[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH v2 04/10] block.c: bdrv_replace_child_noperm: first remove th
From: |
Emanuele Giuseppe Esposito |
Subject: |
Re: [PATCH v2 04/10] block.c: bdrv_replace_child_noperm: first remove the child, and then call ->detach() |
Date: |
Wed, 16 Mar 2022 10:13:45 +0100 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101 Thunderbird/91.2.0 |
Unfortunately this patch is not safe: theoretically ->detach can call
bdrv_unapply_subtree_drain, and if it polls, will can call a bh that
for example reads the graph, finding it in an inconsistent state, since
it is between the two writes QLIST_REMOVE(child, next_parent); and
QLIST_REMOVE(child, next);
Please ignore it.
This patch could eventually go in the subtree_drain serie, if we decide
to go in that direction.
Emanuele
Am 14/03/2022 um 14:18 schrieb Emanuele Giuseppe Esposito:
> Doing the opposite can make ->detach() (more precisely
> bdrv_unapply_subtree_drain() in bdrv_child_cb_detach) undo the subtree_drain
> just performed to protect the removal of the child from the graph,
> thus making the fully-enabled assert_bdrv_graph_writable fail.
>
> Note that assert_bdrv_graph_writable is not yet fully enabled.
>
> Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
> ---
> block.c | 19 ++++++++++++++-----
> 1 file changed, 14 insertions(+), 5 deletions(-)
>
> diff --git a/block.c b/block.c
> index 372f16f4a0..d870ba5393 100644
> --- a/block.c
> +++ b/block.c
> @@ -1448,6 +1448,11 @@ static void bdrv_child_cb_attach(BdrvChild *child)
> bdrv_apply_subtree_drain(child, bs);
> }
>
> +/*
> + * Unapply the drain in the whole child subtree, as
> + * it has been already detached, and then remove from
> + * child->opaque->children.
> + */
> static void bdrv_child_cb_detach(BdrvChild *child)
> {
> BlockDriverState *bs = child->opaque;
> @@ -2864,14 +2869,18 @@ static void bdrv_replace_child_noperm(BdrvChild
> **childp,
> }
>
> if (old_bs) {
> - /* Detach first so that the recursive drain sections coming from
> @child
> - * are already gone and we only end the drain sections that came from
> - * elsewhere. */
> + /* First remove child from child->bs->parents list */
> + assert_bdrv_graph_writable(old_bs);
> + QLIST_REMOVE(child, next_parent);
> + /*
> + * Then call ->detach() cb.
> + * In child_of_bds case, update the child parent
> + * (child->opaque) ->children list and
> + * remove any drain left in the child subtree.
> + */
> if (child->klass->detach) {
> child->klass->detach(child);
> }
> - assert_bdrv_graph_writable(old_bs);
> - QLIST_REMOVE(child, next_parent);
> }
>
> child->bs = new_bs;
>
- [PATCH v2 07/10] test-bdrv-drain.c: remove test_detach_by_parent_cb(), (continued)
- [PATCH v2 07/10] test-bdrv-drain.c: remove test_detach_by_parent_cb(), Emanuele Giuseppe Esposito, 2022/03/14
- [PATCH v2 06/10] test-bdrv-drain.c: adapt test to support additional subtree drains, Emanuele Giuseppe Esposito, 2022/03/14
- [PATCH v2 05/10] block.c: bdrv_replace_child_noperm: first call ->attach(), and then add child, Emanuele Giuseppe Esposito, 2022/03/14
- [PATCH v2 01/10] drains: create bh only when polling, Emanuele Giuseppe Esposito, 2022/03/14
- [PATCH v2 09/10] child_job_drained_poll: override polling condition only when in home thread, Emanuele Giuseppe Esposito, 2022/03/14
- [PATCH v2 10/10] tests/qemu-iotests/030: test_stream_parallel should use auto_finalize=False, Emanuele Giuseppe Esposito, 2022/03/14
- [PATCH v2 02/10] bdrv_parent_drained_begin_single: handle calls from coroutine context, Emanuele Giuseppe Esposito, 2022/03/14
- [PATCH v2 08/10] tests/unit/test-bdrv-drain.c: graph setup functions can't run in coroutines, Emanuele Giuseppe Esposito, 2022/03/14
- [PATCH v2 04/10] block.c: bdrv_replace_child_noperm: first remove the child, and then call ->detach(), Emanuele Giuseppe Esposito, 2022/03/14
- Re: [PATCH v2 04/10] block.c: bdrv_replace_child_noperm: first remove the child, and then call ->detach(),
Emanuele Giuseppe Esposito <=