|
The machine was initially a Windows box, and has 2 x motherboard IDE interfaces (ide0 and ide1)
Originally it had an ATAPI CDROM as master on ide0, and a single IDE disk as master on ide1.
This has subsequently had an experimental LVM (Logical Volume Management) Linux config and so had had the CDROM disconnected and a second IDE disk installed instead (as master on ide0)
It all worked a few weeks ago, but for some reason it wouldn't now boot as it was - which was strange, I was sure I left it in a usable state.
Essentially, it would hang at this point in the boot process:
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide1 at 0x170-0x177,0x376 on irq 15
As is the nature of this kind of problem, Googling was no use at all. Several reports of similar symptoms but no authoritative conclusions - reinforcing that old saw:
Go not to Google for knowledge, for it will say both "Yes" and "No" and "Did you mean riutroutubgrouting?"
So I...
Disconnect everything IDE related - machine will now boot from a RedHat 7.3 bootnet.img floppy, but it's not much good as I can't install RH on to thin air.
Start reattaching devices one by one - two disks, disk and the CDROM as slave on same bus, one disk and CDROM as separate masters - all to no avail. It seems that anything on ide1 causes the machine to fail to boot.
Naturally, I suspect ide1 is fried.
So I concentrate on ide0 - one disk only, disk and CDROM, CDROM only, different CDROM only, other disk, different cables, different ordering on the cables, different jumper settings on the disks.... nope.
Next, have a look at the BIOS - it's an ancient Award BIOS. Poke this, poke that, change 'enable' to 'disable' for lots of things. Reset to factory defaults, start again a bit more logically. Eventually get a result with a single disk as jumpered master and only the device on ide0 - it'll boot from the bootnet.img floppy. Achieved by enabling 'Assign IRQ for VGA" or similar.
With a working single disk, I proceed to do a network install of RH - which works fine. In the process discover that the RedHat 7.3 installer is apparently unable to autopartition a device that's previously been part of an LVM set (had to manually delete the old partition setup with disk druid).
Reboot - success!
Except I've got a messages file containing:
PCI: Found IRQ 11 for device 00:01.1
IRQ routing conflict for 00:01.1, have irq 14, want irq 11
and
hda: dma_intr: status=0x51 { DriveReady SeekComplete Error }
hda: dma_intr: error=0x84 { DriveStatusError BadCRC }
Swap around a couple of cards without success - there'a a PCI VGA adapter and an old ISA EtherLink III, plus an on-board 'Plug and Play AudioDrive' chip. Poke around the BIOS again without any success.
Uh oh - back to no booting again, this time we hang at a different point, just after "Enabling swap space".
In sheer frustration, I pull everything out again and plug the disk into the supposedly fritzed ide1 instead of ide0. Much to my surprise, everything now works fine (once I've changed hda to hdc in grub.conf and /etc/fstab)
A few cautious reboots, resetting of BIOS to defaults again (why does the memory count run three times by default? Who knows - but enabling the quick power on self test BIOS setting gets it back to a single count again), replacing the EtherLink III and everything is still well - what's more, no IRQ conflicts or dma errors.
Add *anything* on ide0, however, and it all falls apart.
So it was ide0 that was fritzed all the time.
As to what killed it in the first place? No idea. I probably plugged the CDROM cable in upside down when I finished the LVM experiments, found it wouldn't boot and thought 'hell with it, I'll fix it later'. What *is* the point of a keyed connector plug if the socket doesn't care either way?
Definitely a case of "How much are your IDE interfaces? A fiver each and 8 quid for two.... Here's 3 quid, I'll have the other one."
--
simon
|
