From mjtg@cus.cam.ac.uk

Out of memory!
Out of memory!
Callback called exit at bin/cjc.stats line 1, <IN> chunk 9.
END failed--cleanup aborted at bin/cjc.stats line 1, <IN> chunk 9.

a) The message "Out of memory!" is not listed in perldiag in
  perl5.005 onwards, tho' it is still listed in perl5.004_05 . Has
  the entry been removed erroneously, or should the message have been
  replaced by

  "Out of memory during request for %s" ?

b) I don't understand the explanation for "Callback called exit"​:

  (F) A subroutine invoked from an external package via
  perl_call_sv() exited by calling exit.

  My program is pure Perl, with no "external package" involved,
  and no explicit use of "exit".

c) Similarly, I don't understand the last message​:

  END failed--cleanup aborted
  (F) An untrapped exception was raised while executing an
  END subroutine. The interpreter is immediately exited.

  My program contains no END {} subroutine.

Presumably the last two really mean "Perl's internal data structures
are in a twist", and perhaps indicate bugs in Perl. Is it worth
seeing if I can generate a smaller test case?

And the original problem seems to be another example of "debugger causing
space leak", which I previously reported long long ago (but without a
useful test case). The program uses a constant 3784K when not under
debug, and a constant 5256K under debug if no breakpoints are set.
But if conditional breakpoints are set, the size expands steadily until
swap runs out. Research on this continues.

Mike Guy

% perl -V
Summary of my perl5 (5.0 patchlevel 5 subversion 3) configuration​:
  Platform​:
  osname=solaris, osvers=2.6, archname=sun4-solaris
  uname=''
  hint=previous, useposix=true, d_sigaction=define
  usethreads=undef useperlio=undef d_sfio=undef
  Compiler​:
  cc='gcc', optimize='-O', gccversion=2.7.2.3
  cppflags='-I/usr/local/include -I/opt/local/include -DREG_INFTY=22786'
  ccflags ='-I/usr/local/include -I/opt/local/include -DREG_INFTY=22786'
  stdchar='unsigned char', d_stdstdio=define, usevfork=false
  intsize=4, longsize=4, ptrsize=4, doublesize=8
  d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=16
  alignbytes=8, usemymalloc=y, prototype=define
  Linker and Libraries​:
  ld='gcc', ldflags =' -L/usr/local/lib -L/opt/local/lib'
  libpth=/usr/local/lib /opt/local/lib /lib /usr/lib /usr/ccs/lib
  libs=-lsocket -lnsl -ldl -lm -lc -lcrypt
  libc=/lib/libc.so, so=so, useshrplib=false, libperl=libperl.a
  Dynamic Linking​:
  dlsrc=dl_dlopen.xs, dlext=so, d_dlsymun=undef, ccdlflags=' '
  cccdlflags='-fpic', lddlflags='-G -L/usr/local/lib -L/opt/local/lib'

Characteristics of this binary (from libperl)​:
  Built under solaris
  Compiled at Mar 30 1999 14​:22​:18
  @​INC​:
  /home/mjtg/perl5.005_03/lib
  /home/mjtg/perl5.005_03/lib
  /home/mjtg/perl5.005_03/lib
  .

From @gsar

On Thu, 30 Sep 1999 14​:31​:28 BST, "M.J.T. Guy" wrote​:

a) The message "Out of memory!" is not listed in perldiag in
perl5.005 onwards, tho' it is still listed in perl5.004_05 . Has
the entry been removed erroneously, [...]

I think so.

[two other errors]

Presumably the last two really mean "Perl's internal data structures
are in a twist", and perhaps indicate bugs in Perl. Is it worth
seeing if I can generate a smaller test case?

Please do.

Sarathy
gsar@​activestate.com

From @gsar

On Thu, 30 Sep 1999 14​:31​:28 BST, "M.J.T. Guy" wrote​:

a) The message "Out of memory!" is not listed in perldiag in
perl5.005 onwards, tho' it is still listed in perl5.004_05 . Has
the entry been removed erroneously, or should the message have been
replaced by

  "Out of memory during request for %s"      ?

FYI.

Sarathy
gsar@​activestate.com

-----------------------------------8<-----------------------------------
Change 4264 by gsar@auger on 1999/09/30 17:59:26

	re-add missing "Out of memory!" entry

Affected files ...

