bug#9723: 24.0.50; Emacs Clipboard crash

[Joseph Jones]

Disconnected == Remote Desktop session. I was connected from home and then when 
I came in to the office and connected locally it was crashed.

Sorry, I don't have it in crashed state. I do have it running in the debugger 
so if it happens again I can grab more info. I am constantly connecting 
remotely from home so if that is the issue hopefully it will crash again.

As for reproducibility, this is the first time I've seen this. I have yet to 
crash it again like this.

-----Original Message-----
From: Eli Zaretskii [mailto:eliz@gnu.org] 
Sent: Monday, January 23, 2012 9:31 PM
To: Joseph Jones
Cc: 9723@debbugs.gnu.org
Subject: Re: bug#9723: 24.0.50; Emacs Clipboard crash

> From: Joseph Jones <josejones@expedia.com>
> CC: "9723@debbugs.gnu.org" <9723@debbugs.gnu.org>
> Date: Mon, 23 Jan 2012 18:23:10 -0800
> 
> OK, it crashed for me again using the latest version from the link below.

This is not the same crash as what you had when this bug was opened.
It's something entirely different.  It is clearly related to the display 
engine, see below.

> This time I wasn't doing anything other than running a compile. I was 
> disconnected from the machine and when I connected remotely this is what I 
> found.

Sorry, I don't understand: what does it mean "disconnected from the machine"?  
Are you running Emacs through a remote desktop of some kind (mstsc.exe or some 
such)?  If so, this definitely warrants a separate bug report.  In any case, 
please describe in more detail what happened around the crash.

> Program received signal SIGTRAP, Trace/breakpoint trap.
> [Switching to Thread 10208.0x2420]
> 0x7d61002e in strchr () from C:\WINDOWS\system32\ntdll.dll
> (gdb) bt
> #0  0x7d61002e in strchr () from C:\WINDOWS\system32\ntdll.dll
> #1  0x0114f10a in w32_abort () at w32fns.c:7210
> #2  0x010ea4c8 in adjust_glyph_matrix (w=0x3903a00, matrix=0x38ed500, 
> x=0, y=0, dim=...) at dispnew.c:493

This is an assertion violation here:

      left = margin_glyphs_to_reserve (w, dim.width, w->left_margin_cols);
      right = margin_glyphs_to_reserve (w, dim.width, w->right_margin_cols);
      xassert (left >= 0 && right >= 0);

I hope you still have this crash in the debugger; if so, please tell what are 
the values of `left' and `right' (in thread #1 and in frame #2).

It's likely that display dimensions cannot be figured out for a disconnected 
remote session.  This is what this crash is about.

Can you reproduce this problem at will, by deliberately disconnecting and then 
reconnecting?