call!
<DRM> What use is a phone call… if you are unable to speak?
(as seen on /.)

Re: grub-probe fails to find PC partition due to Apple disklabel

[Chris Knadle]

On Monday 14 April 2008, Robert Millan wrote:
> Hi Chris,
>
> Initially, all partmaps were arch-specific, so that "pc" arch had msdos
> and gpt partmaps, etc.  This avoided the problem;  so maybe what we could
> do is probe for "native" partmaps first, and then for all the rest?  I can
> think of two ways of doing this:
>
>   - Explicitly probe for native partmaps for each arch (#ifdef
> MACHINE_BIOS, etc), then probe for all.  This is ugly and also inefficient
> (some partmaps are probed twice).
>   - Change the partmap order in partmap.lst (can be complicated because of
> the build system), and make the prober follow the order in that file.  But,
> uhm, how would we access the file in the first place, before partmaps are
> probed? How does it work _now_ actually?
>
> another way would be to add sanity checks in the prober, so that a partmap
> is never identified as "good" when one of its partitions are out of bounds
> or so. But then again:
>
>   - Failing completely for an otherwise perfectly-sane partmap just because
> one partition is out of bounds sounds like a bad idea.
>   - It doesn't completely fix the problem anyway.
>
> Comments?

   The thought I had on this is that the search order for particular partition
maps probably needs to be arbitrary; i.e. there probably isn't a particular
search order that would cover this type of situation for all architectures
that grub2 functions on.  And as you mentioned, there probably isn't a good
way of specifying that search order before grub2 is installed.

   Going back to probing for native partmaps first...  hmm.  Is there any case
where that may create the opposite problem?  i.e. would it be possible for a
disk in a portable drive that's been used for both an Apple and an IBM PC but
which is actually currently partitioned for an Apple may be falsely detected
as being set up for an IBM PC?  I don't know.
   If there was a design decision to move away from doing this then there was
likely a reason for doing so; perhaps there was a similar situation where a
more flexible design was required.  i.e. my first thought would be to look
into why the design decision was made because there may be use cases where the
current design is required.

   Instead of probing for partmaps in a particular order, it seems that
solving for this might best be done by probing for all of the possible
partmaps for the architecture and trying to figure out which one best fits. 
grub2's grub-probe is clearly already doing this; _however_, somehow the PC
partmap probe is subtly different depending on whether partmap/apple.c is
included in the search.

The debug output files --
with partmap/apple.c detection:
http://bugs.debian.org/cgi-bin/bugreport.cgi?msg=15;filename=grub-probe_0.97-36.txt;att=1;bug=475718
without partmap/apple.c detection:
http://bugs.debian.org/cgi-bin/bugreport.cgi?msg=23;filename=grub-probe_noapple.txt;att=1;bug=475718
       
Comparing between the two outputs:

with partmap/apple.c detection:
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
   partmap/pc.c:141: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
   partmap/pc.c:141: partition 1: flag 0x0, type 0x0, start 0x0, len 0x0
   partmap/pc.c:141: partition 2: flag 0x0, type 0x0, start 0x0, len 0x0
   partmap/pc.c:141: partition 3: flag 0x0, type 0x0, start 0x0, len 0x0

without partmap/apple.c detection:
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
   partmap/pc.c:141: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
   partmap/pc.c:141: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
   partmap/pc.c:141: partition 2: flag 0x0, type 0x83, start 0x4c92c06, len 0x274c3c5
   partmap/pc.c:141: partition 3: flag 0x0, type 0x5, start 0x73defcb, len 0xb639af6


Look at the detection of partition 0 between the two; both the type and the
len numbers are different.  So I'm wondering, why would pc partmap detection
be different if the apple partmap was searched first?

   -- Chris

-- 

Chris Knadle
Chris.Knadle@coredump.us

Re: grub-probe fails to find PC partition due to Apple disklabel

[Pavel Roskin]

On Mon, 2008-04-14 at 10:21 -0400, Chris Knadle wrote:

>    Instead of probing for partmaps in a particular order, it seems that
> solving for this might best be done by probing for all of the possible
> partmaps for the architecture and trying to figure out which one best fits. 
> grub2's grub-probe is clearly already doing this; _however_, somehow the PC
> partmap probe is subtly different depending on whether partmap/apple.c is
> included in the search.

Apple and PC partition maps can coexist.  The place occupied by the PC
partition and the PC magic is not used by the Apple partition.

The way to differentiate them would be to check at the first sector.  If
it starts with 0x45 0x52, then it's an Apple partition.  A PC partition
would have an executable entry there.  The Apple magic would decode to:

inc    %bp
push   %dx

It's highly unlikely to be used in the first two instructions of the PC
boot code.

partmap/apple.c starts reading with the second sector (unsigned pos =
GRUB_DISK_SECTOR_SIZE).  Perhaps it should read the first sector to
check the magic.

-- 
Regards,
Pavel Roskin

Re: grub-probe fails to find PC partition due to Apple disklabel

[Robert Millan]

On Mon, Apr 14, 2008 at 01:20:49PM -0400, Pavel Roskin wrote:
> On Mon, 2008-04-14 at 10:21 -0400, Chris Knadle wrote:
> 
> >    Instead of probing for partmaps in a particular order, it seems that
> > solving for this might best be done by probing for all of the possible
> > partmaps for the architecture and trying to figure out which one best fits. 
> > grub2's grub-probe is clearly already doing this; _however_, somehow the PC
> > partmap probe is subtly different depending on whether partmap/apple.c is
> > included in the search.
> 
> Apple and PC partition maps can coexist.  The place occupied by the PC
> partition and the PC magic is not used by the Apple partition.
> 
> The way to differentiate them would be to check at the first sector.  If
> it starts with 0x45 0x52, then it's an Apple partition.  A PC partition
> would have an executable entry there.  The Apple magic would decode to:
> 
> inc    %bp
> push   %dx
> 
> It's highly unlikely to be used in the first two instructions of the PC
> boot code.
> 
> partmap/apple.c starts reading with the second sector (unsigned pos =
> GRUB_DISK_SECTOR_SIZE).  Perhaps it should read the first sector to
> check the magic.

Are you sure checking for the code part would be a good idea?  One could be
using i8088 bootstrap code in an Apple first sector, or vice-versa.

Btw, how does Linux handle this?  Maybe this issue has been beaten to death
already.

-- 
Robert Millan

<GPLv2> I know my rights; I want my phone call!
<DRM> What use is a phone call… if you are unable to speak?
(as seen on /.)

Re: grub-probe fails to find PC partition due to Apple disklabel

[Pavel Roskin]

On Tue, 2008-04-15 at 15:15 +0200, Robert Millan wrote:

> > partmap/apple.c starts reading with the second sector (unsigned pos =
> > GRUB_DISK_SECTOR_SIZE).  Perhaps it should read the first sector to
> > check the magic.
> 
> Are you sure checking for the code part would be a good idea?  One could be
> using i8088 bootstrap code in an Apple first sector, or vice-versa.

Apple partition table requires the magic in the first two bytes of the
first sector, followed by the sector size.  Even if those instructions
are survivable by x86 processors, executing data is a very bad hack.

> Btw, how does Linux handle this?  Maybe this issue has been beaten to death
> already.

fs/partitions/mac.c checks the magic in the first sector in
mac_partition()

fs/partitions/msdos.c doesn't check for the Apple magic, but it makes
some other sanity checks.  It rejects the partition table if any boot
flags is not 0x00 or 0x80.  It also refuses to accept the table if there
is an EFI GPT partition.

To fix the issue at hand (PC partition mistaken as Apple), all we need
is to check the Apple magic in the Apple partition code.

-- 
Regards,
Pavel Roskin

Re: grub-probe fails to find PC partition due to Apple disklabel

[Chris Knadle]

On Monday 14 April 2008, Pavel Roskin wrote:
> On Mon, 2008-04-14 at 10:21 -0400, Chris Knadle wrote:
> >    Instead of probing for partmaps in a particular order, it seems that
> > solving for this might best be done by probing for all of the possible
> > partmaps for the architecture and trying to figure out which one best
> > fits. grub2's grub-probe is clearly already doing this; _however_,
> > somehow the PC partmap probe is subtly different depending on whether
> > partmap/apple.c is included in the search.
>
> Apple and PC partition maps can coexist.  The place occupied by the PC
> partition and the PC magic is not used by the Apple partition.
>
> The way to differentiate them would be to check at the first sector.  If
> it starts with 0x45 0x52, then it's an Apple partition.  A PC partition
> would have an executable entry there.

   I can verify that this disk does not have 0x45 0x52 in the first sector.  
To check I used 'hexedit /dev/hde' as root, and I've copy-pasted the relevant 
text of the first 2120 bytes.

   Rather than attach it (some mailing lists don't appreciate attachments -- 
let me know if you'd prefer an attachment), I've placed the above file at the 
link below if you want to have a look.  Regardless of the fact that there are 
bits of text like "Apple", "Apple_partition_map", "Apple_Free", "Apple_HFS", 
etc, I assure you that this disk is not in an Apple and has been in my 
Desktop IBM PC for six months as the main boot disk.

   I regularly build and install new Linux kernels on this disk, and grub + 
update-grub handled all of those very well up until this point, which in 
itself is amazing given that the first couple of sectors seem to be full of 
false information as far as reading it as a human being goes.  ;-)

      ftp://ftp.coredump.us/grub-debug/Linux-apple-disk_2120bytes.txt

   -- Chris

-- 

Chris Knadle
Chris.Knadle@coredump.us

Re: grub-probe fails to find PC partition due to Apple disklabel

[Chris Knadle]

Felix Zielcke asked me about the Apple/PC hybrid disk problem, so I'm trying 
to figure out how to fix it.  Sorry for letting this fall through the cracks; 
this is from three months ago.

On Monday 14 April 2008, Pavel Roskin wrote:
> On Mon, 2008-04-14 at 10:21 -0400, Chris Knadle wrote:
> >    Instead of probing for partmaps in a particular order, it seems that
> > solving for this might best be done by probing for all of the possible
> > partmaps for the architecture and trying to figure out which one best
> > fits. grub2's grub-probe is clearly already doing this; _however_,
> > somehow the PC partmap probe is subtly different depending on whether
> > partmap/apple.c is included in the search.
>
> Apple and PC partition maps can coexist.  The place occupied by the
> PC partition and the PC magic is not used by the Apple partition. 

   Yeah, I see -- this includes the Apple magic of 0xAA55 that's placed in 
Sector 0,  so searching for that won't help.
 
> The way to differentiate them would be to check at the first sector.  If
> it starts with 0x45 0x52, then it's an Apple partition.  A PC partition
> would have an executable entry there.  The Apple magic would decode to:
>
> inc    %bp
> push   %dx
>
> It's highly unlikely to be used in the first two instructions of the PC
> boot code.
>
> partmap/apple.c starts reading with the second sector (unsigned pos =
> GRUB_DISK_SECTOR_SIZE).  Perhaps it should read the first sector to
> check the magic.

   I see where you got 0x4552 from; it's the HFS+ magic descriptor.  For the 
moment I'm using the following #define for that:
   #define GRUB_APPLE_FIRST_SECTOR_MAGIC 0x4552


   I'm trying to figure out how to get grub to check the first sector.  Mainly 
I'm asking for hints.  I think it involves running a new instance of 
grub_disk_read for the first sector (sector 0 of the partition) within the 
apple_partition_map_iterate function -- something like:

 grub_disk_read(&raw, 0, 0, sizeof (struct grub_apple_part), (char *)&apart))

   But then I want to check the first block against 0x4552, and 
apart.first_phys_block is 32-bits.  Since I want the first (high) 16 bits, I 
think I need to make the comparison in a way such as:

   if ((apart.first_phys_block & 0xFFFF0000) == (0x4552 << 16))
   

   Is this vaguely on the right track?

   Thanks.
   -- Chris

-- 

Chris Knadle
Chris.Knadle@coredump.us

Re: grub-probe fails to find PC partition due to Apple disklabel

[Pavel Roskin]

Quoting Chris Knadle <Chris.Knadle@coredump.us>:

>    I'm trying to figure out how to get grub to check the first   
> sector.  Mainly
> I'm asking for hints.  I think it involves running a new instance of
> grub_disk_read for the first sector (sector 0 of the partition) within the
> apple_partition_map_iterate function -- something like:
>
>  grub_disk_read(&raw, 0, 0, sizeof (struct grub_apple_part), (char *)&apart))

I'm not sure grub_apple_part would describe the first sector  
correctly.  Besides, you only need two bytes.  Why reuse "apart"?  You  
can use a 16-bit variable and read those two bytes there.

>    But then I want to check the first block against 0x4552, and
> apart.first_phys_block is 32-bits.

Why first_phys_block?  The first field is 16 bit wide and is called "magic".

> Since I want the first (high) 16 bits, I
> think I need to make the comparison in a way such as:
>
>    if ((apart.first_phys_block & 0xFFFF0000) == (0x4552 << 16))

That won't work on little endian systems, including ordinary PC.  You  
are assuming big endian byte order here.  Instead, the data from the  
disk should be converted to CPU-endian from whatever it is and  
compared to the magic value.

-- 
Regards,
Pavel Roskin

Re: grub-probe fails to find PC partition due to Apple disklabel

[Chris Knadle]

Thanks for your pointsers, Pavel; they were all helpful.

On Tuesday 22 July 2008, Pavel Roskin wrote:
> Quoting Chris Knadle <Chris.Knadle@coredump.us>:
> > Since I want the first (high) 16 bits, I
> > think I need to make the comparison in a way such as:
> >
> >    if ((apart.first_phys_block & 0xFFFF0000) == (0x4552 << 16))
>
> That won't work on little endian systems, including ordinary PC.  You
> are assuming big endian byte order here.  Instead, the data from the
> disk should be converted to CPU-endian from whatever it is and
> compared to the magic value.

   Yes, I did suspect that problem; so this needs a #ifdef for CPU 
architecture and choosing the appropriate conversion with one of the 
functions in include/grub/type.h -- okay.

   The next problem is that I'm not able to figure out how to get partition.c 
to "give up" the Apple_partition_map that is detected, and to re-detect for 
others.  Right now part_map_iterate() detects an Apple partition based on 
Sector 0 [which is unavoidable, since Sector 0 is not going to have an HFS+ 
magic number], after which it becomes "fixated" on it.  Calling 
grub_partition_map_unregister() does lead to secceding in detection of a 
pc_partition_map, but then still isn't fully successful in a grub-probe.

   Below is a patch based on the git tree if you want to see where I'm 
currently at.

-- 

Chris Knadle
Chris.Knadle@coredump.us

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

diff --git a/partmap/apple.c b/partmap/apple.c
index 4e5c481..3786301 100644
--- a/partmap/apple.c
+++ b/partmap/apple.c
@@ -23,6 +23,7 @@
 #include <grub/partition.h>

 #define GRUB_APPLE_PART_MAGIC  0x504D
+#define GRUB_APPLE_HFSPLUS_MAGIC 0x4552

 struct grub_apple_part
 {
@@ -102,6 +103,7 @@ apple_partition_map_iterate (grub_disk_t disk,
   struct grub_disk raw;
   int partno = 0;
   unsigned pos = GRUB_DISK_SECTOR_SIZE;
+  grub_uint16_t magic;

   /* Enforce raw disk access.  */
   raw = *disk;
@@ -111,6 +113,21 @@ apple_partition_map_iterate (grub_disk_t disk,

   for (;;)
     {
+
+      /* load + check partition magic number */
+      if (grub_disk_read (&raw, 0, 0, sizeof(grub_uint16_t), (char *) 
&magic))
+        return grub_errno;
+      if (grub_be_to_cpu16(magic) != GRUB_APPLE_HFSPLUS_MAGIC  && partno > 
0 ) {
+         grub_dprintf("partition",
+            "non-Apple partition %d: bad magic (found 0x%x; wanted 0x%x\n",
+            partno, grub_be_to_cpu16 (magic),
+            GRUB_APPLE_HFSPLUS_MAGIC);
+         grub_errno = 1;
+         grub_partition_map_unregister(part.partmap);
+         return 1;
+      }
+
+
       if (grub_disk_read (&raw, pos / GRUB_DISK_SECTOR_SIZE,
                          pos % GRUB_DISK_SECTOR_SIZE,
                          sizeof (struct grub_apple_part),  (char *) &apart))

Re: grub-probe fails to find PC partition due to Apple disklabel

[Chris Knadle]

On Wednesday 23 July 2008, Chris Knadle wrote:
> Right now part_map_iterate() detects an Apple partition based on Sector 0
> [which is unavoidable, since Sector 0 is not going to have an HFS+ magic
> number]

   Sorry, I meant *partition 0*, not Sector 0...

   -- Chris

-- 

Chris Knadle
Chris.Knadle@coredump.us

Re: grub-probe fails to find PC partition due to Apple disklabel

[Robert Millan]

On Wed, Jul 23, 2008 at 04:47:42PM -0400, Chris Knadle wrote:
> On Wednesday 23 July 2008, Chris Knadle wrote:
> > Right now part_map_iterate() detects an Apple partition based on Sector 0
> > [which is unavoidable, since Sector 0 is not going to have an HFS+ magic
> > number]
> 
>    Sorry, I meant *partition 0*, not Sector 0...

Assuming the problem is the missing check for the other magic number in
sector 0 (of the whole disk), I just sent a patch in:

  http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=475718#70

which supposedly fixes this problem.  Please test!

-- 
Robert Millan

<GPLv2> I know my rights; I want my phone call!
<DRM> What good is a phone call… if you are unable to speak?
(as seen on /.)

Re: grub-probe fails to find PC partition due to Apple disklabel

[Chris Knadle]

On Friday 25 July 2008, Robert Millan wrote:
> On Wed, Jul 23, 2008 at 04:47:42PM -0400, Chris Knadle wrote:
> > On Wednesday 23 July 2008, Chris Knadle wrote:
> > > Right now part_map_iterate() detects an Apple partition based on Sector
> > > 0 [which is unavoidable, since Sector 0 is not going to have an HFS+
> > > magic number]
> >
> >    Sorry, I meant *partition 0*, not Sector 0...
>
> Assuming the problem is the missing check for the other magic number in
> sector 0 (of the whole disk), I just sent a patch in:
>
>   http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=475718#70
>
> which supposedly fixes this problem.  Please test!

   Just tested -- nope, grub-probe still can't find the device.  Have a look 
at the following snippit of a 'grub-probe -vv /dev/hde3' output to see the 
interesting behavior this generates -- note that the first two bytes on this 
disk /dev/hde (i.e. the first two bytes on the disk, which should be part of 
the JUMP VECTOR to Grub) are 0xFC31.  So this:

   kern/partition.c:106: Detecting apple_partition_map...
   kern/disk.c:368: Reading `hd0'...
   partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
   kern/partition.c:112: apple_partition_map detection failed.

means that this patch is trying to match the HFS magic number on the first 
sector of the drive, when that doesn't have a filesystem magic number at 
all.  :-(  Do you see the problem -- or am I wrong about this?  If I'm 
correct (and I really hope I'm not), it means that there's no good way of 
discarding Apple partition detection based solely on Sector 0 of the drive -- 
which was already suspected anyway, since Apple and PC partition maps can 
coexist.

I'm going to attach the full 'grub-probe' output and a dd of the first 1024 
bytes of /dev/hde to the Debian bug report.

  -- Chris

-- 

Chris Knadle
Chris.Knadle@coredump.us

Re: grub-probe fails to find PC partition due to Apple disklabel

[Felix Zielcke]

Am Freitag, den 25.07.2008, 19:09 -0400 schrieb Chris Knadle:


>    Just tested -- nope, grub-probe still can't find the device.

> So this:
> 
>    kern/partition.c:106: Detecting apple_partition_map...
>    kern/disk.c:368: Reading `hd0'...
>    partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
>    kern/partition.c:112: apple_partition_map detection failed.
> 
> means that this patch is trying to match the HFS magic number on the first 
> sector of the drive, when that doesn't have a filesystem magic number at 
> all.

kern/partition.c:117: pc_partition_map detection succeeded.

Just after the snippet you wrote.
So I think it should work.
Maybe you didn't test this right?
Wipe a disk, then make an apple partition, and then make msdos
partitions and make a filesystem on it.
Just like it probable was as you reported the bug initially.

Btw. I just read the whole story again and I was wrong saying that
grub-legacy won't have a fix for this.
grub-legacy now uses grub-probe so if grub-probe works then both are
fixed for you.

Re: grub-probe fails to find PC partition due to Apple disklabel

[Felix Zielcke]

Am Sonntag, den 27.07.2008, 21:53 +0200 schrieb Felix Zielcke:

> Wipe a disk, then make an apple partition, and then make msdos
> partitions and make a filesystem on it.
> Just like it probable was as you reported the bug initially.
> 

Urm I forgot:
probable better if you make also an apple filesystem on it
and check your device.map that it's right and has an try for the /dev/
you're testing with.
grub-mkdevicemap makes a new one if you don't want to run grub-install
--recheck
The best would be a real test, not something constructed
So grabbing a harddisk used on an apple and then overwriting it with a
msdos partition scheme and linux filesystem and having linux installed
on it
So basicly as I already said in my mail before:
Almost the same situation as you had as you reported the bug :)

Re: grub-probe fails to find PC partition due to Apple disklabel

[Chris Knadle]

[-- Attachment #1: Type: text/plain, Size: 2333 bytes --]

On Sunday 27 July 2008, Felix Zielcke wrote:
> Am Freitag, den 25.07.2008, 19:09 -0400 schrieb Chris Knadle:
> >    Just tested -- nope, grub-probe still can't find the device.
> >
> > So this:
> >
> >    kern/partition.c:106: Detecting apple_partition_map...
> >    kern/disk.c:368: Reading `hd0'...
> >    partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
> >    kern/partition.c:112: apple_partition_map detection failed.
> >
> > means that this patch is trying to match the HFS magic number on the
> > first sector of the drive, when that doesn't have a filesystem magic
> > number at all.
>
> kern/partition.c:117: pc_partition_map detection succeeded.
>
> Just after the snippet you wrote.
> So I think it should work.
> Maybe you didn't test this right?

   I went back and looked at the bug report again.  Early on Robert suggested 
removing Apple partition detection, after which 'grub-probe -d /dev/hde2' 
succeeds (ext2).  [It's not in the bug report, but 'grub-probe -d /dev/hde3' 
also succeeds (xfs)].  So I'm satisfied that at least the overall testing 
method is fine.

   However on the last test I did make an error and left out the -d option to 
grub-probe, and that's required for the grub-probe to succeed, so 
unfortunately the output I posted is erroneous.
   A new one is attached which has Robert's patch applied -- it does succeed 
in finding the partition, but unfortunately we can't fix the problem this way 
because I believe it makes Apple partition detection *always* fail.

> Wipe a disk, then make an apple partition, and then make msdos
> partitions and make a filesystem on it.
> Just like it probable was as you reported the bug initially.

   Duplicating the problem is fine, but I don't see how this changes the 
testing method.  I'm probably willing to duplicate this on another system 
later, although independent verification would probably be better.  Right now 
I'm more interested in focusing on trying to remove Apple partmap detection 
for the PC architecture, since we already know that will fix the problem.

> Btw. I just read the whole story again and I was wrong saying that
> grub-legacy won't have a fix for this.
> grub-legacy now uses grub-probe so if grub-probe works then both are
> fixed for you.

   Right.

   -- Chris

-- 

Chris Knadle
Chris.Knadle@coredump.us

[-- Attachment #2: apple-patch_partmap-detect.txt --]
[-- Type: text/plain, Size: 23411 bytes --]

