signal 11! Ack!

[Kevin Buettner]

On Dec 8,  1:20am, Todd Hought wrote:

> ok, I'm rebuilding my machine, and everything is going along just fine,
> when during the compile of the i2c package, so that I can use
> lm-sensors, dies with signal 11 from gcc.
> I check google.
> Google brings up many many articles, most of which say that I probably
> have bad ram.
> I install memtest86.
> Sure enough, my 256M stick is bad.
> I take out the ram, and run on the other 128 that was in the box.
> I go to recompile i2c, and still the signal 11 rears its ugly head.
[...]
> But what could be the problem?

If the SIGSEGV is consistently reproducible, it's likely that it's
*not* bad memory.  Instead, I'd suspect a gcc bug.  You might want to
upgrade or downgrade gcc to see if that helps.

If you feel up to debugging this problem yourself, take the gcc
command line that's giving you the SIGSEGV, and run it by hand from
the shell.  Do it a couple of times to make sure that the SIGSEV is
reproducible.

Then run the gcc command again, this time adding "-v -save-temps" (but
without the double quotes) to gcc's command line options.  This will
show the command line used by each step in that particular compile. 
After that, using the command line of the failing command, you can use
GDB to debug that component in the toolchain.  (You'll find it helpful
to have source code available for gcc to do this.)

[If you send me the .i file produced from running "gcc -v -save-temps ...",
I'll try to reproduce the problem on my machines.]

Kevin