[PATCH AUTOSEL 4.19 312/671] NFS: Don't interrupt file writeout due to fatal errors

[PATCH AUTOSEL 4.19 312/671] NFS: Don't interrupt file writeout due to fatal errors

[Sasha Levin]

From: Trond Myklebust <trondmy@gmail.com>

[ Upstream commit 14bebe3c90b326d2a0df78aed5e9de090c71d878 ]

When flushing out dirty pages, the fact that we may hit fatal errors
is not a reason to stop writeback. Those errors are reported through
fsync(), not through the flush mechanism.

Fixes: a6598813a4c5b ("NFS: Don't write back further requests if there...")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 fs/nfs/write.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index 117ffd90419e..e27637fa0f79 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -646,7 +646,7 @@ static int nfs_page_async_flush(struct nfs_pageio_descriptor *pgio,
 	return ret;
 out_launder:
 	nfs_write_error_remove_page(req);
-	return ret;
+	return 0;
 }
 
 static int nfs_do_writepage(struct page *page, struct writeback_control *wbc,
-- 
2.20.1

[PATCH AUTOSEL 4.19 445/671] xprtrdma: Fix use-after-free in rpcrdma_post_recvs

[Sasha Levin]

From: Chuck Lever <chuck.lever@oracle.com>

[ Upstream commit 2d0abe36cf13fb7b577949fd1539326adddcc9bc ]

Dereference wr->next /before/ the memory backing wr has been
released. This issue was found by code inspection. It is not
expected to be a significant problem because it is in an error
path that is almost never executed.

Fixes: 7c8d9e7c8863 ("xprtrdma: Move Receive posting to ... ")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 net/sunrpc/xprtrdma/verbs.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c
index 5ddbf227e7c6..2a1d8ec7f706 100644
--- a/net/sunrpc/xprtrdma/verbs.c
+++ b/net/sunrpc/xprtrdma/verbs.c
@@ -1558,10 +1558,11 @@ rpcrdma_post_recvs(struct rpcrdma_xprt *r_xprt, bool temp)
 	rc = ib_post_recv(r_xprt->rx_ia.ri_id->qp, wr,
 			  (const struct ib_recv_wr **)&bad_wr);
 	if (rc) {
-		for (wr = bad_wr; wr; wr = wr->next) {
+		for (wr = bad_wr; wr;) {
 			struct rpcrdma_rep *rep;
 
 			rep = container_of(wr, struct rpcrdma_rep, rr_recv_wr);
+			wr = wr->next;
 			rpcrdma_recv_buffer_put(rep);
 			--count;
 		}
-- 
2.20.1

[PATCH AUTOSEL 4.19 628/671] SUNRPC: Fix svcauth_gss_proxy_init()

[Sasha Levin]

From: Chuck Lever <chuck.lever@oracle.com>

[ Upstream commit 5866efa8cbfbadf3905072798e96652faf02dbe8 ]

gss_read_proxy_verf() assumes things about the XDR buffer containing
the RPC Call that are not true for buffers generated by
svc_rdma_recv().

RDMA's buffers look more like what the upper layer generates for
sending: head is a kmalloc'd buffer; it does not point to a page
whose contents are contiguous with the first page in the buffers'
page array. The result is that ACCEPT_SEC_CONTEXT via RPC/RDMA has
stopped working on Linux NFS servers that use gssproxy.

This does not affect clients that use only TCP to send their
ACCEPT_SEC_CONTEXT operation (that's all Linux clients). Other
clients, like Solaris NFS clients, send ACCEPT_SEC_CONTEXT on the
same transport as they send all other NFS operations. Such clients
can send ACCEPT_SEC_CONTEXT via RPC/RDMA.

I thought I had found every direct reference in the server RPC code
to the rqstp->rq_pages field.

Bug found at the 2019 Westford NFS bake-a-thon.

Fixes: 3316f0631139 ("svcrdma: Persistently allocate and DMA- ... ")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Bill Baker <bill.baker@oracle.com>
Reviewed-by: Simo Sorce <simo@redhat.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 net/sunrpc/auth_gss/svcauth_gss.c | 84 +++++++++++++++++++++++--------
 1 file changed, 63 insertions(+), 21 deletions(-)

diff --git a/net/sunrpc/auth_gss/svcauth_gss.c b/net/sunrpc/auth_gss/svcauth_gss.c
index 1a65f88d021a..24c7a1e2bd34 100644
--- a/net/sunrpc/auth_gss/svcauth_gss.c
+++ b/net/sunrpc/auth_gss/svcauth_gss.c
@@ -1054,24 +1054,32 @@ gss_read_verf(struct rpc_gss_wire_cred *gc,
 	return 0;
 }
 
-/* Ok this is really heavily depending on a set of semantics in
- * how rqstp is set up by svc_recv and pages laid down by the
- * server when reading a request. We are basically guaranteed that
- * the token lays all down linearly across a set of pages, starting
- * at iov_base in rq_arg.head[0] which happens to be the first of a
- * set of pages stored in rq_pages[].
- * rq_arg.head[0].iov_base will provide us the page_base to pass
- * to the upcall.
- */
-static inline int
-gss_read_proxy_verf(struct svc_rqst *rqstp,
-		    struct rpc_gss_wire_cred *gc, __be32 *authp,
-		    struct xdr_netobj *in_handle,
-		    struct gssp_in_token *in_token)
+static void gss_free_in_token_pages(struct gssp_in_token *in_token)
 {
-	struct kvec *argv = &rqstp->rq_arg.head[0];
 	u32 inlen;
-	int res;
+	int i;
+
+	i = 0;
+	inlen = in_token->page_len;
+	while (inlen) {
+		if (in_token->pages[i])
+			put_page(in_token->pages[i]);
+		inlen -= inlen > PAGE_SIZE ? PAGE_SIZE : inlen;
+	}
+
+	kfree(in_token->pages);
+	in_token->pages = NULL;
+}
+
+static int gss_read_proxy_verf(struct svc_rqst *rqstp,
+			       struct rpc_gss_wire_cred *gc, __be32 *authp,
+			       struct xdr_netobj *in_handle,
+			       struct gssp_in_token *in_token)
+{
+	struct kvec *argv = &rqstp->rq_arg.head[0];
+	unsigned int page_base, length;
+	int pages, i, res;
+	size_t inlen;
 
 	res = gss_read_common_verf(gc, argv, authp, in_handle);
 	if (res)
@@ -1081,10 +1089,36 @@ gss_read_proxy_verf(struct svc_rqst *rqstp,
 	if (inlen > (argv->iov_len + rqstp->rq_arg.page_len))
 		return SVC_DENIED;
 
-	in_token->pages = rqstp->rq_pages;
-	in_token->page_base = (ulong)argv->iov_base & ~PAGE_MASK;
+	pages = DIV_ROUND_UP(inlen, PAGE_SIZE);
+	in_token->pages = kcalloc(pages, sizeof(struct page *), GFP_KERNEL);
+	if (!in_token->pages)
+		return SVC_DENIED;
+	in_token->page_base = 0;
 	in_token->page_len = inlen;
+	for (i = 0; i < pages; i++) {
+		in_token->pages[i] = alloc_page(GFP_KERNEL);
+		if (!in_token->pages[i]) {
+			gss_free_in_token_pages(in_token);
+			return SVC_DENIED;
+		}
+	}
 
+	length = min_t(unsigned int, inlen, argv->iov_len);
+	memcpy(page_address(in_token->pages[0]), argv->iov_base, length);
+	inlen -= length;
+
+	i = 1;
+	page_base = rqstp->rq_arg.page_base;
+	while (inlen) {
+		length = min_t(unsigned int, inlen, PAGE_SIZE);
+		memcpy(page_address(in_token->pages[i]),
+		       page_address(rqstp->rq_arg.pages[i]) + page_base,
+		       length);
+
+		inlen -= length;
+		page_base = 0;
+		i++;
+	}
 	return 0;
 }
 
@@ -1259,8 +1293,11 @@ static int svcauth_gss_proxy_init(struct svc_rqst *rqstp,
 		break;
 	case GSS_S_COMPLETE:
 		status = gss_proxy_save_rsc(sn->rsc_cache, &ud, &handle);
-		if (status)
+		if (status) {
+			pr_info("%s: gss_proxy_save_rsc failed (%d)\n",
+				__func__, status);
 			goto out;
+		}
 		cli_handle.data = (u8 *)&handle;
 		cli_handle.len = sizeof(handle);
 		break;
@@ -1271,15 +1308,20 @@ static int svcauth_gss_proxy_init(struct svc_rqst *rqstp,
 
 	/* Got an answer to the upcall; use it: */
 	if (gss_write_init_verf(sn->rsc_cache, rqstp,
-				&cli_handle, &ud.major_status))
+				&cli_handle, &ud.major_status)) {
+		pr_info("%s: gss_write_init_verf failed\n", __func__);
 		goto out;
+	}
 	if (gss_write_resv(resv, PAGE_SIZE,
 			   &cli_handle, &ud.out_token,
-			   ud.major_status, ud.minor_status))
+			   ud.major_status, ud.minor_status)) {
+		pr_info("%s: gss_write_resv failed\n", __func__);
 		goto out;
+	}
 
 	ret = SVC_COMPLETE;
 out:
+	gss_free_in_token_pages(&ud.in_token);
 	gssp_free_upcall_data(&ud);
 	return ret;
 }
-- 
2.20.1