qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Max Reitz <mreitz@redhat.com>
To: qemu-block@nongnu.org
Cc: Kevin Wolf <kwolf@redhat.com>, Alberto Garcia <berto@igalia.com>,
	John Snow <jsnow@redhat.com>,
	Markus Armbruster <armbru@redhat.com>,
	qemu-devel@nongnu.org, Stefan Hajnoczi <stefanha@redhat.com>,
	Paolo Bonzini <pbonzini@redhat.com>,
	Max Reitz <mreitz@redhat.com>
Subject: [Qemu-devel] [PATCH v7 22/24] block: Rewrite bdrv_close_all()
Date: Mon,  9 Nov 2015 23:39:31 +0100	[thread overview]
Message-ID: <1447108773-6836-23-git-send-email-mreitz@redhat.com> (raw)
In-Reply-To: <1447108773-6836-1-git-send-email-mreitz@redhat.com>

This patch rewrites bdrv_close_all(): Until now, all root BDSs have been
force-closed. This is bad because it can lead to cached data not being
flushed to disk.

Instead, try to make all reference holders relinquish their reference
voluntarily:

1. All BlockBackend users are handled by making all BBs simply eject
   their BDS tree. Since a BDS can never be on top of a BB, this will
   not cause any of the issues as seen with the force-closing of BDSs.
   The references will be relinquished and any further access to the BB
   will fail gracefully.
2. All BDSs which are owned by the monitor itself (because they do not
   have a BB) are relinquished next.
3. Besides BBs and the monitor, block jobs and other BDSs are the only
   things left that can hold a reference to BDSs. After every remaining
   block job has been canceled, there should not be any BDSs left (and
   the loop added here will always terminate (as long as NDEBUG is not
   defined), because either all_bdrv_states will be empty or there will
   not be any block job left to cancel, failing the assertion).

Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
---
 block.c | 45 +++++++++++++++++++++++++++++++++++++--------
 1 file changed, 37 insertions(+), 8 deletions(-)

diff --git a/block.c b/block.c
index b935dff..e3d5aea 100644
--- a/block.c
+++ b/block.c
@@ -1905,9 +1905,7 @@ static void bdrv_close(BlockDriverState *bs)
 {
     BdrvAioNotifier *ban, *ban_next;
 
-    if (bs->job) {
-        block_job_cancel_sync(bs->job);
-    }
+    assert(!bs->job);
 
     /* Disable I/O limits and drain all pending throttled requests */
     if (bs->throttle_state) {
@@ -1971,13 +1969,44 @@ static void bdrv_close(BlockDriverState *bs)
 void bdrv_close_all(void)
 {
     BlockDriverState *bs;
+    AioContext *aio_context;
+    int original_refcount = 0;
 
-    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
-        AioContext *aio_context = bdrv_get_aio_context(bs);
+    /* Drop references from requests still in flight, such as canceled block
+     * jobs whose AIO context has not been polled yet */
+    bdrv_drain_all();
 
-        aio_context_acquire(aio_context);
-        bdrv_close(bs);
-        aio_context_release(aio_context);
+    blockdev_close_all_bdrv_states();
+    blk_remove_all_bs();
+
+    /* Cancel all block jobs */
+    while (!QTAILQ_EMPTY(&all_bdrv_states)) {
+        QTAILQ_FOREACH(bs, &all_bdrv_states, bs_list) {
+            aio_context = bdrv_get_aio_context(bs);
+
+            aio_context_acquire(aio_context);
+            if (bs->job) {
+                /* So we can safely query the current refcount */
+                bdrv_ref(bs);
+                original_refcount = bs->refcnt;
+
+                block_job_cancel_sync(bs->job);
+                aio_context_release(aio_context);
+                break;
+            }
+            aio_context_release(aio_context);
+        }
+
+        /* All the remaining BlockDriverStates are referenced directly or
+         * indirectly from block jobs, so there needs to be at least one BDS
+         * directly used by a block job */
+        assert(bs);
+
+        /* Wait for the block job to release its reference */
+        while (bs->refcnt >= original_refcount) {
+            aio_poll(aio_context, true);
+        }
+        bdrv_unref(bs);
     }
 }
 
-- 
2.6.2

  parent reply	other threads:[~2015-11-09 22:40 UTC|newest]

