Merge branch 'separate-smc-parameter-settings-from-tcp-sysctls'

Wen Gu says:

====================
Separate SMC parameter settings from TCP sysctls

SMC shares some sysctls with TCP, but considering the difference
between these two protocols, it may not be very suitable for SMC
to reuse TCP parameter settings in some cases, such as keepalive
time or buffer size.

So this patch set aims to introduce some SMC specific sysctls to
independently and flexibly set the parameters that suit SMC.
====================

Link: https://lore.kernel.org/r/1663667542-119851-1-git-send-email-guwen@linux.alibaba.comSigned-off-by: Paolo Abeni <pabeni@redhat.com>

Documentation/networking/smc-sysctl.rst

@@ -34,3 +34,28 @@ smcr_buf_type - INTEGER
         - 1 - Use virtually contiguous buffers
         - 2 - Mixed use of the two types. Try physically contiguous buffers first.
           If not available, use virtually contiguous buffers then.
+
+smcr_testlink_time - INTEGER
+	How frequently SMC-R link sends out TEST_LINK LLC messages to confirm
+	viability, after the last activity of connections on it. Value 0 means
+	disabling TEST_LINK.
+
+	Default: 30 seconds.
+
+wmem - INTEGER
+	Initial size of send buffer used by SMC sockets.
+	The default value inherits from net.ipv4.tcp_wmem[1].
+
+	The minimum value is 16KiB and there is no hard limit for max value, but
+	only allowed 512KiB for SMC-R and 1MiB for SMC-D.
+
+	Default: 16K
+
+rmem - INTEGER
+	Initial size of receive buffer (RMB) used by SMC sockets.
+	The default value inherits from net.ipv4.tcp_rmem[1].
+
+	The minimum value is 16KiB and there is no hard limit for max value, but
+	only allowed 512KiB for SMC-R and 1MiB for SMC-D.
+
+	Default: 128K

include/net/netns/smc.h

@@ -19,5 +19,8 @@ struct netns_smc {
 #endif
 	unsigned int			sysctl_autocorking_size;
 	unsigned int			sysctl_smcr_buf_type;
+	int				sysctl_smcr_testlink_time;
+	int				sysctl_wmem;
+	int				sysctl_rmem;
 };
 #endif

net/smc/af_smc.c

@@ -379,6 +379,8 @@ static struct sock *smc_sock_alloc(struct net *net, struct socket *sock,
 	sk->sk_state = SMC_INIT;
 	sk->sk_destruct = smc_destruct;
 	sk->sk_protocol = protocol;
+	WRITE_ONCE(sk->sk_sndbuf, READ_ONCE(net->smc.sysctl_wmem));
+	WRITE_ONCE(sk->sk_rcvbuf, READ_ONCE(net->smc.sysctl_rmem));
 	smc = smc_sk(sk);
 	INIT_WORK(&smc->tcp_listen_work, smc_tcp_listen_work);
 	INIT_WORK(&smc->connect_work, smc_connect_work);
@@ -3253,9 +3255,6 @@ static int __smc_create(struct net *net, struct socket *sock, int protocol,
 		smc->clcsock = clcsock;
 	}
 
-	smc->sk.sk_sndbuf = max(smc->clcsock->sk->sk_sndbuf, SMC_BUF_MIN_SIZE);
-	smc->sk.sk_rcvbuf = max(smc->clcsock->sk->sk_rcvbuf, SMC_BUF_MIN_SIZE);
-
 out:
 	return rc;
 }

net/smc/smc_core.c

@@ -2307,10 +2307,10 @@ static int __smc_buf_create(struct smc_sock *smc, bool is_smcd, bool is_rmb)
 
 	if (is_rmb)
 		/* use socket recv buffer size (w/o overhead) as start value */
-		sk_buf_size = smc->sk.sk_rcvbuf / 2;
+		sk_buf_size = smc->sk.sk_rcvbuf;
 	else
 		/* use socket send buffer size (w/o overhead) as start value */
-		sk_buf_size = smc->sk.sk_sndbuf / 2;
+		sk_buf_size = smc->sk.sk_sndbuf;
 
