emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Hello, I just built a fresh trunk checkout on my WinXP machine. The 
binaries seem to work fine (though the annoying coding system warning is 
still present). I used --enable-font-backend during the configure step. 
However, those binaries crashes immediately under Windows Vista.
You can do emacs --version and emacs -help, the expected output is 
displayed. However, emacs -Q crashes immediately (and without -Q).

Under cygwin (on my vista machine), I tried:
$ strace emacs -Q
just to see what strace would say:
--- Process 3660, exception C00000005 at 77DF986B

was the only output.

- Eric Lilja

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Jason Rumney]

Eric Lilja wrote:
> Hello, I just built a fresh trunk checkout on my WinXP machine. The 
> binaries seem to work fine (though the annoying coding system warning 
> is still present). I used --enable-font-backend during the configure 
> step. However, those binaries crashes immediately under Windows Vista.
> You can do emacs --version and emacs -help, the expected output is 
> displayed. However, emacs -Q crashes immediately (and without -Q).
Please run emacs under gdb to find where it crashes.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Jason Rumney wrote:
> Please run emacs under gdb to find where it crashes.
> 
> 
> 
> 

Does this help?

$ gdb emacs
GNU gdb 6.5.50.20060706-cvs (cygwin-special)
Copyright (C) 2006 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain 
conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "i686-pc-cygwin"...(no debugging symbols found)

(gdb) run
Starting program: /cygdrive/c/emacs-23/bin/emacs.exe
Loaded symbols for /cygdrive/c/Windows/system32/ntdll.dll
Loaded symbols for /cygdrive/c/Windows/system32/kernel32.dll
Loaded symbols for /cygdrive/c/Windows/system32/msvcrt.dll
Loaded symbols for /cygdrive/c/Windows/system32/advapi32.dll
Loaded symbols for /cygdrive/c/Windows/system32/rpcrt4.dll
Loaded symbols for 
/cygdrive/c/Windows/WinSxS/x86_microsoft.windows.common-controls_6595b64144ccf1df_6.0.6000.16386_none_5d07289e07e1d100/comctl32.d
ll
Loaded symbols for /cygdrive/c/Windows/system32/gdi32.dll
Loaded symbols for /cygdrive/c/Windows/system32/user32.dll
Loaded symbols for /cygdrive/c/Windows/system32/shlwapi.dll
Loaded symbols for /cygdrive/c/Windows/system32/comdlg32.dll
Loaded symbols for /cygdrive/c/Windows/system32/shell32.dll
Loaded symbols for /cygdrive/c/Windows/system32/mpr.dll
Loaded symbols for /cygdrive/c/Windows/system32/ole32.dll
Loaded symbols for /cygdrive/c/Windows/system32/winmm.dll
Loaded symbols for /cygdrive/c/Windows/system32/oleaut32.dll
Loaded symbols for /cygdrive/c/Windows/system32/oleacc.dll
Loaded symbols for /cygdrive/c/Windows/system32/winspool.drv

Program received signal SIGSEGV, Segmentation fault.
0x77df986b in ?? ()
(gdb) bt
#0  0x77df986b in ?? ()
#1  0x01080f6b in init_user_info ()
#2  0x01083754 in init_environment ()
#3  0x0100260e in main ()
(gdb)

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Jason Rumney]

Eric Lilja wrote:
> Program received signal SIGSEGV, Segmentation fault.
> 0x77df986b in ?? ()
> (gdb) bt
> #0  0x77df986b in ?? ()
> #1  0x01080f6b in init_user_info ()
> #2  0x01083754 in init_environment ()
> #3  0x0100260e in main ()
> (gdb)

It helps a bit. I don't know whether its the absence of source/object 
files on the machine where this is being debugged, the flags used to 
build Emacs, or an incompatibility between mingw and cygwin but the 
backtrace is missing file and line number information. The file I can 
work out (w32.c), but if the line number was there it would be more 
apparent what was going wrong.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Jason Rumney skrev:
> Eric Lilja wrote:
>> Program received signal SIGSEGV, Segmentation fault.
>> 0x77df986b in ?? ()
>> (gdb) bt
>> #0  0x77df986b in ?? ()
>> #1  0x01080f6b in init_user_info ()
>> #2  0x01083754 in init_environment ()
>> #3  0x0100260e in main ()
>> (gdb)
> 
> It helps a bit. I don't know whether its the absence of source/object 
> files on the machine where this is being debugged, the flags used to 
> build Emacs, or an incompatibility between mingw and cygwin but the 
> backtrace is missing file and line number information. The file I can 
> work out (w32.c), but if the line number was there it would be more 
> apparent what was going wrong.
> 
> 
> 
> 
> 
> 

I just remembered that my build script passes --no-debug during the 
configure step. Would it be worthwhile to rebuild without it?

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Jason Rumney]

Eric Lilja wrote:
> I just remembered that my build script passes --no-debug during the 
> configure step. Would it be worthwhile to rebuild without it?

Yes. Using --no-debug for a development copy is generally a bad idea. 
Even for release builds, the savings in size and speed are insignificant 
on modern hardware.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Jason Rumney skrev:
> Eric Lilja wrote:
>> I just remembered that my build script passes --no-debug during the 
>> configure step. Would it be worthwhile to rebuild without it?
> 
> Yes. Using --no-debug for a development copy is generally a bad idea. 
> Even for release builds, the savings in size and speed are insignificant 
> on modern hardware.
> 
> 
> 
> 
> 

Here we go. Also, this time I added -Q when performing "run" under gdb:
Program received signal SIGSEGV, Segmentation fault.
0x77df986b in ?? ()
(gdb) bt
#0  0x77df986b in ?? ()
#1  0x01080f6b in init_user_info () at w32.c:619
#2  0x01083754 in init_environment (argv=0x8828f0) at w32.c:1297
#3  0x0100260e in main (argc=2, argv=0x8828f0) at emacs.c:1506
(gdb)

At least we have line numbers now.

- Eric

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eli Zaretskii]

> From: Eric Lilja <mindcooler@gmail.com>
> Date: Mon, 11 Feb 2008 14:10:52 +0100
> 
> Here we go. Also, this time I added -Q when performing "run" under gdb:
> Program received signal SIGSEGV, Segmentation fault.
> 0x77df986b in ?? ()
> (gdb) bt
> #0  0x77df986b in ?? ()
> #1  0x01080f6b in init_user_info () at w32.c:619
> #2  0x01083754 in init_environment (argv=0x8828f0) at w32.c:1297
> #3  0x0100260e in main (argc=2, argv=0x8828f0) at emacs.c:1506
> (gdb)
> 
> At least we have line numbers now.

