Date:   Wed, 9 Nov 2022 17:10:10 +0800
From:   "D. Wythe" <alibuda@...ux.alibaba.com>
To:     Wenjia Zhang <wenjia@...ux.ibm.com>,
        Jan Karcher <jaka@...ux.ibm.com>, kgraul@...ux.ibm.com
Cc:     kuba@...nel.org, davem@...emloft.net, netdev@...r.kernel.org,
        linux-s390@...r.kernel.org, linux-rdma@...r.kernel.org
Subject: Re: [PATCH net-next v4 00/10] optimize the parallelism of SMC-R
 connections


Hi Wenjia and Jan,

I'm not sure whether my guess is right, I need some help from you. I guess the smcd_ops register_dmb()
is not thread-safe, after I remove the lock, different connections might get the same sba_idx, which will cause
the connection to be lost in the map(smcd->conn). If so, the CDC message carrying close/abort information cannot be
distributed to the correct connection, then the connection remains in link group abnormally.

/* Set a connection using this DMBE. */
void smc_ism_set_conn(struct smc_connection *conn)
{
	unsigned long flags;

	spin_lock_irqsave(&conn->lgr->smcd->lock, flags);
	conn->lgr->smcd->conn[conn->rmb_desc->sba_idx] = conn;
	spin_unlock_irqrestore(&conn->lgr->smcd->lock, flags);
}


struct smcd_ops {

	int (*register_dmb)(struct smcd_dev *dev, struct smcd_dmb *dmb);
}


On 11/7/22 7:05 PM, D. Wythe wrote:
> 
> Hi Wenjia,
> 
> Thanks a lot for your information, before that we thought you did PATCH test one by one,
> now I think I have found the root cause, and I will release a new version to fix this
> soon as possible.
> 
> Best Wishes.
> D. Wythe
> 
> On 11/2/22 9:55 PM, Wenjia Zhang wrote:
>>
>>
>> On 01.11.22 08:22, D. Wythe wrote:
>>>
>>> Hi Jan,
>>>
>>> Our team conducted some code reviews over this, but unfortunately no obvious problems were found. Hence
>>> we are waiting for Tony Lu's virtual SMC-D device to test, which is expected to come in this week.  Before that,
>>> I wonder if your tests are running separately on separate PATCH? If so, I would like to please you to test
>>> the first PATCH and the second PATCH together. I doubt that the problem repaired by the second PATCH
>>> is the cause of this issues.
>>>
>>> Best Wishes.
>>> D. Wythe
>>>
>>
>> Hi D. Wythe,
>>
>> We did test the series of the patches as a whole. That would be great if you could use Tony's virtual device to test SMC-D. By the way, I'll put your patches in our CI, let's see if it can find something.
>>
>> Best,
>> Wenjia
>>>
>>> On 10/24/22 9:11 PM, Jan Karcher wrote:
>>>> Hi D. Wythe,
>>>>
>>>> I re-run the tests with your fix.
>>>> SMC-R works fine now. For SMC-D we still have the following problem. It is kind of the same as i reported in v2 but even weirder:
>>>>
>>>> smc stats:
>>>>
>>>> t8345011
>>>> SMC-D Connections Summary
>>>>    Total connections handled          2465
>>>> SMC-R Connections Summary
>>>>    Total connections handled           232
>>>>
>>>> t8345010
>>>> SMC-D Connections Summary
>>>>    Total connections handled          2290
>>>> SMC-R Connections Summary
>>>>    Total connections handled           231
>>>>
>>>>
>>>> smc linkgroups:
>>>>
>>>> t8345011
>>>> [root@...45011 ~]# smcr linkgroup
>>>> LG-ID    LG-Role  LG-Type  VLAN  #Conns  PNET-ID
>>>> 00000400 SERV     SYM         0       0  NET25
>>>> [root@...45011 ~]# smcd linkgroup
>>>> LG-ID    VLAN  #Conns  PNET-ID
>>>> 00000300    0      16  NET25
>>>>
>>>> t8345010
>>>> [root@...45010 tela-kernel]# smcr linkgroup
>>>> LG-ID    LG-Role  LG-Type  VLAN  #Conns  PNET-ID
>>>> 00000400 CLNT     SYM         0       0  NET25
>>>> [root@...45010 tela-kernel]# smcd linkgroup
>>>> LG-ID    VLAN  #Conns  PNET-ID
>>>> 00000300    0       1  NET25
>>>>
>>>>
>>>> smcss:
>>>>
>>>> t8345011
>>>> [root@...45011 ~]# smcss
>>>> State          UID   Inode   Local Address           Peer Address Intf Mode
>>>>
>>>> t8345010
>>>> [root@...45010 tela-kernel]# smcss
>>>> State          UID   Inode   Local Address           Peer Address Intf Mode
>>>>
>>>>
>>>> lsmod:
>>>>
>>>> t8345011
>>>> [root@...45011 ~]# lsmod | grep smc
>>>> smc                   225280  18 ism,smc_diag
>>>> t8345010
>>>> [root@...45010 tela-kernel]# lsmod | grep smc
>>>> smc                   225280  3 ism,smc_diag
>>>>
>>>> Also smc_dbg and netstat do not show any more information on this problem. We only see in the dmesg that the code seems to build up SMC-R linkgroups even tho we are running the SMC-D tests.
>>>> NOTE: we disabled the syncookies for the tests.
>>>>
>>>> dmesg:
>>>>
>>>> t8345011
>>>> smc-tests: test_smcapp_torture_test started
>>>> kernel: TCP: request_sock_TCP: Possible SYN flooding on port 22465. Dropping request.  Check SNMP counters.
>>>> kernel: smc: SMC-R lg 00000400 net 1 link added: id 00000401, peerid 00000401, ibdev mlx5_0, ibport 1
>>>> kernel: smc: SMC-R lg 00000400 net 1 state changed: SINGLE, pnetid NET25
>>>> kernel: smc: SMC-R lg 00000400 net 1 link added: id 00000402, peerid 00000402, ibdev mlx5_1, ibport 1
>>>> kernel: smc: SMC-R lg 00000400 net 1 state changed: SYMMETRIC, pnetid NET25
>>>>
>>>> t8345010
>>>> smc-tests: test_smcapp_torture_test started
>>>> kernel: smc: SMC-R lg 00000400 net 1 link added: id 00000401, peerid 00000401, ibdev mlx5_0, ibport 1
>>>> kernel: smc: SMC-R lg 00000400 net 1 state changed: SINGLE, pnetid NET25
>>>> kernel: smc: SMC-R lg 00000400 net 1 link added: id 00000402, peerid 00000402, ibdev mlx5_1, ibport 1
>>>> kernel: smc: SMC-R lg 00000400 net 1 state changed: SYMMETRIC, pnetid NET25
>>>>
>>>> If this output does not help and if you want us to look deeper into it feel free to let us know and we can debug further.
>>>>
>>>> On 23/10/2022 14:43, D.Wythe wrote:
>>>>> From: "D.Wythe" <alibuda@...ux.alibaba.com>
>>>>>
>>>>> This patch set attempts to optimize the parallelism of SMC-R connections,
>>>>> mainly to reduce unnecessary blocking on locks, and to fix exceptions that
>>>>> occur after thoses optimization.
>>>>>
>>>>> According to Off-CPU graph, SMC worker's off-CPU as that:
>>>>>
>>>>> smc_close_passive_work                  (1.09%)
>>>>>          smcr_buf_unuse                  (1.08%)
>>>>>                  smc_llc_flow_initiate   (1.02%)
>>>>>
>>>>> smc_listen_work                         (48.17%)
>>>>>          __mutex_lock.isra.11            (47.96%)
>>>>>
>>>>>
>>>>> An ideal SMC-R connection process should only block on the IO events
>>>>> of the network, but it's quite clear that the SMC-R connection now is
>>>>> queued on the lock most of the time.
>>>>>
