dscp ranges
----------
This property controls which dscp values the processes in a cgroup are
allowed to use. A process in a cgroup will receive an EACCES error if it
tries to do any of these things:
* set a socket's IP_TOS option to a value whose dscp field (bits 7:2) is
outside the range
* use a socket to send a message in which the IP_TOS ancillary data is
set to a value whose dscp field is outside the range
This property is exposed to userspace through the 'net.dscp_ranges' file,
similar to the bind and listen port ranges.
Tested: wrote python to attempt to setsockopt the IP_TOS option to a
value with an out-of-range dscp field, and expect a failure
Signed-off-by: Anoop Naravaram <anaravaram@xxxxxxxxxx>
---
Documentation/cgroup-v1/net.txt | 14 ++++++++++++++
include/net/net_cgroup.h | 6 ++++++
net/core/net_cgroup.c | 34 ++++++++++++++++++++++++++++++++--
net/ipv4/ip_sockglue.c | 13 +++++++++++++
net/ipv6/datagram.c | 9 +++++++++
net/ipv6/ipv6_sockglue.c | 8 ++++++++
6 files changed, 82 insertions(+), 2 deletions(-)
diff --git a/Documentation/cgroup-v1/net.txt b/Documentation/cgroup-v1/net.txt
index a14fd1c..ea2f1db 100644
--- a/Documentation/cgroup-v1/net.txt
+++ b/Documentation/cgroup-v1/net.txt
@@ -30,6 +30,20 @@ This property is exposed to userspace through the 'net.listen_port_ranges' file,
as ranges of ports that the processes can listen on (as described in the HOW TO
INTERACT WITH RANGES FILES section).
+dscp ranges
+-----------
+This property controls which dscp values the processes in a cgroup are
+allowed to use. A process in a cgroup will receive an EACCES error if it
+tries to do any of these things:
+* set a socket's IP_TOS option to a value whose dscp field (bits 7:2) is
+ outside the range
+* use a socket to send a message in which the IP_TOS ancillary data is
+ set to a value whose dscp field is outside the range
+
+This property is exposed to userspace through the 'net.dscp_ranges' file, as
+ranges of dscp values that the process can use (as described in the HOW TO
+INTERACT WITH RANGES FILES section).
+
udp port usage and limit
------------------------
This property controls the limit of udp ports that can be used by the
diff --git a/include/net/net_cgroup.h b/include/net/net_cgroup.h
index 25a9def..d89e98d 100644
--- a/include/net/net_cgroup.h
+++ b/include/net/net_cgroup.h
@@ -23,6 +23,7 @@
enum {
NETCG_LISTEN_RANGES,
NETCG_BIND_RANGES,
+ NETCG_DSCP_RANGES,
NETCG_NUM_RANGE_TYPES
};
@@ -73,6 +74,7 @@ struct net_cgroup {
bool net_cgroup_bind_allowed(u16 port);
bool net_cgroup_listen_allowed(u16 port);
+bool net_cgroup_dscp_allowed(u8 dscp);
bool net_cgroup_acquire_udp_port(void);
void net_cgroup_release_udp_port(void);
@@ -85,6 +87,10 @@ static inline bool net_cgroup_listen_allowed(u16 port)
{
return true;
}
+static inline bool net_cgroup_dscp_allowed(u8 dscp)
+{
+ return true;
+}
static inline bool net_cgroup_acquire_udp_port(void)
{
return true;
diff --git a/net/core/net_cgroup.c b/net/core/net_cgroup.c
index 2f58e13..73dc5e7 100644
--- a/net/core/net_cgroup.c
+++ b/net/core/net_cgroup.c
@@ -21,6 +21,9 @@
#define MIN_PORT_VALUE 0
#define MAX_PORT_VALUE 65535
+#define MIN_DSCP_VALUE 0
+#define MAX_DSCP_VALUE 63
+
/* Deriving MAX_ENTRIES from MAX_WRITE_SIZE as a rough estimate */
#define MAX_ENTRIES ((MAX_WRITE_SIZE - offsetof(struct net_ranges, range)) / \
BYTES_PER_ENTRY)
@@ -161,7 +164,10 @@ cgrp_css_alloc(struct cgroup_subsys_state *parent_css)
MIN_PORT_VALUE, MAX_PORT_VALUE) ||
alloc_init_net_ranges(
&netcg->whitelists[NETCG_LISTEN_RANGES],
- MIN_PORT_VALUE, MAX_PORT_VALUE)) {
+ MIN_PORT_VALUE, MAX_PORT_VALUE) ||
+ alloc_init_net_ranges(
+ &netcg->whitelists[NETCG_DSCP_RANGES],
+ MIN_DSCP_VALUE, MAX_DSCP_VALUE)) {
free_net_cgroup(netcg);
/* if any of these cause an error, return ENOMEM */
return ERR_PTR(-ENOMEM);
@@ -178,7 +184,11 @@ cgrp_css_alloc(struct cgroup_subsys_state *parent_css)
alloc_copy_net_ranges(
&netcg->whitelists[NETCG_LISTEN_RANGES],
MIN_PORT_VALUE, MAX_PORT_VALUE,
- &parent_netcg->whitelists[NETCG_LISTEN_RANGES])) {
+ &parent_netcg->whitelists[NETCG_LISTEN_RANGES]) ||
+ alloc_copy_net_ranges(
+ &netcg->whitelists[NETCG_DSCP_RANGES],
+ MIN_DSCP_VALUE, MAX_DSCP_VALUE,
+ &parent_netcg->whitelists[NETCG_DSCP_RANGES])) {
free_net_cgroup(netcg);
/* if any of these cause an error, return ENOMEM */
return ERR_PTR(-ENOMEM);
@@ -237,6 +247,12 @@ bool net_cgroup_listen_allowed(u16 port)
}
EXPORT_SYMBOL_GPL(net_cgroup_listen_allowed);
+bool net_cgroup_dscp_allowed(u8 dscp)
+{
+ return net_cgroup_value_allowed(dscp, NETCG_DSCP_RANGES);
+}
+EXPORT_SYMBOL_GPL(net_cgroup_dscp_allowed);
+
static s64 net_udp_read_s64(struct cgroup_subsys_state *css, struct cftype *cft)
{
struct net_cgroup *netcg = css_to_net_cgroup(css);
@@ -634,6 +650,20 @@ static struct cftype ss_files[] = {
.max_write_len = MAX_WRITE_SIZE,
},
{
+ .name = "dscp_ranges",
+ .flags = CFTYPE_ONLY_ON_ROOT,
+ .seq_show = net_read_ranges,
+ .private = NETCG_DSCP_RANGES,
+ },
+ {
+ .name = "dscp_ranges",
+ .flags = CFTYPE_NOT_ON_ROOT,
+ .seq_show = net_read_ranges,
+ .write = net_write_ranges,
+ .private = NETCG_DSCP_RANGES,
+ .max_write_len = MAX_WRITE_SIZE,
+ },
+ {
.name = "udp_limit",
.flags = CFTYPE_ONLY_ON_ROOT,
.read_s64 = net_udp_read_s64,
diff --git a/net/ipv4/ip_sockglue.c b/net/ipv4/ip_sockglue.c
index 71a52f4d..71a4297 100644
--- a/net/ipv4/ip_sockglue.c
+++ b/net/ipv4/ip_sockglue.c
@@ -42,6 +42,7 @@
#include <net/transp_v6.h>
#endif
#include <net/ip_fib.h>
+#include <net/net_cgroup.h>
#include <linux/errqueue.h>
#include <asm/uaccess.h>
@@ -289,6 +290,11 @@ int ip_cmsg_send(struct sock *sk, struct msghdr *msg, struct ipcm_cookie *ipc,
val = *(int *)CMSG_DATA(cmsg);
if (val < 0 || val > 255)
return -EINVAL;
+ /* val is 8-bit tos, we need to rightshift 2 to get the
+ * 6-bit dscp field
+ */
+ if (!net_cgroup_dscp_allowed(val >> 2))
+ return -EACCES;
ipc->tos = val;
ipc->priority = rt_tos2priority(ipc->tos);
break;
@@ -727,6 +733,13 @@ static int do_ip_setsockopt(struct sock *sk, int level,
val &= ~INET_ECN_MASK;
val |= inet->tos & INET_ECN_MASK;
}
+ /* val is 8-bit tos, we need to rightshift 2 to get the
+ * 6-bit dscp field
+ */
+ if (!net_cgroup_dscp_allowed(val >> 2)) {
+ err = -EACCES;
+ break;
+ }
if (inet->tos != val) {
inet->tos = val;
sk->sk_priority = rt_tos2priority(val);
diff --git a/net/ipv6/datagram.c b/net/ipv6/datagram.c
index 37874e2..9053b83 100644
--- a/net/ipv6/datagram.c
+++ b/net/ipv6/datagram.c
@@ -34,6 +34,7 @@
#include <linux/errqueue.h>
#include <asm/uaccess.h>
+#include <net/net_cgroup.h>
static bool ipv6_mapped_addr_any(const struct in6_addr *a)
{
@@ -973,6 +974,14 @@ int ip6_datagram_send_ctl(struct net *net, struct sock *sk,
if (tc < -1 || tc > 0xff)
goto exit_f;
+ /* tc is 8-bit tclass, we need to rightshift 2 to get
+ * the 6-bit dscp field
+ */
+ if (!net_cgroup_dscp_allowed(tc >> 2)) {
+ err = -EACCES;
+ goto exit_f;
+ }
+
err = 0;
ipc6->tclass = tc;
diff --git a/net/ipv6/ipv6_sockglue.c b/net/ipv6/ipv6_sockglue.c
index a9895e1..eac3f88 100644
--- a/net/ipv6/ipv6_sockglue.c
+++ b/net/ipv6/ipv6_sockglue.c
@@ -52,6 +52,7 @@
#include <net/udplite.h>
#include <net/xfrm.h>
#include <net/compat.h>
+#include <net/net_cgroup.h>
#include <asm/uaccess.h>
@@ -339,6 +340,13 @@ static int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
/* RFC 3542, 6.5: default traffic class of 0x0 */
if (val == -1)
val = 0;
+ /* val is 8-bit tclass, we need to rightshift 2 to get the 6-bit
+ * dscp field
+ */
+ if (!net_cgroup_dscp_allowed(val >> 2)) {
+ retv = -EACCES;
+ break;
+ }
np->tclass = val;
retv = 0;
break;
--
2.8.0.rc3.226.g39d4020
--
To unsubscribe from this list: send the line "unsubscribe linux-doc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