# ./grub-probe -vv -d /dev/hde3 2>&1

kern/disk.c:220: Opening `hd0'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0'...
grub-probe: info: opening the device `/dev/hde'
kern/disk.c:368: Reading `hd0'...
kern/disk.c:368: Reading `hd0'...
kern/disk.c:368: Reading `hd0'...
kern/disk.c:312: Closing `hd0'.
kern/disk.c:220: Opening `hd0'...
grub-probe: info: the size of hd0 is 312581808
kern/partition.c:106: Detecting gpt_partition_map...
kern/disk.c:368: Reading `hd0'...
kern/partition.c:112: gpt_partition_map detection failed.
kern/partition.c:106: Detecting apple_partition_map...
kern/disk.c:368: Reading `hd0'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/partition.c:112: apple_partition_map detection failed.
kern/partition.c:106: Detecting pc_partition_map...
kern/disk.c:368: Reading `hd0'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
kern/partition.c:117: pc_partition_map detection succeeded.
kern/disk.c:368: Reading `hd0'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
kern/disk.c:220: Opening `hd0,1'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0,1'...
kern/disk.c:368: Reading `hd0,1'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/disk.c:368: Reading `hd0,1'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
kern/disk.c:368: Reading `hd0,1'...
grub-probe: info: opening the device `/dev/hde1'
grub-probe: info: opening the device `/dev/hde1'
kern/disk.c:368: Reading `hd0,1'...
grub-probe: info: opening the device `/dev/hde1'
kern/disk.c:368: Reading `hd0,1'...
kern/disk.c:368: Reading `hd0,1'...
kern/disk.c:312: Closing `hd0,1'.
partmap/pc.c:143: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
kern/disk.c:220: Opening `hd0,2'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0,2'...
kern/disk.c:368: Reading `hd0,2'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/disk.c:368: Reading `hd0,2'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
partmap/pc.c:143: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
kern/disk.c:368: Reading `hd0,2'...
grub-probe: info: opening the device `/dev/hde2'
grub-probe: info: opening the device `/dev/hde2'
kern/disk.c:368: Reading `hd0,2'...
grub-probe: info: opening the device `/dev/hde2'
grub-probe: info: opening the device `/dev/hde2'
kern/disk.c:368: Reading `hd0,2'...
grub-probe: info: opening the device `/dev/hde2'
grub-probe: info: opening the device `/dev/hde2'
kern/disk.c:368: Reading `hd0,2'...
grub-probe: info: opening the device `/dev/hde2'
grub-probe: info: opening the device `/dev/hde2'
kern/disk.c:312: Closing `hd0,2'.
partmap/pc.c:143: partition 2: flag 0x0, type 0x83, start 0x4c92c06, len 0x274c3c5
kern/disk.c:220: Opening `hd0,3'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0,3'...
kern/disk.c:368: Reading `hd0,3'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/disk.c:368: Reading `hd0,3'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
partmap/pc.c:143: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
partmap/pc.c:143: partition 2: flag 0x0, type 0x83, start 0x4c92c06, len 0x274c3c5
kern/disk.c:368: Reading `hd0,3'...
grub-probe: info: opening the device `/dev/hde3'
grub-probe: info: opening the device `/dev/hde3'
kern/disk.c:368: Reading `hd0,3'...
grub-probe: info: opening the device `/dev/hde3'
grub-probe: info: opening the device `/dev/hde3'
kern/disk.c:368: Reading `hd0,3'...
grub-probe: info: opening the device `/dev/hde3'
kern/disk.c:368: Reading `hd0,3'...
kern/disk.c:312: Closing `hd0,3'.
partmap/pc.c:143: partition 3: flag 0x0, type 0x5, start 0x73defcb, len 0xb639af6
kern/disk.c:368: Reading `hd0'...
grub-probe: info: opening the device `/dev/hde'
partmap/pc.c:143: partition 0: flag 0x0, type 0x82, start 0x73df00a, len 0x686ceb
kern/disk.c:220: Opening `hd0,5'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0,5'...
kern/disk.c:368: Reading `hd0,5'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/disk.c:368: Reading `hd0,5'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
partmap/pc.c:143: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
partmap/pc.c:143: partition 2: flag 0x0, type 0x83, start 0x4c92c06, len 0x274c3c5
partmap/pc.c:143: partition 3: flag 0x0, type 0x5, start 0x73defcb, len 0xb639af6
kern/disk.c:368: Reading `hd0,5'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x82, start 0x73df00a, len 0x686ceb
kern/disk.c:368: Reading `hd0,5'...
grub-probe: info: opening the device `/dev/hde5'
grub-probe: info: opening the device `/dev/hde5'
kern/disk.c:368: Reading `hd0,5'...
grub-probe: info: opening the device `/dev/hde5'
grub-probe: info: opening the device `/dev/hde5'
kern/disk.c:368: Reading `hd0,5'...
grub-probe: info: opening the device `/dev/hde5'
grub-probe: info: opening the device `/dev/hde5'
kern/disk.c:368: Reading `hd0,5'...
grub-probe: info: opening the device `/dev/hde5'
grub-probe: info: opening the device `/dev/hde5'
kern/disk.c:312: Closing `hd0,5'.
partmap/pc.c:143: partition 1: flag 0x0, type 0x5, start 0x7a65cf5, len 0xafb2dcc
partmap/pc.c:143: partition 2: flag 0x0, type 0x0, start 0x73defcb, len 0x0
partmap/pc.c:143: partition 3: flag 0x0, type 0x0, start 0x73defcb, len 0x0
kern/disk.c:368: Reading `hd0'...
grub-probe: info: opening the device `/dev/hde'
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x7a65d34, len 0xafb2d8d
kern/disk.c:220: Opening `hd0,6'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0,6'...
kern/disk.c:368: Reading `hd0,6'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/disk.c:368: Reading `hd0,6'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
partmap/pc.c:143: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
partmap/pc.c:143: partition 2: flag 0x0, type 0x83, start 0x4c92c06, len 0x274c3c5
partmap/pc.c:143: partition 3: flag 0x0, type 0x5, start 0x73defcb, len 0xb639af6
kern/disk.c:368: Reading `hd0,6'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x82, start 0x73df00a, len 0x686ceb
partmap/pc.c:143: partition 1: flag 0x0, type 0x5, start 0x7a65cf5, len 0xafb2dcc
partmap/pc.c:143: partition 2: flag 0x0, type 0x0, start 0x73defcb, len 0x0
partmap/pc.c:143: partition 3: flag 0x0, type 0x0, start 0x73defcb, len 0x0
kern/disk.c:368: Reading `hd0,6'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x7a65d34, len 0xafb2d8d
kern/disk.c:368: Reading `hd0,6'...
grub-probe: info: opening the device `/dev/hde6'
grub-probe: info: opening the device `/dev/hde6'
kern/disk.c:368: Reading `hd0,6'...
grub-probe: info: opening the device `/dev/hde6'
grub-probe: info: opening the device `/dev/hde6'
kern/disk.c:368: Reading `hd0,6'...
grub-probe: info: opening the device `/dev/hde6'
grub-probe: info: opening the device `/dev/hde6'
kern/disk.c:368: Reading `hd0,6'...
grub-probe: info: opening the device `/dev/hde6'
grub-probe: info: opening the device `/dev/hde6'
kern/disk.c:312: Closing `hd0,6'.
partmap/pc.c:143: partition 1: flag 0x0, type 0x0, start 0x7a65cf5, len 0x0
partmap/pc.c:143: partition 2: flag 0x0, type 0x0, start 0x7a65cf5, len 0x0
partmap/pc.c:143: partition 3: flag 0x0, type 0x0, start 0x7a65cf5, len 0x0
kern/disk.c:312: Closing `hd0'.
kern/disk.c:220: Opening `hd1'...
grub-probe: info: the size of hd1 is 976773168
kern/disk.c:368: Reading `hd1'...
grub-probe: info: opening the device `/dev/hdg'
kern/disk.c:368: Reading `hd1'...
kern/disk.c:368: Reading `hd1'...
kern/disk.c:368: Reading `hd1'...
kern/disk.c:312: Closing `hd1'.
kern/disk.c:220: Opening `hd1'...
grub-probe: info: the size of hd1 is 976773168
kern/partition.c:106: Detecting gpt_partition_map...
kern/disk.c:368: Reading `hd1'...
kern/partition.c:112: gpt_partition_map detection failed.
kern/partition.c:106: Detecting apple_partition_map...
kern/disk.c:368: Reading `hd1'...
partmap/apple.c:129: bad magic (found 0x0; wanted 0x4552
kern/partition.c:112: apple_partition_map detection failed.
kern/partition.c:106: Detecting pc_partition_map...
kern/disk.c:368: Reading `hd1'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
kern/partition.c:117: pc_partition_map detection succeeded.
kern/disk.c:368: Reading `hd1'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
kern/disk.c:220: Opening `hd1,1'...
grub-probe: info: the size of hd1 is 976773168
kern/disk.c:368: Reading `hd1,1'...
kern/disk.c:368: Reading `hd1,1'...
partmap/apple.c:129: bad magic (found 0x0; wanted 0x4552
kern/disk.c:368: Reading `hd1,1'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
kern/disk.c:368: Reading `hd1,1'...
grub-probe: info: opening the device `/dev/hdg1'
grub-probe: info: opening the device `/dev/hdg1'
kern/disk.c:368: Reading `hd1,1'...
grub-probe: info: opening the device `/dev/hdg1'
kern/disk.c:368: Reading `hd1,1'...
kern/disk.c:368: Reading `hd1,1'...
kern/disk.c:312: Closing `hd1,1'.
partmap/pc.c:143: partition 1: flag 0x0, type 0x0, start 0x0, len 0x0
partmap/pc.c:143: partition 2: flag 0x0, type 0x0, start 0x0, len 0x0
partmap/pc.c:143: partition 3: flag 0x0, type 0x0, start 0x0, len 0x0
kern/disk.c:312: Closing `hd1'.
kern/disk.c:220: Opening `hd2'...
grub-probe: info: the size of hd2 is 976773168
kern/disk.c:368: Reading `hd2'...
grub-probe: info: opening the device `/dev/hdh'
kern/disk.c:368: Reading `hd2'...
kern/disk.c:368: Reading `hd2'...
kern/disk.c:368: Reading `hd2'...
kern/disk.c:312: Closing `hd2'.
kern/disk.c:220: Opening `hd2'...
grub-probe: info: the size of hd2 is 976773168
kern/partition.c:106: Detecting gpt_partition_map...
kern/disk.c:368: Reading `hd2'...
kern/partition.c:112: gpt_partition_map detection failed.
kern/partition.c:106: Detecting apple_partition_map...
kern/disk.c:368: Reading `hd2'...
partmap/apple.c:129: bad magic (found 0x0; wanted 0x4552
kern/partition.c:112: apple_partition_map detection failed.
kern/partition.c:106: Detecting pc_partition_map...
kern/disk.c:368: Reading `hd2'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
kern/partition.c:117: pc_partition_map detection succeeded.
kern/disk.c:368: Reading `hd2'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
kern/disk.c:220: Opening `hd2,1'...
grub-probe: info: the size of hd2 is 976773168
kern/disk.c:368: Reading `hd2,1'...
kern/disk.c:368: Reading `hd2,1'...
partmap/apple.c:129: bad magic (found 0x0; wanted 0x4552
kern/disk.c:368: Reading `hd2,1'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
kern/disk.c:368: Reading `hd2,1'...
grub-probe: info: opening the device `/dev/hdh1'
grub-probe: info: opening the device `/dev/hdh1'
kern/disk.c:368: Reading `hd2,1'...
grub-probe: info: opening the device `/dev/hdh1'
kern/disk.c:368: Reading `hd2,1'...
kern/disk.c:368: Reading `hd2,1'...
kern/disk.c:312: Closing `hd2,1'.
partmap/pc.c:143: partition 1: flag 0x0, type 0x0, start 0x0, len 0x0
partmap/pc.c:143: partition 2: flag 0x0, type 0x0, start 0x0, len 0x0
partmap/pc.c:143: partition 3: flag 0x0, type 0x0, start 0x0, len 0x0
kern/disk.c:312: Closing `hd2'.
disk/raid.c:361: Scanning for RAID devices
kern/disk.c:220: Opening `hd0'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0'...
grub-probe: info: opening the device `/dev/hde'
kern/disk.c:312: Closing `hd0'.
kern/disk.c:220: Opening `hd0'...
grub-probe: info: the size of hd0 is 312581808
kern/partition.c:106: Detecting gpt_partition_map...
kern/disk.c:368: Reading `hd0'...
kern/partition.c:112: gpt_partition_map detection failed.
kern/partition.c:106: Detecting apple_partition_map...
kern/disk.c:368: Reading `hd0'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/partition.c:112: apple_partition_map detection failed.
kern/partition.c:106: Detecting pc_partition_map...
kern/disk.c:368: Reading `hd0'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
kern/partition.c:117: pc_partition_map detection succeeded.
kern/disk.c:368: Reading `hd0'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
disk/raid.c:361: Scanning for RAID devices
kern/disk.c:220: Opening `hd0,1'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0,1'...
kern/disk.c:368: Reading `hd0,1'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/disk.c:368: Reading `hd0,1'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
kern/disk.c:368: Reading `hd0,1'...
grub-probe: info: opening the device `/dev/hde1'
grub-probe: info: opening the device `/dev/hde1'
kern/disk.c:312: Closing `hd0,1'.
partmap/pc.c:143: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
disk/raid.c:361: Scanning for RAID devices
kern/disk.c:220: Opening `hd0,2'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0,2'...
kern/disk.c:368: Reading `hd0,2'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/disk.c:368: Reading `hd0,2'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
partmap/pc.c:143: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
kern/disk.c:368: Reading `hd0,2'...
grub-probe: info: opening the device `/dev/hde2'
grub-probe: info: opening the device `/dev/hde2'
kern/disk.c:312: Closing `hd0,2'.
partmap/pc.c:143: partition 2: flag 0x0, type 0x83, start 0x4c92c06, len 0x274c3c5
disk/raid.c:361: Scanning for RAID devices
kern/disk.c:220: Opening `hd0,3'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0,3'...
kern/disk.c:368: Reading `hd0,3'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/disk.c:368: Reading `hd0,3'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
partmap/pc.c:143: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
partmap/pc.c:143: partition 2: flag 0x0, type 0x83, start 0x4c92c06, len 0x274c3c5
kern/disk.c:368: Reading `hd0,3'...
grub-probe: info: opening the device `/dev/hde3'
grub-probe: info: opening the device `/dev/hde3'
kern/disk.c:312: Closing `hd0,3'.
partmap/pc.c:143: partition 3: flag 0x0, type 0x5, start 0x73defcb, len 0xb639af6
kern/disk.c:368: Reading `hd0'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x82, start 0x73df00a, len 0x686ceb
disk/raid.c:361: Scanning for RAID devices
kern/disk.c:220: Opening `hd0,5'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0,5'...
kern/disk.c:368: Reading `hd0,5'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/disk.c:368: Reading `hd0,5'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
partmap/pc.c:143: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
partmap/pc.c:143: partition 2: flag 0x0, type 0x83, start 0x4c92c06, len 0x274c3c5
partmap/pc.c:143: partition 3: flag 0x0, type 0x5, start 0x73defcb, len 0xb639af6
kern/disk.c:368: Reading `hd0,5'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x82, start 0x73df00a, len 0x686ceb
kern/disk.c:368: Reading `hd0,5'...
grub-probe: info: opening the device `/dev/hde5'
grub-probe: info: opening the device `/dev/hde5'
kern/disk.c:312: Closing `hd0,5'.
partmap/pc.c:143: partition 1: flag 0x0, type 0x5, start 0x7a65cf5, len 0xafb2dcc
partmap/pc.c:143: partition 2: flag 0x0, type 0x0, start 0x73defcb, len 0x0
partmap/pc.c:143: partition 3: flag 0x0, type 0x0, start 0x73defcb, len 0x0
kern/disk.c:368: Reading `hd0'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x7a65d34, len 0xafb2d8d
disk/raid.c:361: Scanning for RAID devices
kern/disk.c:220: Opening `hd0,6'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0,6'...
kern/disk.c:368: Reading `hd0,6'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/disk.c:368: Reading `hd0,6'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
partmap/pc.c:143: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
partmap/pc.c:143: partition 2: flag 0x0, type 0x83, start 0x4c92c06, len 0x274c3c5
partmap/pc.c:143: partition 3: flag 0x0, type 0x5, start 0x73defcb, len 0xb639af6
kern/disk.c:368: Reading `hd0,6'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x82, start 0x73df00a, len 0x686ceb
partmap/pc.c:143: partition 1: flag 0x0, type 0x5, start 0x7a65cf5, len 0xafb2dcc
partmap/pc.c:143: partition 2: flag 0x0, type 0x0, start 0x73defcb, len 0x0
partmap/pc.c:143: partition 3: flag 0x0, type 0x0, start 0x73defcb, len 0x0
kern/disk.c:368: Reading `hd0,6'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x7a65d34, len 0xafb2d8d
kern/disk.c:368: Reading `hd0,6'...
grub-probe: info: opening the device `/dev/hde6'
grub-probe: info: opening the device `/dev/hde6'
kern/disk.c:312: Closing `hd0,6'.
partmap/pc.c:143: partition 1: flag 0x0, type 0x0, start 0x7a65cf5, len 0x0
partmap/pc.c:143: partition 2: flag 0x0, type 0x0, start 0x7a65cf5, len 0x0
partmap/pc.c:143: partition 3: flag 0x0, type 0x0, start 0x7a65cf5, len 0x0
kern/disk.c:312: Closing `hd0'.
disk/raid.c:361: Scanning for RAID devices
kern/disk.c:220: Opening `hd1'...
grub-probe: info: the size of hd1 is 976773168
kern/disk.c:368: Reading `hd1'...
grub-probe: info: opening the device `/dev/hdg'
kern/disk.c:312: Closing `hd1'.
kern/disk.c:220: Opening `hd1'...
grub-probe: info: the size of hd1 is 976773168
kern/partition.c:106: Detecting gpt_partition_map...
kern/disk.c:368: Reading `hd1'...
kern/partition.c:112: gpt_partition_map detection failed.
kern/partition.c:106: Detecting apple_partition_map...
kern/disk.c:368: Reading `hd1'...
partmap/apple.c:129: bad magic (found 0x0; wanted 0x4552
kern/partition.c:112: apple_partition_map detection failed.
kern/partition.c:106: Detecting pc_partition_map...
kern/disk.c:368: Reading `hd1'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
kern/partition.c:117: pc_partition_map detection succeeded.
kern/disk.c:368: Reading `hd1'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
disk/raid.c:361: Scanning for RAID devices
kern/disk.c:220: Opening `hd1,1'...
grub-probe: info: the size of hd1 is 976773168
kern/disk.c:368: Reading `hd1,1'...
kern/disk.c:368: Reading `hd1,1'...
partmap/apple.c:129: bad magic (found 0x0; wanted 0x4552
kern/disk.c:368: Reading `hd1,1'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
kern/disk.c:368: Reading `hd1,1'...
grub-probe: info: opening the device `/dev/hdg1'
grub-probe: info: opening the device `/dev/hdg1'
kern/disk.c:312: Closing `hd1,1'.
partmap/pc.c:143: partition 1: flag 0x0, type 0x0, start 0x0, len 0x0
partmap/pc.c:143: partition 2: flag 0x0, type 0x0, start 0x0, len 0x0
partmap/pc.c:143: partition 3: flag 0x0, type 0x0, start 0x0, len 0x0
kern/disk.c:312: Closing `hd1'.
disk/raid.c:361: Scanning for RAID devices
kern/disk.c:220: Opening `hd2'...
grub-probe: info: the size of hd2 is 976773168
kern/disk.c:368: Reading `hd2'...
grub-probe: info: opening the device `/dev/hdh'
kern/disk.c:312: Closing `hd2'.
kern/disk.c:220: Opening `hd2'...
grub-probe: info: the size of hd2 is 976773168
kern/partition.c:106: Detecting gpt_partition_map...
kern/disk.c:368: Reading `hd2'...
kern/partition.c:112: gpt_partition_map detection failed.
kern/partition.c:106: Detecting apple_partition_map...
kern/disk.c:368: Reading `hd2'...
partmap/apple.c:129: bad magic (found 0x0; wanted 0x4552
kern/partition.c:112: apple_partition_map detection failed.
kern/partition.c:106: Detecting pc_partition_map...
kern/disk.c:368: Reading `hd2'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
kern/partition.c:117: pc_partition_map detection succeeded.
kern/disk.c:368: Reading `hd2'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
disk/raid.c:361: Scanning for RAID devices
kern/disk.c:220: Opening `hd2,1'...
grub-probe: info: the size of hd2 is 976773168
kern/disk.c:368: Reading `hd2,1'...
kern/disk.c:368: Reading `hd2,1'...
partmap/apple.c:129: bad magic (found 0x0; wanted 0x4552
kern/disk.c:368: Reading `hd2,1'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x83, start 0x3f, len 0x3a384c02
kern/disk.c:368: Reading `hd2,1'...
grub-probe: info: opening the device `/dev/hdh1'
grub-probe: info: opening the device `/dev/hdh1'
kern/disk.c:312: Closing `hd2,1'.
partmap/pc.c:143: partition 1: flag 0x0, type 0x0, start 0x0, len 0x0
partmap/pc.c:143: partition 2: flag 0x0, type 0x0, start 0x0, len 0x0
partmap/pc.c:143: partition 3: flag 0x0, type 0x0, start 0x0, len 0x0
kern/disk.c:312: Closing `hd2'.
grub-probe: info: /dev/hde3 starts from 80292870
grub-probe: info: opening the device hd0
kern/disk.c:220: Opening `hd0'...
grub-probe: info: the size of hd0 is 312581808
kern/partition.c:106: Detecting gpt_partition_map...
kern/disk.c:368: Reading `hd0'...
kern/partition.c:112: gpt_partition_map detection failed.
kern/partition.c:106: Detecting apple_partition_map...
kern/disk.c:368: Reading `hd0'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/partition.c:112: apple_partition_map detection failed.
kern/partition.c:106: Detecting pc_partition_map...
kern/disk.c:368: Reading `hd0'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
kern/partition.c:117: pc_partition_map detection succeeded.
kern/disk.c:368: Reading `hd0'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
grub-probe: info: DOS partition 0 starts from 63
partmap/pc.c:143: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
grub-probe: info: DOS partition 1 starts from 80084025
partmap/pc.c:143: partition 2: flag 0x0, type 0x83, start 0x4c92c06, len 0x274c3c5
grub-probe: info: DOS partition 2 starts from 80292870
grub-probe: info: opening hd0,3
kern/disk.c:220: Opening `hd0,3'...
grub-probe: info: the size of hd0 is 312581808
kern/disk.c:368: Reading `hd0,3'...
kern/disk.c:368: Reading `hd0,3'...
partmap/apple.c:129: bad magic (found 0xfc31; wanted 0x4552
kern/disk.c:368: Reading `hd0,3'...
partmap/pc.c:143: partition 0: flag 0x0, type 0x7, start 0x3f, len 0x4c5fbfa
partmap/pc.c:143: partition 1: flag 0x80, type 0x83, start 0x4c5fc39, len 0x32fcd
partmap/pc.c:143: partition 2: flag 0x0, type 0x83, start 0x4c92c06, len 0x274c3c5
kern/fs.c:83: Detecting afs...
kern/disk.c:368: Reading `hd0,3'...
kern/disk.c:368: Reading `hd0,3'...
kern/fs.c:89: afs detection failed.
kern/fs.c:83: Detecting xfs...
kern/disk.c:368: Reading `hd0,3'...
grub-probe: info: opening the device `/dev/hde3'
grub-probe: info: opening the device `/dev/hde3'
kern/disk.c:368: Reading `hd0,3'...
grub-probe: info: opening the device `/dev/hde3'
kern/disk.c:368: Reading `hd0,3'...
grub-probe: info: opening the device `/dev/hde3'
grub-probe: info: opening the device `/dev/hde3'
kern/disk.c:368: Reading `hd0,3'...
xfs
kern/disk.c:312: Closing `hd0,3'.

