diff options
Diffstat (limited to 'net/sunrpc/xprtrdma')
-rw-r--r-- | net/sunrpc/xprtrdma/svc_rdma_marshal.c | 32 | ||||
-rw-r--r-- | net/sunrpc/xprtrdma/svc_rdma_recvfrom.c | 34 | ||||
-rw-r--r-- | net/sunrpc/xprtrdma/svc_rdma_sendto.c | 28 | ||||
-rw-r--r-- | net/sunrpc/xprtrdma/svc_rdma_transport.c | 17 |
4 files changed, 58 insertions, 53 deletions
diff --git a/net/sunrpc/xprtrdma/svc_rdma_marshal.c b/net/sunrpc/xprtrdma/svc_rdma_marshal.c index 765bca47c74d..0ba9887f3e22 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_marshal.c +++ b/net/sunrpc/xprtrdma/svc_rdma_marshal.c @@ -145,19 +145,32 @@ static __be32 *decode_reply_array(__be32 *va, __be32 *vaend) return (__be32 *)&ary->wc_array[nchunks]; } -int svc_rdma_xdr_decode_req(struct rpcrdma_msg *rmsgp, struct svc_rqst *rqstp) +/** + * svc_rdma_xdr_decode_req - Parse incoming RPC-over-RDMA header + * @rq_arg: Receive buffer + * + * On entry, xdr->head[0].iov_base points to first byte in the + * RPC-over-RDMA header. + * + * On successful exit, head[0] points to first byte past the + * RPC-over-RDMA header. For RDMA_MSG, this is the RPC message. + * The length of the RPC-over-RDMA header is returned. + */ +int svc_rdma_xdr_decode_req(struct xdr_buf *rq_arg) { + struct rpcrdma_msg *rmsgp; __be32 *va, *vaend; unsigned int len; u32 hdr_len; /* Verify that there's enough bytes for header + something */ - if (rqstp->rq_arg.len <= RPCRDMA_HDRLEN_ERR) { + if (rq_arg->len <= RPCRDMA_HDRLEN_ERR) { dprintk("svcrdma: header too short = %d\n", - rqstp->rq_arg.len); + rq_arg->len); return -EINVAL; } + rmsgp = (struct rpcrdma_msg *)rq_arg->head[0].iov_base; if (rmsgp->rm_vers != rpcrdma_version) { dprintk("%s: bad version %u\n", __func__, be32_to_cpu(rmsgp->rm_vers)); @@ -189,10 +202,10 @@ int svc_rdma_xdr_decode_req(struct rpcrdma_msg *rmsgp, struct svc_rqst *rqstp) be32_to_cpu(rmsgp->rm_body.rm_padded.rm_thresh); va = &rmsgp->rm_body.rm_padded.rm_pempty[4]; - rqstp->rq_arg.head[0].iov_base = va; + rq_arg->head[0].iov_base = va; len = (u32)((unsigned long)va - (unsigned long)rmsgp); - rqstp->rq_arg.head[0].iov_len -= len; - if (len > rqstp->rq_arg.len) + rq_arg->head[0].iov_len -= len; + if (len > rq_arg->len) return -EINVAL; return len; default: @@ -205,7 +218,7 @@ int svc_rdma_xdr_decode_req(struct rpcrdma_msg *rmsgp, struct svc_rqst *rqstp) * chunk list and a reply chunk list. */ va = &rmsgp->rm_body.rm_chunks[0]; - vaend = (__be32 *)((unsigned long)rmsgp + rqstp->rq_arg.len); + vaend = (__be32 *)((unsigned long)rmsgp + rq_arg->len); va = decode_read_list(va, vaend); if (!va) { dprintk("svcrdma: failed to decode read list\n"); @@ -222,10 +235,9 @@ int svc_rdma_xdr_decode_req(struct rpcrdma_msg *rmsgp, struct svc_rqst *rqstp) return -EINVAL; } - rqstp->rq_arg.head[0].iov_base = va; + rq_arg->head[0].iov_base = va; hdr_len = (unsigned long)va - (unsigned long)rmsgp; - rqstp->rq_arg.head[0].iov_len -= hdr_len; - + rq_arg->head[0].iov_len -= hdr_len; return hdr_len; } diff --git a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c index fbe7444e7de6..2c25606f2561 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c +++ b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c @@ -447,10 +447,8 @@ static int rdma_read_chunks(struct svcxprt_rdma *xprt, head->arg.len = rqstp->rq_arg.len; head->arg.buflen = rqstp->rq_arg.buflen; - ch = (struct rpcrdma_read_chunk *)&rmsgp->rm_body.rm_chunks[0]; - position = be32_to_cpu(ch->rc_position); - /* RDMA_NOMSG: RDMA READ data should land just after RDMA RECV data */ + position = be32_to_cpu(ch->rc_position); if (position == 0) { head->arg.pages = &head->pages[0]; page_offset = head->byte_len; @@ -488,7 +486,7 @@ static int rdma_read_chunks(struct svcxprt_rdma *xprt, if (page_offset & 3) { u32 pad = 4 - (page_offset & 3); - head->arg.page_len += pad; + head->arg.tail[0].iov_len += pad; head->arg.len += pad; head->arg.buflen += pad; page_offset += pad; @@ -510,11 +508,10 @@ static int rdma_read_chunks(struct svcxprt_rdma *xprt, return ret; } -static int rdma_read_complete(struct svc_rqst *rqstp, - struct svc_rdma_op_ctxt *head) +static void rdma_read_complete(struct svc_rqst *rqstp, + struct svc_rdma_op_ctxt *head) { int page_no; - int ret; /* Copy RPC pages */ for (page_no = 0; page_no < head->count; page_no++) { @@ -550,23 +547,6 @@ static int rdma_read_complete(struct svc_rqst *rqstp, rqstp->rq_arg.tail[0] = head->arg.tail[0]; rqstp->rq_arg.len = head->arg.len; rqstp->rq_arg.buflen = head->arg.buflen; - - /* Free the context */ - svc_rdma_put_context(head, 0); - - /* XXX: What should this be? */ - rqstp->rq_prot = IPPROTO_MAX; - svc_xprt_copy_addrs(rqstp, rqstp->rq_xprt); - - ret = rqstp->rq_arg.head[0].iov_len - + rqstp->rq_arg.page_len - + rqstp->rq_arg.tail[0].iov_len; - dprintk("svcrdma: deferred read ret=%d, rq_arg.len=%u, " - "rq_arg.head[0].iov_base=%p, rq_arg.head[0].iov_len=%zu\n", - ret, rqstp->rq_arg.len, rqstp->rq_arg.head[0].iov_base, - rqstp->rq_arg.head[0].iov_len); - - return ret; } /* By convention, backchannel calls arrive via rdma_msg type @@ -624,7 +604,8 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp) dto_q); list_del_init(&ctxt->dto_q); spin_unlock_bh(&rdma_xprt->sc_rq_dto_lock); - return rdma_read_complete(rqstp, ctxt); + rdma_read_complete(rqstp, ctxt); + goto complete; } else if (!list_empty(&rdma_xprt->sc_rq_dto_q)) { ctxt = list_entry(rdma_xprt->sc_rq_dto_q.next, struct svc_rdma_op_ctxt, @@ -655,7 +636,7 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp) /* Decode the RDMA header. */ rmsgp = (struct rpcrdma_msg *)rqstp->rq_arg.head[0].iov_base; - ret = svc_rdma_xdr_decode_req(rmsgp, rqstp); + ret = svc_rdma_xdr_decode_req(&rqstp->rq_arg); if (ret < 0) goto out_err; if (ret == 0) @@ -682,6 +663,7 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp) return 0; } +complete: ret = rqstp->rq_arg.head[0].iov_len + rqstp->rq_arg.page_len + rqstp->rq_arg.tail[0].iov_len; diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c index 4f1b1c4f45f9..54d533300620 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c @@ -463,25 +463,21 @@ static int send_reply(struct svcxprt_rdma *rdma, struct svc_rqst *rqstp, struct page *page, struct rpcrdma_msg *rdma_resp, - struct svc_rdma_op_ctxt *ctxt, struct svc_rdma_req_map *vec, int byte_count) { + struct svc_rdma_op_ctxt *ctxt; struct ib_send_wr send_wr; u32 xdr_off; int sge_no; int sge_bytes; int page_no; int pages; - int ret; - - ret = svc_rdma_repost_recv(rdma, GFP_KERNEL); - if (ret) { - svc_rdma_put_context(ctxt, 0); - return -ENOTCONN; - } + int ret = -EIO; /* Prepare the context */ + ctxt = svc_rdma_get_context(rdma); + ctxt->direction = DMA_TO_DEVICE; ctxt->pages[0] = page; ctxt->count = 1; @@ -565,8 +561,7 @@ static int send_reply(struct svcxprt_rdma *rdma, err: svc_rdma_unmap_dma(ctxt); svc_rdma_put_context(ctxt, 1); - pr_err("svcrdma: failed to send reply, rc=%d\n", ret); - return -EIO; + return ret; } void svc_rdma_prep_reply_hdr(struct svc_rqst *rqstp) @@ -585,7 +580,6 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) int ret; int inline_bytes; struct page *res_page; - struct svc_rdma_op_ctxt *ctxt; struct svc_rdma_req_map *vec; dprintk("svcrdma: sending response for rqstp=%p\n", rqstp); @@ -598,8 +592,6 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) rp_ary = svc_rdma_get_reply_array(rdma_argp, wr_ary); /* Build an req vec for the XDR */ - ctxt = svc_rdma_get_context(rdma); - ctxt->direction = DMA_TO_DEVICE; vec = svc_rdma_get_req_map(rdma); ret = svc_rdma_map_xdr(rdma, &rqstp->rq_res, vec, wr_ary != NULL); if (ret) @@ -635,7 +627,12 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) inline_bytes -= ret; } - ret = send_reply(rdma, rqstp, res_page, rdma_resp, ctxt, vec, + /* Post a fresh Receive buffer _before_ sending the reply */ + ret = svc_rdma_post_recv(rdma, GFP_KERNEL); + if (ret) + goto err1; + + ret = send_reply(rdma, rqstp, res_page, rdma_resp, vec, inline_bytes); if (ret < 0) goto err1; @@ -648,7 +645,8 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) put_page(res_page); err0: svc_rdma_put_req_map(rdma, vec); - svc_rdma_put_context(ctxt, 0); + pr_err("svcrdma: Could not send reply, err=%d. Closing transport.\n", + ret); set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags); return -ENOTCONN; } diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c index 90668969d559..dd9440137834 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_transport.c +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c @@ -789,7 +789,7 @@ static struct svc_xprt *svc_rdma_create(struct svc_serv *serv, int ret; dprintk("svcrdma: Creating RDMA socket\n"); - if (sa->sa_family != AF_INET) { + if ((sa->sa_family != AF_INET) && (sa->sa_family != AF_INET6)) { dprintk("svcrdma: Address family %d is not supported.\n", sa->sa_family); return ERR_PTR(-EAFNOSUPPORT); } @@ -805,6 +805,16 @@ static struct svc_xprt *svc_rdma_create(struct svc_serv *serv, goto err0; } + /* Allow both IPv4 and IPv6 sockets to bind a single port + * at the same time. + */ +#if IS_ENABLED(CONFIG_IPV6) + ret = rdma_set_afonly(listen_id, 1); + if (ret) { + dprintk("svcrdma: rdma_set_afonly failed = %d\n", ret); + goto err1; + } +#endif ret = rdma_bind_addr(listen_id, sa); if (ret) { dprintk("svcrdma: rdma_bind_addr failed = %d\n", ret); @@ -1073,7 +1083,7 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt) newxprt->sc_dev_caps |= SVCRDMA_DEVCAP_READ_W_INV; /* Post receive buffers */ - for (i = 0; i < newxprt->sc_rq_depth; i++) { + for (i = 0; i < newxprt->sc_max_requests; i++) { ret = svc_rdma_post_recv(newxprt, GFP_KERNEL); if (ret) { dprintk("svcrdma: failure posting receive buffers\n"); @@ -1170,6 +1180,9 @@ static void __svc_rdma_free(struct work_struct *work) dprintk("svcrdma: %s(%p)\n", __func__, rdma); + if (rdma->sc_qp && !IS_ERR(rdma->sc_qp)) + ib_drain_qp(rdma->sc_qp); + /* We should only be called from kref_put */ if (atomic_read(&xprt->xpt_ref.refcount) != 0) pr_err("svcrdma: sc_xprt still in use? (%d)\n", |