* Re: [Qemu-devel] [PATCH v3 05/12] block: Introduce "drained begin/end" API
@ 2015-10-18 23:22 Fam Zheng
0 siblings, 0 replies; 2+ messages in thread
From: Fam Zheng @ 2015-10-18 23:22 UTC (permalink / raw)
To: Fam Zheng; +Cc: kwolf, pbonzini, stefanha, qemu-devel, qemu-block
I should have updated bdrv_move_feature_bits as Kevin pointed out in the previous revision. I'll rebase on to the pull req that removes bdrv_swap anyway.
Thanks,
Fam
^ permalink raw reply [flat|nested] 2+ messages in thread
* [Qemu-devel] [PATCH v3 00/12] block: Protect nested event loop with bdrv_drained_begin and bdrv_drained_end
@ 2015-10-16 10:12 Fam Zheng
2015-10-16 10:12 ` [Qemu-devel] [PATCH v3 05/12] block: Introduce "drained begin/end" API Fam Zheng
0 siblings, 1 reply; 2+ messages in thread
From: Fam Zheng @ 2015-10-16 10:12 UTC (permalink / raw)
To: qemu-devel; +Cc: kwolf, pbonzini, stefanha, qemu-block
v3: Call bdrv_drain unconditionally in bdrv_drained_begin.
Document the internal I/O implications between bdrv_drain_begin and end.
The nested aio_poll()'s in block layer has a bug that new r/w requests from
ioeventfds and nbd exports are processed, which might break the caller's
semantics (qmp_transaction) or even pointers (bdrv_reopen).
Fam Zheng (12):
aio: Add "is_external" flag for event handlers
nbd: Mark fd handlers client type as "external"
dataplane: Mark host notifiers' client type as "external"
aio: introduce aio_{disable,enable}_external
block: Introduce "drained begin/end" API
block: Add "drained begin/end" for transactional external snapshot
block: Add "drained begin/end" for transactional backup
block: Add "drained begin/end" for transactional blockdev-backup
block: Add "drained begin/end" for internal snapshot
block: Introduce BlockDriver.bdrv_drain callback
qed: Implement .bdrv_drain
tests: Add test case for aio_disable_external
aio-posix.c | 9 ++++-
aio-win32.c | 8 +++-
async.c | 3 +-
block.c | 2 +
block/curl.c | 14 ++++---
block/io.c | 23 +++++++++++-
block/iscsi.c | 9 ++---
block/linux-aio.c | 5 ++-
block/nbd-client.c | 10 +++--
block/nfs.c | 17 ++++-----
block/qed.c | 7 ++++
block/sheepdog.c | 38 ++++++++++++-------
block/ssh.c | 5 ++-
block/win32-aio.c | 5 ++-
blockdev.c | 27 +++++++++++---
hw/block/dataplane/virtio-blk.c | 5 ++-
hw/scsi/virtio-scsi-dataplane.c | 22 +++++++----
include/block/aio.h | 39 ++++++++++++++++++++
include/block/block.h | 24 ++++++++++++
include/block/block_int.h | 8 ++++
iohandler.c | 3 +-
nbd.c | 4 +-
tests/test-aio.c | 82 ++++++++++++++++++++++++++++-------------
23 files changed, 276 insertions(+), 93 deletions(-)
--
2.4.3
^ permalink raw reply [flat|nested] 2+ messages in thread
* [Qemu-devel] [PATCH v3 05/12] block: Introduce "drained begin/end" API
2015-10-16 10:12 [Qemu-devel] [PATCH v3 00/12] block: Protect nested event loop with bdrv_drained_begin and bdrv_drained_end Fam Zheng
@ 2015-10-16 10:12 ` Fam Zheng
0 siblings, 0 replies; 2+ messages in thread
From: Fam Zheng @ 2015-10-16 10:12 UTC (permalink / raw)
To: qemu-devel; +Cc: kwolf, pbonzini, stefanha, qemu-block
The semantics is that after bdrv_drained_begin(bs), bs will not get new external
requests until the matching bdrv_drained_end(bs).
Signed-off-by: Fam Zheng <famz@redhat.com>
---
block.c | 2 ++
block/io.c | 17 +++++++++++++++++
include/block/block.h | 19 +++++++++++++++++++
include/block/block_int.h | 2 ++
4 files changed, 40 insertions(+)
diff --git a/block.c b/block.c
index 1f90b47..9b28a07 100644
--- a/block.c
+++ b/block.c
@@ -2058,6 +2058,8 @@ static void bdrv_move_feature_fields(BlockDriverState *bs_dest,
bs_dest->device_list = bs_src->device_list;
bs_dest->blk = bs_src->blk;
+ bs_dest->quiesce_counter = bs_src->quiesce_counter;
+
memcpy(bs_dest->op_blockers, bs_src->op_blockers,
sizeof(bs_dest->op_blockers));
}
diff --git a/block/io.c b/block/io.c
index 17293c3..a331a19 100644
--- a/block/io.c
+++ b/block/io.c
@@ -2618,3 +2618,20 @@ void bdrv_flush_io_queue(BlockDriverState *bs)
}
bdrv_start_throttled_reqs(bs);
}
+
+void bdrv_drained_begin(BlockDriverState *bs)
+{
+ if (!bs->quiesce_counter++) {
+ aio_disable_external(bdrv_get_aio_context(bs));
+ }
+ bdrv_drain(bs);
+}
+
+void bdrv_drained_end(BlockDriverState *bs)
+{
+ assert(bs->quiesce_counter > 0);
+ if (--bs->quiesce_counter > 0) {
+ return;
+ }
+ aio_enable_external(bdrv_get_aio_context(bs));
+}
diff --git a/include/block/block.h b/include/block/block.h
index 2dd6630..c4f6eef 100644
--- a/include/block/block.h
+++ b/include/block/block.h
@@ -619,4 +619,23 @@ void bdrv_flush_io_queue(BlockDriverState *bs);
BlockAcctStats *bdrv_get_stats(BlockDriverState *bs);
+/**
+ * bdrv_drained_begin:
+ *
+ * Begin a quiesced section for exclusive access to the BDS, by disabling
+ * external request sources including NBD server and device model. Note that
+ * this doesn't block timers or coroutines from submitting more requests, which
+ * means block_job_pause is still necessary.
+ *
+ * This function can be recursive.
+ */
+void bdrv_drained_begin(BlockDriverState *bs);
+
+/**
+ * bdrv_drained_end:
+ *
+ * End a quiescent section started by bdrv_drained_begin().
+ */
+void bdrv_drained_end(BlockDriverState *bs);
+
#endif
diff --git a/include/block/block_int.h b/include/block/block_int.h
index 14ad4c3..7c58221 100644
--- a/include/block/block_int.h
+++ b/include/block/block_int.h
@@ -456,6 +456,8 @@ struct BlockDriverState {
/* threshold limit for writes, in bytes. "High water mark". */
uint64_t write_threshold_offset;
NotifierWithReturn write_threshold_notifier;
+
+ int quiesce_counter;
};
--
2.4.3
^ permalink raw reply related [flat|nested] 2+ messages in thread
end of thread, other threads:[~2015-10-18 23:36 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2015-10-18 23:22 [Qemu-devel] [PATCH v3 05/12] block: Introduce "drained begin/end" API Fam Zheng
-- strict thread matches above, loose matches on Subject: below --
2015-10-16 10:12 [Qemu-devel] [PATCH v3 00/12] block: Protect nested event loop with bdrv_drained_begin and bdrv_drained_end Fam Zheng
2015-10-16 10:12 ` [Qemu-devel] [PATCH v3 05/12] block: Introduce "drained begin/end" API Fam Zheng
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).