Re: __DEATH__ of SMP

Matthias Urlichs (smurf@work.smurf.noris.de)
23 Apr 1997 12:50:48 +0200

"Richard B. Johnson" <root@analogic.com> writes:
> >
> > Have you tried any of the AIC7XXX experimental patches?
> > ftp://ftp.pcnet.com/users/eischen/Linux
> > Daniel Eischen is doing a wonderful job.
> >
> Nope. The current driver seems to work fine unless there is a SCSI
> Bus error at which time all bets are off. Since I don't have any
> external devices on this machine, I don't get any SCSI Bus errors.
>
Our main news+cache server (four 4-GB disks) crashed about twice per week
with at least one kernel stack overflow or similar Aieee during the hour
before the crash. There was no indication at all about any SCSI flakiness.

I fixed that by patching the kernel to use two pages for the kernel stack.
No more crashes since. The only remaining problem is that the de4x5 network
driver hangs the card when it happens not to get free memory. :-( :-( :-(

An aic7xxx driver that doesn't eat so much stack _will_ be an improvement.
Still better would be a release of GCC 2.8.0 that does better stack layout
(2.6.x was _much_ better). In fact, did anybody with access to gcc
snapshots check this?

-- 
Q:  How many Christians does it take to change a light bulb?
A:  Three, but they're really only one.

-- 
Matthias Urlichs         \  noris network GmbH  /  Xlink-POP Nürnberg 
Schleiermacherstraße 12   \   Linux+Internet   /   EMail: urlichs@noris.de
90491 Nürnberg (Germany)   \    Consulting+Programming+Networking+etc'ing
   PGP: 1024/4F578875   1B 89 E2 1C 43 EA 80 44  15 D2 29 CF C6 C7 E0 DE
       Click <A HREF="http://info.noris.de/~smurf/finger">here</A>.    42