heap_modifytuple: repl is \-62

Solaris x86 v2.6 is broken...has been for weeks now, but to date, nobody
seems able to come up with a solution, or a suggestion on further
debugging, so once more I bring it up :)

The first question is, of course, is anyone else seeing this using Solaris
x86?

If not...*where* should I try to debug this?  It just hangs there once I
hit this, process CPU time remains static...


> ./initdb --pglib=/loc/pgsql/lib --pgdata=/loc/pgsql/data
initdb: using /loc/pgsql/lib/local1_template1.bki.source as input to
	create the template database. 
initdb: using /loc/pgsql/lib/global1.bki.source as input to create the
	global classes. 
initdb: using /loc/pgsql/lib/pg_hba.conf.sample as the host-based
	authentication control file. 

We are initializing the database system with username marc (uid=100).
This user will own all the files and must also own the server process.

initdb: creating template database in /loc/pgsql/data/base/template1
Running: postgres -boot -C -F -D/loc/pgsql/data -Q template1
ERROR:  heap_modifytuple: repl is \-62
ERROR:  heap_modifytuple: repl is \-62



for (attoff = 0;
     attoff < numberOfAttributes;
     attoff += 1)
{
	if (repl[attoff] == ' ')
	{
		value[attoff] =
			heap_getattr(tuple,
					 AttrOffsetGetAttrNumber(attoff),
					 RelationGetTupleDescriptor(relation),
					 &isNull);
		nulls[attoff] = (isNull) ? 'n' : ' ';
	}
	else if (repl[attoff] != 'r')
	{
		elog(ERROR, "heap_modifytuple: repl is \\%3d", repl[attoff]);
	}
	else
	{						/* == 'r' */
		value[attoff] = replValue[attoff];
		nulls[attoff] = replNull[attoff];
	}
}