Re: grub-probe fails to find PC partition due to Apple disklabel

[Felix Zielcke]

Am Montag, den 28.07.2008, 14:41 -0400 schrieb Chris Knadle:

> 
>    However on the last test I did make an error and left out the -d option to 
> grub-probe, and that's required for the grub-probe to succeed, so 
> unfortunately the output I posted is erroneous.
>    A new one is attached which has Robert's patch applied -- it does succeed 
> in finding the partition, but unfortunately we can't fix the problem this way 
> because I believe it makes Apple partition detection *always* fail.
> 
Urgs, yeah didn't see that.
You have to use -d if you give a device.
Why do you think that with this it will always fail?
Maybe you can try this out? :)

Re: grub-probe fails to find PC partition due to Apple disklabel

[Chris Knadle]

On Monday 28 July 2008, Felix Zielcke wrote:
> Am Montag, den 28.07.2008, 14:41 -0400 schrieb Chris Knadle:
> >    However on the last test I did make an error and left out the -d
> > option to grub-probe, and that's required for the grub-probe to succeed,
> > so unfortunately the output I posted is erroneous.
> >    A new one is attached which has Robert's patch applied -- it does
> > succeed in finding the partition, but unfortunately we can't fix the
> > problem this way because I believe it makes Apple partition detection
> > *always* fail.
>
> Urgs, yeah didn't see that.
> You have to use -d if you give a device.
> Why do you think that with this it will always fail?

