[Fwd: Re: gcc3.3.2, kernel-2.6.6, and Mandrake 10.0 compiling problem.]

From: The Viking
Date: Mon Jun 21 2004 - 17:09:39 EST



Does this happen randomly, not in the same place each time?
If yes, this usually happens with flakey hardware.
I had this problem just yesterday. My box survives memtest86
and cpuburn, yet gcc segfaults within minute or two.
Relaxing DDR timing (Trcd: 2->3) eliminated segvs.

OTOH, note that there exist K6's with nasty bug:
in some very rare circumstances (you have to have more than 32MB or RAM,
you must modify data lying exactly N*32MB away from your current
instruction pointer, etc...) CPU erroneously execute an instruction
twice. As you can imagine, that will make your computer very unhappy.

As I said, I've done all of that. My guess is: it's not a hardware problem, it seems to be consistently at exactly the same place every single time.
I have no way of measuring whether my compiler modifies data exactly Nx32Mb away...and as I have 192Mb, I don't know if that's even the fault, though I've got a K6-II@500MHz (running at 533MHz, more stable that way- keeps behaving strangely at 500MHz)
I'm still scratching my head about it, frankly. I finally found my local Mandrake kernel source tree, so I'll see if the toolchain even builds that. If it doesn't, then I'll have to hunt in other areas for answers.

Cheers, and thanks for your comments.
The Viking
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/