From: Ondřej Surý <ondrej@isc.org>
Date: Tue, 27 Jun 2023 06:26:12 +0000 (+0200)
Subject: Call rcu_barrier() five times in the isc__mem_destroy()
X-Git-Tag: v9.19.16~14^2~1
X-Git-Url: http://git.ipfire.org/gitweb.cgi?a=commitdiff_plain;h=c1821ccf920a44d67284be462bb0ca6729235ef1;p=thirdparty%2Fbind9.git

Call rcu_barrier() five times in the isc__mem_destroy()

Because rcu_barrier() needs to be called as many times as the number of
nested call_rcu() calls (call_rcu() calls made from call_rcu thread),
and currently there's no mechanism to detect whether there are more
call_rcu callbacks scheduled, we simply call the rcu_barrier() multiple
times.  The overhead is negligible and it prevents rare assertion
failures caused by the check for memory leaks in isc__mem_destroy().
---

diff --git a/lib/isc/mem.c b/lib/isc/mem.c
index 593b6108904..b03e1122d0b 100644
--- a/lib/isc/mem.c
+++ b/lib/isc/mem.c
@@ -582,9 +582,25 @@ isc__mem_destroy(isc_mem_t **ctxp FLARG) {
 
 	/*
 	 * wait for asynchronous memory reclamation to complete
-	 * before checking for memory leaks
+	 * before checking for memory leaks.
+	 *
+	 * Because rcu_barrier() needs to be called as many times
+	 * as the number of nested call_rcu() calls (call_rcu()
+	 * calls made from call_rcu thread), and currently there's
+	 * no mechanism to detect whether there are more call_rcu
+	 * callbacks scheduled, we simply call the rcu_barrier()
+	 * multiple times.  The overhead is negligible and it
+	 * prevents rare assertion failures caused by the check
+	 * for memory leaks below.
+	 *
+	 * If there's more nested call_rcu() calls than five levels,
+	 * we are doing something horribly wrong...
 	 */
 	rcu_barrier();
+	rcu_barrier();
+	rcu_barrier();
+	rcu_barrier();
+	rcu_barrier();
 
 #if ISC_MEM_TRACKLINES
 	if ((ctx->debugging & ISC_MEM_DEBUGTRACE) != 0) {