Re: [PATCH v2] ceph: force a cap message when a deferred revoke can't be acked immediately

Alex Markuze <[email protected]>
Newsgroups org.kernel.vger.ceph-devel
Message-ID <[email protected]>
Hi Max,

Two issues in v2:

The clear_bit(CEPH_I_FLUSH_FORCE_BIT) in ceph_check_caps() fires whenever
any cap flush completes on the inode, not only when the deferred-revoke cap
message itself is done. A concurrent cap flush for a different cap clears
the flag prematurely, before the deferred revoke is handled.

Also, ci->i_ceph_flags |= CEPH_I_FLUSH_FORCE is a non-atomic
read-modify-write — every other flag on i_ceph_flags uses set_bit() /
clear_bit(). This one should be
set_bit(CEPH_I_FLUSH_FORCE_BIT, &ci->i_ceph_flags).

Could you respin?

--
Alex Markuze
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.