bug#13759: 24.3.50; bootstrap failed; bootstrap-emacs segfaults

bug#13759: 24.3.50; bootstrap failed; bootstrap-emacs segfaults

[Katsumi Yamaoka]

In GNU Emacs 24.3.50.1 (i686-pc-cygwin, X toolkit, Xaw3d scroll bars)
 of 2013-02-15 on localhost
(Bzr revision: 111784 jan.h.d@swipnet.se-20130214190112-35htkvic2zp1bwr2)
Windowing system distributor `The Cygwin/X Project', version 11.0.11302000
Configured using:
 `configure --verbose --with-x-toolkit=lucid --without-imagemagick
 --without-dbus --without-gconf --without-gsettings'

Can't build Emacs trunk on Cygwin (daily updated) on Windows 7:
[...]
Static heap usage: 9541024 of 13631488 bytes
43466 pure bytes used
cd ../lisp; make -w compile-first EMACS=/Work/emacs/src/bootstrap-emacs.exe
make[3]: Entering directory `/Work/emacs/lisp'
Compiling /Work/emacs/lisp/emacs-lisp/macroexp.el
Fatal error 11: Segmentation faultMakefile:243:\
 recipe for target `/Work/emacs/lisp/emacs-lisp/macroexp.elc' failed

Here's a backtrace:

% gdb src/bootstrap-emacs.exe
GNU gdb (GDB) 7.5.50.20120815-cvs (cygwin-special)
[...]
Reading symbols from /Work/emacs/src/bootstrap-emacs.exe...done.
(gdb) r
Starting program: /Work/emacs/src/bootstrap-emacs.exe
[New Thread 9364.0x1a3c]
[New Thread 9364.0x1d18]

Program received signal SIGSEGV, Segmentation fault.
mmap_alloc (var=0xf26d18, nbytes=15273) at buffer.c:4906
4906            r->next->prev = r;
(gdb) bt
#0  mmap_alloc (var=0xf26d18, nbytes=15273) at buffer.c:4906
#1  0x004d07b9 in mmap_realloc (nbytes=15273, var=0xf26d18) at buffer.c:4945
#2  enlarge_buffer_text (b=0xf26c00, delta=0) at buffer.c:5053
#3  0x004d1e16 in init_buffer () at buffer.c:5301
#4  0x004af6dd in main (argc=1, argv=0x22ac40) at emacs.c:1305
(gdb) 

Is any other information required?  Thanks.

bug#13759: 24.3.50; bootstrap failed; bootstrap-emacs segfaults

[Daniel Colascione]

[-- Attachment #1: Type: text/plain, Size: 3303 bytes --]

On 2/18/2013 4:06 PM, Katsumi Yamaoka wrote:
> Program received signal SIGSEGV, Segmentation fault.
> mmap_alloc (var=0xf26d18, nbytes=15273) at buffer.c:4906
> 4906            r->next->prev = r;
> (gdb) bt
> #0  mmap_alloc (var=0xf26d18, nbytes=15273) at buffer.c:4906
> #1  0x004d07b9 in mmap_realloc (nbytes=15273, var=0xf26d18) at buffer.c:4945
> #2  enlarge_buffer_text (b=0xf26c00, delta=0) at buffer.c:5053
> #3  0x004d1e16 in init_buffer () at buffer.c:5301
> #4  0x004af6dd in main (argc=1, argv=0x22ac40) at emacs.c:1305
> (gdb) 
> 
> Is any other information required?  Thanks.

The problem is that mmap_regions, the head of a linked list of memory
allocations we've made for buffers, is non-NULL in the dumped Emacs, but
obviously points to garbage that was valid only before Emacs was dumped. The
call to mmap_set_vars in Fdump_emacs is supposed to zero these variables before
we dump Emacs, but mmap_regions becomes non-NULL inside Fdump_emacs:

gdb) where
#0  mmap_alloc (var=0x7ca918, nbytes=6389) at buffer.c:4909
#1  0x004b0039 in mmap_realloc (nbytes=6389, var=0x7ca918) at buffer.c:4945
#2  enlarge_buffer_text (b=0x7ca800, delta=0) at buffer.c:5053
#3  0x004b0148 in set_buffer_internal_1 (b=0x7ca800) at buffer.c:2104
#4  0x004b068b in set_buffer_internal (b=<optimized out>) at buffer.h:1090
#5  Fset_buffer (buffer_or_name=8169477) at buffer.c:2195
#6  0x0041e47c in message_dolog (m=0x287110 "Static heap usage: 11072128 of
13631488 bytes\rC", nbytes=45, nlflag=true,
    multibyte=false) at xdisp.c:9389
#7  0x00430d50 in message3 (m=17698433) at xdisp.c:9598
#8  0x0054da7e in report_sheap_usage (die_if_pure_storage_exceeded=1) at sheap.c:94
#9  0x0052ec2a in unexec (outfile=0x121f934 "/users/dancol/edev/cyg/src/emacs",
infile=0x121f968 "/users/dancol/edev/cyg/src/temacs")
    at unexcw.c:287
#10 0x0048de13 in Fdump_emacs (filename=17698561, symfile=17698449) at emacs.c:2056
#11 0x004fb913 in eval_sub (form=11000526) at eval.c:2043
#12 0x004fbc8e in Fprogn (args=11000262) at eval.c:359
#13 0x004fba80 in eval_sub (form=10618670) at eval.c:1988
#14 0x004fba80 in eval_sub (form=10273462) at eval.c:1988
#15 0x0051cbd6 in readevalloop (readcharfun=10344786, stream=0x7ca60c,
sourcename=10459185, printflag=false, unibyte=10283034,
    readfun=10283034, start=10283034, end=10283034) at lread.c:1855
#16 0x0051d8c4 in Fload (file=10400673, noerror=10283034, nomessage=10283034,
nosuffix=10283034, must_suffix=10283034) at lread.c:1326
#17 0x004fb8b9 in eval_sub (form=10268918) at eval.c:2054
#18 0x004fd8b2 in Feval (form=10268918, lexical=10283034) at eval.c:1902
#19 0x0049240c in top_level_2 () at keyboard.c:1182
#20 0x004faac4 in internal_condition_case (bfun=0x4923f0 <top_level_2>,
handlers=10325114, hfun=0x493c90 <cmd_error>) at eval.c:1193
#21 0x00492c30 in top_level_1 (ignore=10283034) at keyboard.c:1190
#22 0x004fa9ea in internal_catch (tag=10314994, func=0x492bd0 <top_level_1>,
arg=10283034) at eval.c:964
#23 0x004937a8 in command_loop () at keyboard.c:1151
#24 recursive_edit_1 () at keyboard.c:785
#25 0x00493ab1 in Frecursive_edit () at keyboard.c:849
#26 0x0048f662 in main (argc=<optimized out>, argv=0x28ac20) at emacs.c:1556

I don't know why we suddenly started seeing this behavior.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 260 bytes --]

bug#13759: 24.3.50; bootstrap failed; bootstrap-emacs segfaults

[Daniel Colascione]

[-- Attachment #1: Type: text/plain, Size: 1910 bytes --]

On 2/19/2013 9:12 PM, Daniel Colascione wrote:
> I don't know why we suddenly started seeing this behavior.

Revision 111820, "Prefer `message1' over `message'.", caused us to start calling
message1 instead of message in sheap.c. message1 ends up calling message_dolog,
which tries to allocate a *Messages* buffer. Maybe Fdump_emacs should bind
Vmessage_max_log?

Anyway, I've installed the following patch to trunk, which fixes the problem. If
we want to go back to using message1, let's somehow make sure it doesn't log.

revno: 111830
committer: Daniel Colascione <dancol@dancol.org>
branch nick: cyg
timestamp: Tue 2013-02-19 21:43:53 -0800
message:
  2013-02-19  Daniel Colascione <dancol@dancol.org>

          * sheap.c (report_sheap_usage): Use message, not message1, so
         that we don't try to create a buffer while we're in the middle
         of dumping Emacs.  Explain why.
diff:
=== modified file 'src/ChangeLog'
--- src/ChangeLog       2013-02-19 03:29:28 +0000
+++ src/ChangeLog       2013-02-20 05:43:53 +0000
@@ -1,3 +1,9 @@
+2013-02-19  Daniel Colascione <dancol@dancol.org>
+
+       * sheap.c (report_sheap_usage): Use message, not message1, so
+       that we don't try to create a buffer while we're in the middle
+       of dumping Emacs.  Explain why.
+
 2013-02-19  Eli Zaretskii  <eliz@gnu.org>

        * w32proc.c (new_child): Avoid leaking handles if the subprocess

=== modified file 'src/sheap.c'
--- src/sheap.c 2013-02-17 16:49:27 +0000
+++ src/sheap.c 2013-02-20 05:43:53 +0000
@@ -91,5 +91,8 @@
   char buf[200];
   sprintf (buf, "Static heap usage: %d of %d bytes",
           bss_sbrk_ptr - bss_sbrk_buffer, STATIC_HEAP_SIZE);
-  message1 (buf);
+  /* Don't change this call to message1! message1 can log
+     messages, and at this point, we're not allowed to create
+     buffers.  */
+  message ("%s", buf);
 }



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 260 bytes --]

bug#13759: 24.3.50; bootstrap failed; bootstrap-emacs segfaults

[Katsumi Yamaoka]

Daniel Colascione wrote:
[...]
> Anyway, I've installed the following patch to trunk, which
> fixes the problem.  If we want to go back to using message1,
> let's somehow make sure it doesn't log.

