linux/net/ipv4/udp.c

2309 lines
58 KiB
C
Raw Normal View History

/*
* INET An implementation of the TCP/IP protocol suite for the LINUX
* operating system. INET is implemented using the BSD Socket
* interface as the means of communication with the user level.
*
* The User Datagram Protocol (UDP).
*
* Authors: Ross Biro
* Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
* Arnt Gulbrandsen, <agulbra@nvg.unit.no>
* Alan Cox, <alan@lxorguk.ukuu.org.uk>
* Hirokazu Takahashi, <taka@valinux.co.jp>
*
* Fixes:
* Alan Cox : verify_area() calls
* Alan Cox : stopped close while in use off icmp
* messages. Not a fix but a botch that
* for udp at least is 'valid'.
* Alan Cox : Fixed icmp handling properly
* Alan Cox : Correct error for oversized datagrams
* Alan Cox : Tidied select() semantics.
* Alan Cox : udp_err() fixed properly, also now
* select and read wake correctly on errors
* Alan Cox : udp_send verify_area moved to avoid mem leak
* Alan Cox : UDP can count its memory
* Alan Cox : send to an unknown connection causes
* an ECONNREFUSED off the icmp, but
* does NOT close.
* Alan Cox : Switched to new sk_buff handlers. No more backlog!
* Alan Cox : Using generic datagram code. Even smaller and the PEEK
* bug no longer crashes it.
* Fred Van Kempen : Net2e support for sk->broadcast.
* Alan Cox : Uses skb_free_datagram
* Alan Cox : Added get/set sockopt support.
* Alan Cox : Broadcasting without option set returns EACCES.
* Alan Cox : No wakeup calls. Instead we now use the callbacks.
* Alan Cox : Use ip_tos and ip_ttl
* Alan Cox : SNMP Mibs
* Alan Cox : MSG_DONTROUTE, and 0.0.0.0 support.
* Matt Dillon : UDP length checks.
* Alan Cox : Smarter af_inet used properly.
* Alan Cox : Use new kernel side addressing.
* Alan Cox : Incorrect return on truncated datagram receive.
* Arnt Gulbrandsen : New udp_send and stuff
* Alan Cox : Cache last socket
* Alan Cox : Route cache
* Jon Peatfield : Minor efficiency fix to sendto().
* Mike Shaver : RFC1122 checks.
* Alan Cox : Nonblocking error fix.
* Willy Konynenberg : Transparent proxying support.
* Mike McLagan : Routing by source
* David S. Miller : New socket lookup architecture.
* Last socket cache retained as it
* does have a high hit rate.
* Olaf Kirch : Don't linearise iovec on sendmsg.
* Andi Kleen : Some cleanups, cache destination entry
* for connect.
* Vitaly E. Lavrov : Transparent proxy revived after year coma.
* Melvin Smith : Check msg_name not msg_namelen in sendto(),
* return ENOTCONN for unconnected sockets (POSIX)
* Janos Farkas : don't deliver multi/broadcasts to a different
* bound-to-device socket
* Hirokazu Takahashi : HW checksumming for outgoing UDP
* datagrams.
* Hirokazu Takahashi : sendfile() on UDP works now.
* Arnaldo C. Melo : convert /proc/net/udp to seq_file
* YOSHIFUJI Hideaki @USAGI and: Support IPV6_V6ONLY socket option, which
* Alexey Kuznetsov: allow both IPv4 and IPv6 sockets to bind
* a single port at the same time.
* Derek Atkins <derek@ihtfp.com>: Add Encapulation Support
* James Chapman : Add L2TP encapsulation type.
*
*
* 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.
*/
#define pr_fmt(fmt) "UDP: " fmt
#include <asm/uaccess.h>
#include <asm/ioctls.h>
#include <linux/bootmem.h>
#include <linux/highmem.h>
#include <linux/swap.h>
#include <linux/types.h>
#include <linux/fcntl.h>
#include <linux/module.h>
#include <linux/socket.h>
#include <linux/sockios.h>
#include <linux/igmp.h>
#include <linux/in.h>
#include <linux/errno.h>
#include <linux/timer.h>
#include <linux/mm.h>
#include <linux/inet.h>
#include <linux/netdevice.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
#include <net/tcp_states.h>
#include <linux/skbuff.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <net/net_namespace.h>
#include <net/icmp.h>
#include <net/route.h>
#include <net/checksum.h>
#include <net/xfrm.h>
#include <trace/events/udp.h>
#include <linux/static_key.h>
#include <trace/events/skb.h>
[NET]: Supporting UDP-Lite (RFC 3828) in Linux This is a revision of the previously submitted patch, which alters the way files are organized and compiled in the following manner: * UDP and UDP-Lite now use separate object files * source file dependencies resolved via header files net/ipv{4,6}/udp_impl.h * order of inclusion files in udp.c/udplite.c adapted accordingly [NET/IPv4]: Support for the UDP-Lite protocol (RFC 3828) This patch adds support for UDP-Lite to the IPv4 stack, provided as an extension to the existing UDPv4 code: * generic routines are all located in net/ipv4/udp.c * UDP-Lite specific routines are in net/ipv4/udplite.c * MIB/statistics support in /proc/net/snmp and /proc/net/udplite * shared API with extensions for partial checksum coverage [NET/IPv6]: Extension for UDP-Lite over IPv6 It extends the existing UDPv6 code base with support for UDP-Lite in the same manner as per UDPv4. In particular, * UDPv6 generic and shared code is in net/ipv6/udp.c * UDP-Litev6 specific extensions are in net/ipv6/udplite.c * MIB/statistics support in /proc/net/snmp6 and /proc/net/udplite6 * support for IPV6_ADDRFORM * aligned the coding style of protocol initialisation with af_inet6.c * made the error handling in udpv6_queue_rcv_skb consistent; to return `-1' on error on all error cases * consolidation of shared code [NET]: UDP-Lite Documentation and basic XFRM/Netfilter support The UDP-Lite patch further provides * API documentation for UDP-Lite * basic xfrm support * basic netfilter support for IPv4 and IPv6 (LOG target) Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-27 19:10:57 +00:00
#include "udp_impl.h"
struct udp_table udp_table __read_mostly;
EXPORT_SYMBOL(udp_table);
long sysctl_udp_mem[3] __read_mostly;
EXPORT_SYMBOL(sysctl_udp_mem);
int sysctl_udp_rmem_min __read_mostly;
EXPORT_SYMBOL(sysctl_udp_rmem_min);
int sysctl_udp_wmem_min __read_mostly;
EXPORT_SYMBOL(sysctl_udp_wmem_min);
atomic_long_t udp_memory_allocated;
EXPORT_SYMBOL(udp_memory_allocated);
#define MAX_UDP_PORTS 65536
#define PORTS_PER_CHAIN (MAX_UDP_PORTS / UDP_HTABLE_SIZE_MIN)
static int udp_lib_lport_inuse(struct net *net, __u16 num,
const struct udp_hslot *hslot,
unsigned long *bitmap,
struct sock *sk,
int (*saddr_comp)(const struct sock *sk1,
const struct sock *sk2),
unsigned int log)
{
struct sock *sk2;
struct hlist_nulls_node *node;
sk_nulls_for_each(sk2, node, &hslot->head)
if (net_eq(sock_net(sk2), net) &&
sk2 != sk &&
(bitmap || udp_sk(sk2)->udp_port_hash == num) &&
(!sk2->sk_reuse || !sk->sk_reuse) &&
(!sk2->sk_bound_dev_if || !sk->sk_bound_dev_if ||
sk2->sk_bound_dev_if == sk->sk_bound_dev_if) &&
(*saddr_comp)(sk, sk2)) {
if (bitmap)
__set_bit(udp_sk(sk2)->udp_port_hash >> log,
bitmap);
else
return 1;
}
return 0;
}
/*
* Note: we still hold spinlock of primary hash chain, so no other writer
* can insert/delete a socket with local_port == num
*/
static int udp_lib_lport_inuse2(struct net *net, __u16 num,
struct udp_hslot *hslot2,
struct sock *sk,
int (*saddr_comp)(const struct sock *sk1,
const struct sock *sk2))
{
struct sock *sk2;
struct hlist_nulls_node *node;
int res = 0;
spin_lock(&hslot2->lock);
udp_portaddr_for_each_entry(sk2, node, &hslot2->head)
if (net_eq(sock_net(sk2), net) &&
sk2 != sk &&
(udp_sk(sk2)->udp_port_hash == num) &&
(!sk2->sk_reuse || !sk->sk_reuse) &&
(!sk2->sk_bound_dev_if || !sk->sk_bound_dev_if ||
sk2->sk_bound_dev_if == sk->sk_bound_dev_if) &&
(*saddr_comp)(sk, sk2)) {
res = 1;
break;
}
spin_unlock(&hslot2->lock);
return res;
}
/**
* udp_lib_get_port - UDP/-Lite port lookup for IPv4 and IPv6
*
* @sk: socket struct in question
* @snum: port number to look up
* @saddr_comp: AF-dependent comparison of bound local IP addresses
* @hash2_nulladdr: AF-dependent hash value in secondary hash chains,
* with NULL address
*/
int udp_lib_get_port(struct sock *sk, unsigned short snum,
int (*saddr_comp)(const struct sock *sk1,
const struct sock *sk2),
unsigned int hash2_nulladdr)
{
struct udp_hslot *hslot, *hslot2;
struct udp_table *udptable = sk->sk_prot->h.udp_table;
int error = 1;
struct net *net = sock_net(sk);
if (!snum) {
int low, high, remaining;
unsigned int rand;
unsigned short first, last;
DECLARE_BITMAP(bitmap, PORTS_PER_CHAIN);
inet_get_local_port_range(&low, &high);
remaining = (high - low) + 1;
rand = net_random();
first = (((u64)rand * remaining) >> 32) + low;
/*
* force rand to be an odd multiple of UDP_HTABLE_SIZE
*/
rand = (rand | 1) * (udptable->mask + 1);
last = first + udptable->mask + 1;
do {
hslot = udp_hashslot(udptable, net, first);
bitmap_zero(bitmap, PORTS_PER_CHAIN);
spin_lock_bh(&hslot->lock);
udp_lib_lport_inuse(net, snum, hslot, bitmap, sk,
saddr_comp, udptable->log);
snum = first;
/*
* Iterate on all possible values of snum for this hash.
* Using steps of an odd multiple of UDP_HTABLE_SIZE
* give us randomization and full range coverage.
