From owner-svn-src-all@freebsd.org Tue Sep 8 08:41:08 2015 Return-Path: Delivered-To: svn-src-all@mailman.ysv.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) by mailman.ysv.freebsd.org (Postfix) with ESMTP id 2B961A00531; Tue, 8 Sep 2015 08:41:08 +0000 (UTC) (envelope-from kib@FreeBSD.org) Received: from repo.freebsd.org (repo.freebsd.org [IPv6:2001:1900:2254:2068::e6a:0]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (Client did not present a certificate) by mx1.freebsd.org (Postfix) with ESMTPS id 1CA2C1E03; Tue, 8 Sep 2015 08:41:08 +0000 (UTC) (envelope-from kib@FreeBSD.org) Received: from repo.freebsd.org ([127.0.1.70]) by repo.freebsd.org (8.15.2/8.15.2) with ESMTP id t888f7I1082187; Tue, 8 Sep 2015 08:41:07 GMT (envelope-from kib@FreeBSD.org) Received: (from kib@localhost) by repo.freebsd.org (8.15.2/8.15.2/Submit) id t888f76o082185; Tue, 8 Sep 2015 08:41:07 GMT (envelope-from kib@FreeBSD.org) Message-Id: <201509080841.t888f76o082185@repo.freebsd.org> X-Authentication-Warning: repo.freebsd.org: kib set sender to kib@FreeBSD.org using -f From: Konstantin Belousov Date: Tue, 8 Sep 2015 08:41:07 +0000 (UTC) To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: svn commit: r287556 - head/lib/libthr/thread X-SVN-Group: head MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-BeenThere: svn-src-all@freebsd.org X-Mailman-Version: 2.1.20 Precedence: list List-Id: "SVN commit messages for the entire src tree \(except for " user" and " projects" \)" List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 08 Sep 2015 08:41:08 -0000 Author: kib Date: Tue Sep 8 08:41:07 2015 New Revision: 287556 URL: https://svnweb.freebsd.org/changeset/base/287556 Log: In the pthread_once(), if the initializer has already run, then the calling thread is supposed to see accesses issued by the initializer. This means that the read of the once_control->state variable should have an acquire semantic. Use atomic_thread_fence_acq() when the value read is ONCE_DONE, instead of straightforward atomic_load_acq(), to only put a barrier when needed (*). On the other hand, the updates of the once_control->state with the intermediate progress state do not need to synchronize with other state accesses, remove _acq suffix. Reviewed by: alc (previous version) Suggested by: alc (*) Sponsored by: The FreeBSD Foundation MFC after: 1 week Modified: head/lib/libthr/thread/thr_once.c Modified: head/lib/libthr/thread/thr_once.c ============================================================================== --- head/lib/libthr/thread/thr_once.c Tue Sep 8 08:06:20 2015 (r287555) +++ head/lib/libthr/thread/thr_once.c Tue Sep 8 08:41:07 2015 (r287556) @@ -68,13 +68,15 @@ _pthread_once(pthread_once_t *once_contr for (;;) { state = once_control->state; - if (state == ONCE_DONE) + if (state == ONCE_DONE) { + atomic_thread_fence_acq(); return (0); + } if (state == ONCE_NEVER_DONE) { - if (atomic_cmpset_acq_int(&once_control->state, state, ONCE_IN_PROGRESS)) + if (atomic_cmpset_int(&once_control->state, state, ONCE_IN_PROGRESS)) break; } else if (state == ONCE_IN_PROGRESS) { - if (atomic_cmpset_acq_int(&once_control->state, state, ONCE_WAIT)) + if (atomic_cmpset_int(&once_control->state, state, ONCE_WAIT)) _thr_umtx_wait_uint(&once_control->state, ONCE_WAIT, NULL, 0); } else if (state == ONCE_WAIT) { _thr_umtx_wait_uint(&once_control->state, state, NULL, 0);