Re: [gentoo-user] gcc seg faults very often

[Raphael Melo de Oliveira Bastos Sales]

Hi Richard,

    First of all, thanks for replying. I'll test the swap partition
with the command you sent. But how do I test the memory? Is there any
way to do it? I think I configured CMOS to do a memory check during
start up.

    Also, I'm using -march=i686 -O2 -fomit-frame-pointer, since I
don't know if a higher value is compatible with my AMD Sempron.

2005/8/22, Richard Fish <bigfish@××××××××××.org>:
> Raphael Melo de Oliveira Bastos Sales wrote:
>
> >Hi there,
> >
> >   I'm using gcc-3.3.5.20050130-r1 and I keep getting segmentation
> >fault every now and then. Since it doesn't work, I can't compile an
> >earlier, more stable version. I hope it isn't a hardware failure,
> >because the warranty on my new computer just ended. It is probably not
> >an memory error, because I'm not using -pipe. I checked the filesystem
> >and no corruption was found.
> >
> >
>
> Don't be so sure.  -pipe doesn't add that much additional memory
> overhead, in fact, only a few pages used as an IO buffer between the
> processes.  The process of compiling itself is very tough on memory,
> reading and writing to various locations in rapid succession.
>
> I would say memory is the most likely problem, but it could be
> overheating or power supply problems also.
>
> >   Anybody had this type of error too? If so, how did you handle it?
> >Are there any tools to check the hard drive's surface for flaws?
> >
> >
>
> Bad disk blocks are almost certainly not the issue, as you would end up
> with IO errors during the compilation, not segfaults.  Well, I guess if
> your swap had bad blocks, you might get a segfault...
>
> Anyway, "dd if=/dev/hda of=/dev/null bs=4k" will test readability of
> your entire disk.  It doesn't test the validity of your data though...
>
> -Richard
>
> --
> gentoo-user@g.o mailing list
>
>

-- 
gentoo-user@g.o mailing list