From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:37611) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YuYYf-0006lr-Ly for qemu-devel@nongnu.org; Mon, 18 May 2015 23:51:10 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1YuYYe-0005OJ-Kt for qemu-devel@nongnu.org; Mon, 18 May 2015 23:51:09 -0400 From: Fam Zheng Date: Tue, 19 May 2015 11:49:45 +0000 Message-Id: <1432036186-29903-13-git-send-email-famz@redhat.com> In-Reply-To: <1432036186-29903-1-git-send-email-famz@redhat.com> References: <1432036186-29903-1-git-send-email-famz@redhat.com> Subject: [Qemu-devel] [PATCH v4 12/13] block: Block "device IO" during bdrv_drain and bdrv_drain_all List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: Kevin Wolf , qemu-block@nongnu.org, Juan Quintela , Jeff Cody , "Michael S. Tsirkin" , mreitz@redhat.com, Stefan Hajnoczi , Amit Shah , Paolo Bonzini We don't want new requests from guest, so block the operation around the nested poll. It also avoids looping forever when iothread is submitting a lot of requests. Signed-off-by: Fam Zheng --- block/io.c | 22 ++++++++++++++++++++-- 1 file changed, 20 insertions(+), 2 deletions(-) diff --git a/block/io.c b/block/io.c index 1ce62c4..b23a83f 100644 --- a/block/io.c +++ b/block/io.c @@ -286,12 +286,21 @@ static bool bdrv_drain_one(BlockDriverState *bs) * * Note that unlike bdrv_drain_all(), the caller must hold the BlockDriverState * AioContext. + * + * Devices are paused to avoid looping forever because otherwise they could + * keep submitting more requests. */ void bdrv_drain(BlockDriverState *bs) { + Error *blocker = NULL; + + error_setg(&blocker, "bdrv_drain in progress"); + bdrv_op_block(bs, BLOCK_OP_TYPE_DEVICE_IO, blocker); while (bdrv_drain_one(bs)) { /* Keep iterating */ } + bdrv_op_unblock(bs, BLOCK_OP_TYPE_DEVICE_IO, blocker); + error_free(blocker); } /* @@ -303,14 +312,20 @@ void bdrv_drain(BlockDriverState *bs) * Note that completion of an asynchronous I/O operation can trigger any * number of other I/O operations on other devices---for example a coroutine * can be arbitrarily complex and a constant flow of I/O can come until the - * coroutine is complete. Because of this, it is not possible to have a - * function to drain a single device's I/O queue. + * coroutine is complete. Because of this, we must call bdrv_drain_one in a + * loop. + * + * We explicitly pause block jobs and devices to prevent them from submitting + * more requests. */ void bdrv_drain_all(void) { /* Always run first iteration so any pending completion BHs run */ bool busy = true; BlockDriverState *bs = NULL; + Error *blocker = NULL; + + error_setg(&blocker, "bdrv_drain_all in progress"); while ((bs = bdrv_next(bs))) { AioContext *aio_context = bdrv_get_aio_context(bs); @@ -319,6 +334,7 @@ void bdrv_drain_all(void) if (bs->job) { block_job_pause(bs->job); } + bdrv_op_block(bs, BLOCK_OP_TYPE_DEVICE_IO, blocker); aio_context_release(aio_context); } @@ -343,8 +359,10 @@ void bdrv_drain_all(void) if (bs->job) { block_job_resume(bs->job); } + bdrv_op_unblock(bs, BLOCK_OP_TYPE_DEVICE_IO, blocker); aio_context_release(aio_context); } + error_free(blocker); } /** -- 2.4.1