*/
do {
if (low <= snum && snum <= high &&
!test_bit(snum >> udptable->log, bitmap) &&
!inet_is_reserved_local_port(snum))
goto found;
snum += rand;
} while (snum != first);
spin_unlock_bh(&hslot->lock);
} while (++first != last);
goto fail;
} else {
hslot = udp_hashslot(udptable, net, snum);
spin_lock_bh(&hslot->lock);
if (hslot->count > 10) {
int exist;
unsigned int slot2 = udp_sk(sk)->udp_portaddr_hash ^ snum;
slot2 &= udptable->mask;
hash2_nulladdr &= udptable->mask;
hslot2 = udp_hashslot2(udptable, slot2);
if (hslot->count < hslot2->count)
goto scan_primary_hash;
exist = udp_lib_lport_inuse2(net, snum, hslot2,
sk, saddr_comp);
if (!exist && (hash2_nulladdr != slot2)) {
hslot2 = udp_hashslot2(udptable, hash2_nulladdr);
exist = udp_lib_lport_inuse2(net, snum, hslot2,
sk, saddr_comp);
}
if (exist)
goto fail_unlock;
else
goto found;
}
scan_primary_hash:
if (udp_lib_lport_inuse(net, snum, hslot, NULL, sk,
saddr_comp, 0))
goto fail_unlock;
}
found:
inet_sk(sk)->inet_num = snum;
udp_sk(sk)->udp_port_hash = snum;
udp_sk(sk)->udp_portaddr_hash ^= snum;
if (sk_unhashed(sk)) {
sk_nulls_add_node_rcu(sk, &hslot->head);
hslot->count++;
sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
hslot2 = udp_hashslot2(udptable, udp_sk(sk)->udp_portaddr_hash);
spin_lock(&hslot2->lock);
hlist_nulls_add_head_rcu(&udp_sk(sk)->udp_portaddr_node,
&hslot2->head);
hslot2->count++;
spin_unlock(&hslot2->lock);
}
error = 0;
fail_unlock:
spin_unlock_bh(&hslot->lock);
fail:
return error;
}
EXPORT_SYMBOL(udp_lib_get_port);
static int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
{
struct inet_sock *inet1 = inet_sk(sk1), *inet2 = inet_sk(sk2);
return (!ipv6_only_sock(sk2) &&
(!inet1->inet_rcv_saddr || !inet2->inet_rcv_saddr ||
inet1->inet_rcv_saddr == inet2->inet_rcv_saddr));
}
static unsigned int udp4_portaddr_hash(struct net *net, __be32 saddr,
unsigned int port)
{
return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
}
int udp_v4_get_port(struct sock *sk, unsigned short snum)
{
unsigned int hash2_nulladdr =
udp4_portaddr_hash(sock_net(sk), htonl(INADDR_ANY), snum);
unsigned int hash2_partial =
udp4_portaddr_hash(sock_net(sk), inet_sk(sk)->inet_rcv_saddr, 0);
/* precompute partial secondary hash */
udp_sk(sk)->udp_portaddr_hash = hash2_partial;
return udp_lib_get_port(sk, snum, ipv4_rcv_saddr_equal, hash2_nulladdr);
}
static inline int compute_score(struct sock *sk, struct net *net, __be32 saddr,
unsigned short hnum,
__be16 sport, __be32 daddr, __be16 dport, int dif)
{
int score = -1;
if (net_eq(sock_net(sk), net) && udp_sk(sk)->udp_port_hash == hnum &&
!ipv6_only_sock(sk)) {
struct inet_sock *inet = inet_sk(sk);
score = (sk->sk_family == PF_INET ? 1 : 0);
if (inet->inet_rcv_saddr) {
if (inet->inet_rcv_saddr != daddr)
return -1;
score += 2;
}
if (inet->inet_daddr) {
if (inet->inet_daddr != saddr)
return -1;
score += 2;
}
if (inet->inet_dport) {
if (inet->inet_dport != sport)
return -1;
score += 2;
}
if (sk->sk_bound_dev_if) {
if (sk->sk_bound_dev_if != dif)
return -1;
score += 2;
}
}
return score;
}
/*
* In this second variant, we check (daddr, dport) matches (inet_rcv_sadd, inet_num)
*/
#define SCORE2_MAX (1 + 2 + 2 + 2)
static inline int compute_score2(struct sock *sk, struct net *net,
__be32 saddr, __be16 sport,
__be32 daddr, unsigned int hnum, int dif)
{
int score = -1;
if (net_eq(sock_net(sk), net) && !ipv6_only_sock(sk)) {
struct inet_sock *inet = inet_sk(sk);
if (inet->inet_rcv_saddr != daddr)
return -1;
if (inet->inet_num != hnum)
return -1;
score = (sk->sk_family == PF_INET ? 1 : 0);
if (inet->inet_daddr) {
if (inet->inet_daddr != saddr)
return -1;
score += 2;
}
if (inet->inet_dport) {
if (inet->inet_dport != sport)
return -1;
score += 2;
}
if (sk->sk_bound_dev_if) {
if (sk->sk_bound_dev_if != dif)
return -1;
score += 2;
}
}
return score;
}
/* called with read_rcu_lock() */
static struct sock *udp4_lib_lookup2(struct net *net,
__be32 saddr, __be16 sport,
__be32 daddr, unsigned int hnum, int dif,
struct udp_hslot *hslot2, unsigned int slot2)
{
struct sock *sk, *result;
struct hlist_nulls_node *node;
int score, badness;
begin:
result = NULL;
badness = -1;
udp_portaddr_for_each_entry_rcu(sk, node, &hslot2->head) {
score = compute_score2(sk, net, saddr, sport,
daddr, hnum, dif);
if (score > badness) {
result = sk;
badness = score;
if (score == SCORE2_MAX)
goto exact_match;
}
}
/*
* if the nulls value we got at the end of this lookup is
* not the expected one, we must restart lookup.
* We probably met an item that was moved to another chain.
*/
if (get_nulls_value(node) != slot2)
goto begin;
if (result) {
exact_match:
if (unlikely(!atomic_inc_not_zero_hint(&result->sk_refcnt, 2)))
result = NULL;
else if (unlikely(compute_score2(result, net, saddr, sport,
daddr, hnum, dif) < badness)) {
sock_put(result);
goto begin;
}
}
return result;
}
/* UDP is nearly always wildcards out the wazoo, it makes no sense to try
* harder than this. -DaveM
*/
struct sock *__udp4_lib_lookup(struct net *net, __be32 saddr,
__be16 sport, __be32 daddr, __be16 dport,
int dif, struct udp_table *udptable)
{
udp: RCU handling for Unicast packets. Goals are : 1) Optimizing handling of incoming Unicast UDP frames, so that no memory writes should happen in the fast path. Note: Multicasts and broadcasts still will need to take a lock, because doing a full lockless lookup in this case is difficult. 2) No expensive operations in the socket bind/unhash phases : - No expensive synchronize_rcu() calls. - No added rcu_head in socket structure, increasing memory needs, but more important, forcing us to use call_rcu() calls, that have the bad property of making sockets structure cold. (rcu grace period between socket freeing and its potential reuse make this socket being cold in CPU cache). David did a previous patch using call_rcu() and noticed a 20% impact on TCP connection rates. Quoting Cristopher Lameter : "Right. That results in cacheline cooldown. You'd want to recycle the object as they are cache hot on a per cpu basis. That is screwed up by the delayed regular rcu processing. We have seen multiple regressions due to cacheline cooldown. The only choice in cacheline hot sensitive areas is to deal with the complexity that comes with SLAB_DESTROY_BY_RCU or give up on RCU." - Because udp sockets are allocated from dedicated kmem_cache, use of SLAB_DESTROY_BY_RCU can help here. Theory of operation : --------------------- As the lookup is lockfree (using rcu_read_lock()/rcu_read_unlock()), special attention must be taken by readers and writers. Use of SLAB_DESTROY_BY_RCU is tricky too, because a socket can be freed, reused, inserted in a different chain or in worst case in the same chain while readers could do lookups in the same time. In order to avoid loops, a reader must check each socket found in a chain really belongs to the chain the reader was traversing. If it finds a mismatch, lookup must start again at the begining. This *restart* loop is the reason we had to use rdlock for the multicast case, because we dont want to send same message several times to the same socket. We use RCU only for fast path. Thus, /proc/net/udp still takes spinlocks. Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-29 09:11:14 +00:00
struct sock *sk, *result;
struct hlist_nulls_node *node;
unsigned short hnum = ntohs(dport);
unsigned int hash2, slot2, slot = udp_hashfn(net, hnum, udptable->mask);
struct udp_hslot *hslot2, *hslot = &udptable->hash[slot];
udp: RCU handling for Unicast packets. Goals are : 1) Optimizing handling of incoming Unicast UDP frames, so that no memory writes should happen in the fast path. Note: Multicasts and broadcasts still will need to take a lock, because doing a full lockless lookup in this case is difficult. 2) No expensive operations in the socket bind/unhash phases : - No expensive synchronize_rcu() calls. - No added rcu_head in socket structure, increasing memory needs, but more important, forcing us to use call_rcu() calls, that have the bad property of making sockets structure cold. (rcu grace period between socket freeing and its potential reuse make this socket being cold in CPU cache). David did a previous patch using call_rcu() and noticed a 20% impact on TCP connection rates. Quoting Cristopher Lameter : "Right. That results in cacheline cooldown. You'd want to recycle the object as they are cache hot on a per cpu basis. That is screwed up by the delayed regular rcu processing. We have seen multiple regressions due to cacheline cooldown. The only choice in cacheline hot sensitive areas is to deal with the complexity that comes with SLAB_DESTROY_BY_RCU or give up on RCU." - Because udp sockets are allocated from dedicated kmem_cache, use of SLAB_DESTROY_BY_RCU can help here. Theory of operation : --------------------- As the lookup is lockfree (using rcu_read_lock()/rcu_read_unlock()), special attention must be taken by readers and writers. Use of SLAB_DESTROY_BY_RCU is tricky too, because a socket can be freed, reused, inserted in a different chain or in worst case in the same chain while readers could do lookups in the same time. In order to avoid loops, a reader must check each socket found in a chain really belongs to the chain the reader was traversing. If it finds a mismatch, lookup must start again at the begining. This *restart* loop is the reason we had to use rdlock for the multicast case, because we dont want to send same message several times to the same socket. We use RCU only for fast path. Thus, /proc/net/udp still takes spinlocks. Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-29 09:11:14 +00:00
int score, badness;
udp: RCU handling for Unicast packets. Goals are : 1) Optimizing handling of incoming Unicast UDP frames, so that no memory writes should happen in the fast path. Note: Multicasts and broadcasts still will need to take a lock, because doing a full lockless lookup in this case is difficult. 2) No expensive operations in the socket bind/unhash phases : - No expensive synchronize_rcu() calls. - No added rcu_head in socket structure, increasing memory needs, but more important, forcing us to use call_rcu() calls, that have the bad property of making sockets structure cold. (rcu grace period between socket freeing and its potential reuse make this socket being cold in CPU cache). David did a previous patch using call_rcu() and noticed a 20% impact on TCP connection rates. Quoting Cristopher Lameter : "Right. That results in cacheline cooldown. You'd want to recycle the object as they are cache hot on a per cpu basis. That is screwed up by the delayed regular rcu processing. We have seen multiple regressions due to cacheline cooldown. The only choice in cacheline hot sensitive areas is to deal with the complexity that comes with SLAB_DESTROY_BY_RCU or give up on RCU." - Because udp sockets are allocated from dedicated kmem_cache, use of SLAB_DESTROY_BY_RCU can help here. Theory of operation : --------------------- As the lookup is lockfree (using rcu_read_lock()/rcu_read_unlock()), special attention must be taken by readers and writers. Use of SLAB_DESTROY_BY_RCU is tricky too, because a socket can be freed, reused, inserted in a different chain or in worst case in the same chain while readers could do lookups in the same time. In order to avoid loops, a reader must check each socket found in a chain really belongs to the chain the reader was traversing. If it finds a mismatch, lookup must start again at the begining. This *restart* loop is the reason we had to use rdlock for the multicast case, because we dont want to send same message several times to the same socket. We use RCU only for fast path. Thus, /proc/net/udp still takes spinlocks. Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-29 09:11:14 +00:00
rcu_read_lock();
if (hslot->count > 10) {
hash2 = udp4_portaddr_hash(net, daddr, hnum);
slot2 = hash2 & udptable->mask;
hslot2 = &udptable->hash2[slot2];
if (hslot->count < hslot2->count)
goto begin;
result = udp4_lib_lookup2(net, saddr, sport,
daddr, hnum, dif,
hslot2, slot2);
if (!result) {
hash2 = udp4_portaddr_hash(net, htonl(INADDR_ANY), hnum);
slot2 = hash2 & udptable->mask;
hslot2 = &udptable->hash2[slot2];
if (hslot->count < hslot2->count)
goto begin;
result = udp4_lib_lookup2(net, saddr, sport,
htonl(INADDR_ANY), hnum, dif,
hslot2, slot2);
}
rcu_read_unlock();
return result;
}
udp: RCU handling for Unicast packets. Goals are : 1) Optimizing handling of incoming Unicast UDP frames, so that no memory writes should happen in the fast path. Note: Multicasts and broadcasts still will need to take a lock, because doing a full lockless lookup in this case is difficult. 2) No expensive operations in the socket bind/unhash phases : - No expensive synchronize_rcu() calls. - No added rcu_head in socket structure, increasing memory needs, but more important, forcing us to use call_rcu() calls, that have the bad property of making sockets structure cold. (rcu grace period between socket freeing and its potential reuse make this socket being cold in CPU cache). David did a previous patch using call_rcu() and noticed a 20% impact on TCP connection rates. Quoting Cristopher Lameter : "Right. That results in cacheline cooldown. You'd want to recycle the object as they are cache hot on a per cpu basis. That is screwed up by the delayed regular rcu processing. We have seen multiple regressions due to cacheline cooldown. The only choice in cacheline hot sensitive areas is to deal with the complexity that comes with SLAB_DESTROY_BY_RCU or give up on RCU." - Because udp sockets are allocated from dedicated kmem_cache, use of SLAB_DESTROY_BY_RCU can help here. Theory of operation : --------------------- As the lookup is lockfree (using rcu_read_lock()/rcu_read_unlock()), special attention must be taken by readers and writers. Use of SLAB_DESTROY_BY_RCU is tricky too, because a socket can be freed, reused, inserted in a different chain or in worst case in the same chain while readers could do lookups in the same time. In order to avoid loops, a reader must check each socket found in a chain really belongs to the chain the reader was traversing. If it finds a mismatch, lookup must start again at the begining. This *restart* loop is the reason we had to use rdlock for the multicast case, because we dont want to send same message several times to the same socket. We use RCU only for fast path. Thus, /proc/net/udp still takes spinlocks. Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-29 09:11:14 +00:00
begin:
result = NULL;
badness = -1;
sk_nulls_for_each_rcu(sk, node, &hslot->head) {
score = compute_score(sk, net, saddr, hnum, sport,
daddr, dport, dif);
if (score > badness) {
result = sk;
badness = score;
}
}
/*
* if the nulls value we got at the end of this lookup is
* not the expected one, we must restart lookup.
* We probably met an item that was moved to another chain.
*/
if (get_nulls_value(node) != slot)
goto begin;
udp: RCU handling for Unicast packets. Goals are : 1) Optimizing handling of incoming Unicast UDP frames, so that no memory writes should happen in the fast path. Note: Multicasts and broadcasts still will need to take a lock, because doing a full lockless lookup in this case is difficult. 2) No expensive operations in the socket bind/unhash phases : - No expensive synchronize_rcu() calls. - No added rcu_head in socket structure, increasing memory needs, but more important, forcing us to use call_rcu() calls, that have the bad property of making sockets structure cold. (rcu grace period between socket freeing and its potential reuse make this socket being cold in CPU cache). David did a previous patch using call_rcu() and noticed a 20% impact on TCP connection rates. Quoting Cristopher Lameter : "Right. That results in cacheline cooldown. You'd want to recycle the object as they are cache hot on a per cpu basis. That is screwed up by the delayed regular rcu processing. We have seen multiple regressions due to cacheline cooldown. The only choice in cacheline hot sensitive areas is to deal with the complexity that comes with SLAB_DESTROY_BY_RCU or give up on RCU." - Because udp sockets are allocated from dedicated kmem_cache, use of SLAB_DESTROY_BY_RCU can help here. Theory of operation : --------------------- As the lookup is lockfree (using rcu_read_lock()/rcu_read_unlock()), special attention must be taken by readers and writers. Use of SLAB_DESTROY_BY_RCU is tricky too, because a socket can be freed, reused, inserted in a different chain or in worst case in the same chain while readers could do lookups in the same time. In order to avoid loops, a reader must check each socket found in a chain really belongs to the chain the reader was traversing. If it finds a mismatch, lookup must start again at the begining. This *restart* loop is the reason we had to use rdlock for the multicast case, because we dont want to send same message several times to the same socket. We use RCU only for fast path. Thus, /proc/net/udp still takes spinlocks. Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-29 09:11:14 +00:00
if (result) {
if (unlikely(!atomic_inc_not_zero_hint(&result->sk_refcnt, 2)))
udp: RCU handling for Unicast packets. Goals are : 1) Optimizing handling of incoming Unicast UDP frames, so that no memory writes should happen in the fast path. Note: Multicasts and broadcasts still will need to take a lock, because doing a full lockless lookup in this case is difficult. 2) No expensive operations in the socket bind/unhash phases : - No expensive synchronize_rcu() calls. - No added rcu_head in socket structure, increasing memory needs, but more important, forcing us to use call_rcu() calls, that have the bad property of making sockets structure cold. (rcu grace period between socket freeing and its potential reuse make this socket being cold in CPU cache). David did a previous patch using call_rcu() and noticed a 20% impact on TCP connection rates. Quoting Cristopher Lameter : "Right. That results in cacheline cooldown. You'd want to recycle the object as they are cache hot on a per cpu basis. That is screwed up by the delayed regular rcu processing. We have seen multiple regressions due to cacheline cooldown. The only choice in cacheline hot sensitive areas is to deal with the complexity that comes with SLAB_DESTROY_BY_RCU or give up on RCU." - Because udp sockets are allocated from dedicated kmem_cache, use of SLAB_DESTROY_BY_RCU can help here. Theory of operation : --------------------- As the lookup is lockfree (using rcu_read_lock()/rcu_read_unlock()), special attention must be taken by readers and writers. Use of SLAB_DESTROY_BY_RCU is tricky too, because a socket can be freed, reused, inserted in a different chain or in worst case in the same chain while readers could do lookups in the same time. In order to avoid loops, a reader must check each socket found in a chain really belongs to the chain the reader was traversing. If it finds a mismatch, lookup must start again at the begining. This *restart* loop is the reason we had to use rdlock for the multicast case, because we dont want to send same message several times to the same socket. We use RCU only for fast path. Thus, /proc/net/udp still takes spinlocks. Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-29 09:11:14 +00:00
result = NULL;
else if (unlikely(compute_score(result, net, saddr, hnum, sport,
daddr, dport, dif) < badness)) {
sock_put(result);
goto begin;
}
}
rcu_read_unlock();
return result;
}
EXPORT_SYMBOL_GPL(__udp4_lib_lookup);
static inline struct sock *__udp4_lib_lookup_skb(struct sk_buff *skb,
__be16 sport, __be16 dport,
struct udp_table *udptable)
{
struct sock *sk;
const struct iphdr *iph = ip_hdr(skb);
if (unlikely(sk = skb_steal_sock(skb)))
return sk;
else
return __udp4_lib_lookup(dev_net(skb_dst(skb)->dev), iph->saddr, sport,
iph->daddr, dport, inet_iif(skb),
udptable);
}
struct sock *udp4_lib_lookup(struct net *net, __be32 saddr, __be16 sport,
__be32 daddr, __be16 dport, int dif)
{
return __udp4_lib_lookup(net, saddr, sport, daddr, dport, dif, &udp_table);
}
EXPORT_SYMBOL_GPL(udp4_lib_lookup);
static inline struct sock *udp_v4_mcast_next(struct net *net, struct sock *sk,
__be16 loc_port, __be32 loc_addr,
__be16 rmt_port, __be32 rmt_addr,
int dif)
{
struct hlist_nulls_node *node;
struct sock *s = sk;
unsigned short hnum = ntohs(loc_port);
sk_nulls_for_each_from(s, node) {
struct inet_sock *inet = inet_sk(s);
if (!net_eq(sock_net(s), net) ||
udp_sk(s)->udp_port_hash != hnum ||
(inet->inet_daddr && inet->inet_daddr != rmt_addr) ||
(inet->inet_dport != rmt_port && inet->inet_dport) ||
(inet->inet_rcv_saddr &&
inet->inet_rcv_saddr != loc_addr) ||
ipv6_only_sock(s) ||
(s->sk_bound_dev_if && s->sk_bound_dev_if != dif))
continue;
if (!ip_mc_sf_allow(s, loc_addr, rmt_addr, dif))
continue;
goto found;
}
s = NULL;
found:
return s;
}
/*
* This routine is called by the ICMP module when it gets some
* sort of error condition. If err < 0 then the socket should
* be closed and the error returned to the user. If err > 0
* it's just the icmp type << 8 | icmp code.
* Header points to the ip header of the error packet. We move
* on past this. Then (as it used to claim before adjustment)
* header points to the first 8 bytes of the udp header. We need
* to find the appropriate port.
*/
void __udp4_lib_err(struct sk_buff *skb, u32 info, struct udp_table *udptable)
{
struct inet_sock *inet;
const struct iphdr *iph = (const struct iphdr *)skb->data;
struct udphdr *uh = (struct udphdr *)(skb->data+(iph->ihl<<2));
const int type = icmp_hdr(skb)->type;
const int code = icmp_hdr(skb)->code;
struct sock *sk;
int harderr;
int err;
struct net *net = dev_net(skb->dev);
sk = __udp4_lib_lookup(net, iph->daddr, uh->dest,
iph->saddr, uh->source, skb->dev->ifindex, udptable);
if (sk == NULL) {
ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
return; /* No socket for error */
}
err = 0;
harderr = 0;
inet = inet_sk(sk);
switch (type) {
default:
case ICMP_TIME_EXCEEDED:
err = EHOSTUNREACH;
break;
case ICMP_SOURCE_QUENCH:
goto out;
case ICMP_PARAMETERPROB:
err = EPROTO;
harderr = 1;
break;
case ICMP_DEST_UNREACH:
if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
ipv4_sk_update_pmtu(skb, sk, info);
if (inet->pmtudisc != IP_PMTUDISC_DONT) {
err = EMSGSIZE;
harderr = 1;
break;
}
goto out;
}
err = EHOSTUNREACH;
if (code <= NR_ICMP_UNREACH) {
harderr = icmp_err_convert[code].fatal;
err = icmp_err_convert[code].errno;
}
break;
case ICMP_REDIRECT:
ipv4_sk_redirect(skb, sk);
break;
}
/*
* RFC1122: OK. Passes ICMP errors back to application, as per
* 4.1.3.3.
*/
if (!inet->recverr) {
if (!harderr || sk->sk_state != TCP_ESTABLISHED)
goto out;
} else
ip_icmp_error(sk, skb, err, uh->dest, info, (u8 *)(uh+1));
sk->sk_err = err;
sk->sk_error_report(sk);
out:
sock_put(sk);
}
void udp_err(struct sk_buff *skb, u32 info)
{
__udp4_lib_err(skb, info, &udp_table);
}
/*
* Throw away all pending data and cancel the corking. Socket is locked.
*/
void udp_flush_pending_frames(struct sock *sk)
{
struct udp_sock *up = udp_sk(sk);
if (up->pending) {
up->len = 0;
up->pending = 0;
ip_flush_pending_frames(sk);
}
}
EXPORT_SYMBOL(udp_flush_pending_frames);
/**
* udp4_hwcsum - handle outgoing HW checksumming
* @skb: sk_buff containing the filled-in UDP header
* (checksum field must be zeroed out)
* @src: source IP address
* @dst: destination IP address
*/
static void udp4_hwcsum(struct sk_buff *skb, __be32 src, __be32 dst)
{
struct udphdr *uh = udp_hdr(skb);
struct sk_buff *frags = skb_shinfo(skb)->frag_list;
int offset = skb_transport_offset(skb);
int len = skb->len - offset;
int hlen = len;
__wsum csum = 0;
if (!frags) {
/*
* Only one fragment on the socket.
*/
skb->csum_start = skb_transport_header(skb) - skb->head;
skb->csum_offset = offsetof(struct udphdr, check);
uh->check = ~csum_tcpudp_magic(src, dst, len,
IPPROTO_UDP, 0);
} else {
/*
* HW-checksum won't work as there are two or more
* fragments on the socket so that all csums of sk_buffs
* should be together
*/
do {
csum = csum_add(csum, frags->csum);
hlen -= frags->len;
} while ((frags = frags->next));
csum = skb_checksum(skb, offset, hlen, csum);
skb->ip_summed = CHECKSUM_NONE;
uh->check = csum_tcpudp_magic(src, dst, len, IPPROTO_UDP, csum);
if (uh->check == 0)
uh->check = CSUM_MANGLED_0;
}
}
static int udp_send_skb(struct sk_buff *skb, struct flowi4 *fl4)
{
struct sock *sk = skb->sk;
struct inet_sock *inet = inet_sk(sk);
struct udphdr *uh;
int err = 0;
int is_udplite = IS_UDPLITE(sk);
int offset = skb_transport_offset(skb);
int len = skb->len - offset;
__wsum csum = 0;
/*
* Create a UDP header
*/
uh = udp_hdr(skb);
uh->source = inet->inet_sport;
uh->dest = fl4->fl4_dport;
uh->len = htons(len);
uh->check = 0;
if (is_udplite) /* UDP-Lite */
csum = udplite_csum(skb);
else if (sk->sk_no_check == UDP_CSUM_NOXMIT) { /* UDP csum disabled */
skb->ip_summed = CHECKSUM_NONE;
goto send;
} else if (skb->ip_summed == CHECKSUM_PARTIAL) { /* UDP hardware csum */
udp4_hwcsum(skb, fl4->saddr, fl4->daddr);
goto send;
} else
csum = udp_csum(skb);
/* add protocol-dependent pseudo-header */
uh->check = csum_tcpudp_magic(fl4->saddr, fl4->daddr, len,
sk->sk_protocol, csum);
if (uh->check == 0)
uh->check = CSUM_MANGLED_0;
send:
err = ip_send_skb(sock_net(sk), skb);
if (err) {
if (err == -ENOBUFS && !inet->recverr) {
UDP_INC_STATS_USER(sock_net(sk),
UDP_MIB_SNDBUFERRORS, is_udplite);
err = 0;
}
} else
UDP_INC_STATS_USER(sock_net(sk),
UDP_MIB_OUTDATAGRAMS, is_udplite);
return err;
}
/*
* Push out all pending data as one UDP datagram. Socket is locked.
*/
static int udp_push_pending_frames(struct sock *sk)
{
struct udp_sock *up = udp_sk(sk);
struct inet_sock *inet = inet_sk(sk);
struct flowi4 *fl4 = &inet->cork.fl.u.ip4;
struct sk_buff *skb;
int err = 0;
skb = ip_finish_skb(sk, fl4);
if (!skb)
goto out;
err = udp_send_skb(skb, fl4);
out:
up->len = 0;
up->pending = 0;
return err;
}
int udp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
size_t len)
{
struct inet_sock *inet = inet_sk(sk);
struct udp_sock *up = udp_sk(sk);
struct flowi4 fl4_stack;
struct flowi4 *fl4;
int ulen = len;
struct ipcm_cookie ipc;
struct rtable *rt = NULL;
int free = 0;
int connected = 0;
__be32 daddr, faddr, saddr;
__be16 dport;
u8 tos;
int err, is_udplite = IS_UDPLITE(sk);
int corkreq = up->corkflag || msg->msg_flags&MSG_MORE;
int (*getfrag)(void *, char *, int, int, int, struct sk_buff *);
struct sk_buff *skb;
struct ip_options_data opt_copy;
if (len > 0xFFFF)
return -EMSGSIZE;
/*
* Check the flags.
*/
if (msg->msg_flags & MSG_OOB) /* Mirror BSD error message compatibility */
return -EOPNOTSUPP;
ipc.opt = NULL;
ipc.tx_flags = 0;
getfrag = is_udplite ? udplite_getfrag : ip_generic_getfrag;
fl4 = &inet->cork.fl.u.ip4;
if (up->pending) {
/*
* There are pending frames.
* The socket lock must be held while it's corked.
*/
lock_sock(sk);
if (likely(up->pending)) {
if (unlikely(up->pending != AF_INET)) {
release_sock(sk);
return -EINVAL;
}
goto do_append_data;
}
release_sock(sk);
}
ulen += sizeof(struct udphdr);
/*
* Get and verify the address.
*/
if (msg->msg_name) {
struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
if (msg->msg_namelen < sizeof(*usin))
return -EINVAL;
if (usin->sin_family != AF_INET) {
if (usin->sin_family != AF_UNSPEC)
return -EAFNOSUPPORT;
}
daddr = usin->sin_addr.s_addr;
dport = usin->sin_port;
if (dport == 0)
return -EINVAL;
} else {
if (sk->sk_state != TCP_ESTABLISHED)
return -EDESTADDRREQ;
daddr = inet->inet_daddr;
dport = inet->inet_dport;
/* Open fast path for connected socket.
Route will not be used, if at least one option is set.
