[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Qemu-devel] [PATCH v2 00/11] Fix transactional snapshot with virtio-blk
[Qemu-devel] [PATCH v2 00/11] Fix transactional snapshot with virtio-blk dataplane and NBD export
Wed, 13 May 2015 17:28:08 +0000
Changes from RFC:
- Add op blocker listener in nbd server.
- Add other transaction types.
- Only notify listeners when changing from/to empty. (Paolo)
Reported by Paolo.
Unlike the iohandler in main loop, iothreads currently process the event
notifier used by virtio-blk ioeventfd in nested aio_poll. This is dangerous
without proper protection, because guest requests could sneak to block layer
where they mustn't.
For example, a QMP transaction may involve multiple bdrv_drain_all() in
handling the list of AioContext it works on. If an aio_poll in one of the
bdrv_drain_all() happens to process a guest VQ kick, and dispatches the
ioeventfd event to virtio-blk, a new guest write is then submitted, and voila,
the transaction semantics is violated.
This series avoids this problem by disabling virtio-blk handlers during
bdrv_drain_all() and transactions.
- Patches 1~3 add the block layer op blocker change notifier code.
- Patches 4,5 secure virtio-blk dataplane.
- Patch 6 secures nbd export.
- Patch 7~10 protect each transaction type from being voilated by new IO
generated in nested aio_poll.
- Patch 11 protects bdrv_drain and bdrv_drain_all.
virtio-scsi-dataplane will be a bit more complicated, but still doable. It
would probably need one more interface abstraction between scsi-disk, scsi-bus
Although other devices don't have a pause/resume callback yet, the
blk_check_request, which returns -EBUSY if "device io" op blocker is set, could
hopefully cover most cases already.
Timers and block jobs also generate IO, but it should be fine as long as they
don't change guest visible data, which is true AFAICT. Besides, bdrv_drain_all
already pauses block jobs.
Fam Zheng (11):
block: Add op blocker type "device IO"
block: Add op blocker notifier list
block-backend: Add blk_op_blocker_add_notifier
virtio-blk: Move complete_request to 'ops' structure
virtio-blk: Don't handle output when there is "device IO" op blocker
nbd-server: Clear "can_read" when "device io" blocker is set
blockdev: Block device IO during internal snapshot transaction
blockdev: Block device IO during external snapshot transaction
blockdev: Block device IO during drive-backup transaction
blockdev: Block device IO during blockdev-backup transaction
block: Block "device IO" during bdrv_drain and bdrv_drain_all
block.c | 28 +++++++++++++++++
block/block-backend.c | 10 ++++++
block/io.c | 12 +++++++
blockdev.c | 49 ++++++++++++++++++++++++-----
hw/block/dataplane/virtio-blk.c | 36 ++++++++++++++++++---
hw/block/virtio-blk.c | 69 +++++++++++++++++++++++++++++++++++++++--
include/block/block.h | 9 ++++++
include/block/block_int.h | 3 ++
include/hw/virtio/virtio-blk.h | 17 ++++++++--
include/sysemu/block-backend.h | 2 ++
nbd.c | 18 +++++++++++
11 files changed, 235 insertions(+), 18 deletions(-)
[Qemu-devel] [PATCH v2 02/11] block: Add op blocker notifier list, Fam Zheng, 2015/05/13
[Qemu-devel] [PATCH v2 04/11] virtio-blk: Move complete_request to 'ops' structure, Fam Zheng, 2015/05/13
[Qemu-devel] [PATCH v2 03/11] block-backend: Add blk_op_blocker_add_notifier, Fam Zheng, 2015/05/13
- [Qemu-devel] [PATCH v2 00/11] Fix transactional snapshot with virtio-blk dataplane and NBD export,
Fam Zheng <=