[2] xprtrdma: Fix XDRBUF_SPARSE_PAGES support

Commit Message

Chuck Lever Dec. 5, 2020, 6:21 p.m. UTC

Olga K. observed that rpcrdma_marsh_req() allocates sparse pages
only when it has determined that a Reply chunk is necessary. There
are plenty of cases where no Reply chunk is needed, but the
XDRBUF_SPARSE_PAGES flag is set. The result would be a crash in
rpcrdma_inline_fixup() when it tries to copy parts of the received
Reply into a missing page.

To avoid crashing, handle sparse page allocation up front.

Until XATTR support was added, this issue did not appear often
because the only SPARSE_PAGES consumer always expected a reply large
enough to always require a Reply chunk.

Reported-by: Olga Kornievskaia <[e-mail]>
Signed-off-by: Chuck Lever <[e-mail]>
Cc: <[e-mail]>
---
 net/sunrpc/xprtrdma/rpc_rdma.c |   42 +++++++++++++++++++++++++++++++---------
 1 file changed, 33 insertions(+), 9 deletions(-)

Changes since RFC:
- Ensure xdr_buf_pagecount() is defined in rpc_rdma.c
- noinline the sparse page allocator -- it's an uncommon path

Patch

diff --git a/net/sunrpc/xprtrdma/rpc_rdma.c b/net/sunrpc/xprtrdma/rpc_rdma.c
index 0f5120c7668f..3ce7dcce5fb9 100644
--- a/net/sunrpc/xprtrdma/rpc_rdma.c
+++ b/net/sunrpc/xprtrdma/rpc_rdma.c
@@ -49,6 +49,7 @@ 
 
 #include <linux/highmem.h>
 
+#include <linux/sunrpc/xdr.h>
 #include <linux/sunrpc/svc_rdma.h>
 
 #include "xprt_rdma.h"
@@ -179,6 +180,32 @@  rpcrdma_nonpayload_inline(const struct rpcrdma_xprt *r_xprt,
 		r_xprt->rx_ep->re_max_inline_recv;
 }
 
+/* ACL likes to be lazy in allocating pages. For TCP, these
+ * pages can be allocated during receive processing. Not true
+ * for RDMA, which must always provision receive buffers
+ * up front.
+ */
+static noinline int
+rpcrdma_alloc_sparse_pages(struct rpc_rqst *rqst)
+{
+	struct xdr_buf *xb = &rqst->rq_rcv_buf;
+	struct page **ppages;
+	int len;
+
+	len = xb->page_len;
+	ppages = xb->pages + xdr_buf_pagecount(xb);
+	while (len > 0) {
+		if (!*ppages)
+			*ppages = alloc_page(GFP_NOWAIT | __GFP_NOWARN);
+		if (!*ppages)
+			return -ENOBUFS;
+		ppages++;
+		len -= PAGE_SIZE;
+	}
+
+	return 0;
+}
+
 /* Split @vec on page boundaries into SGEs. FMR registers pages, not
  * a byte range. Other modes coalesce these SGEs into a single MR
  * when they can.
@@ -233,15 +260,6 @@  rpcrdma_convert_iovs(struct rpcrdma_xprt *r_xprt, struct xdr_buf *xdrbuf,
 	ppages = xdrbuf->pages + (xdrbuf->page_base >> PAGE_SHIFT);
 	page_base = offset_in_page(xdrbuf->page_base);
 	while (len) {
-		/* ACL likes to be lazy in allocating pages - ACLs
-		 * are small by default but can get huge.
-		 */
-		if (unlikely(xdrbuf->flags & XDRBUF_SPARSE_PAGES)) {
-			if (!*ppages)
-				*ppages = alloc_page(GFP_NOWAIT | __GFP_NOWARN);
-			if (!*ppages)
-				return -ENOBUFS;
-		}
 		seg->mr_page = *ppages;
 		seg->mr_offset = (char *)page_base;
 		seg->mr_len = min_t(u32, PAGE_SIZE - page_base, len);
@@ -867,6 +885,12 @@  rpcrdma_marshal_req(struct rpcrdma_xprt *r_xprt, struct rpc_rqst *rqst)
 	__be32 *p;
 	int ret;
 
+	if (unlikely(rqst->rq_rcv_buf.flags & XDRBUF_SPARSE_PAGES)) {
+		ret = rpcrdma_alloc_sparse_pages(rqst);
+		if (ret)
+			return ret;
+	}
+
 	rpcrdma_set_xdrlen(&req->rl_hdrbuf, 0);
 	xdr_init_encode(xdr, &req->rl_hdrbuf, rdmab_data(req->rl_rdmabuf),
 			rqst);