linux/net/unix/af_unix.c

2458 lines
57 KiB
C
Raw Normal View History

/*
* NET4: Implementation of BSD Unix domain sockets.
*
* Authors: Alan Cox, <alan@lxorguk.ukuu.org.uk>
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*
* Fixes:
* Linus Torvalds : Assorted bug cures.
* Niibe Yutaka : async I/O support.
* Carsten Paeth : PF_UNIX check, address fixes.
* Alan Cox : Limit size of allocated blocks.
* Alan Cox : Fixed the stupid socketpair bug.
* Alan Cox : BSD compatibility fine tuning.
* Alan Cox : Fixed a bug in connect when interrupted.
* Alan Cox : Sorted out a proper draft version of
* file descriptor passing hacked up from
* Mike Shaver's work.
* Marty Leisner : Fixes to fd passing
* Nick Nevin : recvmsg bugfix.
* Alan Cox : Started proper garbage collector
* Heiko EiBfeldt : Missing verify_area check
* Alan Cox : Started POSIXisms
* Andreas Schwab : Replace inode by dentry for proper
* reference counting
* Kirk Petersen : Made this a module
* Christoph Rohland : Elegant non-blocking accept/connect algorithm.
* Lots of bug fixes.
* Alexey Kuznetosv : Repaired (I hope) bugs introduces
* by above two patches.
* Andrea Arcangeli : If possible we block in connect(2)
* if the max backlog of the listen socket
* is been reached. This won't break
* old apps and it will avoid huge amount
* of socks hashed (this for unix_gc()
* performances reasons).
* Security fix that limits the max
* number of socks to 2*max_files and
* the number of skb queueable in the
* dgram receiver.
* Artur Skawina : Hash function optimizations
* Alexey Kuznetsov : Full scale SMP. Lot of bugs are introduced 8)
* Malcolm Beattie : Set peercred for socketpair
* Michal Ostrowski : Module initialization cleanup.
* Arnaldo C. Melo : Remove MOD_{INC,DEC}_USE_COUNT,
* the core infrastructure is doing that
* for all net proto families now (2.5.69+)
*
*
* Known differences from reference BSD that was tested:
*
* [TO FIX]
* ECONNREFUSED is not returned from one end of a connected() socket to the
* other the moment one end closes.
* fstat() doesn't return st_dev=0, and give the blksize as high water mark
* and a fake inode identifier (nor the BSD first socket fstat twice bug).
* [NOT TO FIX]
* accept() returns a path name even if the connecting socket has closed
* in the meantime (BSD loses the path and gives up).
* accept() returns 0 length path for an unbound connector. BSD returns 16
* and a null first byte in the path (but not for gethost/peername - BSD bug ??)
* socketpair(...SOCK_RAW..) doesn't panic the kernel.
* BSD af_unix apparently has connect forgetting to block properly.
* (need to check this with the POSIX spec in detail)
*
* Differences from 2.0.0-11-... (ANK)
* Bug fixes and improvements.
* - client shutdown killed server socket.
* - removed all useless cli/sti pairs.
*
* Semantic changes/extensions.
* - generic control message passing.
* - SCM_CREDENTIALS control message.
* - "Abstract" (not FS based) socket bindings.
* Abstract names are sequences of bytes (not zero terminated)
* started by 0, so that this name space does not intersect
* with BSD names.
*/
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/signal.h>
#include <linux/sched.h>
#include <linux/errno.h>
#include <linux/string.h>
#include <linux/stat.h>
#include <linux/dcache.h>
#include <linux/namei.h>
#include <linux/socket.h>
#include <linux/un.h>
#include <linux/fcntl.h>
#include <linux/termios.h>
#include <linux/sockios.h>
#include <linux/net.h>
#include <linux/in.h>
#include <linux/fs.h>
#include <linux/slab.h>
#include <asm/uaccess.h>
#include <linux/skbuff.h>
#include <linux/netdevice.h>
#include <net/net_namespace.h>
#include <net/sock.h>
#include <net/tcp_states.h>
#include <net/af_unix.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <net/scm.h>
#include <linux/init.h>
#include <linux/poll.h>
#include <linux/rtnetlink.h>
#include <linux/mount.h>
#include <net/checksum.h>
#include <linux/security.h>
#include <linux/freezer.h>
struct hlist_head unix_socket_table[2 * UNIX_HASH_SIZE];
EXPORT_SYMBOL_GPL(unix_socket_table);
DEFINE_SPINLOCK(unix_table_lock);
EXPORT_SYMBOL_GPL(unix_table_lock);
fs: allow for more than 2^31 files Robin Holt tried to boot a 16TB system and found af_unix was overflowing a 32bit value : <quote> We were seeing a failure which prevented boot. The kernel was incapable of creating either a named pipe or unix domain socket. This comes down to a common kernel function called unix_create1() which does: atomic_inc(&unix_nr_socks); if (atomic_read(&unix_nr_socks) > 2 * get_max_files()) goto out; The function get_max_files() is a simple return of files_stat.max_files. files_stat.max_files is a signed integer and is computed in fs/file_table.c's files_init(). n = (mempages * (PAGE_SIZE / 1024)) / 10; files_stat.max_files = n; In our case, mempages (total_ram_pages) is approx 3,758,096,384 (0xe0000000). That leaves max_files at approximately 1,503,238,553. This causes 2 * get_max_files() to integer overflow. </quote> Fix is to let /proc/sys/fs/file-nr & /proc/sys/fs/file-max use long integers, and change af_unix to use an atomic_long_t instead of atomic_t. get_max_files() is changed to return an unsigned long. get_nr_files() is changed to return a long. unix_nr_socks is changed from atomic_t to atomic_long_t, while not strictly needed to address Robin problem. Before patch (on a 64bit kernel) : # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max -18446744071562067968 After patch: # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max 2147483648 # cat /proc/sys/fs/file-nr 704 0 2147483648 Reported-by: Robin Holt <holt@sgi.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: David Miller <davem@davemloft.net> Reviewed-by: Robin Holt <holt@sgi.com> Tested-by: Robin Holt <holt@sgi.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-10-27 05:22:44 +08:00
static atomic_long_t unix_nr_socks;
static struct hlist_head *unix_sockets_unbound(void *addr)
{
unsigned long hash = (unsigned long)addr;
hash ^= hash >> 16;
hash ^= hash >> 8;
hash %= UNIX_HASH_SIZE;
return &unix_socket_table[UNIX_HASH_SIZE + hash];
}
#define UNIX_ABSTRACT(sk) (unix_sk(sk)->addr->hash < UNIX_HASH_SIZE)
[AF_UNIX]: Datagram getpeersec This patch implements an API whereby an application can determine the label of its peer's Unix datagram sockets via the auxiliary data mechanism of recvmsg. Patch purpose: This patch enables a security-aware application to retrieve the security context of the peer of a Unix datagram socket. The application can then use this security context to determine the security context for processing on behalf of the peer who sent the packet. Patch design and implementation: The design and implementation is very similar to the UDP case for INET sockets. Basically we build upon the existing Unix domain socket API for retrieving user credentials. Linux offers the API for obtaining user credentials via ancillary messages (i.e., out of band/control messages that are bundled together with a normal message). To retrieve the security context, the application first indicates to the kernel such desire by setting the SO_PASSSEC option via getsockopt. Then the application retrieves the security context using the auxiliary data mechanism. An example server application for Unix datagram socket should look like this: toggle = 1; toggle_len = sizeof(toggle); setsockopt(sockfd, SOL_SOCKET, SO_PASSSEC, &toggle, &toggle_len); recvmsg(sockfd, &msg_hdr, 0); if (msg_hdr.msg_controllen > sizeof(struct cmsghdr)) { cmsg_hdr = CMSG_FIRSTHDR(&msg_hdr); if (cmsg_hdr->cmsg_len <= CMSG_LEN(sizeof(scontext)) && cmsg_hdr->cmsg_level == SOL_SOCKET && cmsg_hdr->cmsg_type == SCM_SECURITY) { memcpy(&scontext, CMSG_DATA(cmsg_hdr), sizeof(scontext)); } } sock_setsockopt is enhanced with a new socket option SOCK_PASSSEC to allow a server socket to receive security context of the peer. Testing: We have tested the patch by setting up Unix datagram client and server applications. We verified that the server can retrieve the security context using the auxiliary data mechanism of recvmsg. Signed-off-by: Catherine Zhang <cxzhang@watson.ibm.com> Acked-by: Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-06-30 03:27:47 +08:00
#ifdef CONFIG_SECURITY_NETWORK
static void unix_get_secdata(struct scm_cookie *scm, struct sk_buff *skb)
[AF_UNIX]: Datagram getpeersec This patch implements an API whereby an application can determine the label of its peer's Unix datagram sockets via the auxiliary data mechanism of recvmsg. Patch purpose: This patch enables a security-aware application to retrieve the security context of the peer of a Unix datagram socket. The application can then use this security context to determine the security context for processing on behalf of the peer who sent the packet. Patch design and implementation: The design and implementation is very similar to the UDP case for INET sockets. Basically we build upon the existing Unix domain socket API for retrieving user credentials. Linux offers the API for obtaining user credentials via ancillary messages (i.e., out of band/control messages that are bundled together with a normal message). To retrieve the security context, the application first indicates to the kernel such desire by setting the SO_PASSSEC option via getsockopt. Then the application retrieves the security context using the auxiliary data mechanism. An example server application for Unix datagram socket should look like this: toggle = 1; toggle_len = sizeof(toggle); setsockopt(sockfd, SOL_SOCKET, SO_PASSSEC, &toggle, &toggle_len); recvmsg(sockfd, &msg_hdr, 0); if (msg_hdr.msg_controllen > sizeof(struct cmsghdr)) { cmsg_hdr = CMSG_FIRSTHDR(&msg_hdr); if (cmsg_hdr->cmsg_len <= CMSG_LEN(sizeof(scontext)) && cmsg_hdr->cmsg_level == SOL_SOCKET && cmsg_hdr->cmsg_type == SCM_SECURITY) { memcpy(&scontext, CMSG_DATA(cmsg_hdr), sizeof(scontext)); } } sock_setsockopt is enhanced with a new socket option SOCK_PASSSEC to allow a server socket to receive security context of the peer. Testing: We have tested the patch by setting up Unix datagram client and server applications. We verified that the server can retrieve the security context using the auxiliary data mechanism of recvmsg. Signed-off-by: Catherine Zhang <cxzhang@watson.ibm.com> Acked-by: Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-06-30 03:27:47 +08:00
{
memcpy(UNIXSID(skb), &scm->secid, sizeof(u32));
[AF_UNIX]: Datagram getpeersec This patch implements an API whereby an application can determine the label of its peer's Unix datagram sockets via the auxiliary data mechanism of recvmsg. Patch purpose: This patch enables a security-aware application to retrieve the security context of the peer of a Unix datagram socket. The application can then use this security context to determine the security context for processing on behalf of the peer who sent the packet. Patch design and implementation: The design and implementation is very similar to the UDP case for INET sockets. Basically we build upon the existing Unix domain socket API for retrieving user credentials. Linux offers the API for obtaining user credentials via ancillary messages (i.e., out of band/control messages that are bundled together with a normal message). To retrieve the security context, the application first indicates to the kernel such desire by setting the SO_PASSSEC option via getsockopt. Then the application retrieves the security context using the auxiliary data mechanism. An example server application for Unix datagram socket should look like this: toggle = 1; toggle_len = sizeof(toggle); setsockopt(sockfd, SOL_SOCKET, SO_PASSSEC, &toggle, &toggle_len); recvmsg(sockfd, &msg_hdr, 0); if (msg_hdr.msg_controllen > sizeof(struct cmsghdr)) { cmsg_hdr = CMSG_FIRSTHDR(&msg_hdr); if (cmsg_hdr->cmsg_len <= CMSG_LEN(sizeof(scontext)) && cmsg_hdr->cmsg_level == SOL_SOCKET && cmsg_hdr->cmsg_type == SCM_SECURITY) { memcpy(&scontext, CMSG_DATA(cmsg_hdr), sizeof(scontext)); } } sock_setsockopt is enhanced with a new socket option SOCK_PASSSEC to allow a server socket to receive security context of the peer. Testing: We have tested the patch by setting up Unix datagram client and server applications. We verified that the server can retrieve the security context using the auxiliary data mechanism of recvmsg. Signed-off-by: Catherine Zhang <cxzhang@watson.ibm.com> Acked-by: Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-06-30 03:27:47 +08:00
}
static inline void unix_set_secdata(struct scm_cookie *scm, struct sk_buff *skb)
{
scm->secid = *UNIXSID(skb);
[AF_UNIX]: Datagram getpeersec This patch implements an API whereby an application can determine the label of its peer's Unix datagram sockets via the auxiliary data mechanism of recvmsg. Patch purpose: This patch enables a security-aware application to retrieve the security context of the peer of a Unix datagram socket. The application can then use this security context to determine the security context for processing on behalf of the peer who sent the packet. Patch design and implementation: The design and implementation is very similar to the UDP case for INET sockets. Basically we build upon the existing Unix domain socket API for retrieving user credentials. Linux offers the API for obtaining user credentials via ancillary messages (i.e., out of band/control messages that are bundled together with a normal message). To retrieve the security context, the application first indicates to the kernel such desire by setting the SO_PASSSEC option via getsockopt. Then the application retrieves the security context using the auxiliary data mechanism. An example server application for Unix datagram socket should look like this: toggle = 1; toggle_len = sizeof(toggle); setsockopt(sockfd, SOL_SOCKET, SO_PASSSEC, &toggle, &toggle_len); recvmsg(sockfd, &msg_hdr, 0); if (msg_hdr.msg_controllen > sizeof(struct cmsghdr)) { cmsg_hdr = CMSG_FIRSTHDR(&msg_hdr); if (cmsg_hdr->cmsg_len <= CMSG_LEN(sizeof(scontext)) && cmsg_hdr->cmsg_level == SOL_SOCKET && cmsg_hdr->cmsg_type == SCM_SECURITY) { memcpy(&scontext, CMSG_DATA(cmsg_hdr), sizeof(scontext)); } } sock_setsockopt is enhanced with a new socket option SOCK_PASSSEC to allow a server socket to receive security context of the peer. Testing: We have tested the patch by setting up Unix datagram client and server applications. We verified that the server can retrieve the security context using the auxiliary data mechanism of recvmsg. Signed-off-by: Catherine Zhang <cxzhang@watson.ibm.com> Acked-by: Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-06-30 03:27:47 +08:00
}
#else
static inline void unix_get_secdata(struct scm_cookie *scm, struct sk_buff *skb)
[AF_UNIX]: Datagram getpeersec This patch implements an API whereby an application can determine the label of its peer's Unix datagram sockets via the auxiliary data mechanism of recvmsg. Patch purpose: This patch enables a security-aware application to retrieve the security context of the peer of a Unix datagram socket. The application can then use this security context to determine the security context for processing on behalf of the peer who sent the packet. Patch design and implementation: The design and implementation is very similar to the UDP case for INET sockets. Basically we build upon the existing Unix domain socket API for retrieving user credentials. Linux offers the API for obtaining user credentials via ancillary messages (i.e., out of band/control messages that are bundled together with a normal message). To retrieve the security context, the application first indicates to the kernel such desire by setting the SO_PASSSEC option via getsockopt. Then the application retrieves the security context using the auxiliary data mechanism. An example server application for Unix datagram socket should look like this: toggle = 1; toggle_len = sizeof(toggle); setsockopt(sockfd, SOL_SOCKET, SO_PASSSEC, &toggle, &toggle_len); recvmsg(sockfd, &msg_hdr, 0); if (msg_hdr.msg_controllen > sizeof(struct cmsghdr)) { cmsg_hdr = CMSG_FIRSTHDR(&msg_hdr); if (cmsg_hdr->cmsg_len <= CMSG_LEN(sizeof(scontext)) && cmsg_hdr->cmsg_level == SOL_SOCKET && cmsg_hdr->cmsg_type == SCM_SECURITY) { memcpy(&scontext, CMSG_DATA(cmsg_hdr), sizeof(scontext)); } } sock_setsockopt is enhanced with a new socket option SOCK_PASSSEC to allow a server socket to receive security context of the peer. Testing: We have tested the patch by setting up Unix datagram client and server applications. We verified that the server can retrieve the security context using the auxiliary data mechanism of recvmsg. Signed-off-by: Catherine Zhang <cxzhang@watson.ibm.com> Acked-by: Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-06-30 03:27:47 +08:00
{ }
static inline void unix_set_secdata(struct scm_cookie *scm, struct sk_buff *skb)
{ }
#endif /* CONFIG_SECURITY_NETWORK */
/*
* SMP locking strategy:
* hash table is protected with spinlock unix_table_lock
* each socket state is protected by separate spin lock.
*/
static inline unsigned int unix_hash_fold(__wsum n)
{
unsigned int hash = (__force unsigned int)n;
hash ^= hash>>16;
hash ^= hash>>8;
return hash&(UNIX_HASH_SIZE-1);
}
#define unix_peer(sk) (unix_sk(sk)->peer)
static inline int unix_our_peer(struct sock *sk, struct sock *osk)
{
return unix_peer(osk) == sk;
}
static inline int unix_may_send(struct sock *sk, struct sock *osk)
{
return unix_peer(osk) == NULL || unix_our_peer(sk, osk);
}
af_unix: fix 'poll for write'/ connected DGRAM sockets The unix_dgram_sendmsg routine implements a (somewhat crude) form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for connected PF_UNIX non-stream sockets when the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_sendmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-18 13:28:05 +08:00
static inline int unix_recvq_full(struct sock const *sk)
{
return skb_queue_len(&sk->sk_receive_queue) > sk->sk_max_ack_backlog;
}
struct sock *unix_peer_get(struct sock *s)
{
struct sock *peer;
unix_state_lock(s);
peer = unix_peer(s);
if (peer)
sock_hold(peer);
unix_state_unlock(s);
return peer;
}
EXPORT_SYMBOL_GPL(unix_peer_get);
static inline void unix_release_addr(struct unix_address *addr)
{
if (atomic_dec_and_test(&addr->refcnt))
kfree(addr);
}
/*
* Check unix socket name:
* - should be not zero length.
* - if started by not zero, should be NULL terminated (FS object)
* - if started by zero, it is abstract name.
*/
static int unix_mkname(struct sockaddr_un *sunaddr, int len, unsigned int *hashp)
{
if (len <= sizeof(short) || len > sizeof(*sunaddr))
return -EINVAL;
if (!sunaddr || sunaddr->sun_family != AF_UNIX)
return -EINVAL;
if (sunaddr->sun_path[0]) {
/*
* This may look like an off by one error but it is a bit more
* subtle. 108 is the longest valid AF_UNIX path for a binding.
* sun_path[108] doesn't as such exist. However in kernel space
* we are guaranteed that it is a valid memory location in our
* kernel address buffer.
*/
((char *)sunaddr)[len] = 0;
len = strlen(sunaddr->sun_path)+1+sizeof(short);
return len;
}
*hashp = unix_hash_fold(csum_partial(sunaddr, len, 0));
return len;
}
static void __unix_remove_socket(struct sock *sk)
{
sk_del_node_init(sk);
}
static void __unix_insert_socket(struct hlist_head *list, struct sock *sk)
{
WARN_ON(!sk_unhashed(sk));
sk_add_node(sk, list);
}
static inline void unix_remove_socket(struct sock *sk)
{
spin_lock(&unix_table_lock);
__unix_remove_socket(sk);
spin_unlock(&unix_table_lock);
}
static inline void unix_insert_socket(struct hlist_head *list, struct sock *sk)
{
spin_lock(&unix_table_lock);
__unix_insert_socket(list, sk);
spin_unlock(&unix_table_lock);
}
static struct sock *__unix_find_socket_byname(struct net *net,
struct sockaddr_un *sunname,
int len, int type, unsigned int hash)
{
struct sock *s;
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
sk_for_each(s, &unix_socket_table[hash ^ type]) {
struct unix_sock *u = unix_sk(s);
if (!net_eq(sock_net(s), net))
continue;
if (u->addr->len == len &&
!memcmp(u->addr->name, sunname, len))
goto found;
}
s = NULL;
found:
return s;
}
static inline struct sock *unix_find_socket_byname(struct net *net,
struct sockaddr_un *sunname,
int len, int type,
unsigned int hash)
{
struct sock *s;
spin_lock(&unix_table_lock);
s = __unix_find_socket_byname(net, sunname, len, type, hash);
if (s)
sock_hold(s);
spin_unlock(&unix_table_lock);
return s;
}
af_unix: Allow connecting to sockets in other network namespaces. Remove the restriction that only allows connecting to a unix domain socket identified by unix path that is in the same network namespace. Crossing network namespaces is always tricky and we did not support this at first, because of a strict policy of don't mix the namespaces. Later after Pavel proposed this we did not support this because no one had performed the audit to make certain using unix domain sockets across namespaces is safe. What fundamentally makes connecting to af_unix sockets in other namespaces is safe is that you have to have the proper permissions on the unix domain socket inode that lives in the filesystem. If you want strict isolation you just don't create inodes where unfriendlys can get at them, or with permissions that allow unfriendlys to open them. All nicely handled for us by the mount namespace and other standard file system facilities. I looked through unix domain sockets and they are a very controlled environment so none of the work that goes on in dev_forward_skb to make crossing namespaces safe appears needed, we are not loosing controll of the skb and so do not need to set up the skb to look like it is comming in fresh from the outside world. Further the fields in struct unix_skb_parms should not have any problems crossing network namespaces. Now that we handle SCM_CREDENTIALS in a way that gives useable values across namespaces. There does not appear to be any operational problems with encouraging the use of unix domain sockets across containers either. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Acked-by: Daniel Lezcano <daniel.lezcano@free.fr> Acked-by: Pavel Emelyanov <xemul@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-06-13 11:35:48 +08:00
static struct sock *unix_find_socket_byinode(struct inode *i)
{
struct sock *s;
spin_lock(&unix_table_lock);
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
sk_for_each(s,
&unix_socket_table[i->i_ino & (UNIX_HASH_SIZE - 1)]) {
struct dentry *dentry = unix_sk(s)->path.dentry;
if (dentry && dentry->d_inode == i) {
sock_hold(s);
goto found;
}
}
s = NULL;
found:
spin_unlock(&unix_table_lock);
return s;
}
static inline int unix_writable(struct sock *sk)
{
return (atomic_read(&sk->sk_wmem_alloc) << 2) <= sk->sk_sndbuf;
}
static void unix_write_space(struct sock *sk)
{
net: sock_def_readable() and friends RCU conversion sk_callback_lock rwlock actually protects sk->sk_sleep pointer, so we need two atomic operations (and associated dirtying) per incoming packet. RCU conversion is pretty much needed : 1) Add a new structure, called "struct socket_wq" to hold all fields that will need rcu_read_lock() protection (currently: a wait_queue_head_t and a struct fasync_struct pointer). [Future patch will add a list anchor for wakeup coalescing] 2) Attach one of such structure to each "struct socket" created in sock_alloc_inode(). 3) Respect RCU grace period when freeing a "struct socket_wq" 4) Change sk_sleep pointer in "struct sock" by sk_wq, pointer to "struct socket_wq" 5) Change sk_sleep() function to use new sk->sk_wq instead of sk->sk_sleep 6) Change sk_has_sleeper() to wq_has_sleeper() that must be used inside a rcu_read_lock() section. 7) Change all sk_has_sleeper() callers to : - Use rcu_read_lock() instead of read_lock(&sk->sk_callback_lock) - Use wq_has_sleeper() to eventually wakeup tasks. - Use rcu_read_unlock() instead of read_unlock(&sk->sk_callback_lock) 8) sock_wake_async() is modified to use rcu protection as well. 9) Exceptions : macvtap, drivers/net/tun.c, af_unix use integrated "struct socket_wq" instead of dynamically allocated ones. They dont need rcu freeing. Some cleanups or followups are probably needed, (possible sk_callback_lock conversion to a spinlock for example...). Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-29 19:01:49 +08:00
struct socket_wq *wq;
rcu_read_lock();
if (unix_writable(sk)) {
net: sock_def_readable() and friends RCU conversion sk_callback_lock rwlock actually protects sk->sk_sleep pointer, so we need two atomic operations (and associated dirtying) per incoming packet. RCU conversion is pretty much needed : 1) Add a new structure, called "struct socket_wq" to hold all fields that will need rcu_read_lock() protection (currently: a wait_queue_head_t and a struct fasync_struct pointer). [Future patch will add a list anchor for wakeup coalescing] 2) Attach one of such structure to each "struct socket" created in sock_alloc_inode(). 3) Respect RCU grace period when freeing a "struct socket_wq" 4) Change sk_sleep pointer in "struct sock" by sk_wq, pointer to "struct socket_wq" 5) Change sk_sleep() function to use new sk->sk_wq instead of sk->sk_sleep 6) Change sk_has_sleeper() to wq_has_sleeper() that must be used inside a rcu_read_lock() section. 7) Change all sk_has_sleeper() callers to : - Use rcu_read_lock() instead of read_lock(&sk->sk_callback_lock) - Use wq_has_sleeper() to eventually wakeup tasks. - Use rcu_read_unlock() instead of read_unlock(&sk->sk_callback_lock) 8) sock_wake_async() is modified to use rcu protection as well. 9) Exceptions : macvtap, drivers/net/tun.c, af_unix use integrated "struct socket_wq" instead of dynamically allocated ones. They dont need rcu freeing. Some cleanups or followups are probably needed, (possible sk_callback_lock conversion to a spinlock for example...). Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-29 19:01:49 +08:00
wq = rcu_dereference(sk->sk_wq);
if (wq_has_sleeper(wq))
wake_up_interruptible_sync_poll(&wq->wait,
POLLOUT | POLLWRNORM | POLLWRBAND);
sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
}
net: sock_def_readable() and friends RCU conversion sk_callback_lock rwlock actually protects sk->sk_sleep pointer, so we need two atomic operations (and associated dirtying) per incoming packet. RCU conversion is pretty much needed : 1) Add a new structure, called "struct socket_wq" to hold all fields that will need rcu_read_lock() protection (currently: a wait_queue_head_t and a struct fasync_struct pointer). [Future patch will add a list anchor for wakeup coalescing] 2) Attach one of such structure to each "struct socket" created in sock_alloc_inode(). 3) Respect RCU grace period when freeing a "struct socket_wq" 4) Change sk_sleep pointer in "struct sock" by sk_wq, pointer to "struct socket_wq" 5) Change sk_sleep() function to use new sk->sk_wq instead of sk->sk_sleep 6) Change sk_has_sleeper() to wq_has_sleeper() that must be used inside a rcu_read_lock() section. 7) Change all sk_has_sleeper() callers to : - Use rcu_read_lock() instead of read_lock(&sk->sk_callback_lock) - Use wq_has_sleeper() to eventually wakeup tasks. - Use rcu_read_unlock() instead of read_unlock(&sk->sk_callback_lock) 8) sock_wake_async() is modified to use rcu protection as well. 9) Exceptions : macvtap, drivers/net/tun.c, af_unix use integrated "struct socket_wq" instead of dynamically allocated ones. They dont need rcu freeing. Some cleanups or followups are probably needed, (possible sk_callback_lock conversion to a spinlock for example...). Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-29 19:01:49 +08:00
rcu_read_unlock();
}
/* When dgram socket disconnects (or changes its peer), we clear its receive
* queue of packets arrived from previous peer. First, it allows to do
* flow control based only on wmem_alloc; second, sk connected to peer
* may receive messages only from that peer. */
static void unix_dgram_disconnected(struct sock *sk, struct sock *other)
{
if (!skb_queue_empty(&sk->sk_receive_queue)) {
skb_queue_purge(&sk->sk_receive_queue);
wake_up_interruptible_all(&unix_sk(sk)->peer_wait);
/* If one link of bidirectional dgram pipe is disconnected,
* we signal error. Messages are lost. Do not make this,
* when peer was not connected to us.
*/
if (!sock_flag(other, SOCK_DEAD) && unix_peer(other) == sk) {
other->sk_err = ECONNRESET;
other->sk_error_report(other);
}
}
}
static void unix_sock_destructor(struct sock *sk)
{
struct unix_sock *u = unix_sk(sk);
skb_queue_purge(&sk->sk_receive_queue);
WARN_ON(atomic_read(&sk->sk_wmem_alloc));
WARN_ON(!sk_unhashed(sk));
WARN_ON(sk->sk_socket);
if (!sock_flag(sk, SOCK_DEAD)) {
printk(KERN_INFO "Attempt to release alive unix socket: %p\n", sk);
return;
}
if (u->addr)
unix_release_addr(u->addr);
fs: allow for more than 2^31 files Robin Holt tried to boot a 16TB system and found af_unix was overflowing a 32bit value : <quote> We were seeing a failure which prevented boot. The kernel was incapable of creating either a named pipe or unix domain socket. This comes down to a common kernel function called unix_create1() which does: atomic_inc(&unix_nr_socks); if (atomic_read(&unix_nr_socks) > 2 * get_max_files()) goto out; The function get_max_files() is a simple return of files_stat.max_files. files_stat.max_files is a signed integer and is computed in fs/file_table.c's files_init(). n = (mempages * (PAGE_SIZE / 1024)) / 10; files_stat.max_files = n; In our case, mempages (total_ram_pages) is approx 3,758,096,384 (0xe0000000). That leaves max_files at approximately 1,503,238,553. This causes 2 * get_max_files() to integer overflow. </quote> Fix is to let /proc/sys/fs/file-nr & /proc/sys/fs/file-max use long integers, and change af_unix to use an atomic_long_t instead of atomic_t. get_max_files() is changed to return an unsigned long. get_nr_files() is changed to return a long. unix_nr_socks is changed from atomic_t to atomic_long_t, while not strictly needed to address Robin problem. Before patch (on a 64bit kernel) : # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max -18446744071562067968 After patch: # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max 2147483648 # cat /proc/sys/fs/file-nr 704 0 2147483648 Reported-by: Robin Holt <holt@sgi.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: David Miller <davem@davemloft.net> Reviewed-by: Robin Holt <holt@sgi.com> Tested-by: Robin Holt <holt@sgi.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-10-27 05:22:44 +08:00
atomic_long_dec(&unix_nr_socks);
local_bh_disable();
sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
local_bh_enable();
#ifdef UNIX_REFCNT_DEBUG
fs: allow for more than 2^31 files Robin Holt tried to boot a 16TB system and found af_unix was overflowing a 32bit value : <quote> We were seeing a failure which prevented boot. The kernel was incapable of creating either a named pipe or unix domain socket. This comes down to a common kernel function called unix_create1() which does: atomic_inc(&unix_nr_socks); if (atomic_read(&unix_nr_socks) > 2 * get_max_files()) goto out; The function get_max_files() is a simple return of files_stat.max_files. files_stat.max_files is a signed integer and is computed in fs/file_table.c's files_init(). n = (mempages * (PAGE_SIZE / 1024)) / 10; files_stat.max_files = n; In our case, mempages (total_ram_pages) is approx 3,758,096,384 (0xe0000000). That leaves max_files at approximately 1,503,238,553. This causes 2 * get_max_files() to integer overflow. </quote> Fix is to let /proc/sys/fs/file-nr & /proc/sys/fs/file-max use long integers, and change af_unix to use an atomic_long_t instead of atomic_t. get_max_files() is changed to return an unsigned long. get_nr_files() is changed to return a long. unix_nr_socks is changed from atomic_t to atomic_long_t, while not strictly needed to address Robin problem. Before patch (on a 64bit kernel) : # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max -18446744071562067968 After patch: # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max 2147483648 # cat /proc/sys/fs/file-nr 704 0 2147483648 Reported-by: Robin Holt <holt@sgi.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: David Miller <davem@davemloft.net> Reviewed-by: Robin Holt <holt@sgi.com> Tested-by: Robin Holt <holt@sgi.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-10-27 05:22:44 +08:00
printk(KERN_DEBUG "UNIX %p is destroyed, %ld are still alive.\n", sk,
atomic_long_read(&unix_nr_socks));
#endif
}
static void unix_release_sock(struct sock *sk, int embrion)
{
struct unix_sock *u = unix_sk(sk);
struct path path;
struct sock *skpair;
struct sk_buff *skb;
int state;
unix_remove_socket(sk);
/* Clear state */
unix_state_lock(sk);
sock_orphan(sk);
sk->sk_shutdown = SHUTDOWN_MASK;
path = u->path;
u->path.dentry = NULL;
u->path.mnt = NULL;
state = sk->sk_state;
sk->sk_state = TCP_CLOSE;
unix_state_unlock(sk);
wake_up_interruptible_all(&u->peer_wait);
skpair = unix_peer(sk);
if (skpair != NULL) {
if (sk->sk_type == SOCK_STREAM || sk->sk_type == SOCK_SEQPACKET) {
unix_state_lock(skpair);
/* No more writes */
skpair->sk_shutdown = SHUTDOWN_MASK;
if (!skb_queue_empty(&sk->sk_receive_queue) || embrion)
skpair->sk_err = ECONNRESET;
unix_state_unlock(skpair);
skpair->sk_state_change(skpair);
sk_wake_async(skpair, SOCK_WAKE_WAITD, POLL_HUP);
}
sock_put(skpair); /* It may now die */
unix_peer(sk) = NULL;
}
/* Try to flush out this socket. Throw out buffers at least */
while ((skb = skb_dequeue(&sk->sk_receive_queue)) != NULL) {
if (state == TCP_LISTEN)
unix_release_sock(skb->sk, 1);
/* passed fds are erased in the kfree_skb hook */
kfree_skb(skb);
}
if (path.dentry)
path_put(&path);
sock_put(sk);
/* ---- Socket is dead now and most probably destroyed ---- */
/*
* Fixme: BSD difference: In BSD all sockets connected to us get
* ECONNRESET and we die on the spot. In Linux we behave
* like files and pipes do and wait for the last
* dereference.
*
* Can't we simply set sock->err?
*
* What the above comment does talk about? --ANK(980817)
*/
if (unix_tot_inflight)
unix_gc(); /* Garbage collect fds */
}
static void init_peercred(struct sock *sk)
{
put_pid(sk->sk_peer_pid);
if (sk->sk_peer_cred)
put_cred(sk->sk_peer_cred);
sk->sk_peer_pid = get_pid(task_tgid(current));
sk->sk_peer_cred = get_current_cred();
}
static void copy_peercred(struct sock *sk, struct sock *peersk)
{
put_pid(sk->sk_peer_pid);
if (sk->sk_peer_cred)
put_cred(sk->sk_peer_cred);
sk->sk_peer_pid = get_pid(peersk->sk_peer_pid);
sk->sk_peer_cred = get_cred(peersk->sk_peer_cred);
}
static int unix_listen(struct socket *sock, int backlog)
{
int err;
struct sock *sk = sock->sk;
struct unix_sock *u = unix_sk(sk);
struct pid *old_pid = NULL;
err = -EOPNOTSUPP;
if (sock->type != SOCK_STREAM && sock->type != SOCK_SEQPACKET)
goto out; /* Only stream/seqpacket sockets accept */
err = -EINVAL;
if (!u->addr)
goto out; /* No listens on an unbound socket */
unix_state_lock(sk);
if (sk->sk_state != TCP_CLOSE && sk->sk_state != TCP_LISTEN)
goto out_unlock;
if (backlog > sk->sk_max_ack_backlog)
wake_up_interruptible_all(&u->peer_wait);
sk->sk_max_ack_backlog = backlog;
sk->sk_state = TCP_LISTEN;
/* set credentials so connect can copy them */
init_peercred(sk);
err = 0;
out_unlock:
unix_state_unlock(sk);
put_pid(old_pid);
out:
return err;
}
static int unix_release(struct socket *);
static int unix_bind(struct socket *, struct sockaddr *, int);
static int unix_stream_connect(struct socket *, struct sockaddr *,
int addr_len, int flags);
static int unix_socketpair(struct socket *, struct socket *);
static int unix_accept(struct socket *, struct socket *, int);
static int unix_getname(struct socket *, struct sockaddr *, int *, int);
static unsigned int unix_poll(struct file *, struct socket *, poll_table *);
af_unix: fix 'poll for write'/connected DGRAM sockets For n:1 'datagram connections' (eg /dev/log), the unix_dgram_sendmsg routine implements a form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for PF_UNIX non-stream sockets connected to server sockets dealing with (potentially) multiple clients if the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_recvmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-28 10:34:18 +08:00
static unsigned int unix_dgram_poll(struct file *, struct socket *,
poll_table *);
static int unix_ioctl(struct socket *, unsigned int, unsigned long);
static int unix_shutdown(struct socket *, int);
static int unix_stream_sendmsg(struct kiocb *, struct socket *,
struct msghdr *, size_t);
static int unix_stream_recvmsg(struct kiocb *, struct socket *,
struct msghdr *, size_t, int);
static int unix_dgram_sendmsg(struct kiocb *, struct socket *,
struct msghdr *, size_t);
static int unix_dgram_recvmsg(struct kiocb *, struct socket *,
struct msghdr *, size_t, int);
static int unix_dgram_connect(struct socket *, struct sockaddr *,
int, int);
static int unix_seqpacket_sendmsg(struct kiocb *, struct socket *,
struct msghdr *, size_t);
static int unix_seqpacket_recvmsg(struct kiocb *, struct socket *,
struct msghdr *, size_t, int);
static void unix_set_peek_off(struct sock *sk, int val)
{
struct unix_sock *u = unix_sk(sk);
mutex_lock(&u->readlock);
sk->sk_peek_off = val;
mutex_unlock(&u->readlock);
}
static const struct proto_ops unix_stream_ops = {
.family = PF_UNIX,
.owner = THIS_MODULE,
.release = unix_release,
.bind = unix_bind,
.connect = unix_stream_connect,
.socketpair = unix_socketpair,
.accept = unix_accept,
.getname = unix_getname,
.poll = unix_poll,
.ioctl = unix_ioctl,
.listen = unix_listen,
.shutdown = unix_shutdown,
.setsockopt = sock_no_setsockopt,
.getsockopt = sock_no_getsockopt,
.sendmsg = unix_stream_sendmsg,
.recvmsg = unix_stream_recvmsg,
.mmap = sock_no_mmap,
.sendpage = sock_no_sendpage,
.set_peek_off = unix_set_peek_off,
};
static const struct proto_ops unix_dgram_ops = {
.family = PF_UNIX,
.owner = THIS_MODULE,
.release = unix_release,
.bind = unix_bind,
.connect = unix_dgram_connect,
.socketpair = unix_socketpair,
.accept = sock_no_accept,
.getname = unix_getname,
af_unix: fix 'poll for write'/connected DGRAM sockets For n:1 'datagram connections' (eg /dev/log), the unix_dgram_sendmsg routine implements a form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for PF_UNIX non-stream sockets connected to server sockets dealing with (potentially) multiple clients if the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_recvmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-28 10:34:18 +08:00
.poll = unix_dgram_poll,
.ioctl = unix_ioctl,
.listen = sock_no_listen,
.shutdown = unix_shutdown,
.setsockopt = sock_no_setsockopt,
.getsockopt = sock_no_getsockopt,
.sendmsg = unix_dgram_sendmsg,
.recvmsg = unix_dgram_recvmsg,
.mmap = sock_no_mmap,
.sendpage = sock_no_sendpage,
.set_peek_off = unix_set_peek_off,
};
static const struct proto_ops unix_seqpacket_ops = {
.family = PF_UNIX,
.owner = THIS_MODULE,
.release = unix_release,
.bind = unix_bind,
.connect = unix_stream_connect,
.socketpair = unix_socketpair,
.accept = unix_accept,
.getname = unix_getname,
af_unix: fix 'poll for write'/connected DGRAM sockets For n:1 'datagram connections' (eg /dev/log), the unix_dgram_sendmsg routine implements a form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for PF_UNIX non-stream sockets connected to server sockets dealing with (potentially) multiple clients if the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_recvmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-28 10:34:18 +08:00
.poll = unix_dgram_poll,
.ioctl = unix_ioctl,
.listen = unix_listen,
.shutdown = unix_shutdown,
.setsockopt = sock_no_setsockopt,
.getsockopt = sock_no_getsockopt,
.sendmsg = unix_seqpacket_sendmsg,
.recvmsg = unix_seqpacket_recvmsg,
.mmap = sock_no_mmap,
.sendpage = sock_no_sendpage,
.set_peek_off = unix_set_peek_off,
};
static struct proto unix_proto = {
.name = "UNIX",
.owner = THIS_MODULE,
.obj_size = sizeof(struct unix_sock),
};
/*
* AF_UNIX sockets do not interact with hardware, hence they
* dont trigger interrupts - so it's safe for them to have
* bh-unsafe locking for their sk_receive_queue.lock. Split off
* this special lock-class by reinitializing the spinlock key:
*/
static struct lock_class_key af_unix_sk_receive_queue_lock_key;
static struct sock *unix_create1(struct net *net, struct socket *sock)
{
struct sock *sk = NULL;
struct unix_sock *u;
fs: allow for more than 2^31 files Robin Holt tried to boot a 16TB system and found af_unix was overflowing a 32bit value : <quote> We were seeing a failure which prevented boot. The kernel was incapable of creating either a named pipe or unix domain socket. This comes down to a common kernel function called unix_create1() which does: atomic_inc(&unix_nr_socks); if (atomic_read(&unix_nr_socks) > 2 * get_max_files()) goto out; The function get_max_files() is a simple return of files_stat.max_files. files_stat.max_files is a signed integer and is computed in fs/file_table.c's files_init(). n = (mempages * (PAGE_SIZE / 1024)) / 10; files_stat.max_files = n; In our case, mempages (total_ram_pages) is approx 3,758,096,384 (0xe0000000). That leaves max_files at approximately 1,503,238,553. This causes 2 * get_max_files() to integer overflow. </quote> Fix is to let /proc/sys/fs/file-nr & /proc/sys/fs/file-max use long integers, and change af_unix to use an atomic_long_t instead of atomic_t. get_max_files() is changed to return an unsigned long. get_nr_files() is changed to return a long. unix_nr_socks is changed from atomic_t to atomic_long_t, while not strictly needed to address Robin problem. Before patch (on a 64bit kernel) : # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max -18446744071562067968 After patch: # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max 2147483648 # cat /proc/sys/fs/file-nr 704 0 2147483648 Reported-by: Robin Holt <holt@sgi.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: David Miller <davem@davemloft.net> Reviewed-by: Robin Holt <holt@sgi.com> Tested-by: Robin Holt <holt@sgi.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-10-27 05:22:44 +08:00
atomic_long_inc(&unix_nr_socks);
if (atomic_long_read(&unix_nr_socks) > 2 * get_max_files())
goto out;
sk = sk_alloc(net, PF_UNIX, GFP_KERNEL, &unix_proto);
if (!sk)
goto out;
sock_init_data(sock, sk);
lockdep_set_class(&sk->sk_receive_queue.lock,
&af_unix_sk_receive_queue_lock_key);
sk->sk_write_space = unix_write_space;
sk->sk_max_ack_backlog = net->unx.sysctl_max_dgram_qlen;
sk->sk_destruct = unix_sock_destructor;
u = unix_sk(sk);
u->path.dentry = NULL;
u->path.mnt = NULL;
spin_lock_init(&u->lock);
atomic_long_set(&u->inflight, 0);
INIT_LIST_HEAD(&u->link);
mutex_init(&u->readlock); /* single task reading lock */
init_waitqueue_head(&u->peer_wait);
unix_insert_socket(unix_sockets_unbound(sk), sk);
out:
if (sk == NULL)
fs: allow for more than 2^31 files Robin Holt tried to boot a 16TB system and found af_unix was overflowing a 32bit value : <quote> We were seeing a failure which prevented boot. The kernel was incapable of creating either a named pipe or unix domain socket. This comes down to a common kernel function called unix_create1() which does: atomic_inc(&unix_nr_socks); if (atomic_read(&unix_nr_socks) > 2 * get_max_files()) goto out; The function get_max_files() is a simple return of files_stat.max_files. files_stat.max_files is a signed integer and is computed in fs/file_table.c's files_init(). n = (mempages * (PAGE_SIZE / 1024)) / 10; files_stat.max_files = n; In our case, mempages (total_ram_pages) is approx 3,758,096,384 (0xe0000000). That leaves max_files at approximately 1,503,238,553. This causes 2 * get_max_files() to integer overflow. </quote> Fix is to let /proc/sys/fs/file-nr & /proc/sys/fs/file-max use long integers, and change af_unix to use an atomic_long_t instead of atomic_t. get_max_files() is changed to return an unsigned long. get_nr_files() is changed to return a long. unix_nr_socks is changed from atomic_t to atomic_long_t, while not strictly needed to address Robin problem. Before patch (on a 64bit kernel) : # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max -18446744071562067968 After patch: # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max 2147483648 # cat /proc/sys/fs/file-nr 704 0 2147483648 Reported-by: Robin Holt <holt@sgi.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: David Miller <davem@davemloft.net> Reviewed-by: Robin Holt <holt@sgi.com> Tested-by: Robin Holt <holt@sgi.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-10-27 05:22:44 +08:00
atomic_long_dec(&unix_nr_socks);
else {
local_bh_disable();
sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
local_bh_enable();
}
return sk;
}
static int unix_create(struct net *net, struct socket *sock, int protocol,
int kern)
{
if (protocol && protocol != PF_UNIX)
return -EPROTONOSUPPORT;
sock->state = SS_UNCONNECTED;
switch (sock->type) {
case SOCK_STREAM:
sock->ops = &unix_stream_ops;
break;
/*
* Believe it or not BSD has AF_UNIX, SOCK_RAW though
* nothing uses it.
*/
case SOCK_RAW:
sock->type = SOCK_DGRAM;
case SOCK_DGRAM:
sock->ops = &unix_dgram_ops;
break;
case SOCK_SEQPACKET:
sock->ops = &unix_seqpacket_ops;
break;
default:
return -ESOCKTNOSUPPORT;
}
return unix_create1(net, sock) ? 0 : -ENOMEM;
}
static int unix_release(struct socket *sock)
{
struct sock *sk = sock->sk;
if (!sk)
return 0;
unix_release_sock(sk, 0);
sock->sk = NULL;
return 0;
}
static int unix_autobind(struct socket *sock)
{
struct sock *sk = sock->sk;
struct net *net = sock_net(sk);
struct unix_sock *u = unix_sk(sk);
static u32 ordernum = 1;
struct unix_address *addr;
int err;
unsigned int retries = 0;
mutex_lock(&u->readlock);
err = 0;
if (u->addr)
goto out;
err = -ENOMEM;
addr = kzalloc(sizeof(*addr) + sizeof(short) + 16, GFP_KERNEL);
if (!addr)
goto out;
addr->name->sun_family = AF_UNIX;
atomic_set(&addr->refcnt, 1);
retry:
addr->len = sprintf(addr->name->sun_path+1, "%05x", ordernum) + 1 + sizeof(short);
addr->hash = unix_hash_fold(csum_partial(addr->name, addr->len, 0));
spin_lock(&unix_table_lock);
ordernum = (ordernum+1)&0xFFFFF;
if (__unix_find_socket_byname(net, addr->name, addr->len, sock->type,
addr->hash)) {
spin_unlock(&unix_table_lock);
/*
* __unix_find_socket_byname() may take long time if many names
* are already in use.
*/
cond_resched();
/* Give up if all names seems to be in use. */
if (retries++ == 0xFFFFF) {
err = -ENOSPC;
kfree(addr);
goto out;
}
goto retry;
}
addr->hash ^= sk->sk_type;
__unix_remove_socket(sk);
u->addr = addr;
__unix_insert_socket(&unix_socket_table[addr->hash], sk);
spin_unlock(&unix_table_lock);
err = 0;
out: mutex_unlock(&u->readlock);
return err;
}
static struct sock *unix_find_other(struct net *net,
struct sockaddr_un *sunname, int len,
int type, unsigned int hash, int *error)
{
struct sock *u;
struct path path;
int err = 0;
if (sunname->sun_path[0]) {
struct inode *inode;
err = kern_path(sunname->sun_path, LOOKUP_FOLLOW, &path);
if (err)
goto fail;
inode = path.dentry->d_inode;
err = inode_permission(inode, MAY_WRITE);
if (err)
goto put_fail;
err = -ECONNREFUSED;
if (!S_ISSOCK(inode->i_mode))
goto put_fail;
af_unix: Allow connecting to sockets in other network namespaces. Remove the restriction that only allows connecting to a unix domain socket identified by unix path that is in the same network namespace. Crossing network namespaces is always tricky and we did not support this at first, because of a strict policy of don't mix the namespaces. Later after Pavel proposed this we did not support this because no one had performed the audit to make certain using unix domain sockets across namespaces is safe. What fundamentally makes connecting to af_unix sockets in other namespaces is safe is that you have to have the proper permissions on the unix domain socket inode that lives in the filesystem. If you want strict isolation you just don't create inodes where unfriendlys can get at them, or with permissions that allow unfriendlys to open them. All nicely handled for us by the mount namespace and other standard file system facilities. I looked through unix domain sockets and they are a very controlled environment so none of the work that goes on in dev_forward_skb to make crossing namespaces safe appears needed, we are not loosing controll of the skb and so do not need to set up the skb to look like it is comming in fresh from the outside world. Further the fields in struct unix_skb_parms should not have any problems crossing network namespaces. Now that we handle SCM_CREDENTIALS in a way that gives useable values across namespaces. There does not appear to be any operational problems with encouraging the use of unix domain sockets across containers either. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Acked-by: Daniel Lezcano <daniel.lezcano@free.fr> Acked-by: Pavel Emelyanov <xemul@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-06-13 11:35:48 +08:00
u = unix_find_socket_byinode(inode);
if (!u)
goto put_fail;
if (u->sk_type == type)
touch_atime(&path);
path_put(&path);
err = -EPROTOTYPE;
if (u->sk_type != type) {
sock_put(u);
goto fail;
}
} else {
err = -ECONNREFUSED;
u = unix_find_socket_byname(net, sunname, len, type, hash);
if (u) {
struct dentry *dentry;
dentry = unix_sk(u)->path.dentry;
if (dentry)
touch_atime(&unix_sk(u)->path);
} else
goto fail;
}
return u;
put_fail:
path_put(&path);
fail:
*error = err;
return NULL;
}
static int unix_mknod(const char *sun_path, umode_t mode, struct path *res)
{
struct dentry *dentry;
struct path path;
int err = 0;
/*
* Get the parent directory, calculate the hash for last
* component.
*/
dentry = kern_path_create(AT_FDCWD, sun_path, &path, 0);
err = PTR_ERR(dentry);
if (IS_ERR(dentry))
return err;
/*
* All right, let's create it.
*/
err = security_path_mknod(&path, dentry, mode, 0);
if (!err) {
err = vfs_mknod(path.dentry->d_inode, dentry, mode, 0);
if (!err) {
res->mnt = mntget(path.mnt);
res->dentry = dget(dentry);
}
}
done_path_create(&path, dentry);
return err;
}
static int unix_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
{
struct sock *sk = sock->sk;
struct net *net = sock_net(sk);
struct unix_sock *u = unix_sk(sk);
struct sockaddr_un *sunaddr = (struct sockaddr_un *)uaddr;
char *sun_path = sunaddr->sun_path;
int err;
unsigned int hash;
struct unix_address *addr;
struct hlist_head *list;
err = -EINVAL;
if (sunaddr->sun_family != AF_UNIX)
goto out;
if (addr_len == sizeof(short)) {
err = unix_autobind(sock);
goto out;
}
err = unix_mkname(sunaddr, addr_len, &hash);
if (err < 0)
goto out;
addr_len = err;
mutex_lock(&u->readlock);
err = -EINVAL;
if (u->addr)
goto out_up;
err = -ENOMEM;
addr = kmalloc(sizeof(*addr)+addr_len, GFP_KERNEL);
if (!addr)
goto out_up;
memcpy(addr->name, sunaddr, addr_len);
addr->len = addr_len;
addr->hash = hash ^ sk->sk_type;
atomic_set(&addr->refcnt, 1);
if (sun_path[0]) {
struct path path;
umode_t mode = S_IFSOCK |
(SOCK_INODE(sock)->i_mode & ~current_umask());
err = unix_mknod(sun_path, mode, &path);
if (err) {
if (err == -EEXIST)
err = -EADDRINUSE;
unix_release_addr(addr);
goto out_up;
}
addr->hash = UNIX_HASH_SIZE;
hash = path.dentry->d_inode->i_ino & (UNIX_HASH_SIZE-1);
spin_lock(&unix_table_lock);
u->path = path;
list = &unix_socket_table[hash];
} else {
spin_lock(&unix_table_lock);
err = -EADDRINUSE;
if (__unix_find_socket_byname(net, sunaddr, addr_len,
sk->sk_type, hash)) {
unix_release_addr(addr);
goto out_unlock;
}
list = &unix_socket_table[addr->hash];
}
err = 0;
__unix_remove_socket(sk);
u->addr = addr;
__unix_insert_socket(list, sk);
out_unlock:
spin_unlock(&unix_table_lock);
out_up:
mutex_unlock(&u->readlock);
out:
return err;
}
static void unix_state_double_lock(struct sock *sk1, struct sock *sk2)
{
if (unlikely(sk1 == sk2) || !sk2) {
unix_state_lock(sk1);
return;
}
if (sk1 < sk2) {
unix_state_lock(sk1);
unix_state_lock_nested(sk2);
} else {
unix_state_lock(sk2);
unix_state_lock_nested(sk1);
}
}
static void unix_state_double_unlock(struct sock *sk1, struct sock *sk2)
{
if (unlikely(sk1 == sk2) || !sk2) {
unix_state_unlock(sk1);
return;
}
unix_state_unlock(sk1);
unix_state_unlock(sk2);
}
static int unix_dgram_connect(struct socket *sock, struct sockaddr *addr,
int alen, int flags)
{
struct sock *sk = sock->sk;
struct net *net = sock_net(sk);
struct sockaddr_un *sunaddr = (struct sockaddr_un *)addr;
struct sock *other;
unsigned int hash;
int err;
if (addr->sa_family != AF_UNSPEC) {
err = unix_mkname(sunaddr, alen, &hash);
if (err < 0)
goto out;
alen = err;
if (test_bit(SOCK_PASSCRED, &sock->flags) &&
!unix_sk(sk)->addr && (err = unix_autobind(sock)) != 0)
goto out;
restart:
other = unix_find_other(net, sunaddr, alen, sock->type, hash, &err);
if (!other)
goto out;
unix_state_double_lock(sk, other);
/* Apparently VFS overslept socket death. Retry. */
if (sock_flag(other, SOCK_DEAD)) {
unix_state_double_unlock(sk, other);
sock_put(other);
goto restart;
}
err = -EPERM;
if (!unix_may_send(sk, other))
goto out_unlock;
err = security_unix_may_send(sk->sk_socket, other->sk_socket);
if (err)
goto out_unlock;
} else {
/*
* 1003.1g breaking connected state with AF_UNSPEC
*/
other = NULL;
unix_state_double_lock(sk, other);
}
/*
* If it was connected, reconnect.
*/
if (unix_peer(sk)) {
struct sock *old_peer = unix_peer(sk);
unix_peer(sk) = other;
unix_state_double_unlock(sk, other);
if (other != old_peer)
unix_dgram_disconnected(sk, old_peer);
sock_put(old_peer);
} else {
unix_peer(sk) = other;
unix_state_double_unlock(sk, other);
}
return 0;
out_unlock:
unix_state_double_unlock(sk, other);
sock_put(other);
out:
return err;
}
static long unix_wait_for_peer(struct sock *other, long timeo)
{
struct unix_sock *u = unix_sk(other);
int sched;
DEFINE_WAIT(wait);
prepare_to_wait_exclusive(&u->peer_wait, &wait, TASK_INTERRUPTIBLE);
sched = !sock_flag(other, SOCK_DEAD) &&
!(other->sk_shutdown & RCV_SHUTDOWN) &&
af_unix: fix 'poll for write'/ connected DGRAM sockets The unix_dgram_sendmsg routine implements a (somewhat crude) form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for connected PF_UNIX non-stream sockets when the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_sendmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-18 13:28:05 +08:00
unix_recvq_full(other);
unix_state_unlock(other);
if (sched)
timeo = schedule_timeout(timeo);
finish_wait(&u->peer_wait, &wait);
return timeo;
}
static int unix_stream_connect(struct socket *sock, struct sockaddr *uaddr,
int addr_len, int flags)
{
struct sockaddr_un *sunaddr = (struct sockaddr_un *)uaddr;
struct sock *sk = sock->sk;
struct net *net = sock_net(sk);
struct unix_sock *u = unix_sk(sk), *newu, *otheru;
struct sock *newsk = NULL;
struct sock *other = NULL;
struct sk_buff *skb = NULL;
unsigned int hash;
int st;
int err;
long timeo;
err = unix_mkname(sunaddr, addr_len, &hash);
if (err < 0)
goto out;
addr_len = err;
if (test_bit(SOCK_PASSCRED, &sock->flags) && !u->addr &&
(err = unix_autobind(sock)) != 0)
goto out;
timeo = sock_sndtimeo(sk, flags & O_NONBLOCK);
/* First of all allocate resources.
If we will make it after state is locked,
we will have to recheck all again in any case.