Yes.  Can you please step into get_sid_sub_authority_count, and see
what happens there?  Does it fail to load advapi32.dll, perhaps?  Or
does it succeed, but the return value of get_sid_sub_authority_count
is NULL, and then dereferencing it in init_user_info crashes?

Thanks.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Eli Zaretskii wrote:
>> From: Eric Lilja <mindcooler@gmail.com>
>> Date: Mon, 11 Feb 2008 14:10:52 +0100
>>
>> Here we go. Also, this time I added -Q when performing "run" under gdb:
>> Program received signal SIGSEGV, Segmentation fault.
>> 0x77df986b in ?? ()
>> (gdb) bt
>> #0  0x77df986b in ?? ()
>> #1  0x01080f6b in init_user_info () at w32.c:619
>> #2  0x01083754 in init_environment (argv=0x8828f0) at w32.c:1297
>> #3  0x0100260e in main (argc=2, argv=0x8828f0) at emacs.c:1506
>> (gdb)
>>
>> At least we have line numbers now.
> 
> Yes.  Can you please step into get_sid_sub_authority_count, and see
> what happens there?  Does it fail to load advapi32.dll, perhaps?  Or
> does it succeed, but the return value of get_sid_sub_authority_count
> is NULL, and then dereferencing it in init_user_info crashes?
> 
> Thanks.
> 
> 
> 

I'm doing a fresh build right now (I wonder if I should have disabled 
the new font backend this time to see what happened, oh well). Could you 
explain how exactly I would do that in gdb? If I copy the source along 
with the binaries, will gdb be able to find it and I can put a 
breakpoint in there? I usually copy what gets put in the directory when 
I do a make install.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eli Zaretskii]

> From: Eric Lilja <mindcooler@gmail.com>
> Date: Mon, 11 Feb 2008 23:57:00 +0100
> 
> I'm doing a fresh build right now (I wonder if I should have disabled 
> the new font backend this time to see what happened, oh well).

The default is to build with font backend disabled, so if you didn't
_enable_ it, you are building without it.

> Could you explain how exactly I would do that in gdb?

From the src directory:

  gdb ./oo-spd/i386/emacs.exe
  (gdb) break init_user_info
  (gdb) run -Q

Then, when it stops in init_user_info, type:

  (gdb) until 619

(619 is the number of the line where it calls
get_sid_sub_authority_count.)

When it stops at line 619, type "step" repeatedly to step through
get_sid_sub_authority_count, and see what happens there.

If you compiled with optimizations, it could be that
get_sid_sub_authority_count is inlined, and you will have difficulties
stepping through it.  If that happens, please recompile without
optimizations (edit gmake.defs to remove the -U2 compiler switch, then
run configure, make, make install to rebuild).

> If I copy the source along with the binaries, will gdb be able to
> find it and I can put a breakpoint in there?

If it doesn't find the sources, use the `dir' command to tell GDB
where to find the sources.  This command accepts a name of a directory
and adds it to the list of directories where GDB looks for sources.

Btw, is that a 32-bit Vista or a 64-bit one?

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Eli Zaretskii wrote:
>> From: Eric Lilja <mindcooler@gmail.com>
>> Date: Mon, 11 Feb 2008 23:57:00 +0100
>>
>> I'm doing a fresh build right now (I wonder if I should have disabled 
>> the new font backend this time to see what happened, oh well).
> 
> The default is to build with font backend disabled, so if you didn't
> _enable_ it, you are building without it.

Yes, sorry.

> 
>> Could you explain how exactly I would do that in gdb?
> 
> From the src directory:
> 
>   gdb ./oo-spd/i386/emacs.exe
>   (gdb) break init_user_info
>   (gdb) run -Q
> 
> Then, when it stops in init_user_info, type:
> 
>   (gdb) until 619
> 
> (619 is the number of the line where it calls
> get_sid_sub_authority_count.)
> 
> When it stops at line 619, type "step" repeatedly to step through
> get_sid_sub_authority_count, and see what happens there.
> 
> If you compiled with optimizations, it could be that
> get_sid_sub_authority_count is inlined, and you will have difficulties
> stepping through it.  If that happens, please recompile without
> optimizations (edit gmake.defs to remove the -U2 compiler switch, then
> run configure, make, make install to rebuild).
> 
>> If I copy the source along with the binaries, will gdb be able to
>> find it and I can put a breakpoint in there?
> 
> If it doesn't find the sources, use the `dir' command to tell GDB
> where to find the sources.  This command accepts a name of a directory
> and adds it to the list of directories where GDB looks for sources.
> 
> Btw, is that a 32-bit Vista or a 64-bit one?
> 
> 
> 

It's a 32-bit vista. Here's my new gdb session:

$ gdb ./oo-spd/i386/emacs.exe
GNU gdb 6.5.50.20060706-cvs (cygwin-special)
Copyright (C) 2006 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain 
conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "i686-pc-cygwin"...
Environment variable "DISPLAY" not defined.
TERM = cygwin
Breakpoint 1 at 0x114ad94: file w32fns.c, line 9317.
Breakpoint 2 at 0x10a01d4: file sysdep.c, line 1319.
(gdb) break init_user_info
Breakpoint 3 at 0x1080d4c: file w32.c, line 598.
(gdb) run -Q
Starting program: /home/mindcooler/emacs/src/oo-spd/i386/emacs.exe -Q
Loaded symbols for /cygdrive/c/Windows/system32/ntdll.dll
Loaded symbols for /cygdrive/c/Windows/system32/kernel32.dll
Loaded symbols for /cygdrive/c/Windows/system32/msvcrt.dll
Loaded symbols for /cygdrive/c/Windows/system32/advapi32.dll
Loaded symbols for /cygdrive/c/Windows/system32/rpcrt4.dll
Loaded symbols for 
/cygdrive/c/Windows/WinSxS/x86_microsoft.windows.common-controls_6595b64144ccf1df_6.0.6000.16386_none_5d07289e07e1d100/comctl32.d
ll
Loaded symbols for /cygdrive/c/Windows/system32/gdi32.dll
Loaded symbols for /cygdrive/c/Windows/system32/user32.dll
Loaded symbols for /cygdrive/c/Windows/system32/shlwapi.dll
Loaded symbols for /cygdrive/c/Windows/system32/comdlg32.dll
Loaded symbols for /cygdrive/c/Windows/system32/shell32.dll
Loaded symbols for /cygdrive/c/Windows/system32/mpr.dll
Loaded symbols for /cygdrive/c/Windows/system32/ole32.dll
Loaded symbols for /cygdrive/c/Windows/system32/winmm.dll
Loaded symbols for /cygdrive/c/Windows/system32/oleaut32.dll
Loaded symbols for /cygdrive/c/Windows/system32/oleacc.dll
Loaded symbols for /cygdrive/c/Windows/system32/winspool.drv

Breakpoint 3, init_user_info () at w32.c:598
598       DWORD        length = sizeof (name), dlength = sizeof 
(domain), trash;
(gdb) until 619
init_user_info () at w32.c:619
619               DWORD n_subauthorities =
(gdb) step
get_sid_sub_authority_count (pSid=0x82f288) at w32.c:388
388     {
(gdb) step
392       if (is_windows_9x () == TRUE)
(gdb) step
is_windows_9x () at w32.c:179
179       if (g_b_init_is_windows_9x == 0)
(gdb) step
190     }
(gdb) step
get_sid_sub_authority_count (pSid=0x82f288) at w32.c:396
396       if (g_b_init_get_sid_sub_authority_count == 0)
(gdb) step
404       if (s_pfn_Get_Sid_Sub_Authority_Count == NULL)
(gdb) step
408       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
(gdb) step
409     }
(gdb) step
get_sid_sub_authority_count (pSid=0x0) at w32.c:408
408       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
(gdb) step
0x77df986b in ntdll!_fltused ()
(gdb) step
Single stepping until exit from function ntdll!_fltused,
which has no line number information.

Program received signal SIGSEGV, Segmentation fault.
0x77df9871 in ntdll!_fltused ()
(gdb)

Hope this helps!

- Eric

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Richard Stallman]

Since we're talking about running Emacs under the nastiest PC
operating system ever seen, I think it is relevant to direct
people's attention to BadVista.org.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eli Zaretskii]

> From: Eric Lilja <mindcooler@gmail.com>
> Date: Tue, 12 Feb 2008 10:49:29 +0100
> 
> init_user_info () at w32.c:619
> 619               DWORD n_subauthorities =
> (gdb) step
> get_sid_sub_authority_count (pSid=0x82f288) at w32.c:388
> 388     {
> (gdb) step
> 392       if (is_windows_9x () == TRUE)
> (gdb) step
> is_windows_9x () at w32.c:179
> 179       if (g_b_init_is_windows_9x == 0)
> (gdb) step
> 190     }
> (gdb) step
> get_sid_sub_authority_count (pSid=0x82f288) at w32.c:396
> 396       if (g_b_init_get_sid_sub_authority_count == 0)
> (gdb) step
> 404       if (s_pfn_Get_Sid_Sub_Authority_Count == NULL)
> (gdb) step
> 408       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
> (gdb) step
> 409     }
> (gdb) step
> get_sid_sub_authority_count (pSid=0x0) at w32.c:408
> 408       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
> (gdb) step
> 0x77df986b in ntdll!_fltused ()
> (gdb) step
> Single stepping until exit from function ntdll!_fltused,
> which has no line number information.
> 
> Program received signal SIGSEGV, Segmentation fault.
> 0x77df9871 in ntdll!_fltused ()
> (gdb)

Thanks.

Would it be possible for you to build Emacs without the change below,
and then see if it works on Vista?  TIA

  2008-01-26  Eli Zaretskii  <eliz@gnu.org>

	  * w32.c (g_b_init_get_sid_sub_authority)
	  (g_b_init_get_sid_sub_authority_count): New static variables.
	  (GetSidSubAuthority_Proc, GetSidSubAuthorityCount_Proc): New typedefs.
	  (get_sid_sub_authority, get_sid_sub_authority_count): New functions.
	  (init_user_info): Use them to retrieve uid and gid.
	  Use 500/513, the Windows defaults, as Administrator's uid/gid.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Eli Zaretskii skrev:
>> From: Eric Lilja <mindcooler@gmail.com>
>> Date: Tue, 12 Feb 2008 10:49:29 +0100
>>
>> init_user_info () at w32.c:619
>> 619               DWORD n_subauthorities =
>> (gdb) step
>> get_sid_sub_authority_count (pSid=0x82f288) at w32.c:388
>> 388     {
>> (gdb) step
>> 392       if (is_windows_9x () == TRUE)
>> (gdb) step
>> is_windows_9x () at w32.c:179
>> 179       if (g_b_init_is_windows_9x == 0)
>> (gdb) step
>> 190     }
>> (gdb) step
>> get_sid_sub_authority_count (pSid=0x82f288) at w32.c:396
>> 396       if (g_b_init_get_sid_sub_authority_count == 0)
>> (gdb) step
>> 404       if (s_pfn_Get_Sid_Sub_Authority_Count == NULL)
>> (gdb) step
>> 408       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
>> (gdb) step
>> 409     }
>> (gdb) step
>> get_sid_sub_authority_count (pSid=0x0) at w32.c:408
>> 408       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
>> (gdb) step
>> 0x77df986b in ntdll!_fltused ()
>> (gdb) step
>> Single stepping until exit from function ntdll!_fltused,
>> which has no line number information.
>>
>> Program received signal SIGSEGV, Segmentation fault.
>> 0x77df9871 in ntdll!_fltused ()
>> (gdb)
> 
> Thanks.
> 
> Would it be possible for you to build Emacs without the change below,
> and then see if it works on Vista?  TIA
> 
>   2008-01-26  Eli Zaretskii  <eliz@gnu.org>
> 
> 	  * w32.c (g_b_init_get_sid_sub_authority)
> 	  (g_b_init_get_sid_sub_authority_count): New static variables.
> 	  (GetSidSubAuthority_Proc, GetSidSubAuthorityCount_Proc): New typedefs.
> 	  (get_sid_sub_authority, get_sid_sub_authority_count): New functions.
> 	  (init_user_info): Use them to retrieve uid and gid.
> 	  Use 500/513, the Windows defaults, as Administrator's uid/gid.
> 
> 
> 
> 

Yes, if I just knew a clean way of undoing that change.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Juanma Barranquero]

On Feb 13, 2008 11:57 PM, Eric Lilja <mindcooler@gmail.com> wrote:

> Yes, if I just knew a clean way of undoing that change.

You can try updating to a previous version, for example, in src/ do

  cvs update -r 1.122 -p w32.c > w32.c

(the -p and redirection is to avoid the change being sticky, i.e.,
w32.c to be pegged to revision 1.122).

             Juanma

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Juanma Barranquero wrote:
> On Feb 13, 2008 11:57 PM, Eric Lilja <mindcooler@gmail.com> wrote:
> 
>> Yes, if I just knew a clean way of undoing that change.
> 
> You can try updating to a previous version, for example, in src/ do
> 
>   cvs update -r 1.122 -p w32.c > w32.c
> 
> (the -p and redirection is to avoid the change being sticky, i.e.,
> w32.c to be pegged to revision 1.122).
> 
>              Juanma
> 
> 
> 

Thanks, I'll try that as soon as the error introduced to coding.c is 
fixed so I can bootstrap again. :)

- Eric

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Juanma Barranquero]

On Feb 14, 2008 1:44 AM, Eric Lilja <mindcooler@gmail.com> wrote:

> Thanks, I'll try that as soon as the error introduced to coding.c is
> fixed so I can bootstrap again. :)

Just change the ASLOTs to AREF in the CODING_ATTR_* defines of coding.h

             Juanma

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Kenichi Handa]

In article <fp02t8$5cl$1@ger.gmane.org>, Eric Lilja <mindcooler@gmail.com> writes:

> Thanks, I'll try that as soon as the error introduced to coding.c is 
> fixed so I can bootstrap again. :)

Please explain what error is introduced to coding.c.  I'm
sorry but I have not followed the discussion of this
subject.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Juanma Barranquero wrote:
> On Feb 14, 2008 1:44 AM, Eric Lilja <mindcooler@gmail.com> wrote:
> 
>> Thanks, I'll try that as soon as the error introduced to coding.c is
>> fixed so I can bootstrap again. :)
> 
> Just change the ASLOTs to AREF in the CODING_ATTR_* defines of coding.h
> 
>              Juanma
> 
> 
> 

Thanks, bootstrapping now and enjoying a cup of tea while I wait.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Eric Lilja wrote:
> Juanma Barranquero wrote:
>> On Feb 14, 2008 1:44 AM, Eric Lilja <mindcooler@gmail.com> wrote:
>>
>>> Thanks, I'll try that as soon as the error introduced to coding.c is
>>> fixed so I can bootstrap again. :)
>>
>> Just change the ASLOTs to AREF in the CODING_ATTR_* defines of coding.h
>>
>>              Juanma
>>
>>
>>
> 
> Thanks, bootstrapping now and enjoying a cup of tea while I wait.
> 
> 
> 
> 

I am happy to report that Emacs does indeed not crash on Vista after 
reverting the changes to w32.c, using the method suggested by Juanma. 
Hopefully a proper fix (because I assume the changes I reverted have 
some purpose too) isn't too hard to do.

- Eric

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Jason Rumney]

Eric Lilja wrote:
> Eric Lilja wrote:
>> Juanma Barranquero wrote:
>>> On Feb 14, 2008 1:44 AM, Eric Lilja <mindcooler@gmail.com> wrote:
>>>
>>>> Thanks, I'll try that as soon as the error introduced to coding.c is
>>>> fixed so I can bootstrap again. :)
>>>
>>> Just change the ASLOTs to AREF in the CODING_ATTR_* defines of coding.h
>>>
>>>              Juanma
>>>
>>>
>>>
>>
>> Thanks, bootstrapping now and enjoying a cup of tea while I wait.
>>
>>
>>
>>
>
> I am happy to report that Emacs does indeed not crash on Vista after 
> reverting the changes to w32.c, using the method suggested by Juanma. 
> Hopefully a proper fix (because I assume the changes I reverted have 
> some purpose too) isn't too hard to do.

One possibility is that user_sid is not large enough to hold the sub 
authorities on Vista. One way to check this is to look at the value of 
trash in init_user_info just before the crash occurs. That should 
contain the size that was required. This code could all be written more 
robustly, checking the size and using alloca to temporarily allocate a 
larger buffer where necessary.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eli Zaretskii]

> From: Eric Lilja <mindcooler@gmail.com>
> Date: Thu, 14 Feb 2008 02:18:36 +0100
> 
> I am happy to report that Emacs does indeed not crash on Vista after 
> reverting the changes to w32.c, using the method suggested by Juanma. 
> Hopefully a proper fix (because I assume the changes I reverted have 
> some purpose too) isn't too hard to do.

Could you please compile the program below and run it on Vista with
some file name (a name of any existing file) as a single argument?
Then please post here everything it prints.  That might give me enough
information to fix the code.

Thanks.

-------------------------- cut here --------------------
#include <stdio.h>
#include <windows.h>
#include <lmcons.h>
#include <accctrl.h>
#include <aclapi.h>
#include <tchar.h>

BOOL GetTextualSid(
    PSID pSid,            // binary SID
    LPTSTR TextualSid,    // buffer for Textual representation of SID
    LPDWORD lpdwBufferLen // required/provided TextualSid buffersize
    )
{
    PSID_IDENTIFIER_AUTHORITY psia;
    DWORD dwSubAuthorities;
    DWORD dwSidRev=SID_REVISION;
    DWORD dwCounter;
    DWORD dwSidSize;

    // Validate the binary SID.

    if(!IsValidSid(pSid)) return FALSE;

    // Get the identifier authority value from the SID.

    psia = GetSidIdentifierAuthority(pSid);

    // Get the number of subauthorities in the SID.

    dwSubAuthorities = *GetSidSubAuthorityCount(pSid);

    // Compute the buffer length.
    // S-SID_REVISION- + IdentifierAuthority- + subauthorities- + NULL

    dwSidSize=(15 + 12 + (12 * dwSubAuthorities) + 1) * sizeof(TCHAR);

    // Check input buffer length.
    // If too small, indicate the proper size and set the last error.

    if (*lpdwBufferLen < dwSidSize)
    {
        *lpdwBufferLen = dwSidSize;
        SetLastError(ERROR_INSUFFICIENT_BUFFER);
        return FALSE;
    }

