2.5.30 LILO FreeBSD partition problems

2.5.30 LILO FreeBSD partition problems

[Skip Ford]

While running 2.5.30 I receive this error when running LILO with a
FreeBSD partition in lilo.conf

  Device 0x0300: Invalid partition table, 3rd entry
    3D address:     1/0/530 (534240)
    Linear address: 1/14/8446 (8514450)

I removed the fbsd entry and LILO had no problems.  I then booted
to 2.4 and readded the fbsd partition and it installed fine.

The problem seems similar to what Ingo reported for 2.5.29 however on my
system the only problem is the freebsd partition.  Also, 2.5.29 worked
fine for me.

-- 
Skip

Re: 2.5.30 LILO FreeBSD partition problems

[Andries Brouwer]

On Sat, Aug 03, 2002 at 07:00:21PM -0400, Skip Ford wrote:

> While running 2.5.30 I receive this error when running LILO with a
> FreeBSD partition in lilo.conf
> 
>   Device 0x0300: Invalid partition table, 3rd entry
>     3D address:     1/0/530 (534240)
>     Linear address: 1/14/8446 (8514450)
> 
> I removed the fbsd entry and LILO had no problems.  I then booted
> to 2.4 and readded the fbsd partition and it installed fine.

Which LILO version is this?
What do cfdisk -Ps /dev/hda and cfdisk -Pt /dev/hda say?
What are the kernel boot messages for this disk
(dmesg | grep hda), both for 2.5.29 and 2.5.30?

Andries

[lilo-22.3.1 does not print such messages.
Must be from the part_verify() in some older LILO.
Probably some LILO option like "ignore-table" or "linear" or "lba32"
would help. But it is interesting to see where this 1/0/530 comes from.]

Re: 2.5.30 LILO FreeBSD partition problems

[Skip Ford]

Andries Brouwer wrote:
> On Sat, Aug 03, 2002 at 07:00:21PM -0400, Skip Ford wrote:
> 
> > While running 2.5.30 I receive this error when running LILO with a
> > FreeBSD partition in lilo.conf
> > 
> >   Device 0x0300: Invalid partition table, 3rd entry
> >     3D address:     1/0/530 (534240)
> >     Linear address: 1/14/8446 (8514450)
> > 
> > I removed the fbsd entry and LILO had no problems.  I then booted
> > to 2.4 and readded the fbsd partition and it installed fine.
> 
> Which LILO version is this?

21.4-3

> What do cfdisk -Ps /dev/hda 

Partition Table for /dev/hda

            First    Last
 # Type     Sector   Sector   Offset  Length   Filesystem Type (ID)   Flags
-- ------- -------- --------- ------ --------- ---------------------- ---------
 1 Primary        0  8193149      63  8193150  Win95 FAT32 (0B)       Boot (80)
 2 Primary  8193150  8514449       0   321300  Linux native (83)      None (00)
 3 Primary  8514450 26957069       0 18442620  BSD/386 (A5)           None (00)
 4 Primary 26957070 39873329       0 12916260  Extended (05)          None (00)
 5 Logical 26957070 35150219      63  8193150  Linux native (83)      None (00)
 6 Logical 35150220 35471519      63   321300  Linux native (83)      None (00)
 7 Logical 35471520 35567909      63    96390  Linux swap (82)        None (00)
 8 Logical 35567910 39873329      63  4305420  Linux native (83)      None (00)


> and cfdisk -Pt /dev/hda say?

Partition Table for /dev/hda

         ---Starting---      ----Ending----    Start Number of
 # Flags Head Sect Cyl   ID  Head Sect Cyl    Sector  Sectors
-- ----- ---- ---- ---- ---- ---- ---- ---- -------- ---------
 1  0x80    1    1    0 0x0B  254   63  509       63   8193087
 2  0x00    0    1  510 0x83  254   63  529  8193150    321300
 3  0x00    0    1  530 0xA5  254   63 1023  8514450  18442620
 4  0x00  254   63 1023 0x05  254   63 1023 26957070  12916260
 5  0x00  254   63 1023 0x83  254   63 1023       63   8193087
 6  0x00  254   63 1023 0x83  254   63 1023       63    321237
 7  0x00  254   63 1023 0x82  254   63 1023       63     96327
 8  0x00  254   63 1023 0x83  254   63 1023       63   4305357


> What are the kernel boot messages for this disk
> (dmesg | grep hda), both for 2.5.29

hda: Maxtor 2B020H1, DISK drive
 hda: 39876480 sectors w/2048KiB Cache, CHS=39560/16/63, UDMA(33)
 hda: [PTBL] [2482/255/63] hda1 hda2 hda3 hda4 < hda5 hda6 hda7 hda8 >
 hda0: <bsd: hda9 hda10 hda11 hda12 hda13 hda14 >

> and 2.5.30?

hda: Maxtor 2B020H1, DISK drive
 hda: 39876480 sectors w/2048KiB Cache, CHS=39560/16/63, UDMA(33)
 hda: hda1 hda2 hda3 hda4 < hda5 hda6 hda7 hda8 >
 hda3: <bsd: hda9 hda10 hda11 hda12 hda13 hda14 >

> Probably some LILO option like "ignore-table" or "linear" or "lba32"
> would help. But it is interesting to see where this 1/0/530 comes from.]

lba32 and linear each didn't work.  I didn't try ignore-table.  I see
the difference above between .29 and .30 with [PTBL] but I don't know
what it means.

-- 
Skip

Re: 2.5.30 LILO FreeBSD partition problems

[Andries Brouwer]

On Sat, Aug 03, 2002 at 08:17:14PM -0400, Skip Ford wrote:

> > Which LILO version is this?
> 
> 21.4-3
> 
> > What are the kernel boot messages for this disk
> 
> 2.5.29:
> hda: Maxtor 2B020H1, DISK drive
>  hda: 39876480 sectors w/2048KiB Cache, CHS=39560/16/63, UDMA(33)
>  hda: [PTBL] [2482/255/63] hda1 hda2 hda3 hda4 < hda5 hda6 hda7 hda8 >
>
> 2.5.30:
>  hda: hda1 hda2 hda3 hda4 < hda5 hda6 hda7 hda8 >
> 
> > Probably some LILO option like "ignore-table" or "linear" or "lba32"
> > would help.
> 
> lba32 and linear each didn't work.  I didn't try ignore-table.  I see
> the difference above between .29 and .30 with [PTBL] but I don't know
> what it means.

LILO wants to check that what it sees in the partition table
corresponds to the truth. The kernel has first decided that
the truth is CHS=39560/16/63, but then looks at the partition table
and sees that some programs are going to be unhappy if it reveals
this truth, and instead it returns what it finds in the partition
table. That is what this "[PTBL] [2482/255/63]" means.

Since 2.5.30 the kernel no longer looks at the partition table to
guess what would make fdisk and lilo happy. These programs can
(and do) look at the partition table themselves and need no
kernel help. Now fdisk already knows this, but maybe lilo doesn't.

If you want the kernel to provide lilo with some random numbers,
add boot parameters to the kernel invocation: hda=C,H,S.
If you want lilo to ignore the discrepancy between what the
disk said and the kernel repeated and what it sees in the
partition table, try the global option "ignore-table".
(I never tried, this is just from the man page.)

If lilo tries to check 3D addresses in spite of the fact that
it got a "linear" or "lba32" option, then that is a lilo bug,
especially if it aborts on a difference.
Let me cc the maintainer - maybe John Coffman.

Andries

Re: 2.5.30 LILO FreeBSD partition problems

[Skip Ford]

Andries Brouwer wrote:
> On Sat, Aug 03, 2002 at 07:00:21PM -0400, Skip Ford wrote:
> 
> > While running 2.5.30 I receive this error when running LILO with a
> > FreeBSD partition in lilo.conf
> > 
> >   Device 0x0300: Invalid partition table, 3rd entry
> >     3D address:     1/0/530 (534240)
> >     Linear address: 1/14/8446 (8514450)
> > 
> > I removed the fbsd entry and LILO had no problems.  I then booted
> > to 2.4 and readded the fbsd partition and it installed fine.
> 
> [snip]
>
> If you want the kernel to provide lilo with some random numbers,
> add boot parameters to the kernel invocation: hda=C,H,S.
> If you want lilo to ignore the discrepancy between what the
> disk said and the kernel repeated and what it sees in the
> partition table, try the global option "ignore-table".
> (I never tried, this is just from the man page.)

Ok.  I finally got the nerve to try "ignore-table" and it seems to work.
LILO successfully installs and all partitions successfully boot.

It still prints the above ugly message every time though.

> If lilo tries to check 3D addresses in spite of the fact that
> it got a "linear" or "lba32" option, then that is a lilo bug,
> especially if it aborts on a difference.

-- 
Skip