From: xiubli@redhat.com
To: ceph-devel@vger.kernel.org
Cc: Xiubo Li <xiubli@redhat.com>
Subject: [PATCH v2] ceph: fail the request if the peer MDS doesn't support getvxattr op
Date: Wed, 3 Aug 2022 17:18:44 +0800 [thread overview]
Message-ID: <20220803091844.608743-1-xiubli@redhat.com> (raw)
From: Xiubo Li <xiubli@redhat.com>
Just fail the request instead sending the request out, or the peer
MDS will crash.
URL: https://tracker.ceph.com/issues/56529
Signed-off-by: Xiubo Li <xiubli@redhat.com>
---
V2:
- Sync to the https://github.com/ceph/ceph/pull/47063. Will return -EOPNOTSUPP
directly instead of falling back to old method.
fs/ceph/inode.c | 1 +
fs/ceph/mds_client.c | 11 +++++++++++
fs/ceph/mds_client.h | 6 +++++-
fs/ceph/xattr.c | 6 +-----
4 files changed, 18 insertions(+), 6 deletions(-)
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index 79ff197c7cc5..cfa0b550eef2 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -2607,6 +2607,7 @@ int ceph_do_getvxattr(struct inode *inode, const char *name, void *value,
goto out;
}
+ req->r_feature_needed = CEPHFS_FEATURE_OP_GETVXATTR;
req->r_path2 = kstrdup(name, GFP_NOFS);
if (!req->r_path2) {
err = -ENOMEM;
diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index 598012ddc401..e3683305445c 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -2501,6 +2501,7 @@ ceph_mdsc_create_request(struct ceph_mds_client *mdsc, int op, int mode)
INIT_LIST_HEAD(&req->r_unsafe_dir_item);
INIT_LIST_HEAD(&req->r_unsafe_target_item);
req->r_fmode = -1;
+ req->r_feature_needed = -1;
kref_init(&req->r_kref);
RB_CLEAR_NODE(&req->r_node);
INIT_LIST_HEAD(&req->r_wait);
@@ -3255,6 +3256,16 @@ static void __do_request(struct ceph_mds_client *mdsc,
dout("do_request mds%d session %p state %s\n", mds, session,
ceph_session_state_name(session->s_state));
+
+ /*
+ * The old ceph will crash the MDSs when see unknown OPs
+ */
+ if (req->r_feature_needed > 0 &&
+ !test_bit(req->r_feature_needed, &session->s_features)) {
+ err = -EOPNOTSUPP;
+ goto out_session;
+ }
+
if (session->s_state != CEPH_MDS_SESSION_OPEN &&
session->s_state != CEPH_MDS_SESSION_HUNG) {
/*
diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h
index e15ee2858fef..728b7d72bf76 100644
--- a/fs/ceph/mds_client.h
+++ b/fs/ceph/mds_client.h
@@ -31,8 +31,9 @@ enum ceph_feature_type {
CEPHFS_FEATURE_METRIC_COLLECT,
CEPHFS_FEATURE_ALTERNATE_NAME,
CEPHFS_FEATURE_NOTIFY_SESSION_STATE,
+ CEPHFS_FEATURE_OP_GETVXATTR,
- CEPHFS_FEATURE_MAX = CEPHFS_FEATURE_NOTIFY_SESSION_STATE,
+ CEPHFS_FEATURE_MAX = CEPHFS_FEATURE_OP_GETVXATTR,
};
#define CEPHFS_FEATURES_CLIENT_SUPPORTED { \
@@ -45,6 +46,7 @@ enum ceph_feature_type {
CEPHFS_FEATURE_METRIC_COLLECT, \
CEPHFS_FEATURE_ALTERNATE_NAME, \
CEPHFS_FEATURE_NOTIFY_SESSION_STATE, \
+ CEPHFS_FEATURE_OP_GETVXATTR, \
}
/*
@@ -352,6 +354,8 @@ struct ceph_mds_request {
long long r_dir_ordered_cnt;
int r_readdir_cache_idx;
+ int r_feature_needed;
+
struct ceph_cap_reservation r_caps_reservation;
};
diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c
index b10d459c2326..5f29c1f08de1 100644
--- a/fs/ceph/xattr.c
+++ b/fs/ceph/xattr.c
@@ -983,11 +983,7 @@ ssize_t __ceph_getxattr(struct inode *inode, const char *name, void *value,
}
return err;
} else {
- err = ceph_do_getvxattr(inode, name, value, size);
- /* this would happen with a new client and old server combo */
- if (err == -EOPNOTSUPP)
- err = -ENODATA;
- return err;
+ return ceph_do_getvxattr(inode, name, value, size);
}
handle_non_vxattrs:
req_mask = __get_request_mask(inode);
--
2.36.0.rc1
next reply other threads:[~2022-08-03 9:18 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-08-03 9:18 xiubli [this message]
2022-08-08 2:26 ` [PATCH v2] ceph: fail the request if the peer MDS doesn't support getvxattr op Xiubo Li
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=20220803091844.608743-1-xiubli@redhat.com \
--to=xiubli@redhat.com \
--cc=ceph-devel@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.