Re: /proc problems: 1.3.6[46]
Jared Mauch (jared@wolverine.hq.cic.net)
Wed, 21 Feb 1996 11:24:12 -0500 (EST)
Arthur D. Jerijian spewed:
>
> On Tue, 20 Feb 1996, David Engberg wrote:
>
> >
> > I decided that my 486 was too slow for C++ development, so went out and
> > bought a P5-133 system with a Triton chipset PCI motherboard last weekend...
> >
> > Both 1.2.11 and 1.2.13 exhibited a horrible corrupting problem where I
> > would work for a while, then gcc would spew some internal error before
> > taking out virtual memory and, after enough restarts, my root
> > filesystem.
>
> Eeeck. It looks like you have some severe memory problem. The
> Linux FAQ contains a pointer to the GCC Sig11 FAQ web site. I'd suggest
> you try to determine which of your SIMMs is the guilty one.
>
> > So I tried several of the more recent 1.3.* series, which mention
> > corrupting PCI chipset fixes in the 'config' process. I haven't seen the
> > corrupting problem, but now emacs bombs in its memory manager (xemacs
> > works ok) and 'ps u' and 'top' both die with a floating exception:
> >
> > > ps u
> > USER PID %CPU %MEM SIZE RSS TTY STAT START TIME COMMAND
> > Floating exception
>
> This is a software problem. Linux 1.3.4x-1.3.5x added a new
> field, "cached", to /proc/meminfo. You might want to upgrade to
> procps-0.99, available at
tsx-11.mit.edu, not sunsite, sunsite is the most loaded place
to get linux stuff from.
tsx-11.mit.edu:/pub/linux/BETA/procps
- jared