[EMAIL PROTECTED] (Thalia L. Hooker) wrote
> Anyways, my next step was to 'make zImage'. I kept getting Sig 11 errors
> though not at the same point of compilation:
> gcc: Internal compiler error: program cc1 got fatal signal 11
...
> After reading the FAQ, it seemed that my problems were typical of the ones
> described by other users. I have tried at least 30 times starting over
> from:
> make dep ; make clean
> make zImage
>
> So, has any one had the same thing happen to them and did it turn out to be
> a RAM problem? Or, how were you able to solve it?
I seem to have (nearly) the same problem on my AMD K90 system,
except the error I usually get is "segmentation fault". Again, it
happens at random places during a kernel compile. (I delete the
last .obj file and restart the compile with 'make zImage' and the
compile continues. Eventually, I make it all the way through with
a successful compile.) I also run into it perhaps 1 out of 3 times
when I do a large LaTeX compile, again at random places in the
compile.
I am inclined to believe that the sig 11 and the segmentation
faults are indeed symptoms of bad hardware. Last May, I spent
several days swapping RAM SIMMs around, hoping to get a good
set that would cure the kernel compile symptom, but with no
luck. I have tentatively concluded that my problem is not
bad RAM but some other hardware problem. My next step, when
I have the strength to face it, is to pull all the non-essential
cards and try again. I also have a few additional SIMMs to
try, just in case all the others I tried were bad. If pulling
the cards doesn't solve it, I will see if I can jump the
motherboard to reduce the CPU clock speed and/or bus speed.
I already tried turning off the CPU and motherboard caches,
without that solving it.
This really is a fascinating topic, I think. How can
we (at least I) get any work done when I don't have full
confidence in my working platform? Well, on this same
machine, W95 (often) and plain DOS (sometimes) lock up
on me. I spend more time in Linux, though, and it
essentially never locks up on me, but does bomb out on
kernel or LaTeX compiles. At least under Linux, the
error is caught gracefully and I continue to have a command
prompt and working operating system, whereas W95 is
extremely rude in requiring the machine to be powered off.
So, I would say, from my experience on this one machine,
that, given the same level of faulty hardware, Linux is
much more reliable.
Now, another interesting point: if indeed the problems
I've experienced on this machine are really _hardware_
problems, and even though W95 cannot cope with those
failures gracefully, is it possible that W95 runs
reliably given flawless hardware? That is, is it possible
that W95 gets a worse reputation than it deserves due
to unsuspected bad hardware? Perhaps I will take to
recommending that customers purchase only Hewlett-Packward
PCs. Would that solve it? (At least, Hewlett-Packward
seems to have a good reputation for quality hardware.)
-- Frank
[EMAIL PROTECTED]
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
[EMAIL PROTECTED] .
Trouble? e-mail to [EMAIL PROTECTED] .
