My basic set-up:
2 WD ide drives
1 Quantum atlas drive.
DPT PCI card.
Motherboard: Tyan Tomcat I, P90, 512kb cache
(with tag ram chip), latest bios (v3.02 award).
/dev/hda1: MSDOS
/dev/hda2: /
/dev/sda1: /usr
/dev/sda4: /home
/dev/sda2: swap
/dev/sda3: swap
System works fine with 64Mb. No problems at all.
Changed to 128Mb. Set cache at 512Mb in BIOS.
(Options:
L2 Cache Cacheable Size: 64Mb or 512Mb)
Following upgrade to 128Mb:
Boots in MSDOS and windows fine with 128Mb.
kernel 2.0.17. Relevant lines from lilo.conf:
image = /vmlinux
label = linux
append = "mcd=0x300,9 ether=5,0x340,0,0,eth0 mem=128M"
Result:
System recognized memory of 128Mb.
after line
VFS mounted root (ext2file system) readonly
(MUCH _clicking_ from hda)
hda: read_intr: status=0x59 {DriveReady SeekComplete DR (lost this) Error}
error=0x10 {SectorId Not Found}, LBAsector=267458767, sector 98330
hda: read_intr: status=0x59 {DriveReady SeekComplete DR Error}
error=0x10 {SectorId Not Found}, LBAsector=267458767, sector 98330
hda: read_intr: status=0x59 {DriveReady SeekComplete DR. Error}
error=0x10 {SectorId Not Found}, LBAsector=267458767, sector 98330
hda: read_intr: status=0x59 {DriveReady SeekComplete DR Error}
error=0x10 {SectorId Not Found}, LBAsector=267458767, sector 98330
ide0:reset:success
hda: read_intr: status=0x59 {DriveReady SeekComplete DR Error}
error=0x10 {SectorId Not Found}, LBAsector=267458767, sector 98330
end_request: I/O error, dev 3.02, sector 98330
Kernel panic: ext2_read_inode: unable to read i-node block
-inode=12022,block=49165
Booting with line: linux mem=64M
Everthing OK.
Next I booted with an older kernel, 2.0.7
After VFS mounted root, here is output:
hda: irq timeout: status 0x58 {Drive Ready SeekComplete DR}
ide0: reset:success
hda: irq timeout: status 0x50 {Drive Ready SeekComplete DR}
hda: irq timeout: status 0x50 {Drive Ready SeekComplete DR}
hda: irq timeout: status 0x50 {Drive Ready SeekComplete DR}
hda: irq timeout: status 0x58 {Drive Ready SeekComplete DR}
ide0:reset:success
hda: irq timeout: status 0x50 {Drive Ready SeekComplete DR}
end_request: I/O error, dev 3.02, sector 10
Kernel panic:EXT2-fs panic (device 03.02):
ext2_read_inode: unable to read i-node block
-inode=2,block 5
After this, it seemed it might be time to move "/" over to the
SCSI drive which I did. After editing lilo.conf and fstab
carefully, I removed /dev/hda2,3,4 /dev/hdb from fstab.
Result:
With mem=128M set in lilo.conf and/or on command line,
system hangs after
VFS: Mounted root (ext2 filesystem) readonly.
With booting off the SCSI, lilo now gives the message
ide: probable bad entry for /dev/hdc
ide: to fix it run: /usr/src/linux/scripts/MAKEDEV.ide
Warning: BIOS drive 0x82 may not be accessible.
Warning or no warning, everything seemed to work (apart from
getting the extra memory of course).
Back to BIOS. After setting everything to defaults,
I finally found out that setting cache back to 64Mb limit fixed everything.
Comments:
I don't believe there is a problem with the extra RAM.
(In fact it was a replacement for earlier faulty RAM
which even the BIOS threw up on.)
I can believe there might be a problem with motherboard but
the errors do seem odd to say the least.
I can believe there is a kernel problem.
I can also believe there is a BIOS problem.
Interesting, I changed to the Tyan board after failing to upgrade
a Micronics board to 128Mb - here even the BIOS wouldn't
recognize more than 96Mb even though the board was
supposedly capable of handling 192Mb.
I would appreciate some help or suggestions.
Mike Field