Re: buffer.c/buffer.h: How to add new buffer-local variables?

[Keith David Bershatsky <esq@lawlist.com>]

[-- Attachment #1: Type: text/plain, Size: 1143 bytes --]

Increasing the eassert value in field_relpos, and increasing the value of MAX_PER_BUFFER_VARS, are sufficient to build and use Emacs for the NS and X11 ports.

The w32 port of Emacs, however, crashes when adding one too many local-variables in buffer.c/h that breaks the camel's back:

Program received signal SIGSEGV, Segmentation fault.
0x01189584 in vector_marked_p (v=0xbaadf008) at alloc.c:3980
3980      return XVECTOR_MARKED_P (v);
(gdb) bt
#0  0x01189584 in vector_marked_p (v=0xbaadf008) at alloc.c:3980
#1  0x0118dc21 in mark_object (arg=...) at alloc.c:6684
#2  0x0118d2d4 in mark_vectorlike (header=0x5ab4b98) at alloc.c:6361
#3  0x0118d510 in mark_buffer (buffer=0x5ab4b98) at alloc.c:6428
#4  0x0118dd1f in mark_object (arg=...) at alloc.c:6717
#5  0x0118d69d in mark_localized_symbol (ptr=0x4a86c40) at alloc.c:6481

Attached is a gdb backtrace and a minimal diff working example that causes the w32 port to crash during garbage collection.  To reproduce the crash on the w32 port of Emacs, launch the GUI Emacs and at the *GNU Emacs* welcome screen, and hold down the right arrow key until a garbage collection occurs.

Keith


[-- Attachment #2: 001.diff --]
[-- Type: application/diff, Size: 12095 bytes --]

[-- Attachment #3: gdb_001.txt --]
[-- Type: application/txt, Size: 66888 bytes --]