From: Peter Maydell <peter.maydell@linaro.org>
To: Kevin Wolf <kwolf@redhat.com>
Cc: QEMU Developers <qemu-devel@nongnu.org>,
Qemu-block <qemu-block@nongnu.org>
Subject: Re: [PULL 30/39] block: bdrv_reopen_multiple: refresh permissions on updated graph
Date: Fri, 30 Apr 2021 23:38:42 +0100 [thread overview]
Message-ID: <CAFEAcA_u4X+Cq6b0UtPeLXF7-7ugiMmOeb1u3UMMqTP-W0-MQA@mail.gmail.com> (raw)
In-Reply-To: <20210430105147.125840-31-kwolf@redhat.com>
On Fri, 30 Apr 2021 at 11:53, Kevin Wolf <kwolf@redhat.com> wrote:
>
> From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
>
> Move bdrv_reopen_multiple to new paradigm of permission update:
> first update graph relations, then do refresh the permissions.
>
> We have to modify reopen process in file-posix driver: with new scheme
> we don't have prepared permissions in raw_reopen_prepare(), so we
> should reconfigure fd in raw_check_perm(). Still this seems more native
> and simple anyway.
>
> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
> Reviewed-by: Kevin Wolf <kwolf@redhat.com>
> Message-Id: <20210428151804.439460-31-vsementsov@virtuozzo.com>
> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Hi; Coverity thinks this change introduced a resource leak
(CID 1452772):
> @@ -4271,6 +4270,9 @@ int bdrv_reopen_multiple(BlockReopenQueue *bs_queue, Error **errp)
> {
> int ret = -1;
> BlockReopenQueueEntry *bs_entry, *next;
> + Transaction *tran = tran_new();
> + g_autoptr(GHashTable) found = NULL;
> + g_autoptr(GSList) refresh_list = NULL;
Now we allocate a new Transaction at the start of the function...
>
> assert(bs_queue != NULL);
>
...but in the code between these two hunks there is this:
QTAILQ_FOREACH(bs_entry, bs_queue, entry) {
ret = bdrv_flush(bs_entry->state.bs);
if (ret < 0) {
error_setg_errno(errp, -ret, "Error flushing drive");
goto cleanup;
}
}
which jumps to 'cleanup' on failure...
> - if (ret == 0) {
> - QTAILQ_FOREACH_REVERSE(bs_entry, bs_queue, entry) {
> - BlockDriverState *bs = bs_entry->state.bs;
> + ret = 0;
> + goto cleanup;
>
> - if (bs->drv->bdrv_reopen_commit_post)
> - bs->drv->bdrv_reopen_commit_post(&bs_entry->state);
> +abort:
> + tran_abort(tran);
> + QTAILQ_FOREACH_SAFE(bs_entry, bs_queue, entry, next) {
> + if (bs_entry->prepared) {
> + bdrv_reopen_abort(&bs_entry->state);
> }
> + qobject_unref(bs_entry->state.explicit_options);
> + qobject_unref(bs_entry->state.options);
> }
> +
> cleanup:
> QTAILQ_FOREACH_SAFE(bs_entry, bs_queue, entry, next) {
> - if (ret) {
> - if (bs_entry->prepared) {
> - bdrv_reopen_abort(&bs_entry->state);
> - }
> - qobject_unref(bs_entry->state.explicit_options);
> - qobject_unref(bs_entry->state.options);
> - }
> - if (bs_entry->state.new_backing_bs) {
> - bdrv_unref(bs_entry->state.new_backing_bs);
> - }
> g_free(bs_entry);
> }
> g_free(bs_queue);
...and the 'cleanup' label doesn't free the Transaction.
An easy fix would be to move the call to tran_new() down to
below the loop that calls bdrv_flush().
thanks
-- PMM
next prev parent reply other threads:[~2021-04-30 22:40 UTC|newest]
Thread overview: 43+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-04-30 10:51 [PULL 00/39] Block layer patches Kevin Wolf
2021-04-30 10:51 ` [PULL 01/39] tests/test-bdrv-graph-mod: add test_parallel_exclusive_write Kevin Wolf
2021-04-30 10:51 ` [PULL 02/39] tests/test-bdrv-graph-mod: add test_parallel_perm_update Kevin Wolf
2021-04-30 10:51 ` [PULL 03/39] tests/test-bdrv-graph-mod: add test_append_greedy_filter Kevin Wolf
2021-04-30 10:51 ` [PULL 04/39] block: bdrv_append(): don't consume reference Kevin Wolf
2021-04-30 10:51 ` [PULL 05/39] block: BdrvChildClass: add .get_parent_aio_context handler Kevin Wolf
2021-04-30 10:51 ` [PULL 06/39] block: drop ctx argument from bdrv_root_attach_child Kevin Wolf
2021-04-30 10:51 ` [PULL 07/39] block: make bdrv_reopen_{prepare,commit,abort} private Kevin Wolf
2021-04-30 10:51 ` [PULL 08/39] util: add transactions.c Kevin Wolf
2021-04-30 10:51 ` [PULL 09/39] block: bdrv_refresh_perms: check for parents permissions conflict Kevin Wolf
2021-04-30 10:51 ` [PULL 10/39] block: refactor bdrv_child* permission functions Kevin Wolf
2021-04-30 10:51 ` [PULL 11/39] block: rewrite bdrv_child_try_set_perm() using bdrv_refresh_perms() Kevin Wolf
2021-04-30 10:51 ` [PULL 12/39] block: inline bdrv_child_*() permission functions calls Kevin Wolf
2021-04-30 10:51 ` [PULL 13/39] block: use topological sort for permission update Kevin Wolf
2021-04-30 10:51 ` [PULL 14/39] block: add bdrv_drv_set_perm transaction action Kevin Wolf
2021-04-30 10:51 ` [PULL 15/39] block: add bdrv_list_* permission update functions Kevin Wolf
2021-04-30 10:51 ` [PULL 16/39] block: add bdrv_replace_child_safe() transaction action Kevin Wolf
2021-04-30 10:51 ` [PULL 17/39] block: fix bdrv_replace_node_common Kevin Wolf
2021-04-30 10:51 ` [PULL 18/39] block: add bdrv_attach_child_common() transaction action Kevin Wolf
2021-04-30 22:33 ` Peter Maydell
2021-04-30 10:51 ` [PULL 19/39] block: add bdrv_attach_child_noperm() " Kevin Wolf
2021-04-30 10:51 ` [PULL 20/39] block: split out bdrv_replace_node_noperm() Kevin Wolf
2021-04-30 10:51 ` [PULL 21/39] block: adapt bdrv_append() for inserting filters Kevin Wolf
2021-04-30 10:51 ` [PULL 22/39] block: add bdrv_remove_filter_or_cow transaction action Kevin Wolf
2021-04-30 10:51 ` [PULL 23/39] block: introduce bdrv_drop_filter() Kevin Wolf
2021-04-30 10:51 ` [PULL 24/39] block/backup-top: drop .active Kevin Wolf
2021-04-30 10:51 ` [PULL 25/39] block: drop ignore_children for permission update functions Kevin Wolf
2021-04-30 10:51 ` [PULL 26/39] block: make bdrv_unset_inherits_from to be a transaction action Kevin Wolf
2021-04-30 10:51 ` [PULL 27/39] block: make bdrv_refresh_limits() " Kevin Wolf
2021-04-30 10:51 ` [PULL 28/39] block: add bdrv_set_backing_noperm() " Kevin Wolf
2021-04-30 10:51 ` [PULL 29/39] block: bdrv_reopen_multiple(): move bdrv_flush to separate pre-prepare Kevin Wolf
2021-04-30 10:51 ` [PULL 30/39] block: bdrv_reopen_multiple: refresh permissions on updated graph Kevin Wolf
2021-04-30 22:38 ` Peter Maydell [this message]
2021-04-30 10:51 ` [PULL 31/39] block: drop unused permission update functions Kevin Wolf
2021-04-30 10:51 ` [PULL 32/39] block: inline bdrv_check_perm_common() Kevin Wolf
2021-04-30 10:51 ` [PULL 33/39] block: inline bdrv_replace_child() Kevin Wolf
2021-04-30 10:51 ` [PULL 34/39] block: refactor bdrv_child_set_perm_safe() transaction action Kevin Wolf
2021-04-30 10:51 ` [PULL 35/39] block: rename bdrv_replace_child_safe() to bdrv_replace_child() Kevin Wolf
2021-04-30 10:51 ` [PULL 36/39] block: refactor bdrv_node_check_perm() Kevin Wolf
2021-04-30 10:51 ` [PULL 37/39] block: Add BDRV_O_NO_SHARE for blk_new_open() Kevin Wolf
2021-04-30 10:51 ` [PULL 38/39] qemu-img convert: Unshare write permission for source Kevin Wolf
2021-04-30 10:51 ` [PULL 39/39] vhost-user-blk: Fail gracefully on too large queue size Kevin Wolf
2021-04-30 15:00 ` [PULL 00/39] 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=CAFEAcA_u4X+Cq6b0UtPeLXF7-7ugiMmOeb1u3UMMqTP-W0-MQA@mail.gmail.com \
--to=peter.maydell@linaro.org \
--cc=kwolf@redhat.com \
--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 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).