kernel-ark/include/net/scm.h
Eric Dumazet 16e5726269 af_unix: dont send SCM_CREDENTIALS by default
Since commit 7361c36c52 (af_unix: Allow credentials to work across
user and pid namespaces) af_unix performance dropped a lot.

This is because we now take a reference on pid and cred in each write(),
and release them in read(), usually done from another process,
eventually from another cpu. This triggers false sharing.

# Events: 154K cycles
#
# Overhead  Command       Shared Object        Symbol
# ........  .......  ..................  .........................
#
    10.40%  hackbench  [kernel.kallsyms]   [k] put_pid
     8.60%  hackbench  [kernel.kallsyms]   [k] unix_stream_recvmsg
     7.87%  hackbench  [kernel.kallsyms]   [k] unix_stream_sendmsg
     6.11%  hackbench  [kernel.kallsyms]   [k] do_raw_spin_lock
     4.95%  hackbench  [kernel.kallsyms]   [k] unix_scm_to_skb
     4.87%  hackbench  [kernel.kallsyms]   [k] pid_nr_ns
     4.34%  hackbench  [kernel.kallsyms]   [k] cred_to_ucred
     2.39%  hackbench  [kernel.kallsyms]   [k] unix_destruct_scm
     2.24%  hackbench  [kernel.kallsyms]   [k] sub_preempt_count
     1.75%  hackbench  [kernel.kallsyms]   [k] fget_light
     1.51%  hackbench  [kernel.kallsyms]   [k]
__mutex_lock_interruptible_slowpath
     1.42%  hackbench  [kernel.kallsyms]   [k] sock_alloc_send_pskb

This patch includes SCM_CREDENTIALS information in a af_unix message/skb
only if requested by the sender, [man 7 unix for details how to include
ancillary data using sendmsg() system call]

Note: This might break buggy applications that expected SCM_CREDENTIAL
from an unaware write() system call, and receiver not using SO_PASSCRED
socket option.

If SOCK_PASSCRED is set on source or destination socket, we still
include credentials for mere write() syscalls.

Performance boost in hackbench : more than 50% gain on a 16 thread
machine (2 quad-core cpus, 2 threads per core)

hackbench 20 thread 2000

4.228 sec instead of 9.102 sec

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-09-28 13:29:50 -04:00

130 lines
3.1 KiB
C

#ifndef __LINUX_NET_SCM_H
#define __LINUX_NET_SCM_H
#include <linux/limits.h>
#include <linux/net.h>
#include <linux/security.h>
#include <linux/pid.h>
#include <linux/nsproxy.h>
/* Well, we should have at least one descriptor open
* to accept passed FDs 8)
*/
#define SCM_MAX_FD 253
struct scm_fp_list {
struct list_head list;
short count;
short max;
struct file *fp[SCM_MAX_FD];
};
struct scm_cookie {
struct pid *pid; /* Skb credentials */
const struct cred *cred;
struct scm_fp_list *fp; /* Passed files */
struct ucred creds; /* Skb credentials */
#ifdef CONFIG_SECURITY_NETWORK
u32 secid; /* Passed security ID */
#endif
};
extern void scm_detach_fds(struct msghdr *msg, struct scm_cookie *scm);
extern void scm_detach_fds_compat(struct msghdr *msg, struct scm_cookie *scm);
extern int __scm_send(struct socket *sock, struct msghdr *msg, struct scm_cookie *scm);
extern void __scm_destroy(struct scm_cookie *scm);
extern struct scm_fp_list * scm_fp_dup(struct scm_fp_list *fpl);
#ifdef CONFIG_SECURITY_NETWORK
static __inline__ void unix_get_peersec_dgram(struct socket *sock, struct scm_cookie *scm)
{
security_socket_getpeersec_dgram(sock, NULL, &scm->secid);
}
#else
static __inline__ void unix_get_peersec_dgram(struct socket *sock, struct scm_cookie *scm)
{ }
#endif /* CONFIG_SECURITY_NETWORK */
static __inline__ void scm_set_cred(struct scm_cookie *scm,
struct pid *pid, const struct cred *cred)
{
scm->pid = get_pid(pid);
scm->cred = cred ? get_cred(cred) : NULL;
cred_to_ucred(pid, cred, &scm->creds);
}
static __inline__ void scm_destroy_cred(struct scm_cookie *scm)
{
put_pid(scm->pid);
scm->pid = NULL;
if (scm->cred)
put_cred(scm->cred);
scm->cred = NULL;
}
static __inline__ void scm_destroy(struct scm_cookie *scm)
{
scm_destroy_cred(scm);
if (scm && scm->fp)
__scm_destroy(scm);
}
static __inline__ int scm_send(struct socket *sock, struct msghdr *msg,
struct scm_cookie *scm)
{
memset(scm, 0, sizeof(*scm));
unix_get_peersec_dgram(sock, scm);
if (msg->msg_controllen <= 0)
return 0;
return __scm_send(sock, msg, scm);
}
#ifdef CONFIG_SECURITY_NETWORK
static inline void scm_passec(struct socket *sock, struct msghdr *msg, struct scm_cookie *scm)
{
char *secdata;
u32 seclen;
int err;
if (test_bit(SOCK_PASSSEC, &sock->flags)) {
err = security_secid_to_secctx(scm->secid, &secdata, &seclen);
if (!err) {
put_cmsg(msg, SOL_SOCKET, SCM_SECURITY, seclen, secdata);
security_release_secctx(secdata, seclen);
}
}
}
#else
static inline void scm_passec(struct socket *sock, struct msghdr *msg, struct scm_cookie *scm)
{ }
#endif /* CONFIG_SECURITY_NETWORK */
static __inline__ void scm_recv(struct socket *sock, struct msghdr *msg,
struct scm_cookie *scm, int flags)
{
if (!msg->msg_control) {
if (test_bit(SOCK_PASSCRED, &sock->flags) || scm->fp)
msg->msg_flags |= MSG_CTRUNC;
scm_destroy(scm);
return;
}
if (test_bit(SOCK_PASSCRED, &sock->flags))
put_cmsg(msg, SOL_SOCKET, SCM_CREDENTIALS, sizeof(scm->creds), &scm->creds);
scm_destroy_cred(scm);
scm_passec(sock, msg, scm);
if (!scm->fp)
return;
scm_detach_fds(msg, scm);
}
#endif /* __LINUX_NET_SCM_H */