FAILED: patch "[PATCH] SUNRPC: svcauth_gss: enforce krb5 token minimum length" failed to apply to 5.15-stable tree

FAILED: patch "[PATCH] SUNRPC: svcauth_gss: enforce krb5 token minimum length" failed to apply to 5.15-stable tree

[gregkh]

The patch below does not apply to the 5.15-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable@vger.kernel.org>.

To reproduce the conflict and resubmit, you may use the following commands:

git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-5.15.y
git checkout FETCH_HEAD
git cherry-pick -x a919c5c88769cf8fb3ec071e6078d830bf512489
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<stable@vger.kernel.org>' --in-reply-to '2026090352-bamboo-relic-6231@gregkh' --subject-prefix 'PATCH 5.15.y' 'HEAD^..'

Possible dependencies:



thanks,

greg k-h

------------------ original commit in Linus's tree ------------------

From a919c5c88769cf8fb3ec071e6078d830bf512489 Mon Sep 17 00:00:00 2001
From: Chris Mason <clm@meta.com>
Date: Sat, 23 May 2026 21:02:10 -0400
Subject: [PATCH] SUNRPC: svcauth_gss: enforce krb5 token minimum length

svcauth_gss_unwrap_priv() validates only an upper bound on the
wire-supplied opaque length before handing the buffer to
gss_unwrap():

    if (len > xdr_stream_remaining(xdr))
            goto unwrap_failed;
    offset = xdr_stream_pos(xdr);
    ...
    maj_stat = gss_unwrap(ctx, offset, offset + len, buf);

The wire value `len` flows unchanged as the upper bound into the
krb5 unwrap path, so a len in [0, 16] passes this check and is
handed to gss_unwrap(). For a krb5 v2 context that lands in
gss_krb5_unwrap_v2(), which reads the 16-byte RFC 4121 token
header fields at ptr+4 and ptr+6 and then calls rotate_left()
before any integrity check. With a sub-header length the header
reads run past the token, and _rotate_left()'s `shift %= buf->len`
path can divide by zero when buf->len has been driven to zero by
the truncated token. A header-only token (len == 16) is equally
invalid: with a non-zero RRC field and the opaque blob ending at
the XDR buffer boundary, rotate_left() builds a zero-length
subbuffer, reaching the same division.

Reject the token at the server entry point before it reaches the
krb5 unwrap core. A valid sealed RFC 4121 token must contain
the 16-byte header plus at least some encrypted payload.

Fix by adding a minimum-length check immediately after the
existing upper-bound check:

    if (len <= GSS_KRB5_TOK_HDR_LEN)
            goto unwrap_failed;

Fixes: 7c9fdcfb1b64 ("[PATCH] knfsd: svcrpc: gss: server-side implementation of rpcsec_gss privacy")
Cc: stable@vger.kernel.org
Assisted-by: kres (claude-opus-4-7)
Signed-off-by: Chris Mason <clm@meta.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Link: https://patch.msgid.link/20260524010213.557424-2-cel@kernel.org
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

diff --git a/net/sunrpc/auth_gss/svcauth_gss.c b/net/sunrpc/auth_gss/svcauth_gss.c
index d14209031e18..8e8aceb31270 100644
--- a/net/sunrpc/auth_gss/svcauth_gss.c
+++ b/net/sunrpc/auth_gss/svcauth_gss.c
@@ -949,6 +949,8 @@ svcauth_gss_unwrap_priv(struct svc_rqst *rqstp, u32 seq, struct gss_ctx *ctx)
 	}
 	if (len > xdr_stream_remaining(xdr))
 		goto unwrap_failed;
+	if (len <= GSS_KRB5_TOK_HDR_LEN)
+		goto unwrap_failed;
 	offset = xdr_stream_pos(xdr);
 
 	saved_len = buf->len;

[PATCH 5.15.y 1/4] SUNRPC: Fix crasher in unwrap_integ_data()

[Sasha Levin]

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

[ Upstream commit 4dd9daa9124aad3c3d4ceca4f1d417cc62d59156 ]

If a zero length is passed to kmalloc() it returns 0x10, which is
not a valid address. gss_verify_mic() subsequently crashes when it
attempts to dereference that pointer.

Instead of allocating this memory on every call based on an
untrusted size value, use a piece of dynamically-allocated scratch
memory that is always available.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Stable-dep-of: a919c5c88769 ("SUNRPC: svcauth_gss: enforce krb5 token minimum length")
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 net/sunrpc/auth_gss/svcauth_gss.c | 55 ++++++++++++++++++-------------
 1 file changed, 32 insertions(+), 23 deletions(-)

diff --git a/net/sunrpc/auth_gss/svcauth_gss.c b/net/sunrpc/auth_gss/svcauth_gss.c
index 1b8285e83df6d..b91dd85470a89 100644
--- a/net/sunrpc/auth_gss/svcauth_gss.c
+++ b/net/sunrpc/auth_gss/svcauth_gss.c
@@ -49,11 +49,36 @@
 #include <linux/sunrpc/svcauth.h>
 #include <linux/sunrpc/svcauth_gss.h>
 #include <linux/sunrpc/cache.h>
+#include <linux/sunrpc/gss_krb5.h>
 
 #include <trace/events/rpcgss.h>
 
 #include "gss_rpc_upcall.h"
 
+/*
+ * Unfortunately there isn't a maximum checksum size exported via the
+ * GSS API. Manufacture one based on GSS mechanisms supported by this
+ * implementation.
+ */
+#define GSS_MAX_CKSUMSIZE (GSS_KRB5_TOK_HDR_LEN + GSS_KRB5_MAX_CKSUM_LEN)
+
+/*
+ * This value may be increased in the future to accommodate other
+ * usage of the scratch buffer.
+ */
+#define GSS_SCRATCH_SIZE GSS_MAX_CKSUMSIZE
+
+struct gss_svc_data {
+	/* decoded gss client cred: */
+	struct rpc_gss_wire_cred	clcred;
+	/* save a pointer to the beginning of the encoded verifier,
+	 * for use in encryption/checksumming in svcauth_gss_release: */
+	__be32				*verf_start;
+	struct rsc			*rsci;
+
+	/* for temporary results */
+	u8				gsd_scratch[GSS_SCRATCH_SIZE];
+};
 
 /* The rpcsec_init cache is used for mapping RPCSEC_GSS_{,CONT_}INIT requests
  * into replies.
@@ -887,13 +912,11 @@ read_u32_from_xdr_buf(struct xdr_buf *buf, int base, u32 *obj)
 static int
 unwrap_integ_data(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq, struct gss_ctx *ctx)
 {
+	struct gss_svc_data *gsd = rqstp->rq_auth_data;
 	u32 integ_len, rseqno, maj_stat;
-	int stat = -EINVAL;
 	struct xdr_netobj mic;
 	struct xdr_buf integ_buf;
 
-	mic.data = NULL;
-
 	/* NFS READ normally uses splice to send data in-place. However
 	 * the data in cache can change after the reply's MIC is computed
 	 * but before the RPC reply is sent. To prevent the client from
@@ -917,11 +940,9 @@ unwrap_integ_data(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq, struct g
 	/* copy out mic... */
 	if (read_u32_from_xdr_buf(buf, integ_len, &mic.len))
 		goto unwrap_failed;
-	if (mic.len > RPC_MAX_AUTH_SIZE)
-		goto unwrap_failed;
-	mic.data = kmalloc(mic.len, GFP_KERNEL);
-	if (!mic.data)
+	if (mic.len > sizeof(gsd->gsd_scratch))
 		goto unwrap_failed;
+	mic.data = gsd->gsd_scratch;
 	if (read_bytes_from_xdr_buf(buf, integ_len + 4, mic.data, mic.len))
 		goto unwrap_failed;
 	maj_stat = gss_verify_mic(ctx, &integ_buf, &mic);
@@ -932,20 +953,17 @@ unwrap_integ_data(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq, struct g
 		goto bad_seqno;
 	/* trim off the mic and padding at the end before returning */
 	xdr_buf_trim(buf, round_up_to_quad(mic.len) + 4);
-	stat = 0;
-out:
-	kfree(mic.data);
-	return stat;
+	return 0;
 
 unwrap_failed:
 	trace_rpcgss_svc_unwrap_failed(rqstp);
-	goto out;
+	return -EINVAL;
 bad_seqno:
 	trace_rpcgss_svc_seqno_bad(rqstp, seq, rseqno);
-	goto out;
+	return -EINVAL;
 bad_mic:
 	trace_rpcgss_svc_mic(rqstp, maj_stat);
-	goto out;
+	return -EINVAL;
 }
 
 static inline int
@@ -1023,15 +1041,6 @@ unwrap_priv_data(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq, struct gs
 	return -EINVAL;
 }
 
-struct gss_svc_data {
-	/* decoded gss client cred: */
-	struct rpc_gss_wire_cred	clcred;
-	/* save a pointer to the beginning of the encoded verifier,
-	 * for use in encryption/checksumming in svcauth_gss_release: */
-	__be32				*verf_start;
-	struct rsc			*rsci;
-};
-
 static int
 svcauth_gss_set_client(struct svc_rqst *rqstp)
 {
-- 
2.53.0

[PATCH 5.15.y 2/4] SUNRPC: Rename automatic variables in unwrap_priv_data()

[Sasha Levin]

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

[ Upstream commit f4a59e822f8116baf51a9fc86590e4391bd542d2 ]

Clean up: To help orient readers, name the stack variables to match
the XDR field names.

Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Stable-dep-of: a919c5c88769 ("SUNRPC: svcauth_gss: enforce krb5 token minimum length")
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 net/sunrpc/auth_gss/svcauth_gss.c | 38 ++++++++++++++++++++-----------
 1 file changed, 25 insertions(+), 13 deletions(-)

diff --git a/net/sunrpc/auth_gss/svcauth_gss.c b/net/sunrpc/auth_gss/svcauth_gss.c
index b91dd85470a89..7ba523335aefa 100644
--- a/net/sunrpc/auth_gss/svcauth_gss.c
+++ b/net/sunrpc/auth_gss/svcauth_gss.c
@@ -983,16 +983,28 @@ fix_priv_head(struct xdr_buf *buf, int pad)
 	}
 }
 
+/*
+ * RFC 2203, Section 5.3.2.3
+ *
+ *	struct rpc_gss_priv_data {
+ *		opaque databody_priv<>
+ *	};
+ *
+ *	struct rpc_gss_data_t {
+ *		unsigned int seq_num;
+ *		proc_req_arg_t arg;
+ *	};
+ */
 static int
-unwrap_priv_data(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq, struct gss_ctx *ctx)
+svcauth_gss_unwrap_priv(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq,
+			struct gss_ctx *ctx)
 {
-	u32 priv_len, maj_stat;
+	u32 len, seq_num, maj_stat;
 	int pad, remaining_len, offset;
-	u32 rseqno;
 
 	clear_bit(RQ_SPLICE_OK, &rqstp->rq_flags);
 
-	priv_len = svc_getnl(&buf->head[0]);
+	len = svc_getnl(&buf->head[0]);
 	if (rqstp->rq_deferred) {
 		/* Already decrypted last time through! The sequence number
 		 * check at out_seq is unnecessary but harmless: */
@@ -1002,14 +1014,14 @@ unwrap_priv_data(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq, struct gs
 	 * request to the end, where head[0].iov_len is just the bytes
 	 * not yet read from the head, so these two values are different: */
 	remaining_len = total_buf_len(buf);
-	if (priv_len > remaining_len)
+	if (len > remaining_len)
 		goto unwrap_failed;
-	pad = remaining_len - priv_len;
+	pad = remaining_len - len;
 	buf->len -= pad;
 	fix_priv_head(buf, pad);
 
-	maj_stat = gss_unwrap(ctx, 0, priv_len, buf);
-	pad = priv_len - buf->len;
+	maj_stat = gss_unwrap(ctx, 0, len, buf);
+	pad = len - buf->len;
 	/* The upper layers assume the buffer is aligned on 4-byte boundaries.
 	 * In the krb5p case, at least, the data ends up offset, so we need to
 	 * move it around. */
@@ -1025,8 +1037,8 @@ unwrap_priv_data(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq, struct gs
 	if (maj_stat != GSS_S_COMPLETE)
 		goto bad_unwrap;
 out_seq:
-	rseqno = svc_getnl(&buf->head[0]);
-	if (rseqno != seq)
+	seq_num = svc_getnl(&buf->head[0]);
+	if (seq_num != seq)
 		goto bad_seqno;
 	return 0;
 
@@ -1034,7 +1046,7 @@ unwrap_priv_data(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq, struct gs
 	trace_rpcgss_svc_unwrap_failed(rqstp);
 	return -EINVAL;
 bad_seqno:
-	trace_rpcgss_svc_seqno_bad(rqstp, seq, rseqno);
+	trace_rpcgss_svc_seqno_bad(rqstp, seq, seq_num);
 	return -EINVAL;
 bad_unwrap:
 	trace_rpcgss_svc_unwrap(rqstp, maj_stat);
@@ -1653,8 +1665,8 @@ svcauth_gss_accept(struct svc_rqst *rqstp)
 			/* placeholders for length and seq. number: */
 			svc_putnl(resv, 0);
 			svc_putnl(resv, 0);
-			if (unwrap_priv_data(rqstp, &rqstp->rq_arg,
-					gc->gc_seq, rsci->mechctx))
+			if (svcauth_gss_unwrap_priv(rqstp, &rqstp->rq_arg,
+						    gc->gc_seq, rsci->mechctx))
 				goto garbage_args;
 			rqstp->rq_auth_slack = RPC_MAX_AUTH_SIZE * 2;
 			break;
-- 
2.53.0

[PATCH 5.15.y 3/4] SUNRPC: Convert unwrap_priv_data() to use xdr_stream

[Sasha Levin]

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

[ Upstream commit 42140718ea26c47fb06c679451a6aa6703545136 ]

Done as part of hardening the server-side RPC header decoding path.

Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Stable-dep-of: a919c5c88769 ("SUNRPC: svcauth_gss: enforce krb5 token minimum length")
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 include/linux/sunrpc/xdr.h        |  1 -
 net/sunrpc/auth_gss/svcauth_gss.c | 64 ++++++++++---------------------
 net/sunrpc/xdr.c                  |  7 ----
 3 files changed, 21 insertions(+), 51 deletions(-)

diff --git a/include/linux/sunrpc/xdr.h b/include/linux/sunrpc/xdr.h
index 98e197376a0d9..add7174cb245a 100644
--- a/include/linux/sunrpc/xdr.h
+++ b/include/linux/sunrpc/xdr.h
@@ -188,7 +188,6 @@ xdr_adjust_iovec(struct kvec *iov, __be32 *p)
 /*
  * XDR buffer helper functions
  */
-extern void xdr_shift_buf(struct xdr_buf *, size_t);
 extern void xdr_buf_from_iov(const struct kvec *, struct xdr_buf *);
 extern int xdr_buf_subsegment(const struct xdr_buf *, struct xdr_buf *, unsigned int, unsigned int);
 extern void xdr_buf_trim(struct xdr_buf *, unsigned int);
diff --git a/net/sunrpc/auth_gss/svcauth_gss.c b/net/sunrpc/auth_gss/svcauth_gss.c
index 7ba523335aefa..afcfdb2f84b6c 100644
--- a/net/sunrpc/auth_gss/svcauth_gss.c
+++ b/net/sunrpc/auth_gss/svcauth_gss.c
@@ -972,17 +972,6 @@ total_buf_len(struct xdr_buf *buf)
 	return buf->head[0].iov_len + buf->page_len + buf->tail[0].iov_len;
 }
 
-static void
-fix_priv_head(struct xdr_buf *buf, int pad)
-{
-	if (buf->page_len == 0) {
-		/* We need to adjust head and buf->len in tandem in this
-		 * case to make svc_defer() work--it finds the original
-		 * buffer start using buf->len - buf->head[0].iov_len. */
-		buf->head[0].iov_len -= pad;
-	}
-}
-
 /*
  * RFC 2203, Section 5.3.2.3
  *
@@ -995,49 +984,37 @@ fix_priv_head(struct xdr_buf *buf, int pad)
  *		proc_req_arg_t arg;
  *	};
  */
-static int
-svcauth_gss_unwrap_priv(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq,
-			struct gss_ctx *ctx)
+static noinline_for_stack int
+svcauth_gss_unwrap_priv(struct svc_rqst *rqstp, u32 seq, struct gss_ctx *ctx)
 {
-	u32 len, seq_num, maj_stat;
-	int pad, remaining_len, offset;
+	struct xdr_stream *xdr = &rqstp->rq_arg_stream;
+	u32 len, maj_stat, seq_num, offset;
+	struct xdr_buf *buf = xdr->buf;
+	unsigned int saved_len;
 
 	clear_bit(RQ_SPLICE_OK, &rqstp->rq_flags);
 
-	len = svc_getnl(&buf->head[0]);
+	if (xdr_stream_decode_u32(xdr, &len) < 0)
+		goto unwrap_failed;
 	if (rqstp->rq_deferred) {
 		/* Already decrypted last time through! The sequence number
 		 * check at out_seq is unnecessary but harmless: */
 		goto out_seq;
 	}
-	/* buf->len is the number of bytes from the original start of the
-	 * request to the end, where head[0].iov_len is just the bytes
-	 * not yet read from the head, so these two values are different: */
-	remaining_len = total_buf_len(buf);
-	if (len > remaining_len)
+	if (len > xdr_stream_remaining(xdr))
 		goto unwrap_failed;
-	pad = remaining_len - len;
-	buf->len -= pad;
-	fix_priv_head(buf, pad);
-
-	maj_stat = gss_unwrap(ctx, 0, len, buf);
-	pad = len - buf->len;
-	/* The upper layers assume the buffer is aligned on 4-byte boundaries.
-	 * In the krb5p case, at least, the data ends up offset, so we need to
-	 * move it around. */
-	/* XXX: This is very inefficient.  It would be better to either do
-	 * this while we encrypt, or maybe in the receive code, if we can peak
-	 * ahead and work out the service and mechanism there. */
-	offset = xdr_pad_size(buf->head[0].iov_len);
-	if (offset) {
-		buf->buflen = RPCSVC_MAXPAYLOAD;
-		xdr_shift_buf(buf, offset);
-		fix_priv_head(buf, pad);
-	}
+	offset = xdr_stream_pos(xdr);
+
+	saved_len = buf->len;
+	maj_stat = gss_unwrap(ctx, offset, offset + len, buf);
 	if (maj_stat != GSS_S_COMPLETE)
 		goto bad_unwrap;
+	xdr->nwords -= XDR_QUADLEN(saved_len - buf->len);
+
 out_seq:
-	seq_num = svc_getnl(&buf->head[0]);
+	/* gss_unwrap() decrypted the sequence number. */
+	if (xdr_stream_decode_u32(xdr, &seq_num) < 0)
+		goto unwrap_failed;
 	if (seq_num != seq)
 		goto bad_seqno;
 	return 0;
@@ -1665,8 +1642,9 @@ svcauth_gss_accept(struct svc_rqst *rqstp)
 			/* placeholders for length and seq. number: */
 			svc_putnl(resv, 0);
 			svc_putnl(resv, 0);
-			if (svcauth_gss_unwrap_priv(rqstp, &rqstp->rq_arg,
-						    gc->gc_seq, rsci->mechctx))
+			svcxdr_init_decode(rqstp);
+			if (svcauth_gss_unwrap_priv(rqstp, gc->gc_seq,
+						    rsci->mechctx))
 				goto garbage_args;
 			rqstp->rq_auth_slack = RPC_MAX_AUTH_SIZE * 2;
 			break;
diff --git a/net/sunrpc/xdr.c b/net/sunrpc/xdr.c
index b227d0c8471ff..34cd38e757d6f 100644
--- a/net/sunrpc/xdr.c
+++ b/net/sunrpc/xdr.c
@@ -835,13 +835,6 @@ static unsigned int xdr_shrink_pagelen(struct xdr_buf *buf, unsigned int len)
 	return shift;
 }
 
-void
-xdr_shift_buf(struct xdr_buf *buf, size_t len)
-{
-	xdr_shrink_bufhead(buf, buf->head->iov_len - len);
-}
-EXPORT_SYMBOL_GPL(xdr_shift_buf);
-
 /**
  * xdr_stream_pos - Return the current offset from the start of the xdr_stream
  * @xdr: pointer to struct xdr_stream
-- 
2.53.0

[PATCH 5.15.y 4/4] SUNRPC: svcauth_gss: enforce krb5 token minimum length

[Sasha Levin]

From: Chris Mason <clm@meta.com>

[ Upstream commit a919c5c88769cf8fb3ec071e6078d830bf512489 ]

svcauth_gss_unwrap_priv() validates only an upper bound on the
wire-supplied opaque length before handing the buffer to
gss_unwrap():

    if (len > xdr_stream_remaining(xdr))
            goto unwrap_failed;
    offset = xdr_stream_pos(xdr);
    ...
    maj_stat = gss_unwrap(ctx, offset, offset + len, buf);

The wire value `len` flows unchanged as the upper bound into the
krb5 unwrap path, so a len in [0, 16] passes this check and is
handed to gss_unwrap(). For a krb5 v2 context that lands in
gss_krb5_unwrap_v2(), which reads the 16-byte RFC 4121 token
header fields at ptr+4 and ptr+6 and then calls rotate_left()
before any integrity check. With a sub-header length the header
reads run past the token, and _rotate_left()'s `shift %= buf->len`
path can divide by zero when buf->len has been driven to zero by
the truncated token. A header-only token (len == 16) is equally
invalid: with a non-zero RRC field and the opaque blob ending at
the XDR buffer boundary, rotate_left() builds a zero-length
subbuffer, reaching the same division.

Reject the token at the server entry point before it reaches the
krb5 unwrap core. A valid sealed RFC 4121 token must contain
the 16-byte header plus at least some encrypted payload.

Fix by adding a minimum-length check immediately after the
existing upper-bound check:

    if (len <= GSS_KRB5_TOK_HDR_LEN)
            goto unwrap_failed;

Fixes: 7c9fdcfb1b64 ("[PATCH] knfsd: svcrpc: gss: server-side implementation of rpcsec_gss privacy")
Cc: stable@vger.kernel.org
Assisted-by: kres (claude-opus-4-7)
Signed-off-by: Chris Mason <clm@meta.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Link: https://patch.msgid.link/20260524010213.557424-2-cel@kernel.org
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 net/sunrpc/auth_gss/svcauth_gss.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/net/sunrpc/auth_gss/svcauth_gss.c b/net/sunrpc/auth_gss/svcauth_gss.c
index afcfdb2f84b6c..77a61bdd1d5ed 100644
--- a/net/sunrpc/auth_gss/svcauth_gss.c
+++ b/net/sunrpc/auth_gss/svcauth_gss.c
@@ -1003,6 +1003,8 @@ svcauth_gss_unwrap_priv(struct svc_rqst *rqstp, u32 seq, struct gss_ctx *ctx)
 	}
 	if (len > xdr_stream_remaining(xdr))
 		goto unwrap_failed;
+	if (len <= GSS_KRB5_TOK_HDR_LEN)
+		goto unwrap_failed;
 	offset = xdr_stream_pos(xdr);
 
 	saved_len = buf->len;
-- 
2.53.0