[gnu.emacs.help] Segmentation Faults in GNU Emacs 18.55 on decstation 3100

rep@genrad.com (Pete Peterson) (01/05/91)

We have run into a problem which appears to manifest itself when editing
large (8 Mb) files using GNU Emacs 18.55 on the pmax (decstation 3100).
Editing the same or larger files on a Sun[34] or VAX causes no problems.
The problem shows up regardless of whether you're running from an xterm or
a standard terminal login and may occur during startup or may occur after a
few editing commands.

The problem shows up as a message "Fatal error (11)." (segmentation fault)
then a delay of several minutes followed by a standard unix segmentation
fault -- core dumped message.  

The emacs was built using Dec's standard C compiler with optimization
turned off since I had found the optimization to produce broken code when I
had compiled perl for the pmax.

I used s-bsd4-2.h.  The m-pmax.h that came with the distribution didn't
quite work, since m-mips.h pointed to a nonexistent crt1.o and had time.h
in the wrong place for our decstation ultrix system.  I made m-pmax.h from
m-mips.h with the following changes.

 /* #define BIG_ENDIAN */

 /* #define START_FILES pre-crt0.o /usr/lib/crt1.o */
 #define START_FILES pre-crt0.o /usr/lib/crt0.o

 /* #include <bsd/sys/time.h> */
 #include <sys/time.h>

 #undef LIBS_MACHINE
 #undef BIG_ENDIAN
...........................................................................
Is this a solved problem and if so, could somebody describe the fix?  I
have no contact with decstations myself except for building emacs and perl
for them, so I know virtually nothing about their architecture or the
quirks of the DEC compiler.

---------------------------------------------------------------------------

An unrelated question: I have seen references to gnu.emacs.help which we
don't seem to get, although we get the following:
 gnu.announce, gnu.chess, gnu.config, gnu.emacs.bug, gnu.emacs.gnus,
 gnu.emacs.lisp.manual, gnu.emacs.vms, gnu.emacs, gnu.g++.bug, gnu.g++,
 gnu.gcc.bug, gnu.gcc, gnu.gdb.bug, gnu.ghostscript.bug gnu.test,
 gnu.utils.bug

Does gnu.emacs.help come from a different place from the other gnu.xxx
groups?  Our news guru tells me we aren't doing anything to exclude that
particular group.

	pete peterson
	rep@genrad.com
	{decvax,linus,wjh12,mit-eddie,masscomp}!genrad!rep