    // Add 'S' prefix and revision number to the string.

    dwSidSize=_stprintf(TextualSid,
              TEXT("S-%lu-"), dwSidRev );

    // Add a SID identifier authority to the string.

    if ( (psia->Value[0] != 0) || (psia->Value[1] != 0) )
    {
        dwSidSize+=_stprintf(TextualSid + lstrlen(TextualSid),
                    TEXT("0x%02hx%02hx%02hx%02hx%02hx%02hx"),
                    (USHORT)psia->Value[0],
                    (USHORT)psia->Value[1],
                    (USHORT)psia->Value[2],
                    (USHORT)psia->Value[3],
                    (USHORT)psia->Value[4],
                    (USHORT)psia->Value[5]);
    }
    else
    {
        dwSidSize+=_stprintf(TextualSid + lstrlen(TextualSid),
                    TEXT("%lu"),
                    (ULONG)(psia->Value[5]      )   +
                    (ULONG)(psia->Value[4] <<  8)   +
                    (ULONG)(psia->Value[3] << 16)   +
                    (ULONG)(psia->Value[2] << 24)   );
    }

    // Add SID subauthorities to the string.
    //
    for (dwCounter=0 ; dwCounter < dwSubAuthorities ; dwCounter++)
    {
        dwSidSize+=_stprintf(TextualSid + dwSidSize,
                    TEXT("-%lu"),
                    *GetSidSubAuthority(pSid, dwCounter) );
    }

    return TRUE;
}

int
main (int argc, char *argv[])
{
  SECURITY_INFORMATION si =
    OWNER_SECURITY_INFORMATION
    | GROUP_SECURITY_INFORMATION
    | DACL_SECURITY_INFORMATION;
  /* The magic value 244 is returned by GetFileSecurity if one uses
     less storage for the three flags above.  */
  SECURITY_DESCRIPTOR  *sd = malloc(244);
  const char *arg = (argv[1] ? argv[1] : argv[0]);
  DWORD len_needed;

  if (!GetFileSecurity (arg, si, sd, 244, &len_needed))
    fprintf (stderr, "GetFileSecurity: %lu (%lu/%u)\n",
	     GetLastError (), len_needed, 244);
  else
    {
      SID *psid, *pgsid;
      BOOL deflt, deflt1;
      char name[UNLEN + 1], gname[UNLEN + 1];
      SID_NAME_USE snu, sng;
      char domain[1024];
      DWORD nl = UNLEN + 1, gnl = UNLEN + 1;
      DWORD dl = 1024, gdl = 1024;
      char tusid[1024], tgsid[1024];
      DWORD tsidlen = sizeof(tusid);
      BOOL daclpresent;
      PACL pdacl;

      if (!GetSecurityDescriptorOwner (sd, (PSID *)&psid, &deflt))
	fprintf (stderr, "GetSecurityDescriptorOwner: %lu\n", GetLastError ());
      else if (!LookupAccountSid (NULL, psid, name, &nl,
				  domain, &dl, &snu))
	fprintf (stderr, "LookupAccountSid: %lu\n",
		 GetLastError ());
      else if (!GetTextualSid (psid, tusid, &tsidlen))
	fprintf (stderr, "GetTextualSid (owner): %lu\n", GetLastError ());
      else if (!GetSecurityDescriptorGroup (sd, (PSID *)&pgsid, &deflt))
	fprintf (stderr, "GetSecurityDescriptorGroup: %lu\n", GetLastError ());
      else if (!LookupAccountSid (NULL, pgsid, gname, &gnl,
				  domain, &gdl, &sng))
	fprintf (stderr, "LookupAccountSid: %lu\n",
		 GetLastError ());
      else if (!GetTextualSid (pgsid, tgsid, &tsidlen))
	fprintf (stderr, "GetTextualSid (group): %lu\n", GetLastError ());
      else if (!GetSecurityDescriptorDacl (sd, &daclpresent, &pdacl, &deflt1))
	fprintf (stderr, "GetSecurityDescriptorDacl: %lu\n", GetLastError ());
      else if (!daclpresent)
	fprintf (stderr, "GetSecurityDescriptorDacl: NO DACL\n");
      else
	{
	  ACCESS_MASK umask, gmask;
	  DWORD err;
	  TRUSTEE trustee = {
	    NULL, NO_MULTIPLE_TRUSTEE, TRUSTEE_IS_SID, TRUSTEE_IS_USER, NULL
	  };

	  trustee.ptstrName = (LPTSTR)psid;
	  if ((err = GetEffectiveRightsFromAcl (pdacl, &trustee, &umask)) != ERROR_SUCCESS)
	    fprintf (stderr, "GetEffectiveRightsFromAcl (owner): %lu\n", err);
	  else if (trustee.TrusteeType = TRUSTEE_IS_GROUP,
		   trustee.ptstrName = (LPTSTR)pgsid,
		   (err = GetEffectiveRightsFromAcl (pdacl, &trustee, &gmask)) != ERROR_SUCCESS)
	    fprintf (stderr, "GetEffectiveRightsFromAcl (group): %lu\n", err);
	  else
	    printf ("%s:\n         dflt=%d owner=%s grp=%s domain=%s use=%d\n         Ownr SID=%s 0x%lx\n         Grp  SID=%s 0x%lx\n",
		    arg, deflt, name, gname, domain, snu, tusid, umask, tgsid, gmask);
	}
    }
  return 0;
}

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eli Zaretskii]

> Date: Thu, 14 Feb 2008 08:32:17 +0000
> From: Jason Rumney <jasonr@gnu.org>
> Cc: emacs-devel@gnu.org
> 
> > I am happy to report that Emacs does indeed not crash on Vista after 
> > reverting the changes to w32.c, using the method suggested by Juanma. 
> > Hopefully a proper fix (because I assume the changes I reverted have 
> > some purpose too) isn't too hard to do.
> 
> One possibility is that user_sid is not large enough to hold the sub 
> authorities on Vista. One way to check this is to look at the value of 
> trash in init_user_info just before the crash occurs. That should 
> contain the size that was required. This code could all be written more 
> robustly, checking the size and using alloca to temporarily allocate a 
> larger buffer where necessary.

I never liked the way that code casts structures to `PSID *', as in

      lookup_account_sid (NULL, *((PSID *) user_sid), name, &length,

This smells of abusing some object by a subtly incompatible pointer.
But that's how that part of the code was written before me, and it
worked on XP and saved me yet another dance with a function that needs
to be called via a pointer after loading a DLL.

Hopefully, the program I sent earlier, if it doesn't crash on Vista,
will give me enough information to rewrite that code in a more robust
way.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Eli Zaretskii wrote:
>> From: Eric Lilja <mindcooler@gmail.com>
>> Date: Thu, 14 Feb 2008 02:18:36 +0100
>>
>> I am happy to report that Emacs does indeed not crash on Vista after 
>> reverting the changes to w32.c, using the method suggested by Juanma. 
>> Hopefully a proper fix (because I assume the changes I reverted have 
>> some purpose too) isn't too hard to do.
> 
> Could you please compile the program below and run it on Vista with
> some file name (a name of any existing file) as a single argument?
> Then please post here everything it prints.  That might give me enough
> information to fix the code.
> 
> Thanks.
> 

I had to comment out #include <tchar.h> to get it to compile under 
cygwin's gcc in mingw mode. Here's the output from both mingw and visual 
c++ 9:
mingw:
$ ./runme.exe source.c
source.c:
          dflt=0 owner=Administrat÷rer grp=None domain=mindcool-laptop use=4
          Ownr SID=S-1-5-32-544 0x1f01ff
          Grp  SID=S-1-5-21-2313625805-1149499212-3830963393-513 0x1201ff



vs9:
$ ./source.exe ../../source.c
../../source.c:
          dflt=0 owner=Administrat÷rer grp=None domain=mindcool-laptop use=4
          Ownr SID=S-1-5-32-544 0x1f01ff
          Grp  SID=S-1-5-21-2313625805-1149499212-3830963393-513 0x1201ff

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Hanchrow]

>>>>> "Eric" == Eric Lilja <mindcooler@gmail.com> writes:

    Eric> However, those binaries crashes immediately under Windows
    Eric> Vista.  

(Greetings, fellow co-conspirator (http://catb.org/~esr/ecsl/))

What do you actually see when it crashes?  I ask because _sometimes_
when I start Emacs, I see a popup that says something like "Emacs has
stopped working".  The popup has a progress bar which moves left and
right to give the illusion that Windows is looking for a solution to the
problem.

If you're seeing the same thing, then maybe the fix for your problem
will also fix mine ...

-- 
If you're trying to choose between two theories and one gives
you an excuse for being lazy, the other one is probably right.
        -- Paul Graham

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eli Zaretskii]

> From: Eric Lilja <mindcooler@gmail.com>
> Date: Thu, 14 Feb 2008 21:12:06 +0100
> 
> mingw:
> $ ./runme.exe source.c
> source.c:
>           dflt=0 owner=Administrat÷rer grp=None domain=mindcool-laptop use=4
>           Ownr SID=S-1-5-32-544 0x1f01ff
>           Grp  SID=S-1-5-21-2313625805-1149499212-3830963393-513 0x1201ff
> 
> 
> 
> vs9:
> $ ./source.exe ../../source.c
> ../../source.c:
>           dflt=0 owner=Administrat÷rer grp=None domain=mindcool-laptop use=4
>           Ownr SID=S-1-5-32-544 0x1f01ff
>           Grp  SID=S-1-5-21-2313625805-1149499212-3830963393-513 0x1201ff

Thanks, I will try to fix w32.c when I have time, hopefully soon.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eli Zaretskii]

> Date: Thu, 14 Feb 2008 23:44:36 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: emacs-devel@gnu.org
> 
> Thanks, I will try to fix w32.c when I have time, hopefully soon.

Could you please try again with the latest version in CVS?

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Eli Zaretskii wrote:
>> Date: Thu, 14 Feb 2008 23:44:36 +0200
>> From: Eli Zaretskii <eliz@gnu.org>
>> Cc: emacs-devel@gnu.org
>>
>> Thanks, I will try to fix w32.c when I have time, hopefully soon.
> 
> Could you please try again with the latest version in CVS?
> 
> 
> 

Still crashes I'm afraid.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eli Zaretskii]

> From: Eric Lilja <mindcooler@gmail.com>
> Date: Sat, 16 Feb 2008 23:10:23 +0100
> 
> Eli Zaretskii wrote:
> >> Date: Thu, 14 Feb 2008 23:44:36 +0200
> >> From: Eli Zaretskii <eliz@gnu.org>
> >> Cc: emacs-devel@gnu.org
> >>
> >> Thanks, I will try to fix w32.c when I have time, hopefully soon.
> > 
> > Could you please try again with the latest version in CVS?
> > 
> > 
> > 
> 
> Still crashes I'm afraid.

Hmm... there's something I'm missing here, obviously.

Could you please again step through the call to
get_sid_sub_authority_count, and print the value of pSid at each step?
The command "display/x pSid" should do that; please type this command
as soon as you step into get_sid_sub_authority_count, and then
continue with the "next" command until the crash.  I also need the to
see value of user_token.User.Sid inside init_user_info, immediately
before the call to get_sid_sub_authority_count.

Thanks, and sorry for the mess.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Eli Zaretskii wrote:
>> From: Eric Lilja <mindcooler@gmail.com>
>> Date: Sat, 16 Feb 2008 23:10:23 +0100
>>
>> Eli Zaretskii wrote:
>>>> Date: Thu, 14 Feb 2008 23:44:36 +0200
>>>> From: Eli Zaretskii <eliz@gnu.org>
>>>> Cc: emacs-devel@gnu.org
>>>>
>>>> Thanks, I will try to fix w32.c when I have time, hopefully soon.
>>> Could you please try again with the latest version in CVS?
>>>
>>>
>>>
>> Still crashes I'm afraid.
> 
> Hmm... there's something I'm missing here, obviously.
> 
> Could you please again step through the call to
> get_sid_sub_authority_count, and print the value of pSid at each step?
> The command "display/x pSid" should do that; please type this command
> as soon as you step into get_sid_sub_authority_count, and then
> continue with the "next" command until the crash.  I also need the to
> see value of user_token.User.Sid inside init_user_info, immediately
> before the call to get_sid_sub_authority_count.
> 
> Thanks, and sorry for the mess.
> 
> 
> 

As you know, I'm not very good at gdb. I had some problems when 
debugging this (and here I'm refering to the "Cannot find bounds of 
current function", see below). Hope this log helps, if not, I can 
attempt to redo it in a proper way, with your guidance.


