Skip to contentSkip to conversation

scm: provide full privilege set via SCM_PRIVILEGE

PATCH In review Casey Schaufler 3 participants 9 messages Jan 17, 2011 – Feb 15, 2011

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 scm: provide full privilege set via SCM_PRIVILEGE

Subject: [PATCH] scm: provide full privilege set via SCM_PRIVILEGE

The SCM mechanism currently provides interfaces for delivering the uid/gid and the "security context" (LSM information) of the peer on a UDS socket. All of the security credential information is available, but there is no interface available to obtain it. Further, the existing interfaces require that the user chose between the uid/gid and the context as the existing interfaces are exclusive.

This patch introduces an additional interface that provides a complete set of security information from the peer credential. No additional work is required to provide the information internally, it is all being passed, just not exposed.

---

 include/asm-generic/socket.h |    1 +
 include/linux/net.h          |    1 +
 include/linux/socket.h       |    1 +
 include/net/scm.h            |   80 +++++++++++++++++++++++++++++++++++++++++-
 net/core/sock.c              |   11 ++++++
 5 files changed, 93 insertions(+), 1 deletions(-)
include/asm-generic/socket.h +1 -0
index 9a6115e..7aa8e84 100644
@@ -64,4 +64,5 @@
 #define SO_DOMAIN		39
 
 #define SO_RXQ_OVFL             40
+#define SO_PASSPRIV		41
 #endif /* __ASM_GENERIC_SOCKET_H */
include/linux/net.h +1 -0
index 16faa13..159a929 100644
@@ -71,6 +71,7 @@ struct net;
 #define SOCK_NOSPACE		2
 #define SOCK_PASSCRED		3
 #define SOCK_PASSSEC		4
+#define SOCK_PASSPRIV		5
 
 #ifndef ARCH_HAS_SOCKET_TYPES
 /**
include/linux/socket.h +1 -0
index 86b652f..e9cfd68 100644
@@ -147,6 +147,7 @@ static inline struct cmsghdr * cmsg_nxthdr (struct msghdr *__msg, struct cmsghdr
 #define	SCM_RIGHTS	0x01		/* rw: access rights (array of int) */
 #define SCM_CREDENTIALS 0x02		/* rw: struct ucred		*/
 #define SCM_SECURITY	0x03		/* rw: security label		*/
+#define SCM_PRIVILEGES  0x04		/* rw: privilege set		*/
 
 struct ucred {
 	__u32	pid;
include/net/scm.h +79 -1

97 lines not on this page · raw message

net/core/sock.c +11 -0
index fb60801..f134126 100644
@@ -725,6 +725,13 @@ set_rcvbuf:
 		else
 			clear_bit(SOCK_PASSSEC, &sock->flags);
 		break;
+
+	case SO_PASSPRIV:
+		if (valbool)
+			set_bit(SOCK_PASSPRIV, &sock->flags);
+		else
+			clear_bit(SOCK_PASSPRIV, &sock->flags);
+		break;
 	case SO_MARK:
 		if (!capable(CAP_NET_ADMIN))
 			ret = -EPERM;
@@ -950,6 +957,10 @@ int sock_getsockopt(struct socket *sock, int level, int optname,
 		v.val = test_bit(SOCK_PASSSEC, &sock->flags) ? 1 : 0;
 		break;
 
+	case SO_PASSPRIV:
+		v.val = test_bit(SOCK_PASSPRIV, &sock->flags) ? 1 : 0;
+		break;
+
 	case SO_PEERSEC:
 		return security_socket_getpeersec_stream(sock, optval, optlen, len);
 
#2 lore raw
In reply to Casey Schaufler
Quoted 10 lines · Casey Schaufler

Subject: [PATCH] scm: provide full privilege set via SCM_PRIVILEGE

The SCM mechanism currently provides interfaces for delivering the uid/gid and the "security context" (LSM information) of the peer on a UDS socket. All of the security credential information is available, but there is no interface available to obtain it. Further, the existing interfaces require that the user chose between the uid/gid and the context as the existing interfaces are exclusive.

Quoted 4 lines

This patch introduces an additional interface that provides a complete set of security information from the peer credential. No additional work is required to provide the information internally, it is all being passed, just not exposed.

In ascii text?
A bitmap in hex?

Maybe it is just me, but this seems harder to deal with than if the data had been transferred in binary.

Eric

Quoted 177 lines

---

 include/asm-generic/socket.h |    1 +
 include/linux/net.h          |    1 +
 include/linux/socket.h       |    1 +
 include/net/scm.h            |   80 +++++++++++++++++++++++++++++++++++++++++-
 net/core/sock.c              |   11 ++++++
 5 files changed, 93 insertions(+), 1 deletions(-)
include/asm-generic/socket.h +1 -0
index 9a6115e..7aa8e84 100644
@@ -64,4 +64,5 @@
 #define SO_DOMAIN		39
 
 #define SO_RXQ_OVFL             40
+#define SO_PASSPRIV		41
 #endif /* __ASM_GENERIC_SOCKET_H */
include/linux/net.h +1 -0
index 16faa13..159a929 100644
@@ -71,6 +71,7 @@ struct net;
 #define SOCK_NOSPACE		2
 #define SOCK_PASSCRED		3
 #define SOCK_PASSSEC		4
+#define SOCK_PASSPRIV		5
 
 #ifndef ARCH_HAS_SOCKET_TYPES
 /**
include/linux/socket.h +1 -0
index 86b652f..e9cfd68 100644
@@ -147,6 +147,7 @@ static inline struct cmsghdr * cmsg_nxthdr (struct msghdr *__msg, struct cmsghdr
 #define	SCM_RIGHTS	0x01		/* rw: access rights (array of int) */
 #define SCM_CREDENTIALS 0x02		/* rw: struct ucred		*/
 #define SCM_SECURITY	0x03		/* rw: security label		*/
+#define SCM_PRIVILEGES  0x04		/* rw: privilege set		*/
 
 struct ucred {
 	__u32	pid;
include/net/scm.h +79 -1
index 3165650..4b8db21 100644
@@ -101,6 +101,83 @@ static inline void scm_passec(struct socket *sock, struct msghdr *msg, struct sc
 { }
 #endif /* CONFIG_SECURITY_NETWORK */
 
+static __inline__ void scm_passpriv(struct socket *sock, struct msghdr *msg,
+				struct scm_cookie *scm)
+{
+	const struct cred *credp = scm->cred;
+	const struct group_info *gip;
+	char *result;
+	char *cp;
+	int i;
+#ifdef CONFIG_SECURITY_NETWORK
+	char *secdata;
+	u32 seclen;
+	int err;
+#endif /* CONFIG_SECURITY_NETWORK */
+
+	if (!test_bit(SOCK_PASSPRIV, &sock->flags))
+		return;
+
+	gip = credp->group_info;
+
+	/*
+	 * uid + euid + gid + egid + group-list + capabilities
+	 *     + "uid=" + "euid=" + "gid=" + "egid=" + "grps="
+	 *     + "cap-e=" + "cap-p=" + "cap-i="
+	 * 10  + 10   + 10  + 10   + (ngrps * 10) + ecap + pcap + icap
+	 *     + 4 + 5 + 4 + 5 + 5 + 6 + 6 + 6
+	 */
+	i = ((4 + gip->ngroups) * 11) + (3 * (_KERNEL_CAPABILITY_U32S * 8 + 1))
+		+ 41;
+
+#ifdef CONFIG_SECURITY_NETWORK
+	err = security_secid_to_secctx(scm->secid, &secdata, &seclen);
+	if (!err)
+		/*
+		 * " context="
+		 */
+		i += seclen + 10;
+#endif /* CONFIG_SECURITY_NETWORK */
+
+	result = kzalloc(i, GFP_KERNEL);
+	if (result == NULL)
+		return;
+
+	cp = result + sprintf(result, "euid=%d uid=%d egid=%d gid=%d",
+				credp->euid, credp->uid,
+				credp->egid, credp->gid);
+
+	if (gip != NULL && gip->ngroups > 0) {
+		cp += sprintf(cp, " grps=%d", GROUP_AT(gip, 0));
+		for (i = 1 ; i < gip->ngroups; i++)
+			cp += sprintf(cp, ",%d", GROUP_AT(gip, i));
+	}
+
+	cp += sprintf(cp, " cap-e=");
+	CAP_FOR_EACH_U32(i)
+		cp += sprintf(cp, "%08x", credp->cap_effective.cap[i]);
+	cp += sprintf(cp, " cap-p=");
+	CAP_FOR_EACH_U32(i)
+		cp += sprintf(cp, "%08x", credp->cap_permitted.cap[i]);
+	cp += sprintf(cp, " cap-i=");
+	CAP_FOR_EACH_U32(i)
+		cp += sprintf(cp, "%08x", credp->cap_inheritable.cap[i]);
+
+#ifdef CONFIG_SECURITY_NETWORK
+	cp += sprintf(cp, " context=");
+	strncpy(cp, secdata, seclen);
+	cp += seclen;
+	*cp = '\0';
+
+	security_release_secctx(secdata, seclen);
+#endif /* CONFIG_SECURITY_NETWORK */
+
+	put_cmsg(msg, SOL_SOCKET, SCM_PRIVILEGES, strlen(result)+1, result);
+
+	kfree(result);
+}
+
+
 static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 				struct scm_cookie *scm, int flags)
 {
@@ -114,6 +191,8 @@ static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 	if (test_bit(SOCK_PASSCRED, &sock->flags))
 		put_cmsg(msg, SOL_SOCKET, SCM_CREDENTIALS, sizeof(scm->creds), &scm->creds);
 
+	scm_passpriv(sock, msg, scm);
+
 	scm_destroy_cred(scm);
 
 	scm_passec(sock, msg, scm);
@@ -124,6 +203,5 @@ static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 	scm_detach_fds(msg, scm);
 }
 
-
 #endif /* __LINUX_NET_SCM_H */
 
net/core/sock.c +11 -0
index fb60801..f134126 100644
@@ -725,6 +725,13 @@ set_rcvbuf:
 		else
 			clear_bit(SOCK_PASSSEC, &sock->flags);
 		break;
+
+	case SO_PASSPRIV:
+		if (valbool)
+			set_bit(SOCK_PASSPRIV, &sock->flags);
+		else
+			clear_bit(SOCK_PASSPRIV, &sock->flags);
+		break;
 	case SO_MARK:
 		if (!capable(CAP_NET_ADMIN))
 			ret = -EPERM;
@@ -950,6 +957,10 @@ int sock_getsockopt(struct socket *sock, int level, int optname,
 		v.val = test_bit(SOCK_PASSSEC, &sock->flags) ? 1 : 0;
 		break;
 
+	case SO_PASSPRIV:
+		v.val = test_bit(SOCK_PASSPRIV, &sock->flags) ? 1 : 0;
+		break;
+
 	case SO_PEERSEC:
 		return security_socket_getpeersec_stream(sock, optval, optlen, len);
 
#3 lore raw
Quoted 12 lines · Eric W. Biederman
Quoted 10 lines · Casey Schaufler

Subject: [PATCH] scm: provide full privilege set via SCM_PRIVILEGE

The SCM mechanism currently provides interfaces for delivering the uid/gid and the "security context" (LSM information) of the peer on a UDS socket. All of the security credential information is available, but there is no interface available to obtain it. Further, the existing interfaces require that the user chose between the uid/gid and the context as the existing interfaces are exclusive.

Quoted 5 lines
Quoted 4 lines

This patch introduces an additional interface that provides a complete set of security information from the peer credential. No additional work is required to provide the information internally, it is all being passed, just not exposed.

In ascii text?

As is commonly done in /proc interfaces.

Quoted 1 line

A bitmap in hex?

As is done in /proc/<pid>/status. I seriously doubt that anyone would want the kernel doing the capability set to text conversion.

Quoted 2 lines

Maybe it is just me, but this seems harder to deal with than if the data had been transferred in binary.

There are a couple of issues with passing a binary structure in the modern cred case. First is the capability set, which has been proven to grow over time. Sure, it took a while to get past 32 bits, and hopefully will never go beyond 64, but given the long term problems caused by 16 bit uids (some of us still remember) I would hate to get bitten by this in my old age. Second is the LSM specific security context, which may not be there at all and if it is the size will depend on the LSM in use.

There are classic C language techniques for dealing with both of these issues, and I've used them enough times to want to avoid them where possible. This is the same logic that the aforementioned /proc interface implementers have been using for some time. And while there are problems with formatting, passing and parsing a string they pale in comparison to maintaining multiple versions of kernel interface structures that are themselves variable depending on the kernel configuration.

Quoted 179 lines

Eric

Quoted 176 lines

---

 include/asm-generic/socket.h |    1 +
 include/linux/net.h          |    1 +
 include/linux/socket.h       |    1 +
 include/net/scm.h            |   80 +++++++++++++++++++++++++++++++++++++++++-
 net/core/sock.c              |   11 ++++++
 5 files changed, 93 insertions(+), 1 deletions(-)
include/asm-generic/socket.h +1 -0
index 9a6115e..7aa8e84 100644
@@ -64,4 +64,5 @@
 #define SO_DOMAIN		39
 
 #define SO_RXQ_OVFL             40
+#define SO_PASSPRIV		41
 #endif /* __ASM_GENERIC_SOCKET_H */
include/linux/net.h +1 -0
index 16faa13..159a929 100644
@@ -71,6 +71,7 @@ struct net;
 #define SOCK_NOSPACE		2
 #define SOCK_PASSCRED		3
 #define SOCK_PASSSEC		4
+#define SOCK_PASSPRIV		5
 
 #ifndef ARCH_HAS_SOCKET_TYPES
 /**
include/linux/socket.h +1 -0
index 86b652f..e9cfd68 100644
@@ -147,6 +147,7 @@ static inline struct cmsghdr * cmsg_nxthdr (struct msghdr *__msg, struct cmsghdr
 #define	SCM_RIGHTS	0x01		/* rw: access rights (array of int) */
 #define SCM_CREDENTIALS 0x02		/* rw: struct ucred		*/
 #define SCM_SECURITY	0x03		/* rw: security label		*/
+#define SCM_PRIVILEGES  0x04		/* rw: privilege set		*/
 
 struct ucred {
 	__u32	pid;
include/net/scm.h +79 -1
index 3165650..4b8db21 100644
@@ -101,6 +101,83 @@ static inline void scm_passec(struct socket *sock, struct msghdr *msg, struct sc
 { }
 #endif /* CONFIG_SECURITY_NETWORK */
 
+static __inline__ void scm_passpriv(struct socket *sock, struct msghdr *msg,
+				struct scm_cookie *scm)
+{
+	const struct cred *credp = scm->cred;
+	const struct group_info *gip;
+	char *result;
+	char *cp;
+	int i;
+#ifdef CONFIG_SECURITY_NETWORK
+	char *secdata;
+	u32 seclen;
+	int err;
+#endif /* CONFIG_SECURITY_NETWORK */
+
+	if (!test_bit(SOCK_PASSPRIV, &sock->flags))
+		return;
+
+	gip = credp->group_info;
+
+	/*
+	 * uid + euid + gid + egid + group-list + capabilities
+	 *     + "uid=" + "euid=" + "gid=" + "egid=" + "grps="
+	 *     + "cap-e=" + "cap-p=" + "cap-i="
+	 * 10  + 10   + 10  + 10   + (ngrps * 10) + ecap + pcap + icap
+	 *     + 4 + 5 + 4 + 5 + 5 + 6 + 6 + 6
+	 */
+	i = ((4 + gip->ngroups) * 11) + (3 * (_KERNEL_CAPABILITY_U32S * 8 + 1))
+		+ 41;
+
+#ifdef CONFIG_SECURITY_NETWORK
+	err = security_secid_to_secctx(scm->secid, &secdata, &seclen);
+	if (!err)
+		/*
+		 * " context="
+		 */
+		i += seclen + 10;
+#endif /* CONFIG_SECURITY_NETWORK */
+
+	result = kzalloc(i, GFP_KERNEL);
+	if (result == NULL)
+		return;
+
+	cp = result + sprintf(result, "euid=%d uid=%d egid=%d gid=%d",
+				credp->euid, credp->uid,
+				credp->egid, credp->gid);
+
+	if (gip != NULL && gip->ngroups > 0) {
+		cp += sprintf(cp, " grps=%d", GROUP_AT(gip, 0));
+		for (i = 1 ; i < gip->ngroups; i++)
+			cp += sprintf(cp, ",%d", GROUP_AT(gip, i));
+	}
+
+	cp += sprintf(cp, " cap-e=");
+	CAP_FOR_EACH_U32(i)
+		cp += sprintf(cp, "%08x", credp->cap_effective.cap[i]);
+	cp += sprintf(cp, " cap-p=");
+	CAP_FOR_EACH_U32(i)
+		cp += sprintf(cp, "%08x", credp->cap_permitted.cap[i]);
+	cp += sprintf(cp, " cap-i=");
+	CAP_FOR_EACH_U32(i)
+		cp += sprintf(cp, "%08x", credp->cap_inheritable.cap[i]);
+
+#ifdef CONFIG_SECURITY_NETWORK
+	cp += sprintf(cp, " context=");
+	strncpy(cp, secdata, seclen);
+	cp += seclen;
+	*cp = '\0';
+
+	security_release_secctx(secdata, seclen);
+#endif /* CONFIG_SECURITY_NETWORK */
+
+	put_cmsg(msg, SOL_SOCKET, SCM_PRIVILEGES, strlen(result)+1, result);
+
+	kfree(result);
+}
+
+
 static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 				struct scm_cookie *scm, int flags)
 {
@@ -114,6 +191,8 @@ static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 	if (test_bit(SOCK_PASSCRED, &sock->flags))
 		put_cmsg(msg, SOL_SOCKET, SCM_CREDENTIALS, sizeof(scm->creds), &scm->creds);
 