*/
err = -ENOMEM;
/* create new sock for complete connection */
newsk = unix_create1(sock_net(sk), NULL);
if (newsk == NULL)
goto out;
/* Allocate skb for sending to listening sock */
skb = sock_wmalloc(newsk, 1, 0, GFP_KERNEL);
if (skb == NULL)
goto out;
restart:
/* Find listening sock. */
other = unix_find_other(net, sunaddr, addr_len, sk->sk_type, hash, &err);
if (!other)
goto out;
/* Latch state of peer */
unix_state_lock(other);
/* Apparently VFS overslept socket death. Retry. */
if (sock_flag(other, SOCK_DEAD)) {
unix_state_unlock(other);
sock_put(other);
goto restart;
}
err = -ECONNREFUSED;
if (other->sk_state != TCP_LISTEN)
goto out_unlock;
AF_UNIX: Fix deadlock on connecting to shutdown socket I found a deadlock bug in UNIX domain socket, which makes able to DoS attack against the local machine by non-root users. How to reproduce: 1. Make a listening AF_UNIX/SOCK_STREAM socket with an abstruct namespace(*), and shutdown(2) it. 2. Repeat connect(2)ing to the listening socket from the other sockets until the connection backlog is full-filled. 3. connect(2) takes the CPU forever. If every core is taken, the system hangs. PoC code: (Run as many times as cores on SMP machines.) int main(void) { int ret; int csd; int lsd; struct sockaddr_un sun; /* make an abstruct name address (*) */ memset(&sun, 0, sizeof(sun)); sun.sun_family = PF_UNIX; sprintf(&sun.sun_path[1], "%d", getpid()); /* create the listening socket and shutdown */ lsd = socket(AF_UNIX, SOCK_STREAM, 0); bind(lsd, (struct sockaddr *)&sun, sizeof(sun)); listen(lsd, 1); shutdown(lsd, SHUT_RDWR); /* connect loop */ alarm(15); /* forcely exit the loop after 15 sec */ for (;;) { csd = socket(AF_UNIX, SOCK_STREAM, 0); ret = connect(csd, (struct sockaddr *)&sun, sizeof(sun)); if (-1 == ret) { perror("connect()"); break; } puts("Connection OK"); } return 0; } (*) Make sun_path[0] = 0 to use the abstruct namespace. If a file-based socket is used, the system doesn't deadlock because of context switches in the file system layer. Why this happens: Error checks between unix_socket_connect() and unix_wait_for_peer() are inconsistent. The former calls the latter to wait until the backlog is processed. Despite the latter returns without doing anything when the socket is shutdown, the former doesn't check the shutdown state and just retries calling the latter forever. Patch: The patch below adds shutdown check into unix_socket_connect(), so connect(2) to the shutdown socket will return -ECONREFUSED. Signed-off-by: Tomoki Sekiyama <tomoki.sekiyama.qu@hitachi.com> Signed-off-by: Masanori Yoshida <masanori.yoshida.tv@hitachi.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-10-19 14:17:37 +08:00
if (other->sk_shutdown & RCV_SHUTDOWN)
goto out_unlock;
af_unix: fix 'poll for write'/ connected DGRAM sockets The unix_dgram_sendmsg routine implements a (somewhat crude) form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for connected PF_UNIX non-stream sockets when the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_sendmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-18 13:28:05 +08:00
if (unix_recvq_full(other)) {
err = -EAGAIN;
if (!timeo)
goto out_unlock;
timeo = unix_wait_for_peer(other, timeo);
err = sock_intr_errno(timeo);
if (signal_pending(current))
goto out;
sock_put(other);
goto restart;
}
/* Latch our state.
It is tricky place. We need to grab our state lock and cannot
drop lock on peer. It is dangerous because deadlock is
possible. Connect to self case and simultaneous
attempt to connect are eliminated by checking socket
state. other is TCP_LISTEN, if sk is TCP_LISTEN we
check this before attempt to grab lock.
Well, and we have to recheck the state after socket locked.
*/
st = sk->sk_state;
switch (st) {
case TCP_CLOSE:
/* This is ok... continue with connect */
break;
case TCP_ESTABLISHED:
/* Socket is already connected */
err = -EISCONN;
goto out_unlock;
default:
err = -EINVAL;
goto out_unlock;
}
unix_state_lock_nested(sk);
if (sk->sk_state != st) {
unix_state_unlock(sk);
unix_state_unlock(other);
sock_put(other);
goto restart;
}
err = security_unix_stream_connect(sk, other, newsk);
if (err) {
unix_state_unlock(sk);
goto out_unlock;
}
/* The way is open! Fastly set all the necessary fields... */
sock_hold(sk);
unix_peer(newsk) = sk;
newsk->sk_state = TCP_ESTABLISHED;
newsk->sk_type = sk->sk_type;
init_peercred(newsk);
newu = unix_sk(newsk);
RCU_INIT_POINTER(newsk->sk_wq, &newu->peer_wq);
otheru = unix_sk(other);
/* copy address information from listening to new sock*/
if (otheru->addr) {
atomic_inc(&otheru->addr->refcnt);
newu->addr = otheru->addr;
}
if (otheru->path.dentry) {
path_get(&otheru->path);
newu->path = otheru->path;
}
/* Set credentials */
copy_peercred(sk, other);
sock->state = SS_CONNECTED;
sk->sk_state = TCP_ESTABLISHED;
sock_hold(newsk);
smp_mb__after_atomic_inc(); /* sock_hold() does an atomic_inc() */
unix_peer(sk) = newsk;
unix_state_unlock(sk);
/* take ten and and send info to listening sock */
spin_lock(&other->sk_receive_queue.lock);
__skb_queue_tail(&other->sk_receive_queue, skb);
spin_unlock(&other->sk_receive_queue.lock);
unix_state_unlock(other);
other->sk_data_ready(other, 0);
sock_put(other);
return 0;
out_unlock:
if (other)
unix_state_unlock(other);
out:
kfree_skb(skb);
if (newsk)
unix_release_sock(newsk, 0);
if (other)
sock_put(other);
return err;
}
static int unix_socketpair(struct socket *socka, struct socket *sockb)
{
struct sock *ska = socka->sk, *skb = sockb->sk;
/* Join our sockets back to back */
sock_hold(ska);
sock_hold(skb);
unix_peer(ska) = skb;
unix_peer(skb) = ska;
init_peercred(ska);
init_peercred(skb);
if (ska->sk_type != SOCK_DGRAM) {
ska->sk_state = TCP_ESTABLISHED;
skb->sk_state = TCP_ESTABLISHED;
socka->state = SS_CONNECTED;
sockb->state = SS_CONNECTED;
}
return 0;
}
static int unix_accept(struct socket *sock, struct socket *newsock, int flags)
{
struct sock *sk = sock->sk;
struct sock *tsk;
struct sk_buff *skb;
int err;
err = -EOPNOTSUPP;
if (sock->type != SOCK_STREAM && sock->type != SOCK_SEQPACKET)
goto out;
err = -EINVAL;
if (sk->sk_state != TCP_LISTEN)
goto out;
/* If socket state is TCP_LISTEN it cannot change (for now...),
* so that no locks are necessary.
*/
skb = skb_recv_datagram(sk, 0, flags&O_NONBLOCK, &err);
if (!skb) {
/* This means receive shutdown. */
if (err == 0)
err = -EINVAL;
goto out;
}
tsk = skb->sk;
skb_free_datagram(sk, skb);
wake_up_interruptible(&unix_sk(sk)->peer_wait);
/* attach accepted sock to socket */
unix_state_lock(tsk);
newsock->state = SS_CONNECTED;
sock_graft(tsk, newsock);
unix_state_unlock(tsk);
return 0;
out:
return err;
}
static int unix_getname(struct socket *sock, struct sockaddr *uaddr, int *uaddr_len, int peer)
{
struct sock *sk = sock->sk;
struct unix_sock *u;
DECLARE_SOCKADDR(struct sockaddr_un *, sunaddr, uaddr);
int err = 0;
if (peer) {
sk = unix_peer_get(sk);
err = -ENOTCONN;
if (!sk)
goto out;
err = 0;
} else {
sock_hold(sk);
}
u = unix_sk(sk);
unix_state_lock(sk);
if (!u->addr) {
sunaddr->sun_family = AF_UNIX;
sunaddr->sun_path[0] = 0;
*uaddr_len = sizeof(short);
} else {
struct unix_address *addr = u->addr;
*uaddr_len = addr->len;
memcpy(sunaddr, addr->name, *uaddr_len);
}
unix_state_unlock(sk);
sock_put(sk);
out:
return err;
}
static void unix_detach_fds(struct scm_cookie *scm, struct sk_buff *skb)
{
int i;
scm->fp = UNIXCB(skb).fp;
UNIXCB(skb).fp = NULL;
for (i = scm->fp->count-1; i >= 0; i--)
unix_notinflight(scm->fp->fp[i]);
}
static void unix_destruct_scm(struct sk_buff *skb)
{
struct scm_cookie scm;
memset(&scm, 0, sizeof(scm));
scm.pid = UNIXCB(skb).pid;
if (UNIXCB(skb).fp)
unix_detach_fds(&scm, skb);
/* Alas, it calls VFS */
/* So fscking what? fput() had been SMP-safe since the last Summer */
scm_destroy(&scm);
sock_wfree(skb);
}
#define MAX_RECURSION_LEVEL 4
static int unix_attach_fds(struct scm_cookie *scm, struct sk_buff *skb)
{
int i;
unsigned char max_level = 0;
int unix_sock_count = 0;
for (i = scm->fp->count - 1; i >= 0; i--) {
struct sock *sk = unix_get_socket(scm->fp->fp[i]);
if (sk) {
unix_sock_count++;
max_level = max(max_level,
unix_sk(sk)->recursion_level);
}
}
if (unlikely(max_level > MAX_RECURSION_LEVEL))
return -ETOOMANYREFS;
/*
* Need to duplicate file references for the sake of garbage
* collection. Otherwise a socket in the fps might become a
* candidate for GC while the skb is not yet queued.
*/
UNIXCB(skb).fp = scm_fp_dup(scm->fp);
if (!UNIXCB(skb).fp)
return -ENOMEM;
if (unix_sock_count) {
for (i = scm->fp->count - 1; i >= 0; i--)
unix_inflight(scm->fp->fp[i]);
}
return max_level;
}
static int unix_scm_to_skb(struct scm_cookie *scm, struct sk_buff *skb, bool send_fds)
{
int err = 0;
af_unix: dont send SCM_CREDENTIALS by default Since commit 7361c36c5224 (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>
2011-09-19 13:52:27 +08:00
UNIXCB(skb).pid = get_pid(scm->pid);
UNIXCB(skb).uid = scm->creds.uid;
UNIXCB(skb).gid = scm->creds.gid;
UNIXCB(skb).fp = NULL;
if (scm->fp && send_fds)
err = unix_attach_fds(scm, skb);
skb->destructor = unix_destruct_scm;
return err;
}
af_unix: dont send SCM_CREDENTIALS by default Since commit 7361c36c5224 (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>
2011-09-19 13:52:27 +08:00
/*
* Some apps rely on write() giving SCM_CREDENTIALS
* We include credentials if source or destination socket
* asserted SOCK_PASSCRED.
*/
static void maybe_add_creds(struct sk_buff *skb, const struct socket *sock,
const struct sock *other)
{
if (UNIXCB(skb).pid)
af_unix: dont send SCM_CREDENTIALS by default Since commit 7361c36c5224 (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>
2011-09-19 13:52:27 +08:00
return;
if (test_bit(SOCK_PASSCRED, &sock->flags) ||
!other->sk_socket ||
test_bit(SOCK_PASSCRED, &other->sk_socket->flags)) {
af_unix: dont send SCM_CREDENTIALS by default Since commit 7361c36c5224 (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>
2011-09-19 13:52:27 +08:00
UNIXCB(skb).pid = get_pid(task_tgid(current));
current_uid_gid(&UNIXCB(skb).uid, &UNIXCB(skb).gid);
af_unix: dont send SCM_CREDENTIALS by default Since commit 7361c36c5224 (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>
2011-09-19 13:52:27 +08:00
}
}
/*
* Send AF_UNIX data.
*/
static int unix_dgram_sendmsg(struct kiocb *kiocb, struct socket *sock,
struct msghdr *msg, size_t len)
{
struct sock_iocb *siocb = kiocb_to_siocb(kiocb);
struct sock *sk = sock->sk;
struct net *net = sock_net(sk);
struct unix_sock *u = unix_sk(sk);
struct sockaddr_un *sunaddr = msg->msg_name;
struct sock *other = NULL;
int namelen = 0; /* fake GCC */
int err;
unsigned int hash;
struct sk_buff *skb;
long timeo;
struct scm_cookie tmp_scm;
int max_level;
int data_len = 0;
if (NULL == siocb->scm)
siocb->scm = &tmp_scm;
wait_for_unix_gc();
err = scm_send(sock, msg, siocb->scm, false);
if (err < 0)
return err;
err = -EOPNOTSUPP;
if (msg->msg_flags&MSG_OOB)
goto out;
if (msg->msg_namelen) {
err = unix_mkname(sunaddr, msg->msg_namelen, &hash);
if (err < 0)
goto out;
namelen = err;
} else {
sunaddr = NULL;
err = -ENOTCONN;
other = unix_peer_get(sk);
if (!other)
goto out;
}
if (test_bit(SOCK_PASSCRED, &sock->flags) && !u->addr
&& (err = unix_autobind(sock)) != 0)
goto out;
err = -EMSGSIZE;
if (len > sk->sk_sndbuf - 32)
goto out;
if (len > SKB_MAX_ALLOC)
data_len = min_t(size_t,
len - SKB_MAX_ALLOC,
MAX_SKB_FRAGS * PAGE_SIZE);
skb = sock_alloc_send_pskb(sk, len - data_len, data_len,
msg->msg_flags & MSG_DONTWAIT, &err,
PAGE_ALLOC_COSTLY_ORDER);
if (skb == NULL)
goto out;
err = unix_scm_to_skb(siocb->scm, skb, true);
if (err < 0)
goto out_free;
max_level = err + 1;
unix_get_secdata(siocb->scm, skb);
[AF_UNIX]: Datagram getpeersec This patch implements an API whereby an application can determine the label of its peer's Unix datagram sockets via the auxiliary data mechanism of recvmsg. Patch purpose: This patch enables a security-aware application to retrieve the security context of the peer of a Unix datagram socket. The application can then use this security context to determine the security context for processing on behalf of the peer who sent the packet. Patch design and implementation: The design and implementation is very similar to the UDP case for INET sockets. Basically we build upon the existing Unix domain socket API for retrieving user credentials. Linux offers the API for obtaining user credentials via ancillary messages (i.e., out of band/control messages that are bundled together with a normal message). To retrieve the security context, the application first indicates to the kernel such desire by setting the SO_PASSSEC option via getsockopt. Then the application retrieves the security context using the auxiliary data mechanism. An example server application for Unix datagram socket should look like this: toggle = 1; toggle_len = sizeof(toggle); setsockopt(sockfd, SOL_SOCKET, SO_PASSSEC, &toggle, &toggle_len); recvmsg(sockfd, &msg_hdr, 0); if (msg_hdr.msg_controllen > sizeof(struct cmsghdr)) { cmsg_hdr = CMSG_FIRSTHDR(&msg_hdr); if (cmsg_hdr->cmsg_len <= CMSG_LEN(sizeof(scontext)) && cmsg_hdr->cmsg_level == SOL_SOCKET && cmsg_hdr->cmsg_type == SCM_SECURITY) { memcpy(&scontext, CMSG_DATA(cmsg_hdr), sizeof(scontext)); } } sock_setsockopt is enhanced with a new socket option SOCK_PASSSEC to allow a server socket to receive security context of the peer. Testing: We have tested the patch by setting up Unix datagram client and server applications. We verified that the server can retrieve the security context using the auxiliary data mechanism of recvmsg. Signed-off-by: Catherine Zhang <cxzhang@watson.ibm.com> Acked-by: Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-06-30 03:27:47 +08:00
skb_put(skb, len - data_len);
skb->data_len = data_len;
skb->len = len;
err = skb_copy_datagram_from_iovec(skb, 0, msg->msg_iov, 0, len);
if (err)
goto out_free;
timeo = sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT);
restart:
if (!other) {
err = -ECONNRESET;
if (sunaddr == NULL)
goto out_free;
other = unix_find_other(net, sunaddr, namelen, sk->sk_type,
hash, &err);
if (other == NULL)
goto out_free;
}
2011-01-18 14:39:15 +08:00
if (sk_filter(other, skb) < 0) {
/* Toss the packet but do not return any error to the sender */
err = len;
goto out_free;
}
unix_state_lock(other);
err = -EPERM;
if (!unix_may_send(sk, other))
goto out_unlock;
if (sock_flag(other, SOCK_DEAD)) {
/*
* Check with 1003.1g - what should
* datagram error
*/
unix_state_unlock(other);
sock_put(other);
err = 0;
unix_state_lock(sk);
if (unix_peer(sk) == other) {
unix_peer(sk) = NULL;
unix_state_unlock(sk);
unix_dgram_disconnected(sk, other);
sock_put(other);
err = -ECONNREFUSED;
} else {
unix_state_unlock(sk);
}
other = NULL;
if (err)
goto out_free;
goto restart;
}
err = -EPIPE;
if (other->sk_shutdown & RCV_SHUTDOWN)
goto out_unlock;
if (sk->sk_type != SOCK_SEQPACKET) {
err = security_unix_may_send(sk->sk_socket, other->sk_socket);
if (err)
goto out_unlock;
}
af_unix: fix 'poll for write'/ connected DGRAM sockets The unix_dgram_sendmsg routine implements a (somewhat crude) form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for connected PF_UNIX non-stream sockets when the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_sendmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-18 13:28:05 +08:00
if (unix_peer(other) != sk && unix_recvq_full(other)) {
if (!timeo) {
err = -EAGAIN;
goto out_unlock;
}
timeo = unix_wait_for_peer(other, timeo);
err = sock_intr_errno(timeo);
if (signal_pending(current))
goto out_free;
goto restart;
}
if (sock_flag(other, SOCK_RCVTSTAMP))
__net_timestamp(skb);
af_unix: dont send SCM_CREDENTIALS by default Since commit 7361c36c5224 (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>
2011-09-19 13:52:27 +08:00
maybe_add_creds(skb, sock, other);
skb_queue_tail(&other->sk_receive_queue, skb);
if (max_level > unix_sk(other)->recursion_level)
unix_sk(other)->recursion_level = max_level;
unix_state_unlock(other);
other->sk_data_ready(other, len);
sock_put(other);
scm_destroy(siocb->scm);
return len;
out_unlock:
unix_state_unlock(other);
out_free:
kfree_skb(skb);
out:
if (other)
sock_put(other);
scm_destroy(siocb->scm);
return err;
}
/* We use paged skbs for stream sockets, and limit occupancy to 32768
* bytes, and a minimun of a full page.
*/
#define UNIX_SKB_FRAGS_SZ (PAGE_SIZE << get_order(32768))
static int unix_stream_sendmsg(struct kiocb *kiocb, struct socket *sock,
struct msghdr *msg, size_t len)
{
struct sock_iocb *siocb = kiocb_to_siocb(kiocb);
struct sock *sk = sock->sk;
struct sock *other = NULL;
int err, size;
struct sk_buff *skb;
int sent = 0;
struct scm_cookie tmp_scm;
bool fds_sent = false;
int max_level;
int data_len;
if (NULL == siocb->scm)
siocb->scm = &tmp_scm;
wait_for_unix_gc();
err = scm_send(sock, msg, siocb->scm, false);
if (err < 0)
return err;
err = -EOPNOTSUPP;
if (msg->msg_flags&MSG_OOB)
goto out_err;
if (msg->msg_namelen) {
err = sk->sk_state == TCP_ESTABLISHED ? -EISCONN : -EOPNOTSUPP;
goto out_err;
} else {
err = -ENOTCONN;
other = unix_peer(sk);
if (!other)
goto out_err;
}
if (sk->sk_shutdown & SEND_SHUTDOWN)
goto pipe_err;
while (sent < len) {
size = len - sent;
/* Keep two messages in the pipe so it schedules better */
size = min_t(int, size, (sk->sk_sndbuf >> 1) - 64);
/* allow fallback to order-0 allocations */
size = min_t(int, size, SKB_MAX_HEAD(0) + UNIX_SKB_FRAGS_SZ);
data_len = max_t(int, 0, size - SKB_MAX_HEAD(0));
skb = sock_alloc_send_pskb(sk, size - data_len, data_len,
msg->msg_flags & MSG_DONTWAIT, &err,
get_order(UNIX_SKB_FRAGS_SZ));
if (!skb)
goto out_err;
/* Only send the fds in the first buffer */
err = unix_scm_to_skb(siocb->scm, skb, !fds_sent);
if (err < 0) {
kfree_skb(skb);
goto out_err;
}
max_level = err + 1;
fds_sent = true;
skb_put(skb, size - data_len);
skb->data_len = data_len;
skb->len = size;
err = skb_copy_datagram_from_iovec(skb, 0, msg->msg_iov,
sent, size);
if (err) {
kfree_skb(skb);
goto out_err;
}
unix_state_lock(other);
if (sock_flag(other, SOCK_DEAD) ||
(other->sk_shutdown & RCV_SHUTDOWN))
goto pipe_err_free;
af_unix: dont send SCM_CREDENTIALS by default Since commit 7361c36c5224 (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>
2011-09-19 13:52:27 +08:00
maybe_add_creds(skb, sock, other);
skb_queue_tail(&other->sk_receive_queue, skb);
if (max_level > unix_sk(other)->recursion_level)
unix_sk(other)->recursion_level = max_level;
unix_state_unlock(other);
other->sk_data_ready(other, size);
sent += size;
}
scm_destroy(siocb->scm);
siocb->scm = NULL;
return sent;
pipe_err_free:
unix_state_unlock(other);
kfree_skb(skb);
pipe_err:
if (sent == 0 && !(msg->msg_flags&MSG_NOSIGNAL))
send_sig(SIGPIPE, current, 0);
err = -EPIPE;
out_err:
scm_destroy(siocb->scm);
siocb->scm = NULL;
return sent ? : err;
}
static int unix_seqpacket_sendmsg(struct kiocb *kiocb, struct socket *sock,
struct msghdr *msg, size_t len)
{
int err;
struct sock *sk = sock->sk;
err = sock_error(sk);
if (err)
return err;
if (sk->sk_state != TCP_ESTABLISHED)
return -ENOTCONN;
if (msg->msg_namelen)
msg->msg_namelen = 0;
return unix_dgram_sendmsg(kiocb, sock, msg, len);
}
static int unix_seqpacket_recvmsg(struct kiocb *iocb, struct socket *sock,
struct msghdr *msg, size_t size,
int flags)
{
struct sock *sk = sock->sk;
if (sk->sk_state != TCP_ESTABLISHED)
return -ENOTCONN;
return unix_dgram_recvmsg(iocb, sock, msg, size, flags);
}
static void unix_copy_addr(struct msghdr *msg, struct sock *sk)
{
struct unix_sock *u = unix_sk(sk);
msg->msg_namelen = 0;
if (u->addr) {
msg->msg_namelen = u->addr->len;
memcpy(msg->msg_name, u->addr->name, u->addr->len);
}
}
static int unix_dgram_recvmsg(struct kiocb *iocb, struct socket *sock,
struct msghdr *msg, size_t size,
int flags)
{
struct sock_iocb *siocb = kiocb_to_siocb(iocb);
struct scm_cookie tmp_scm;
struct sock *sk = sock->sk;
struct unix_sock *u = unix_sk(sk);
int noblock = flags & MSG_DONTWAIT;
struct sk_buff *skb;
int err;
int peeked, skip;
err = -EOPNOTSUPP;
if (flags&MSG_OOB)
goto out;
msg->msg_namelen = 0;
err = mutex_lock_interruptible(&u->readlock);
if (err) {
err = sock_intr_errno(sock_rcvtimeo(sk, noblock));
goto out;
}
skip = sk_peek_offset(sk, flags);
skb = __skb_recv_datagram(sk, flags, &peeked, &skip, &err);
[UNIX]: EOF on non-blocking SOCK_SEQPACKET I am not absolutely sure whether this actually is a bug (as in: I've got no clue what the standards say or what other implementations do), but at least I was pretty surprised when I noticed that a recv() on a non-blocking unix domain socket of type SOCK_SEQPACKET (which is connection oriented, after all) where the remote end has closed the connection returned -1 (EAGAIN) rather than 0 to indicate end of file. This is a test case: | #include <sys/types.h> | #include <unistd.h> | #include <sys/socket.h> | #include <sys/un.h> | #include <fcntl.h> | #include <string.h> | #include <stdlib.h> | | int main(){ | int sock; | struct sockaddr_un addr; | char buf[4096]; | int pfds[2]; | | pipe(pfds); | sock=socket(PF_UNIX,SOCK_SEQPACKET,0); | addr.sun_family=AF_UNIX; | strcpy(addr.sun_path,"/tmp/foobar_testsock"); | bind(sock,(struct sockaddr *)&addr,sizeof(addr)); | listen(sock,1); | if(fork()){ | close(sock); | sock=socket(PF_UNIX,SOCK_SEQPACKET,0); | connect(sock,(struct sockaddr *)&addr,sizeof(addr)); | fcntl(sock,F_SETFL,fcntl(sock,F_GETFL)|O_NONBLOCK); | close(pfds[1]); | read(pfds[0],buf,sizeof(buf)); | recv(sock,buf,sizeof(buf),0); // <-- this one | }else accept(sock,NULL,NULL); | exit(0); | } If you try it, make sure /tmp/foobar_testsock doesn't exist. The marked recv() returns -1 (EAGAIN) on 2.6.23.9. Below you find a patch that fixes that. Signed-off-by: Florian Zumbiehl <florz@florz.de> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2007-11-29 20:19:23 +08:00
if (!skb) {
unix_state_lock(sk);
/* Signal EOF on disconnected non-blocking SEQPACKET socket. */
if (sk->sk_type == SOCK_SEQPACKET && err == -EAGAIN &&
(sk->sk_shutdown & RCV_SHUTDOWN))
err = 0;
unix_state_unlock(sk);
goto out_unlock;
[UNIX]: EOF on non-blocking SOCK_SEQPACKET I am not absolutely sure whether this actually is a bug (as in: I've got no clue what the standards say or what other implementations do), but at least I was pretty surprised when I noticed that a recv() on a non-blocking unix domain socket of type SOCK_SEQPACKET (which is connection oriented, after all) where the remote end has closed the connection returned -1 (EAGAIN) rather than 0 to indicate end of file. This is a test case: | #include <sys/types.h> | #include <unistd.h> | #include <sys/socket.h> | #include <sys/un.h> | #include <fcntl.h> | #include <string.h> | #include <stdlib.h> | | int main(){ | int sock; | struct sockaddr_un addr; | char buf[4096]; | int pfds[2]; | | pipe(pfds); | sock=socket(PF_UNIX,SOCK_SEQPACKET,0); | addr.sun_family=AF_UNIX; | strcpy(addr.sun_path,"/tmp/foobar_testsock"); | bind(sock,(struct sockaddr *)&addr,sizeof(addr)); | listen(sock,1); | if(fork()){ | close(sock); | sock=socket(PF_UNIX,SOCK_SEQPACKET,0); | connect(sock,(struct sockaddr *)&addr,sizeof(addr)); | fcntl(sock,F_SETFL,fcntl(sock,F_GETFL)|O_NONBLOCK); | close(pfds[1]); | read(pfds[0],buf,sizeof(buf)); | recv(sock,buf,sizeof(buf),0); // <-- this one | }else accept(sock,NULL,NULL); | exit(0); | } If you try it, make sure /tmp/foobar_testsock doesn't exist. The marked recv() returns -1 (EAGAIN) on 2.6.23.9. Below you find a patch that fixes that. Signed-off-by: Florian Zumbiehl <florz@florz.de> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2007-11-29 20:19:23 +08:00
}
wake_up_interruptible_sync_poll(&u->peer_wait,
POLLOUT | POLLWRNORM | POLLWRBAND);
if (msg->msg_name)
unix_copy_addr(msg, skb->sk);
if (size > skb->len - skip)
size = skb->len - skip;
else if (size < skb->len - skip)
msg->msg_flags |= MSG_TRUNC;
err = skb_copy_datagram_iovec(skb, skip, msg->msg_iov, size);
if (err)
goto out_free;
if (sock_flag(sk, SOCK_RCVTSTAMP))
__sock_recv_timestamp(msg, sk, skb);
if (!siocb->scm) {
siocb->scm = &tmp_scm;
memset(&tmp_scm, 0, sizeof(tmp_scm));
}
scm_set_cred(siocb->scm, UNIXCB(skb).pid, UNIXCB(skb).uid, UNIXCB(skb).gid);
[AF_UNIX]: Datagram getpeersec This patch implements an API whereby an application can determine the label of its peer's Unix datagram sockets via the auxiliary data mechanism of recvmsg. Patch purpose: This patch enables a security-aware application to retrieve the security context of the peer of a Unix datagram socket. The application can then use this security context to determine the security context for processing on behalf of the peer who sent the packet. Patch design and implementation: The design and implementation is very similar to the UDP case for INET sockets. Basically we build upon the existing Unix domain socket API for retrieving user credentials. Linux offers the API for obtaining user credentials via ancillary messages (i.e., out of band/control messages that are bundled together with a normal message). To retrieve the security context, the application first indicates to the kernel such desire by setting the SO_PASSSEC option via getsockopt. Then the application retrieves the security context using the auxiliary data mechanism. An example server application for Unix datagram socket should look like this: toggle = 1; toggle_len = sizeof(toggle); setsockopt(sockfd, SOL_SOCKET, SO_PASSSEC, &toggle, &toggle_len); recvmsg(sockfd, &msg_hdr, 0); if (msg_hdr.msg_controllen > sizeof(struct cmsghdr)) { cmsg_hdr = CMSG_FIRSTHDR(&msg_hdr); if (cmsg_hdr->cmsg_len <= CMSG_LEN(sizeof(scontext)) && cmsg_hdr->cmsg_level == SOL_SOCKET && cmsg_hdr->cmsg_type == SCM_SECURITY) { memcpy(&scontext, CMSG_DATA(cmsg_hdr), sizeof(scontext)); } } sock_setsockopt is enhanced with a new socket option SOCK_PASSSEC to allow a server socket to receive security context of the peer. Testing: We have tested the patch by setting up Unix datagram client and server applications. We verified that the server can retrieve the security context using the auxiliary data mechanism of recvmsg. Signed-off-by: Catherine Zhang <cxzhang@watson.ibm.com> Acked-by: Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-06-30 03:27:47 +08:00
unix_set_secdata(siocb->scm, skb);
if (!(flags & MSG_PEEK)) {
if (UNIXCB(skb).fp)
unix_detach_fds(siocb->scm, skb);
sk_peek_offset_bwd(sk, skb->len);
} else {
/* It is questionable: on PEEK we could:
- do not return fds - good, but too simple 8)
- return fds, and do not return them on read (old strategy,
apparently wrong)
- clone fds (I chose it for now, it is the most universal
solution)
POSIX 1003.1g does not actually define this clearly
at all. POSIX 1003.1g doesn't define a lot of things
clearly however!
*/
sk_peek_offset_fwd(sk, size);
if (UNIXCB(skb).fp)
siocb->scm->fp = scm_fp_dup(UNIXCB(skb).fp);
}
err = (flags & MSG_TRUNC) ? skb->len - skip : size;
scm_recv(sock, msg, siocb->scm, flags);
out_free:
skb_free_datagram(sk, skb);
out_unlock:
mutex_unlock(&u->readlock);
out:
return err;
}
/*
* Sleep until more data has arrived. But check for races..
*/
static long unix_stream_data_wait(struct sock *sk, long timeo,
struct sk_buff *last)
{
DEFINE_WAIT(wait);
unix_state_lock(sk);
for (;;) {
prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
if (skb_peek_tail(&sk->sk_receive_queue) != last ||
sk->sk_err ||
(sk->sk_shutdown & RCV_SHUTDOWN) ||
signal_pending(current) ||
!timeo)
break;
set_bit(SOCK_ASYNC_WAITDATA, &sk->sk_socket->flags);
unix_state_unlock(sk);
timeo = freezable_schedule_timeout(timeo);
unix_state_lock(sk);
clear_bit(SOCK_ASYNC_WAITDATA, &sk->sk_socket->flags);
}
finish_wait(sk_sleep(sk), &wait);
unix_state_unlock(sk);
return timeo;
}
static unsigned int unix_skb_len(const struct sk_buff *skb)
{
return skb->len - UNIXCB(skb).consumed;
}
static int unix_stream_recvmsg(struct kiocb *iocb, struct socket *sock,
struct msghdr *msg, size_t size,
int flags)
{
struct sock_iocb *siocb = kiocb_to_siocb(iocb);
struct scm_cookie tmp_scm;
struct sock *sk = sock->sk;
struct unix_sock *u = unix_sk(sk);
struct sockaddr_un *sunaddr = msg->msg_name;
int copied = 0;
int check_creds = 0;
int target;
int err = 0;
long timeo;
int skip;
err = -EINVAL;
if (sk->sk_state != TCP_ESTABLISHED)
goto out;
err = -EOPNOTSUPP;
if (flags&MSG_OOB)
goto out;
target = sock_rcvlowat(sk, flags&MSG_WAITALL, size);
timeo = sock_rcvtimeo(sk, flags&MSG_DONTWAIT);
msg->msg_namelen = 0;
/* Lock the socket to prevent queue disordering
* while sleeps in memcpy_tomsg
*/
if (!siocb->scm) {
siocb->scm = &tmp_scm;
memset(&tmp_scm, 0, sizeof(tmp_scm));
}
err = mutex_lock_interruptible(&u->readlock);
if (err) {
err = sock_intr_errno(timeo);
goto out;
}
do {
int chunk;
struct sk_buff *skb, *last;
unix_state_lock(sk);
last = skb = skb_peek(&sk->sk_receive_queue);
again:
if (skb == NULL) {
unix_sk(sk)->recursion_level = 0;
if (copied >= target)
goto unlock;
/*
* POSIX 1003.1g mandates this order.
*/
err = sock_error(sk);
if (err)
goto unlock;
if (sk->sk_shutdown & RCV_SHUTDOWN)
goto unlock;
unix_state_unlock(sk);
err = -EAGAIN;
if (!timeo)
break;
mutex_unlock(&u->readlock);
timeo = unix_stream_data_wait(sk, timeo, last);
if (signal_pending(current)
|| mutex_lock_interruptible(&u->readlock)) {
err = sock_intr_errno(timeo);
goto out;
}
continue;
unlock:
unix_state_unlock(sk);
break;
}
skip = sk_peek_offset(sk, flags);
while (skip >= unix_skb_len(skb)) {
skip -= unix_skb_len(skb);
last = skb;
skb = skb_peek_next(skb, &sk->sk_receive_queue);
if (!skb)
goto again;
}
unix_state_unlock(sk);
if (check_creds) {
/* Never glue messages from different writers */
if ((UNIXCB(skb).pid != siocb->scm->pid) ||
!uid_eq(UNIXCB(skb).uid, siocb->scm->creds.uid) ||
!gid_eq(UNIXCB(skb).gid, siocb->scm->creds.gid))
break;
} else if (test_bit(SOCK_PASSCRED, &sock->flags)) {
/* Copy credentials */
scm_set_cred(siocb->scm, UNIXCB(skb).pid, UNIXCB(skb).uid, UNIXCB(skb).gid);
check_creds = 1;
}
/* Copy address just once */
if (sunaddr) {
unix_copy_addr(msg, skb->sk);
sunaddr = NULL;
}
chunk = min_t(unsigned int, unix_skb_len(skb) - skip, size);
if (skb_copy_datagram_iovec(skb, UNIXCB(skb).consumed + skip,
msg->msg_iov, chunk)) {
if (copied == 0)
copied = -EFAULT;
break;
}
copied += chunk;
size -= chunk;
/* Mark read part of skb as used */
if (!(flags & MSG_PEEK)) {
UNIXCB(skb).consumed += chunk;
sk_peek_offset_bwd(sk, chunk);
if (UNIXCB(skb).fp)
unix_detach_fds(siocb->scm, skb);
if (unix_skb_len(skb))
break;
skb_unlink(skb, &sk->sk_receive_queue);
consume_skb(skb);
if (siocb->scm->fp)
break;
} else {
/* It is questionable, see note in unix_dgram_recvmsg.
*/
if (UNIXCB(skb).fp)
siocb->scm->fp = scm_fp_dup(UNIXCB(skb).fp);
sk_peek_offset_fwd(sk, chunk);
break;
}
} while (size);
mutex_unlock(&u->readlock);
scm_recv(sock, msg, siocb->scm, flags);
out:
return copied ? : err;
}
static int unix_shutdown(struct socket *sock, int mode)
{
struct sock *sk = sock->sk;
struct sock *other;
if (mode < SHUT_RD || mode > SHUT_RDWR)
return -EINVAL;
/* This maps:
* SHUT_RD (0) -> RCV_SHUTDOWN (1)
* SHUT_WR (1) -> SEND_SHUTDOWN (2)
* SHUT_RDWR (2) -> SHUTDOWN_MASK (3)
*/
++mode;
unix_state_lock(sk);
sk->sk_shutdown |= mode;
other = unix_peer(sk);
if (other)
sock_hold(other);
unix_state_unlock(sk);
sk->sk_state_change(sk);
if (other &&
(sk->sk_type == SOCK_STREAM || sk->sk_type == SOCK_SEQPACKET)) {
int peer_mode = 0;
if (mode&RCV_SHUTDOWN)
peer_mode |= SEND_SHUTDOWN;
if (mode&SEND_SHUTDOWN)
peer_mode |= RCV_SHUTDOWN;
unix_state_lock(other);
other->sk_shutdown |= peer_mode;
unix_state_unlock(other);
other->sk_state_change(other);
if (peer_mode == SHUTDOWN_MASK)
sk_wake_async(other, SOCK_WAKE_WAITD, POLL_HUP);
else if (peer_mode & RCV_SHUTDOWN)
sk_wake_async(other, SOCK_WAKE_WAITD, POLL_IN);
}
if (other)
sock_put(other);
return 0;
}
long unix_inq_len(struct sock *sk)
{
struct sk_buff *skb;
long amount = 0;
if (sk->sk_state == TCP_LISTEN)
return -EINVAL;
spin_lock(&sk->sk_receive_queue.lock);
if (sk->sk_type == SOCK_STREAM ||
sk->sk_type == SOCK_SEQPACKET) {
skb_queue_walk(&sk->sk_receive_queue, skb)
amount += unix_skb_len(skb);
} else {
skb = skb_peek(&sk->sk_receive_queue);
if (skb)
amount = skb->len;
}
spin_unlock(&sk->sk_receive_queue.lock);
return amount;
}
EXPORT_SYMBOL_GPL(unix_inq_len);
long unix_outq_len(struct sock *sk)
{
return sk_wmem_alloc_get(sk);
}
EXPORT_SYMBOL_GPL(unix_outq_len);
static int unix_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
{
struct sock *sk = sock->sk;
long amount = 0;
int err;
switch (cmd) {
case SIOCOUTQ:
amount = unix_outq_len(sk);
err = put_user(amount, (int __user *)arg);
break;
case SIOCINQ:
amount = unix_inq_len(sk);
if (amount < 0)
err = amount;
else
err = put_user(amount, (int __user *)arg);
break;
default:
err = -ENOIOCTLCMD;
break;
}
return err;
}
static unsigned int unix_poll(struct file *file, struct socket *sock, poll_table *wait)
{
struct sock *sk = sock->sk;
unsigned int mask;
sock_poll_wait(file, sk_sleep(sk), wait);
mask = 0;
/* exceptional events? */
if (sk->sk_err)
mask |= POLLERR;
if (sk->sk_shutdown == SHUTDOWN_MASK)
mask |= POLLHUP;
if (sk->sk_shutdown & RCV_SHUTDOWN)
mask |= POLLRDHUP | POLLIN | POLLRDNORM;
/* readable? */
if (!skb_queue_empty(&sk->sk_receive_queue))
mask |= POLLIN | POLLRDNORM;
/* Connection-based need to check for termination and startup */
if ((sk->sk_type == SOCK_STREAM || sk->sk_type == SOCK_SEQPACKET) &&
sk->sk_state == TCP_CLOSE)
mask |= POLLHUP;
/*
* we set writable also when the other side has shut down the
* connection. This prevents stuck sockets.
*/
if (unix_writable(sk))
mask |= POLLOUT | POLLWRNORM | POLLWRBAND;
return mask;
}
af_unix: fix 'poll for write'/connected DGRAM sockets For n:1 'datagram connections' (eg /dev/log), the unix_dgram_sendmsg routine implements a form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for PF_UNIX non-stream sockets connected to server sockets dealing with (potentially) multiple clients if the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_recvmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-28 10:34:18 +08:00
static unsigned int unix_dgram_poll(struct file *file, struct socket *sock,
poll_table *wait)
af_unix: fix 'poll for write'/ connected DGRAM sockets The unix_dgram_sendmsg routine implements a (somewhat crude) form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for connected PF_UNIX non-stream sockets when the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_sendmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-18 13:28:05 +08:00
{
af_unix: fix 'poll for write'/connected DGRAM sockets For n:1 'datagram connections' (eg /dev/log), the unix_dgram_sendmsg routine implements a form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for PF_UNIX non-stream sockets connected to server sockets dealing with (potentially) multiple clients if the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_recvmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-28 10:34:18 +08:00
struct sock *sk = sock->sk, *other;
unsigned int mask, writable;
af_unix: fix 'poll for write'/ connected DGRAM sockets The unix_dgram_sendmsg routine implements a (somewhat crude) form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for connected PF_UNIX non-stream sockets when the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_sendmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-18 13:28:05 +08:00
sock_poll_wait(file, sk_sleep(sk), wait);
af_unix: fix 'poll for write'/ connected DGRAM sockets The unix_dgram_sendmsg routine implements a (somewhat crude) form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for connected PF_UNIX non-stream sockets when the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_sendmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-18 13:28:05 +08:00
mask = 0;
/* exceptional events? */
if (sk->sk_err || !skb_queue_empty(&sk->sk_error_queue))
mask |= POLLERR |
(sock_flag(sk, SOCK_SELECT_ERR_QUEUE) ? POLLPRI : 0);
af_unix: fix 'poll for write'/ connected DGRAM sockets The unix_dgram_sendmsg routine implements a (somewhat crude) form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for connected PF_UNIX non-stream sockets when the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_sendmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-18 13:28:05 +08:00
if (sk->sk_shutdown & RCV_SHUTDOWN)
mask |= POLLRDHUP | POLLIN | POLLRDNORM;
af_unix: fix 'poll for write'/ connected DGRAM sockets The unix_dgram_sendmsg routine implements a (somewhat crude) form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for connected PF_UNIX non-stream sockets when the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_sendmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-18 13:28:05 +08:00
if (sk->sk_shutdown == SHUTDOWN_MASK)
mask |= POLLHUP;
/* readable? */
if (!skb_queue_empty(&sk->sk_receive_queue))
af_unix: fix 'poll for write'/ connected DGRAM sockets The unix_dgram_sendmsg routine implements a (somewhat crude) form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for connected PF_UNIX non-stream sockets when the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_sendmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-18 13:28:05 +08:00
mask |= POLLIN | POLLRDNORM;
/* Connection-based need to check for termination and startup */
if (sk->sk_type == SOCK_SEQPACKET) {
if (sk->sk_state == TCP_CLOSE)
mask |= POLLHUP;
/* connection hasn't started yet? */
if (sk->sk_state == TCP_SYN_SENT)
return mask;
}
/* No write status requested, avoid expensive OUT tests. */
poll: add poll_requested_events() and poll_does_not_wait() functions In some cases the poll() implementation in a driver has to do different things depending on the events the caller wants to poll for. An example is when a driver needs to start a DMA engine if the caller polls for POLLIN, but doesn't want to do that if POLLIN is not requested but instead only POLLOUT or POLLPRI is requested. This is something that can happen in the video4linux subsystem among others. Unfortunately, the current epoll/poll/select implementation doesn't provide that information reliably. The poll_table_struct does have it: it has a key field with the event mask. But once a poll() call matches one or more bits of that mask any following poll() calls are passed a NULL poll_table pointer. Also, the eventpoll implementation always left the key field at ~0 instead of using the requested events mask. This was changed in eventpoll.c so the key field now contains the actual events that should be polled for as set by the caller. The solution to the NULL poll_table pointer is to set the qproc field to NULL in poll_table once poll() matches the events, not the poll_table pointer itself. That way drivers can obtain the mask through a new poll_requested_events inline. The poll_table_struct can still be NULL since some kernel code calls it internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In that case poll_requested_events() returns ~0 (i.e. all events). Very rarely drivers might want to know whether poll_wait will actually wait. If another earlier file descriptor in the set already matched the events the caller wanted to wait for, then the kernel will return from the select() call without waiting. This might be useful information in order to avoid doing expensive work. A new helper function poll_does_not_wait() is added that drivers can use to detect this situation. This is now used in sock_poll_wait() in include/net/sock.h. This was the only place in the kernel that needed this information. Drivers should no longer access any of the poll_table internals, but use the poll_requested_events() and poll_does_not_wait() access functions instead. In order to enforce that the poll_table fields are now prepended with an underscore and a comment was added warning against using them directly. This required a change in unix_dgram_poll() in unix/af_unix.c which used the key field to get the requested events. It's been replaced by a call to poll_requested_events(). For qproc it was especially important to change its name since the behavior of that field changes with this patch since this function pointer can now be NULL when that wasn't possible in the past. Any driver accessing the qproc or key fields directly will now fail to compile. Some notes regarding the correctness of this patch: the driver's poll() function is called with a 'struct poll_table_struct *wait' argument. This pointer may or may not be NULL, drivers can never rely on it being one or the other as that depends on whether or not an earlier file descriptor in the select()'s fdset matched the requested events. There are only three things a driver can do with the wait argument: 1) obtain the key field: events = wait ? wait->key : ~0; This will still work although it should be replaced with the new poll_requested_events() function (which does exactly the same). This will now even work better, since wait is no longer set to NULL unnecessarily. 2) use the qproc callback. This could be deadly since qproc can now be NULL. Renaming qproc should prevent this from happening. There are no kernel drivers that actually access this callback directly, BTW. 3) test whether wait == NULL to determine whether poll would return without waiting. This is no longer sufficient as the correct test is now wait == NULL || wait->_qproc == NULL. However, the worst that can happen here is a slight performance hit in the case where wait != NULL and wait->_qproc == NULL. In that case the driver will assume that poll_wait() will actually add the fd to the set of waiting file descriptors. Of course, poll_wait() will not do that since it tests for wait->_qproc. This will not break anything, though. There is only one place in the whole kernel where this happens (sock_poll_wait() in include/net/sock.h) and that code will be replaced by a call to poll_does_not_wait() in the next patch. Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait() actually waiting. The next file descriptor from the set might match the event mask and thus any possible waits will never happen. Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Reviewed-by: Jonathan Corbet <corbet@lwn.net> Reviewed-by: Al Viro <viro@zeniv.linux.org.uk> Cc: Davide Libenzi <davidel@xmailserver.org> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Cc: Mauro Carvalho Chehab <mchehab@infradead.org> Cc: David Miller <davem@davemloft.net> Cc: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 06:02:27 +08:00
if (!(poll_requested_events(wait) & (POLLWRBAND|POLLWRNORM|POLLOUT)))
return mask;
af_unix: fix 'poll for write'/connected DGRAM sockets For n:1 'datagram connections' (eg /dev/log), the unix_dgram_sendmsg routine implements a form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for PF_UNIX non-stream sockets connected to server sockets dealing with (potentially) multiple clients if the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_recvmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-28 10:34:18 +08:00
writable = unix_writable(sk);
other = unix_peer_get(sk);
if (other) {
if (unix_peer(other) != sk) {
sock_poll_wait(file, &unix_sk(other)->peer_wait, wait);
if (unix_recvq_full(other))
writable = 0;
af_unix: fix 'poll for write'/connected DGRAM sockets For n:1 'datagram connections' (eg /dev/log), the unix_dgram_sendmsg routine implements a form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for PF_UNIX non-stream sockets connected to server sockets dealing with (potentially) multiple clients if the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_recvmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-28 10:34:18 +08:00
}
sock_put(other);
af_unix: fix 'poll for write'/connected DGRAM sockets For n:1 'datagram connections' (eg /dev/log), the unix_dgram_sendmsg routine implements a form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for PF_UNIX non-stream sockets connected to server sockets dealing with (potentially) multiple clients if the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_recvmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-28 10:34:18 +08:00
}
if (writable)
af_unix: fix 'poll for write'/ connected DGRAM sockets The unix_dgram_sendmsg routine implements a (somewhat crude) form of receiver-imposed flow control by comparing the length of the receive queue of the 'peer socket' with the max_ack_backlog value stored in the corresponding sock structure, either blocking the thread which caused the send-routine to be called or returning EAGAIN. This routine is used by both SOCK_DGRAM and SOCK_SEQPACKET sockets. The poll-implementation for these socket types is datagram_poll from core/datagram.c. A socket is deemed to be writeable by this routine when the memory presently consumed by datagrams owned by it is less than the configured socket send buffer size. This is always wrong for connected PF_UNIX non-stream sockets when the abovementioned receive queue is currently considered to be full. 'poll' will then return, indicating that the socket is writeable, but a subsequent write result in EAGAIN, effectively causing an (usual) application to 'poll for writeability by repeated send request with O_NONBLOCK set' until it has consumed its time quantum. The change below uses a suitably modified variant of the datagram_poll routines for both type of PF_UNIX sockets, which tests if the recv-queue of the peer a socket is connected to is presently considered to be 'full' as part of the 'is this socket writeable'-checking code. The socket being polled is additionally put onto the peer_wait wait queue associated with its peer, because the unix_dgram_sendmsg routine does a wake up on this queue after a datagram was received and the 'other wakeup call' is done implicitly as part of skb destruction, meaning, a process blocked in poll because of a full peer receive queue could otherwise sleep forever if no datagram owned by its socket was already sitting on this queue. Among this change is a small (inline) helper routine named 'unix_recvq_full', which consolidates the actual testing code (in three different places) into a single location. Signed-off-by: Rainer Weikusat <rweikusat@mssgmbh.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-18 13:28:05 +08:00
mask |= POLLOUT | POLLWRNORM | POLLWRBAND;
else
set_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags);
return mask;
}
#ifdef CONFIG_PROC_FS
#define BUCKET_SPACE (BITS_PER_LONG - (UNIX_HASH_BITS + 1) - 1)
#define get_bucket(x) ((x) >> BUCKET_SPACE)
#define get_offset(x) ((x) & ((1L << BUCKET_SPACE) - 1))
#define set_bucket_offset(b, o) ((b) << BUCKET_SPACE | (o))
static struct sock *unix_from_bucket(struct seq_file *seq, loff_t *pos)
{
unsigned long offset = get_offset(*pos);
unsigned long bucket = get_bucket(*pos);
struct sock *sk;
unsigned long count = 0;
for (sk = sk_head(&unix_socket_table[bucket]); sk; sk = sk_next(sk)) {
if (sock_net(sk) != seq_file_net(seq))
continue;
if (++count == offset)
break;
}
return sk;
}
static struct sock *unix_next_socket(struct seq_file *seq,
struct sock *sk,
loff_t *pos)
{
unsigned long bucket;
while (sk > (struct sock *)SEQ_START_TOKEN) {
sk = sk_next(sk);
if (!sk)
goto next_bucket;
if (sock_net(sk) == seq_file_net(seq))
return sk;
}
do {
sk = unix_from_bucket(seq, pos);
if (sk)
return sk;
next_bucket:
bucket = get_bucket(*pos) + 1;
*pos = set_bucket_offset(bucket, 1);
} while (bucket < ARRAY_SIZE(unix_socket_table));
return NULL;
}
static void *unix_seq_start(struct seq_file *seq, loff_t *pos)
__acquires(unix_table_lock)
{
spin_lock(&unix_table_lock);
if (!*pos)
return SEQ_START_TOKEN;
if (get_bucket(*pos) >= ARRAY_SIZE(unix_socket_table))
return NULL;
return unix_next_socket(seq, NULL, pos);
}
static void *unix_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
++*pos;
return unix_next_socket(seq, v, pos);
}
static void unix_seq_stop(struct seq_file *seq, void *v)
__releases(unix_table_lock)
{
spin_unlock(&unix_table_lock);
}
static int unix_seq_show(struct seq_file *seq, void *v)
{
if (v == SEQ_START_TOKEN)
seq_puts(seq, "Num RefCount Protocol Flags Type St "
"Inode Path\n");
else {
struct sock *s = v;
struct unix_sock *u = unix_sk(s);
unix_state_lock(s);
net: convert %p usage to %pK The %pK format specifier is designed to hide exposed kernel pointers, specifically via /proc interfaces. Exposing these pointers provides an easy target for kernel write vulnerabilities, since they reveal the locations of writable structures containing easily triggerable function pointers. The behavior of %pK depends on the kptr_restrict sysctl. If kptr_restrict is set to 0, no deviation from the standard %p behavior occurs. If kptr_restrict is set to 1, the default, if the current user (intended to be a reader via seq_printf(), etc.) does not have CAP_SYSLOG (currently in the LSM tree), kernel pointers using %pK are printed as 0's. If kptr_restrict is set to 2, kernel pointers using %pK are printed as 0's regardless of privileges. Replacing with 0's was chosen over the default "(null)", which cannot be parsed by userland %p, which expects "(nil)". The supporting code for kptr_restrict and %pK are currently in the -mm tree. This patch converts users of %p in net/ to %pK. Cases of printing pointers to the syslog are not covered, since this would eliminate useful information for postmortem debugging and the reading of the syslog is already optionally protected by the dmesg_restrict sysctl. Signed-off-by: Dan Rosenberg <drosenberg@vsecurity.com> Cc: James Morris <jmorris@namei.org> Cc: Eric Dumazet <eric.dumazet@gmail.com> Cc: Thomas Graf <tgraf@infradead.org> Cc: Eugene Teo <eugeneteo@kernel.org> Cc: Kees Cook <kees.cook@canonical.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: David S. Miller <davem@davemloft.net> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Eric Paris <eparis@parisplace.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-05-23 20:17:35 +08:00
seq_printf(seq, "%pK: %08X %08X %08X %04X %02X %5lu",
s,
atomic_read(&s->sk_refcnt),
0,
s->sk_state == TCP_LISTEN ? __SO_ACCEPTCON : 0,
s->sk_type,
s->sk_socket ?
(s->sk_state == TCP_ESTABLISHED ? SS_CONNECTED : SS_UNCONNECTED) :
(s->sk_state == TCP_ESTABLISHED ? SS_CONNECTING : SS_DISCONNECTING),
sock_i_ino(s));
if (u->addr) {
int i, len;
seq_putc(seq, ' ');
i = 0;
len = u->addr->len - sizeof(short);
if (!UNIX_ABSTRACT(s))
len--;
else {
seq_putc(seq, '@');
i++;
}
for ( ; i < len; i++)
seq_putc(seq, u->addr->name->sun_path[i]);
}
unix_state_unlock(s);
seq_putc(seq, '\n');
}
return 0;
}
static const struct seq_operations unix_seq_ops = {
.start = unix_seq_start,
.next = unix_seq_next,
.stop = unix_seq_stop,
.show = unix_seq_show,
};
static int unix_seq_open(struct inode *inode, struct file *file)
{
return seq_open_net(inode, file, &unix_seq_ops,
sizeof(struct seq_net_private));
}
static const struct file_operations unix_seq_fops = {
.owner = THIS_MODULE,
.open = unix_seq_open,
.read = seq_read,
.llseek = seq_lseek,
.release = seq_release_net,
};
#endif
static const struct net_proto_family unix_family_ops = {
.family = PF_UNIX,
.create = unix_create,
.owner = THIS_MODULE,
};
static int __net_init unix_net_init(struct net *net)
{
int error = -ENOMEM;
net->unx.sysctl_max_dgram_qlen = 10;
if (unix_sysctl_register(net))
goto out;
#ifdef CONFIG_PROC_FS
if (!proc_create("unix", 0, net->proc_net, &unix_seq_fops)) {
unix_sysctl_unregister(net);
goto out;
}
#endif
error = 0;
out:
return error;
}
static void __net_exit unix_net_exit(struct net *net)
{
unix_sysctl_unregister(net);
remove_proc_entry("unix", net->proc_net);
}
static struct pernet_operations unix_net_ops = {
.init = unix_net_init,
.exit = unix_net_exit,
};
static int __init af_unix_init(void)
{
int rc = -1;
BUILD_BUG_ON(sizeof(struct unix_skb_parms) > FIELD_SIZEOF(struct sk_buff, cb));
rc = proto_register(&unix_proto, 1);
if (rc != 0) {
printk(KERN_CRIT "%s: Cannot create unix_sock SLAB cache!\n",
__func__);
goto out;
}
sock_register(&unix_family_ops);
register_pernet_subsys(&unix_net_ops);
out:
return rc;
}
static void __exit af_unix_exit(void)
{
sock_unregister(PF_UNIX);
proto_unregister(&unix_proto);
unregister_pernet_subsys(&unix_net_ops);
}
/* Earlier than device_initcall() so that other drivers invoking
request_module() don't end up in a loop when modprobe tries
to use a UNIX socket. But later than subsys_initcall() because
we depend on stuff initialised there */
fs_initcall(af_unix_init);
module_exit(af_unix_exit);
MODULE_LICENSE("GPL");
MODULE_ALIAS_NETPROTO(PF_UNIX);