File 2891-erts-Assert-that-GC-should-not-be-disabled... of Package erlang

Overview Repositories Revisions Requests Users Attributes Meta

File 2891-erts-Assert-that-GC-should-not-be-disabled-twice.patch of Package erlang (Revision b1dff5cbf87755f120bfe931af793e05)

Currently displaying revision b1dff5cbf87755f120bfe931af793e05 , Show latest

From 2571cbf025e164989e15502da57de4d5e339f07d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?John=20H=C3=B6gberg?= <john@erlang.org>
Date: Fri, 3 May 2019 09:54:15 +0200
Subject: [PATCH 1/3] erts: Assert that GC should not be disabled twice

---
 erts/emulator/beam/erl_process.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/erts/emulator/beam/erl_process.c b/erts/emulator/beam/erl_process.c
index 1f6adb98ef..e69a9b8813 100644
--- a/erts/emulator/beam/erl_process.c
+++ b/erts/emulator/beam/erl_process.c
@@ -12035,8 +12035,13 @@ erts_set_gc_state(Process *c_p, int enable)
     ERTS_SMP_LC_ASSERT(ERTS_PROC_LOCK_MAIN == erts_proc_lc_my_proc_locks(c_p));
 
     if (!enable) {
-	c_p->flags |= F_DISABLE_GC;
-	return 0;
+        /* Strictly speaking it's not illegal to disable the GC when it's
+         * already disabled, but we risk enabling the GC prematurely if (for
+         * example) a BIF were to blindly disable it when trapping and then
+         * re-enable it before returning its result. */
+        ASSERT(!(c_p->flags & F_DISABLE_GC));
+        c_p->flags |= F_DISABLE_GC;
+        return 0;
     }
 
     c_p->flags &= ~F_DISABLE_GC;
-- 
2.16.4

Places

File 2891-erts-Assert-that-GC-should-not-be-disabled-twice.patch of Package erlang (Revision b1dff5cbf87755f120bfe931af793e05)

Places