[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <89f76eef-42e3-1469-5329-bd6512490bf6@redhat.com>
Date: Mon, 28 Mar 2022 19:01:11 +0800
From: Xiubo Li <xiubli@...hat.com>
To: Jeff Layton <jlayton@...nel.org>
Cc: idryomov@...il.com, vshankar@...hat.com,
ceph-devel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] ceph: remove unused CEPH_MDS_LEASE_RELEASE related code
On 3/28/22 6:28 PM, Jeff Layton wrote:
> On Mon, 2022-03-28 at 10:25 +0800, xiubli@...hat.com wrote:
>> From: Xiubo Li <xiubli@...hat.com>
>>
>> The ceph_mdsc_lease_release() has been removed by commit(8aa152c77890)
>> and the CEPH_MDS_LEASE_RELEASE will never be used.
>>
> Like it says in Documentation/process/5.Posting.rst:
>
> "...and please provide both the commit ID and the title when citing
> commits"
>
> You might want to reword this with something like:
>
> "ceph_mdsc_lease_release was removed by commit 8aa152c77890 (ceph:
> remove ceph_mdsc_lease_release). ceph_mdsc_lease_send_msg will never
> call this function with CEPH_MDS_LEASE_RELEASE."
Sure, will fix it.
>> Signed-off-by: Xiubo Li <xiubli@...hat.com>
>> ---
>> fs/ceph/mds_client.c | 6 ------
>> 1 file changed, 6 deletions(-)
>>
>> diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
>> index 459c6f23915f..a89ee866ebbb 100644
>> --- a/fs/ceph/mds_client.c
>> +++ b/fs/ceph/mds_client.c
>> @@ -4424,12 +4424,6 @@ void ceph_mdsc_lease_send_msg(struct ceph_mds_session *session,
>> memcpy((void *)(lease + 1) + 4,
>> dentry->d_name.name, dentry->d_name.len);
>> spin_unlock(&dentry->d_lock);
>> - /*
>> - * if this is a preemptive lease RELEASE, no need to
>> - * flush request stream, since the actual request will
>> - * soon follow.
>> - */
>> - msg->more_to_follow = (action == CEPH_MDS_LEASE_RELEASE);
>>
>> ceph_con_send(&session->s_con, msg);
>> }
> It might be possible to trim this function back further. There's only
> one caller and it always calls this with the same "action" value. Still,
> this looks fine...
Yeah, I checked the commit history in both kernel and ceph, it seems
never used since this was introduced. Maybe I missed something.
The user space clients is use the CEPH_MDS_LEASE_RELEASE as the lease
revoke ack. In kernel we are using the CEPH_MDS_LEASE_REVOKE_ACK instead.
Thanks.
>
> Reviewed-by: Jeff Layton <jlayton@...nel.org>
>
Powered by blists - more mailing lists