   Because the test that the patch does is to check for an HFS+ filesystem 
magic number against the first 2 bytes on the drive, which contains a jump 
vector for the Grub stage1 binary.  i.e. in order for the test to pass, Grub 
has to be unbootable.

> Maybe you can try this out? :)

   I normally don't run Apple machines.  I have an old 400MHz iMac G3 (the 
type that is housed inside of a monitor) available to me, but replacing the 
hard disk requires a somewhat messy disassembly effort.  I also have an 
external Firewire enclosure I could theoretically try this with, but I don't 
know how I'd get the G3 to attempt to boot on it.
   I'd much rather be looking at C code than spending the next two days 
futzing with an old Apple box.  :-/

   -- Chris

-- 

Chris Knadle
Chris.Knadle@coredump.us

Re: grub-probe fails to find PC partition due to Apple disklabel

[Robert Millan]

On Mon, Jul 28, 2008 at 03:25:00PM -0400, Chris Knadle wrote:
> On Monday 28 July 2008, Felix Zielcke wrote:
> > Am Montag, den 28.07.2008, 14:41 -0400 schrieb Chris Knadle:
> > >    However on the last test I did make an error and left out the -d
> > > option to grub-probe, and that's required for the grub-probe to succeed,
> > > so unfortunately the output I posted is erroneous.
> > >    A new one is attached which has Robert's patch applied -- it does
> > > succeed in finding the partition, but unfortunately we can't fix the
> > > problem this way because I believe it makes Apple partition detection
> > > *always* fail.
> >
> > Urgs, yeah didn't see that.
> > You have to use -d if you give a device.
> > Why do you think that with this it will always fail?
> 
>    Because the test that the patch does is to check for an HFS+ filesystem 
> magic number against the first 2 bytes on the drive, which contains a jump 
> vector for the Grub stage1 binary.  i.e. in order for the test to pass, Grub 
> has to be unbootable.

My check has nothing to do with HFS+, it's based on the header magic number
(0x4552), which differs from the partition magic (0x504D) but also from the
HFS+ magic (0x482B, 0x4858).

Please could you test it against an apple partmap?

-- 
Robert Millan

  The DRM opt-in fallacy: "Your data belongs to us. We will decide when (and
  how) you may access your data; but nobody's threatening your freedom: we
  still allow you to remove your data and not access it at all."

Re: grub-probe fails to find PC partition due to Apple disklabel

[Robert Millan]

On Mon, Jul 28, 2008 at 11:20:26PM +0200, Robert Millan wrote:
> On Mon, Jul 28, 2008 at 03:25:00PM -0400, Chris Knadle wrote:
> > On Monday 28 July 2008, Felix Zielcke wrote:
> > > Am Montag, den 28.07.2008, 14:41 -0400 schrieb Chris Knadle:
> > > >    However on the last test I did make an error and left out the -d
> > > > option to grub-probe, and that's required for the grub-probe to succeed,
> > > > so unfortunately the output I posted is erroneous.
> > > >    A new one is attached which has Robert's patch applied -- it does
> > > > succeed in finding the partition, but unfortunately we can't fix the
> > > > problem this way because I believe it makes Apple partition detection
> > > > *always* fail.
> > >
> > > Urgs, yeah didn't see that.
> > > You have to use -d if you give a device.
> > > Why do you think that with this it will always fail?
> > 
> >    Because the test that the patch does is to check for an HFS+ filesystem 
> > magic number against the first 2 bytes on the drive, which contains a jump 
> > vector for the Grub stage1 binary.  i.e. in order for the test to pass, Grub 
> > has to be unbootable.
> 
> My check has nothing to do with HFS+, it's based on the header magic number
> (0x4552), which differs from the partition magic (0x504D) but also from the
> HFS+ magic (0x482B, 0x4858).
> 
> Please could you test it against an apple partmap?

Actually, it was simple for me to test, since Parted supports apple/mac
partition tables.  The test I added succeeds with the partition I created,
so it must be good.

I just checked that in.  Thanks a lot for the long & tedious debugging process.

-- 
Robert Millan

  The DRM opt-in fallacy: "Your data belongs to us. We will decide when (and
  how) you may access your data; but nobody's threatening your freedom: we
  still allow you to remove your data and not access it at all."

Re: grub-probe fails to find PC partition due to Apple disklabel

[Chris Knadle]

On Monday 28 July 2008, Robert Millan wrote:
> On Mon, Jul 28, 2008 at 03:25:00PM -0400, Chris Knadle wrote:
> > On Monday 28 July 2008, Felix Zielcke wrote:
> > > Am Montag, den 28.07.2008, 14:41 -0400 schrieb Chris Knadle:
> > > >    However on the last test I did make an error and left out the -d
> > > > option to grub-probe, and that's required for the grub-probe to
> > > > succeed, so unfortunately the output I posted is erroneous.
> > > >    A new one is attached which has Robert's patch applied -- it does
> > > > succeed in finding the partition, but unfortunately we can't fix the
> > > > problem this way because I believe it makes Apple partition detection
> > > > *always* fail.
> > >
> > > Urgs, yeah didn't see that.
> > > You have to use -d if you give a device.
> > > Why do you think that with this it will always fail?
> >
> >    Because the test that the patch does is to check for an HFS+
> > filesystem magic number against the first 2 bytes on the drive, which
> > contains a jump vector for the Grub stage1 binary.  i.e. in order for the
> > test to pass, Grub has to be unbootable.
>
> My check has nothing to do with HFS+, it's based on the header magic number
> (0x4552), which differs from the partition magic (0x504D) but also from the
> HFS+ magic (0x482B, 0x4858).
>
> Please could you test it against an apple partmap?

   I just tested this on the iMac with a different HD and installing Debian 
Lenny Beta2 (since I couldn't find a download for Etch) -- took forever to do 
but turns out it was worth it -- your patch WORKS!  :-)  Congratulations.

   This also means I misunderstood several things.  I must have misread what 
exactly the 0x4552 magic number pertained to.  Ugh.

   Thanks very much for hanging in there and fixing this.

   -- Chris   

-- 

Chris Knadle
Chris.Knadle@coredump.us