bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

1. emacs -Q

2. M-: (+ 1997 120 400)

Emacs crashes systematically.  I was trying to repro the behavior
reported for bug #21131, and fell upon this immediately.


In GNU Emacs 25.1.50.1 (i686-pc-mingw32)
 of 2015-12-10
Repository revision: 6148555ee5a3d0139ae517803718b3e0357933c7
Windowing system distributor `Microsoft Corp.', version 6.1.7601
Configured using:
 `configure --prefix=/c/Devel/emacs/snapshot/trunk --enable-checking=yes
 --enable-check-lisp-object-type --without-compress-install 'CFLAGS=-Og
 -ggdb3' LDFLAGS=-Lc:/Devel/emacs/lib 'CPPFLAGS=-DGC_MCHECK=1
 -Ic:/Devel/emacs/include''

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Sun, 10 Jul 2016 09:17:30 -0700 (PDT)
> From: Drew Adams <drew.adams@oracle.com>
> 
> 1. emacs -Q
> 
> 2. M-: (+ 1997 120 400)
> 
> Emacs crashes systematically.

It doesn't crash here.  Maybe you have some font installed that
triggers this crash.

Can someone else reproduce this?

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Clément Pit--Claudel]

[-- Attachment #1.1: Type: text/plain, Size: 478 bytes --]

On 2016-07-10 18:41, Eli Zaretskii wrote:
> Can someone else reproduce this?

I can't. The font thing sounds like a smart guess.
Drew, does M-: (insert 2517) cause a crash too?

In GNU Emacs 25.1.50.1 (x86_64-unknown-linux-gnu, GTK+ Version 3.10.8)
 of 2016-06-26 built on clem-w50-mint
Repository revision: 431437b6593320dc5a7a8aac9c911c778a656117
Windowing system distributor 'The X.Org Foundation', version 11.0.11501000
System Description:	Linux Mint 17.3 Rosa


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> > Can someone else reproduce this?
> 
> I can't. The font thing sounds like a smart guess.
> Drew, does M-: (insert 2517) cause a crash too?

Yes!

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> > 1. emacs -Q
> > 2. M-: (+ 1997 120 400)
> > Emacs crashes systematically.
> 
> It doesn't crash here.  Maybe you have some font installed that
> triggers this crash.  Can someone else reproduce this?

It's also possible that this was fixed.  The build is a bit old
(it's all I have).

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Sun, 10 Jul 2016 09:45:19 -0700 (PDT)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
> > > 1. emacs -Q
> > > 2. M-: (+ 1997 120 400)
> > > Emacs crashes systematically.
> > 
> > It doesn't crash here.  Maybe you have some font installed that
> > triggers this crash.  Can someone else reproduce this?
> 
> It's also possible that this was fixed.  The build is a bit old
> (it's all I have).

I have now tried binaries from around the date your build was made,
and they don't crash here.  The system where I tried doesn't have a
font that supports the characters whose codepoint is the result of the
evaluation, u+09D5.

That character belongs to the Bengali block.  Can you see which
font(s) you have installed that support that block?  Also, does Emacs
display the Bengali greeting in etc/HELLO?  If it does, you can use
"C-u C-x =" on that part of HELLO to show the font Emacs uses.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> > > > 1. emacs -Q
> > > > 2. M-: (+ 1997 120 400)
> > > > Emacs crashes systematically.
> > >
> > > It doesn't crash here.  Maybe you have some font installed that
> > > triggers this crash.  Can someone else reproduce this?
> >
> > It's also possible that this was fixed.  The build is a bit old
> > (it's all I have).
> 
> I have now tried binaries from around the date your build was made,
> and they don't crash here.

OK, great.  Thanks for doing that.

> The system where I tried doesn't have a font that supports the
> characters whose codepoint is the result of the evaluation, u+09D5.
> That character belongs to the Bengali block.
>
> Can you see which font(s) you have installed that support that block?

I have no idea how to do that.  If it is easy and quick to do, I can
try.  Please let me know what to do.

> Also, does Emacs display the Bengali greeting in etc/HELLO?  If it
> does, you can use "C-u C-x =" on that part of HELLO to show the font
> Emacs uses.

First, `C-h h' also crashes immediately.

Second, trying to visit file etc/HELLO also crashes immediately.

HTH.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Sun, 10 Jul 2016 10:28:17 -0700 (PDT)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
> > The system where I tried doesn't have a font that supports the
> > characters whose codepoint is the result of the evaluation, u+09D5.
> > That character belongs to the Bengali block.
> >
> > Can you see which font(s) you have installed that support that block?
> 
> I have no idea how to do that.  If it is easy and quick to do, I can
> try.  Please let me know what to do.

This page says how to display all the install fonts:

  http://superuser.com/questions/866189/deterministic-way-to-find-which-installed-fonts-support-a-given-unicode-characte

This page lists all the fonts that come with Windows, including which
scripts each font supports:

  https://en.wikipedia.org/wiki/List_of_typefaces_included_with_Microsoft_Windows

By comparing the list of installed fonts with the list of fonts that
comes with your version of Windows, you can identify the fonts that
didn't come out of the box -- I'm guessing that one of them is the
culprit.

This page lists fonts that support Bengali:

  http://salrc.uchicago.edu/resources/fonts/available/bengali/

Can you tell which ones are installed on your system?

On my system, Emacs uses Vrinda to display Bengali, and it doesn't
crash for HELLO.

> > Also, does Emacs display the Bengali greeting in etc/HELLO?  If it
> > does, you can use "C-u C-x =" on that part of HELLO to show the font
> > Emacs uses.
> 
> First, `C-h h' also crashes immediately.
> 
> Second, trying to visit file etc/HELLO also crashes immediately.

Can you remove the Bengali parts from HELLO and try again?

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> > First, `C-h h' also crashes immediately.
> > Second, trying to visit file etc/HELLO also crashes immediately.
> 
> Can you remove the Bengali parts from HELLO and try again?

I removed the Bengali line using another editor, and tried again.
Emacs still crashed.

I will now go looking for the Bengali fonts I have installed...

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> This page says how to display all the install fonts:
>   http://superuser.com/questions/866189/deterministic-way-to-find-which-
> installed-fonts-support-a-given-unicode-characte

I used this command from that page:
reg query "HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Fonts"

> This page lists fonts that support Bengali:
>   http://salrc.uchicago.edu/resources/fonts/available/bengali/
> 
> Can you tell which ones are installed on your system?

    Shonar Bangla (TrueType)    REG_SZ    Shonar.ttf
    Shonar Bangla Bold (TrueType)    REG_SZ    Shonarb.ttf

I see no other matches between those two lists.

HTH.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Sun, 10 Jul 2016 11:19:01 -0700 (PDT)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
>     Shonar Bangla (TrueType)    REG_SZ    Shonar.ttf
>     Shonar Bangla Bold (TrueType)    REG_SZ    Shonarb.ttf
> 
> I see no other matches between those two lists.

Hmm... these are stock Windows fonts, so unlikely to cause the crash.

I guess we will have to detect the problematic line(s) in HELLO one by
one.  Please start "emacs -Q", then type "C-x 2", and shrink the upper
window to just 5 lines.  Then type "C-h H" in that window -- this
shouldn't crash, as the first 5 lines of HELLO are pure ASCII.

Now click on the bottom arrow of the scroll bar of that window one
click at a time.  Each click should show one more line.  (Some clicks
might take a couple of seconds to react, as Emacs searches for the
fonts it needs.)  When the problematic line is about to enter the
view, the click to scroll that line into the window will crash Emacs.
Please tell what was the last line you were able to display before the
crash.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Sun, 10 Jul 2016 11:19:01 -0700 (PDT)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
>     Shonar Bangla (TrueType)    REG_SZ    Shonar.ttf
>     Shonar Bangla Bold (TrueType)    REG_SZ    Shonarb.ttf
> 
> I see no other matches between those two lists.

Strange, don't you see Virana?

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> Hmm... these are stock Windows fonts, so unlikely to cause the crash.
> 
> I guess we will have to detect the problematic line(s) in HELLO one by
> one.  Please start "emacs -Q", then type "C-x 2", and shrink the upper
> window to just 5 lines.  Then type "C-h H" in that window -- this
> shouldn't crash, as the first 5 lines of HELLO are pure ASCII.
> 
> Now click on the bottom arrow of the scroll bar of that window one
> click at a time.  Each click should show one more line.  (Some clicks
> might take a couple of seconds to react, as Emacs searches for the
> fonts it needs.)  When the problematic line is about to enter the
> view, the click to scroll that line into the window will crash Emacs.
> Please tell what was the last line you were able to display before the
> crash.

Middle/Near East was successful.  Clicking once more crashed Emacs.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Sun, 10 Jul 2016 12:46:07 -0700 (PDT)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
> > Now click on the bottom arrow of the scroll bar of that window one
> > click at a time.  Each click should show one more line.  (Some clicks
> > might take a couple of seconds to react, as Emacs searches for the
> > fonts it needs.)  When the problematic line is about to enter the
> > view, the click to scroll that line into the window will crash Emacs.
> > Please tell what was the last line you were able to display before the
> > crash.
> 
> Middle/Near East was successful.  Clicking once more crashed Emacs.

Looks like your problem is more basic than just one font.  Do other
(older) Emacs versions also crash trying to display HELLO?

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Mon, 11 Jul 2016 17:26:03 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: 23935@debbugs.gnu.org
> 
> > Date: Sun, 10 Jul 2016 12:46:07 -0700 (PDT)
> > From: Drew Adams <drew.adams@oracle.com>
> > Cc: 23935@debbugs.gnu.org
> > 
> > Middle/Near East was successful.  Clicking once more crashed Emacs.
> 
> Looks like your problem is more basic than just one font.  Do other
> (older) Emacs versions also crash trying to display HELLO?

One more thing to try is start Emacs like this:

  emacs -Q -xrm Emacs.fontBackend:gdi

and then see if "C-h H" crashes Emacs.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> Any news on this one?

No, and I doubt I will ever get around to doing all that is (apparently) required for this.  Please close it, if that helps you, or keep it open, in case someone else reproduces it easily.  Thanks for spending some time on this.  I don't have more time for it; sorry.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> >     Shonar Bangla (TrueType)    REG_SZ    Shonar.ttf
> >     Shonar Bangla Bold (TrueType)    REG_SZ    Shonarb.ttf
> >
> > I see no other matches between those two lists.
> 
> Strange, don't you see Virana?

Yes, sorry. I see (only) this one:

    Vrinda Bold (TrueType)    REG_SZ    vrindab.ttf


Thx.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> > Middle/Near East was successful.  Clicking once more crashed Emacs.
> 
> Looks like your problem is more basic than just one font.  Do other
> (older) Emacs versions also crash trying to display HELLO?

No. Emacs 24.5 works fine (C-h h).  So do Emacs 23.4.1 and 22.3.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> > > Middle/Near East was successful.  Clicking once more crashed Emacs.
> >
> > Looks like your problem is more basic than just one font.  Do other
> > (older) Emacs versions also crash trying to display HELLO?
> 
> One more thing to try is start Emacs like this:
> 
>   emacs -Q -xrm Emacs.fontBackend:gdi
> 
> and then see if "C-h H" crashes Emacs.

No, it does not crash (I used runemacs instead of emacs - let me know
if that invalidates the test).

This, BTW, is what it shows for Bengali:

Bengali (বাংলা)	নমস্কার

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Mon, 11 Jul 2016 17:52:07 -0700 (PDT)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
> > > > Middle/Near East was successful.  Clicking once more crashed Emacs.
> > >
> > > Looks like your problem is more basic than just one font.  Do other
> > > (older) Emacs versions also crash trying to display HELLO?
> > 
> > One more thing to try is start Emacs like this:
> > 
> >   emacs -Q -xrm Emacs.fontBackend:gdi
> > 
> > and then see if "C-h H" crashes Emacs.
> 
> No, it does not crash (I used runemacs instead of emacs - let me know
> if that invalidates the test).

I don't think this invalidates the test, but please also try with
emacs.

It sounds like the problem is with the Uniscribe DLL, usp10.dll.  How
many of those do you have on your system, and in what directories?
Also, do you see the crashes in the Windows Event Viewer (under
Application)?  If so, please post the details of one such log entry.

Finally, if you can get hold of a later binary of Emacs 25, please try
that as well.  I'd like to be sure we are not chasing a wild goose
here.

Thanks.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> > > > > Middle/Near East was successful.  Clicking once more crashed Emacs.
> > > >
> > > > Looks like your problem is more basic than just one font.  Do other
> > > > (older) Emacs versions also crash trying to display HELLO?
> > >
> > > One more thing to try is start Emacs like this:
> > >   emacs -Q -xrm Emacs.fontBackend:gdi
> > > and then see if "C-h H" crashes Emacs.
> >
> > No, it does not crash (I used runemacs instead of emacs - let me know
> > if that invalidates the test).
> 
> I don't think this invalidates the test, but please also try with
> emacs.

Same thing - no crash.

> It sounds like the problem is with the Uniscribe DLL, usp10.dll.  How
> many of those do you have on your system, and in what directories?

I guess I don't have any of them.
Searching C: and D: drives for usp10.dll found nothing.

> Also, do you see the crashes in the Windows Event Viewer (under
> Application)?  If so, please post the details of one such log entry.

Dunno what that means.  You will need to give specific instructions.

> Finally, if you can get hold of a later binary of Emacs 25, please try
> that as well.  I'd like to be sure we are not chasing a wild goose
> here.
> 
> Thanks.

The latest build I have is this pretest, which also crashes using C-h H
from runemacs -Q, but which does not crash with C-h H from emacs -Q -xrm Emacs.fontBackend:gdi.

In GNU Emacs 25.0.94.1 (x86_64-w64-mingw32)
 of 2016-05-25 built on KAEL
Windowing system distributor 'Microsoft Corp.', version 6.1.7601
Configured using:
 'configure --prefix=/tmp/emacs --without-imagemagick
 --enable-checking=yes,glyphs --enable-check-lisp-object-type
 'CFLAGS=-Og -gdwarf-4 -g3''

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Tue, 12 Jul 2016 13:51:50 +0000 (UTC)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
> > It sounds like the problem is with the Uniscribe DLL, usp10.dll.  How
> > many of those do you have on your system, and in what directories?
> 
> I guess I don't have any of them.
> Searching C: and D: drives for usp10.dll found nothing.

This cannot be true.  Don't you have one in C:\WINDOWS\SysWOW64 and
another one in C:\WINDOWS\system32?  If you didn't, I think all other
Emacs versions would have crashed.

This DLL comes with Windows out of the box, so it must be somewhere.

Do you have Process Explorer installed?  If not, can you install it
from the Microsoft Sysinternals site,
https://technet.microsoft.com/en-us/sysinternals/processexplorer?
That program can show the DLLs loaded by Emacs when it runs, so you
can see whether Emacs loads the DLL, and its full absolute file name.

> > Also, do you see the crashes in the Windows Event Viewer (under
> > Application)?  If so, please post the details of one such log entry.
> 
> Dunno what that means.  You will need to give specific instructions.

Right-click "Computer" and select "Manage".  Expand "Event Viewer",
then expand "Windows Logs", then click on Application.  After a few
seconds, you will be presented with a list of events; look for ones
labeled "Error" around the time of a crash.  Double-clicking on such a
line will show the details of the crash.

> The latest build I have is this pretest, which also crashes using C-h H
> from runemacs -Q, but which does not crash with C-h H from emacs -Q -xrm Emacs.fontBackend:gdi.

Very strange.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> > > It sounds like the problem is with the Uniscribe DLL, usp10.dll.  How
> > > many of those do you have on your system, and in what directories?
> >
> > I guess I don't have any of them.
> > Searching C: and D: drives for usp10.dll found nothing.
> 
> This cannot be true.  Don't you have one in C:\WINDOWS\SysWOW64 and
> another one in C:\WINDOWS\system32?  If you didn't, I think all other
> Emacs versions would have crashed.

Yes, I have it in both places.  But for some reason search (from
Windows Explorer) does not find it, even when searching just those
folders.

> Do you have Process Explorer installed?

Not that I know of.  How would I know?

> If not, can you install it from the Microsoft Sysinternals site,
> https://technet.microsoft.com/en-us/sysinternals/processexplorer?

It seems not.  That page says:

  Runs on:
  Client: Windows Vista and higher (Including IA64).
  Server: Windows Server 2008 and higher (Including IA64).

I have Windows 7 (64-bit).

> > > Also, do you see the crashes in the Windows Event Viewer (under
> > > Application)?  If so, please post the details of one such log entry.
> >
> > Dunno what that means.  You will need to give specific instructions.
> 
> Right-click "Computer" and select "Manage".  Expand "Event Viewer",
> then expand "Windows Logs", then click on Application.  After a few
> seconds, you will be presented with a list of events; look for ones
> labeled "Error" around the time of a crash.  Double-clicking on such a
> line will show the details of the crash.

Nope. The only errors I see listed there from July (when I did these
tests) are for Google Update and a VPN executable.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Tue, 12 Jul 2016 14:33:30 +0000 (UTC)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
> > Do you have Process Explorer installed?
> 
> Not that I know of.  How would I know?
> 
> > If not, can you install it from the Microsoft Sysinternals site,
> > https://technet.microsoft.com/en-us/sysinternals/processexplorer?
> 
> It seems not.  That page says:
> 
>   Runs on:
>   Client: Windows Vista and higher (Including IA64).
>   Server: Windows Server 2008 and higher (Including IA64).
> 
> I have Windows 7 (64-bit).

Your system is a Client, and Windows 7 is higher than Vista.  So you
are OK.  (I use that program on Windows 7 as well.)

I think it's important to know which Uniscribe DLL is loaded by Emacs
on your machine.

> > Right-click "Computer" and select "Manage".  Expand "Event Viewer",
> > then expand "Windows Logs", then click on Application.  After a few
> > seconds, you will be presented with a list of events; look for ones
> > labeled "Error" around the time of a crash.  Double-clicking on such a
> > line will show the details of the crash.
> 
> Nope. The only errors I see listed there from July (when I did these
> tests) are for Google Update and a VPN executable.

Even more strange.

When Emacs crashes, does it pop up some dialog?  If so, which dialog?

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

[-- Attachment #1: Type: text/plain, Size: 1260 bytes --]

> > > Do you have Process Explorer installed?
> 
> I think it's important to know which Uniscribe DLL is loaded by Emacs
> on your machine.

I installed it.

I started Emacs -Q. I see the emacs.exe process in Process Explorer,
but I don't see any usp10, and I don't see any Uniscribe.  The only
processes listed whose names begin with "u" are unsecapp.exe and
UdaterUI.exe (McAfee).

I tried Show Process Tree, but didn't see anything there either.

I tried Find > Process Explorer Search > Handle or DLL substring,
but that also did not find ups10 or uniscribe or Uniscribe. Perhaps
it has another name?

> > > Right-click "Computer" and select "Manage".  Expand "Event Viewer",
> > > then expand "Windows Logs", then click on Application.  After a few
> > > seconds, you will be presented with a list of events; look for ones
> > > labeled "Error" around the time of a crash.  Double-clicking on such a
> > > line will show the details of the crash.
> >
> > Nope. The only errors I see listed there from July (when I did these
> > tests) are for Google Update and a VPN executable.
> 
> Even more strange.
> When Emacs crashes, does it pop up some dialog?  If so, which dialog?

See attached. #2 is shown after I click No in #1.

[-- Attachment #2: throw-emacs-crash-dialog-2.png --]
[-- Type: image/png, Size: 19792 bytes --]

[-- Attachment #3: throw-emacs-crash-dialog-1.png --]
[-- Type: image/png, Size: 17427 bytes --]

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Tue, 12 Jul 2016 16:09:50 +0000 (UTC)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
> I started Emacs -Q. I see the emacs.exe process in Process Explorer,
> but I don't see any usp10, and I don't see any Uniscribe.  The only
> processes listed whose names begin with "u" are unsecapp.exe and
> UdaterUI.exe (McAfee).

It's not in the process list.

Click View on the menu bar, and then make sure "Show Lower Pane" is
checked; if not, click it so it is.  Then Click View->Lower Pane View
and select "DLLs".  Next, click on the line of the Emacs process, and
the lower pane will show all of the DLLs that are loaded into the
Emacs process.  Look for usp10.dll, and tell what is its full absolute
file name and its Version.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> Click View on the menu bar, and then make sure "Show Lower Pane" is
> checked; if not, click it so it is.  Then Click View->Lower Pane View
> and select "DLLs".  Next, click on the line of the Emacs process, and
> the lower pane will show all of the DLLs that are loaded into the
> Emacs process.  Look for usp10.dll, and tell what is its full absolute
> file name and its Version.

Sure enough; thanks for the lesson.

  usp10.dll
  Uniscribe Unicode script processor
  Microsoft Corporation
  C:\Windows\SysWOW64\usp10.dll

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Tue, 12 Jul 2016 19:57:02 +0000 (UTC)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
>   usp10.dll
>   Uniscribe Unicode script processor
>   Microsoft Corporation
>   C:\Windows\SysWOW64\usp10.dll

It should also show its version, can you tell what that is?

Thanks.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> >   usp10.dll
> >   Uniscribe Unicode script processor
> >   Microsoft Corporation
> >   C:\Windows\SysWOW64\usp10.dll
> 
> It should also show its version, can you tell what that is?

There are only those 4 columns: Name, Description, Company Name, and Path.

However, if I right-click on that line and choose Properties
then I see this on the Image tab:

Version: 1.626.7601.19054
Build Time: Tue Nov 03 10:54:42 2015

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Tue, 12 Jul 2016 20:23:59 -0700 (PDT)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
> Version: 1.626.7601.19054
> Build Time: Tue Nov 03 10:54:42 2015

Hmm, I have an earlier version on my Windows 7 machines.  This one
seems to come with some security update.  But it's still legit, I
think.

At this point, I think the only practical way forward is to run Emacs
under GDB and show the backtrace from the crash.  Which versions of
GDB do you have on that system?  (Run "gdb --config" to see the
answer.)

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> Hmm, I have an earlier version on my Windows 7 machines.  This one
> seems to come with some security update.  But it's still legit, I
> think.
> 
> At this point, I think the only practical way forward is to run Emacs
> under GDB and show the backtrace from the crash.  Which versions of
> GDB do you have on that system?  (Run "gdb --config" to see the
> answer.)

Gdb is not in my path anymore, so cmd.exe does not recognize that.
But I have an old version.

If you want me to use GDB I'm afraid you will need to give me
simple, clear instructions, step-by-step, including the download.

If it's simple and I don't spend hours picking it up, a dividend
might be that I could use it with other Emacs crashes - assuming
the instructions apply there also.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Wed, 13 Jul 2016 08:02:19 -0700 (PDT)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
> Gdb is not in my path anymore, so cmd.exe does not recognize that.
> But I have an old version.

What does that old version say when you invoke it "gdb --version"?  It
might be good enough.

> If you want me to use GDB I'm afraid you will need to give me
> simple, clear instructions, step-by-step, including the download.

If your existing old version fits the bill, I think it's even easier,
no?

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

> > Gdb is not in my path anymore, so cmd.exe does not recognize that.
> > But I have an old version.
> 
> What does that old version say when you invoke it "gdb --version"?  It
> might be good enough.

I have two folders, for two versions: GDB-7.2 and GDB-7.10.

If I try gdb.exe --version in the first of these folders, a cmd window
opens and immediately closes.  Apparently this version does not work
in its present form on this machine (?).

When I do the same in the second folder, a cmd window named gdb.exe
opens, and then I get an error msg window:

 gdb.exe - System Error

 The program can't start because python26.dll is missing from your
 computer. Try reinstalling the program to fix this problem.

 [OK]

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Wed, 13 Jul 2016 08:33:27 -0700 (PDT)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
> When I do the same in the second folder, a cmd window named gdb.exe
> opens, and then I get an error msg window:
> 
>  gdb.exe - System Error
> 
>  The program can't start because python26.dll is missing from your
>  computer. Try reinstalling the program to fix this problem.

Can you install Python from this site:

  https://www.python.org/download/releases/2.6.6/

?  Then GDB 7.10 you have should work.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Drew Adams]

>   https://www.python.org/download/releases/2.6.6/

That says:

"Python 2.6.6 has been replaced by a newer security-fix only source release of Python. Please download Python 2.6.8 instead."

This is my work computer, so I'm reluctant to pick up something
that needed a security fix.  How about using 2.6.8?

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

This also rang a bell, so I searched old bug reports.  I sent
you a message on 2010/Dec/09, where I said this:

----

> > > It's enough to type "gdb RET" at the cmd prompt.  But to 
> > > attach to a running Emacs process, it is more convenient
> > > to type "gdb -p PID RET",
> > > where PID is the Emacs process ID, a number.

Hi Eli,

I tried that just now, when I got a crash dialog box. I cd'd to the dir where I have the files you sent me, then did `gdb -p 5692' (the PID).

I got a Windows error box, with title `gdb.exe - Unable to Locate Component' and with text "This application has failed to start because libexpat-1.dll was not found. Re-installing the application may fix this problem. [OK]"

In the directory where I put the files you sent (which is in my PATH), I have these files (that you sent):

.gdbinit
gdb-python27.exe
gdb.exe
gdbserver.exe

----

It seems I didn't get a reply from you.  But I do find the file
gdb-python27.exe in the other GDB directory, for version 7.2.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Wed, 13 Jul 2016 09:55:49 -0700 (PDT)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 23935@debbugs.gnu.org
> 
> >   https://www.python.org/download/releases/2.6.6/
> 
> That says:
> 
> "Python 2.6.6 has been replaced by a newer security-fix only source release of Python. Please download Python 2.6.8 instead."
> 
> This is my work computer, so I'm reluctant to pick up something
> that needed a security fix.  How about using 2.6.8?

You can try, it will most probably work OK.

Please also note these installation notes from the ezwinports site:

  If you install this anywhere but D:\usr, you will need to set 2
  environment variables:

    set GUILE_LOAD_PATH=x:\foo\share\guile\2.0
    set GUILE_LOAD_COMPILED_PATH=x:\foo\lib\guile\2.0\ccache

  where "x:\foo" is the directory from which you unzipped the GDB
  binary zip file.  These variables are required for the Guile support
  in GDB to be able to initialize itself.

  If GDB complains at startup about being unable to load Python
  modules, you may need to set the PYTHONPATH environment variable to
  point to your Python 2.6.6 installation, like this:

    set PYTHONPATH=C:\Python26;C:\Python26\Lib;C:\Python26\;C:\Python26\DLLs;C:\Python26\Lib\lib-tk;C:\Python26\Lib\site-packages

  (assuming you installed Python in C:\Python26).

> This also rang a bell, so I searched old bug reports.  I sent
> you a message on 2010/Dec/09, where I said this:
> 
> ----
> 
> > > > It's enough to type "gdb RET" at the cmd prompt.  But to 
> > > > attach to a running Emacs process, it is more convenient
> > > > to type "gdb -p PID RET",
> > > > where PID is the Emacs process ID, a number.
> 
> Hi Eli,
> 
> I tried that just now, when I got a crash dialog box. I cd'd to the dir where I have the files you sent me, then did `gdb -p 5692' (the PID).
> 
> I got a Windows error box, with title `gdb.exe - Unable to Locate Component' and with text "This application has failed to start because libexpat-1.dll was not found. Re-installing the application may fix this problem. [OK]"
> 
> In the directory where I put the files you sent (which is in my PATH), I have these files (that you sent):
> 
> .gdbinit
> gdb-python27.exe
> gdb.exe
> gdbserver.exe
> 
> ----
> 
> It seems I didn't get a reply from you.  But I do find the file
> gdb-python27.exe in the other GDB directory, for version 7.2.

The binary zip distribution of GDB 7.10 that you have includes all the
dependency DLLs, with the sole exception of Python.  So once you
install Python, I don't expect any additional problems with missing
DLLs.  (You can also download the latest GDB 7.11 from the same site.)

One more caveat: this GDB 7.10/7.11 can only debug 32-bit programs, so
you can only use it with the binaries that identify themselves as
i686-pc-mingw32, not x86_64-w64-mingw64.  If the i686-pc-mingw32 one
is stripped, i.e. without debugging symbols, we will have to arrange
fore you to download a binary with symbols.  (The .exe binary with
symbols is much larger, around 70MB.)  But first, let's see if GDB
works there.

bug#23935: 25.1.50; simple reproducible crash for emacs -Q: `M-: (+ 1997 120 400)'

[Eli Zaretskii]

> Date: Wed, 13 Jul 2016 20:24:27 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: 23935@debbugs.gnu.org
> 
> > Date: Wed, 13 Jul 2016 09:55:49 -0700 (PDT)
> > From: Drew Adams <drew.adams@oracle.com>
> > Cc: 23935@debbugs.gnu.org
> > 
> > >   https://www.python.org/download/releases/2.6.6/
> > 
> > That says:
> > 
> > "Python 2.6.6 has been replaced by a newer security-fix only source release of Python. Please download Python 2.6.8 instead."
> > 
> > This is my work computer, so I'm reluctant to pick up something
> > that needed a security fix.  How about using 2.6.8?
> 
> You can try, it will most probably work OK.
> 
> Please also note these installation notes from the ezwinports site:

Any news on this one?