Skip to contentSkip to conversation

netlink: Use copy_to_user() for optval in netlink_getsockopt().

PATCH Superseded Kuniyuki Iwashima 3 participants 5 messages Apr 20, 2023 – Apr 21, 2023

Subsystems (2)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 netlink: Use copy_to_user() for optval in netlink_getsockopt().

Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require more than int as length.

The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char.

Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval.

To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user().

Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page.

  $ man 7 netlink
  ...
  Socket options
       To set or get a netlink socket option, call getsockopt(2) to read
       or setsockopt(2) to write the option with the option level argument
       set to SOL_NETLINK.  Unless otherwise noted, optval is a pointer to
       an int.
Fixes 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups")
Fixes be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option")
Fixes 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag")
Fixes 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option")
Fixes 2d4bc93368f5 ("netlink: extended ACK reporting")
Fixes 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps")

---
v2:
* Keep the len check and setting

v1: https://lore.kernel.org/netdev/[email protected]/

---
 net/netlink/af_netlink.c | 61 +++++++++++-----------------------------
 1 file changed, 17 insertions(+), 44 deletions(-)
net/netlink/af_netlink.c +17 -44
index f365dfdd672d..5c0d17b3984c 100644
@@ -1742,7 +1742,7 @@ static int netlink_getsockopt(struct socket *sock, int level, int optname,
 {
 	struct sock *sk = sock->sk;
 	struct netlink_sock *nlk = nlk_sk(sk);
-	int len, val, err;
+	int len, val;
 
 	if (level != SOL_NETLINK)
 		return -ENOPROTOOPT;
@@ -1753,40 +1753,27 @@ static int netlink_getsockopt(struct socket *sock, int level, int optname,
 		return -EINVAL;
 
 	switch (optname) {
-	case NETLINK_PKTINFO:
+	case NETLINK_LIST_MEMBERSHIPS:
+		break;
+	default:
 		if (len < sizeof(int))
 			return -EINVAL;
 		len = sizeof(int);
+	}
+
+	switch (optname) {
+	case NETLINK_PKTINFO:
 		val = nlk->flags & NETLINK_F_RECV_PKTINFO ? 1 : 0;
-		if (put_user(len, optlen) ||
-		    put_user(val, optval))
-			return -EFAULT;
-		err = 0;
 		break;
 	case NETLINK_BROADCAST_ERROR:
-		if (len < sizeof(int))
-			return -EINVAL;
-		len = sizeof(int);
 		val = nlk->flags & NETLINK_F_BROADCAST_SEND_ERROR ? 1 : 0;
-		if (put_user(len, optlen) ||
-		    put_user(val, optval))
-			return -EFAULT;
-		err = 0;
 		break;
 	case NETLINK_NO_ENOBUFS:
-		if (len < sizeof(int))
-			return -EINVAL;
-		len = sizeof(int);
 		val = nlk->flags & NETLINK_F_RECV_NO_ENOBUFS ? 1 : 0;
-		if (put_user(len, optlen) ||
-		    put_user(val, optval))
-			return -EFAULT;
-		err = 0;
 		break;
 	case NETLINK_LIST_MEMBERSHIPS: {
-		int pos, idx, shift;
+		int pos, idx, shift, err = 0;
 
-		err = 0;
 		netlink_lock_table();
 		for (pos = 0; pos * 8 < nlk->ngroups; pos += sizeof(u32)) {
 			if (len - pos < sizeof(u32))
@@ -1803,40 +1790,26 @@ static int netlink_getsockopt(struct socket *sock, int level, int optname,
 		if (put_user(ALIGN(nlk->ngroups / 8, sizeof(u32)), optlen))
 			err = -EFAULT;
 		netlink_unlock_table();
-		break;
+		return err;
 	}
 	case NETLINK_CAP_ACK:
-		if (len < sizeof(int))
-			return -EINVAL;
-		len = sizeof(int);
 		val = nlk->flags & NETLINK_F_CAP_ACK ? 1 : 0;
-		if (put_user(len, optlen) ||
-		    put_user(val, optval))
-			return -EFAULT;
-		err = 0;
 		break;
 	case NETLINK_EXT_ACK:
-		if (len < sizeof(int))
-			return -EINVAL;
-		len = sizeof(int);
 		val = nlk->flags & NETLINK_F_EXT_ACK ? 1 : 0;
-		if (put_user(len, optlen) || put_user(val, optval))
-			return -EFAULT;
-		err = 0;
 		break;
 	case NETLINK_GET_STRICT_CHK:
-		if (len < sizeof(int))
-			return -EINVAL;
-		len = sizeof(int);
 		val = nlk->flags & NETLINK_F_STRICT_CHK ? 1 : 0;
-		if (put_user(len, optlen) || put_user(val, optval))
-			return -EFAULT;
-		err = 0;
 		break;
 	default:
-		err = -ENOPROTOOPT;
+		return -ENOPROTOOPT;
 	}
-	return err;
+
+	if (put_user(len, optlen) ||
+	    copy_to_user(optval, &val, len))
+		return -EFAULT;
+
+	return 0;
 }
 
 static void netlink_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb)
Signature · 1 line
2.30.2
maintainer
#1 lore raw
Quoted 28 lines · On Thu, 20 Apr 2023 16:33:51 -0700 Kuniyuki Iwashima wrote:
net/netlink/af_netlink.c +5 -3
index f365dfdd672d..5c0d17b3984c 100644
@@ -1742,7 +1742,7 @@ static int netlink_getsockopt(struct socket *sock, int level, int optname,
 {
 	struct sock *sk = sock->sk;
 	struct netlink_sock *nlk = nlk_sk(sk);
-	int len, val, err;
+	int len, val;
 
 	if (level != SOL_NETLINK)
 		return -ENOPROTOOPT;
@@ -1753,40 +1753,27 @@ static int netlink_getsockopt(struct socket *sock, int level, int optname,
 		return -EINVAL;
 
 	switch (optname) {
-	case NETLINK_PKTINFO:
+	case NETLINK_LIST_MEMBERSHIPS:
+		break;
+	default:
 		if (len < sizeof(int))
 			return -EINVAL;
 		len = sizeof(int);
+	}
 
 
-		err = 0;

net/netlink/af_netlink.c static int netlink_getsockopt(struct socket *sock, int level, int optname, · line 1772

Quoted 10 lines
net/netlink/af_netlink.c +5 -2
 		break;
 	default:
-		err = -ENOPROTOOPT;
+		return -ENOPROTOOPT;
 	}
-	return err;
+
+	if (put_user(len, optlen) ||
+	    copy_to_user(optval, &val, len))
+		return -EFAULT;

Maybe this is a nit pick but we'd unnecessarily change the return value to unknown opts when len < 4, from -ENOPROTOOPT to -EINVAL, right?

How about we do:

	unsigned int flag;

	flag = 0;
	switch (optname) {
	case NETLINK_PKTINFO:
		flag = NETLINK_F_RECV_PKTINFO;
 		break;
 	case NETLINK_BROADCAST_ERROR:
		flag = NETLINK_F_BROADCAST_SEND_ERROR;
		break;
	...
	case NETLINK_LIST_MEMBERSHIPS: {
	...
	default:
		return -ENOPROTOOPT;
	}

	if (flag) {
		if (len < sizeof(int))
			return -EINVAL;
		len = sizeof(int);
 		val = nlk->flags & flag ? 1 : 0;
		if (put_user(len, optlen) || 
		    copy_to_user(optval, &val, len))
			return -EFAULT;
		...
	}
#2 lore raw
From:   Jakub Kicinski <[email protected]>
Date:   Thu, 20 Apr 2023 20:33:24 -0700
Quoted 28 lines · On Thu, 20 Apr 2023 16:33:51 -0700 Kuniyuki Iwashima wrote:
Quoted 25 lines
net/netlink/af_netlink.c +5 -2
index f365dfdd672d..5c0d17b3984c 100644
@@ -1742,7 +1742,7 @@ static int netlink_getsockopt(struct socket *sock, int level, int optname,
 {
 	struct sock *sk = sock->sk;
 	struct netlink_sock *nlk = nlk_sk(sk);
-	int len, val, err;
+	int len, val;
 
 	if (level != SOL_NETLINK)
 		return -ENOPROTOOPT;
@@ -1753,40 +1753,27 @@ static int netlink_getsockopt(struct socket *sock, int level, int optname,
 		return -EINVAL;
 
 	switch (optname) {
-	case NETLINK_PKTINFO:
+	case NETLINK_LIST_MEMBERSHIPS:
+		break;
+	default:
 		if (len < sizeof(int))
 			return -EINVAL;
 		len = sizeof(int);
+	}
Quoted 14 lines

net/netlink/af_netlink.c static int netlink_getsockopt(struct socket *sock, int level, int optname, · line 1810

Quoted 11 lines
net/netlink/af_netlink.c +5 -3
-		err = 0;
 		break;
 	default:
-		err = -ENOPROTOOPT;
+		return -ENOPROTOOPT;
 	}
-	return err;
+
+	if (put_user(len, optlen) ||
+	    copy_to_user(optval, &val, len))
+		return -EFAULT;

Maybe this is a nit pick but we'd unnecessarily change the return value to unknown opts when len < 4, from -ENOPROTOOPT to -EINVAL, right?

Exactly, we should not change it.

Quoted 2 lines

How about we do:

This is cleaner, will do in v3.

Thanks!

Quoted 28 lines
	unsigned int flag;

	flag = 0;
	switch (optname) {
	case NETLINK_PKTINFO:
		flag = NETLINK_F_RECV_PKTINFO;
net/netlink/af_netlink.c +0 -0
 		break;
 	case NETLINK_BROADCAST_ERROR:
		flag = NETLINK_F_BROADCAST_SEND_ERROR;
		break;
	...
	case NETLINK_LIST_MEMBERSHIPS: {
	...
	default:
		return -ENOPROTOOPT;
	}

	if (flag) {
		if (len < sizeof(int))
			return -EINVAL;
		len = sizeof(int);
 		val = nlk->flags & flag ? 1 : 0;
		if (put_user(len, optlen) || 
		    copy_to_user(optval, &val, len))
			return -EFAULT;
		...
	}
#3 lore raw
In reply to Kuniyuki Iwashima
Quoted 3 lines · Kuniyuki Iwashima

Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require more than int as length.

Nit: not "more than" but "at least" (and sizeof(int), I guess).

Quoted 7 lines

The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char.

Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval.

Maybe as a future thing, we should make the getsockopt method prototype have void here, so this kind of thing becomes a compilation error? That affects a fair number I guess, though I can't think of any socket options that really _should_ be just a char, so if it fails anywhere that might uncover additional bugs (and potentially avoid future ones)?

johannes

#4 lore raw
From:   Johannes Berg <[email protected]>
Date:   Fri, 21 Apr 2023 09:56:37 +0200
Quoted 6 lines · Kuniyuki Iwashima
Quoted 3 lines

Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require more than int as length.

Nit: not "more than" but "at least" (and sizeof(int), I guess).

Will change in v3.

Quoted 9 lines
Quoted 8 lines

The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char.

Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval.

Quoted 5 lines

Maybe as a future thing, we should make the getsockopt method prototype have void here, so this kind of thing becomes a compilation error? That affects a fair number I guess, though I can't think of any socket options that really _should_ be just a char, so if it fails anywhere that might uncover additional bugs (and potentially avoid future ones)?

Ah, cool, we can uncover the same issue easily by doing so and fix it unless the handler accepts char.

Thanks!