qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Stefan Hajnoczi <stefanha@redhat.com>
To: qemu-devel@nongnu.org
Cc: "Coiby Xu" <Coiby.Xu@gmail.com>,
	"Marcel Apfelbaum" <marcel.apfelbaum@gmail.com>,
	"Michael S. Tsirkin" <mst@redhat.com>,
	"Richard W.M. Jones" <rjones@redhat.com>,
	"Peter Xu" <peterx@redhat.com>,
	xen-devel@lists.xenproject.org, "Kevin Wolf" <kwolf@redhat.com>,
	"Ronnie Sahlberg" <ronniesahlberg@gmail.com>,
	"Stefano Stabellini" <sstabellini@kernel.org>,
	"Richard Henderson" <richard.henderson@linaro.org>,
	"Stefan Hajnoczi" <stefanha@redhat.com>,
	"Julia Suvorova" <jusual@redhat.com>,
	"Hanna Reitz" <hreitz@redhat.com>,
	"Leonardo Bras" <leobras@redhat.com>,
	eesposit@redhat.com, "Fam Zheng" <fam@euphon.net>,
	"Aarushi Mehta" <mehta.aaru20@gmail.com>,
	"David Woodhouse" <dwmw2@infradead.org>,
	"Xie Yongji" <xieyongji@bytedance.com>,
	"Stefano Garzarella" <sgarzare@redhat.com>,
	qemu-block@nongnu.org, "Eduardo Habkost" <eduardo@habkost.net>,
	"Paul Durrant" <paul@xen.org>, "Stefan Weil" <sw@weilnetz.de>,
	"Anthony Perard" <anthony.perard@citrix.com>,
	"Daniel P. Berrangé" <berrange@redhat.com>,
	"Peter Lieven" <pl@kamp.de>,
	"Paolo Bonzini" <pbonzini@redhat.com>,
	"Philippe Mathieu-Daudé" <philmd@linaro.org>,
	"Juan Quintela" <quintela@redhat.com>
Subject: [PATCH v6 01/20] block-backend: split blk_do_set_aio_context()
Date: Tue, 16 May 2023 15:02:19 -0400	[thread overview]
Message-ID: <20230516190238.8401-2-stefanha@redhat.com> (raw)
In-Reply-To: <20230516190238.8401-1-stefanha@redhat.com>

blk_set_aio_context() is not fully transactional because
blk_do_set_aio_context() updates blk->ctx outside the transaction. Most
of the time this goes unnoticed but a BlockDevOps.drained_end() callback
that invokes blk_get_aio_context() fails assert(ctx == blk->ctx). This
happens because blk->ctx is only assigned after
BlockDevOps.drained_end() is called and we're in an intermediate state
where BlockDrvierState nodes already have the new context and the
BlockBackend still has the old context.

Making blk_set_aio_context() fully transactional solves this assertion
failure because the BlockBackend's context is updated as part of the
transaction (before BlockDevOps.drained_end() is called).

Split blk_do_set_aio_context() in order to solve this assertion failure.
This helper function actually serves two different purposes:
1. It drives blk_set_aio_context().
2. It responds to BdrvChildClass->change_aio_ctx().

Get rid of the helper function. Do #1 inside blk_set_aio_context() and
do #2 inside blk_root_set_aio_ctx_commit(). This simplifies the code.

The only drawback of the fully transactional approach is that
blk_set_aio_context() must contend with blk_root_set_aio_ctx_commit()
being invoked as part of the AioContext change propagation. This can be
solved by temporarily setting blk->allow_aio_context_change to true.

Future patches call blk_get_aio_context() from
BlockDevOps->drained_end(), so this patch will become necessary.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
---
 block/block-backend.c | 71 +++++++++++++++++--------------------------
 1 file changed, 28 insertions(+), 43 deletions(-)

diff --git a/block/block-backend.c b/block/block-backend.c
index ca537cd0ad..68087437ac 100644
--- a/block/block-backend.c
+++ b/block/block-backend.c
@@ -2411,52 +2411,31 @@ static AioContext *blk_aiocb_get_aio_context(BlockAIOCB *acb)
     return blk_get_aio_context(blk_acb->blk);
 }
 
