From: Max Reitz <mreitz@redhat.com>
To: qemu-block@nongnu.org
Cc: Kevin Wolf <kwolf@redhat.com>,
qemu-devel@nongnu.org, Max Reitz <mreitz@redhat.com>
Subject: [PATCH v3 09/33] block: Add generic bdrv_inherited_options()
Date: Tue, 18 Feb 2020 13:42:18 +0100 [thread overview]
Message-ID: <20200218124242.584644-10-mreitz@redhat.com> (raw)
In-Reply-To: <20200218124242.584644-1-mreitz@redhat.com>
After the series this patch belongs to, we want to have a common
BdrvChildClass that encompasses all of child_file, child_format, and
child_backing. Such a single class needs a single .inherit_options()
implementation, and this patch introduces it.
The next patch will show how the existing implementations can fall back
to it just by passing appropriate BdrvChildRole and parent_is_format
values.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
---
block.c | 84 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 84 insertions(+)
diff --git a/block.c b/block.c
index c33f0e9b42..9179b9b604 100644
--- a/block.c
+++ b/block.c
@@ -998,6 +998,90 @@ static void bdrv_temp_snapshot_options(int *child_flags, QDict *child_options,
*child_flags &= ~BDRV_O_NATIVE_AIO;
}
+/*
+ * Returns the options and flags that a generic child of a BDS should
+ * get, based on the given options and flags for the parent BDS.
+ */
+static void __attribute__((unused))
+ bdrv_inherited_options(BdrvChildRole role, bool parent_is_format,
+ int *child_flags, QDict *child_options,
+ int parent_flags, QDict *parent_options)
+{
+ int flags = parent_flags;
+
+ /*
+ * First, decide whether to set, clear, or leave BDRV_O_PROTOCOL.
+ * Generally, the question to answer is: Should this child be
+ * format-probed by default?
+ */
+
+ /*
+ * Pure and non-filtered data children of non-format nodes should
+ * be probed by default (even when the node itself has BDRV_O_PROTOCOL
+ * set). This only affects a very limited set of drivers (namely
+ * quorum and blkverify when this comment was written).
+ * Force-clear BDRV_O_PROTOCOL then.
+ */
+ if (!parent_is_format &&
+ (role & (BDRV_CHILD_DATA | BDRV_CHILD_METADATA |
+ BDRV_CHILD_FILTERED)) ==
+ BDRV_CHILD_DATA)
+ {
+ flags &= ~BDRV_O_PROTOCOL;
+ }
+
+ /*
+ * All children of format nodes (except for COW children) and all
+ * metadata children in general should never be format-probed.
+ * Force-set BDRV_O_PROTOCOL then.
+ */
+ if ((parent_is_format && !(role & BDRV_CHILD_COW)) ||
+ (role & BDRV_CHILD_METADATA))
+ {
+ flags |= BDRV_O_PROTOCOL;
+ }
+
+ /*
+ * If the cache mode isn't explicitly set, inherit direct and no-flush from
+ * the parent.
+ */
+ qdict_copy_default(child_options, parent_options, BDRV_OPT_CACHE_DIRECT);
+ qdict_copy_default(child_options, parent_options, BDRV_OPT_CACHE_NO_FLUSH);
+ qdict_copy_default(child_options, parent_options, BDRV_OPT_FORCE_SHARE);
+
+ if (role & BDRV_CHILD_COW) {
+ /* backing files are always opened read-only */
+ qdict_set_default_str(child_options, BDRV_OPT_READ_ONLY, "on");
+ qdict_set_default_str(child_options, BDRV_OPT_AUTO_READ_ONLY, "off");
+ } else {
+ /* Inherit the read-only option from the parent if it's not set */
+ qdict_copy_default(child_options, parent_options, BDRV_OPT_READ_ONLY);
+ qdict_copy_default(child_options, parent_options,
+ BDRV_OPT_AUTO_READ_ONLY);
+ }
+
+ if (parent_is_format && !(role & BDRV_CHILD_COW)) {
+ /*
+ * Our format drivers take care to send flushes and respect
+ * unmap policy, so we can default to enable both on lower
+ * layers regardless of the corresponding parent options.
+ */
+ qdict_set_default_str(child_options, BDRV_OPT_DISCARD, "unmap");
+ }
+
+ /* Clear flags that only apply to the top layer */
+ flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING | BDRV_O_COPY_ON_READ);
+
+ if (role & BDRV_CHILD_METADATA) {
+ flags &= ~BDRV_O_NO_IO;
+ }
+ if (role & BDRV_CHILD_COW) {
+ flags &= ~BDRV_O_TEMPORARY;
+ }
+
+ *child_flags = flags;
+}
+
/*
* Returns the options and flags that bs->file should get if a protocol driver
* is expected, based on the given options and flags for the parent BDS
--
2.24.1
next prev parent reply other threads:[~2020-02-18 12:49 UTC|newest]
Thread overview: 63+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-02-18 12:42 [PATCH v3 00/33] block: Introduce real BdrvChildRole Max Reitz
2020-02-18 12:42 ` [PATCH v3 01/33] block: Add BlockDriver.is_format Max Reitz
2020-02-18 12:42 ` [PATCH v3 02/33] block: Rename BdrvChildRole to BdrvChildClass Max Reitz
2020-02-18 12:42 ` [PATCH v3 03/33] block: Add BdrvChildRole and BdrvChildRoleBits Max Reitz
2020-02-18 13:05 ` Eric Blake
2020-05-05 11:19 ` Kevin Wolf
2020-05-05 11:59 ` Max Reitz
2020-05-05 12:54 ` Kevin Wolf
2020-05-05 13:20 ` Max Reitz
2020-05-05 13:38 ` Kevin Wolf
2020-02-18 12:42 ` [PATCH v3 04/33] block: Add BdrvChildRole to BdrvChild Max Reitz
2020-02-18 13:06 ` Eric Blake
2020-02-18 12:42 ` [PATCH v3 05/33] block: Pass BdrvChildRole to bdrv_child_perm() Max Reitz
2020-02-18 12:42 ` [PATCH v3 06/33] block: Pass BdrvChildRole to .inherit_options() Max Reitz
2020-02-18 12:42 ` [PATCH v3 07/33] block: Pass parent_is_format " Max Reitz
2020-02-18 12:42 ` [PATCH v3 08/33] block: Rename bdrv_inherited_options() Max Reitz
2020-02-18 12:42 ` Max Reitz [this message]
2020-05-06 10:37 ` [PATCH v3 09/33] block: Add generic bdrv_inherited_options() Kevin Wolf
2020-05-06 13:11 ` Kevin Wolf
2020-05-07 9:18 ` Max Reitz
2020-05-07 8:49 ` Max Reitz
2020-05-07 11:19 ` Kevin Wolf
2020-05-07 11:34 ` Max Reitz
2020-02-18 12:42 ` [PATCH v3 10/33] block: Use bdrv_inherited_options() Max Reitz
2020-02-18 12:42 ` [PATCH v3 11/33] block: Unify bdrv_child_cb_attach() Max Reitz
2020-02-18 12:42 ` [PATCH v3 12/33] block: Unify bdrv_child_cb_detach() Max Reitz
2020-05-06 12:41 ` Kevin Wolf
2020-05-07 9:09 ` Max Reitz
2020-02-18 12:42 ` [PATCH v3 13/33] block: Add child_of_bds Max Reitz
2020-05-06 12:59 ` Kevin Wolf
2020-02-18 12:42 ` [PATCH v3 14/33] block: Distinguish paths in *_format_default_perms Max Reitz
2020-02-18 12:42 ` [PATCH v3 15/33] block: Pull out bdrv_default_perms_for_backing() Max Reitz
2020-05-06 13:21 ` Kevin Wolf
2020-05-07 9:19 ` Max Reitz
2020-02-18 12:42 ` [PATCH v3 16/33] block: Pull out bdrv_default_perms_for_storage() Max Reitz
2020-02-18 12:42 ` [PATCH v3 17/33] block: Split bdrv_default_perms_for_storage() Max Reitz
2020-02-18 12:42 ` [PATCH v3 18/33] block: Add bdrv_default_perms() Max Reitz
2020-05-06 13:47 ` Kevin Wolf
2020-05-07 9:26 ` Max Reitz
2020-02-18 12:42 ` [PATCH v3 19/33] raw-format: Split raw_read_options() Max Reitz
2020-02-18 12:42 ` [PATCH v3 20/33] block: Switch child_format users to child_of_bds Max Reitz
2020-02-18 13:10 ` Eric Blake
2020-02-18 12:42 ` [PATCH v3 21/33] block: Drop child_format Max Reitz
2020-02-18 12:42 ` [PATCH v3 22/33] block: Make backing files child_of_bds children Max Reitz
2020-05-06 16:37 ` Kevin Wolf
2020-05-07 9:28 ` Max Reitz
2020-02-18 12:42 ` [PATCH v3 23/33] block: Drop child_backing Max Reitz
2020-02-18 12:42 ` [PATCH v3 24/33] block: Make format drivers use child_of_bds Max Reitz
2020-02-18 12:42 ` [PATCH v3 25/33] block: Make filter " Max Reitz
2020-02-18 12:42 ` [PATCH v3 26/33] block: Use child_of_bds in remaining places Max Reitz
2020-05-06 17:04 ` Kevin Wolf
2020-05-07 9:33 ` Max Reitz
2020-05-07 11:32 ` Kevin Wolf
2020-02-18 12:42 ` [PATCH v3 27/33] tests: Use child_of_bds instead of child_file Max Reitz
2020-02-18 12:42 ` [PATCH v3 28/33] block: Use bdrv_default_perms() Max Reitz
2020-02-18 12:42 ` [PATCH v3 29/33] block: Make bdrv_filter_default_perms() static Max Reitz
2020-02-18 12:42 ` [PATCH v3 30/33] block: Drop bdrv_format_default_perms() Max Reitz
2020-02-18 12:42 ` [PATCH v3 31/33] block: Drop child_file Max Reitz
2020-02-18 12:42 ` [PATCH v3 32/33] block: Pass BdrvChildRole in remaining cases Max Reitz
2020-05-06 17:13 ` Kevin Wolf
2020-05-07 9:36 ` Max Reitz
2020-05-07 11:40 ` Kevin Wolf
2020-02-18 12:42 ` [PATCH v3 33/33] block: Drop @child_class from bdrv_child_perm() Max Reitz
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=20200218124242.584644-10-mreitz@redhat.com \
--to=mreitz@redhat.com \
--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).