From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:51866) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1fUxHQ-0004sm-VE for qemu-devel@nongnu.org; Mon, 18 Jun 2018 12:45:26 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1fUxHP-0006Hb-Mz for qemu-devel@nongnu.org; Mon, 18 Jun 2018 12:45:24 -0400 From: Kevin Wolf Date: Mon, 18 Jun 2018 18:44:41 +0200 Message-Id: <20180618164504.24488-13-kwolf@redhat.com> In-Reply-To: <20180618164504.24488-1-kwolf@redhat.com> References: <20180618164504.24488-1-kwolf@redhat.com> Subject: [Qemu-devel] [PULL 12/35] block: Don't poll in parent drain callbacks List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-block@nongnu.org Cc: kwolf@redhat.com, qemu-devel@nongnu.org bdrv_do_drained_begin() is only safe if we have a single BDRV_POLL_WHILE() after quiescing all affected nodes. We cannot allow that parent callbacks introduce a nested polling loop that could cause graph changes while we're traversing the graph. Split off bdrv_do_drained_begin_quiesce(), which only quiesces a single node without waiting for its requests to complete. These requests will be waited for in the BDRV_POLL_WHILE() call down the call chain. Signed-off-by: Kevin Wolf --- include/block/block.h | 9 +++++++++ block.c | 2 +- block/io.c | 24 ++++++++++++++++-------- 3 files changed, 26 insertions(+), 9 deletions(-) diff --git a/include/block/block.h b/include/block/block.h index 254ed2e4c9..067d24cc4a 100644 --- a/include/block/block.h +++ b/include/block/block.h @@ -591,6 +591,15 @@ bool bdrv_drain_poll(BlockDriverState *bs, bool recursive, void bdrv_drained_begin(BlockDriverState *bs); /** + * bdrv_do_drained_begin_quiesce: + * + * Quiesces a BDS like bdrv_drained_begin(), but does not wait for already + * running requests to complete. + */ +void bdrv_do_drained_begin_quiesce(BlockDriverState *bs, + BdrvChild *parent); + +/** * Like bdrv_drained_begin, but recursively begins a quiesced section for * exclusive access to all child nodes as well. */ diff --git a/block.c b/block.c index 80abd3c2ae..50f8e3dc3b 100644 --- a/block.c +++ b/block.c @@ -818,7 +818,7 @@ static char *bdrv_child_get_parent_desc(BdrvChild *c) static void bdrv_child_cb_drained_begin(BdrvChild *child) { BlockDriverState *bs = child->opaque; - bdrv_drained_begin(bs); + bdrv_do_drained_begin_quiesce(bs, NULL); } static bool bdrv_child_cb_drained_poll(BdrvChild *child) diff --git a/block/io.c b/block/io.c index a413841bfc..ffb273708a 100644 --- a/block/io.c +++ b/block/io.c @@ -286,15 +286,10 @@ static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs, assert(data.done); } -void bdrv_do_drained_begin(BlockDriverState *bs, bool recursive, - BdrvChild *parent, bool poll) +void bdrv_do_drained_begin_quiesce(BlockDriverState *bs, + BdrvChild *parent) { - BdrvChild *child, *next; - - if (qemu_in_coroutine()) { - bdrv_co_yield_to_drain(bs, true, recursive, parent, poll); - return; - } + assert(!qemu_in_coroutine()); /* Stop things in parent-to-child order */ if (atomic_fetch_inc(&bs->quiesce_counter) == 0) { @@ -303,6 +298,19 @@ void bdrv_do_drained_begin(BlockDriverState *bs, bool recursive, bdrv_parent_drained_begin(bs, parent); bdrv_drain_invoke(bs, true); +} + +static void bdrv_do_drained_begin(BlockDriverState *bs, bool recursive, + BdrvChild *parent, bool poll) +{ + BdrvChild *child, *next; + + if (qemu_in_coroutine()) { + bdrv_co_yield_to_drain(bs, true, recursive, parent, poll); + return; + } + + bdrv_do_drained_begin_quiesce(bs, parent); if (recursive) { bs->recursive_quiesce_counter++; -- 2.13.6