... //depot/perl/pod/perldiag.pod#97 edit

Differences ...

==== //depot/perl/pod/perldiag.pod#97 (text) ====
Index: perl/pod/perldiag.pod
--- perl/pod/perldiag.pod.~1~	Thu Sep 30 10:59:30 1999
+++ perl/pod/perldiag.pod	Thu Sep 30 10:59:30 1999
@@ -2141,6 +2141,12 @@
 For example, if you say "*foo *foo" it will be interpreted as
 if you said "*foo * 'foo'".
 
+=item Out of memory!
+
+(X) The malloc() function returned 0, indicating there was insufficient
+remaining memory (or virtual memory) to satisfy the request.  Perl
+has no option but to exit immediately.
+
 =item Out of memory for yacc stack
 
 (F) The yacc parser wanted to grow its stack so it could continue parsing,
End of Patch.

From [Unknown Contact. See original ticket]

On Thu, 30 Sep 1999 at 11​:18​:05 -0700, Gurusamy Sarathy wrote​:

+=item Out of memory!
+
+(X) The malloc() function returned 0, indicating there was insufficient
+remaining memory (or virtual memory) to satisfy the request. Perl
+has no option but to exit immediately.

Should $^M be mentioned here?

Ian

From @gsar

On Thu, 30 Sep 1999 23​:20​:43 BST, Ian Phillipps wrote​:

On Thu, 30 Sep 1999 at 11​:18​:05 -0700, Gurusamy Sarathy wrote​:

+=item Out of memory!
+
+(X) The malloc() function returned 0, indicating there was insufficient
+remaining memory (or virtual memory) to satisfy the request. Perl
+has no option but to exit immediately.

Should $^M be mentioned here?

$^M doesn't change what happens--Perl always exits after emitting the
message.

Sarathy
gsar@​activestate.com

From [Unknown Contact. See original ticket]

Ian Phillipps writes​:

On Thu, 30 Sep 1999 at 11​:18​:05 -0700, Gurusamy Sarathy wrote​:

+=item Out of memory!
+
+(X) The malloc() function returned 0, indicating there was insufficient
+remaining memory (or virtual memory) to satisfy the request. Perl
+has no option but to exit immediately.

Should $^M be mentioned here?

Definitely not! This message comes from non-Perl-malloc. $^M is
active with Perl-malloc only.

Ilya

From @gsar

On Thu, 30 Sep 1999 21​:21​:15 EDT, Ilya Zakharevich wrote​:

+=item Out of memory!
[...]
This message comes from non-Perl-malloc.

malloc.c​:1016​: PerlIO_puts(PerlIO_stderr(),"Out of memory!\n");

Sarathy
gsar@​activestate.com

From [Unknown Contact. See original ticket]

Ilya Zakharevich <ilya@​math.ohio-state.edu> wrote

Definitely not! This message comes from non-Perl-malloc. $^M is
active with Perl-malloc only.

Errr... quoting from my original post​:

  alignbytes=8, usemymalloc=y, prototype=define
  ^^^^^^^^^^^^^

I presume this misconception is the reason why the message was erroneously
removed from perldiag.

Mike Guy

From [Unknown Contact. See original ticket]

On Fri, Oct 01, 1999 at 12​:49​:14AM -0700, Gurusamy Sarathy wrote​:

On Thu, 30 Sep 1999 21​:21​:15 EDT, Ilya Zakharevich wrote​:

+=item Out of memory!
[...]
This message comes from non-Perl-malloc.

malloc.c​:1016​: PerlIO_puts(PerlIO_stderr(),"Out of memory!\n");

????!!!!! Who wrote this junk?! I count until three, then start to
shoot every third guy around!

Ilya

P.S. I think this code is not reachable without exercising some of
  compile-time options. Nevertheless, it could be improved.

From [Unknown Contact. See original ticket]

Gurusamy Sarathy <gsar@​ActiveState.com> wrote

Presumably the last two really mean "Perl's internal data structures
are in a twist", and perhaps indicate bugs in Perl. Is it worth
seeing if I can generate a smaller test case?

Please do.

Turns out I can't reproduce either effect. Sorry.

Mike Guy