-static int blk_do_set_aio_context(BlockBackend *blk, AioContext *new_context,
-                                  bool update_root_node, Error **errp)
-{
-    BlockDriverState *bs = blk_bs(blk);
-    ThrottleGroupMember *tgm = &blk->public.throttle_group_member;
-    int ret;
-
-    if (bs) {
-        bdrv_ref(bs);
-
-        if (update_root_node) {
-            /*
-             * update_root_node MUST be false for blk_root_set_aio_ctx_commit(),
-             * as we are already in the commit function of a transaction.
-             */
-            ret = bdrv_try_change_aio_context(bs, new_context, blk->root, errp);
-            if (ret < 0) {
-                bdrv_unref(bs);
-                return ret;
-            }
-        }
-        /*
-         * Make blk->ctx consistent with the root node before we invoke any
-         * other operations like drain that might inquire blk->ctx
-         */
-        blk->ctx = new_context;
-        if (tgm->throttle_state) {
-            bdrv_drained_begin(bs);
-            throttle_group_detach_aio_context(tgm);
-            throttle_group_attach_aio_context(tgm, new_context);
-            bdrv_drained_end(bs);
-        }
-
-        bdrv_unref(bs);
-    } else {
-        blk->ctx = new_context;
-    }
-
-    return 0;
-}
-
 int blk_set_aio_context(BlockBackend *blk, AioContext *new_context,
                         Error **errp)
 {
+    bool old_allow_change;
+    BlockDriverState *bs = blk_bs(blk);
+    int ret;
+
     GLOBAL_STATE_CODE();
-    return blk_do_set_aio_context(blk, new_context, true, errp);
+
+    if (!bs) {
+        blk->ctx = new_context;
+        return 0;
+    }
+
+    bdrv_ref(bs);
+
+    old_allow_change = blk->allow_aio_context_change;
+    blk->allow_aio_context_change = true;
+
+    ret = bdrv_try_change_aio_context(bs, new_context, NULL, errp);
+
+    blk->allow_aio_context_change = old_allow_change;
+
+    bdrv_unref(bs);
+    return ret;
 }
 
 typedef struct BdrvStateBlkRootContext {
@@ -2468,8 +2447,14 @@ static void blk_root_set_aio_ctx_commit(void *opaque)
 {
     BdrvStateBlkRootContext *s = opaque;
     BlockBackend *blk = s->blk;
+    AioContext *new_context = s->new_ctx;
+    ThrottleGroupMember *tgm = &blk->public.throttle_group_member;
 
-    blk_do_set_aio_context(blk, s->new_ctx, false, &error_abort);
+    blk->ctx = new_context;
+    if (tgm->throttle_state) {
+        throttle_group_detach_aio_context(tgm);
+        throttle_group_attach_aio_context(tgm, new_context);
+    }
 }
 
 static TransactionActionDrv set_blk_root_context = {
-- 
2.40.1



  reply	other threads:[~2023-05-16 19:05 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-05-16 19:02 [PATCH v6 00/20] block: remove aio_disable_external() API Stefan Hajnoczi
2023-05-16 19:02 ` Stefan Hajnoczi [this message]
2023-05-16 19:02 ` [PATCH v6 02/20] hw/qdev: introduce qdev_is_realized() helper Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 03/20] virtio-scsi: avoid race between unplug and transport event Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 04/20] virtio-scsi: stop using aio_disable_external() during unplug Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 05/20] util/vhost-user-server: rename refcount to in_flight counter Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 06/20] block/export: wait for vhost-user-blk requests when draining Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 07/20] block/export: stop using is_external in vhost-user-blk server Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 08/20] hw/xen: do not use aio_set_fd_handler(is_external=true) in xen_xenstore Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 09/20] block: add blk_in_drain() API Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 10/20] block: drain from main loop thread in bdrv_co_yield_to_drain() Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 11/20] xen-block: implement BlockDevOps->drained_begin() Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 12/20] hw/xen: do not set is_external=true on evtchn fds Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 13/20] block/export: rewrite vduse-blk drain code Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 14/20] block/export: don't require AioContext lock around blk_exp_ref/unref() Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 15/20] block/fuse: do not set is_external=true on FUSE fd Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 16/20] virtio: make it possible to detach host notifier from any thread Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 17/20] virtio-blk: implement BlockDevOps->drained_begin() Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 18/20] virtio-scsi: " Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 19/20] virtio: do not set is_external=true on host notifiers Stefan Hajnoczi
2023-05-16 19:02 ` [PATCH v6 20/20] aio: remove aio_disable_external() API Stefan Hajnoczi
2023-05-30 16:24 ` [PATCH v6 00/20] block: " Kevin Wolf

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=20230516190238.8401-2-stefanha@redhat.com \
    --to=stefanha@redhat.com \
    --cc=Coiby.Xu@gmail.com \
    --cc=anthony.perard@citrix.com \
    --cc=berrange@redhat.com \
    --cc=dwmw2@infradead.org \
    --cc=eduardo@habkost.net \
    --cc=eesposit@redhat.com \
    --cc=fam@euphon.net \
    --cc=hreitz@redhat.com \
    --cc=jusual@redhat.com \
    --cc=kwolf@redhat.com \
    --cc=leobras@redhat.com \
    --cc=marcel.apfelbaum@gmail.com \
    --cc=mehta.aaru20@gmail.com \
    --cc=mst@redhat.com \
    --cc=paul@xen.org \
    --cc=pbonzini@redhat.com \
    --cc=peterx@redhat.com \
    --cc=philmd@linaro.org \
    --cc=pl@kamp.de \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=quintela@redhat.com \
    --cc=richard.henderson@linaro.org \
    --cc=rjones@redhat.com \
    --cc=ronniesahlberg@gmail.com \
    --cc=sgarzare@redhat.com \
    --cc=sstabellini@kernel.org \
    --cc=sw@weilnetz.de \
    --cc=xen-devel@lists.xenproject.org \
    --cc=xieyongji@bytedance.com \
    /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 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).