Netdev List
 help / color / mirror / Atom feed
From: Stephen Hemminger <stephen@networkplumber.org>
To: netdev@vger.kernel.org
Cc: dsahern@kernel.org, Stephen Hemminger <stephen@networkplumber.org>
Subject: [PATCH iproute2 04/11] ss: size the TCPDIAG_FILE buffer to the record
Date: Thu, 12 Nov 2026 11:59:17 -0800	[thread overview]
Message-ID: <20261112200046.2164391-5-stephen@networkplumber.org> (raw)
In-Reply-To: <20261112200046.2164391-1-stephen@networkplumber.org>

tcp_show_netlink_file() read each record into a 16K stack buffer
using the length from the file, so a record over 16K, which the
kernel can produce (e.g. many TCP-MD5 keys), overflowed the stack.

Read into a heap buffer grown to the record instead. A length below
a netlink header is still rejected, since the size arithmetic would
underflow.

Reported-by: Anthropic OSS scanner (ANT-2026-RV0SH3JH)
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
---
 misc/ss.c | 40 +++++++++++++++++++++++++++++++++++-----
 1 file changed, 35 insertions(+), 5 deletions(-)

diff --git a/misc/ss.c b/misc/ss.c
index 4462a479..3c538a4e 100644
--- a/misc/ss.c
+++ b/misc/ss.c
@@ -3923,8 +3923,9 @@ Exit:
 
 static int tcp_show_netlink_file(struct filter *f)
 {
+	struct nlmsghdr *h = NULL;
+	size_t buflen = 0;
 	FILE	*fp;
-	char	buf[16384];
 	int	err = -1;
 
 	if ((fp = fopen(getenv("TCPDIAG_FILE"), "r")) == NULL) {
@@ -3935,11 +3936,11 @@ static int tcp_show_netlink_file(struct filter *f)
 	while (1) {
 		int err2;
 		size_t status, nitems;
-		struct nlmsghdr *h = (struct nlmsghdr *)buf;
+		struct nlmsghdr hdr;
 		struct sockstat s = {};
 
-		status = fread(buf, 1, sizeof(*h), fp);
-		if (status != sizeof(*h)) {
+		status = fread(&hdr, 1, sizeof(hdr), fp);
+		if (status != sizeof(hdr)) {
 			if (ferror(fp))
 				perror("Reading header from $TCPDIAG_FILE");
 			if (feof(fp))
@@ -3947,7 +3948,29 @@ static int tcp_show_netlink_file(struct filter *f)
 			break;
 		}
 
-		nitems = NLMSG_ALIGN(h->nlmsg_len - sizeof(*h));
+		/* NLMSG_ALIGN() is u32 arithmetic and wraps near UINT_MAX,
+		 * which would size the buffer below the record.
+		 */
+		if (hdr.nlmsg_len < sizeof(hdr) ||
+		    hdr.nlmsg_len > UINT32_MAX - NLMSG_ALIGNTO + 1) {
+			fprintf(stderr, "Malformed message in $TCPDIAG_FILE: len=%u\n",
+				hdr.nlmsg_len);
+			break;
+		}
+
+		if (NLMSG_ALIGN(hdr.nlmsg_len) > buflen) {
+			void *nbuf = realloc(h, NLMSG_ALIGN(hdr.nlmsg_len));
+
+			if (!nbuf) {
+				perror("Reading $TCPDIAG_FILE");
+				break;
+			}
+			h = nbuf;
+			buflen = NLMSG_ALIGN(hdr.nlmsg_len);
+		}
+		*h = hdr;
+
+		nitems = NLMSG_ALIGN(h->nlmsg_len) - sizeof(*h);
 		status = fread(h+1, 1, nitems, fp);
 
 		if (status != nitems) {
@@ -3976,6 +3999,12 @@ static int tcp_show_netlink_file(struct filter *f)
 			break;
 		}
 
+		if (h->nlmsg_len < NLMSG_LENGTH(sizeof(struct inet_diag_msg))) {
+			fprintf(stderr, "Malformed message in $TCPDIAG_FILE: len=%u\n",
+				h->nlmsg_len);
+			break;
+		}
+
 		parse_diag_msg(h, &s);
 		s.type = IPPROTO_TCP;
 
@@ -3989,6 +4018,7 @@ static int tcp_show_netlink_file(struct filter *f)
 		}
 	}
 
+	free(h);
 	fclose(fp);
 	return err;
 }
-- 
2.53.0


  parent reply	other threads:[~2026-11-12 20:00 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-12 19:59 [PATCH iproute2 00/11] ss fixes from an OSS scan Stephen Hemminger
2026-11-12 19:59 ` [PATCH iproute2 01/11] ss: fix double close on sock_diag send failure Stephen Hemminger
2026-11-12 19:59 ` [PATCH iproute2 02/11] ss: drop /proc fallback for socket listing Stephen Hemminger
2026-11-12 19:59 ` [PATCH iproute2 03/11] ss: escape names chosen by the socket owner Stephen Hemminger
2026-11-12 19:59 ` Stephen Hemminger [this message]
2026-11-12 19:59 ` [PATCH iproute2 05/11] ss: replay all dumps in TCPDIAG_FILE Stephen Hemminger
2026-11-12 19:59 ` [PATCH iproute2 06/11] ss: bound state and type name lookups Stephen Hemminger
2026-11-12 19:59 ` [PATCH iproute2 07/11] ss: fix rpcbind service cache Stephen Hemminger
2026-11-12 19:59 ` [PATCH iproute2 08/11] ss: read socket SELinux context only for -z Stephen Hemminger
2026-11-12 19:59 ` [PATCH iproute2 09/11] ss: check owner before naming a netlink socket Stephen Hemminger
2026-11-12 19:59 ` [PATCH iproute2 10/11] ss: keep following events after ENOBUFS Stephen Hemminger
2026-11-12 19:59 ` [PATCH iproute2 11/11] ss: report a failed SCTP or MPTCP dump Stephen Hemminger
  -- strict thread matches above, loose matches on Subject: below --
2026-11-09 21:25 [PATCH iproute2 00/11] ss fixes from an OSS scan Stephen Hemminger
2026-11-09 21:25 ` [PATCH iproute2 04/11] ss: size the TCPDIAG_FILE buffer to the record Stephen Hemminger

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20261112200046.2164391-5-stephen@networkplumber.org \
    --to=stephen@networkplumber.org \
    --cc=dsahern@kernel.org \
    --cc=netdev@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox