Return-Path: Received: from mail-qg0-f46.google.com ([209.85.192.46]:34946 "EHLO mail-qg0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751492AbbGMQbj (ORCPT ); Mon, 13 Jul 2015 12:31:39 -0400 Subject: [PATCH v2 12/14] xprtrdma: Fix large NFS SYMLINK calls From: Chuck Lever To: linux-rdma@vger.kernel.org, linux-nfs@vger.kernel.org Date: Mon, 13 Jul 2015 12:31:36 -0400 Message-ID: <20150713163136.17630.5572.stgit@manet.1015granger.net> In-Reply-To: <20150713160617.17630.97475.stgit@manet.1015granger.net> References: <20150713160617.17630.97475.stgit@manet.1015granger.net> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Sender: linux-nfs-owner@vger.kernel.org List-ID: Repair how rpcrdma_marshal_req() chooses which RDMA message type to use for large non-WRITE operations so that it picks RDMA_NOMSG in the correct situations, and sets up the marshaling logic to SEND only the RPC/RDMA header. Large NFSv2 SYMLINK requests now use RDMA_NOMSG calls. The Linux NFS server XDR decoder for NFSv2 SYMLINK does not handle having the pathname argument arrive in a separate buffer. The decoder could be fixed, but this is simpler and RDMA_NOMSG can be used in a variety of other situations. Ensure that the Linux client continues to use "RDMA_MSG + read list" when sending large NFSv3 SYMLINK requests, which is more efficient than using RDMA_NOMSG. Large NFSv4 CREATE(NF4LNK) requests are changed to use "RDMA_MSG + read list" just like NFSv3 (see Section 5 of RFC 5667). Before, these did not work at all. Signed-off-by: Chuck Lever --- fs/nfs/nfs3xdr.c | 1 + fs/nfs/nfs4xdr.c | 4 +++- net/sunrpc/xprtrdma/rpc_rdma.c | 25 ++++++++++++++++--------- 3 files changed, 20 insertions(+), 10 deletions(-) diff --git a/fs/nfs/nfs3xdr.c b/fs/nfs/nfs3xdr.c index 9b04c2e..267126d 100644 --- a/fs/nfs/nfs3xdr.c +++ b/fs/nfs/nfs3xdr.c @@ -1103,6 +1103,7 @@ static void nfs3_xdr_enc_symlink3args(struct rpc_rqst *req, { encode_diropargs3(xdr, args->fromfh, args->fromname, args->fromlen); encode_symlinkdata3(xdr, args); + xdr->buf->flags |= XDRBUF_WRITE; } /* diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c index 558cd65d..c42459e 100644 --- a/fs/nfs/nfs4xdr.c +++ b/fs/nfs/nfs4xdr.c @@ -1154,7 +1154,9 @@ static void encode_create(struct xdr_stream *xdr, const struct nfs4_create_arg * case NF4LNK: p = reserve_space(xdr, 4); *p = cpu_to_be32(create->u.symlink.len); - xdr_write_pages(xdr, create->u.symlink.pages, 0, create->u.symlink.len); + xdr_write_pages(xdr, create->u.symlink.pages, 0, + create->u.symlink.len); + xdr->buf->flags |= XDRBUF_WRITE; break; case NF4BLK: case NF4CHR: diff --git a/net/sunrpc/xprtrdma/rpc_rdma.c b/net/sunrpc/xprtrdma/rpc_rdma.c index 1dd48f2..2721586 100644 --- a/net/sunrpc/xprtrdma/rpc_rdma.c +++ b/net/sunrpc/xprtrdma/rpc_rdma.c @@ -475,21 +475,24 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) * * o If the total request is under the inline threshold, all ops * are sent as inline. - * o Large non-write ops are sent with the entire message as a - * single read chunk (protocol 0-position special case). * o Large write ops transmit data as read chunk(s), header as * inline. + * o Large non-write ops are sent with the entire message as a + * single read chunk (protocol 0-position special case). * - * Note: the NFS code sending down multiple argument segments - * implies the op is a write. - * TBD check NFSv4 setacl + * This assumes that the upper layer does not present a request + * that both has a data payload, and whose non-data arguments + * by themselves are larger than the inline threshold. */ - if (rpcrdma_args_inline(rqst)) + if (rpcrdma_args_inline(rqst)) { rtype = rpcrdma_noch; - else if (rqst->rq_snd_buf.page_len == 0) - rtype = rpcrdma_areadch; - else + } else if (rqst->rq_snd_buf.flags & XDRBUF_WRITE) { rtype = rpcrdma_readch; + } else { + headerp->rm_type = htonl(RDMA_NOMSG); + rtype = rpcrdma_areadch; + rpclen = 0; + } /* The following simplification is not true forever */ if (rtype != rpcrdma_noch && wtype == rpcrdma_replych) @@ -546,6 +549,10 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) req->rl_send_iov[0].length = hdrlen; req->rl_send_iov[0].lkey = rdmab_lkey(req->rl_rdmabuf); + req->rl_niovs = 1; + if (rtype == rpcrdma_areadch) + return 0; + req->rl_send_iov[1].addr = rdmab_addr(req->rl_sendbuf); req->rl_send_iov[1].length = rpclen; req->rl_send_iov[1].lkey = rdmab_lkey(req->rl_sendbuf);