From: Kevin Wolf <kwolf@redhat.com>
To: qemu-block@nongnu.org
Cc: kwolf@redhat.com, peter.maydell@linaro.org, qemu-devel@nongnu.org
Subject: [Qemu-devel] [PULL 01/24] block: Poll after drain on attaching a node
Date: Tue, 10 Jul 2018 17:42:41 +0200 [thread overview]
Message-ID: <20180710154304.18304-2-kwolf@redhat.com> (raw)
In-Reply-To: <20180710154304.18304-1-kwolf@redhat.com>
Commit dcf94a23b1 ('block: Don't poll in parent drain callbacks')
removed polling in bdrv_child_cb_drained_begin() on the grounds that the
original bdrv_drain() already will poll and BdrvChildRole.drained_begin
calls must not cause graph changes (and therefore must not call
aio_poll() or the recursion through the graph will break.
This reasoning is correct for calls through bdrv_do_drained_begin().
However, BdrvChildRole.drained_begin is also called when a node that is
already in a drained section (i.e. bdrv_do_drained_begin() has already
returned and therefore can't poll any more) is attached to a new parent.
In this case, we must explicitly poll to have all requests completed
before the drained new child can be attached to the parent.
In bdrv_replace_child_noperm(), we know that we're not inside the
recursion of bdrv_do_drained_begin() because graph changes are not
allowed there, and bdrv_replace_child_noperm() is a graph change. The
call of BdrvChildRole.drained_begin() must therefore be followed by a
BDRV_POLL_WHILE() that waits for the completion of requests.
Reported-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
---
include/block/block.h | 8 ++++++++
include/block/block_int.h | 3 +++
block.c | 2 +-
block/io.c | 26 ++++++++++++++++++++------
4 files changed, 32 insertions(+), 7 deletions(-)
diff --git a/include/block/block.h b/include/block/block.h
index bc76b1e59f..706ef009ad 100644
--- a/include/block/block.h
+++ b/include/block/block.h
@@ -569,6 +569,14 @@ void bdrv_parent_drained_begin(BlockDriverState *bs, BdrvChild *ignore,
bool ignore_bds_parents);
/**
+ * bdrv_parent_drained_begin_single:
+ *
+ * Begin a quiesced section for the parent of @c. If @poll is true, wait for
+ * any pending activity to cease.
+ */
+void bdrv_parent_drained_begin_single(BdrvChild *c, bool poll);
+
+/**
* bdrv_parent_drained_end:
*
* End a quiesced section of all users of @bs. This is part of
diff --git a/include/block/block_int.h b/include/block/block_int.h
index af71b414be..81cd3db7a9 100644
--- a/include/block/block_int.h
+++ b/include/block/block_int.h
@@ -606,6 +606,9 @@ struct BdrvChildRole {
* requests after returning from .drained_begin() until .drained_end() is
* called.
*
+ * These functions must not change the graph (and therefore also must not
+ * call aio_poll(), which could change the graph indirectly).
+ *
* Note that this can be nested. If drained_begin() was called twice, new
* I/O is allowed only after drained_end() was called twice, too.
*/
diff --git a/block.c b/block.c
index ac8b3a3511..a2fe05ea96 100644
--- a/block.c
+++ b/block.c
@@ -2066,7 +2066,7 @@ static void bdrv_replace_child_noperm(BdrvChild *child,
}
assert(num >= 0);
for (i = 0; i < num; i++) {
- child->role->drained_begin(child);
+ bdrv_parent_drained_begin_single(child, true);
}
}
diff --git a/block/io.c b/block/io.c
index 1a2272fad3..038449f81f 100644
--- a/block/io.c
+++ b/block/io.c
@@ -52,9 +52,7 @@ void bdrv_parent_drained_begin(BlockDriverState *bs, BdrvChild *ignore,
if (c == ignore || (ignore_bds_parents && c->role->parent_is_bds)) {
continue;
}
- if (c->role->drained_begin) {
- c->role->drained_begin(c);
- }
+ bdrv_parent_drained_begin_single(c, false);
}
}
@@ -73,6 +71,14 @@ void bdrv_parent_drained_end(BlockDriverState *bs, BdrvChild *ignore,
}
}
+static bool bdrv_parent_drained_poll_single(BdrvChild *c)
+{
+ if (c->role->drained_poll) {
+ return c->role->drained_poll(c);
+ }
+ return false;
+}
+
static bool bdrv_parent_drained_poll(BlockDriverState *bs, BdrvChild *ignore,
bool ignore_bds_parents)
{
@@ -83,14 +89,22 @@ static bool bdrv_parent_drained_poll(BlockDriverState *bs, BdrvChild *ignore,
if (c == ignore || (ignore_bds_parents && c->role->parent_is_bds)) {
continue;
}
- if (c->role->drained_poll) {
- busy |= c->role->drained_poll(c);
- }
+ busy |= bdrv_parent_drained_poll_single(c);
}
return busy;
}
+void bdrv_parent_drained_begin_single(BdrvChild *c, bool poll)
+{
+ if (c->role->drained_begin) {
+ c->role->drained_begin(c);
+ }
+ if (poll) {
+ BDRV_POLL_WHILE(c->bs, bdrv_parent_drained_poll_single(c));
+ }
+}
+
static void bdrv_merge_limits(BlockLimits *dst, const BlockLimits *src)
{
dst->opt_transfer = MAX(dst->opt_transfer, src->opt_transfer);
--
2.13.6
next prev parent reply other threads:[~2018-07-10 15:43 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-07-10 15:42 [Qemu-devel] [PULL 00/24] Block layer patches Kevin Wolf
2018-07-10 15:42 ` Kevin Wolf [this message]
2018-07-10 15:42 ` [Qemu-devel] [PULL 02/24] test-bdrv-drain: Test bdrv_append() to drained node Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 03/24] block: Fix copy-on-read crash with partial final cluster Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 04/24] iotests: 222: Don't run with luks Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 05/24] block/io: fix copy_range Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 06/24] block: split flags in copy_range Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 07/24] block: add BDRV_REQ_SERIALISING flag Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 08/24] block/backup: fix fleecing scheme: use serialized writes Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 09/24] qapi/block-core.json: Add missing documentation for blklogwrites log-append option Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 10/24] block/blklogwrites: Make sure the log sector size is not too small Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 11/24] Revert "block: Remove dead deprecation warning code" Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 12/24] Revert "block: Remove deprecated -drive option serial" Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 13/24] Revert "block: Remove deprecated -drive option addr" Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 14/24] Revert "block: Remove deprecated -drive geometry options" Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 15/24] block: Prefix file driver trace points with "file_" Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 16/24] block: Add copy offloading trace points Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 17/24] block: Use BdrvChild to discard Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 18/24] block: Use uint64_t for BdrvTrackedRequest byte fields Kevin Wolf
2018-07-10 15:42 ` [Qemu-devel] [PULL 19/24] block: Extract common write req handling Kevin Wolf
2018-07-10 15:43 ` [Qemu-devel] [PULL 20/24] block: Fix handling of image enlarging write Kevin Wolf
2018-07-10 15:43 ` [Qemu-devel] [PULL 21/24] block: Use common req handling for discard Kevin Wolf
2018-07-10 15:43 ` [Qemu-devel] [PULL 22/24] block: Use common req handling in copy offloading Kevin Wolf
2018-07-10 15:43 ` [Qemu-devel] [PULL 23/24] block: Fix bdrv_co_truncate overlap check Kevin Wolf
2018-07-10 15:43 ` [Qemu-devel] [PULL 24/24] block: Use common write req handling in truncate Kevin Wolf
2018-07-10 17:19 ` [Qemu-devel] [PULL 00/24] Block layer patches Peter Maydell
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180710154304.18304-2-kwolf@redhat.com \
--to=kwolf@redhat.com \
--cc=peter.maydell@linaro.org \
--cc=qemu-block@nongnu.org \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.