I'd like to help narrow things down (on my system at least).
2.1.104 ran great for a week or so. The only patch I can remember adding was
a test of Paul's pci-ne2k driver. The kernel was mostly modular (which is
not normal for me, I usually do monolithic with "perhaps I might use this one
day" things as modules, this one however was totally modular except the bare
essentials like ext2). Oh, silly me, it was 2.1.104-pre1 not the actual .104
Anyway I recompiled with the proper .104 and Alan and Bill's patches, did my
"normal" kernel compile (mostly monolithic).
This was the kernel that had the problems.
It ran okay for about 20 mins or so, I think it really bit the dust after I
tried compiling 2.1.105 (which I got the announcement for moments after booting
this new kernel.. isn't that always the way?;) and php-3.0 simultaneously.
My box only has 32Mb of ram so that was pretty much more than used up.
I stopped the compile of PHP, but nothing cleared up. I even stopped the
kernel compile, did a few syncs to clear the cache, it was still slow as a dog
and the report of "free" said I was 15Mb into swap, which I really shouldn't
be for the processes that were running. Still none of those processes were
actually using that much memory, and as I said 2/3 of them had the 'W' flag.
So something definately was amiss. I still have that kernel around, so if
somebody could suggest some useful debugging output I could get with sysrq
keys or whatever I'll be happy to boot into the "broken" kernel and get it.
I eventually rebooted the system, finished off the kernel compile (which
started to bog down a bit towards the end) and then another reboot into
2.1.105+ac which has been running fine since (and I've given it a thrashing
with swap...although that doesn't take much on a 32Mb box)
-- Matthew Hawkins <matt@goldweb.com.au> | WWW: http://www.goldweb.com.au/~matt/ | "Do not taunt happy fun troll." UID 0 @ Goldweb Internet +61262530059 | PGP: 1024/273E35E1 - 01 8D 6C 62 4C D1 05 3D 0F 59 5B E3 81 9F 59 B9- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu