> On 6 Apr 2017, at 01:21, Jeff Layton <jlayton@xxxxxxxxxx> wrote:
>
> On Wed, 2017-04-05 at 09:30 +0800, Yan, Zheng wrote:
>> __unregister_session() free the session if it drops the last
>> reference. We should grab an extra reference if we want to use
>> session after __unregister_session().
>>
>> Signed-off-by: "Yan, Zheng" <zyan@xxxxxxxxxx>
>> ---
>> fs/ceph/mds_client.c | 10 ++++++++--
>> 1 file changed, 8 insertions(+), 2 deletions(-)
>>
>> diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
>> index 163f0d3..bf765a8 100644
>> --- a/fs/ceph/mds_client.c
>> +++ b/fs/ceph/mds_client.c
>> @@ -2658,8 +2658,10 @@ static void handle_session(struct ceph_mds_session *session,
>> seq = le64_to_cpu(h->seq);
>>
>> mutex_lock(&mdsc->mutex);
>> - if (op == CEPH_SESSION_CLOSE)
>> + if (op == CEPH_SESSION_CLOSE) {
>> + get_session(session);
>> __unregister_session(mdsc, session);
>> + }
>> /* FIXME: this ttl calculation is generous */
>> session->s_ttl = jiffies + HZ*mdsc->mdsmap->m_session_autoclose;
>> mutex_unlock(&mdsc->mutex);
>> @@ -2748,6 +2750,8 @@ static void handle_session(struct ceph_mds_session *session,
>> kick_requests(mdsc, mds);
>> mutex_unlock(&mdsc->mutex);
>> }
>> + if (op == CEPH_SESSION_CLOSE)
>> + ceph_put_mds_session(session);
>> return;
>>
>> bad:
>> @@ -3148,8 +3152,10 @@ static void check_new_map(struct ceph_mds_client *mdsc,
>> if (s->s_state == CEPH_MDS_SESSION_OPENING) {
>> /* the session never opened, just close it
>> * out now */
>> - __wake_requests(mdsc, &s->s_waiting);
>> + get_session(s);
>> __unregister_session(mdsc, s);
>> + __wake_requests(mdsc, &s->s_waiting);
>> + ceph_put_mds_session(s);
>
> What about this last bit? Why do we need to __wake_requests after
> __unregister_session here? If not for that change then you wouldn't
> need to take the extra reference here, AFAICS.
wake_requests() calls __do_request(). I think do_request may choose the session we want to unregister.
>
>> } else {
>> /* just close it */
>> mutex_unlock(&mdsc->mutex);
>
> Reviewed-by: Jeff Layton <jlayton@xxxxxxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
