Mercurial > emacs
view src/s/aix4-2.h @ 96196:568f14ebb684
* Makefile.in ${SUBDIR}: Pass additional BOOTSTRAPEMACS argument.
* Makefile.in (${lisp} ${SOME_MACHINE_LISP}, ../lisp/loaddefs.el):
Use $(BOOTSTRAPEMACS) rather than witness-emacs.
(bootstrap-emacs${EXEEXT}): Merge witness-emacs into it.
(witness-emacs): Remove.
(lisp, shortlisp): Move loaddefs.el earlier.
(mostlyclean): Forget about witness-emacs.
* Makefile.in (COMPILE_FIRST): Trim down and add autoload.el.
author | Stefan Monnier <monnier@iro.umontreal.ca> |
---|---|
date | Sun, 22 Jun 2008 13:57:28 +0000 |
parents | 43a8bd744de4 |
children | f1d13e615070 ddedcecb18ef |
line wrap: on
line source
#include "aix4-1.h" #undef ALIGN_DATA_RELOC /* On AIX Emacs uses the gmalloc.c malloc implementation. But given the way this system works, libc functions that return malloced memory use the libc malloc implementation. Calling xfree or xrealloc on the results of such functions results in a crash. One solution for this could be to define SYSTEM_MALLOC here, but that does not currently work on this system. It is possible to completely override the malloc implementation on AIX, but that involves putting the malloc functions in a shared library and setting the MALLOCTYPE environment variable to point to tha shared library. Emacs currently calls xrealloc on the results of get_current_dir name, to avoid a crash just use the Emacs implementation for that function. */ #define BROKEN_GET_CURRENT_DIR_NAME 1 /* aix3-1.h defined _NO_PROTO, probably to work around an AIX compiler that did not handle prototypes. On (at least) AIX 5.2, this causes the proper prototype to be thrown away for lseek64, so compiled Lisp files do not load correctly and compilation fails. The AIX compiler should have learned about function prototypes long ago, so we can probably go ahead and undefine _NO_PROTO. However, if someone can demonstrate that this problem still exists for AIX 4, this should be moved into a new file (aix5.h). */ #undef _NO_PROTO /* arch-tag: 38fe75ea-6aef-42bd-8449-bc34d921a562 (do not change this comment) */