libceph: optionally use bounce buffer on recv path in crc mode

Both msgr1 and msgr2 in crc mode are zero copy in the sense that
message data is read from the socket directly into the destination
buffer.  We assume that the destination buffer is stable (i.e. remains
unchanged while it is being read to) though.  Otherwise, CRC errors
ensue:

  libceph: read_partial_message 0000000048edf8ad data crc 1063286393 != exp. 228122706
  libceph: osd1 (1)192.168.122.1:6843 bad crc/signature

  libceph: bad data crc, calculated 57958023, expected 1805382778
  libceph: osd2 (2)192.168.122.1:6876 integrity error, bad crc

Introduce rxbounce option to enable use of a bounce buffer when
receiving message data.  In particular this is needed if a mapped
image is a Windows VM disk, passed to QEMU.  Windows has a system-wide
"dummy" page that may be mapped into the destination buffer (potentially
more than once into the same buffer) by the Windows Memory Manager in
an effort to generate a single large I/O [1][2].  QEMU makes a point of
preserving overlap relationships when cloning I/O vectors, so krbd gets
exposed to this behaviour.

[1] "What Is Really in That MDL?"
    https://docs.microsoft.com/en-us/previous-versions/windows/hardware/design/dn614012(v=vs.85)
[2] https://blogs.msmvps.com/kernelmustard/2005/05/04/dummy-pages/

URL: https://bugzilla.redhat.com/show_bug.cgi?id=1973317
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>

1 files changed, 44 insertions, 14 deletions

diff --git a/net/ceph/messenger_v2.c b/net/ceph/messenger_v2.c
index 2ea00489e691..c81379f93ad5 100644
--- a/net/ceph/messenger_v2.c
+++ b/net/ceph/messenger_v2.c
@@ -1753,7 +1753,7 @@ static int prepare_read_control_remainder(struct ceph_connection *con)
 	return 0;
 }
 
-static void prepare_read_data(struct ceph_connection *con)
+static int prepare_read_data(struct ceph_connection *con)
 {
 	struct bio_vec bv;
 
@@ -1762,23 +1762,55 @@ static void prepare_read_data(struct ceph_connection *con)
 				  data_len(con->in_msg));
 
 	get_bvec_at(&con->v2.in_cursor, &bv);
-	set_in_bvec(con, &bv);
+	if (ceph_test_opt(from_msgr(con->msgr), RXBOUNCE)) {
+		if (unlikely(!con->bounce_page)) {
+			con->bounce_page = alloc_page(GFP_NOIO);
+			if (!con->bounce_page) {
+				pr_err("failed to allocate bounce page\n");
+				return -ENOMEM;
+			}
+		}
+
+		bv.bv_page = con->bounce_page;
+		bv.bv_offset = 0;
+		set_in_bvec(con, &bv);
+	} else {
+		set_in_bvec(con, &bv);
+	}
 	con->v2.in_state = IN_S_PREPARE_READ_DATA_CONT;
+	return 0;
 }
 
 static void prepare_read_data_cont(struct ceph_connection *con)
 {
 	struct bio_vec bv;
 
-	con->in_data_crc = ceph_crc32c_page(con->in_data_crc,
-					    con->v2.in_bvec.bv_page,
-					    con->v2.in_bvec.bv_offset,
-					    con->v2.in_bvec.bv_len);
+	if (ceph_test_opt(from_msgr(con->msgr), RXBOUNCE)) {
+		con->in_data_crc = crc32c(con->in_data_crc,
+					  page_address(con->bounce_page),
+					  con->v2.in_bvec.bv_len);
+
+		get_bvec_at(&con->v2.in_cursor, &bv);
+		memcpy_to_page(bv.bv_page, bv.bv_offset,
+			       page_address(con->bounce_page),
+			       con->v2.in_bvec.bv_len);
+	} else {
+		con->in_data_crc = ceph_crc32c_page(con->in_data_crc,
+						    con->v2.in_bvec.bv_page,
+						    con->v2.in_bvec.bv_offset,
+						    con->v2.in_bvec.bv_len);
+	}
 
 	ceph_msg_data_advance(&con->v2.in_cursor, con->v2.in_bvec.bv_len);
 	if (con->v2.in_cursor.total_resid) {
 		get_bvec_at(&con->v2.in_cursor, &bv);
-		set_in_bvec(con, &bv);
+		if (ceph_test_opt(from_msgr(con->msgr), RXBOUNCE)) {
+			bv.bv_page = con->bounce_page;
+			bv.bv_offset = 0;
+			set_in_bvec(con, &bv);
+		} else {
+			set_in_bvec(con, &bv);
+		}
 		WARN_ON(con->v2.in_state != IN_S_PREPARE_READ_DATA_CONT);
 		return;
 	}
@@ -1791,14 +1823,13 @@ static void prepare_read_data_cont(struct ceph_connection *con)
 	con->v2.in_state = IN_S_HANDLE_EPILOGUE;
 }
 
-static void prepare_read_tail_plain(struct ceph_connection *con)
+static int prepare_read_tail_plain(struct ceph_connection *con)
 {
 	struct ceph_msg *msg = con->in_msg;
 
 	if (!front_len(msg) && !middle_len(msg)) {
 		WARN_ON(!data_len(msg));
-		prepare_read_data(con);
-		return;
+		return prepare_read_data(con);
 	}
 
 	reset_in_kvecs(con);
@@ -1817,6 +1848,7 @@ static void prepare_read_tail_plain(struct ceph_connection *con)
 		add_in_kvec(con, con->v2.in_buf, CEPH_EPILOGUE_PLAIN_LEN);
 		con->v2.in_state = IN_S_HANDLE_EPILOGUE;
 	}
+	return 0;
 }
 
 static void prepare_read_enc_page(struct ceph_connection *con)
@@ -2699,8 +2731,7 @@ static int __handle_control(struct ceph_connection *con, void *p)
 	if (con_secure(con))
 		return prepare_read_tail_secure(con);
 
-	prepare_read_tail_plain(con);
-	return 0;
+	return prepare_read_tail_plain(con);
 }
 
 static int handle_preamble(struct ceph_connection *con)
@@ -2856,8 +2887,7 @@ static int populate_in_iter(struct ceph_connection *con)
 			ret = handle_control_remainder(con);
 			break;
 		case IN_S_PREPARE_READ_DATA:
-			prepare_read_data(con);
-			ret = 0;
+			ret = prepare_read_data(con);
 			break;
 		case IN_S_PREPARE_READ_DATA_CONT:
 			prepare_read_data_cont(con);