*/
connected = 1;
}
ipc.addr = inet->inet_saddr;
ipc.oif = sk->sk_bound_dev_if;
err = sock_tx_timestamp(sk, &ipc.tx_flags);
if (err)
return err;
if (msg->msg_controllen) {
err = ip_cmsg_send(sock_net(sk), msg, &ipc);
if (err)
return err;
if (ipc.opt)
free = 1;
connected = 0;
}
if (!ipc.opt) {
struct ip_options_rcu *inet_opt;
rcu_read_lock();
inet_opt = rcu_dereference(inet->inet_opt);
if (inet_opt) {
memcpy(&opt_copy, inet_opt,
sizeof(*inet_opt) + inet_opt->opt.optlen);
ipc.opt = &opt_copy.opt;
}
rcu_read_unlock();
}
saddr = ipc.addr;
ipc.addr = faddr = daddr;
if (ipc.opt && ipc.opt->opt.srr) {
if (!daddr)
return -EINVAL;
faddr = ipc.opt->opt.faddr;
connected = 0;
}
tos = RT_TOS(inet->tos);
if (sock_flag(sk, SOCK_LOCALROUTE) ||
(msg->msg_flags & MSG_DONTROUTE) ||
(ipc.opt && ipc.opt->opt.is_strictroute)) {
tos |= RTO_ONLINK;
connected = 0;
}
if (ipv4_is_multicast(daddr)) {
if (!ipc.oif)
ipc.oif = inet->mc_index;
if (!saddr)
saddr = inet->mc_addr;
connected = 0;
ipv4: Implement IP_UNICAST_IF socket option. The IP_UNICAST_IF feature is needed by the Wine project. This patch implements the feature by setting the outgoing interface in a similar fashion to that of IP_MULTICAST_IF. A separate option is needed to handle this feature since the existing options do not provide all of the characteristics required by IP_UNICAST_IF, a summary is provided below. SO_BINDTODEVICE: * SO_BINDTODEVICE requires administrative privileges, IP_UNICAST_IF does not. From reading some old mailing list articles my understanding is that SO_BINDTODEVICE requires administrative privileges because it can override the administrator's routing settings. * The SO_BINDTODEVICE option restricts both outbound and inbound traffic, IP_UNICAST_IF only impacts outbound traffic. IP_PKTINFO: * Since IP_PKTINFO and IP_UNICAST_IF are independent options, implementing IP_UNICAST_IF with IP_PKTINFO will likely break some applications. * Implementing IP_UNICAST_IF on top of IP_PKTINFO significantly complicates the Wine codebase and reduces the socket performance (doing this requires a lot of extra communication between the "server" and "user" layers). bind(): * bind() does not work on broadcast packets, IP_UNICAST_IF is specifically intended to work with broadcast packets. * Like SO_BINDTODEVICE, bind() restricts both outbound and inbound traffic. Signed-off-by: Erich E. Hoover <ehoover@mines.edu> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-02-08 09:11:07 +00:00
} else if (!ipc.oif)
ipc.oif = inet->uc_index;
if (connected)
rt = (struct rtable *)sk_dst_check(sk, 0);
if (rt == NULL) {
struct net *net = sock_net(sk);
fl4 = &fl4_stack;
flowi4_init_output(fl4, ipc.oif, sk->sk_mark, tos,
RT_SCOPE_UNIVERSE, sk->sk_protocol,
inet_sk_flowi_flags(sk)|FLOWI_FLAG_CAN_SLEEP,
faddr, saddr, dport, inet->inet_sport);
security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
rt = ip_route_output_flow(net, fl4, sk);
if (IS_ERR(rt)) {
err = PTR_ERR(rt);
rt = NULL;
if (err == -ENETUNREACH)
IP_INC_STATS_BH(net, IPSTATS_MIB_OUTNOROUTES);
goto out;
}
err = -EACCES;
if ((rt->rt_flags & RTCF_BROADCAST) &&
!sock_flag(sk, SOCK_BROADCAST))
goto out;
if (connected)
sk_dst_set(sk, dst_clone(&rt->dst));
}
if (msg->msg_flags&MSG_CONFIRM)
goto do_confirm;
back_from_confirm:
saddr = fl4->saddr;
if (!ipc.addr)
daddr = ipc.addr = fl4->daddr;
/* Lockless fast path for the non-corking case. */
if (!corkreq) {
skb = ip_make_skb(sk, fl4, getfrag, msg->msg_iov, ulen,
sizeof(struct udphdr), &ipc, &rt,
msg->msg_flags);
err = PTR_ERR(skb);
if (skb && !IS_ERR(skb))
err = udp_send_skb(skb, fl4);
goto out;
}
lock_sock(sk);
if (unlikely(up->pending)) {
/* The socket is already corked while preparing it. */
/* ... which is an evident application bug. --ANK */
release_sock(sk);
LIMIT_NETDEBUG(KERN_DEBUG pr_fmt("cork app bug 2\n"));
err = -EINVAL;
goto out;
}
/*
* Now cork the socket to pend data.
*/
fl4 = &inet->cork.fl.u.ip4;
fl4->daddr = daddr;
fl4->saddr = saddr;
fl4->fl4_dport = dport;
fl4->fl4_sport = inet->inet_sport;
up->pending = AF_INET;
do_append_data:
up->len += ulen;
err = ip_append_data(sk, fl4, getfrag, msg->msg_iov, ulen,
sizeof(struct udphdr), &ipc, &rt,
corkreq ? msg->msg_flags|MSG_MORE : msg->msg_flags);
if (err)
udp_flush_pending_frames(sk);
else if (!corkreq)
err = udp_push_pending_frames(sk);
else if (unlikely(skb_queue_empty(&sk->sk_write_queue)))
up->pending = 0;
release_sock(sk);
out:
ip_rt_put(rt);
if (free)
kfree(ipc.opt);
if (!err)
return len;
/*
* ENOBUFS = no kernel mem, SOCK_NOSPACE = no sndbuf space. Reporting
* ENOBUFS might not be good (it's not tunable per se), but otherwise
* we don't have a good statistic (IpOutDiscards but it can be too many
* things). We could add another new stat but at least for now that
* seems like overkill.
*/
if (err == -ENOBUFS || test_bit(SOCK_NOSPACE, &sk->sk_socket->flags)) {
UDP_INC_STATS_USER(sock_net(sk),
UDP_MIB_SNDBUFERRORS, is_udplite);
}
return err;
do_confirm:
dst_confirm(&rt->dst);
if (!(msg->msg_flags&MSG_PROBE) || len)
goto back_from_confirm;
err = 0;
goto out;
}
EXPORT_SYMBOL(udp_sendmsg);
int udp_sendpage(struct sock *sk, struct page *page, int offset,
size_t size, int flags)
{
struct inet_sock *inet = inet_sk(sk);
struct udp_sock *up = udp_sk(sk);
int ret;
if (!up->pending) {
struct msghdr msg = { .msg_flags = flags|MSG_MORE };
/* Call udp_sendmsg to specify destination address which
* sendpage interface can't pass.
* This will succeed only when the socket is connected.
*/
ret = udp_sendmsg(NULL, sk, &msg, 0);
if (ret < 0)
return ret;
}
lock_sock(sk);
if (unlikely(!up->pending)) {
release_sock(sk);
LIMIT_NETDEBUG(KERN_DEBUG pr_fmt("udp cork app bug 3\n"));
return -EINVAL;
}
ret = ip_append_page(sk, &inet->cork.fl.u.ip4,
page, offset, size, flags);
if (ret == -EOPNOTSUPP) {
release_sock(sk);
return sock_no_sendpage(sk->sk_socket, page, offset,
size, flags);
}
if (ret < 0) {
udp_flush_pending_frames(sk);
goto out;
}
up->len += size;
if (!(up->corkflag || (flags&MSG_MORE)))
ret = udp_push_pending_frames(sk);
if (!ret)
ret = size;
out:
release_sock(sk);
return ret;
}
/**
* first_packet_length - return length of first packet in receive queue
* @sk: socket
*
* Drops all bad checksum frames, until a valid one is found.
* Returns the length of found skb, or 0 if none is found.
*/
static unsigned int first_packet_length(struct sock *sk)
{
struct sk_buff_head list_kill, *rcvq = &sk->sk_receive_queue;
struct sk_buff *skb;
unsigned int res;
__skb_queue_head_init(&list_kill);
spin_lock_bh(&rcvq->lock);
while ((skb = skb_peek(rcvq)) != NULL &&
udp_lib_checksum_complete(skb)) {
UDP_INC_STATS_BH(sock_net(sk), UDP_MIB_INERRORS,
IS_UDPLITE(sk));
atomic_inc(&sk->sk_drops);
__skb_unlink(skb, rcvq);
__skb_queue_tail(&list_kill, skb);
}
res = skb ? skb->len : 0;
spin_unlock_bh(&rcvq->lock);
if (!skb_queue_empty(&list_kill)) {
bool slow = lock_sock_fast(sk);
__skb_queue_purge(&list_kill);
sk_mem_reclaim_partial(sk);
unlock_sock_fast(sk, slow);
}
return res;
}
/*
* IOCTL requests applicable to the UDP protocol
*/
int udp_ioctl(struct sock *sk, int cmd, unsigned long arg)
{
switch (cmd) {
case SIOCOUTQ:
{
int amount = sk_wmem_alloc_get(sk);
return put_user(amount, (int __user *)arg);
}
case SIOCINQ:
{
unsigned int amount = first_packet_length(sk);
if (amount)
/*
* We will only return the amount
* of this packet since that is all
* that will be read.
*/
amount -= sizeof(struct udphdr);
return put_user(amount, (int __user *)arg);
}
default:
return -ENOIOCTLCMD;
}
return 0;
}
EXPORT_SYMBOL(udp_ioctl);
/*
* This should be easy, if there is something there we
* return it, otherwise we block.
*/
int udp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
size_t len, int noblock, int flags, int *addr_len)
{
struct inet_sock *inet = inet_sk(sk);
struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
struct sk_buff *skb;
unsigned int ulen, copied;
int peeked, off = 0;
int err;
int is_udplite = IS_UDPLITE(sk);
bool slow;
/*
* Check any passed addresses
*/
if (addr_len)
*addr_len = sizeof(*sin);
if (flags & MSG_ERRQUEUE)
return ip_recv_error(sk, msg, len);
try_again:
skb = __skb_recv_datagram(sk, flags | (noblock ? MSG_DONTWAIT : 0),
&peeked, &off, &err);
if (!skb)
goto out;
ulen = skb->len - sizeof(struct udphdr);
copied = len;
if (copied > ulen)
copied = ulen;
else if (copied < ulen)
msg->msg_flags |= MSG_TRUNC;
/*
* If checksum is needed at all, try to do it while copying the
* data. If the data is truncated, or if we only want a partial
* coverage checksum (UDP-Lite), do it before the copy.
*/
if (copied < ulen || UDP_SKB_CB(skb)->partial_cov) {
if (udp_lib_checksum_complete(skb))
goto csum_copy_err;
}
if (skb_csum_unnecessary(skb))
err = skb_copy_datagram_iovec(skb, sizeof(struct udphdr),
msg->msg_iov, copied);
else {
err = skb_copy_and_csum_datagram_iovec(skb,
sizeof(struct udphdr),
msg->msg_iov);
if (err == -EINVAL)
goto csum_copy_err;
}
if (unlikely(err)) {
trace_kfree_skb(skb, udp_recvmsg);
if (!peeked) {
atomic_inc(&sk->sk_drops);
UDP_INC_STATS_USER(sock_net(sk),
UDP_MIB_INERRORS, is_udplite);
}
goto out_free;
}
if (!peeked)
UDP_INC_STATS_USER(sock_net(sk),
UDP_MIB_INDATAGRAMS, is_udplite);
net: Generalize socket rx gap / receive queue overflow cmsg Create a new socket level option to report number of queue overflows Recently I augmented the AF_PACKET protocol to report the number of frames lost on the socket receive queue between any two enqueued frames. This value was exported via a SOL_PACKET level cmsg. AFter I completed that work it was requested that this feature be generalized so that any datagram oriented socket could make use of this option. As such I've created this patch, It creates a new SOL_SOCKET level option called SO_RXQ_OVFL, which when enabled exports a SOL_SOCKET level cmsg that reports the nubmer of times the sk_receive_queue overflowed between any two given frames. It also augments the AF_PACKET protocol to take advantage of this new feature (as it previously did not touch sk->sk_drops, which this patch uses to record the overflow count). Tested successfully by me. Notes: 1) Unlike my previous patch, this patch simply records the sk_drops value, which is not a number of drops between packets, but rather a total number of drops. Deltas must be computed in user space. 2) While this patch currently works with datagram oriented protocols, it will also be accepted by non-datagram oriented protocols. I'm not sure if thats agreeable to everyone, but my argument in favor of doing so is that, for those protocols which aren't applicable to this option, sk_drops will always be zero, and reporting no drops on a receive queue that isn't used for those non-participating protocols seems reasonable to me. This also saves us having to code in a per-protocol opt in mechanism. 3) This applies cleanly to net-next assuming that commit 977750076d98c7ff6cbda51858bb5a5894a9d9ab (my af packet cmsg patch) is reverted Signed-off-by: Neil Horman <nhorman@tuxdriver.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-10-12 20:26:31 +00:00
sock_recv_ts_and_drops(msg, sk, skb);
/* Copy the address. */
if (sin) {
sin->sin_family = AF_INET;
sin->sin_port = udp_hdr(skb)->source;
sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
}
if (inet->cmsg_flags)
ip_cmsg_recv(msg, skb);
err = copied;
if (flags & MSG_TRUNC)
err = ulen;
out_free:
skb_free_datagram_locked(sk, skb);
out:
return err;
csum_copy_err:
slow = lock_sock_fast(sk);
if (!skb_kill_datagram(sk, skb, flags))
UDP_INC_STATS_USER(sock_net(sk), UDP_MIB_INERRORS, is_udplite);
unlock_sock_fast(sk, slow);
if (noblock)
return -EAGAIN;
/* starting over for a new packet */
msg->msg_flags &= ~MSG_TRUNC;
goto try_again;
}
int udp_disconnect(struct sock *sk, int flags)
{
struct inet_sock *inet = inet_sk(sk);
/*
* 1003.1g - break association.
*/
sk->sk_state = TCP_CLOSE;
inet->inet_daddr = 0;
inet->inet_dport = 0;
sock_rps_reset_rxhash(sk);
sk->sk_bound_dev_if = 0;
if (!(sk->sk_userlocks & SOCK_BINDADDR_LOCK))
inet_reset_saddr(sk);
if (!(sk->sk_userlocks & SOCK_BINDPORT_LOCK)) {
sk->sk_prot->unhash(sk);
inet->inet_sport = 0;
}
sk_dst_reset(sk);
return 0;
}
EXPORT_SYMBOL(udp_disconnect);
void udp_lib_unhash(struct sock *sk)
{
if (sk_hashed(sk)) {
struct udp_table *udptable = sk->sk_prot->h.udp_table;
struct udp_hslot *hslot, *hslot2;
hslot = udp_hashslot(udptable, sock_net(sk),
udp_sk(sk)->udp_port_hash);
hslot2 = udp_hashslot2(udptable, udp_sk(sk)->udp_portaddr_hash);
spin_lock_bh(&hslot->lock);
if (sk_nulls_del_node_init_rcu(sk)) {
hslot->count--;
inet_sk(sk)->inet_num = 0;
sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
spin_lock(&hslot2->lock);
hlist_nulls_del_init_rcu(&udp_sk(sk)->udp_portaddr_node);
hslot2->count--;
spin_unlock(&hslot2->lock);
}
spin_unlock_bh(&hslot->lock);
}
}
EXPORT_SYMBOL(udp_lib_unhash);
/*
* inet_rcv_saddr was changed, we must rehash secondary hash
*/
void udp_lib_rehash(struct sock *sk, u16 newhash)
{
if (sk_hashed(sk)) {
struct udp_table *udptable = sk->sk_prot->h.udp_table;
struct udp_hslot *hslot, *hslot2, *nhslot2;
hslot2 = udp_hashslot2(udptable, udp_sk(sk)->udp_portaddr_hash);
nhslot2 = udp_hashslot2(udptable, newhash);
udp_sk(sk)->udp_portaddr_hash = newhash;
if (hslot2 != nhslot2) {
hslot = udp_hashslot(udptable, sock_net(sk),
udp_sk(sk)->udp_port_hash);
/* we must lock primary chain too */
spin_lock_bh(&hslot->lock);
spin_lock(&hslot2->lock);
hlist_nulls_del_init_rcu(&udp_sk(sk)->udp_portaddr_node);
hslot2->count--;
spin_unlock(&hslot2->lock);
spin_lock(&nhslot2->lock);
hlist_nulls_add_head_rcu(&udp_sk(sk)->udp_portaddr_node,
&nhslot2->head);
nhslot2->count++;
spin_unlock(&nhslot2->lock);
spin_unlock_bh(&hslot->lock);
}
}
}
EXPORT_SYMBOL(udp_lib_rehash);
static void udp_v4_rehash(struct sock *sk)
{
u16 new_hash = udp4_portaddr_hash(sock_net(sk),
inet_sk(sk)->inet_rcv_saddr,
inet_sk(sk)->inet_num);
udp_lib_rehash(sk, new_hash);
}
static int __udp_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
{
rfs: Receive Flow Steering This patch implements receive flow steering (RFS). RFS steers received packets for layer 3 and 4 processing to the CPU where the application for the corresponding flow is running. RFS is an extension of Receive Packet Steering (RPS). The basic idea of RFS is that when an application calls recvmsg (or sendmsg) the application's running CPU is stored in a hash table that is indexed by the connection's rxhash which is stored in the socket structure. The rxhash is passed in skb's received on the connection from netif_receive_skb. For each received packet, the associated rxhash is used to look up the CPU in the hash table, if a valid CPU is set then the packet is steered to that CPU using the RPS mechanisms. The convolution of the simple approach is that it would potentially allow OOO packets. If threads are thrashing around CPUs or multiple threads are trying to read from the same sockets, a quickly changing CPU value in the hash table could cause rampant OOO packets-- we consider this a non-starter. To avoid OOO packets, this solution implements two types of hash tables: rps_sock_flow_table and rps_dev_flow_table. rps_sock_table is a global hash table. Each entry is just a CPU number and it is populated in recvmsg and sendmsg as described above. This table contains the "desired" CPUs for flows. rps_dev_flow_table is specific to each device queue. Each entry contains a CPU and a tail queue counter. The CPU is the "current" CPU for a matching flow. The tail queue counter holds the value of a tail queue counter for the associated CPU's backlog queue at the time of last enqueue for a flow matching the entry. Each backlog queue has a queue head counter which is incremented on dequeue, and so a queue tail counter is computed as queue head count + queue length. When a packet is enqueued on a backlog queue, the current value of the queue tail counter is saved in the hash entry of the rps_dev_flow_table. And now the trick: when selecting the CPU for RPS (get_rps_cpu) the rps_sock_flow table and the rps_dev_flow table for the RX queue are consulted. When the desired CPU for the flow (found in the rps_sock_flow table) does not match the current CPU (found in the rps_dev_flow table), the current CPU is changed to the desired CPU if one of the following is true: - The current CPU is unset (equal to RPS_NO_CPU) - Current CPU is offline - The current CPU's queue head counter >= queue tail counter in the rps_dev_flow table. This checks if the queue tail has advanced beyond the last packet that was enqueued using this table entry. This guarantees that all packets queued using this entry have been dequeued, thus preserving in order delivery. Making each queue have its own rps_dev_flow table has two advantages: 1) the tail queue counters will be written on each receive, so keeping the table local to interrupting CPU s good for locality. 2) this allows lockless access to the table-- the CPU number and queue tail counter need to be accessed together under mutual exclusion from netif_receive_skb, we assume that this is only called from device napi_poll which is non-reentrant. This patch implements RFS for TCP and connected UDP sockets. It should be usable for other flow oriented protocols. There are two configuration parameters for RFS. The "rps_flow_entries" kernel init parameter sets the number of entries in the rps_sock_flow_table, the per rxqueue sysfs entry "rps_flow_cnt" contains the number of entries in the rps_dev_flow table for the rxqueue. Both are rounded to power of two. The obvious benefit of RFS (over just RPS) is that it achieves CPU locality between the receive processing for a flow and the applications processing; this can result in increased performance (higher pps, lower latency). The benefits of RFS are dependent on cache hierarchy, application load, and other factors. On simple benchmarks, we don't necessarily see improvement and sometimes see degradation. However, for more complex benchmarks and for applications where cache pressure is much higher this technique seems to perform very well. Below are some benchmark results which show the potential benfit of this patch. The netperf test has 500 instances of netperf TCP_RR test with 1 byte req. and resp. The RPC test is an request/response test similar in structure to netperf RR test ith 100 threads on each host, but does more work in userspace that netperf. e1000e on 8 core Intel No RFS or RPS 104K tps at 30% CPU No RFS (best RPS config): 290K tps at 63% CPU RFS 303K tps at 61% CPU RPC test tps CPU% 50/90/99% usec latency Latency StdDev No RFS/RPS 103K 48% 757/900/3185 4472.35 RPS only: 174K 73% 415/993/2468 491.66 RFS 223K 73% 379/651/1382 315.61 Signed-off-by: Tom Herbert <therbert@google.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-16 23:01:27 +00:00
int rc;
rfs: Receive Flow Steering This patch implements receive flow steering (RFS). RFS steers received packets for layer 3 and 4 processing to the CPU where the application for the corresponding flow is running. RFS is an extension of Receive Packet Steering (RPS). The basic idea of RFS is that when an application calls recvmsg (or sendmsg) the application's running CPU is stored in a hash table that is indexed by the connection's rxhash which is stored in the socket structure. The rxhash is passed in skb's received on the connection from netif_receive_skb. For each received packet, the associated rxhash is used to look up the CPU in the hash table, if a valid CPU is set then the packet is steered to that CPU using the RPS mechanisms. The convolution of the simple approach is that it would potentially allow OOO packets. If threads are thrashing around CPUs or multiple threads are trying to read from the same sockets, a quickly changing CPU value in the hash table could cause rampant OOO packets-- we consider this a non-starter. To avoid OOO packets, this solution implements two types of hash tables: rps_sock_flow_table and rps_dev_flow_table. rps_sock_table is a global hash table. Each entry is just a CPU number and it is populated in recvmsg and sendmsg as described above. This table contains the "desired" CPUs for flows. rps_dev_flow_table is specific to each device queue. Each entry contains a CPU and a tail queue counter. The CPU is the "current" CPU for a matching flow. The tail queue counter holds the value of a tail queue counter for the associated CPU's backlog queue at the time of last enqueue for a flow matching the entry. Each backlog queue has a queue head counter which is incremented on dequeue, and so a queue tail counter is computed as queue head count + queue length. When a packet is enqueued on a backlog queue, the current value of the queue tail counter is saved in the hash entry of the rps_dev_flow_table. And now the trick: when selecting the CPU for RPS (get_rps_cpu) the rps_sock_flow table and the rps_dev_flow table for the RX queue are consulted. When the desired CPU for the flow (found in the rps_sock_flow table) does not match the current CPU (found in the rps_dev_flow table), the current CPU is changed to the desired CPU if one of the following is true: - The current CPU is unset (equal to RPS_NO_CPU) - Current CPU is offline - The current CPU's queue head counter >= queue tail counter in the rps_dev_flow table. This checks if the queue tail has advanced beyond the last packet that was enqueued using this table entry. This guarantees that all packets queued using this entry have been dequeued, thus preserving in order delivery. Making each queue have its own rps_dev_flow table has two advantages: 1) the tail queue counters will be written on each receive, so keeping the table local to interrupting CPU s good for locality. 2) this allows lockless access to the table-- the CPU number and queue tail counter need to be accessed together under mutual exclusion from netif_receive_skb, we assume that this is only called from device napi_poll which is non-reentrant. This patch implements RFS for TCP and connected UDP sockets. It should be usable for other flow oriented protocols. There are two configuration parameters for RFS. The "rps_flow_entries" kernel init parameter sets the number of entries in the rps_sock_flow_table, the per rxqueue sysfs entry "rps_flow_cnt" contains the number of entries in the rps_dev_flow table for the rxqueue. Both are rounded to power of two. The obvious benefit of RFS (over just RPS) is that it achieves CPU locality between the receive processing for a flow and the applications processing; this can result in increased performance (higher pps, lower latency). The benefits of RFS are dependent on cache hierarchy, application load, and other factors. On simple benchmarks, we don't necessarily see improvement and sometimes see degradation. However, for more complex benchmarks and for applications where cache pressure is much higher this technique seems to perform very well. Below are some benchmark results which show the potential benfit of this patch. The netperf test has 500 instances of netperf TCP_RR test with 1 byte req. and resp. The RPC test is an request/response test similar in structure to netperf RR test ith 100 threads on each host, but does more work in userspace that netperf. e1000e on 8 core Intel No RFS or RPS 104K tps at 30% CPU No RFS (best RPS config): 290K tps at 63% CPU RFS 303K tps at 61% CPU RPC test tps CPU% 50/90/99% usec latency Latency StdDev No RFS/RPS 103K 48% 757/900/3185 4472.35 RPS only: 174K 73% 415/993/2468 491.66 RFS 223K 73% 379/651/1382 315.61 Signed-off-by: Tom Herbert <therbert@google.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-16 23:01:27 +00:00
if (inet_sk(sk)->inet_daddr)
sock_rps_save_rxhash(sk, skb);
rfs: Receive Flow Steering This patch implements receive flow steering (RFS). RFS steers received packets for layer 3 and 4 processing to the CPU where the application for the corresponding flow is running. RFS is an extension of Receive Packet Steering (RPS). The basic idea of RFS is that when an application calls recvmsg (or sendmsg) the application's running CPU is stored in a hash table that is indexed by the connection's rxhash which is stored in the socket structure. The rxhash is passed in skb's received on the connection from netif_receive_skb. For each received packet, the associated rxhash is used to look up the CPU in the hash table, if a valid CPU is set then the packet is steered to that CPU using the RPS mechanisms. The convolution of the simple approach is that it would potentially allow OOO packets. If threads are thrashing around CPUs or multiple threads are trying to read from the same sockets, a quickly changing CPU value in the hash table could cause rampant OOO packets-- we consider this a non-starter. To avoid OOO packets, this solution implements two types of hash tables: rps_sock_flow_table and rps_dev_flow_table. rps_sock_table is a global hash table. Each entry is just a CPU number and it is populated in recvmsg and sendmsg as described above. This table contains the "desired" CPUs for flows. rps_dev_flow_table is specific to each device queue. Each entry contains a CPU and a tail queue counter. The CPU is the "current" CPU for a matching flow. The tail queue counter holds the value of a tail queue counter for the associated CPU's backlog queue at the time of last enqueue for a flow matching the entry. Each backlog queue has a queue head counter which is incremented on dequeue, and so a queue tail counter is computed as queue head count + queue length. When a packet is enqueued on a backlog queue, the current value of the queue tail counter is saved in the hash entry of the rps_dev_flow_table. And now the trick: when selecting the CPU for RPS (get_rps_cpu) the rps_sock_flow table and the rps_dev_flow table for the RX queue are consulted. When the desired CPU for the flow (found in the rps_sock_flow table) does not match the current CPU (found in the rps_dev_flow table), the current CPU is changed to the desired CPU if one of the following is true: - The current CPU is unset (equal to RPS_NO_CPU) - Current CPU is offline - The current CPU's queue head counter >= queue tail counter in the rps_dev_flow table. This checks if the queue tail has advanced beyond the last packet that was enqueued using this table entry. This guarantees that all packets queued using this entry have been dequeued, thus preserving in order delivery. Making each queue have its own rps_dev_flow table has two advantages: 1) the tail queue counters will be written on each receive, so keeping the table local to interrupting CPU s good for locality. 2) this allows lockless access to the table-- the CPU number and queue tail counter need to be accessed together under mutual exclusion from netif_receive_skb, we assume that this is only called from device napi_poll which is non-reentrant. This patch implements RFS for TCP and connected UDP sockets. It should be usable for other flow oriented protocols. There are two configuration parameters for RFS. The "rps_flow_entries" kernel init parameter sets the number of entries in the rps_sock_flow_table, the per rxqueue sysfs entry "rps_flow_cnt" contains the number of entries in the rps_dev_flow table for the rxqueue. Both are rounded to power of two. The obvious benefit of RFS (over just RPS) is that it achieves CPU locality between the receive processing for a flow and the applications processing; this can result in increased performance (higher pps, lower latency). The benefits of RFS are dependent on cache hierarchy, application load, and other factors. On simple benchmarks, we don't necessarily see improvement and sometimes see degradation. However, for more complex benchmarks and for applications where cache pressure is much higher this technique seems to perform very well. Below are some benchmark results which show the potential benfit of this patch. The netperf test has 500 instances of netperf TCP_RR test with 1 byte req. and resp. The RPC test is an request/response test similar in structure to netperf RR test ith 100 threads on each host, but does more work in userspace that netperf. e1000e on 8 core Intel No RFS or RPS 104K tps at 30% CPU No RFS (best RPS config): 290K tps at 63% CPU RFS 303K tps at 61% CPU RPC test tps CPU% 50/90/99% usec latency Latency StdDev No RFS/RPS 103K 48% 757/900/3185 4472.35 RPS only: 174K 73% 415/993/2468 491.66 RFS 223K 73% 379/651/1382 315.61 Signed-off-by: Tom Herbert <therbert@google.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-16 23:01:27 +00:00
ipv4: PKTINFO doesnt need dst reference Le lundi 07 novembre 2011 à 15:33 +0100, Eric Dumazet a écrit : > At least, in recent kernels we dont change dst->refcnt in forwarding > patch (usinf NOREF skb->dst) > > One particular point is the atomic_inc(dst->refcnt) we have to perform > when queuing an UDP packet if socket asked PKTINFO stuff (for example a > typical DNS server has to setup this option) > > I have one patch somewhere that stores the information in skb->cb[] and > avoid the atomic_{inc|dec}(dst->refcnt). > OK I found it, I did some extra tests and believe its ready. [PATCH net-next] ipv4: IP_PKTINFO doesnt need dst reference When a socket uses IP_PKTINFO notifications, we currently force a dst reference for each received skb. Reader has to access dst to get needed information (rt_iif & rt_spec_dst) and must release dst reference. We also forced a dst reference if skb was put in socket backlog, even without IP_PKTINFO handling. This happens under stress/load. We can instead store the needed information in skb->cb[], so that only softirq handler really access dst, improving cache hit ratios. This removes two atomic operations per packet, and false sharing as well. On a benchmark using a mono threaded receiver (doing only recvmsg() calls), I can reach 720.000 pps instead of 570.000 pps. IP_PKTINFO is typically used by DNS servers, and any multihomed aware UDP application. Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 07:24:35 +00:00
rc = sock_queue_rcv_skb(sk, skb);
if (rc < 0) {
int is_udplite = IS_UDPLITE(sk);
/* Note that an ENOMEM error is charged twice */
if (rc == -ENOMEM)
UDP_INC_STATS_BH(sock_net(sk), UDP_MIB_RCVBUFERRORS,
is_udplite);
UDP_INC_STATS_BH(sock_net(sk), UDP_MIB_INERRORS, is_udplite);
kfree_skb(skb);
trace_udp_fail_queue_rcv_skb(rc, sk);
return -1;
}
return 0;
}
static struct static_key udp_encap_needed __read_mostly;
void udp_encap_enable(void)
{
if (!static_key_enabled(&udp_encap_needed))
static_key_slow_inc(&udp_encap_needed);
}
EXPORT_SYMBOL(udp_encap_enable);
/* returns:
* -1: error
* 0: success
* >0: "udp encap" protocol resubmission
*
* Note that in the success and error cases, the skb is assumed to
* have either been requeued or freed.
*/
int udp_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
{
struct udp_sock *up = udp_sk(sk);
int rc;
int is_udplite = IS_UDPLITE(sk);
/*
* Charge it to the socket, dropping if the queue is full.
*/
if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
goto drop;
nf_reset(skb);
if (static_key_false(&udp_encap_needed) && up->encap_type) {
int (*encap_rcv)(struct sock *sk, struct sk_buff *skb);
/*
* This is an encapsulation socket so pass the skb to
* the socket's udp_encap_rcv() hook. Otherwise, just
* fall through and pass this up the UDP socket.
