xprtrdma: Pull up sometimes

On some platforms, DMA mapping part of a page is more costly than
copying bytes. Restore the pull-up code and use that when we
think it's going to be faster. The heuristic for now is to pull-up
when the size of the RPC message body fits in the buffer underlying
the head iovec.

Indeed, not involving the I/O MMU can help the RPC/RDMA transport
scale better for tiny I/Os across more RDMA devices. This is because
interaction with the I/O MMU is eliminated, as is handling a Send
completion, for each of these small I/Os. Without the explicit
unmapping, the NIC no longer needs to do a costly internal TLB shoot
down for buffers that are just a handful of bytes.

Signed-off-by: Chuck Lever <[e-mail]>
Signed-off-by: Anna Schumaker <[e-mail]>

include/trace/events/rpcrdma.h

···
 DEFINE_WRCH_EVENT(reply);
 
 TRACE_DEFINE_ENUM(rpcrdma_noch);
+TRACE_DEFINE_ENUM(rpcrdma_noch_pullup);
+TRACE_DEFINE_ENUM(rpcrdma_noch_mapped);
 TRACE_DEFINE_ENUM(rpcrdma_readch);
 TRACE_DEFINE_ENUM(rpcrdma_areadch);
 TRACE_DEFINE_ENUM(rpcrdma_writech);
···
 #define xprtrdma_show_chunktype(x)					\
 		__print_symbolic(x,					\
 				{ rpcrdma_noch, "inline" },		\
+				{ rpcrdma_noch_pullup, "pullup" },	\
+				{ rpcrdma_noch_mapped, "mapped" },	\
 				{ rpcrdma_readch, "read list" },	\
 				{ rpcrdma_areadch, "*read list" },	\
 				{ rpcrdma_writech, "write list" },	\

net/sunrpc/xprtrdma/backchannel.c

···
 	*p = xdr_zero;
 
 	if (rpcrdma_prepare_send_sges(r_xprt, req, RPCRDMA_HDRLEN_MIN,
-				      &rqst->rq_snd_buf, rpcrdma_noch))
+				      &rqst->rq_snd_buf, rpcrdma_noch_pullup))
 		return -EIO;
 
 	trace_xprtrdma_cb_reply(rqst);

net/sunrpc/xprtrdma/rpc_rdma.c

···
 	unsigned int pos;
 	int nsegs;
 
-	if (rtype == rpcrdma_noch)
+	if (rtype == rpcrdma_noch_pullup || rtype == rpcrdma_noch_mapped)
 		goto done;
 
