Mercurial > hg > xemacs-beta

diff src/redisplay.c @ 1123:37bdd24225ef

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
[xemacs-hg @ 2002-11-27 07:15:02 by ben] bug fixes, profiling debugging improvements configure.in: Check for GCC version and only use -Wpacked in v3. .cvsignore: Add .idb, .ilk for MS Windows VC++. cl-macs.el: Document better. cmdloop.el: Removed. Remove nonworking breakpoint-on-error now that debug-on-error works as documented. help.el: Extract out with-displaying-help-buffer into a more general mechanism. lib-complete.el: Support thunks in find-library-source-path. startup.el: Don't catch errors when noninteractive, because that makes stack traces from stack-trace-on-error useless. .cvsignore: Windows shit. alloc.c: Better redisplay-related assert. elhash.c: Comment change. eval.c: Don't generate large warning strings (e.g. backtraces) when they will be discarded. Implement debug-on-error as documented -- it will enter the debugger and crash when an uncaught signal happens noninteractively and we are --debug. Better redisplay-related asserts. frame-msw.c, frame.c, lisp.h, redisplay.c, scrollbar-gtk.c, scrollbar-x.c, signal.c, sysdep.c: Fix up documentation related to QUIT (which CANNOT garbage-collect under any circumstances), and to redisplay critical sections. lread.c: Add load-ignore-out-of-date-elc-files, load-always-display-messages, load-show-full-path-in-messages for more robust package compilation and debugging. profile.c: Overhaul profile code. Change format to include call count and be extensible for further info. Remove call-count-profile-table. Add set-profiling-info. See related profile.el changes (which SHOULD ABSOLUTELY be in the core! Get rid of xemacs-devel and xemacs-base packages *yesterday*!).
author ben
date Wed, 27 Nov 2002 07:15:36 +0000
parents c15f25529e61
children e22b0213b713
line wrap: on
line diff
--- a/src/redisplay.c	Tue Nov 26 22:52:59 2002 +0000
+++ b/src/redisplay.c	Wed Nov 27 07:15:36 2002 +0000
@@ -6564,8 +6564,9 @@
 
   /* The menubar, toolbar, and icon updates must be done before
      enter_redisplay_critical_section is called and we are officially
-     'in_display'.  They may eval lisp code which may call Fsignal.
-     If in_display is set Fsignal will abort. */
+     'in_display'.  They may eval lisp code which may call QUIT.
+     If in_display is set, QUIT will abort (unless the code is wrapped
+     to protect against errors). */
 
 #ifdef HAVE_MENUBARS
   /* Update the menubar.  It is done first since it could change
@@ -6614,21 +6615,25 @@
      following cannot happen:
 
      1) garbage collection
-     2) Lisp code evaluation
-     3) frame size changes
-
-     We ensure (3) by calling enter_redisplay_critical_section(), which
+     2) QUIT
+     3) Any non-local exits
+     4) frame size changes
+
+     We ensure (4) by calling enter_redisplay_critical_section(), which
      will cause any pending frame size changes to get put on hold
-     till after the end of the critical section.  (1) follows
-     automatically if (2) is met.  #### Unfortunately, there are
-     some places where Lisp code can be called within this section.
-     We need to remove them.
-
-     If Fsignal() is called during this critical section, we
-     will abort().
-
-     If garbage collection is called during this critical section,
-     we simply return. #### We should abort instead.
+     till after the end of the critical section.  (2) is required because
+     of the possibility of (3).  (2) and (3) mean that any place that
+     can execute QUIT (e.g. internal_equal()), and especially any place
+     that executes Lisp code, need to be properly wrapped to protect
+     against these possibilities.  This wrapping happens using 
+     call_trapping_problems(..., INHIBIT_GC), or related functions.
+
+     To help debug potential problems, we arrange (when
+     ERROR_CHECK_TRAPPING_PROBLEMS is set) to crash automatically every
+     time we execute QUIT or check to see whether garbage collection is
+     necessary, inside of an unprotected critical section, as well as
+     further checks when we actually Fsignal(), Fthrow(),
+     garbage_collect_1().
 
      #### If a frame-size change does occur we should probably
      actually be preempting redisplay. */