ruby-changes:37603
From: ko1 <ko1@a...>
Date: Sun, 22 Feb 2015 11:00:49 +0900 (JST)
Subject: [ruby-changes:37603] ko1:r49684 (trunk): * gc.c (rb_objspace_call_finalizer): control GC execution during
ko1 2015-02-22 11:00:40 +0900 (Sun, 22 Feb 2015) New Revision: 49684 http://svn.ruby-lang.org/cgi-bin/viewvc.cgi?view=revision&revision=49684 Log: * gc.c (rb_objspace_call_finalizer): control GC execution during force firnalizations at the end of interpreter process. [Bug #10768] 1) Prohibit incremental GC while running Ruby-level finalizers to avoid any danger. 2) Prohibit GC while invoking T_DATA/T_FILE data structure because these operations break object relations consistency. This patch can introduce another memory consuming issue because Ruby-level finalizers can run after (2), GC is disabled. However, basically object consistency was broken at (2) as I described above. So that running Ruby-level finalizers contains danger originally. Because of this point, I need to suggest to remove these 3 lines (invoking remaining finalizers). And add a rule to add that finalizers should not add new finalizers, or say there is no guarantee to invoke finalizers that added by another finalizer. Modified files: trunk/ChangeLog trunk/gc.c Index: ChangeLog =================================================================== --- ChangeLog (revision 49683) +++ ChangeLog (revision 49684) @@ -1,3 +1,24 @@ https://github.com/ruby/ruby/blob/trunk/ChangeLog#L1 +Sun Feb 22 10:43:37 2015 Koichi Sasada <ko1@a...> + + * gc.c (rb_objspace_call_finalizer): control GC execution during + force firnalizations at the end of interpreter process. + [Bug #10768] + + 1) Prohibit incremental GC while running Ruby-level finalizers + to avoid any danger. + 2) Prohibit GC while invoking T_DATA/T_FILE data structure + because these operations break object relations consistency. + + This patch can introduce another memory consuming issue because + Ruby-level finalizers can run after (2), GC is disabled. + However, basically object consistency was broken at (2) as I + described above. So that running Ruby-level finalizers contains + danger originally. Because of this point, I need to suggest to + remove these 3 lines (invoking remaining finalizers). And add a + rule to add that finalizers should not add new finalizers, or + say there is no guarantee to invoke finalizers that added by + another finalizer. + Sun Feb 22 04:07:05 2015 Zachary Scott <e@z...> * ext/openssl/ossl_asn1.c: [DOC] RDoc formatting fixes for Index: gc.c =================================================================== --- gc.c (revision 49683) +++ gc.c (revision 49684) @@ -1209,10 +1209,8 @@ static void heap_page_free(rb_objspace_t https://github.com/ruby/ruby/blob/trunk/gc.c#L1209 void rb_objspace_free(rb_objspace_t *objspace) { -#if 0 if (is_lazy_sweeping(heap_eden)) rb_bug("lazy sweeping underway when freeing object space"); -#endif if (objspace->profile.records) { free(objspace->profile.records); @@ -2568,6 +2566,10 @@ rb_objspace_call_finalizer(rb_objspace_t https://github.com/ruby/ruby/blob/trunk/gc.c#L2566 finalize_deferred(objspace); assert(heap_pages_deferred_final == 0); + gc_rest(objspace); + /* prohibit incremental GC */ + objspace->flags.dont_incremental = 1; + /* force to run finalizer */ while (finalizer_table->num_entries) { struct force_finalize_list *list = 0; @@ -2582,10 +2584,13 @@ rb_objspace_call_finalizer(rb_objspace_t https://github.com/ruby/ruby/blob/trunk/gc.c#L2584 } } - /* finalizers are part of garbage collection */ + /* prohibit GC because force T_DATA finalizers can break an object graph consistency */ + dont_gc = 1; + + /* running data/file finalizers are part of garbage collection */ gc_enter(objspace, "rb_objspace_call_finalizer"); - /* run data object's finalizers */ + /* run data/file object's finalizers */ for (i = 0; i < heap_allocated_pages; i++) { p = heap_pages_sorted[i]->start; pend = p + heap_pages_sorted[i]->total_slots; while (p < pend) { @@ -2625,14 +2630,6 @@ rb_objspace_call_finalizer(rb_objspace_t https://github.com/ruby/ruby/blob/trunk/gc.c#L2630 st_free_table(finalizer_table); finalizer_table = 0; ATOMIC_SET(finalizing, 0); - -#if 0 - /* - * finish any lazy sweeps that may have been started - * when finalizing the objects in the heap - */ - gc_rest(objspace); -#endif } static inline int -- ML: ruby-changes@q... Info: http://www.atdot.net/~ko1/quickml/