diff options
| author | Alan Wu <XrXr@users.noreply.github.com> | 2020-03-11 10:37:14 -0400 |
|---|---|---|
| committer | NARUSE, Yui <naruse@airemix.jp> | 2020-03-30 19:15:11 +0900 |
| commit | b5fa156b7907b8ea1baf8b9f0cb6a66b0fecb3d4 (patch) | |
| tree | e2551a6cc79b035f1a74e82da2703822e4c08086 /ext | |
| parent | 0057fe4063b57b445d6ffc893e442f362e3a19e0 (diff) | |
| download | ruby-b5fa156b7907b8ea1baf8b9f0cb6a66b0fecb3d4.tar.gz | |
Clear all trace events during teardown
Since 0c2d81dada, not all trace events are cleared during VM teardown.
This causes a crash when there is a tracepoint for
`RUBY_INTERNAL_EVENT_GC_EXIT` active during teardown.
The commit looks like a refactoring commit so I think this change was
unintentional.
[Bug #16682]
(cherry picked from commit b385f7670ffa420790bc548747fa4b58c4c5d8f6)
Diffstat (limited to 'ext')
| -rw-r--r-- | ext/-test-/tracepoint/gc_hook.c | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/ext/-test-/tracepoint/gc_hook.c b/ext/-test-/tracepoint/gc_hook.c index 2e695a9fba..54b469dcad 100644 --- a/ext/-test-/tracepoint/gc_hook.c +++ b/ext/-test-/tracepoint/gc_hook.c @@ -73,8 +73,16 @@ set_after_gc_start(VALUE module, VALUE proc) "__set_after_gc_start_tpval__", "__set_after_gc_start_proc__"); } +static VALUE +start_after_gc_exit(VALUE module, VALUE proc) +{ + return set_gc_hook(module, proc, RUBY_INTERNAL_EVENT_GC_EXIT, + "__set_after_gc_exit_tpval__", "__set_after_gc_exit_proc__"); +} + void Init_gc_hook(VALUE module) { rb_define_module_function(module, "after_gc_start_hook=", set_after_gc_start, 1); + rb_define_module_function(module, "after_gc_exit_hook=", start_after_gc_exit, 1); } |