From e6727f39004bd95725342b3b343a14c7d59df07f Mon Sep 17 00:00:00 2001 From: Ursula Braun Date: Mon, 9 Jan 2017 16:55:23 +0100 Subject: [PATCH] smc: send data (through RDMA) copy data to kernel send buffer, and trigger RDMA write Signed-off-by: Ursula Braun Signed-off-by: David S. Miller --- net/smc/Makefile | 2 +- net/smc/af_smc.c | 13 +- net/smc/smc.h | 1 + net/smc/smc_cdc.c | 7 +- net/smc/smc_tx.c | 438 ++++++++++++++++++++++++++++++++++++++++++++++ net/smc/smc_tx.h | 34 ++++ 6 files changed, 491 insertions(+), 4 deletions(-) create mode 100644 net/smc/smc_tx.c create mode 100644 net/smc/smc_tx.h diff --git a/net/smc/Makefile b/net/smc/Makefile index ec0fd0338c6d..fc28d79aec7d 100644 --- a/net/smc/Makefile +++ b/net/smc/Makefile @@ -1,3 +1,3 @@ obj-$(CONFIG_SMC) += smc.o smc-y := af_smc.o smc_pnet.o smc_ib.o smc_clc.o smc_core.o smc_wr.o smc_llc.o -smc-y += smc_cdc.o +smc-y += smc_cdc.o smc_tx.o diff --git a/net/smc/af_smc.c b/net/smc/af_smc.c index 4b1611d46ef7..b62b69c6c718 100644 --- a/net/smc/af_smc.c +++ b/net/smc/af_smc.c @@ -37,6 +37,7 @@ #include "smc_core.h" #include "smc_ib.h" #include "smc_pnet.h" +#include "smc_tx.h" static DEFINE_MUTEX(smc_create_lgr_pending); /* serialize link group * creation @@ -410,6 +411,8 @@ static int smc_connect_rdma(struct smc_sock *smc) } mutex_unlock(&smc_create_lgr_pending); + smc_tx_init(smc); + out_connected: smc_copy_sock_settings_to_clc(smc); smc->sk.sk_state = SMC_ACTIVE; @@ -751,6 +754,8 @@ static void smc_listen_work(struct work_struct *work) goto decline_rdma; } + smc_tx_init(new_smc); + out_connected: sk_refcnt_debug_inc(newsmcsk); newsmcsk->sk_state = SMC_ACTIVE; @@ -924,7 +929,7 @@ static int smc_sendmsg(struct socket *sock, struct msghdr *msg, size_t len) if (smc->use_fallback) rc = smc->clcsock->ops->sendmsg(smc->clcsock, msg, len); else - rc = sock_no_sendmsg(sock, msg, len); + rc = smc_tx_sendmsg(smc, msg, len); out: release_sock(sk); return rc; @@ -1005,6 +1010,12 @@ static unsigned int smc_poll(struct file *file, struct socket *sock, mask |= smc_accept_poll(sk); if (sk->sk_err) mask |= POLLERR; + if (atomic_read(&smc->conn.sndbuf_space)) { + mask |= POLLOUT | POLLWRNORM; + } else { + sk_set_bit(SOCKWQ_ASYNC_NOSPACE, sk); + set_bit(SOCK_NOSPACE, &sk->sk_socket->flags); + } /* for now - to be enhanced in follow-on patch */ } diff --git a/net/smc/smc.h b/net/smc/smc.h index a8d4a68aac30..0c47d84c1db1 100644 --- a/net/smc/smc.h +++ b/net/smc/smc.h @@ -133,6 +133,7 @@ struct smc_connection { atomic_t sndbuf_space; /* remaining space in sndbuf */ u16 tx_cdc_seq; /* sequence # for CDC send */ spinlock_t send_lock; /* protect wr_sends */ + struct work_struct tx_work; /* retry of smc_cdc_msg_send */ struct smc_host_cdc_msg local_rx_ctrl; /* filled during event_handl. * .prod cf. TCP rcv_nxt diff --git a/net/smc/smc_cdc.c b/net/smc/smc_cdc.c index 111d23ff9e17..77fe16967376 100644 --- a/net/smc/smc_cdc.c +++ b/net/smc/smc_cdc.c @@ -14,6 +14,7 @@ #include "smc.h" #include "smc_wr.h" #include "smc_cdc.h" +#include "smc_tx.h" /********************************** send *************************************/ @@ -52,7 +53,7 @@ static void smc_cdc_tx_handler(struct smc_wr_tx_pend_priv *pnd_snd, smc_curs_read(&cdcpend->cursor, cdcpend->conn), cdcpend->conn); } - /* subsequent patch: wake if send buffer space available */ + smc_tx_sndbuf_nonfull(smc); bh_unlock_sock(&smc->sk); } @@ -204,7 +205,9 @@ static void smc_cdc_msg_recv_action(struct smc_sock *smc, /* subsequent patch: terminate connection */ /* piggy backed tx info */ - /* subsequent patch: wake receivers if receive buffer space