>>>>> The goal of this patchset is to achieve our ideal situation where
>>>>> network IO events are blocked for the majority of the connection lifetime.
>>>>>
>>>>> There are three big locks here:
>>>>>
>>>>> 1. smc_client_lgr_pending & smc_server_lgr_pending
>>>>>
>>>>> 2. llc_conf_mutex
>>>>>
>>>>> 3. rmbs_lock & sndbufs_lock
>>>>>
>>>>> And an implementation issue:
>>>>>
>>>>> 1. confirm/delete rkey msg can't be sent concurrently while
>>>>> protocol allows indeed.
>>>>>
>>>>> Unfortunately,The above problems together affect the parallelism of
>>>>> SMC-R connection. If any of them are not solved. our goal cannot
>>>>> be achieved.
>>>>>
>>>>> After this patch set, we can get a quite ideal off-CPU graph as
>>>>> following:
>>>>>
>>>>> smc_close_passive_work                                  (41.58%)
>>>>>          smcr_buf_unuse                                  (41.57%)
>>>>>                  smc_llc_do_delete_rkey                  (41.57%)
>>>>>
>>>>> smc_listen_work                                         (39.10%)
>>>>>          smc_clc_wait_msg                                (13.18%)
>>>>>                  tcp_recvmsg_locked                      (13.18)
>>>>>          smc_listen_find_device                          (25.87%)
>>>>>                  smcr_lgr_reg_rmbs                       (25.87%)
>>>>>                          smc_llc_do_confirm_rkey         (25.87%)
>>>>>
>>>>> We can see that most of the waiting times are waiting for network IO
>>>>> events. This also has a certain performance improvement on our
>>>>> short-lived conenction wrk/nginx benchmark test:
>>>>>
>>>>> +--------------+------+------+-------+--------+------+--------+
>>>>> |conns/qps     |c4    | c8   |  c16  |  c32   | c64  |  c200  |
>>>>> +--------------+------+------+-------+--------+------+--------+
>>>>> |SMC-R before  |9.7k  | 10k  |  10k  |  9.9k  | 9.1k |  8.9k  |
>>>>> +--------------+------+------+-------+--------+------+--------+
>>>>> |SMC-R now     |13k   | 19k  |  18k  |  16k   | 15k  |  12k   |
>>>>> +--------------+------+------+-------+--------+------+--------+
>>>>> |TCP           |15k   | 35k  |  51k  |  80k   | 100k |  162k  |
>>>>> +--------------+------+------+-------+--------+------+--------+
>>>>>
>>>>> The reason why the benefit is not obvious after the number of connections
>>>>> has increased dues to workqueue. If we try to change workqueue to UNBOUND,
>>>>> we can obtain at least 4-5 times performance improvement, reach up to half
>>>>> of TCP. However, this is not an elegant solution, the optimization of it
>>>>> will be much more complicated. But in any case, we will submit relevant
>>>>> optimization patches as soon as possible.
>>>>>
>>>>> Please note that the premise here is that the lock related problem
>>>>> must be solved first, otherwise, no matter how we optimize the workqueue,
>>>>> there won't be much improvement.
>>>>>
>>>>> Because there are a lot of related changes to the code, if you have
>>>>> any questions or suggestions, please let me know.
>>>>>
>>>>> Thanks
>>>>> D. Wythe
>>>>>
>>>>> v1 -> v2:
>>>>>
>>>>> 1. Fix panic in SMC-D scenario
>>>>> 2. Fix lnkc related hashfn calculation exception, caused by operator
>>>>> priority
>>>>> 3. Only wake up one connection if the lnk is not active
>>>>> 4. Delete obsolete unlock logic in smc_listen_work()
>>>>> 5. PATCH format, do Reverse Christmas tree
>>>>> 6. PATCH format, change all xxx_lnk_xxx function to xxx_link_xxx
>>>>> 7. PATCH format, add correct fix tag for the patches for fixes.
>>>>> 8. PATCH format, fix some spelling error
>>>>> 9. PATCH format, rename slow to do_slow
>>>>>
>>>>> v2 -> v3:
>>>>>
>>>>> 1. add SMC-D support, remove the concept of link cluster since SMC-D has
>>>>> no link at all. Replace it by lgr decision maker, who provides suggestions
>>>>> to SMC-D and SMC-R on whether to create new link group.
>>>>>
>>>>> 2. Fix the corruption problem described by PATCH 'fix application
>>>>> data exception' on SMC-D.
>>>>>
>>>>> v3 -> v4:
>>>>>
>>>>> 1. Fix panic caused by uninitialization map.
>>>>>
>>>>> D. Wythe (10):
>>>>>    net/smc: remove locks smc_client_lgr_pending and
>>>>>      smc_server_lgr_pending
>>>>>    net/smc: fix SMC_CLC_DECL_ERR_REGRMB without smc_server_lgr_pending
>>>>>    net/smc: allow confirm/delete rkey response deliver multiplex
>>>>>    net/smc: make SMC_LLC_FLOW_RKEY run concurrently
>>>>>    net/smc: llc_conf_mutex refactor, replace it with rw_semaphore
>>>>>    net/smc: use read semaphores to reduce unnecessary blocking in
>>>>>      smc_buf_create() & smcr_buf_unuse()
>>>>>    net/smc: reduce unnecessary blocking in smcr_lgr_reg_rmbs()
>>>>>    net/smc: replace mutex rmbs_lock and sndbufs_lock with rw_semaphore
>>>>>    net/smc: Fix potential panic dues to unprotected
>>>>>      smc_llc_srv_add_link()
>>>>>    net/smc: fix application data exception
>>>>>
>>>>>   net/smc/af_smc.c   |  70 ++++----
>>>>>   net/smc/smc_core.c | 478 +++++++++++++++++++++++++++++++++++++++++++++++------
>>>>>   net/smc/smc_core.h |  36 +++-
>>>>>   net/smc/smc_llc.c  | 277 ++++++++++++++++++++++---------
>>>>>   net/smc/smc_llc.h  |   6 +
>>>>>   net/smc/smc_wr.c   |  10 --
>>>>>   net/smc/smc_wr.h   |  10 ++
>>>>>   7 files changed, 712 insertions(+), 175 deletions(-)
>>>>>

Powered by blists - more mailing lists