From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:52651) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Zoy2Q-0007WP-3y for qemu-devel@nongnu.org; Wed, 21 Oct 2015 14:23:03 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1Zoy2P-0006UP-84 for qemu-devel@nongnu.org; Wed, 21 Oct 2015 14:23:02 -0400 Date: Wed, 21 Oct 2015 14:22:54 -0400 From: Jeff Cody Message-ID: <20151021182254.GH6466@localhost.localdomain> References: <1445393209-26545-1-git-send-email-famz@redhat.com> <1445393209-26545-10-git-send-email-famz@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1445393209-26545-10-git-send-email-famz@redhat.com> Subject: Re: [Qemu-devel] [Qemu-block] [PATCH v5 09/12] block: Add "drained begin/end" for internal snapshot List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Fam Zheng Cc: kwolf@redhat.com, pbonzini@redhat.com, qemu-block@nongnu.org, qemu-devel@nongnu.org, stefanha@redhat.com On Wed, Oct 21, 2015 at 10:06:46AM +0800, Fam Zheng wrote: > This ensures the atomicity of the transaction by avoiding processing of > external requests such as those from ioeventfd. > > state->bs is assigned right after bdrv_drained_begin. Because it was > used as the flag for deletion or not in abort, now we need a separate > flag - InternalSnapshotState.created. > > Signed-off-by: Fam Zheng > --- > blockdev.c | 8 ++++++-- > 1 file changed, 6 insertions(+), 2 deletions(-) > > diff --git a/blockdev.c b/blockdev.c > index 52f44b2..92c2d0d 100644 > --- a/blockdev.c > +++ b/blockdev.c > @@ -1370,6 +1370,7 @@ typedef struct InternalSnapshotState { > BlockDriverState *bs; > AioContext *aio_context; > QEMUSnapshotInfo sn; > + bool created; > } InternalSnapshotState; > > static void internal_snapshot_prepare(BlkTransactionState *common, > @@ -1407,6 +1408,8 @@ static void internal_snapshot_prepare(BlkTransactionState *common, > /* AioContext is released in .clean() */ > state->aio_context = blk_get_aio_context(blk); > aio_context_acquire(state->aio_context); > + state->bs = blk_bs(blk); > + bdrv_drained_begin(state->bs); > Do you think it'd make sense to move the context acquisition & call to bdrv_drained_begin() below the check for blk_is_available()? That way the assignment of state->bs doesn't happen until after the call to blk_is_available(). It looks like state->created protects us either way, though - but if blk_is_inserted() fails, then there is no need for _abort or _clean to do anything, if we hold off on grabbing the context. I think I am fine with it as is though, too. > if (!blk_is_available(blk)) { > error_setg(errp, QERR_DEVICE_HAS_NO_MEDIUM, device); > @@ -1465,7 +1468,7 @@ static void internal_snapshot_prepare(BlkTransactionState *common, > } > > /* 4. succeed, mark a snapshot is created */ > - state->bs = bs; > + state->created = true; > } > > static void internal_snapshot_abort(BlkTransactionState *common) > @@ -1476,7 +1479,7 @@ static void internal_snapshot_abort(BlkTransactionState *common) > QEMUSnapshotInfo *sn = &state->sn; > Error *local_error = NULL; > > - if (!bs) { > + if (!state->created) { > return; > } > > @@ -1497,6 +1500,7 @@ static void internal_snapshot_clean(BlkTransactionState *common) > common, common); > > if (state->aio_context) { > + bdrv_drained_end(state->bs); > aio_context_release(state->aio_context); > } > } > -- > 2.4.3 > >