emacs-20111024-r106182 Windows Binaries

emacs-20111024-r106182 Windows Binaries

[Christoph Scholtes]

The Emacs Trunk Windows Binaries (bzr revno 106182) have been published in

http://alpha.gnu.org/gnu/emacs/windows/

The binaries were built using the following libraries:
giflib-4.1.4-1
gnutls-2.10.1
jpeg-6b-4
libXpm-3.5.8
libpng-1.4.3-1
tiff-3.8.2-1
zlib-1.2.5-2

Please report any bugs that you come across via M-x report-emacs-bugs,
or email bug-gnu-emacs@gnu.org.  

For questions, email emacs-devel@gnu.org.

Re: emacs-20111024-r106182 Windows Binaries

[Leo]

On 2011-10-25 11:56 +0800, Christoph Scholtes wrote:
> The Emacs Trunk Windows Binaries (bzr revno 106182) have been published in
>
> http://alpha.gnu.org/gnu/emacs/windows/
>
> The binaries were built using the following libraries:
> giflib-4.1.4-1
> gnutls-2.10.1
> jpeg-6b-4
> libXpm-3.5.8
> libpng-1.4.3-1
> tiff-3.8.2-1
> zlib-1.2.5-2

Which GCC version? I have seen numerous crashes with the last windows
build which is compiled with gcc 4.6.1.

Leo

Re: emacs-20111024-r106182 Windows Binaries

[Christoph Scholtes]

Leo <sdl.web@gmail.com> writes:

> On 2011-10-25 11:56 +0800, Christoph Scholtes wrote:
>> The Emacs Trunk Windows Binaries (bzr revno 106182) have been published in
>>
>> http://alpha.gnu.org/gnu/emacs/windows/
>>
>> The binaries were built using the following libraries:
>> giflib-4.1.4-1
>> gnutls-2.10.1
>> jpeg-6b-4
>> libXpm-3.5.8
>> libpng-1.4.3-1
>> tiff-3.8.2-1
>> zlib-1.2.5-2
>
> Which GCC version? I have seen numerous crashes with the last windows
> build which is compiled with gcc 4.6.1.

I used gcc 4.6.1 (tdmmingw32).

What OS? Do you have any backtraces from the crashes?

I have been using the last build on a regular basis on Windows 7 without
problems.

Someone reported crashes on XP with gcc 4.6.1 compiling with -O2, but
the binaries are compiled without optimizations.

Re: emacs-20111024-r106182 Windows Binaries

[Leo]

On 2011-10-25 13:27 +0800, Christoph Scholtes wrote:
[snipped 16 lines]
>> Which GCC version? I have seen numerous crashes with the last windows
>> build which is compiled with gcc 4.6.1.
>
> I used gcc 4.6.1 (tdmmingw32).
>
> What OS? Do you have any backtraces from the crashes?

Microsoft Windows [version 6.1.7601]

> I have been using the last build on a regular basis on Windows 7
>without problems.
>
> Someone reported crashes on XP with gcc 4.6.1 compiling with -O2, but
> the binaries are compiled without optimizations.

Did the OS save the crash reports somewhere (sorry I have been using
windows OS for only 3 weeks and I don't have gdb running)? Otherwise I
can try and get a crash in a day or two.

Leo

Re: emacs-20111024-r106182 Windows Binaries

[Eli Zaretskii]

> From: Leo <sdl.web@gmail.com>
> Date: Tue, 25 Oct 2011 13:46:32 +0800
> Cc: help-emacs-windows@gnu.org
> 
> > Someone reported crashes on XP with gcc 4.6.1 compiling with -O2, but
> > the binaries are compiled without optimizations.
> 
> Did the OS save the crash reports somewhere (sorry I have been using
> windows OS for only 3 weeks and I don't have gdb running)?

Each application crash is recorded in the system log that you can view
using the "Event Viewer" (right-click on "My Computer", select
"Manage", then expand the "Event Viewer" widget and click on
"Application").

> Otherwise I can try and get a crash in a day or two.

Thanks, but there's no need to do that in this case, as the cause of
these crashes is known, see

  http://lists.gnu.org/archive/html/emacs-devel/2011-10/msg00871.html

This is now in etc/PROBLEMS.

However, I would advise to install GDB anyway, for when you need it
the next time.

Re: emacs-20111024-r106182 Windows Binaries

[Leo]

On 2011-10-25 16:00 +0800, Eli Zaretskii wrote:
> Each application crash is recorded in the system log that you can view
> using the "Event Viewer" (right-click on "My Computer", select
> "Manage", then expand the "Event Viewer" widget and click on
> "Application").

<Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
  <System>
    <Provider Name="Application Error" />
    <EventID Qualifiers="0">1000</EventID>
    <Level>2</Level>
    <Task>100</Task>
    <Keywords>0x80000000000000</Keywords>
    <TimeCreated SystemTime="2011-10-24T03:31:19.000000000Z" />
    <EventRecordID>1692</EventRecordID>
    <Channel>Application</Channel>
    <Computer>ZHAOXIN-PC</Computer>
    <Security />
  </System>
  <EventData>
    <Data>emacs.exe</Data>
    <Data>24.0.90.0</Data>
    <Data>4e9e1382</Data>
    <Data>ntdll.dll</Data>
    <Data>6.1.7601.17514</Data>
    <Data>4ce7ba58</Data>
    <Data>c0000029</Data>
    <Data>000907b6</Data>
    <Data>760</Data>
    <Data>01cc91fd0016c7e4</Data>
    <Data>D:\emacs-24.0.90\bin\emacs.exe</Data>
    <Data>C:\windows\SysWOW64\ntdll.dll</Data>
    <Data>a2f3ba8a-fdf0-11e0-9321-c0f8dac4b430</Data>
  </EventData>
</Event>

>
>> Otherwise I can try and get a crash in a day or two.
>
> Thanks, but there's no need to do that in this case, as the cause of
> these crashes is known, see
>
>   http://lists.gnu.org/archive/html/emacs-devel/2011-10/msg00871.html
>
> This is now in etc/PROBLEMS.
>
> However, I would advise to install GDB anyway, for when you need it
> the next time.

OK

Leo

Re: emacs-20111024-r106182 Windows Binaries

[Eli Zaretskii]

> From: Leo <sdl.web@gmail.com>
> Date: Tue, 25 Oct 2011 16:36:52 +0800
> Cc: help-emacs-windows@gnu.org
> 
> On 2011-10-25 16:00 +0800, Eli Zaretskii wrote:
> > Each application crash is recorded in the system log that you can view
> > using the "Event Viewer" (right-click on "My Computer", select
> > "Manage", then expand the "Event Viewer" widget and click on
> > "Application").
> 
> <Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">

The textual description is much better, because it includes
descriptions of the <Data> field, instead of relying on the schema.
There's a "copy to clipboard" button on the "event properties" dialog,
which you can use to paste the textual description into the mail.

>   <EventData>
>     <Data>emacs.exe</Data>
>     <Data>24.0.90.0</Data>
>     <Data>4e9e1382</Data>
>     <Data>ntdll.dll</Data>
>     <Data>6.1.7601.17514</Data>
>     <Data>4ce7ba58</Data>
>     <Data>c0000029</Data>
>     <Data>000907b6</Data>
>     <Data>760</Data>
>     <Data>01cc91fd0016c7e4</Data>
>     <Data>D:\emacs-24.0.90\bin\emacs.exe</Data>
>     <Data>C:\windows\SysWOW64\ntdll.dll</Data>

Yes, that's the same crash that I cited.

Re: emacs-20111024-r106182 Windows Binaries

[Leo]

On 2011-10-25 16:56 +0800, Eli Zaretskii wrote:
> The textual description is much better, because it includes
> descriptions of the <Data> field, instead of relying on the schema.
> There's a "copy to clipboard" button on the "event properties" dialog,
> which you can use to paste the textual description into the mail.

Yes, I used the copy button but then deleted the parts containing
Chinese chars. A full version looks like this:
http://paste.pound-python.org/show/14200/

Leo

Re: emacs-20111024-r106182 Windows Binaries

[Christoph Scholtes]

Eli Zaretskii <eliz@gnu.org> writes:

>>   <EventData>
>>     <Data>emacs.exe</Data>
>>     <Data>24.0.90.0</Data>
>>     <Data>4e9e1382</Data>
>>     <Data>ntdll.dll</Data>
>>     <Data>6.1.7601.17514</Data>
>>     <Data>4ce7ba58</Data>
>>     <Data>c0000029</Data>
>>     <Data>000907b6</Data>
>>     <Data>760</Data>
>>     <Data>01cc91fd0016c7e4</Data>
>>     <Data>D:\emacs-24.0.90\bin\emacs.exe</Data>
>>     <Data>C:\windows\SysWOW64\ntdll.dll</Data>
>
> Yes, that's the same crash that I cited.

Leo said that this happens with the last weekly build I provided. It was
compiled _without_ optimizations. The original report claimed to compile
with `-O2'. 

Just to make sure, is specifying `--no-opt' enough to disable
optimizations when configuring or do I have to specify `-O0'
explicitely?

If this turns out to be a problem with 4.6.1 in general, I will switch
back to an older version. Or should I use `-fno-omit-frame-pointer'?

Christoph

Re: emacs-20111024-r106182 Windows Binaries

[Eli Zaretskii]

> From: Christoph Scholtes <cschol2112@googlemail.com>
> Date: Tue, 25 Oct 2011 06:32:41 -0600
> Cc: help-emacs-windows@gnu.org, Leo <sdl.web@gmail.com>, emacs-devel@gnu.org
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> >>   <EventData>
> >>     <Data>emacs.exe</Data>
> >>     <Data>24.0.90.0</Data>
> >>     <Data>4e9e1382</Data>
> >>     <Data>ntdll.dll</Data>
> >>     <Data>6.1.7601.17514</Data>
> >>     <Data>4ce7ba58</Data>
> >>     <Data>c0000029</Data>
> >>     <Data>000907b6</Data>
> >>     <Data>760</Data>
> >>     <Data>01cc91fd0016c7e4</Data>
> >>     <Data>D:\emacs-24.0.90\bin\emacs.exe</Data>
> >>     <Data>C:\windows\SysWOW64\ntdll.dll</Data>
> >
> > Yes, that's the same crash that I cited.
> 
> Leo said that this happens with the last weekly build I provided. It was
> compiled _without_ optimizations. The original report claimed to compile
> with `-O2'. 

I didn't realize the above crash was with your binaries.  If it was,
it's probably a different problem.  Leo, if you have a way of
reproducing this, please run under GDB and produce a backtrace, or
better yet, show the recipe (and file a bug report).

> Just to make sure, is specifying `--no-opt' enough to disable
> optimizations when configuring or do I have to specify `-O0'
> explicitely?

Specifying --no-opt is enough.

> If this turns out to be a problem with 4.6.1 in general, I will switch
> back to an older version. Or should I use `-fno-omit-frame-pointer'?

With --no-opt, -fno-omit-frame-pointer should have no effect, because
this optimization is turned on by -O1.

Re: emacs-20111024-r106182 Windows Binaries

[Christoph Scholtes]

Eli Zaretskii <eliz@gnu.org> writes:

> I didn't realize the above crash was with your binaries.  If it was,
> it's probably a different problem.  Leo, if you have a way of
> reproducing this, please run under GDB and produce a backtrace, or
> better yet, show the recipe (and file a bug report).

I looked around in the tdm bug tracker but I couldn't find any issues
related to optimization or `-fno-omit-frame-pointer' reported for 4.6.1
so far.

>> Just to make sure, is specifying `--no-opt' enough to disable
>> optimizations when configuring or do I have to specify `-O0'
>> explicitely?
>
> Specifying --no-opt is enough.
>
> With --no-opt, -fno-omit-frame-pointer should have no effect, because
> this optimization is turned on by -O1.

Thanks.

Re: emacs-20111024-r106182 Windows Binaries

[Eli Zaretskii]

> From: Christoph Scholtes <cschol2112@googlemail.com>
> Cc: sdl.web@gmail.com,  emacs-devel@gnu.org
> Date: Tue, 25 Oct 2011 19:06:30 -0600
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > I didn't realize the above crash was with your binaries.  If it was,
> > it's probably a different problem.  Leo, if you have a way of
> > reproducing this, please run under GDB and produce a backtrace, or
> > better yet, show the recipe (and file a bug report).
> 
> I looked around in the tdm bug tracker but I couldn't find any issues
> related to optimization or `-fno-omit-frame-pointer' reported for 4.6.1
> so far.

The problem is not specific to tdm, it's a general problem of any GCC
port to Windows that uses the Microsoft runtime.  I posted here the
URL that describes the bug in more detail.  It only happens with
`-fomit-frame-pointer', i.e. with any optimization at or above -O1 (if
I understand correctly the GCC manual and if the manual is accurate).

Re: emacs-20111024-r106182 Windows Binaries

[Leo]

On 2011-10-25 21:12 +0800, Eli Zaretskii wrote:
> I didn't realize the above crash was with your binaries.  If it was,
> it's probably a different problem.  Leo, if you have a way of
> reproducing this, please run under GDB and produce a backtrace, or
> better yet, show the recipe (and file a bug report).

This may not be related to gcc version since I have also seen it in an
Emacs built with gcc 3.4.

All crashes happened during completion (such as in C-h f or C-h v).

It seems to have something to do this setting in my .emacs:
(setq completion-styles '(basic initials substring partial-completion))

Leo

Re: [h-e-w] emacs-20111024-r106182 Windows Binaries

[Eli Zaretskii]

> From: Leo <sdl.web@gmail.com>
> Date: Wed, 26 Oct 2011 11:25:24 +0800
> Cc: emacs-devel@gnu.org
> 
> This may not be related to gcc version since I have also seen it in an
> Emacs built with gcc 3.4.
> 
> All crashes happened during completion (such as in C-h f or C-h v).
> 
> It seems to have something to do this setting in my .emacs:
> (setq completion-styles '(basic initials substring partial-completion))

If you cannot easily reproduce this, just run Emacs under GDB at all
times.  The debugger stays out of your way as long as Emacs doesn't
crash.