GNU bug report logs - #61076
Change in MBR between parted 3.2 and 3.3

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 61076 in the body.
You can then email your comments to 61076 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Frédéric Martinsons <frederic.martinsons <at> gmail.com>
To: bug-parted <at> gnu.org
Subject: Change in MBR between parted 3.2 and 3.3
Date: Thu, 26 Jan 2023 11:52:36 +0100

Hello,

I have an arm based board with an eMMC card for storage.
I cross compil an GNU/Linux OS (with yocto) for this boardand and I came
across an issue after updating parted to 3.3 and above.

Below are my commands to partition the 4GB storage (2 partitions of 1.8GB,
1 bootable partition of 400MB) :

:~# parted /dev/mmcblk0 mklabel msdos
:~# parted -a none /dev/mmcblk0 unit s mkpart primary ext4 2048 3474431
:~# parted -a none /dev/mmcblk0 unit s mkpart primary ext4 3474432 6946815
:~# parted -a none /dev/mmcblk0 unit s mkpart primary ext4 6946816 7733247
:~# parted /dev/mmcblk0 set 3 boot on

With parted 3.2, I have the following MBR:

:~# hexdump -n 1024 -C /dev/mmcblk0
00000000  fa b8 00 10 8e d0 bc 00  b0 b8 00 00 8e d8 8e c0  |................|
00000010  fb be 00 7c bf 00 06 b9  00 02 f3 a4 ea 21 06 00  |...|.........!..|
00000020  00 be be 07 38 04 75 0b  83 c6 10 81 fe fe 07 75  |....8.u........u|
00000030  f3 eb 16 b4 02 b0 01 bb  00 7c b2 80 8a 74 01 8b  |.........|...t..|
00000040  4c 02 cd 13 ea 00 7c 00  00 eb fe 00 00 00 00 00  |L.....|.........|
00000050  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
*
000001b0  00 00 00 00 00 00 00 00  34 15 3b 6e 00 00 00 20  |........4.;n... |
000001c0  21 00 83 45 2d d8 00 08  00 00 00 fc 34 00 00 45  |!..E-.......4..E|
000001d0  2e d8 83 6a 7a b0 00 04  35 00 00 fc 34 00 80 6a  |...jz...5...4..j|
000001e0  7b b0 83 5e 7d e1 00 00  6a 00 00 00 0c 00 00 00  |{..^}...j.......|
000001f0  00 00 00 00 00 00 00 00  00 00 00 00 00 00 55 aa  |..............U.|
00000200  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
*
00000400

With parted 3.3, I have that:

:~# hexdump -n 1024 -C /dev/mmcblk0
00000000  fa b8 00 10 8e d0 bc 00  b0 b8 00 00 8e d8 8e c0  |................|
00000010  fb be 00 7c bf 00 06 b9  00 02 f3 a4 ea 21 06 00  |...|.........!..|
00000020  00 be be 07 38 04 75 0b  83 c6 10 81 fe fe 07 75  |....8.u........u|
00000030  f3 eb 16 b4 02 b0 01 bb  00 7c b2 80 8a 74 01 8b  |.........|...t..|
00000040  4c 02 cd 13 ea 00 7c 00  00 eb fe 00 00 00 00 00  |L.....|.........|
00000050  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
*
000001b0  00 00 00 00 00 00 00 00  85 1a 7c 02 00 00 00 04  |..........|.....|
000001c0  01 04 83 fe c2 ff 00 08  00 00 00 fc 34 00 00 fe  |............4...|
000001d0  c2 ff 83 fe c2 ff 00 04  35 00 00 fc 34 00 80 fe  |........5...4...|
000001e0  c2 ff 83 fe c2 ff 00 00  6a 00 00 00 0c 00 00 00  |........j.......|
000001f0  00 00 00 00 00 00 00 00  00 00 00 00 00 00 55 aa  |..............U.|
00000200  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
*
00000400

The first and last CHS address of the partition entries are completely different
and for the entry 2 and 3, first address is equal to the last !



Below is more information by parted 3.2:
:~# parted /dev/mmcblk0 print unit s print unit chs print
Model: MMC 004GA0 (sd/mmc)
Disk /dev/mmcblk0: 3959MB
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags:

Number  Start   End     Size    Type     File system  Flags
 1      1049kB  1779MB  1778MB  primary
 2      1779MB  3557MB  1778MB  primary
 3      3557MB  3959MB  403MB   primary  ext4         boot

Model: MMC 004GA0 (sd/mmc)
Disk /dev/mmcblk0: 7733248s
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags:

Number  Start     End       Size      Type     File system  Flags
 1      2048s     3474431s  3472384s  primary
 2      3474432s  6946815s  3472384s  primary
 3      6946816s  7733247s  786432s   primary  ext4         boot

Model: MMC 004GA0 (sd/mmc)
Disk /dev/mmcblk0: 481,94,60
Sector size (logical/physical): 512B/512B
BIOS cylinder,head,sector geometry: 481,255,63.  Each cylinder is 8225kB.
Partition Table: msdos
Disk Flags:

Number  Start       End         Type     File system  Flags
 1      0,32,32     216,69,44   primary
 2      216,69,45   432,106,57  primary
 3      432,106,58  481,94,60   primary  ext4         boot



And the same with parted 3.3:
:~# parted /dev/mmcblk0 print unit s print unit chs print
Model: MMC 004GA0 (sd/mmc)
Disk /dev/mmcblk0: 3959MB
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags:

Number  Start   End     Size    Type     File system  Flags
 1      1049kB  1779MB  1778MB  primary
 2      1779MB  3557MB  1778MB  primary
 3      3557MB  3959MB  403MB   primary  ext4         boot

Model: MMC 004GA0 (sd/mmc)
Disk /dev/mmcblk0: 7733248s
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags:

Number  Start     End       Size      Type     File system  Flags
 1      2048s     3474431s  3472384s  primary
 2      3474432s  6946815s  3472384s  primary
 3      6946816s  7733247s  786432s   primary  ext4         boot

Model: MMC 004GA0 (sd/mmc)
Disk /dev/mmcblk0: 15163,58,1
Sector size (logical/physical): 512B/512B
BIOS cylinder,head,sector geometry: 15163,255,2.  Each cylinder is 261kB.
Partition Table: msdos
Disk Flags:

Number  Start       End         Type     File system  Flags
 1      4,4,0       6812,155,1  primary
 2      6812,156,0  13621,52,1  primary
 3      13621,53,0  15163,58,1  primary  ext4         boot



The main difference I spotted is the BIOS geometry:
  - parted 3.2:   481 cylinder, 255 head, 63 sector (cylinder size 8225 kB)
  - parted 3.3: 15163 cylinder, 255 head,  2 sector (cylinder size  261 kB)

I also tested parted 3.4 and parted 3.5 with the same result.
Nevertheless, the partition table created by parted 3.3 and above is
perfectly usable from what I see.

Long story short, do you know the origin of this discrepancy (I didn't
see nothing
in release not that could explain that though I obviously don't understan all
the mechanics) and if it is possible to come back to the same kind of
MBR generated
by parted 3.2 ?
One additional question arises for my understanding though, how come a partition
with CHS address of the first sector equal to the last one is usable ?

Thanks in advance for all insights/advices you can bring to my issue.

Message #10 received at 61076-close <at> debbugs.gnu.org (full text, mbox):

From: "Brian C. Lane" <bcl <at> redhat.com>
To: Frédéric Martinsons <frederic.martinsons <at> gmail.com>
Cc: 61076-close <at> debbugs.gnu.org
Subject: Re: bug#61076: Change in MBR between parted 3.2 and 3.3
Date: Thu, 26 Jan 2023 08:28:18 -0800

On Thu, Jan 26, 2023 at 11:52:36AM +0100, Frédéric Martinsons wrote:
> Hello,
> 
> I have an arm based board with an eMMC card for storage.
> I cross compil an GNU/Linux OS (with yocto) for this boardand and I came
> across an issue after updating parted to 3.3 and above.

[snip]

> 
> The main difference I spotted is the BIOS geometry:
>   - parted 3.2:   481 cylinder, 255 head, 63 sector (cylinder size 8225 kB)
>   - parted 3.3: 15163 cylinder, 255 head,  2 sector (cylinder size  261 kB)
> 
> I also tested parted 3.4 and parted 3.5 with the same result.
> Nevertheless, the partition table created by parted 3.3 and above is
> perfectly usable from what I see.
> 
> Long story short, do you know the origin of this discrepancy (I didn't
> see nothing
> in release not that could explain that though I obviously don't understan all
> the mechanics) and if it is possible to come back to the same kind of
> MBR generated
> by parted 3.2 ?

This change was introduced by commit
61dd3d4c5eb782eb43caa95342e63727db3f8281, it was needed to fix problems
growing partitions when using SD cards on the Raspberry Pi.

> One additional question arises for my understanding though, how come a partition
> with CHS address of the first sector equal to the last one is usable ?

Well, nothing should be actually using the CHS values these days. So
it's possible that's a bug that doesn't matter in practice, but I'll
have to look into that.

Brian

-- 
Brian C. Lane (PST8PDT) - weldr.io - lorax - parted - pykickstart

Message #13 received at 61076 <at> debbugs.gnu.org (full text, mbox):

From: Frédéric Martinsons <frederic.martinsons <at> gmail.com>
To: 61076 <at> debbugs.gnu.org
Subject: Re: Change in MBR between parted 3.2 and 3.3
Date: Thu, 26 Jan 2023 17:39:24 +0100

> This change was introduced by commit
> 61dd3d4c5eb782eb43caa95342e63727db3f8281, it was needed to fix problems
> growing partitions when using SD cards on the Raspberry Pi.

OK , thank you very much for the explanation. Since I don't use
raspberry Pi among my targets, do you think I can safely run a custom
parted with a revert of 61dd3d4c5eb782eb43caa95342e63727db3f8281 ?

> Well, nothing should be actually using the CHS values these days. So
> it's possible that's a bug that doesn't matter in practice, but I'll
> have to look into that.

OK I understand, since you close the present bug, do you know where I
can have your conclusion when you'll have time to look at it ?

Thanks again for the quick answer by the way.

Message #16 received at 61076 <at> debbugs.gnu.org (full text, mbox):

From: Frédéric Martinsons <frederic.martinsons <at> gmail.com>
To: "Brian C. Lane" <bcl <at> redhat.com>, 61076 <at> debbugs.gnu.org
Subject: Re: bug#61076: Change in MBR between parted 3.2 and 3.3
Date: Fri, 27 Jan 2023 09:14:01 +0100

[Message part 1 (text/plain, inline)]

> > OK , thank you very much for the explanation. Since I don't use
> > raspberry Pi among my targets, do you think I can safely run a custom
> > parted with a revert of 61dd3d4c5eb782eb43caa95342e63727db3f8281 ?

> That's probably fine, as long as you know what you are doing ;)

Ok I'll go this way, thanks.

> > OK I understand, since you close the present bug, do you know where I
> > can have your conclusion when you'll have time to look at it ?
> >
> > Thanks again for the quick answer by the way.

> I'm *hoping* to get a new parted release ready sometime soon, but I'm
> not yet sure when. I have a few other things to review and I'll take a
> look at this at the same time to make sure it's not a bug.

Fine, I'll keep a look to the parted release then ;)
By the way, I come up with a patch (joined) which is a mix of revert
61dd3d4c5eb782eb43caa95342e63727db3f8281 and
taken into account 52360db2f5397b7842d2ed90bf946c5e8fa91750
(which mention some kind of regression too):

Have a nice day

On Thu, 26 Jan 2023 at 19:50, Brian C. Lane <bcl <at> redhat.com> wrote:
>
> On Thu, Jan 26, 2023 at 05:39:24PM +0100, Frédéric Martinsons wrote:
> > > This change was introduced by commit
> > > 61dd3d4c5eb782eb43caa95342e63727db3f8281, it was needed to fix problems
> > > growing partitions when using SD cards on the Raspberry Pi.
> >
> > OK , thank you very much for the explanation. Since I don't use
> > raspberry Pi among my targets, do you think I can safely run a custom
> > parted with a revert of 61dd3d4c5eb782eb43caa95342e63727db3f8281 ?
>
> That's probably fine, as long as you know what you are doing ;)
>
> >
> > > Well, nothing should be actually using the CHS values these days. So
> > > it's possible that's a bug that doesn't matter in practice, but I'll
> > > have to look into that.
> >
> > OK I understand, since you close the present bug, do you know where I
> > can have your conclusion when you'll have time to look at it ?
> >
> > Thanks again for the quick answer by the way.
>
> I'm *hoping* to get a new parted release ready sometime soon, but I'm
> not yet sure when. I have a few other things to review and I'll take a
> look at this at the same time to make sure it's not a bug.
>
> Brian
>
> --
> Brian C. Lane (PST8PDT) - weldr.io - lorax - parted - pykickstart
>

[0001-Get-back-the-old-way-of-getting-device-geometry-info.patch (text/x-patch, attachment)]

Message #19 received at 61076 <at> debbugs.gnu.org (full text, mbox):

From: Frédéric Martinsons <frederic.martinsons <at> gmail.com>
To: "Brian C. Lane" <bcl <at> redhat.com>, 61076 <at> debbugs.gnu.org
Subject: Re: bug#61076: Change in MBR between parted 3.2 and 3.3
Date: Fri, 27 Jan 2023 11:58:22 +0100

After more testing with the mentioned patch applied on parted 3.3, I
still have differences on the MBR
toward parted 3.2, below is what I got now:

:~# hexdump -n 1024 -C /dev/mmcblk0
00000000  fa b8 00 10 8e d0 bc 00  b0 b8 00 00 8e d8 8e c0  |................|
00000010  fb be 00 7c bf 00 06 b9  00 02 f3 a4 ea 21 06 00  |...|.........!..|
00000020  00 be be 07 38 04 75 0b  83 c6 10 81 fe fe 07 75  |....8.u........u|
00000030  f3 eb 16 b4 02 b0 01 bb  00 7c b2 80 8a 74 01 8b  |.........|...t..|
00000040  4c 02 cd 13 ea 00 7c 00  00 eb fe 00 00 00 00 00  |L.....|.........|
00000050  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
*
000001b0  00 00 00 00 00 00 00 00  11 95 8b 42 00 00 00 00  |...........B....|
000001c0  01 20 83 03 d0 ff 00 08  00 00 00 fc 34 00 00 03  |. ..........4...|
000001d0  d0 ff 83 03 d0 ff 00 04  35 00 00 fc 34 00 80 03  |........5...4...|
000001e0  d0 ff 83 03 d0 ff 00 00  6a 00 00 00 0c 00 00 00  |........j.......|
000001f0  00 00 00 00 00 00 00 00  00 00 00 00 00 00 55 aa  |..............U.|
00000200  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|

And the partition table info:
:~# parted /dev/mmcblk0 print unit s print unit chs print
Model: MMC 004GA0 (sd/mmc)
Disk /dev/mmcblk0: 3959MB
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags:

Number  Start   End     Size    Type     File system  Flags
 1      1049kB  1779MB  1778MB  primary
 2      1779MB  3557MB  1778MB  primary
 3      3557MB  3959MB  403MB   primary  ext4         boot

Model: MMC 004GA0 (sd/mmc)
Disk /dev/mmcblk0: 7733248s
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags:

Number  Start     End       Size      Type     File system  Flags
 1      2048s     3474431s  3472384s  primary
 2      3474432s  6946815s  3472384s  primary
 3      6946816s  7733247s  786432s   primary  ext4         boot

