linux/fs/ceph
Xiubo Li 825978fd6a ceph: fix possible deadlock when holding Fwb to get inline_data
1, mount with wsync.
2, create a file with O_RDWR, and the request was sent to mds.0:

   ceph_atomic_open()-->
     ceph_mdsc_do_request(openc)
     finish_open(file, dentry, ceph_open)-->
       ceph_open()-->
         ceph_init_file()-->
           ceph_init_file_info()-->
             ceph_uninline_data()-->
             {
               ...
               if (inline_version == 1 || /* initial version, no data */
                   inline_version == CEPH_INLINE_NONE)
                     goto out_unlock;
               ...
             }

The inline_version will be 1, which is the initial version for the
new create file. And here the ci->i_inline_version will keep with 1,
it's buggy.

3, buffer write to the file immediately:

   ceph_write_iter()-->
     ceph_get_caps(file, need=Fw, want=Fb, ...);
     generic_perform_write()-->
       a_ops->write_begin()-->
         ceph_write_begin()-->
           netfs_write_begin()-->
             netfs_begin_read()-->
               netfs_rreq_submit_slice()-->
                 netfs_read_from_server()-->
                   rreq->netfs_ops->issue_read()-->
                     ceph_netfs_issue_read()-->
                     {
                       ...
                       if (ci->i_inline_version != CEPH_INLINE_NONE &&
                           ceph_netfs_issue_op_inline(subreq))
                         return;
                       ...
                     }
     ceph_put_cap_refs(ci, Fwb);

The ceph_netfs_issue_op_inline() will send a getattr(Fsr) request to
mds.1.

4, then the mds.1 will request the rd lock for CInode::filelock from
the auth mds.0, the mds.0 will do the CInode::filelock state transation
from excl --> sync, but it need to revoke the Fxwb caps back from the
clients.

While the kernel client has aleady held the Fwb caps and waiting for
the getattr(Fsr).

It's deadlock!

URL: https://tracker.ceph.com/issues/55377
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2022-05-25 20:45:14 +02:00
..
acl.c vfs: add rcu argument to ->get_acl() callback 2021-08-18 22:08:24 +02:00
addr.c ceph: fix possible deadlock when holding Fwb to get inline_data 2022-05-25 20:45:14 +02:00
cache.c netfs: Add a netfs inode context 2022-03-18 09:29:05 +00:00
cache.h Netfs prep for write helpers 2022-03-31 15:49:36 -07:00
caps.c ceph: rename unsafe_request_wait() 2022-05-25 20:45:14 +02:00
ceph_frag.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
debugfs.c ceph: use tracked average r/w/m latencies to display metrics in debugfs 2022-03-21 13:35:16 +01:00
dir.c ceph: fix memory leak in ceph_readdir when note_last_dentry returns error 2022-03-21 13:35:16 +01:00
export.c ceph: shut down access to inode when async create fails 2021-11-08 03:29:51 +01:00
file.c ceph: fix setting of xattrs on async created inodes 2022-05-10 09:48:31 +02:00
inode.c ceph: try to choose the auth MDS if possible for getattr 2022-05-25 20:45:14 +02:00
io.c ceph: fix kerneldoc copypasta over ceph_start_io_direct 2021-04-27 23:52:23 +02:00
io.h ceph: add buffered/direct exclusionary locking for reads and writes 2019-09-16 12:06:25 +02:00
ioctl.c ceph: consider inode's last read/write when calculating wanted caps 2020-03-30 12:42:42 +02:00
ioctl.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
Kconfig ceph: conversion to new fscache API 2022-01-11 22:13:01 +00:00
locks.c ceph: eliminate req->r_wait_for_completion from ceph_mds_request 2022-03-01 18:26:36 +01:00
Makefile ceph: add dentry lease metric support 2020-06-01 13:22:51 +02:00
mds_client.c ceph: flush the mdlog for filesystem sync 2022-05-25 20:45:14 +02:00
mds_client.h ceph: add getvxattr op 2022-03-01 18:26:37 +01:00
mdsmap.c ceph: fix mdsmap decode when there are MDS's beyond max_mds 2021-11-08 03:29:52 +01:00
metric.c ceph: include average/stdev r/w/m latency in mds metrics 2022-03-21 13:35:16 +01:00
metric.h ceph: include average/stdev r/w/m latency in mds metrics 2022-03-21 13:35:16 +01:00
quota.c ceph: don't check for quotas on MDS stray dirs 2022-01-13 13:40:07 +01:00
snap.c ceph: misc fix for code style and logs 2022-03-01 18:26:37 +01:00
strings.c ceph: add getvxattr op 2022-03-01 18:26:37 +01:00
super.c ceph: disable updating the atime since cephfs won't maintain it 2022-05-25 20:45:14 +02:00
super.h ceph: try to choose the auth MDS if possible for getattr 2022-05-25 20:45:14 +02:00
util.c ceph: move net/ceph/ceph_fs.c to fs/ceph/util.c 2020-01-27 16:53:40 +01:00
xattr.c ceph: allow ceph.dir.rctime xattr to be updatable 2022-05-25 20:45:13 +02:00