The Tuesday 24 Jun 2014 à 16:08:57 (+0200), Kevin Wolf wrote : > Am 24.06.2014 um 15:32 hat Jeff Cody geschrieben: > > On Tue, Jun 24, 2014 at 10:48:52AM +0800, Fam Zheng wrote: > > > On Mon, 06/23 21:08, Stefan Hajnoczi wrote: > > > > On Thu, Jun 19, 2014 at 12:26:00PM -0400, Jeff Cody wrote: > > > > > On Thu, Jun 19, 2014 at 05:17:16PM +0800, Stefan Hajnoczi wrote: > > > > > > On Tue, Jun 17, 2014 at 05:53:48PM -0400, Jeff Cody wrote: > > > > > > Let's discuss this topic in a sub-thread and figure out what to do > > > > > > for > > > > > > QEMU 2.1. This is an important issue to solve before the release > > > > > > because we can't change QMP command semantics easily later. > > > > > > > > > > > > My questions are: > > > > > > a. How do we fix resize, snapshot-sync, etc? It seems like we need > > > > > > to > > > > > > propagate child op blockers. > > > > > > > > > > > > b. Is it a good idea to perform op blocker checks on the root node? > > > > > > It's inconsistent with resize, snapshot-sync, etc. Permissions > > > > > > in > > > > > > BDS graphs with multiple root nodes (e.g. guest device and NBD > > > > > > run-time server) will be different depending on which root you > > > > > > specify. > > > > > > > > > > I don't think (b) is the ultimate solution. It is used as a stop-gap > > > > > because op blockers in the current implementation is essentially > > > > > analogous to the in-use flag. But is it good enough for 2.1? If > > > > > *everything* checks the topmost node in 2.1, then I think we are OK in > > > > > all cases except where images files share a common BDS. > > > > > > > > Checking op blockers on the root node as a stop-gap is a good idea. > > > > Let's apply it across all commands (e.g. snapshot-sync, resize). > > > > > > > > Fam pointed out that this approach is vulnerable to blockdev-add, where > > > > blockers could be set/checked on an incomplete BDS graph (since you can > > > > add new nodes on top). Do we need to move the blockers up the graph if > > > > a new root node is inserted? > > > > > > My concern is if we allow adding new root on top, it's not easy to know > > > the > > > real root then. > > > > > > To give an example: > > > > > > If we have > > > > > > [base id=""] <- [active id="drive0" blockers=...] > > > > > > When user does > > > > > > (QMP) block-commit device="drive0" ... > > > > > > We should check drive0, which is OK. > > > > > > Then, assume user adds a new root on top, we would take care of moving the > > > blockers: > > > > > > [base id=""] <- [active id="drive0"] <- [active id="drive1" blockers=] > > > > > > At this point, what if user does something on drive0 again? > > > > > > (QMP) block-commit device="drive0" ... > > > > > > The right thing to do is to check blockers on "drive1", since it's the > > > real > > > root now. But how do we know? Do we need to add a back reference pointer > > > ->overlap_hd in BDS, or do we maintain a look up table, or do we search > > > all BDS > > > graphs to figure out? > > > > > > None is easier than if we put the blockers in the bottom BDS, in the first > > > place: > > > > > > [base id="" blockers=...] <- [active id="drive0"] > > > ^^^^^^^^^^^^ > > > > > > > I think you are right. If we place the blocker at the bottom-most > > BDS, then that would be a more restrictive blocker. This may end up > > being more restrictive than needed, but more importantly it should > > make everything safe. > > > > Also, it is an easy change for 2.1 - just call bdrv_find_base(bs), and > > set/check/clear blockers on the returned BDS. > > What does bdrv_find_base() return for e.g. quorum?
This will not work when unblocking a BDS loop like the one formed by drive-mirror when replacing an arbitrary node. > > Kevin > > > > Even if user adds a new root, we don't need to worry about moving > > > blockers, > > > because the bottom is not changed. > > > > > > [base id="" blockers=...] <- [active id="drive0"] <- [active > > > id="drive1"] > > > > > > Checking the blockers are easy, either for drive0 or drive1: just follow > > > the > > > backing chain until getting to the end. > > > > > > Fam >