Model: MMC 004GA0 (sd/mmc)
Disk /dev/mmcblk0: 120831,3,15
Sector size (logical/physical): 512B/512B
BIOS cylinder,head,sector geometry: 120832,4,16.  Each cylinder is 32.8kB.
Partition Table: msdos
Disk Flags:

Number  Start       End          Type     File system  Flags
 1      32,0,0      54287,3,15   primary
 2      54288,0,0   108543,3,15  primary
 3      108544,0,0  120831,3,15  primary  ext4         boot


In parted 3.2, I had the following geometry:
BIOS cylinder,head,sector geometry: 481,255,63.  Each cylinder is 8225kB.

In 3.3 vanilla:
BIOS cylinder,head,sector geometry: 15163,255,2.  Each cylinder is 261kB.

In 3.3 patched:
BIOS cylinder,head,sector geometry: 120832,4,16.  Each cylinder is 32.8kB.

Like you said earlier "nothing should be actually using the CHS values
these days", and indeed
my system could be bootable with these changes, but my problem is that
I have a TPM
chip which check the content of the MBR and refuses to continue
booting process if the content
of it is not what is expected.

Do you see other changes that have been made in parted 3.3 which impact the MBR
content in such a way ?

On Fri, 27 Jan 2023 at 09:14, Frédéric Martinsons
<frederic.martinsons <at> gmail.com> wrote:
>
> > > OK , thank you very much for the explanation. Since I don't use
> > > raspberry Pi among my targets, do you think I can safely run a custom
> > > parted with a revert of 61dd3d4c5eb782eb43caa95342e63727db3f8281 ?
>
> > That's probably fine, as long as you know what you are doing ;)
>
> Ok I'll go this way, thanks.
>
> > > OK I understand, since you close the present bug, do you know where I
> > > can have your conclusion when you'll have time to look at it ?
> > >
> > > Thanks again for the quick answer by the way.
>
> > I'm *hoping* to get a new parted release ready sometime soon, but I'm
> > not yet sure when. I have a few other things to review and I'll take a
> > look at this at the same time to make sure it's not a bug.
>
> Fine, I'll keep a look to the parted release then ;)
> By the way, I come up with a patch (joined) which is a mix of revert
> 61dd3d4c5eb782eb43caa95342e63727db3f8281 and
> taken into account 52360db2f5397b7842d2ed90bf946c5e8fa91750
> (which mention some kind of regression too):
>
> Have a nice day
>
> On Thu, 26 Jan 2023 at 19:50, Brian C. Lane <bcl <at> redhat.com> wrote:
> >
> > On Thu, Jan 26, 2023 at 05:39:24PM +0100, Frédéric Martinsons wrote:
> > > > This change was introduced by commit
> > > > 61dd3d4c5eb782eb43caa95342e63727db3f8281, it was needed to fix problems
> > > > growing partitions when using SD cards on the Raspberry Pi.
> > >
> > > OK , thank you very much for the explanation. Since I don't use
> > > raspberry Pi among my targets, do you think I can safely run a custom
> > > parted with a revert of 61dd3d4c5eb782eb43caa95342e63727db3f8281 ?
> >
> > That's probably fine, as long as you know what you are doing ;)
> >
> > >
> > > > Well, nothing should be actually using the CHS values these days. So
> > > > it's possible that's a bug that doesn't matter in practice, but I'll
> > > > have to look into that.
> > >
> > > OK I understand, since you close the present bug, do you know where I
> > > can have your conclusion when you'll have time to look at it ?
> > >
> > > Thanks again for the quick answer by the way.
> >
> > I'm *hoping* to get a new parted release ready sometime soon, but I'm
> > not yet sure when. I have a few other things to review and I'll take a
> > look at this at the same time to make sure it's not a bug.
> >
> > Brian
> >
> > --
> > Brian C. Lane (PST8PDT) - weldr.io - lorax - parted - pykickstart
> >

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.