public inbox for linux-rdma@vger.kernel.org
 help / color / mirror / Atom feed
From: Zhu Yanjun <yanjun.zhu@linux.dev>
To: Bob Pearson <rpearsonhpe@gmail.com>,
	Zhu Yanjun <yanjun.zhu@intel.com>,
	zyjzyj2000@gmail.com, jgg@ziepe.ca, leon@kernel.org,
	linux-rdma@vger.kernel.org, parav@nvidia.com, lehrer@gmail.com
Cc: Rain River <rain.1986.08.12@gmail.com>
Subject: Re: [PATCHv5 for-rc1 v5 5/8] RDMA/rxe: Replace global variable with sock lookup functions
Date: Wed, 21 Jun 2023 09:30:10 +0800	[thread overview]
Message-ID: <9bf10fb4-1d2a-17c7-55f2-394936465ff4@linux.dev> (raw)
In-Reply-To: <db1aa8dc-04f6-e437-9809-b5e63372d53d@gmail.com>


在 2023/6/21 4:28, Bob Pearson 写道:
> On 4/28/23 04:39, Zhu Yanjun wrote:
>> From: Zhu Yanjun <yanjun.zhu@linux.dev>
>>
>> Originally a global variable is to keep the sock of udp listening
>> on port 4791. In fact, sock lookup functions can be used to get
>> the sock.
>>
>> Tested-by: Rain River <rain.1986.08.12@gmail.com>
>> Signed-off-by: Zhu Yanjun <yanjun.zhu@linux.dev>
>> ---
>>   drivers/infiniband/sw/rxe/rxe.c       |  1 +
>>   drivers/infiniband/sw/rxe/rxe_net.c   | 58 ++++++++++++++++++++-------
>>   drivers/infiniband/sw/rxe/rxe_net.h   |  5 ---
>>   drivers/infiniband/sw/rxe/rxe_verbs.h |  1 +
>>   4 files changed, 45 insertions(+), 20 deletions(-)
>>
>> diff --git a/drivers/infiniband/sw/rxe/rxe.c b/drivers/infiniband/sw/rxe/rxe.c
>> index ebfabc6d6b76..e81c2164d77f 100644
>> --- a/drivers/infiniband/sw/rxe/rxe.c
>> +++ b/drivers/infiniband/sw/rxe/rxe.c
>> @@ -74,6 +74,7 @@ static void rxe_init_device_param(struct rxe_dev *rxe)
>>   			rxe->ndev->dev_addr);
>>   
>>   	rxe->max_ucontext			= RXE_MAX_UCONTEXT;
>> +	rxe->l_sk6				= NULL;
>>   }
>>   
>>   /* initialize port attributes */
>> diff --git a/drivers/infiniband/sw/rxe/rxe_net.c b/drivers/infiniband/sw/rxe/rxe_net.c
>> index 4cc7de7b115b..b56e2c32fbf7 100644
>> --- a/drivers/infiniband/sw/rxe/rxe_net.c
>> +++ b/drivers/infiniband/sw/rxe/rxe_net.c
>> @@ -18,8 +18,6 @@
>>   #include "rxe_net.h"
>>   #include "rxe_loc.h"
>>   
>> -static struct rxe_recv_sockets recv_sockets;
>> -
>>   static struct dst_entry *rxe_find_route4(struct rxe_qp *qp,
>>   					 struct net_device *ndev,
>>   					 struct in_addr *saddr,
>> @@ -51,6 +49,23 @@ static struct dst_entry *rxe_find_route6(struct rxe_qp *qp,
>>   {
>>   	struct dst_entry *ndst;
>>   	struct flowi6 fl6 = { { 0 } };
>> +	struct rxe_dev *rdev;
>> +
>> +	rdev = rxe_get_dev_from_net(ndev);
>> +	if (!rdev->l_sk6) {
>> +		struct sock *sk;
>> +
>> +		rcu_read_lock();
>> +		sk = udp6_lib_lookup(&init_net, NULL, 0, &in6addr_any, htons(ROCE_V2_UDP_DPORT), 0);
> This subroutine (and similar udp4) is currently called for every UD packet. This just adds a lot of
> code to packet processing for UD packets. All to save a global variable. Not sure it's worth it.

Thanks, Bob. I also noticed this problem. 2 variables rxe_sk4 and 
rxe_sk6 are added.

