I've seen these crashes for a long time in XEmacs 21.5, which kept me from using C-c p (toggle-profiling) This trivial patch cures the symptom for me, but I wonder: Are others seeing this as well? Is this the right fix? Best regards! Adrian xemacs-21.5-clean ChangeLog patch: Diff command: cvs -q diff -U 0 Files affected: src/ChangeLog Index: src/ChangeLog =================================================================== RCS file: /pack/xemacscvs/XEmacs/xemacs/src/ChangeLog,v retrieving revision 1.997 diff -u -U0 -r1.997 ChangeLog --- src/ChangeLog 8 Aug 2006 15:26:05 -0000 1.997 +++ src/ChangeLog 15 Aug 2006 17:17:58 -0000 ＠＠ -0,0 +1,9 ＠＠ +2006-08-15 Adrian Aichner <adrian(a)xemacs.org&gt; + + * profile.c (profile_reap_backtrace): Only dereference function + pointer, if not NULL. + xemacs-21.5-clean source patch: Diff command: cvs -f -z3 -q diff -u -w -N Files affected: src/profile.c =================================================================== RCS Index: src/profile.c =================================================================== RCS file: /pack/xemacscvs/XEmacs/xemacs/src/profile.c,v retrieving revision 1.25 diff -u -w -r1.25 profile.c --- src/profile.c 10 Mar 2005 09:12:38 -0000 1.25 +++ src/profile.c 15 Aug 2006 17:16:27 -0000 ＠＠ -253,6 +253,7 ＠＠ handler will record more ticks for the function while the post-amble is executing, and its value will be > our total value. */ bt->function_being_called = 0; + if (bt->function != NULL) { Fputhash (*bt->function, /* This works even when the total_ticks value has overwrapped. Same for total_consing below. */ ＠＠ -265,6 +266,7 ＠＠ + bt->current_total_gc_usage_val), Vtotal_gc_usage_profile_table); } +} void profile_record_just_called (struct backtrace *bt) -- Adrian Aichner mailto:adrian＠xemacs.org http://www.xemacs.org/