From: rrusk@ssd.csd.harris.com (Bob Rusk) Subject: Re: Four small problems (HD, VCs in X, ps/xload, Minicom in xterm) Date: Fri, 30 Oct 1992 13:23:17 GMT
On Thu, 29 Oct 1992 06:04:34 GMT, rmk@rmkhome.UUCP (Rick Kelly) wrote:
>In article <1992Oct28.075855.4644@athena.mit.edu> dthumim@athena.mit.edu (Daniel J Thumim) writes:
>>I've been running Linux on my new 486 system here for a couple of weeks
>>now... quite impressive! (I would *not* have bought a 486 to run DOS;
>>I bought the hardware for linux/X! 8-)
>>Just a couple of minor snags:
>>
>>1. Intermittently, I have the hard disk LED stay lit for a few seconds,
>> everything freezes up, and I get a HD-timeout, HD-controller reset.
>> It's a brand new HD, on a local bus IDE controller. Should this be
>> happening? (and can it be fixed?)
>The main "problem" with localbus systems is that the disk controller and the
>system cpu are contesting for the same bus. When the disk controller is
>accessing memory, the cpu must wait.
>Some people consider a localbus system as a win for MSDOS, but a loser for
>multitasking systems.
Unfortunately, this doesn't seem to be unique to localbus systems. I've got
an old Gateway 2000 386/20 DX with a recent Quantum 240MB IDE drive that
suffers from the same malady under 0.98, although it was fine under 0.96c.
It was really bad with the 0.98 kernels from the bootdisk and from SLS,
and improved somewhat when I built my own kernel from 0.98 sources with
the floating-point and SCSI stuff disabled. It went away almost entirely
when I upgraded from 4MB to 8MB, but it's still there. A couple of times
(before the RAM upgrade), I got three or four of those HD-timeout,
HD-controller reset sequences in a row on a heavily loaded system, and the
system locked up entirely.