 	pos = rqst->rq_snd_buf.head[0].iov_len;
···
 	return false;
 }
 
+/* Copy the tail to the end of the head buffer.
+ */
+static void rpcrdma_pullup_tail_iov(struct rpcrdma_xprt *r_xprt,
+				    struct rpcrdma_req *req,
+				    struct xdr_buf *xdr)
+{
+	unsigned char *dst;
+
+	dst = (unsigned char *)xdr->head[0].iov_base;
+	dst += xdr->head[0].iov_len + xdr->page_len;
+	memmove(dst, xdr->tail[0].iov_base, xdr->tail[0].iov_len);
+	r_xprt->rx_stats.pullup_copy_count += xdr->tail[0].iov_len;
+}
+
+/* Copy pagelist content into the head buffer.
+ */
+static void rpcrdma_pullup_pagelist(struct rpcrdma_xprt *r_xprt,
+				    struct rpcrdma_req *req,
+				    struct xdr_buf *xdr)
+{
+	unsigned int len, page_base, remaining;
+	struct page **ppages;
+	unsigned char *src, *dst;
+
+	dst = (unsigned char *)xdr->head[0].iov_base;
+	dst += xdr->head[0].iov_len;
+	ppages = xdr->pages + (xdr->page_base >> PAGE_SHIFT);
+	page_base = offset_in_page(xdr->page_base);
+	remaining = xdr->page_len;
+	while (remaining) {
+		src = page_address(*ppages);
+		src += page_base;
+		len = min_t(unsigned int, PAGE_SIZE - page_base, remaining);
+		memcpy(dst, src, len);
+		r_xprt->rx_stats.pullup_copy_count += len;
+
+		ppages++;
+		dst += len;
+		remaining -= len;
+		page_base = 0;
+	}
+}
+
+/* Copy the contents of @xdr into @rl_sendbuf and DMA sync it.
+ * When the head, pagelist, and tail are small, a pull-up copy
+ * is considerably less costly than DMA mapping the components
+ * of @xdr.
+ *
+ * Assumptions:
+ *  - the caller has already verified that the total length
+ *    of the RPC Call body will fit into @rl_sendbuf.
+ */
+static bool rpcrdma_prepare_noch_pullup(struct rpcrdma_xprt *r_xprt,
+					struct rpcrdma_req *req,
+					struct xdr_buf *xdr)
+{
+	if (unlikely(xdr->tail[0].iov_len))
+		rpcrdma_pullup_tail_iov(r_xprt, req, xdr);
+
+	if (unlikely(xdr->page_len))
+		rpcrdma_pullup_pagelist(r_xprt, req, xdr);
+
+	/* The whole RPC message resides in the head iovec now */
+	return rpcrdma_prepare_head_iov(r_xprt, req, xdr->len);
+}
+
 static bool rpcrdma_prepare_noch_mapped(struct rpcrdma_xprt *r_xprt,
 					struct rpcrdma_req *req,
 					struct xdr_buf *xdr)
···
 		goto out_unmap;
 
 	switch (rtype) {
-	case rpcrdma_noch:
+	case rpcrdma_noch_pullup:
+		if (!rpcrdma_prepare_noch_pullup(r_xprt, req, xdr))
+			goto out_unmap;
+		break;
+	case rpcrdma_noch_mapped:
 		if (!rpcrdma_prepare_noch_mapped(r_xprt, req, xdr))
 			goto out_unmap;
 		break;
···
 	struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
 	struct xdr_stream *xdr = &req->rl_stream;
 	enum rpcrdma_chunktype rtype, wtype;
+	struct xdr_buf *buf = &rqst->rq_snd_buf;
 	bool ddp_allowed;
 	__be32 *p;
 	int ret;
···
 	 */
 	if (rpcrdma_args_inline(r_xprt, rqst)) {
 		*p++ = rdma_msg;
-		rtype = rpcrdma_noch;
-	} else if (ddp_allowed && rqst->rq_snd_buf.flags & XDRBUF_WRITE) {
+		rtype = buf->len < rdmab_length(req->rl_sendbuf) ?
+			rpcrdma_noch_pullup : rpcrdma_noch_mapped;
+	} else if (ddp_allowed && buf->flags & XDRBUF_WRITE) {
 		*p++ = rdma_msg;
 		rtype = rpcrdma_readch;
 	} else {
···
 		goto out_err;
 
 	ret = rpcrdma_prepare_send_sges(r_xprt, req, req->rl_hdrbuf.len,
-					&rqst->rq_snd_buf, rtype);
+					buf, rtype);
 	if (ret)
 		goto out_err;
 

net/sunrpc/xprtrdma/verbs.c

···
 	for (i = 0; i < buf->rb_max_requests; i++) {
 		struct rpcrdma_req *req;
 
-		req = rpcrdma_req_create(r_xprt, RPCRDMA_V1_DEF_INLINE_SIZE,
+		req = rpcrdma_req_create(r_xprt, RPCRDMA_V1_DEF_INLINE_SIZE * 2,
 					 GFP_KERNEL);
 		if (!req)
 			goto out;

net/sunrpc/xprtrdma/xprt_rdma.h

···
 
 enum rpcrdma_chunktype {
 	rpcrdma_noch = 0,
+	rpcrdma_noch_pullup,
+	rpcrdma_noch_mapped,
 	rpcrdma_readch,
 	rpcrdma_areadch,
 	rpcrdma_writech,