| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <1386264972-5399-26-git-send-email-luis.henriques@canonical.com>
Date: Thu, 5 Dec 2013 17:35:07 +0000
From: Luis Henriques <luis.henriques@...onical.com>
To: linux-kernel@...r.kernel.org, stable@...r.kernel.org,
kernel-team@...ts.ubuntu.com
Cc: Trond Myklebust <Trond.Myklebust@...app.com>,
Luis Henriques <luis.henriques@...onical.com>
Subject: [PATCH 3.5 25/90] SUNRPC: Fix a data corruption issue when retransmitting RPC calls
3.5.7.27 -stable review patch. If anyone has any objections, please let me know.
------------------
From: Trond Myklebust <Trond.Myklebust@...app.com>
commit a6b31d18b02ff9d7915c5898c9b5ca41a798cd73 upstream.
The following scenario can cause silent data corruption when doing
NFS writes. It has mainly been observed when doing database writes
using O_DIRECT.
1) The RPC client uses sendpage() to do zero-copy of the page data.
2) Due to networking issues, the reply from the server is delayed,
and so the RPC client times out.
3) The client issues a second sendpage of the page data as part of
an RPC call retransmission.
4) The reply to the first transmission arrives from the server
_before_ the client hardware has emptied the TCP socket send
buffer.
5) After processing the reply, the RPC state machine rules that
the call to be done, and triggers the completion callbacks.
6) The application notices the RPC call is done, and reuses the
pages to store something else (e.g. a new write).
7) The client NIC drains the TCP socket send buffer. Since the
page data has now changed, it reads a corrupted version of the
initial RPC call, and puts it on the wire.
This patch fixes the problem in the following manner:
The ordering guarantees of TCP ensure that when the server sends a
reply, then we know that the _first_ transmission has completed. Using
zero-copy in that situation is therefore safe.
If a time out occurs, we then send the retransmission using sendmsg()
(i.e. no zero-copy), We then know that the socket contains a full copy of
the data, and so it will retransmit a faithful reproduction even if the
RPC call completes, and the application reuses the O_DIRECT buffer in
the meantime.
Signed-off-by: Trond Myklebust <Trond.Myklebust@...app.com>
Signed-off-by: Luis Henriques <luis.henriques@...onical.com>
---
net/sunrpc/xprtsock.c | 28 +++++++++++++++++++++-------
1 file changed, 21 insertions(+), 7 deletions(-)
diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 79064471..31f981d 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -390,8 +390,10 @@ static int xs_send_kvec(struct socket *sock, struct sockaddr *addr, int addrlen,
return kernel_sendmsg(sock, &msg, NULL, 0, 0);
}
-static int xs_send_pagedata(struct socket *sock, struct xdr_buf *xdr, unsigned int base, int more)
+static int xs_send_pagedata(struct socket *sock, struct xdr_buf *xdr, unsigned int base, int more, bool zerocopy)
{
+ ssize_t (*do_sendpage)(struct socket *sock, struct page *page,
+ int offset, size_t size, int flags);
struct page **ppage;
unsigned int remainder;
int err, sent = 0;
@@ -400,6 +402,9 @@ static int xs_send_pagedata(struct socket *sock, struct xdr_buf *xdr, unsigned i
base += xdr->page_base;
ppage = xdr->pages + (base >> PAGE_SHIFT);
base &= ~PAGE_MASK;
+ do_sendpage = sock->ops->sendpage;
+ if (!zerocopy)
+ do_sendpage = sock_no_sendpage;
for(;;) {
unsigned int len = min_t(unsigned int, PAGE_SIZE - base, remainder);
int flags = XS_SENDMSG_FLAGS;
@@ -407,7 +412,7 @@ static int xs_send_pagedata(struct socket *sock, struct xdr_buf *xdr, unsigned i
remainder -= len;
if (remainder != 0 || more)
flags |= MSG_MORE;
- err = sock->ops->sendpage(sock, *ppage, base, len, flags);
+ err = do_sendpage(sock, *ppage, base, len, flags);
if (remainder == 0 || err != len)
break;
sent += err;
@@ -428,9 +433,10 @@ static int xs_send_pagedata(struct socket *sock, struct xdr_buf *xdr, unsigned i
* @addrlen: UDP only -- length of destination address
* @xdr: buffer containing this request
* @base: starting position in the buffer
+ * @zerocopy: true if it is safe to use sendpage()
*
*/
-static int xs_sendpages(struct socket *sock, struct sockaddr *addr, int addrlen, struct xdr_buf *xdr, unsigned int base)
+static int xs_sendpages(struct socket *sock, struct sockaddr *addr, int addrlen, struct xdr_buf *xdr, unsigned int base, bool zerocopy)
{
unsigned int remainder = xdr->len - base;
int err, sent = 0;
@@ -458,7 +464,7 @@ static int xs_sendpages(struct socket *sock, struct sockaddr *addr, int addrlen,
if (base < xdr->page_len) {
unsigned int len = xdr->page_len - base;
remainder -= len;
- err = xs_send_pagedata(sock, xdr, base, remainder != 0);
+ err = xs_send_pagedata(sock, xdr, base, remainder != 0, zerocopy);
if (remainder == 0 || err != len)
goto out;
sent += err;
@@ -561,7 +567,7 @@ static int xs_local_send_request(struct rpc_task *task)
req->rq_svec->iov_base, req->rq_svec->iov_len);
status = xs_sendpages(transport->sock, NULL, 0,
- xdr, req->rq_bytes_sent);
+ xdr, req->rq_bytes_sent, true);
dprintk("RPC: %s(%u) = %d\n",
__func__, xdr->len - req->rq_bytes_sent, status);
if (likely(status >= 0)) {
@@ -617,7 +623,7 @@ static int xs_udp_send_request(struct rpc_task *task)
status = xs_sendpages(transport->sock,
xs_addr(xprt),
xprt->addrlen, xdr,
- req->rq_bytes_sent);
+ req->rq_bytes_sent, true);
dprintk("RPC: xs_udp_send_request(%u) = %d\n",
xdr->len - req->rq_bytes_sent, status);
@@ -688,6 +694,7 @@ static int xs_tcp_send_request(struct rpc_task *task)
struct rpc_xprt *xprt = req->rq_xprt;
struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
struct xdr_buf *xdr = &req->rq_snd_buf;
+ bool zerocopy = true;
int status;
xs_encode_stream_record_marker(&req->rq_snd_buf);
@@ -695,13 +702,20 @@ static int xs_tcp_send_request(struct rpc_task *task)
xs_pktdump("packet data:",
req->rq_svec->iov_base,
req->rq_svec->iov_len);
+ /* Don't use zero copy if this is a resend. If the RPC call
+ * completes while the socket holds a reference to the pages,
+ * then we may end up resending corrupted data.
+ */
+ if (task->tk_flags & RPC_TASK_SENT)
+ zerocopy = false;
/* Continue transmitting the packet/record. We must be careful
* to cope with writespace callbacks arriving _after_ we have
* called sendmsg(). */
while (1) {
status = xs_sendpages(transport->sock,
- NULL, 0, xdr, req->rq_bytes_sent);
+ NULL, 0, xdr, req->rq_bytes_sent,
+ zerocopy);
dprintk("RPC: xs_tcp_send_request(%u) = %d\n",
xdr->len - req->rq_bytes_sent, status);
--
1.8.3.2
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists