bug#10035: Crash in check_x_frame in w32fns.c

[Eli Zaretskii]

> From: Juanma Barranquero <lekktu@gmail.com>
> Date: Tue, 15 Nov 2011 18:12:47 +0100
> Cc: Christoph Scholtes <cschol2112@googlemail.com>, 10035@debbugs.gnu.org
> 
> I just had another assertion failure in row_equal_p, which seems a
> variant of the bug you fixed. In this case, "pgrowx a" and "pgrowx b"
> do not show the same content, but a->hash == b->hash.

Two different rows having the same hash is not a problem -- the hash
function isn't guaranteed to be perfect.  The problem is that at least
one of the hash values does not match the contents of its glyph row.

> I was just editing elisp code, and the assertion failure happened
> during an isearch for "lets".

The offending row is `a'.  Is the text it shows ("Slot is the name") a
comment, displayed in font-lock-comment-face?  Can you show a
screenshot?

> #2  0x0107bb11 in row_equal_p (b=0x373e968, a=0x5d4f968,
> mouse_face_p=<optimized out>) at dispnew.c:1294
               ^^^^^^^^^^^^^^^
??? Does GCC now optimizes even under -O0?