From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:46647) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1X4oZ9-0005PX-Qj for qemu-devel@nongnu.org; Wed, 09 Jul 2014 05:53:38 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1X4oZ2-0000z8-SB for qemu-devel@nongnu.org; Wed, 09 Jul 2014 05:53:31 -0400 Received: from mail-qa0-x229.google.com ([2607:f8b0:400d:c00::229]:62968) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1X4oZ2-0000yy-Mh for qemu-devel@nongnu.org; Wed, 09 Jul 2014 05:53:24 -0400 Received: by mail-qa0-f41.google.com with SMTP id cm18so6088652qab.28 for ; Wed, 09 Jul 2014 02:53:24 -0700 (PDT) Received: from yakj.usersys.redhat.com (net-2-35-201-190.cust.vodafonedsl.it. [2.35.201.190]) by mx.google.com with ESMTPSA id j97sm31669122qgd.37.2014.07.09.02.53.22 for (version=TLSv1.2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Wed, 09 Jul 2014 02:53:23 -0700 (PDT) Sender: Paolo Bonzini From: Paolo Bonzini Date: Wed, 9 Jul 2014 11:53:01 +0200 Message-Id: <1404899590-24973-2-git-send-email-pbonzini@redhat.com> In-Reply-To: <1404899590-24973-1-git-send-email-pbonzini@redhat.com> References: <1404899590-24973-1-git-send-email-pbonzini@redhat.com> Subject: [Qemu-devel] [PATCH 01/10] AioContext: take bottom halves into account when computing aio_poll timeout List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Right now, QEMU invokes aio_bh_poll before the "poll" phase of aio_poll. It is simpler to do it afterwards and skip the "poll" phase altogether when the OS-dependent parts of AioContext are invoked from GSource. This way, AioContext behaves more similarly when used as a GSource vs. when used as stand-alone. As a start, take bottom halves into account when computing the poll timeout. If a bottom half is ready, do a non-blocking poll. As a side effect, this makes idle bottom halves work with aio_poll; an improvement, but not really an important one since they are deprecated. Signed-off-by: Paolo Bonzini --- aio-posix.c | 2 +- aio-win32.c | 4 ++-- async.c | 32 ++++++++++++++++++-------------- include/block/aio.h | 8 ++++++++ 4 files changed, 29 insertions(+), 17 deletions(-) diff --git a/aio-posix.c b/aio-posix.c index 2eada2e..55706f8 100644 --- a/aio-posix.c +++ b/aio-posix.c @@ -249,7 +249,7 @@ bool aio_poll(AioContext *ctx, bool blocking) /* wait until next event */ ret = qemu_poll_ns((GPollFD *)ctx->pollfds->data, ctx->pollfds->len, - blocking ? timerlistgroup_deadline_ns(&ctx->tlg) : 0); + blocking ? aio_compute_timeout(ctx) : 0); /* if we have any readable fds, dispatch event */ if (ret > 0) { diff --git a/aio-win32.c b/aio-win32.c index c12f61e..fe7ee5b 100644 --- a/aio-win32.c +++ b/aio-win32.c @@ -165,8 +165,8 @@ bool aio_poll(AioContext *ctx, bool blocking) while (count > 0) { int ret; - timeout = blocking ? - qemu_timeout_ns_to_ms(timerlistgroup_deadline_ns(&ctx->tlg)) : 0; + timeout = blocking + ? qemu_timeout_ns_to_ms(aio_compute_timeout(ctx)) : 0; ret = WaitForMultipleObjects(count, events, FALSE, timeout); /* if we have any signaled events, dispatch event */ diff --git a/async.c b/async.c index 34af0b2..ac40eab 100644 --- a/async.c +++ b/async.c @@ -152,39 +152,43 @@ void qemu_bh_delete(QEMUBH *bh) bh->deleted = 1; } -static gboolean -aio_ctx_prepare(GSource *source, gint *timeout) +int +aio_compute_timeout(AioContext *ctx) { - AioContext *ctx = (AioContext *) source; + int64_t deadline; + int timeout = -1; QEMUBH *bh; - int deadline; - /* We assume there is no timeout already supplied */ - *timeout = -1; for (bh = ctx->first_bh; bh; bh = bh->next) { if (!bh->deleted && bh->scheduled) { if (bh->idle) { /* idle bottom halves will be polled at least * every 10ms */ - *timeout = 10; + timeout = 10000000; } else { /* non-idle bottom halves will be executed * immediately */ - *timeout = 0; - return true; + return 0; } } } - deadline = qemu_timeout_ns_to_ms(timerlistgroup_deadline_ns(&ctx->tlg)); + deadline = timerlistgroup_deadline_ns(&ctx->tlg); if (deadline == 0) { - *timeout = 0; - return true; + return 0; } else { - *timeout = qemu_soonest_timeout(*timeout, deadline); + return qemu_soonest_timeout(timeout, deadline); } +} - return false; +static gboolean +aio_ctx_prepare(GSource *source, gint *timeout) +{ + AioContext *ctx = (AioContext *) source; + + /* We assume there is no timeout already supplied */ + *timeout = qemu_timeout_ns_to_ms(aio_compute_timeout(ctx)); + return *timeout == 0; } static gboolean diff --git a/include/block/aio.h b/include/block/aio.h index c23de3c..7eeb961 100644 --- a/include/block/aio.h +++ b/include/block/aio.h @@ -303,4 +303,12 @@ static inline void aio_timer_init(AioContext *ctx, timer_init(ts, ctx->tlg.tl[type], scale, cb, opaque); } +/** + * aio_compute_timeout: + * @ctx: the aio context + * + * Compute the timeout that a blocking aio_poll should use. + */ +int aio_compute_timeout(AioContext *ctx); + #endif -- 1.9.3