available */ + /* trigger sndbuf consumer: RDMA write into peer RMBE and CDC */ + if (diff_cons && smc_tx_prepared_sends(conn)) + smc_tx_sndbuf_nonempty(conn); /* subsequent patch: trigger socket release if connection closed */ diff --git a/net/smc/smc_tx.c b/net/smc/smc_tx.c new file mode 100644 index 000000000000..d86bef6cb681 --- /dev/null +++ b/net/smc/smc_tx.c @@ -0,0 +1,438 @@ +/* + * Shared Memory Communications over RDMA (SMC-R) and RoCE + * + * Manage send buffer. + * Producer: + * Copy user space data into send buffer, if send buffer space available. + * Consumer: + * Trigger RDMA write into RMBE of peer and send CDC, if RMBE space available. + * + * Copyright IBM Corp. 2016 + * + * Author(s): Ursula Braun + */ + +#include +#include +#include +#include + +#include "smc.h" +#include "smc_wr.h" +#include "smc_cdc.h" +#include "smc_tx.h" + +/***************************** sndbuf producer *******************************/ + +/* callback implementation for sk.sk_write_space() + * to wakeup sndbuf producers that blocked with smc_tx_wait_memory(). + * called under sk_socket lock. + */ +static void smc_tx_write_space(struct sock *sk) +{ + struct socket *sock = sk->sk_socket; + struct smc_sock *smc = smc_sk(sk); + struct socket_wq *wq; + + /* similar to sk_stream_write_space */ + if (atomic_read(&smc->conn.sndbuf_space) && sock) { + clear_bit(SOCK_NOSPACE, &sock->flags); + rcu_read_lock(); + wq = rcu_dereference(sk->sk_wq); + if (skwq_has_sleeper(wq)) + wake_up_interruptible_poll(&wq->wait, + POLLOUT | POLLWRNORM | + POLLWRBAND); + if (wq && wq->fasync_list && !(sk->sk_shutdown & SEND_SHUTDOWN)) + sock_wake_async(wq, SOCK_WAKE_SPACE, POLL_OUT); + rcu_read_unlock(); + } +} + +/* Wakeup sndbuf producers that blocked with smc_tx_wait_memory(). + * Cf. tcp_data_snd_check()=>tcp_check_space()=>tcp_new_space(). + */ +void smc_tx_sndbuf_nonfull(struct smc_sock *smc) +{ + if (smc->sk.sk_socket && + test_bit(SOCK_NOSPACE, &smc->sk.sk_socket->flags)) + smc->sk.sk_write_space(&smc->sk); +} + +/* blocks sndbuf producer until at least one byte of free space available */ +static int smc_tx_wait_memory(struct smc_sock *smc, int flags) +{ + DEFINE_WAIT_FUNC(wait, woken_wake_function); + struct smc_connection *conn = &smc->conn; + struct sock *sk = &smc->sk; + bool noblock; + long timeo; + int rc = 0; + + /* similar to sk_stream_wait_memory */ + timeo = sock_sndtimeo(sk, flags & MSG_DONTWAIT); + noblock = timeo ? false : true; + add_wait_queue(sk_sleep(sk), &wait); + while (1) { + sk_set_bit(SOCKWQ_ASYNC_NOSPACE, sk); + if (sk->sk_err || + (sk->sk_shutdown & SEND_SHUTDOWN) || + conn->local_tx_ctrl.conn_state_flags.peer_done_writing) { + rc = -EPIPE; + break; + } + if (conn->local_rx_ctrl.conn_state_flags.peer_conn_abort) { + rc = -ECONNRESET; + break; + } + if (!timeo) { + if (noblock) + set_bit(SOCK_NOSPACE, &sk->sk_socket->flags); + rc = -EAGAIN; + break; + } + if (signal_pending(current)) { + rc = sock_intr_errno(timeo); + break; + } + sk_clear_bit(SOCKWQ_ASYNC_NOSPACE, sk); + if (atomic_read(&conn->sndbuf_space)) + break; /* at least 1 byte of free space available */ + set_bit(SOCK_NOSPACE, &sk->sk_socket->flags); + sk->sk_write_pending++; + sk_wait_event(sk, &timeo, + sk->sk_err || + (sk->sk_shutdown & SEND_SHUTDOWN) || + smc_cdc_rxed_any_close_or_senddone(conn) || + atomic_read(&conn->sndbuf_space), + &wait); + sk->sk_write_pending--; + } + remove_wait_queue(sk_sleep(sk), &wait); + return rc; +} + +/* sndbuf producer: main API called by socket layer. + * called under sock lock. + */ +int smc_tx_sendmsg(struct smc_sock *smc, struct msghdr *msg, size_t len) +{ + size_t copylen, send_done = 0, send_remaining = len; + size_t chunk_len, chunk_off, chunk_len_sum; + struct smc_connection *conn = &smc->conn; + union smc_host_cursor prep; + struct sock *sk = &smc->sk; + char *sndbuf_base; + int tx_cnt_prep; + int writespace; + int rc, chunk; + + /* This should be in poll */ + sk_clear_bit(SOCKWQ_ASYNC_NOSPACE, sk); + + if (sk->sk_err || (sk->sk_shutdown & SEND_SHUTDOWN)) { + rc = -EPIPE; + goto out_err; + } + + while (msg_data_left(msg)) { + if (sk->sk_state == SMC_INIT) + return -ENOTCONN; + if (smc->sk.sk_shutdown & SEND_SHUTDOWN || + conn->local_tx_ctrl.conn_state_flags.peer_conn_abort) + return -EPIPE; + if (smc_cdc_rxed_any_close(conn)) + return send_done ?: -ECONNRESET; + + if (!atomic_read(&conn->sndbuf_space)) { + rc = smc_tx_wait_memory(smc, msg->msg_flags); + if (rc) { + if (send_done) + return send_done; + goto out_err; + } + continue; + } + + /* initialize variables for 1st iteration of subsequent loop */ + /* could be just 1 byte, even after smc_tx_wait_memory above */ + writespace = atomic_read(&conn->sndbuf_space); + /* not more than what user space asked for */ + copylen = min_t(size_t, send_remaining, writespace); + /* determine start of sndbuf */ + sndbuf_base = conn->sndbuf_desc->cpu_addr; + smc_curs_write(&prep, + smc_curs_read(&conn->tx_curs_prep, conn), + conn); + tx_cnt_prep = prep.count; + /* determine chunks where to write into sndbuf */ + /* either unwrapped case, or 1st chunk of wrapped case */ + chunk_len = min_t(size_t, + copylen, conn->sndbuf_size - tx_cnt_prep); + chunk_len_sum = chunk_len; + chunk_off = tx_cnt_prep; + for (chunk = 0; chunk < 2; chunk++) { + rc = memcpy_from_msg(sndbuf_base + chunk_off, + msg, chunk_len); + if (rc) { + if (send_done) + return send_done; + goto out_err; + } + send_done += chunk_len; + send_remaining -= chunk_len; + + if (chunk_len_sum == copylen) + break; /* either on 1st or 2nd iteration */ + /* prepare next (== 2nd) iteration */ + chunk_len = copylen - chunk_len; /* remainder */ + chunk_len_sum += chunk_len; + chunk_off = 0; /* modulo offset in send ring buffer */ + } + /* update cursors */ + smc_curs_add(conn->sndbuf_size, &prep, copylen); + smc_curs_write(&conn->tx_curs_prep, + smc_curs_read(&prep, conn), + conn); + /* increased in send tasklet smc_cdc_tx_handler() */ + smp_mb__before_atomic(); + atomic_sub(copylen, &conn->sndbuf_space); + /* guarantee 0 <= sndbuf_space <= sndbuf_size */ + smp_mb__after_atomic(); + /* since we just produced more new data into sndbuf, + * trigger sndbuf consumer: RDMA write into peer RMBE and CDC + */ + smc_tx_sndbuf_nonempty(conn); + } /* while (msg_data_left(msg)) */ + + return send_done; + +out_err: + rc = sk_stream_error(sk, msg->msg_flags, rc); + /* make sure we wake any epoll edge trigger waiter */ + if (unlikely(rc == -EAGAIN)) + sk->sk_write_space(sk); + return rc; +} + +/***************************** sndbuf consumer *******************************/ + +/* sndbuf consumer: actual data transfer of one target chunk with RDMA write */ +static int smc_tx_rdma_write(struct smc_connection *conn, int peer_rmbe_offset, + int num_sges, struct ib_sge sges[]) +{ + struct smc_link_group *lgr = conn->lgr; + struct ib_send_wr *failed_wr = NULL; + struct ib_rdma_wr rdma_wr; + struct smc_link *link; + int rc; + + memset(&rdma_wr, 0, sizeof(rdma_wr)); + link = &lgr->lnk[SMC_SINGLE_LINK]; + rdma_wr.wr.wr_id = smc_wr_tx_get_next_wr_id(link); + rdma_wr.wr.sg_list = sges; + rdma_wr.wr.num_sge = num_sges; + rdma_wr.wr.opcode = IB_WR_RDMA_WRITE; + rdma_wr.remote_addr = + lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].dma_addr + + /* RMBE within RMB */ + ((conn->peer_conn_idx - 1) * conn->peer_rmbe_size) + + /* offset within RMBE */ + peer_rmbe_offset; + rdma_wr.rkey = lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].rkey; + rc = ib_post_send(link->roce_qp, &rdma_wr.wr, &failed_wr); + if (rc) + conn->local_tx_ctrl.conn_state_flags.peer_conn_abort = 1; + return rc; +} + +/* sndbuf consumer */ +static inline void smc_tx_advance_cursors(struct smc_connection *conn, + union smc_host_cursor *prod, + union smc_host_cursor *sent, + size_t len) +{ + smc_curs_add(conn->peer_rmbe_size, prod, len); + /* increased in recv tasklet smc_cdc_msg_rcv() */ + smp_mb__before_atomic(); + /* data in flight reduces usable snd_wnd */ + atomic_sub(len, &conn->peer_rmbe_space); + /* guarantee 0 <= peer_rmbe_space <= peer_rmbe_size */ + smp_mb__after_atomic(); + smc_curs_add(conn->sndbuf_size, sent, len); +} + +/* sndbuf consumer: prepare all necessary (src&dst) chunks of data transmit; + * usable snd_wnd as max transmit + */ +static int smc_tx_rdma_writes(struct smc_connection *conn) +{ + size_t src_off, src_len, dst_off, dst_len; /* current chunk values */ + size_t len, dst_len_sum, src_len_sum, dstchunk, srcchunk; + union smc_host_cursor sent, prep, prod, cons; + struct ib_sge sges[SMC_IB_MAX_SEND_SGE]; + struct smc_link_group *lgr = conn->lgr; + int to_send, rmbespace; + struct smc_link *link; + int num_sges; + int rc; + + /* source: sndbuf */ + smc_curs_write(&sent, smc_curs_read(&conn->tx_curs_sent, conn), conn); + smc_curs_write(&prep, smc_curs_read(&conn->tx_curs_prep, conn), conn); + /* cf. wmem_alloc - (snd_max - snd_una) */ + to_send = smc_curs_diff(conn->sndbuf_size, &sent, &prep); + if (to_send <= 0) + return 0; + + /* destination: RMBE */ + /* cf. snd_wnd */ + rmbespace = atomic_read(&conn->peer_rmbe_space); + if (rmbespace <= 0) + return 0; + smc_curs_write(&prod, + smc_curs_read(&conn->local_tx_ctrl.prod, conn), + conn); + smc_curs_write(&cons, + smc_curs_read(&conn->local_rx_ctrl.cons, conn), + conn); + + /* if usable snd_wnd closes ask peer to advertise once it opens again */ + conn->local_tx_ctrl.prod_flags.write_blocked = (to_send >= rmbespace); + /* cf. usable snd_wnd */ + len = min(to_send, rmbespace); + + /* initialize variables for first iteration of subsequent nested loop */ + link = &lgr->lnk[SMC_SINGLE_LINK]; + dst_off = prod.count; + if (prod.wrap == cons.wrap) { + /* the filled destination area is unwrapped, + * hence the available free destination space is wrapped + * and we need 2 destination chunks of sum len; start with 1st + * which is limited by what's available in sndbuf + */ + dst_len = min_t(size_t, + conn->peer_rmbe_size - prod.count, len); + } else { + /* the filled destination area is wrapped, + * hence the available free destination space is unwrapped + * and we need a single destination chunk of entire len + */ + dst_len = len; + } + dst_len_sum = dst_len; + src_off = sent.count; + /* dst_len determines the maximum src_len */ + if (sent.count + dst_len <= conn->sndbuf_size) { + /* unwrapped src case: single chunk of entire dst_len */ + src_len = dst_len; + } else { + /* wrapped src case: 2 chunks of sum dst_len; start with 1st: */ + src_len = conn->sndbuf_size - sent.count; + } + src_len_sum = src_len; + for (dstchunk = 0; dstchunk < 2; dstchunk++) { + num_sges = 0; + for (srcchunk = 0; srcchunk < 2; srcchunk++) { + sges[srcchunk].addr = + conn->sndbuf_desc->dma_addr[SMC_SINGLE_LINK] + + src_off; + sges[srcchunk].length = src_len; + sges[srcchunk].lkey = link->roce_pd->local_dma_lkey; + num_sges++; + src_off += src_len; + if (src_off >= conn->sndbuf_size) + src_off -= conn->sndbuf_size; + /* modulo in send ring */ + if (src_len_sum == dst_len) + break; /* either on 1st or 2nd iteration */ + /* prepare next (== 2nd) iteration */ + src_len = dst_len - src_len; /* remainder */ + src_len_sum += src_len; + } + rc = smc_tx_rdma_write(conn, dst_off, num_sges, sges); + if (rc) + return rc; + if (dst_len_sum == len) + break; /* either on 1st or 2nd iteration */ + /* prepare next (== 2nd) iteration */ + dst_off = 0; /* modulo offset in RMBE ring buffer */ + dst_len = len - dst_len; /* remainder */ + dst_len_sum += dst_len; + src_len = min_t(int, + dst_len, conn->sndbuf_size - sent.count); + src_len_sum = src_len; + } + + smc_tx_advance_cursors(conn, &prod, &sent, len); + /* update connection's cursors with advanced local cursors */ + smc_curs_write(&conn->local_tx_ctrl.prod, + smc_curs_read(&prod, conn), + conn); + /* dst: peer RMBE */ + smc_curs_write(&conn->tx_curs_sent, + smc_curs_read(&sent, conn), + conn); + /* src: local sndbuf */ + + return 0; +} + +/* Wakeup sndbuf consumers from any context (IRQ or process) + * since there is more data to transmit; usable snd_wnd as max transmit + */ +int smc_tx_sndbuf_nonempty(struct smc_connection *conn) +{ + struct smc_cdc_tx_pend *pend; + struct smc_wr_buf *wr_buf; + int rc; + + spin_lock_bh(&conn->send_lock); + rc = smc_cdc_get_free_slot(&conn->lgr->lnk[SMC_SINGLE_LINK], &wr_buf, + &pend); + if (rc < 0) { + if (rc == -EBUSY) { + rc = 0; + schedule_work(&conn->tx_work); + } + goto out_unlock; + } + + rc = smc_tx_rdma_writes(conn); + if (rc) { + smc_wr_tx_put_slot(&conn->lgr->lnk[SMC_SINGLE_LINK], + (struct smc_wr_tx_pend_priv *)pend); + goto out_unlock; + } + + rc = smc_cdc_msg_send(conn, wr_buf, pend); + +out_unlock: + spin_unlock_bh(&conn->send_lock); + return rc; +} + +/* Wakeup sndbuf consumers from process context + * since there is more data to transmit + */ +static void smc_tx_work(struct work_struct *work) +{ + struct smc_connection *conn = container_of(work, + struct smc_connection, + tx_work); + struct smc_sock *smc = container_of(conn, struct smc_sock, conn); + + lock_sock(&smc->sk); + smc_tx_sndbuf_nonempty(conn); + release_sock(&smc->sk); +} + +/***************************** send initialize *******************************/ + +/* Initialize send properties on connection establishment. NB: not __init! */ +void smc_tx_init(struct smc_sock *smc) +{ + smc->sk.sk_write_space = smc_tx_write_space; + INIT_WORK(&smc->conn.tx_work, smc_tx_work); + spin_lock_init(&smc->conn.send_lock); +} diff --git a/net/smc/smc_tx.h b/net/smc/smc_tx.h new file mode 100644 index 000000000000..931c66645624 --- /dev/null +++ b/net/smc/smc_tx.h @@ -0,0 +1,34 @@ +/* + * Shared Memory Communications over RDMA (SMC-R) and RoCE + * + * Manage send buffer + * + * Copyright IBM Corp. 2016 + * + * Author(s): Ursula Braun + */ + +#ifndef SMC_TX_H +#define SMC_TX_H + +#include +#include + +#include "smc.h" +#include "smc_cdc.h" + +static inline int smc_tx_prepared_sends(struct smc_connection *conn) +{ + union smc_host_cursor sent, prep; + + smc_curs_write(&sent, smc_curs_read(&conn->tx_curs_sent, conn), conn); + smc_curs_write(&prep, smc_curs_read(&conn->tx_curs_prep, conn), conn); + return smc_curs_diff(conn->sndbuf_size, &sent, &prep); +} + +void smc_tx_init(struct smc_sock *smc); +int smc_tx_sendmsg(struct smc_sock *smc, struct msghdr *msg, size_t len); +int smc_tx_sndbuf_nonempty(struct smc_connection *conn); +void smc_tx_sndbuf_nonfull(struct smc_sock *smc); + +#endif /* SMC_TX_H */