Thanks a lot!  It enables me to play with ELisp again.

bug#13759: 24.3.50; bootstrap failed; bootstrap-emacs segfaults

[Katsumi Yamaoka]

Katsumi Yamaoka wrote:
> Daniel Colascione wrote:
> [...]
>> Anyway, I've installed the following patch to trunk, which
>> fixes the problem.  If we want to go back to using message1,
>> let's somehow make sure it doesn't log.

> Thanks a lot!  It enables me to play with ELisp again.

Stefan Monnier wrote:
> +	* sheap.c (report_sheap_usage): Prefer message1_nolog.

Bootstrap fails again on Cygwin/Windows7:

[...]
gcc -std=gnu99 -c -Demacs -I. -I/Work/emacs/src -I../lib\
 -I/Work/emacs/src/../lib -I/usr/include/freetype2 -D_REENTRANT\
 -I/usr/include/librsvg-2.0 -I/usr/include/glib-2.0\
 -I/usr/lib/glib-2.0/include -I/usr/include/gdk-pixbuf-2.0\
 -I/usr/include/cairo -I/usr/include/libpng15 -I/usr/include/pixman-1\
 -I/usr/include/freetype2 -I/usr/include/libxml2 -I/usr/include/freetype2\
 -MMD -MF deps/sheap.d -MP -I/usr/include/p11-kit-1 -g3 -O2 sheap.c
sheap.c: In function 'report_sheap_usage':
sheap.c:96:3: error: too many arguments to function 'message1_nolog'

What additional info should I provide?  Thanks.

bug#13759: 24.3.50; bootstrap failed; bootstrap-emacs segfaults

[Ken Brown]

On 2/20/2013 6:37 PM, Katsumi Yamaoka wrote:
> Stefan Monnier wrote:
>> +	* sheap.c (report_sheap_usage): Prefer message1_nolog.
>
> Bootstrap fails again on Cygwin/Windows7:
>
> [...]
> gcc -std=gnu99 -c -Demacs -I. -I/Work/emacs/src -I../lib\
>   -I/Work/emacs/src/../lib -I/usr/include/freetype2 -D_REENTRANT\
>   -I/usr/include/librsvg-2.0 -I/usr/include/glib-2.0\
>   -I/usr/lib/glib-2.0/include -I/usr/include/gdk-pixbuf-2.0\
>   -I/usr/include/cairo -I/usr/include/libpng15 -I/usr/include/pixman-1\
>   -I/usr/include/freetype2 -I/usr/include/libxml2 -I/usr/include/freetype2\
>   -MMD -MF deps/sheap.d -MP -I/usr/include/p11-kit-1 -g3 -O2 sheap.c
> sheap.c: In function 'report_sheap_usage':
> sheap.c:96:3: error: too many arguments to function 'message1_nolog'

I've fixed this in bzr revision 111839.

Ken

bug#13759: 24.3.50; bootstrap failed; bootstrap-emacs segfaults

[Katsumi Yamaoka]

Ken Brown wrote:
> On 2/20/2013 6:37 PM, Katsumi Yamaoka wrote:
>> Stefan Monnier wrote:
>>> +	* sheap.c (report_sheap_usage): Prefer message1_nolog.
>>
>> Bootstrap fails again on Cygwin/Windows7:
>>
>> [...]
>> gcc -std=gnu99 -c -Demacs -I. -I/Work/emacs/src -I../lib\
>>   -I/Work/emacs/src/../lib -I/usr/include/freetype2 -D_REENTRANT\
>>   -I/usr/include/librsvg-2.0 -I/usr/include/glib-2.0\
>>   -I/usr/lib/glib-2.0/include -I/usr/include/gdk-pixbuf-2.0\
>>   -I/usr/include/cairo -I/usr/include/libpng15 -I/usr/include/pixman-1\
>>   -I/usr/include/freetype2 -I/usr/include/libxml2 -I/usr/include/freetype2\
>>   -MMD -MF deps/sheap.d -MP -I/usr/include/p11-kit-1 -g3 -O2 sheap.c
>> sheap.c: In function 'report_sheap_usage':
>> sheap.c:96:3: error: too many arguments to function 'message1_nolog'

> I've fixed this in bzr revision 111839.

Confirmed.  Thank you so much.