+	scm_passpriv(sock, msg, scm);
+
 	scm_destroy_cred(scm);
 
 	scm_passec(sock, msg, scm);
@@ -124,6 +203,5 @@ static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 	scm_detach_fds(msg, scm);
 }
 
-
 #endif /* __LINUX_NET_SCM_H */
 
net/core/sock.c +11 -0
index fb60801..f134126 100644
@@ -725,6 +725,13 @@ set_rcvbuf:
 		else
 			clear_bit(SOCK_PASSSEC, &sock->flags);
 		break;
+
+	case SO_PASSPRIV:
+		if (valbool)
+			set_bit(SOCK_PASSPRIV, &sock->flags);
+		else
+			clear_bit(SOCK_PASSPRIV, &sock->flags);
+		break;
 	case SO_MARK:
 		if (!capable(CAP_NET_ADMIN))
 			ret = -EPERM;
@@ -950,6 +957,10 @@ int sock_getsockopt(struct socket *sock, int level, int optname,
 		v.val = test_bit(SOCK_PASSSEC, &sock->flags) ? 1 : 0;
 		break;
 
+	case SO_PASSPRIV:
+		v.val = test_bit(SOCK_PASSPRIV, &sock->flags) ? 1 : 0;
+		break;
+
 	case SO_PEERSEC:
 		return security_socket_getpeersec_stream(sock, optval, optlen, len);
 
#4 lore raw
Quoted 19 lines · Casey Schaufler
Quoted 12 lines · Eric W. Biederman
Quoted 10 lines · Casey Schaufler

Subject: [PATCH] scm: provide full privilege set via SCM_PRIVILEGE

The SCM mechanism currently provides interfaces for delivering the uid/gid and the "security context" (LSM information) of the peer on a UDS socket. All of the security credential information is available, but there is no interface available to obtain it. Further, the existing interfaces require that the user chose between the uid/gid and the context as the existing interfaces are exclusive.

Quoted 5 lines
Quoted 4 lines

This patch introduces an additional interface that provides a complete set of security information from the peer credential. No additional work is required to provide the information internally, it is all being passed, just not exposed.

In ascii text?

Quoted 7 lines

As is commonly done in /proc interfaces.

Quoted 1 line

A bitmap in hex?

As is done in /proc/<pid>/status. I seriously doubt that anyone would want the kernel doing the capability set to text conversion.

But when you have a perfectly good binary interface when reducing the encoding efficiency for effectively no gain.

Quoted 12 lines
Quoted 2 lines

Maybe it is just me, but this seems harder to deal with than if the data had been transferred in binary.

There are a couple of issues with passing a binary structure in the modern cred case. First is the capability set, which has been proven to grow over time. Sure, it took a while to get past 32 bits, and hopefully will never go beyond 64, but given the long term problems caused by 16 bit uids (some of us still remember) I would hate to get bitten by this in my old age. Second is the LSM specific security context, which may not be there at all and if it is the size will depend on the LSM in use.

Sure but you have to use an interface that properly handles variable length binary data, to get to the string. It feels like you are violating the even more classic one value per file rule.

Maybe I am missing something but is there any reason you can't have multiple cmsg types?

Quoted 9 lines

There are classic C language techniques for dealing with both of these issues, and I've used them enough times to want to avoid them where possible. This is the same logic that the aforementioned /proc interface implementers have been using for some time. And while there are problems with formatting, passing and parsing a string they pale in comparison to maintaining multiple versions of kernel interface structures that are themselves variable depending on the kernel configuration.

If you are worrying about variable size structures that vary depending on kernel configuration I pretty certain you are doing it wrong.

The use of sprintf (not snprintf) and the crazy size computation needed for your string also worries me. That part of the implementation appears to be just asking for trouble.

Having a giant function like your scm_passpriv inline in include/net/scm.h also seems very questionable.

I think there is a real impedance mismatch here between the interface you are using and the way you are returning the data.

There is a show stopper bug. You don't translate uid/gid in the receivers user namespace so passing a message between two processes in different user namespaces can pass deceptive credentials. Given that the reason we have the struct cred on unix domain sockets in the first place is that we handled the user namespace conversion issues so we could cross namespaces without security issues not handling that case in a new scm cred is just inexcusable.

On that note Serge and I are slowly working to get credentials to be namespace local as well, so shortly the credentials will also need to be translated as well as just uid's and gid's.

The scm->secid is if I understand correctly a per packet label. Is that what you want here? I thought you were interested in the information off of struct cred.

In principle returning all of the credential should be fine. In practice I think this patch has a lot of poorly thought through details that will be a nightmare to maintain in practice.

Eric

Quoted 177 lines
Quoted 177 lines
Quoted 176 lines

---

 include/asm-generic/socket.h |    1 +
 include/linux/net.h          |    1 +
 include/linux/socket.h       |    1 +
 include/net/scm.h            |   80 +++++++++++++++++++++++++++++++++++++++++-
 net/core/sock.c              |   11 ++++++
 5 files changed, 93 insertions(+), 1 deletions(-)
include/asm-generic/socket.h +1 -0
index 9a6115e..7aa8e84 100644
@@ -64,4 +64,5 @@
 #define SO_DOMAIN		39
 
 #define SO_RXQ_OVFL             40
+#define SO_PASSPRIV		41
 #endif /* __ASM_GENERIC_SOCKET_H */
include/linux/net.h +1 -0
index 16faa13..159a929 100644
@@ -71,6 +71,7 @@ struct net;
 #define SOCK_NOSPACE		2
 #define SOCK_PASSCRED		3
 #define SOCK_PASSSEC		4
+#define SOCK_PASSPRIV		5
 
