From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:35886) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bjPC2-0002Ov-7h for qemu-devel@nongnu.org; Mon, 12 Sep 2016 07:14:31 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1bjPBz-0007oM-1k for qemu-devel@nongnu.org; Mon, 12 Sep 2016 07:14:29 -0400 Received: from mail-wm0-f67.google.com ([74.125.82.67]:33549) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bjPBy-0007gN-R7 for qemu-devel@nongnu.org; Mon, 12 Sep 2016 07:14:26 -0400 Received: by mail-wm0-f67.google.com with SMTP id b187so13088182wme.0 for ; Mon, 12 Sep 2016 04:14:05 -0700 (PDT) Sender: Paolo Bonzini From: Paolo Bonzini Date: Mon, 12 Sep 2016 13:12:38 +0200 Message-Id: <1473678761-8885-14-git-send-email-pbonzini@redhat.com> In-Reply-To: <1473678761-8885-1-git-send-email-pbonzini@redhat.com> References: <1473678761-8885-1-git-send-email-pbonzini@redhat.com> Subject: [Qemu-devel] [PATCH 13/16] cpus-common: simplify locking for start_exclusive/end_exclusive List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: sergey.fedorov@linaro.org, alex.bennee@linaro.org It is not necessary to hold qemu_cpu_list_mutex throughout the exclusive section, because no other exclusive section can run while pending_cpus != 0. exclusive_idle() is called in cpu_exec_start(), and that prevents any CPUs created after start_exclusive() from entering cpu_exec() during an exclusive section. Signed-off-by: Paolo Bonzini --- cpus-common.c | 6 ++++++ docs/tcg-exclusive.promela | 4 +++- include/qom/cpu.h | 4 ---- 3 files changed, 9 insertions(+), 5 deletions(-) diff --git a/cpus-common.c b/cpus-common.c index cabcb9f..12c8e69 100644 --- a/cpus-common.c +++ b/cpus-common.c @@ -191,11 +191,17 @@ void start_exclusive(void) if (pending_cpus > 1) { qemu_cond_wait(&exclusive_cond, &qemu_cpu_list_mutex); } + + /* Can release mutex, no one will enter another exclusive + * section until end_exclusive resets pending_cpus to 0. + */ + qemu_mutex_unlock(&qemu_cpu_list_mutex); } /* Finish an exclusive operation. Releases qemu_cpu_list_mutex. */ void end_exclusive(void) { + qemu_mutex_lock(&qemu_cpu_list_mutex); pending_cpus = 0; qemu_cond_broadcast(&exclusive_resume); qemu_mutex_unlock(&qemu_cpu_list_mutex); diff --git a/docs/tcg-exclusive.promela b/docs/tcg-exclusive.promela index 7793e31..3ef0f34 100644 --- a/docs/tcg-exclusive.promela +++ b/docs/tcg-exclusive.promela @@ -97,9 +97,11 @@ byte has_waiter[N_CPUS]; do \ :: pending_cpus > 1 -> COND_WAIT(exclusive_cond, mutex); \ :: else -> break; \ - od + od; \ + MUTEX_UNLOCK(mutex); #define end_exclusive() \ + MUTEX_LOCK(mutex); \ pending_cpus = 0; \ COND_BROADCAST(exclusive_resume); \ MUTEX_UNLOCK(mutex); diff --git a/include/qom/cpu.h b/include/qom/cpu.h index f872614..934c07a 100644 --- a/include/qom/cpu.h +++ b/include/qom/cpu.h @@ -846,9 +846,6 @@ void cpu_exec_end(CPUState *cpu); * cpu_exec are exited immediately. CPUs that call cpu_exec_start * during the exclusive section go to sleep until this CPU calls * end_exclusive. - * - * Returns with the CPU list lock taken (which nests outside all - * other locks except the BQL). */ void start_exclusive(void); @@ -856,7 +853,6 @@ void start_exclusive(void); * end_exclusive: * * Concludes an exclusive execution section started by start_exclusive. - * Releases the CPU list lock. */ void end_exclusive(void); -- 2.7.4