The 2 functions udp6_lib_lookup and udp4_lib_lookup are removed.

Zhu Yanjun

>
> Bob
>> +		rcu_read_unlock();
>> +		if (!sk) {
>> +			pr_info("file: %s +%d, error\n", __FILE__, __LINE__);
>> +			return (struct dst_entry *)sk;
>> +		}
>> +		__sock_put(sk);
>> +		rdev->l_sk6 = sk->sk_socket;
>> +	}
>> +
>>   
>>   	memset(&fl6, 0, sizeof(fl6));
>>   	fl6.flowi6_oif = ndev->ifindex;
>> @@ -58,8 +73,8 @@ static struct dst_entry *rxe_find_route6(struct rxe_qp *qp,
>>   	memcpy(&fl6.daddr, daddr, sizeof(*daddr));
>>   	fl6.flowi6_proto = IPPROTO_UDP;
>>   
>> -	ndst = ipv6_stub->ipv6_dst_lookup_flow(sock_net(recv_sockets.sk6->sk),
>> -					       recv_sockets.sk6->sk, &fl6,
>> +	ndst = ipv6_stub->ipv6_dst_lookup_flow(dev_net(ndev),
>> +					       rdev->l_sk6->sk, &fl6,
>>   					       NULL);
>>   	if (IS_ERR(ndst)) {
>>   		rxe_dbg_qp(qp, "no route to %pI6\n", daddr);
>> @@ -533,15 +548,33 @@ int rxe_net_add(const char *ibdev_name, struct net_device *ndev)
>>   #define SK_REF_FOR_TUNNEL	2
>>   void rxe_net_del(struct ib_device *dev)
>>   {
>> -	if (refcount_read(&recv_sockets.sk6->sk->sk_refcnt) > SK_REF_FOR_TUNNEL)
>> -		__sock_put(recv_sockets.sk6->sk);
>> +	struct sock *sk;
>> +
>> +	rcu_read_lock();
>> +	sk = udp4_lib_lookup(&init_net, 0, 0, htonl(INADDR_ANY), htons(ROCE_V2_UDP_DPORT), 0);
>> +	rcu_read_unlock();
>> +	if (!sk)
>> +		return;
>> +
>> +	__sock_put(sk);
>> +
>> +	if (refcount_read(&sk->sk_refcnt) > SK_REF_FOR_TUNNEL)
>> +		__sock_put(sk);
>>   	else
>> -		rxe_release_udp_tunnel(recv_sockets.sk6);
>> +		rxe_release_udp_tunnel(sk->sk_socket);
>> +
>> +	rcu_read_lock();
>> +	sk = udp6_lib_lookup(&init_net, NULL, 0, &in6addr_any, htons(ROCE_V2_UDP_DPORT), 0);
>> +	rcu_read_unlock();
>> +	if (!sk)
>> +		return;
>> +
>> +	__sock_put(sk);
>>   
>> -	if (refcount_read(&recv_sockets.sk4->sk->sk_refcnt) > SK_REF_FOR_TUNNEL)
>> -		__sock_put(recv_sockets.sk4->sk);
>> +	if (refcount_read(&sk->sk_refcnt) > SK_REF_FOR_TUNNEL)
>> +		__sock_put(sk);
>>   	else
>> -		rxe_release_udp_tunnel(recv_sockets.sk4);
>> +		rxe_release_udp_tunnel(sk->sk_socket);
>>   }
>>   #undef SK_REF_FOR_TUNNEL
>>   
>> @@ -651,10 +684,8 @@ static int rxe_net_ipv4_init(void)
>>   	sock = rxe_setup_udp_tunnel(&init_net, htons(ROCE_V2_UDP_DPORT), false);
>>   	if (IS_ERR(sock)) {
>>   		pr_err("Failed to create IPv4 UDP tunnel\n");
>> -		recv_sockets.sk4 = NULL;
>>   		return -1;
>>   	}
>> -	recv_sockets.sk4 = sock;
>>   
>>   	return 0;
>>   }
>> @@ -674,17 +705,14 @@ static int rxe_net_ipv6_init(void)
>>   
>>   	sock = rxe_setup_udp_tunnel(&init_net, htons(ROCE_V2_UDP_DPORT), true);
>>   	if (PTR_ERR(sock) == -EAFNOSUPPORT) {
>> -		recv_sockets.sk6 = NULL;
>>   		pr_warn("IPv6 is not supported, can not create a UDPv6 socket\n");
>>   		return 0;
>>   	}
>>   
>>   	if (IS_ERR(sock)) {
>> -		recv_sockets.sk6 = NULL;
>>   		pr_err("Failed to create IPv6 UDP tunnel\n");
>>   		return -1;
>>   	}
>> -	recv_sockets.sk6 = sock;
>>   #endif
>>   	return 0;
>>   }
>> diff --git a/drivers/infiniband/sw/rxe/rxe_net.h b/drivers/infiniband/sw/rxe/rxe_net.h
>> index f48f22f3353b..027b20e1bab6 100644
>> --- a/drivers/infiniband/sw/rxe/rxe_net.h
>> +++ b/drivers/infiniband/sw/rxe/rxe_net.h
>> @@ -11,11 +11,6 @@
>>   #include <net/if_inet6.h>
>>   #include <linux/module.h>
>>   
>> -struct rxe_recv_sockets {
>> -	struct socket *sk4;
>> -	struct socket *sk6;
>> -};
>> -
>>   int rxe_net_add(const char *ibdev_name, struct net_device *ndev);
>>   void rxe_net_del(struct ib_device *dev);
>>   
>> diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.h b/drivers/infiniband/sw/rxe/rxe_verbs.h
>> index c269ae2a3224..ac9bb55820a2 100644
>> --- a/drivers/infiniband/sw/rxe/rxe_verbs.h
>> +++ b/drivers/infiniband/sw/rxe/rxe_verbs.h
>> @@ -396,6 +396,7 @@ struct rxe_dev {
>>   
>>   	struct rxe_port		port;
>>   	struct crypto_shash	*tfm;
>> +	struct socket		*l_sk6;
>>   };
>>   
>>   static inline void rxe_counter_inc(struct rxe_dev *rxe, enum rxe_counters index)