$ gdb ./oo-spd/i386/emacs.exe
GNU gdb 6.5.50.20060706-cvs (cygwin-special)
Copyright (C) 2006 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain 
conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "i686-pc-cygwin"...
Environment variable "DISPLAY" not defined.
TERM = cygwin
Breakpoint 1 at 0x114b0c4: file w32fns.c, line 9317.
Breakpoint 2 at 0x10a0434: file sysdep.c, line 1319.
(gdb) break init_user_info
Breakpoint 3 at 0x1080f9c: file w32.c, line 599.
(gdb) run -Q
Starting program: /home/mindcooler/emacs/src/oo-spd/i386/emacs.exe -Q
Loaded symbols for /cygdrive/c/Windows/system32/ntdll.dll
Loaded symbols for /cygdrive/c/Windows/system32/kernel32.dll
Loaded symbols for /cygdrive/c/Windows/system32/msvcrt.dll
Loaded symbols for /cygdrive/c/Windows/system32/advapi32.dll
Loaded symbols for /cygdrive/c/Windows/system32/rpcrt4.dll
Loaded symbols for 
/cygdrive/c/Windows/WinSxS/x86_microsoft.windows.common-controls_6595b64144ccf1df_6.0.6000.16386_none_5d07289e07e1d100/comctl32.dll
Loaded symbols for /cygdrive/c/Windows/system32/gdi32.dll
Loaded symbols for /cygdrive/c/Windows/system32/user32.dll
Loaded symbols for /cygdrive/c/Windows/system32/shlwapi.dll
Loaded symbols for /cygdrive/c/Windows/system32/comdlg32.dll
Loaded symbols for /cygdrive/c/Windows/system32/shell32.dll
Loaded symbols for /cygdrive/c/Windows/system32/mpr.dll
Loaded symbols for /cygdrive/c/Windows/system32/ole32.dll
Loaded symbols for /cygdrive/c/Windows/system32/winmm.dll
Loaded symbols for /cygdrive/c/Windows/system32/oleaut32.dll
Loaded symbols for /cygdrive/c/Windows/system32/oleacc.dll
Loaded symbols for /cygdrive/c/Windows/system32/winspool.drv

Breakpoint 3, init_user_info () at w32.c:599
599       DWORD        length = sizeof (name), dlength = sizeof 
(domain), trash;
(gdb) n
600       HANDLE       token = NULL;
(gdb) n
606       if (open_process_token (GetCurrentProcess (), TOKEN_QUERY, &token)
(gdb) n
613           strcpy (the_passwd.pw_name, name);
(gdb) s
615           if (stricmp ("administrator", name) == 0)
(gdb) s
624               DWORD n_subauthorities =
(gdb) print user_token.User.Sid
$1 = (PSID) 0x82ea68
(gdb) s
get_sid_sub_authority_count (pSid=0x82ea68) at w32.c:389
389     {
(gdb) display/x pSid
1: /x pSid = 0x82f270
(gdb) n
393       if (is_windows_9x () == TRUE)
1: /x pSid = 0x82ea68
(gdb) n
397       if (g_b_init_get_sid_sub_authority_count == 0)
1: /x pSid = 0x82ea68
(gdb) n
405       if (s_pfn_Get_Sid_Sub_Authority_Count == NULL)
1: /x pSid = 0x82ea68
(gdb) n
409       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
1: /x pSid = 0x82ea68
(gdb) n
410     }
1: /x pSid = 0x82ea68
(gdb) n
get_sid_sub_authority_count (pSid=0x0) at w32.c:409
409       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
1: /x pSid = 0x82f270
(gdb) n
0x77df986b in ?? ()
(gdb) n
Cannot find bounds of current function
(gdb) n
Cannot find bounds of current function
(gdb) n
Cannot find bounds of current function
(gdb) s
Cannot find bounds of current function
(gdb) step out
No symbol "out" in current context.
(gdb) n
Cannot find bounds of current function
(gdb) continue
Continuing.

Program received signal SIGSEGV, Segmentation fault.
0x77df986b in ?? ()
(gdb) bt
#0  0x77df986b in ?? ()
#1  0x010811cd in init_user_info () at w32.c:624
#2  0x010839b0 in init_environment (argv=0x922960) at w32.c:1303
#3  0x0100260e in main (argc=2, argv=0x922960) at emacs.c:1506
(gdb)

This was with a fresh checkout performed just an hour ago.

- Eric

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eli Zaretskii]

> From: Eric Lilja <mindcooler@gmail.com>
> Date: Sun, 17 Feb 2008 14:09:33 +0100
> 
> 405       if (s_pfn_Get_Sid_Sub_Authority_Count == NULL)
> 1: /x pSid = 0x82ea68
> (gdb) n
> 409       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
> 1: /x pSid = 0x82ea68

At this point, i.e. _before_ line 409, please type these commands, and
tell me what they printed:

  p ((SID *)pSid)->Revision
  p ((SID *)pSid)->SubAuthorityCount
  p *((SID *)pSid)->SubAuthority@7

Also, I see that you run Emacs on the Vista as administrator.  Could
you please try running it as an ordinary user, and see if it still
crashes?  For that matter, does it work on XP if you run Emacs as
administrator?

Thanks again for helping me debug this.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Eli Zaretskii wrote:
>> From: Eric Lilja <mindcooler@gmail.com>
>> Date: Sun, 17 Feb 2008 14:09:33 +0100
>>
>> 405       if (s_pfn_Get_Sid_Sub_Authority_Count == NULL)
>> 1: /x pSid = 0x82ea68
>> (gdb) n
>> 409       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
>> 1: /x pSid = 0x82ea68
> 
> At this point, i.e. _before_ line 409, please type these commands, and
> tell me what they printed:
> 
>   p ((SID *)pSid)->Revision
>   p ((SID *)pSid)->SubAuthorityCount
>   p *((SID *)pSid)->SubAuthority@7

Hope I got this right:
(gdb) n
405       if (s_pfn_Get_Sid_Sub_Authority_Count == NULL)
(gdb) p ((SID *)pSid)->Revision
$1 = 1 '\001'
(gdb) p ((SID *)pSid)->SubAuthorityCount
$2 = 5 '\005'
(gdb) p ((SID *)pSid)->SubAuthority@7
$3 = {{21}, {2313625805}, {1149499212}, {3830963393}, {1000}, 
{59246848}, {8579528}}
(gdb)

> 
> Also, I see that you run Emacs on the Vista as administrator.  Could
> you please try running it as an ordinary user, and see if it still
> crashes?  For that matter, does it work on XP if you run Emacs as
> administrator?
> 
> Thanks again for helping me debug this.
> 
> 
> 

I will have to get back to you on that one, I'm not familiar with 
setting up other users under Vista. In fact, I'm not very familiar with 
Vista at all. I just got the laptop that came with it.
However, I am running XP using an administrator account.

