Hi all,
i've tried each of the 3 simms alone and each get the same
result at exactly the same place in the build.

( i didn't do make clean; the 2nd and 3rd times, hope is ok)

Perhaps it's not the RAM.

Perhaps the build is broken?

ideas?


builddir="`pwd`"; cd avcall && make && make check && make install-lib libdir="$b
uilddir" includedir="$builddir"
./minitests > minitests.out
LC_ALL=C uniq -u < minitests.out > minitests.output.sparc--netbsdelf
test '!' -s minitests.output.sparc--netbsdelf
*** Error code 1

thks,
P.









Bernd Sieker wrote:

> On 18.06.06, 21:22:17, Paul (NCC/CS) wrote:
> >
> >
> > thanks.
> >
> > well shell theory is not going that well.
> >
> > tried to build with limits increased for tcsh or using sh,
> > but no different. still can't build lisp.
> >
> > clisp fails minitests and dies.
> > ecl ( not in pkgsrc)  has core dump in ecl_min binary.
> >
> > but it seems to be always in the same place.
> > I thought ram faults would be random.
> >
> >
> > anyway i am now doing builds with 1 ram stick at a time.
> > try to find a culprit.
> >
> > the  first SIMM alone still has the crashes.
> >
> > perhaps that means when i try the next one it will build?
> >
> > if it doesn't what are the chances that both  SIMM's have faults?
>
> There is a slight chance that there is a problem with the CPU Cache;
> a weak bit that only flips occasionally. Over time I've had three
> MBus modules die (all of them SM71 of various revisions). One of
> them only very rarely gave me trouble, that I couldn't get rid of
> by changing memory modules. After exchanging the CPU module all
> was well. I didin't follow the discussion so I don't know what type
> of machine you're using, or if it has swappable CPUs.
>
> >
> >
> > i'm into my second solid week on this!  doh.
> >
> > chs,
> > paul.
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > Martin Husemann wrote:
> >
> > > On Sun, Jun 18, 2006 at 01:53:10PM +1000, Paul (NCC/CS) wrote:
> > > > perhaps i've answered my own question, but i'd
> > > > still be grateful to here opinions on the use of tcsh
> > > > as a root shell for building pkgsrc?
> > >
> > > The default user rlimits are not depending on the shell you use (some are
> > > set depending on your available memory).
> > >
> > > I use tcsh as root shell (on various archs), and have not seen any problems.
> > >
> > > I think I remember some pkgs explicitly testing for low limits and asking
> > > to rise them before starting the build, maybe some pkgs do miss this (or
> > > the needed resources have grown with some update)
> > >
> > > Martin
> >
> > --
> >
> > ____________________________________________
> >
> > Paul
> > Australian Bureau Of Meteorology
> > ____________________________________________
> >
> >
> >
> >
> > "In a world without walls and fences,
> > who needs windows and gates ?"
> >
> >
> >
> >
> >
> >
>
> --
> Bernd Sieker
>
> NetBSD: Use the ENTIRE computer!
>                 -- Andrew Gillham

--

____________________________________________

Paul Seymour           (pts@bom.gov.au)
Senior Programmer
National Climate Centre / Computing Section
Australian Bureau Of Meteorology
____________________________________________




"In a world without walls and fences,
who needs windows and gates ?"