 	for (bufsize_short = smc_compress_bufsize(sk_buf_size, is_smcd, is_rmb);
 	     bufsize_short >= 0; bufsize_short--) {
@@ -2369,7 +2369,7 @@ static int __smc_buf_create(struct smc_sock *smc, bool is_smcd, bool is_rmb)
 	if (is_rmb) {
 		conn->rmb_desc = buf_desc;
 		conn->rmbe_size_short = bufsize_short;
-		smc->sk.sk_rcvbuf = bufsize * 2;
+		smc->sk.sk_rcvbuf = bufsize;
 		atomic_set(&conn->bytes_to_rcv, 0);
 		conn->rmbe_update_limit =
 			smc_rmb_wnd_update_limit(buf_desc->len);
@@ -2377,7 +2377,7 @@ static int __smc_buf_create(struct smc_sock *smc, bool is_smcd, bool is_rmb)
 			smc_ism_set_conn(conn); /* map RMB/smcd_dev to conn */
 	} else {
 		conn->sndbuf_desc = buf_desc;
-		smc->sk.sk_sndbuf = bufsize * 2;
+		smc->sk.sk_sndbuf = bufsize;
 		atomic_set(&conn->sndbuf_space, bufsize);
 	}
 	return 0;

net/smc/smc_llc.c

@@ -2127,7 +2127,7 @@ void smc_llc_lgr_init(struct smc_link_group *lgr, struct smc_sock *smc)
 	init_waitqueue_head(&lgr->llc_flow_waiter);
 	init_waitqueue_head(&lgr->llc_msg_waiter);
 	mutex_init(&lgr->llc_conf_mutex);
-	lgr->llc_testlink_time = READ_ONCE(net->ipv4.sysctl_tcp_keepalive_time);
+	lgr->llc_testlink_time = READ_ONCE(net->smc.sysctl_smcr_testlink_time);
 }
 
 /* called after lgr was removed from lgr_list */

net/smc/smc_llc.h

@@ -19,6 +19,7 @@
 
 #define SMC_LLC_WAIT_FIRST_TIME		(5 * HZ)
 #define SMC_LLC_WAIT_TIME		(2 * HZ)
+#define SMC_LLC_TESTLINK_DEFAULT_TIME	(30 * HZ)
 
 enum smc_llc_reqresp {
 	SMC_LLC_REQ,

net/smc/smc_sysctl.c

@@ -16,8 +16,12 @@
 
 #include "smc.h"
 #include "smc_core.h"
+#include "smc_llc.h"
 #include "smc_sysctl.h"
 
+static int min_sndbuf = SMC_BUF_MIN_SIZE;
+static int min_rcvbuf = SMC_BUF_MIN_SIZE;
+
 static struct ctl_table smc_table[] = {
 	{
 		.procname       = "autocorking_size",
@@ -35,6 +39,29 @@ static struct ctl_table smc_table[] = {
 		.extra1		= SYSCTL_ZERO,
 		.extra2		= SYSCTL_TWO,
 	},
+	{
+		.procname	= "smcr_testlink_time",
+		.data		= &init_net.smc.sysctl_smcr_testlink_time,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_jiffies,
+	},
+	{
+		.procname	= "wmem",
+		.data		= &init_net.smc.sysctl_wmem,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &min_sndbuf,
+	},
+	{
+		.procname	= "rmem",
+		.data		= &init_net.smc.sysctl_rmem,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &min_rcvbuf,
+	},
 	{  }
 };
 
@@ -60,6 +87,9 @@ int __net_init smc_sysctl_net_init(struct net *net)
 
 	net->smc.sysctl_autocorking_size = SMC_AUTOCORKING_DEFAULT_SIZE;
 	net->smc.sysctl_smcr_buf_type = SMCR_PHYS_CONT_BUFS;
+	net->smc.sysctl_smcr_testlink_time = SMC_LLC_TESTLINK_DEFAULT_TIME;
+	WRITE_ONCE(net->smc.sysctl_wmem, READ_ONCE(net->ipv4.sysctl_tcp_wmem[1]));
+	WRITE_ONCE(net->smc.sysctl_rmem, READ_ONCE(net->ipv4.sysctl_tcp_rmem[1]));
 
 	return 0;