 #ifndef ARCH_HAS_SOCKET_TYPES
 /**
include/linux/socket.h +1 -0
index 86b652f..e9cfd68 100644
@@ -147,6 +147,7 @@ static inline struct cmsghdr * cmsg_nxthdr (struct msghdr *__msg, struct cmsghdr
 #define	SCM_RIGHTS	0x01		/* rw: access rights (array of int) */
 #define SCM_CREDENTIALS 0x02		/* rw: struct ucred		*/
 #define SCM_SECURITY	0x03		/* rw: security label		*/
+#define SCM_PRIVILEGES  0x04		/* rw: privilege set		*/
 
 struct ucred {
 	__u32	pid;
include/net/scm.h +79 -1
index 3165650..4b8db21 100644
@@ -101,6 +101,83 @@ static inline void scm_passec(struct socket *sock, struct msghdr *msg, struct sc
 { }
 #endif /* CONFIG_SECURITY_NETWORK */
 
+static __inline__ void scm_passpriv(struct socket *sock, struct msghdr *msg,
+				struct scm_cookie *scm)
+{
+	const struct cred *credp = scm->cred;
+	const struct group_info *gip;
+	char *result;
+	char *cp;
+	int i;
+#ifdef CONFIG_SECURITY_NETWORK
+	char *secdata;
+	u32 seclen;
+	int err;
+#endif /* CONFIG_SECURITY_NETWORK */
+
+	if (!test_bit(SOCK_PASSPRIV, &sock->flags))
+		return;
+
+	gip = credp->group_info;
+
+	/*
+	 * uid + euid + gid + egid + group-list + capabilities
+	 *     + "uid=" + "euid=" + "gid=" + "egid=" + "grps="
+	 *     + "cap-e=" + "cap-p=" + "cap-i="
+	 * 10  + 10   + 10  + 10   + (ngrps * 10) + ecap + pcap + icap
+	 *     + 4 + 5 + 4 + 5 + 5 + 6 + 6 + 6
+	 */
+	i = ((4 + gip->ngroups) * 11) + (3 * (_KERNEL_CAPABILITY_U32S * 8 + 1))
+		+ 41;
+
+#ifdef CONFIG_SECURITY_NETWORK
+	err = security_secid_to_secctx(scm->secid, &secdata, &seclen);
+	if (!err)
+		/*
+		 * " context="
+		 */
+		i += seclen + 10;
+#endif /* CONFIG_SECURITY_NETWORK */
+
+	result = kzalloc(i, GFP_KERNEL);
+	if (result == NULL)
+		return;
+
+	cp = result + sprintf(result, "euid=%d uid=%d egid=%d gid=%d",
+				credp->euid, credp->uid,
+				credp->egid, credp->gid);
+
+	if (gip != NULL && gip->ngroups > 0) {
+		cp += sprintf(cp, " grps=%d", GROUP_AT(gip, 0));
+		for (i = 1 ; i < gip->ngroups; i++)
+			cp += sprintf(cp, ",%d", GROUP_AT(gip, i));
+	}
+
+	cp += sprintf(cp, " cap-e=");
+	CAP_FOR_EACH_U32(i)
+		cp += sprintf(cp, "%08x", credp->cap_effective.cap[i]);
+	cp += sprintf(cp, " cap-p=");
+	CAP_FOR_EACH_U32(i)
+		cp += sprintf(cp, "%08x", credp->cap_permitted.cap[i]);
+	cp += sprintf(cp, " cap-i=");
+	CAP_FOR_EACH_U32(i)
+		cp += sprintf(cp, "%08x", credp->cap_inheritable.cap[i]);
+
+#ifdef CONFIG_SECURITY_NETWORK
+	cp += sprintf(cp, " context=");
+	strncpy(cp, secdata, seclen);
+	cp += seclen;
+	*cp = '\0';
+
+	security_release_secctx(secdata, seclen);
+#endif /* CONFIG_SECURITY_NETWORK */
+
+	put_cmsg(msg, SOL_SOCKET, SCM_PRIVILEGES, strlen(result)+1, result);
+
+	kfree(result);
+}
+
+
 static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 				struct scm_cookie *scm, int flags)
 {
@@ -114,6 +191,8 @@ static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 	if (test_bit(SOCK_PASSCRED, &sock->flags))
 		put_cmsg(msg, SOL_SOCKET, SCM_CREDENTIALS, sizeof(scm->creds), &scm->creds);
 
+	scm_passpriv(sock, msg, scm);
+
 	scm_destroy_cred(scm);
 
 	scm_passec(sock, msg, scm);
@@ -124,6 +203,5 @@ static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 	scm_detach_fds(msg, scm);
 }
 
-
 #endif /* __LINUX_NET_SCM_H */
 
net/core/sock.c +11 -0
index fb60801..f134126 100644
@@ -725,6 +725,13 @@ set_rcvbuf:
 		else
 			clear_bit(SOCK_PASSSEC, &sock->flags);
 		break;
+
+	case SO_PASSPRIV:
+		if (valbool)
+			set_bit(SOCK_PASSPRIV, &sock->flags);
+		else
+			clear_bit(SOCK_PASSPRIV, &sock->flags);
+		break;
 	case SO_MARK:
 		if (!capable(CAP_NET_ADMIN))
 			ret = -EPERM;
@@ -950,6 +957,10 @@ int sock_getsockopt(struct socket *sock, int level, int optname,
 		v.val = test_bit(SOCK_PASSSEC, &sock->flags) ? 1 : 0;
 		break;
 
+	case SO_PASSPRIV:
+		v.val = test_bit(SOCK_PASSPRIV, &sock->flags) ? 1 : 0;
+		break;
+
 	case SO_PEERSEC:
 		return security_socket_getpeersec_stream(sock, optval, optlen, len);
 
#5 lore raw
Quoted 3 lines · Eric W. Biederman
Quoted 1 line · Casey Schaufler

On 1/18/2011 9:45 PM, Eric W. Biederman wrote:

Quite a bit of work to do here, it would appear.

Quoted 17 lines
Quoted 17 lines
Quoted 17 lines
Quoted 10 lines · Casey Schaufler

Subject: [PATCH] scm: provide full privilege set via SCM_PRIVILEGE

The SCM mechanism currently provides interfaces for delivering the uid/gid and the "security context" (LSM information) of the peer on a UDS socket. All of the security credential information is available, but there is no interface available to obtain it. Further, the existing interfaces require that the user chose between the uid/gid and the context as the existing interfaces are exclusive.

Quoted 4 lines

This patch introduces an additional interface that provides a complete set of security information from the peer credential. No additional work is required to provide the information internally, it is all being passed, just not exposed.

In ascii text?

Quoted 8 lines
Quoted 6 lines

As is commonly done in /proc interfaces.

Quoted 1 line

A bitmap in hex?

As is done in /proc/<pid>/status. I seriously doubt that anyone would want the kernel doing the capability set to text conversion.

But when you have a perfectly good binary interface when reducing the encoding efficiency for effectively no gain.

Problem is that with a group list, capability sets, and LSM "context" I don't have a perfectly good binary interface.

Quoted 11 lines
Quoted 11 lines
Quoted 2 lines

Maybe it is just me, but this seems harder to deal with than if the data had been transferred in binary.

There are a couple of issues with passing a binary structure in the modern cred case. First is the capability set, which has been proven to grow over time. Sure, it took a while to get past 32 bits, and hopefully will never go beyond 64, but given the long term problems caused by 16 bit uids (some of us still remember) I would hate to get bitten by this in my old age. Second is the LSM specific security context, which may not be there at all and if it is the size will depend on the LSM in use.

Quoted 3 lines

Sure but you have to use an interface that properly handles variable length binary data, to get to the string. It feels like you are violating the even more classic one value per file rule.

This way leaves all parsing to user space, and uses well understood facilities to create something that isn't going to require internal counters and the like.

Quoted 2 lines

Maybe I am missing something but is there any reason you can't have multiple cmsg types?

Sure, there could be a cmsg for each value, but if you know that you're going to always use them all, why make them separate?

Quoted 9 lines
Quoted 9 lines

There are classic C language techniques for dealing with both of these issues, and I've used them enough times to want to avoid them where possible. This is the same logic that the aforementioned /proc interface implementers have been using for some time. And while there are problems with formatting, passing and parsing a string they pale in comparison to maintaining multiple versions of kernel interface structures that are themselves variable depending on the kernel configuration.

Quoted 3 lines

If you are worrying about variable size structures that vary depending on kernel configuration I pretty certain you are doing it wrong.

I'm seriously more concerned with the maintainability of cap sets. I saw the pain level when they went to 2x32bit and any structure that includes a cap set, especially a user visible interface, is going to be an issue. Not to mention that there's a group list in there as well.

Quoted 3 lines

The use of sprintf (not snprintf) and the crazy size computation needed for your string also worries me. That part of the implementation appears to be just asking for trouble.

snprintf would be a simple change. The computation is just arithmetic.

Quoted 6 lines

Having a giant function like your scm_passpriv inline in include/net/scm.h also seems very questionable.

I think there is a real impedance mismatch here between the interface you are using and the way you are returning the data.

Quoted 7 lines

There is a show stopper bug. You don't translate uid/gid in the receivers user namespace so passing a message between two processes in different user namespaces can pass deceptive credentials. Given that the reason we have the struct cred on unix domain sockets in the first place is that we handled the user namespace conversion issues so we could cross namespaces without security issues not handling that case in a new scm cred is just inexcusable.

The issues with uid and gid translation appear to be easy to fix, assuming that the cred_to_ucred() function provides a reasonable example.

Quoted 3 lines

On that note Serge and I are slowly working to get credentials to be namespace local as well, so shortly the credentials will also need to be translated as well as just uid's and gid's.

Oh dear. Does that include capability sets and LSM blobs?

Quoted 3 lines

The scm->secid is if I understand correctly a per packet label. Is that what you want here? I thought you were interested in the information off of struct cred.

The intent is to pull in the access control information so that the (trusted) application can make decisions based on its own set of criteria. The kernel is making access checks (in SELinux and Smack at any rate) based on the packet label, but it might make sense to provide both.

Quoted 184 lines

In principle returning all of the credential should be fine. In practice I think this patch has a lot of poorly thought through details that will be a nightmare to maintain in practice.

Eric

Quoted 176 lines
Quoted 176 lines
Quoted 176 lines

---

 include/asm-generic/socket.h |    1 +
 include/linux/net.h          |    1 +
 include/linux/socket.h       |    1 +
 include/net/scm.h            |   80 +++++++++++++++++++++++++++++++++++++++++-
 net/core/sock.c              |   11 ++++++
 5 files changed, 93 insertions(+), 1 deletions(-)
include/asm-generic/socket.h +1 -0
index 9a6115e..7aa8e84 100644
@@ -64,4 +64,5 @@
 #define SO_DOMAIN		39
 
 #define SO_RXQ_OVFL             40
+#define SO_PASSPRIV		41
 #endif /* __ASM_GENERIC_SOCKET_H */
include/linux/net.h +1 -0
index 16faa13..159a929 100644
@@ -71,6 +71,7 @@ struct net;
 #define SOCK_NOSPACE		2
 #define SOCK_PASSCRED		3
 #define SOCK_PASSSEC		4
+#define SOCK_PASSPRIV		5
 
 #ifndef ARCH_HAS_SOCKET_TYPES
 /**
include/linux/socket.h +1 -0
index 86b652f..e9cfd68 100644
@@ -147,6 +147,7 @@ static inline struct cmsghdr * cmsg_nxthdr (struct msghdr *__msg, struct cmsghdr
 #define	SCM_RIGHTS	0x01		/* rw: access rights (array of int) */
 #define SCM_CREDENTIALS 0x02		/* rw: struct ucred		*/
 #define SCM_SECURITY	0x03		/* rw: security label		*/
+#define SCM_PRIVILEGES  0x04		/* rw: privilege set		*/
 
 struct ucred {
 	__u32	pid;
include/net/scm.h +79 -1
index 3165650..4b8db21 100644
@@ -101,6 +101,83 @@ static inline void scm_passec(struct socket *sock, struct msghdr *msg, struct sc
 { }
 #endif /* CONFIG_SECURITY_NETWORK */
 
+static __inline__ void scm_passpriv(struct socket *sock, struct msghdr *msg,
+				struct scm_cookie *scm)
+{
+	const struct cred *credp = scm->cred;
+	const struct group_info *gip;
+	char *result;
+	char *cp;
+	int i;
+#ifdef CONFIG_SECURITY_NETWORK
+	char *secdata;
+	u32 seclen;
+	int err;
+#endif /* CONFIG_SECURITY_NETWORK */
+
+	if (!test_bit(SOCK_PASSPRIV, &sock->flags))
+		return;
+
+	gip = credp->group_info;
+
+	/*
+	 * uid + euid + gid + egid + group-list + capabilities
+	 *     + "uid=" + "euid=" + "gid=" + "egid=" + "grps="
+	 *     + "cap-e=" + "cap-p=" + "cap-i="
+	 * 10  + 10   + 10  + 10   + (ngrps * 10) + ecap + pcap + icap
+	 *     + 4 + 5 + 4 + 5 + 5 + 6 + 6 + 6
+	 */
+	i = ((4 + gip->ngroups) * 11) + (3 * (_KERNEL_CAPABILITY_U32S * 8 + 1))
+		+ 41;
+
+#ifdef CONFIG_SECURITY_NETWORK
+	err = security_secid_to_secctx(scm->secid, &secdata, &seclen);
+	if (!err)
+		/*
+		 * " context="
+		 */
+		i += seclen + 10;
+#endif /* CONFIG_SECURITY_NETWORK */
+
+	result = kzalloc(i, GFP_KERNEL);
+	if (result == NULL)
+		return;
+
+	cp = result + sprintf(result, "euid=%d uid=%d egid=%d gid=%d",
+				credp->euid, credp->uid,
+				credp->egid, credp->gid);
+
+	if (gip != NULL && gip->ngroups > 0) {
+		cp += sprintf(cp, " grps=%d", GROUP_AT(gip, 0));
+		for (i = 1 ; i < gip->ngroups; i++)
+			cp += sprintf(cp, ",%d", GROUP_AT(gip, i));
+	}
+
+	cp += sprintf(cp, " cap-e=");
+	CAP_FOR_EACH_U32(i)
+		cp += sprintf(cp, "%08x", credp->cap_effective.cap[i]);
+	cp += sprintf(cp, " cap-p=");
+	CAP_FOR_EACH_U32(i)
+		cp += sprintf(cp, "%08x", credp->cap_permitted.cap[i]);
+	cp += sprintf(cp, " cap-i=");
+	CAP_FOR_EACH_U32(i)
+		cp += sprintf(cp, "%08x", credp->cap_inheritable.cap[i]);
+
+#ifdef CONFIG_SECURITY_NETWORK
+	cp += sprintf(cp, " context=");
+	strncpy(cp, secdata, seclen);
+	cp += seclen;
+	*cp = '\0';
+
+	security_release_secctx(secdata, seclen);
+#endif /* CONFIG_SECURITY_NETWORK */
+
+	put_cmsg(msg, SOL_SOCKET, SCM_PRIVILEGES, strlen(result)+1, result);
+
+	kfree(result);
+}
+
+
 static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 				struct scm_cookie *scm, int flags)
 {
@@ -114,6 +191,8 @@ static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 	if (test_bit(SOCK_PASSCRED, &sock->flags))
 		put_cmsg(msg, SOL_SOCKET, SCM_CREDENTIALS, sizeof(scm->creds), &scm->creds);
 
+	scm_passpriv(sock, msg, scm);
+
 	scm_destroy_cred(scm);
 
 	scm_passec(sock, msg, scm);
@@ -124,6 +203,5 @@ static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 	scm_detach_fds(msg, scm);
 }
 
-
 #endif /* __LINUX_NET_SCM_H */
 
net/core/sock.c +11 -0
index fb60801..f134126 100644
@@ -725,6 +725,13 @@ set_rcvbuf:
 		else
 			clear_bit(SOCK_PASSSEC, &sock->flags);
 		break;
+
+	case SO_PASSPRIV:
+		if (valbool)
+			set_bit(SOCK_PASSPRIV, &sock->flags);
+		else
+			clear_bit(SOCK_PASSPRIV, &sock->flags);
+		break;
 	case SO_MARK:
 		if (!capable(CAP_NET_ADMIN))
 			ret = -EPERM;
@@ -950,6 +957,10 @@ int sock_getsockopt(struct socket *sock, int level, int optname,
 		v.val = test_bit(SOCK_PASSSEC, &sock->flags) ? 1 : 0;
 		break;
 
+	case SO_PASSPRIV:
+		v.val = test_bit(SOCK_PASSPRIV, &sock->flags) ? 1 : 0;
+		break;
+
 	case SO_PEERSEC:
 		return security_socket_getpeersec_stream(sock, optval, optlen, len);
 
#6 lore raw

PATCH 1/1 net: provide capability and group sets via SCM

In reply to Casey Schaufler

Subject: [PATCH] net: provide group lists and capability set via CMSG

Provide the namespace converted group list of the peer process using the SCM mechanism. Provide the capability set of the peer process. The capability set is not namespace converted on the assumption that there is no such conversion available or required.

---
 include/asm-generic/socket.h |    4 ++
 include/linux/net.h          |    2 +
 include/linux/socket.h       |   18 +++++++++
 include/net/scm.h            |   33 +++++++++++++++++
 net/core/sock.c              |   82 ++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 139 insertions(+), 0 deletions(-)
include/asm-generic/socket.h +4 -0
index 9a6115e..fc2d609 100644
@@ -64,4 +64,8 @@
 #define SO_DOMAIN		39
 
 #define SO_RXQ_OVFL             40
+
+#define SO_PASSGROUPS		41
+#define SO_PASSCAPS		42
+
 #endif /* __ASM_GENERIC_SOCKET_H */
include/linux/net.h +2 -0
index 16faa13..929e241 100644
@@ -71,6 +71,8 @@ struct net;
 #define SOCK_NOSPACE		2
 #define SOCK_PASSCRED		3
 #define SOCK_PASSSEC		4
+#define SOCK_PASSGROUPS		5
+#define SOCK_PASSCAPS		6
 
 #ifndef ARCH_HAS_SOCKET_TYPES
 /**
include/linux/socket.h +18 -0
index edbb1d0..63f64f0 100644
@@ -23,6 +23,7 @@ struct __kernel_sockaddr_storage {
 #include <linux/uio.h>			/* iovec support		*/
 #include <linux/types.h>		/* pid_t			*/
 #include <linux/compiler.h>		/* __user			*/
+#include <linux/capability.h>		/* _KERNEL_CAPABILITY_U32S	*/
 
 struct pid;
 struct cred;
@@ -145,6 +146,8 @@ static inline struct cmsghdr * cmsg_nxthdr (struct msghdr *__msg, struct cmsghdr
 #define	SCM_RIGHTS	0x01		/* rw: access rights (array of int) */
 #define SCM_CREDENTIALS 0x02		/* rw: struct ucred		*/
 #define SCM_SECURITY	0x03		/* rw: security label		*/
+#define SCM_GROUPS	0x04		/* rw: group list		*/
+#define SCM_CAPS	0x05		/* rw: capability set		*/
 
 struct ucred {
 	__u32	pid;
@@ -152,6 +155,16 @@ struct ucred {
 	__u32	gid;
 };
 
+struct scm_groups {
+	__u32	count;
+	__u32	gids[0];
+};
+
+struct scm_capabilities {
+	struct __user_cap_data_struct caps[_KERNEL_CAPABILITY_U32S];
+};
+
+
 /* Supported address families. */
 #define AF_UNSPEC	0
 #define AF_UNIX		1	/* Unix domain sockets 		*/
@@ -313,6 +326,11 @@ struct ucred {
 #define IPX_TYPE	1
 
 extern void cred_to_ucred(struct pid *pid, const struct cred *cred, struct ucred *ucred);
+extern void cred_to_scm_groups(const struct cred *cred,
+				struct scm_groups **sgp, int *size);
+extern void cred_to_scm_capabilities(const struct cred *cred,
+					struct scm_capabilities **scp,
+					int *size);
 
 extern int memcpy_fromiovec(unsigned char *kdata, struct iovec *iov, int len);
 extern int memcpy_fromiovecend(unsigned char *kdata, const struct iovec *iov,
include/net/scm.h +33 -0
index 745460f..b263867 100644
@@ -102,6 +102,36 @@ static inline void scm_passec(struct socket *sock, struct msghdr *msg, struct sc
 { }
 #endif /* CONFIG_SECURITY_NETWORK */
 
+static inline void scm_passgroups(struct socket *sock, struct msghdr *msg,
+					struct scm_cookie *scm)
+{
+	struct scm_groups *data;
+	int size = 0;
+
+	if (test_bit(SOCK_PASSGROUPS, &sock->flags)) {
+		cred_to_scm_groups(scm->cred, &data, &size);
+		if (size) {
+			put_cmsg(msg, SOL_SOCKET, SCM_GROUPS, size, data);
+			kfree(data);
+		}
+	}
+}
+
+static inline void scm_passcaps(struct socket *sock, struct msghdr *msg,
+				struct scm_cookie *scm)
+{
+	struct scm_capabilities *data;
+	int size = 0;
+
+	if (test_bit(SOCK_PASSCAPS, &sock->flags)) {
+		cred_to_scm_capabilities(scm->cred, &data, &size);
+		if (size) {
+			put_cmsg(msg, SOL_SOCKET, SCM_CAPS, size, data);
+			kfree(data);
+		}
+	}
+}
+
 static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 				struct scm_cookie *scm, int flags)
 {
@@ -115,6 +145,9 @@ static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
 	if (test_bit(SOCK_PASSCRED, &sock->flags))
 		put_cmsg(msg, SOL_SOCKET, SCM_CREDENTIALS, sizeof(scm->creds), &scm->creds);
 
+	scm_passgroups(sock, msg, scm);
+	scm_passcaps(sock, msg, scm);
+
 	scm_destroy_cred(scm);
 
 	scm_passec(sock, msg, scm);
net/core/sock.c +82 -0

100 lines not on this page · raw message

maintainer
#7 lore raw

Re: [PATCH] net: provide capability and group sets via SCM

From: Casey Schaufler <[email protected]>
Date: Tue, 08 Feb 2011 14:28:27 -0800

Quoted 10 lines

Subject: [PATCH] net: provide group lists and capability set via CMSG

Provide the namespace converted group list of the peer process using the SCM mechanism. Provide the capability set of the peer process. The capability set is not namespace converted on the assumption that there is no such conversion available or required.

You can't just hit the asm-generic header, you have to also hit all of the architectures that don't use the asm-generic header, including sparc, powerpc, mips, s390, arm, alpha, cris, frv, h8300, ia64, m32r, m68k, m68knommu, mn10300, parisc, and xtensa.

#8 lore raw

Re: [PATCH] net: provide capability and group sets via SCM

Quoted 11 lines · David Miller

From: Casey Schaufler <[email protected]>
Date: Tue, 08 Feb 2011 14:28:27 -0800

Quoted 8 lines

Subject: [PATCH] net: provide group lists and capability set via CMSG

Provide the namespace converted group list of the peer process using the SCM mechanism. Provide the capability set of the peer process. The capability set is not namespace converted on the assumption that there is no such conversion available or required.

Quoted 5 lines
Quoted 1 line

You can't just hit the asm-generic header, you have to also hit all of the architectures that don't use the asm-generic header, including sparc, powerpc, mips, s390, arm, alpha, cris, frv, h8300, ia64, m32r, m68k, m68knommu, mn10300, parisc, and xtensa.

OK, I was afraid of that. Updated version will include those. Thank you.

Quoted 7 lines

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/