On Sat, 30 Oct 2004, Martin Husemann wrote:

> On Sat, Oct 30, 2004 at 02:47:24PM +0200, Martin Husemann wrote:
> > Perl configured just fine. I'll update and see if -current /bin/sh has
> > problems.
>
> Sorry, I could not reproduce your problem. I updated my source, rebuild
> kernel and userland, and rebuild perl - no problem at all.
>
> Does it realy crash for you every time you run "make configure" in
> pkgsrc/lang/perl58?

Well, I just updated my sources and rebuilt and reinstalled (from
scratch) everything this morning.

No, it does not crash every time. Just now, make configure finished OK,
but then sh crashed somewhere during the build. After that, just trying
"make install" again without cleaning, it crashed in the same place:
either sh cored or kernel crashed (with the kernel trap 34 in pseg_get).
Then, it suddenly passed that point, but failed a bit later with
some .wrapper script error.

After that I did "make clean" and tried again "make install". Now
sh crashed immediately just before the build said
"let's make sure your kit is complete".

The crash occured in function "environment()", called from
"evalcommand()". Note that all the sh cores I've looked at seem
to always happen when sh is trying to execute some command
via `cmd` (or $(cmd)) mechanism. The traces all look more or less
similar, and the errors occur in environment().


Hmm, here's an odd data point. I did this:

 1. Run make install
 2. sh cored just as I told above
 3. I ran make install twice, and sh cored always in the
    same place
 4. I waited for a moment, and did "cat /usr/bin/* >/dev/null"
    I thought this would mix things a bit in the file cache and
    memory in general.
 5. I ran make install again, and now it passed the point
    where sh cored earlier.

I guess I'm going to try to install 2.0 or even 1.6 and see
if they work at all. I'm starting to be slightly worried
that the hardware might be broken...