Thread overview: 71+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-11-09 22:39 [Qemu-devel] [PATCH v7 for-2.6 00/24] block: Rework bdrv_close_all() Max Reitz
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 01/24] blockdev: Add missing bdrv_unref() in drive-backup Max Reitz
2015-11-12  6:14   ` Fam Zheng
2015-11-18 15:24   ` Kevin Wolf
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 02/24] blockjob: Call bdrv_unref() on creation error Max Reitz
2015-11-12  6:16   ` Fam Zheng
2015-11-18 15:24   ` Kevin Wolf
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 03/24] block: Release dirty bitmaps in bdrv_close() Max Reitz
2015-11-11 21:08   ` John Snow
2015-11-12  6:23   ` Fam Zheng
2015-11-13 22:49     ` John Snow
2015-11-16  1:27       ` Fam Zheng
2015-11-16 17:07         ` John Snow
2015-11-17  4:22           ` Fam Zheng
2015-11-17 17:05             ` [Qemu-devel] Closing Bitmaps (Was: Re: [PATCH v7 03/24] block: Release dirty bitmaps in bdrv_close()) John Snow
2015-11-18  2:29               ` Fam Zheng
2015-11-18 15:47                 ` John Snow
2015-11-18 15:03               ` Kevin Wolf
2015-11-18 15:49                 ` John Snow
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 04/24] iotests: Rename filter_nbd to _filter_nbd in 083 Max Reitz
2015-11-12  6:25   ` Fam Zheng
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 05/24] iotests: Change coding style of " Max Reitz
2015-11-12  6:25   ` Fam Zheng
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 06/24] iotests: Move _filter_nbd into common.filter Max Reitz
2015-11-12  6:26   ` Fam Zheng
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 07/24] iotests: Make _filter_nbd drop log lines Max Reitz
2015-11-12  6:27   ` Fam Zheng
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 08/24] iotests: Make _filter_nbd support more URL types Max Reitz
2015-11-12  6:28   ` Fam Zheng
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 09/24] iotests: Make redirecting qemu's stderr optional Max Reitz
2015-11-12  6:31   ` Fam Zheng
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 10/24] iotests: Add test for eject under NBD server Max Reitz
2015-11-11 21:46   ` John Snow
2015-11-12  6:37   ` Fam Zheng
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 11/24] block: Add BB-BDS remove/insert notifiers Max Reitz
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 12/24] virtio-blk: Functions for op blocker management Max Reitz
2015-11-25 15:57   ` Kevin Wolf
2015-11-25 16:03     ` Max Reitz
2015-11-25 16:18       ` Kevin Wolf
2015-11-25 16:26         ` Max Reitz
2015-11-26  7:48           ` Stefan Hajnoczi
2015-11-26 10:43             ` Kevin Wolf
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 13/24] virtio-scsi: Catch BDS-BB removal/insertion Max Reitz
2015-11-25 16:03   ` Kevin Wolf
2015-11-25 16:08     ` Max Reitz
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 14/24] nbd: Switch from close to eject notifier Max Reitz
2015-11-30 15:36   ` Kevin Wolf
2015-11-30 17:22     ` Max Reitz
2015-12-01 13:16       ` Kevin Wolf
2015-12-02 15:51         ` Max Reitz
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 15/24] block: Remove BDS close notifier Max Reitz
2015-11-30 15:38   ` Kevin Wolf
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 16/24] block: Use blk_remove_bs() in blk_delete() Max Reitz
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 17/24] blockdev: Use blk_remove_bs() in do_drive_del() Max Reitz
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 18/24] block: Make bdrv_close() static Max Reitz
2015-11-12  7:07   ` Fam Zheng
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 19/24] block: Add list of all BlockDriverStates Max Reitz
2015-11-12  7:12   ` Fam Zheng
2015-11-16 16:03     ` Max Reitz
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 20/24] blockdev: Keep track of monitor-owned BDS Max Reitz
2015-11-10  1:25   ` Max Reitz
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 21/24] block: Add blk_remove_all_bs() Max Reitz
2015-11-09 22:39 ` Max Reitz [this message]
2015-11-12  7:34   ` [Qemu-devel] [PATCH v7 22/24] block: Rewrite bdrv_close_all() Fam Zheng
2015-11-16 16:15     ` Max Reitz
2015-11-18  2:52       ` Fam Zheng
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 23/24] iotests: Add test for multiple BB on BDS tree Max Reitz
2015-11-09 22:39 ` [Qemu-devel] [PATCH v7 24/24] iotests: Add test for block jobs and BDS ejection Max Reitz
2015-11-30 16:23   ` Kevin Wolf
2015-11-30 17:44     ` Max Reitz
2015-11-25 16:09 ` [Qemu-devel] [PATCH v7 for-2.6 00/24] block: Rework bdrv_close_all() 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=1447108773-6836-23-git-send-email-mreitz@redhat.com \
    --to=mreitz@redhat.com \
    --cc=armbru@redhat.com \
    --cc=berto@igalia.com \
    --cc=jsnow@redhat.com \
    --cc=kwolf@redhat.com \
    --cc=pbonzini@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).