Hope this information helps in the meantime.

- Eric

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Eric Lilja wrote:
> Eli Zaretskii wrote:
>>> From: Eric Lilja <mindcooler@gmail.com>
>>> Date: Sun, 17 Feb 2008 14:09:33 +0100
>>>
>>> 405       if (s_pfn_Get_Sid_Sub_Authority_Count == NULL)
>>> 1: /x pSid = 0x82ea68
>>> (gdb) n
>>> 409       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
>>> 1: /x pSid = 0x82ea68
>>
>> At this point, i.e. _before_ line 409, please type these commands, and
>> tell me what they printed:
>>
>>   p ((SID *)pSid)->Revision
>>   p ((SID *)pSid)->SubAuthorityCount
>>   p *((SID *)pSid)->SubAuthority@7
> 
> Hope I got this right:
> (gdb) n
> 405       if (s_pfn_Get_Sid_Sub_Authority_Count == NULL)
> (gdb) p ((SID *)pSid)->Revision
> $1 = 1 '\001'
> (gdb) p ((SID *)pSid)->SubAuthorityCount
> $2 = 5 '\005'
> (gdb) p ((SID *)pSid)->SubAuthority@7
> $3 = {{21}, {2313625805}, {1149499212}, {3830963393}, {1000}, 
> {59246848}, {8579528}}
> (gdb)
> 
>>
>> Also, I see that you run Emacs on the Vista as administrator.  Could
>> you please try running it as an ordinary user, and see if it still
>> crashes?  For that matter, does it work on XP if you run Emacs as
>> administrator?
>>
>> Thanks again for helping me debug this.
>>
>>
>>
> 
> I will have to get back to you on that one, I'm not familiar with 
> setting up other users under Vista. In fact, I'm not very familiar with 
> Vista at all. I just got the laptop that came with it.
> However, I am running XP using an administrator account.

I created a normal user account, didn't try debugging or anything, just 
tried to start emacs with -Q and it crashes.

> 
> Hope this information helps in the meantime.
> 
> - Eric
> 
> 
> 
> 

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eli Zaretskii]

> From: Eric Lilja <mindcooler@gmail.com>
> Date: Sun, 17 Feb 2008 16:47:16 +0100
> 
> >> 409       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
> >> 1: /x pSid = 0x82ea68
> > 
> > At this point, i.e. _before_ line 409, please type these commands, and
> > tell me what they printed:
> > 
> >   p ((SID *)pSid)->Revision
> >   p ((SID *)pSid)->SubAuthorityCount
> >   p *((SID *)pSid)->SubAuthority@7
> 
> Hope I got this right:
> (gdb) n
> 405       if (s_pfn_Get_Sid_Sub_Authority_Count == NULL)
> (gdb) p ((SID *)pSid)->Revision
> $1 = 1 '\001'
> (gdb) p ((SID *)pSid)->SubAuthorityCount
> $2 = 5 '\005'
> (gdb) p ((SID *)pSid)->SubAuthority@7
> $3 = {{21}, {2313625805}, {1149499212}, {3830963393}, {1000}, 
> {59246848}, {8579528}}
> (gdb)

Almost: you forgot the leading asterisk in the last command, right
after "p":

  p *((SID *)pSid)->SubAuthority@7

Please tell me what this prints.

The rest looks good: the information is there, it's just that Vista
somehow refuses to produce it, or so it seems.  Hmm...

Thanks.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eli Zaretskii]

> From: Eric Lilja <mindcooler@gmail.com>
> Date: Sun, 17 Feb 2008 16:51:49 +0100
> 
> I created a normal user account, didn't try debugging or anything, just 
> tried to start emacs with -Q and it crashes.

This rejects yet another hypothesis.

Re: emacs 23.0.60.1 built today (11th feb 2008) crashes under Vista

[Eric Lilja]

Eli Zaretskii wrote:
>> From: Eric Lilja <mindcooler@gmail.com>
>> Date: Sun, 17 Feb 2008 16:47:16 +0100
>>
>>>> 409       return (s_pfn_Get_Sid_Sub_Authority_Count (pSid));
>>>> 1: /x pSid = 0x82ea68
>>> At this point, i.e. _before_ line 409, please type these commands, and
>>> tell me what they printed:
>>>
>>>   p ((SID *)pSid)->Revision
>>>   p ((SID *)pSid)->SubAuthorityCount
>>>   p *((SID *)pSid)->SubAuthority@7
>> Hope I got this right:
>> (gdb) n
>> 405       if (s_pfn_Get_Sid_Sub_Authority_Count == NULL)
>> (gdb) p ((SID *)pSid)->Revision
>> $1 = 1 '\001'
>> (gdb) p ((SID *)pSid)->SubAuthorityCount
>> $2 = 5 '\005'
>> (gdb) p ((SID *)pSid)->SubAuthority@7
>> $3 = {{21}, {2313625805}, {1149499212}, {3830963393}, {1000}, 
>> {59246848}, {8579528}}
>> (gdb)
> 
> Almost: you forgot the leading asterisk in the last command, right
> after "p":
> 
>   p *((SID *)pSid)->SubAuthority@7
> 
> Please tell me what this prints.
> 
> The rest looks good: the information is there, it's just that Vista
> somehow refuses to produce it, or so it seems.  Hmm...
> 
> Thanks.
> 
> 
> 

Sorry about that, here's what I get. Quite similar output:
(gdb) p *((SID *)pSid)->SubAuthority@7
$3 = {21, 2313625805, 1149499212, 3830963393, 1000, 60098816, 8579528}

I'm having a close watch on my email to see if there are any more tests 
you would like me to perform. Thank you very much for working on this.

- Eric