I just upgraded to a P5 from an i486 where I used to run 2 SCSI II
hard drives hanging off an AHA2842vl. I substituted the SCSI controller
with an NCR53c810 after having duly recompiled the kernel (1.3.29) and
rebooted on the new kit. Nooo hassles.
I then applied 1.3.30 patch, recompiled but lo ! lilo flunked with
Device 0x0800: Invalid partition table, 2nd entry
3D address: 1/0/320 (761280)
Linear address: 38/18/275 (655360)
('selection' - what a wonderfull tool.)
Zis zhe is not nice ! 'fdisk -l /dev/sda' reports
Disk /dev/sda: 39 heads, 61 sectors, 1018 cylinders
Units = cylinders of 1218048 bytes, blocks of 1024 bytes, counting from 0
Device Boot Start End #cyls #blocks Id System
/dev/sda1 * 0+ 275- 276- 327664 6 DOS 16-bit FAT >=32M
start: (c,h,s) expected (0,0,33) found (0,1,1)
end: (c,h,s) expected (19,18,101) found (63,63,96)
/dev/sda2 275+ 586- 311- 369664 83 Linux native
start: (c,h,s) expected (19,18,102) found (64,0,65)
end: (c,h,s) expected (74,9,173) found (168,63,160)
/dev/sda3 586+ 983- 397- 472064 83 Linux native
start: (c,h,s) expected (74,9,174) found (169,0,129)
end: (c,h,s) expected (215,4,207) found (255,63,224)
partition ends on cylinder 1023, beyond the end of the disk
/dev/sda4 983+ 1018- 36- 41984 82 Linux swap
start: (c,h,s) expected (215,4,208) found (255,63,224)
end: (c,h,s) expected (250,15,239) found (255,63,224)
partition ends on cylinder 1023, beyond the end of the disk
while 'fdisk -l /dev/sdb' saiz
Disk /dev/sdb: 34 heads, 61 sectors, 1021 cylinders
Units = cylinders of 1061888 bytes, blocks of 1024 bytes, counting from 0
Device Boot Start End #cyls #blocks Id System
/dev/sdb1 0+ 493- 494- 511984 6 DOS 16-bit FAT >=32M
start: (c,h,s) expected (0,0,33) found (0,1,1)
end: (c,h,s) expected (237,24,118) found (243,63,96)
/dev/sdb2 493+ 1021- 528- 546816 83 Linux native
start: (c,h,s) expected (237,24,119) found (244,0,65)
end: (c,h,s) expected (253,1,209) found (255,63,224)
partition ends on cylinder 1023, beyond the end of the disk
/dev/sdb3 0 - 0 0 0 Empty
/dev/sdb4 0 - 0 0 0 Empty
(this is fdisk 3.0 - I have not yet had the pleasure of compiling 3.1,
Andries)
and 'df' quips
Filesystem 1024-blocks Used Available Capacity Mounted on
/dev/sda2 366535 287965 74874 79% /
/dev/sda3 466843 461011 1112 100% /home
/dev/sdb2 537837 489358 43011 92% /home1
/dev/sda1 327480 312816 14664 96% /dosc
/dev/sdb1 511712 243024 268688 47% /dos
I got past this problem by reverting temporarily back to i486, AHA2842vl
blah, blah, ..., running lilo, again to P5 stuff, but I am so sure that
this is not my destiny (well, I *hope* not :)
Seems the Adaptec 2842 (and 1542 predecessor) read the geometry different
to the NCR thingy - yet translation on the Adaptecs was never enabled
(well, it was, but then the box fell over, so translation did not work /
I was forced to disable translation) - so it would appear that the NCR
controller translates ?
The BootParams HOWTO (which is amongst the nightly mirrored stuff I keep)
still reckons that there are no boot parameters I can pass on this config.
Well, short of wiping the drives (they're both 1 Gigs) and redoing the
entire installation, what gives ?
(I'd really like an answer soonish - what with Linus cranking out patches
'bi-daily' :)
Oh, I also converted to ELF at the time Linus did, got gcc 2.7.0,
running 1.3.30, Maxtor MXT-1240S and IBM DPES-3108, ...
TIA
Peter Kooiman | Fax : ++27-12-663-4191/2
Paradigm Systems Technology |
Pretoria, South Africa | UUCP : pko@silmaril.UUCP
Voice : ++27-12-663-5340 | DNS : pko@paradigm.co.za