  reply	other threads:[~2023-06-21  1:30 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-04-28  9:39 [PATCHv5 for-rc1 v5 0/8] Fix the problem that rxe can not work in net namespace Zhu Yanjun
2023-04-28  9:39 ` [PATCHv5 for-rc1 v5 1/8] RDMA/rxe: Creating listening sock in newlink function Zhu Yanjun
2023-04-28  9:39 ` [PATCHv5 for-rc1 v5 2/8] RDMA/rxe: Support more rdma links in init_net Zhu Yanjun
2023-04-28  9:39 ` [PATCHv5 for-rc1 v5 3/8] RDMA/nldev: Add dellink function pointer Zhu Yanjun
2023-04-28  9:39 ` [PATCHv5 for-rc1 v5 4/8] RDMA/rxe: Implement dellink in rxe Zhu Yanjun
2023-06-20 20:21   ` Bob Pearson
2023-06-21  2:13     ` Zhu Yanjun
2023-06-21  3:23       ` Bob Pearson
2023-06-21  6:17         ` Zhu Yanjun
2023-06-21 16:24           ` Bob Pearson
2023-06-23  7:19             ` Zhu Yanjun
2023-04-28  9:39 ` [PATCHv5 for-rc1 v5 5/8] RDMA/rxe: Replace global variable with sock lookup functions Zhu Yanjun
2023-06-20 20:28   ` Bob Pearson
2023-06-21  1:30     ` Zhu Yanjun [this message]
2023-04-28  9:39 ` [PATCHv5 for-rc1 v5 6/8] RDMA/rxe: add the support of net namespace Zhu Yanjun
2023-04-28  9:39 ` [PATCHv5 for-rc1 v5 7/8] RDMA/rxe: Add the support of net namespace notifier Zhu Yanjun
2023-04-28  9:39 ` [PATCHv5 for-rc1 v5 8/8] RDMA/rxe: Replace l_sk6 with sk6 in net namespace Zhu Yanjun

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=9bf10fb4-1d2a-17c7-55f2-394936465ff4@linux.dev \
    --to=yanjun.zhu@linux.dev \
    --cc=jgg@ziepe.ca \
    --cc=lehrer@gmail.com \
    --cc=leon@kernel.org \
    --cc=linux-rdma@vger.kernel.org \
    --cc=parav@nvidia.com \
    --cc=rain.1986.08.12@gmail.com \
    --cc=rpearsonhpe@gmail.com \
    --cc=yanjun.zhu@intel.com \
    --cc=zyjzyj2000@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox