From patchwork Tue Oct 31 11:26:05 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Pavel Dovgalyuk X-Patchwork-Id: 117585 Delivered-To: patch@linaro.org Received: by 10.140.22.164 with SMTP id 33csp3780274qgn; Tue, 31 Oct 2017 04:29:32 -0700 (PDT) X-Google-Smtp-Source: ABhQp+QZZRJRAXF5AruhNGzWoJMJkJwFBynGFk7QEe99NLwJUz2BFZSp6nUaAMMXpC6LfR0qoF/I X-Received: by 10.37.104.133 with SMTP id d127mr915492ybc.472.1509449372936; Tue, 31 Oct 2017 04:29:32 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1509449372; cv=none; d=google.com; s=arc-20160816; b=MtAuExGosMZyg0pikEJpqooiRP8st0Q0wx+FFPeqGjiDFKrLUVY6XnrhsVkdF6B8AE Z7Z3rX7JFyEjWQ6uJ5MndYW+xzIR577fQieZ6R8QH0Muxbye9OH7HbPqRsDPhkXydEj6 GmWChvOVr7FkPKnx8dtuxjyZ13JjPeEqvAwoFf9acQ9W3tzb4YEbPpzyEwSKbi2kTbxH /qzUYFaMzBDrOpFfMCLz68dDeT5qfXp1FSSzsTRHCRWRiAzfXavyKvc0/zTui25cBqMQ mFCpFieozbENZDDV/ZVunvbG1Zgkj09VJ0Tlg6F3ajZ5dWej8JGMrNrTvWv8wg66gmuA /X1Q== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:cc:list-subscribe:list-help:list-post:list-archive :list-unsubscribe:list-id:precedence:subject :content-transfer-encoding:mime-version:user-agent:references :in-reply-to:message-id:date:to:from:arc-authentication-results; bh=R2WQnEalgsI8X96AelZsnrDU9uj/n2yJ+Otv0lXWcH8=; b=dIk05R3CRQdb9QjbUHEfrrc/b18eHXCcsPMtwlH5CCqzhXpTv9H0Nle9dDqbE93VYx x7em3GHWVemfBG8mv4Rjv+Xu+PcAWX+V4iKoSdFdw7oedhjLqUxC9ooG2GRvJtByXzC+ PzenJaY+N+HI4AKrwAT3xND0lZchdfLmz5oFkhJUHvxXQlhehgvHasKSAaq5H6e0YcWc rGToOxniXCNqoikedxMQwrBEYawdHKz4QshQwBcmKJREdTpGYvtkkt6iRrD/NplZNzXa lGbRUfO2QOO+7svp2kXPGWRd4+jHpEnM63r8aNY3D4JuwxEn0cSgrLlEJaDPQ8wj8uPf 7P4g== ARC-Authentication-Results: i=1; mx.google.com; spf=pass (google.com: domain of qemu-devel-bounces+patch=linaro.org@nongnu.org designates 2001:4830:134:3::11 as permitted sender) smtp.mailfrom=qemu-devel-bounces+patch=linaro.org@nongnu.org Return-Path: Received: from lists.gnu.org (lists.gnu.org. [2001:4830:134:3::11]) by mx.google.com with ESMTPS id b36si335564ybj.441.2017.10.31.04.29.32 for (version=TLS1 cipher=AES128-SHA bits=128/128); Tue, 31 Oct 2017 04:29:32 -0700 (PDT) Received-SPF: pass (google.com: domain of qemu-devel-bounces+patch=linaro.org@nongnu.org designates 2001:4830:134:3::11 as permitted sender) client-ip=2001:4830:134:3::11; Authentication-Results: mx.google.com; spf=pass (google.com: domain of qemu-devel-bounces+patch=linaro.org@nongnu.org designates 2001:4830:134:3::11 as permitted sender) smtp.mailfrom=qemu-devel-bounces+patch=linaro.org@nongnu.org Received: from localhost ([::1]:44883 helo=lists.gnu.org) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1e9Ujc-0005rH-Az for patch@linaro.org; Tue, 31 Oct 2017 07:29:32 -0400 Received: from eggs.gnu.org ([2001:4830:134:3::10]:50638) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1e9UgK-0003SO-KT for qemu-devel@nongnu.org; Tue, 31 Oct 2017 07:26:13 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1e9UgH-00089P-Ec for qemu-devel@nongnu.org; Tue, 31 Oct 2017 07:26:08 -0400 Received: from mail.ispras.ru ([83.149.199.45]:45146) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1e9UgH-00088B-2v for qemu-devel@nongnu.org; Tue, 31 Oct 2017 07:26:05 -0400 Received: from [127.0.1.1] (unknown [85.142.117.226]) by mail.ispras.ru (Postfix) with ESMTPSA id 3ACD654006A; Tue, 31 Oct 2017 14:26:04 +0300 (MSK) From: Pavel Dovgalyuk To: qemu-devel@nongnu.org Date: Tue, 31 Oct 2017 14:26:05 +0300 Message-ID: <20171031112605.10516.73988.stgit@pasha-VirtualBox> In-Reply-To: <20171031112457.10516.8971.stgit@pasha-VirtualBox> References: <20171031112457.10516.8971.stgit@pasha-VirtualBox> User-Agent: StGit/0.17.1-dirty MIME-Version: 1.0 X-detected-operating-system: by eggs.gnu.org: GNU/Linux 3.x [fuzzy] X-Received-From: 83.149.199.45 Subject: [Qemu-devel] [RFC PATCH 12/26] cpus: push BQL lock to qemu_*_wait_io_event X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.21 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: kwolf@redhat.com, peter.maydell@linaro.org, pavel.dovgaluk@ispras.ru, mst@redhat.com, jasowang@redhat.com, quintela@redhat.com, zuban32s@gmail.com, maria.klimushenkova@ispras.ru, dovgaluk@ispras.ru, kraxel@redhat.com, boost.lists@gmail.com, pbonzini@redhat.com, alex.bennee@linaro.org Errors-To: qemu-devel-bounces+patch=linaro.org@nongnu.org Sender: "Qemu-devel" From: Alex Bennée We only really need to grab the lock for initial setup (so we don't race with the thread-spawning thread). After that we can drop the lock for the whole main loop and only grab it for waiting for IO events. There is a slight wrinkle for the round-robin TCG thread as we also expire timers which needs to be done under BQL as they are in the main-loop. This is stage one of reducing the lock impact as we can drop the requirement of implicit BQL for async work and only grab the lock when we need to sleep on the cpu->halt_cond. Signed-off-by: Alex Bennée Tested-by: Pavel Dovgalyuk --- accel/kvm/kvm-all.c | 4 ---- cpus.c | 27 ++++++++++++++++++++------- target/i386/hax-all.c | 3 +-- 3 files changed, 21 insertions(+), 13 deletions(-) diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c index f290f48..8d1d2c4 100644 --- a/accel/kvm/kvm-all.c +++ b/accel/kvm/kvm-all.c @@ -1857,9 +1857,7 @@ int kvm_cpu_exec(CPUState *cpu) return EXCP_HLT; } - qemu_mutex_unlock_iothread(); cpu_exec_start(cpu); - do { MemTxAttrs attrs; @@ -1989,8 +1987,6 @@ int kvm_cpu_exec(CPUState *cpu) } while (ret == 0); cpu_exec_end(cpu); - qemu_mutex_lock_iothread(); - if (ret < 0) { cpu_dump_state(cpu, stderr, fprintf, CPU_DUMP_CODE); vm_stop(RUN_STATE_INTERNAL_ERROR); diff --git a/cpus.c b/cpus.c index 2eec54f..efde5c1 100644 --- a/cpus.c +++ b/cpus.c @@ -1127,6 +1127,8 @@ static bool qemu_tcg_should_sleep(CPUState *cpu) static void qemu_tcg_wait_io_event(CPUState *cpu) { + qemu_mutex_lock_iothread(); + while (qemu_tcg_should_sleep(cpu)) { stop_tcg_kick_timer(); qemu_cond_wait(cpu->halt_cond, &qemu_global_mutex); @@ -1135,15 +1137,21 @@ static void qemu_tcg_wait_io_event(CPUState *cpu) start_tcg_kick_timer(); qemu_wait_io_event_common(cpu); + + qemu_mutex_unlock_iothread(); } static void qemu_kvm_wait_io_event(CPUState *cpu) { + qemu_mutex_lock_iothread(); + while (cpu_thread_is_idle(cpu)) { qemu_cond_wait(cpu->halt_cond, &qemu_global_mutex); } qemu_wait_io_event_common(cpu); + + qemu_mutex_unlock_iothread(); } static void *qemu_kvm_cpu_thread_fn(void *arg) @@ -1169,6 +1177,8 @@ static void *qemu_kvm_cpu_thread_fn(void *arg) /* signal CPU creation */ cpu->created = true; + qemu_mutex_unlock_iothread(); + qemu_cond_signal(&qemu_cpu_cond); do { @@ -1211,10 +1221,10 @@ static void *qemu_dummy_cpu_thread_fn(void *arg) /* signal CPU creation */ cpu->created = true; + qemu_mutex_unlock_iothread(); qemu_cond_signal(&qemu_cpu_cond); while (1) { - qemu_mutex_unlock_iothread(); do { int sig; r = sigwait(&waitset, &sig); @@ -1225,6 +1235,7 @@ static void *qemu_dummy_cpu_thread_fn(void *arg) } qemu_mutex_lock_iothread(); qemu_wait_io_event_common(cpu); + qemu_mutex_unlock_iothread(); } return NULL; @@ -1313,11 +1324,9 @@ static int tcg_cpu_exec(CPUState *cpu) #ifdef CONFIG_PROFILER ti = profile_getclock(); #endif - qemu_mutex_unlock_iothread(); cpu_exec_start(cpu); ret = cpu_exec(cpu); cpu_exec_end(cpu); - qemu_mutex_lock_iothread(); #ifdef CONFIG_PROFILER tcg_time += profile_getclock() - ti; #endif @@ -1377,6 +1386,7 @@ static void *qemu_tcg_rr_cpu_thread_fn(void *arg) qemu_wait_io_event_common(cpu); } } + qemu_mutex_unlock_iothread(); start_tcg_kick_timer(); @@ -1386,6 +1396,9 @@ static void *qemu_tcg_rr_cpu_thread_fn(void *arg) cpu->exit_request = 1; while (1) { + + qemu_mutex_lock_iothread(); + /* Account partial waits to QEMU_CLOCK_VIRTUAL. */ qemu_account_warp_timer(); @@ -1394,6 +1407,8 @@ static void *qemu_tcg_rr_cpu_thread_fn(void *arg) */ handle_icount_deadline(); + qemu_mutex_unlock_iothread(); + if (!cpu) { cpu = first_cpu; } @@ -1419,9 +1434,7 @@ static void *qemu_tcg_rr_cpu_thread_fn(void *arg) cpu_handle_guest_debug(cpu); break; } else if (r == EXCP_ATOMIC) { - qemu_mutex_unlock_iothread(); cpu_exec_step_atomic(cpu); - qemu_mutex_lock_iothread(); break; } } else if (cpu->stop) { @@ -1462,6 +1475,7 @@ static void *qemu_hax_cpu_thread_fn(void *arg) current_cpu = cpu; hax_init_vcpu(cpu); + qemu_mutex_unlock_iothread(); qemu_cond_signal(&qemu_cpu_cond); while (1) { @@ -1512,6 +1526,7 @@ static void *qemu_tcg_cpu_thread_fn(void *arg) cpu->created = true; cpu->can_do_io = 1; current_cpu = cpu; + qemu_mutex_unlock_iothread(); qemu_cond_signal(&qemu_cpu_cond); /* process any pending work */ @@ -1536,9 +1551,7 @@ static void *qemu_tcg_cpu_thread_fn(void *arg) g_assert(cpu->halted); break; case EXCP_ATOMIC: - qemu_mutex_unlock_iothread(); cpu_exec_step_atomic(cpu); - qemu_mutex_lock_iothread(); default: /* Ignore everything else? */ break; diff --git a/target/i386/hax-all.c b/target/i386/hax-all.c index 3ce6950..99af6bb 100644 --- a/target/i386/hax-all.c +++ b/target/i386/hax-all.c @@ -513,11 +513,10 @@ static int hax_vcpu_hax_exec(CPUArchState *env) hax_vcpu_interrupt(env); - qemu_mutex_unlock_iothread(); cpu_exec_start(cpu); hax_ret = hax_vcpu_run(vcpu); + current_cpu = cpu; cpu_exec_end(cpu); - qemu_mutex_lock_iothread(); /* Simply continue the vcpu_run if system call interrupted */ if (hax_ret == -EINTR || hax_ret == -EAGAIN) {