qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Kevin Wolf <kwolf@redhat.com>
To: qemu-block@nongnu.org
Cc: kwolf@redhat.com, stefanha@redhat.com, hreitz@redhat.com,
	eblake@redhat.com, qemu-devel@nongnu.org
Subject: [PATCH 04/12] block-backend: Fix blk_new_open() for iothreads
Date: Thu, 25 May 2023 14:47:05 +0200	[thread overview]
Message-ID: <20230525124713.401149-5-kwolf@redhat.com> (raw)
In-Reply-To: <20230525124713.401149-1-kwolf@redhat.com>

This fixes blk_new_open() to not assume that bs is in the main context.

In particular, the BlockBackend must be created with the right
AioContext because it will refuse to move to a different context
afterwards. (blk->allow_aio_context_change is false.)

Use this opportunity to use blk_insert_bs() instead of duplicating the
bdrv_root_attach_child() call. This is consistent with what
blk_new_with_bs() does. Add comments to document the locking rules.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
---
 block/block-backend.c | 27 +++++++++++++++++++++------
 1 file changed, 21 insertions(+), 6 deletions(-)

diff --git a/block/block-backend.c b/block/block-backend.c
index 1d89fabd35..dde60e0f71 100644
--- a/block/block-backend.c
+++ b/block/block-backend.c
@@ -389,6 +389,8 @@ BlockBackend *blk_new(AioContext *ctx, uint64_t perm, uint64_t shared_perm)
  * Both sets of permissions can be changed later using blk_set_perm().
  *
  * Return the new BlockBackend on success, null on failure.
+ *
+ * Callers must hold the AioContext lock of @bs.
  */
 BlockBackend *blk_new_with_bs(BlockDriverState *bs, uint64_t perm,
                               uint64_t shared_perm, Error **errp)
@@ -406,11 +408,15 @@ BlockBackend *blk_new_with_bs(BlockDriverState *bs, uint64_t perm,
 
 /*
  * Creates a new BlockBackend, opens a new BlockDriverState, and connects both.
- * The new BlockBackend is in the main AioContext.
+ * By default, the new BlockBackend is in the main AioContext, but if the
+ * parameters connect it with any existing node in a different AioContext, it
+ * may end up there instead.
  *
  * Just as with bdrv_open(), after having called this function the reference to
  * @options belongs to the block layer (even on failure).
  *
+ * Called without holding an AioContext lock.
+ *
  * TODO: Remove @filename and @flags; it should be possible to specify a whole
  * BDS tree just by specifying the @options QDict (or @reference,
  * alternatively). At the time of adding this function, this is not possible,
@@ -422,6 +428,7 @@ BlockBackend *blk_new_open(const char *filename, const char *reference,
 {
     BlockBackend *blk;
     BlockDriverState *bs;
+    AioContext *ctx;
     uint64_t perm = 0;
     uint64_t shared = BLK_PERM_ALL;
 
@@ -451,18 +458,24 @@ BlockBackend *blk_new_open(const char *filename, const char *reference,
         shared = BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE_UNCHANGED;
     }
 
-    blk = blk_new(qemu_get_aio_context(), perm, shared);
     aio_context_acquire(qemu_get_aio_context());
     bs = bdrv_open(filename, reference, options, flags, errp);
     aio_context_release(qemu_get_aio_context());
     if (!bs) {
-        blk_unref(blk);
         return NULL;
     }
 
-    blk->root = bdrv_root_attach_child(bs, "root", &child_root,
-                                       BDRV_CHILD_FILTERED | BDRV_CHILD_PRIMARY,
-                                       perm, shared, blk, errp);
+    /* bdrv_open() could have moved bs to a different AioContext */
+    ctx = bdrv_get_aio_context(bs);
+    blk = blk_new(bdrv_get_aio_context(bs), perm, shared);
+    blk->perm = perm;
+    blk->shared_perm = shared;
+
+    aio_context_acquire(ctx);
+    blk_insert_bs(blk, bs, errp);
+    bdrv_unref(bs);
+    aio_context_release(ctx);
+
     if (!blk->root) {
         blk_unref(blk);
         return NULL;
@@ -903,6 +916,8 @@ void blk_remove_bs(BlockBackend *blk)
 
 /*
  * Associates a new BlockDriverState with @blk.
+ *
+ * Callers must hold the AioContext lock of @bs.
  */
 int blk_insert_bs(BlockBackend *blk, BlockDriverState *bs, Error **errp)
 {
-- 
2.40.1



  parent reply	other threads:[~2023-05-25 12:48 UTC|newest]

Thread overview: 29+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-05-25 12:47 [PATCH 00/12] block: Fix blockdev-create with iothreads Kevin Wolf
2023-05-25 12:47 ` [PATCH 01/12] block-coroutine-wrapper: Take AioContext lock in no_co_wrappers Kevin Wolf
2023-05-25 18:15   ` Stefan Hajnoczi
2023-05-25 12:47 ` [PATCH 02/12] block: Clarify locking rules for bdrv_open(_inherit)() Kevin Wolf
2023-05-25 18:16   ` Stefan Hajnoczi
2023-05-25 12:47 ` [PATCH 03/12] block: Take main AioContext lock when calling bdrv_open() Kevin Wolf
2023-05-25 18:20   ` Stefan Hajnoczi
2023-05-26  8:27     ` Kevin Wolf
2023-05-25 12:47 ` Kevin Wolf [this message]
2023-05-25 18:37   ` [PATCH 04/12] block-backend: Fix blk_new_open() for iothreads Stefan Hajnoczi
2023-05-25 12:47 ` [PATCH 05/12] mirror: Hold main AioContext lock for calling bdrv_open_backing_file() Kevin Wolf
2023-05-25 18:40   ` Stefan Hajnoczi
2023-05-25 12:47 ` [PATCH 06/12] qcow2: Fix open with 'file' in iothread Kevin Wolf
2023-05-25 18:51   ` Stefan Hajnoczi
2023-05-26  8:40     ` Kevin Wolf
2023-05-26  9:04       ` Kevin Wolf
2023-05-25 12:47 ` [PATCH 07/12] raw-format: " Kevin Wolf
2023-05-25 18:52   ` Stefan Hajnoczi
2023-05-25 12:47 ` [PATCH 08/12] copy-before-write: Fix open with child " Kevin Wolf
2023-05-25 18:55   ` Stefan Hajnoczi
2023-05-26 14:30   ` Eric Blake
2023-05-25 12:47 ` [PATCH 09/12] block: Take AioContext lock in bdrv_open_driver() Kevin Wolf
2023-05-25 18:56   ` Stefan Hajnoczi
2023-05-25 12:47 ` [PATCH 10/12] block: Fix AioContext locking in bdrv_insert_node() Kevin Wolf
2023-05-25 18:58   ` Stefan Hajnoczi
2023-05-25 12:47 ` [PATCH 11/12] iotests: Make verify_virtio_scsi_pci_or_ccw() public Kevin Wolf
2023-05-25 18:59   ` Stefan Hajnoczi
2023-05-25 12:47 ` [PATCH 12/12] iotests: Test blockdev-create in iothread Kevin Wolf
2023-05-25 19:01   ` Stefan Hajnoczi

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=20230525124713.401149-5-kwolf@redhat.com \
    --to=kwolf@redhat.com \
    --cc=eblake@redhat.com \
    --cc=hreitz@redhat.com \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=stefanha@redhat.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).