sbrk(2) on PDP 11/73's

[brad at ..UUCP]

/* Written 12:04 pm  Jan 21, 1986 by brad in :net.bugs.2bsd */
/* ---------- "sbrk(2) on PDP 11/73's" ---------- */
System: 2.9BSD from CSS/Harvard with all bug fixes install:

On an PDP 11/73 with 1 MEG of memory 2 RL02, 3 DHV, and 1 DH/DM, plus
a MXV11-B, (CPU is a KDJ11-A). I tracked to bugs which someone might
be able to build upon.

First: Sticky bit programs don't read in from swap.  Haven't looked
much further.

Second: Programs that call sbrk(2) directly and are split i/d. seem
to have a problem.  (Note vi does it sometimes and nroff does it,
binaries work fine on a 11/44, and nroff is compile so that it is
split i/d) The problem is that the program dumps core with a bus
error, but the bus errors don't always happen.

according to adb(1) the program stops at:

	_sbrk+026:	bcc	_sbrk+034

looking at sbrk.s there is no 'bcc', but adb sbrk.s there is. the
line before is this:

	sys	indir {sys	break; 0}

in the sbrk.s file this line is:

	sys	0; 9f

My question, does this mean that the kernel call to 'break', which
turns out to be 'sbreak()' in sys1.c, is the function dieing or
could it be one of the functions that sbreak() calls?

Also does the 22-bit address play an part in this, where as the
11/44 is only 18-bits plus a map?

I am lost on where to go further, any suggestions?

Bradley Smith			UUCP: {cepu,ihnp4,noao,uiucdcs}!bradley!brad
Text Processing			ARPA: cepu!bradley!brad at UCLA-LOCUS
Bradley University		PH: (309) 676-7611 Ext. 446
Peoria, IL 61625
/* End of text from :net.bugs.2bsd */