Re: buildworld failures on STABLE

Eugene Grosbein wrote:
On Sun, Oct 07, 2007 at 11:40:11AM +0200, Roland Smith wrote:

On a remote machine currently with RELENG-6 from 20th. June, with STABLE sources from this morning I get build failures in contrib/ similar to:
<snip>
/usr/src/gnu/usr.bin/cc/cc_int/../../../../contrib/gcc/lcm.c:801: internal compiler error: Segmentation fault: 11
<snip>
Would this indicate a hardware (memory) problem?
Yes. The compiler dying with signal 11 is a typical memory problem.

Any way to test remotely?
There are memory test applications like memtest86+ (http://www.memtest.org/).
You have to boot from it, but it does support a console on a serial port.

There is also ports/sysutils/memtest that may be run from multiuser
if one manages to build it with broken memory :-)
Or it's possible to install a package.

One problem with running such application with OS loaded is that
kernel places strict limit to amount of memory that user-level application
may lock. There is a PR with a patch that allows to raise the limit
so memtest could test most part of free memory:

http://www.FreeBSD.org/cgi/query-pr.cgi?pr=kern/114654

Thanks for the memtest advice, I ran it and memory is definitely hosed.
Sigh, that means 400 kilometers of travel...

Per olof
_______________________________________________
freebsd-stable@xxxxxxxxxxx mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "freebsd-stable-unsubscribe@xxxxxxxxxxx"

Relevant Pages

  • Re: buildworld failures on STABLE
    ... compiler error: Segmentation fault: 11 ... There are memory test applications like memtest86+. ... kernel places strict limit to amount of memory that user-level application ...
    (freebsd-stable)
  • Re: [Lit.] Buffer overruns
    ... >>of physical memory. ... But that doesn't seem to be what the other posters mean by "overcommit" ... their requests don't ...
    (sci.crypt)
  • Re: When to check the return value of malloc
    ... Note yesterday I encountered someone who got a "not enough memory to ... well you draw the line at register dumps so I think this is a matter ... paralyzed like a deer in headlights when confronted with "memory allocation ...
    (comp.lang.c)
  • Re: Exploiting the Perl
    ... CO> snip ... This is what shared memory is for. ... asked about shared ram that stays resident outside of processes. ... CO> resort to slow disk if you need fast memory. ...
    (perl.beginners)
  • Re: Snow Leopard: A 64-bit Observation
    ... You can use manual memory management, use less memory, and just cope. ... short of a commitment or promise to ship 64-bit Photoshop CS5. ... Which Windows does not do nearly as well as OS X. ...
    (comp.sys.mac.advocacy)