From owner-svn-src-all@freebsd.org Sun Dec 31 00:33:29 2017 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 6ACABEB5311; Sun, 31 Dec 2017 00:33:29 +0000 (UTC) (envelope-from mjg@FreeBSD.org) Received: from repo.freebsd.org (repo.freebsd.org [IPv6:2610:1c1:1:6068::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 3438575564; Sun, 31 Dec 2017 00:33:29 +0000 (UTC) (envelope-from mjg@FreeBSD.org) Received: from repo.freebsd.org ([127.0.1.37]) by repo.freebsd.org (8.15.2/8.15.2) with ESMTP id vBV0XS83021985; Sun, 31 Dec 2017 00:33:28 GMT (envelope-from mjg@FreeBSD.org) Received: (from mjg@localhost) by repo.freebsd.org (8.15.2/8.15.2/Submit) id vBV0XS7Y021984; Sun, 31 Dec 2017 00:33:28 GMT (envelope-from mjg@FreeBSD.org) Message-Id: <201712310033.vBV0XS7Y021984@repo.freebsd.org> X-Authentication-Warning: repo.freebsd.org: mjg set sender to mjg@FreeBSD.org using -f From: Mateusz Guzik Date: Sun, 31 Dec 2017 00:33:28 +0000 (UTC) To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: svn commit: r327394 - head/sys/kern X-SVN-Group: head X-SVN-Commit-Author: mjg X-SVN-Commit-Paths: head/sys/kern X-SVN-Commit-Revision: 327394 X-SVN-Commit-Repository: base 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: 2.1.25 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: Sun, 31 Dec 2017 00:33:29 -0000 Author: mjg Date: Sun Dec 31 00:33:28 2017 New Revision: 327394 URL: https://svnweb.freebsd.org/changeset/base/327394 Log: mtx: pre-read the lock value in thread_lock_flags_ Since this function is effectively slow path, if we get here the lock is most likely already taken in which case it is cheaper to not blindly attempt the atomic op. While here move hwpmc probe out of the loop to match other primitives. Modified: head/sys/kern/kern_mutex.c Modified: head/sys/kern/kern_mutex.c ============================================================================== --- head/sys/kern/kern_mutex.c Sun Dec 31 00:31:14 2017 (r327393) +++ head/sys/kern/kern_mutex.c Sun Dec 31 00:33:28 2017 (r327394) @@ -899,6 +899,10 @@ thread_lock_flags_(struct thread *td, int opts, const lock_delay_arg_init(&lda, &mtx_spin_delay); +#ifdef HWPMC_HOOKS + PMC_SOFT_CALL( , , lock, failed); +#endif + #ifdef LOCK_PROFILING doing_lockprof = 1; #elif defined(KDTRACE_HOOKS) @@ -908,22 +912,20 @@ thread_lock_flags_(struct thread *td, int opts, const #endif for (;;) { retry: - v = MTX_UNOWNED; spinlock_enter(); m = td->td_lock; thread_lock_validate(m, opts, file, line); + v = MTX_READ_VALUE(m); for (;;) { - if (_mtx_obtain_lock_fetch(m, &v, tid)) - break; - if (v == MTX_UNOWNED) + if (v == MTX_UNOWNED) { + if (_mtx_obtain_lock_fetch(m, &v, tid)) + break; continue; + } if (v == tid) { m->mtx_recurse++; break; } -#ifdef HWPMC_HOOKS - PMC_SOFT_CALL( , , lock, failed); -#endif lock_profile_obtain_lock_failed(&m->lock_object, &contested, &waittime); /* Give interrupts a chance while we spin. */