From: Max Kellermann <max.kellermann@ionos.com>
To: idryomov@gmail.com, amarkuze@redhat.com,
ceph-devel@vger.kernel.org, linux-kernel@vger.kernel.org
Cc: Max Kellermann <max.kellermann@ionos.com>
Subject: [PATCH v2 3/4] fs/ceph/caps: mark cap remove with RB_CLEAR_NODE() instead of ci=NULL
Date: Mon, 6 Jul 2026 09:38:11 +0200 [thread overview]
Message-ID: <20260706073812.3922732-3-max.kellermann@ionos.com> (raw)
In-Reply-To: <20260706073812.3922732-1-max.kellermann@ionos.com>
__ceph_remove_cap() erases the ceph_cap object from the RB tree, thus
it seems natural to use RB_CLEAR_NODE() / RB_EMPTY_NODE() for the
removal check.
Signed-off-by: Max Kellermann <max.kellermann@ionos.com>
---
v1->v2:
- move code to new patch "add helper function ceph_cap_is_removed()"
- add locking documentation to ceph_cap.ci_node
- fix data race by calling RB_CLEAR_NODE() under s_cap_lock
Signed-off-by: Max Kellermann <max.kellermann@ionos.com>
---
fs/ceph/caps.c | 7 +++++--
fs/ceph/super.h | 16 ++++++++++++++--
2 files changed, 19 insertions(+), 4 deletions(-)
diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index f6f3e817b712..a6a59bc91b79 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -1157,8 +1157,11 @@ static void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release)
cap->session = NULL;
removed = 1;
}
- /* protect backpointer with s_cap_lock: see iterate_session_caps */
- cap->ci = NULL;
+
+ /* protect removal marker with both i_ceph_lock and
+ s_cap_lock, so either one can be used to check for
+ removal */
+ RB_CLEAR_NODE(&cap->ci_node);
/*
* s_cap_reconnect is protected by s_cap_lock. no one changes
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
index 628d153601a5..509839c1db0b 100644
--- a/fs/ceph/super.h
+++ b/fs/ceph/super.h
@@ -203,7 +203,19 @@ struct ceph_fs_client {
*/
struct ceph_cap {
struct ceph_inode_info *ci;
- struct rb_node ci_node; /* per-ci cap tree */
+
+ /**
+ * Per-ci cap tree. Protected with
+ * `ceph_inode_info.i_ceph_lock`.
+ *
+ * Clearing this field with RB_CLEAR_NODE() requires holding
+ * both `ceph_inode_info.i_ceph_lock` and
+ * `ceph_mds_session->s_cap_lock`. Calling RB_EMPTY_NODE()
+ * (via ceph_cap_is_removed()) requires holding at least one
+ * of these.
+ */
+ struct rb_node ci_node;
+
struct ceph_mds_session *session;
struct list_head session_caps; /* per-session caplist */
u64 cap_id; /* unique cap id (mds provided) */
@@ -1280,7 +1292,7 @@ extern void ceph_add_cap(struct inode *inode,
*/
static inline bool ceph_cap_is_removed(const struct ceph_cap *cap)
{
- return !cap->ci;
+ return RB_EMPTY_NODE(&cap->ci_node);
}
extern void ceph_remove_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap,
--
2.47.3
next prev parent reply other threads:[~2026-07-06 7:38 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-07-06 7:38 [PATCH v2 1/4] fs/ceph/caps: make __ceph_remove_cap() static Max Kellermann
2026-07-06 7:38 ` [PATCH v2 2/4] fs/ceph: add helper function ceph_cap_is_removed() Max Kellermann
2026-07-06 7:38 ` Max Kellermann [this message]
2026-07-06 7:38 ` [PATCH v2 4/4] fs/ceph/cap: pass inode pointer around instead of reloading from RAM Max Kellermann
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=20260706073812.3922732-3-max.kellermann@ionos.com \
--to=max.kellermann@ionos.com \
--cc=amarkuze@redhat.com \
--cc=ceph-devel@vger.kernel.org \
--cc=idryomov@gmail.com \
--cc=linux-kernel@vger.kernel.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