* up->encap_rcv() returns the following value:
* =0 if skb was successfully passed to the encap
* handler or was discarded by it.
* >0 if skb should be passed on to UDP.
* <0 if skb should be resubmitted as proto -N
*/
/* if we're overly short, let UDP handle it */
encap_rcv = ACCESS_ONCE(up->encap_rcv);
if (skb->len > sizeof(struct udphdr) && encap_rcv != NULL) {
int ret;
ret = encap_rcv(sk, skb);
if (ret <= 0) {
UDP_INC_STATS_BH(sock_net(sk),
UDP_MIB_INDATAGRAMS,
is_udplite);
return -ret;
}
}
/* FALLTHROUGH -- it's a UDP Packet */
}
/*
* UDP-Lite specific tests, ignored on UDP sockets
*/
if ((is_udplite & UDPLITE_RECV_CC) && UDP_SKB_CB(skb)->partial_cov) {
/*
* MIB statistics other than incrementing the error count are
* disabled for the following two types of errors: these depend
* on the application settings, not on the functioning of the
* protocol stack as such.
*
* RFC 3828 here recommends (sec 3.3): "There should also be a
* way ... to ... at least let the receiving application block
* delivery of packets with coverage values less than a value
* provided by the application."
*/
if (up->pcrlen == 0) { /* full coverage was set */
LIMIT_NETDEBUG(KERN_WARNING "UDPLite: partial coverage %d while full coverage %d requested\n",
UDP_SKB_CB(skb)->cscov, skb->len);
goto drop;
}
/* The next case involves violating the min. coverage requested
* by the receiver. This is subtle: if receiver wants x and x is
* greater than the buffersize/MTU then receiver will complain
* that it wants x while sender emits packets of smaller size y.
* Therefore the above ...()->partial_cov statement is essential.
*/
if (UDP_SKB_CB(skb)->cscov < up->pcrlen) {
LIMIT_NETDEBUG(KERN_WARNING "UDPLite: coverage %d too small, need min %d\n",
UDP_SKB_CB(skb)->cscov, up->pcrlen);
goto drop;
}
}
if (rcu_access_pointer(sk->sk_filter) &&
udp_lib_checksum_complete(skb))
goto drop;
if (sk_rcvqueues_full(sk, skb, sk->sk_rcvbuf))
goto drop;
rc = 0;
ipv4: PKTINFO doesnt need dst reference Le lundi 07 novembre 2011 à 15:33 +0100, Eric Dumazet a écrit : > At least, in recent kernels we dont change dst->refcnt in forwarding > patch (usinf NOREF skb->dst) > > One particular point is the atomic_inc(dst->refcnt) we have to perform > when queuing an UDP packet if socket asked PKTINFO stuff (for example a > typical DNS server has to setup this option) > > I have one patch somewhere that stores the information in skb->cb[] and > avoid the atomic_{inc|dec}(dst->refcnt). > OK I found it, I did some extra tests and believe its ready. [PATCH net-next] ipv4: IP_PKTINFO doesnt need dst reference When a socket uses IP_PKTINFO notifications, we currently force a dst reference for each received skb. Reader has to access dst to get needed information (rt_iif & rt_spec_dst) and must release dst reference. We also forced a dst reference if skb was put in socket backlog, even without IP_PKTINFO handling. This happens under stress/load. We can instead store the needed information in skb->cb[], so that only softirq handler really access dst, improving cache hit ratios. This removes two atomic operations per packet, and false sharing as well. On a benchmark using a mono threaded receiver (doing only recvmsg() calls), I can reach 720.000 pps instead of 570.000 pps. IP_PKTINFO is typically used by DNS servers, and any multihomed aware UDP application. Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 07:24:35 +00:00
ipv4_pktinfo_prepare(skb);
bh_lock_sock(sk);
if (!sock_owned_by_user(sk))
rc = __udp_queue_rcv_skb(sk, skb);
else if (sk_add_backlog(sk, skb, sk->sk_rcvbuf)) {
bh_unlock_sock(sk);
goto drop;
}
bh_unlock_sock(sk);
return rc;
drop:
UDP_INC_STATS_BH(sock_net(sk), UDP_MIB_INERRORS, is_udplite);
atomic_inc(&sk->sk_drops);
kfree_skb(skb);
return -1;
}
static void flush_stack(struct sock **stack, unsigned int count,
struct sk_buff *skb, unsigned int final)
{
unsigned int i;
struct sk_buff *skb1 = NULL;
struct sock *sk;
for (i = 0; i < count; i++) {
sk = stack[i];
if (likely(skb1 == NULL))
skb1 = (i == final) ? skb : skb_clone(skb, GFP_ATOMIC);
if (!skb1) {
atomic_inc(&sk->sk_drops);
UDP_INC_STATS_BH(sock_net(sk), UDP_MIB_RCVBUFERRORS,
IS_UDPLITE(sk));
UDP_INC_STATS_BH(sock_net(sk), UDP_MIB_INERRORS,
IS_UDPLITE(sk));
}
if (skb1 && udp_queue_rcv_skb(sk, skb1) <= 0)
skb1 = NULL;
}
if (unlikely(skb1))
kfree_skb(skb1);
}
/*
* Multicasts and broadcasts go to each listener.
*
* Note: called only from the BH handler context.
*/
static int __udp4_lib_mcast_deliver(struct net *net, struct sk_buff *skb,
struct udphdr *uh,
__be32 saddr, __be32 daddr,
struct udp_table *udptable)
{
struct sock *sk, *stack[256 / sizeof(struct sock *)];
struct udp_hslot *hslot = udp_hashslot(udptable, net, ntohs(uh->dest));
int dif;
unsigned int i, count = 0;
spin_lock(&hslot->lock);
sk = sk_nulls_head(&hslot->head);
dif = skb->dev->ifindex;
sk = udp_v4_mcast_next(net, sk, uh->dest, daddr, uh->source, saddr, dif);
while (sk) {
stack[count++] = sk;
sk = udp_v4_mcast_next(net, sk_nulls_next(sk), uh->dest,
daddr, uh->source, saddr, dif);
if (unlikely(count == ARRAY_SIZE(stack))) {
if (!sk)
break;
flush_stack(stack, count, skb, ~0);
count = 0;
}
}
/*
* before releasing chain lock, we must take a reference on sockets
*/
for (i = 0; i < count; i++)
sock_hold(stack[i]);
spin_unlock(&hslot->lock);
/*
* do the slow work with no lock held
*/
if (count) {
flush_stack(stack, count, skb, count - 1);
for (i = 0; i < count; i++)
sock_put(stack[i]);
} else {
kfree_skb(skb);
}
return 0;
}
/* Initialize UDP checksum. If exited with zero value (success),
* CHECKSUM_UNNECESSARY means, that no more checks are required.
* Otherwise, csum completion requires chacksumming packet body,
* including udp header and folding it to skb->csum.
*/
static inline int udp4_csum_init(struct sk_buff *skb, struct udphdr *uh,
int proto)
{
const struct iphdr *iph;
int err;
UDP_SKB_CB(skb)->partial_cov = 0;
UDP_SKB_CB(skb)->cscov = skb->len;
if (proto == IPPROTO_UDPLITE) {
err = udplite_checksum_init(skb, uh);
if (err)
return err;
}
iph = ip_hdr(skb);
if (uh->check == 0) {
skb->ip_summed = CHECKSUM_UNNECESSARY;
} else if (skb->ip_summed == CHECKSUM_COMPLETE) {
if (!csum_tcpudp_magic(iph->saddr, iph->daddr, skb->len,
proto, skb->csum))
skb->ip_summed = CHECKSUM_UNNECESSARY;
}
if (!skb_csum_unnecessary(skb))
skb->csum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
skb->len, proto, 0);
/* Probably, we should checksum udp header (it should be in cache
* in any case) and data in tiny packets (< rx copybreak).
*/
return 0;
}
/*
* All we need to do is get the socket, and then do a checksum.
*/
int __udp4_lib_rcv(struct sk_buff *skb, struct udp_table *udptable,
int proto)
{
struct sock *sk;
struct udphdr *uh;
unsigned short ulen;
struct rtable *rt = skb_rtable(skb);
__be32 saddr, daddr;
struct net *net = dev_net(skb->dev);
/*
* Validate the packet.
*/
if (!pskb_may_pull(skb, sizeof(struct udphdr)))
goto drop; /* No space for header. */
uh = udp_hdr(skb);
ulen = ntohs(uh->len);
saddr = ip_hdr(skb)->saddr;
daddr = ip_hdr(skb)->daddr;
if (ulen > skb->len)
goto short_packet;
if (proto == IPPROTO_UDP) {
/* UDP validates ulen. */
if (ulen < sizeof(*uh) || pskb_trim_rcsum(skb, ulen))
goto short_packet;
uh = udp_hdr(skb);
}
if (udp4_csum_init(skb, uh, proto))
goto csum_error;
if (rt->rt_flags & (RTCF_BROADCAST|RTCF_MULTICAST))
return __udp4_lib_mcast_deliver(net, skb, uh,
saddr, daddr, udptable);
sk = __udp4_lib_lookup_skb(skb, uh->source, uh->dest, udptable);
if (sk != NULL) {
int ret = udp_queue_rcv_skb(sk, skb);
sock_put(sk);
/* a return value > 0 means to resubmit the input, but
* it wants the return to be -protocol, or 0
*/
if (ret > 0)
return -ret;
return 0;
}
if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb))
goto drop;
nf_reset(skb);
/* No socket. Drop packet silently, if checksum is wrong */
if (udp_lib_checksum_complete(skb))
goto csum_error;
UDP_INC_STATS_BH(net, UDP_MIB_NOPORTS, proto == IPPROTO_UDPLITE);
icmp_send(skb, ICMP_DEST_UNREACH, ICMP_PORT_UNREACH, 0);
/*
* Hmm. We got an UDP packet to a port to which we
* don't wanna listen. Ignore it.
*/
kfree_skb(skb);
return 0;
short_packet:
LIMIT_NETDEBUG(KERN_DEBUG "UDP%s: short packet: From %pI4:%u %d/%d to %pI4:%u\n",
proto == IPPROTO_UDPLITE ? "Lite" : "",
&saddr, ntohs(uh->source),
ulen, skb->len,
&daddr, ntohs(uh->dest));
goto drop;
csum_error:
/*
* RFC1122: OK. Discards the bad packet silently (as far as
* the network is concerned, anyway) as per 4.1.3.4 (MUST).
*/
LIMIT_NETDEBUG(KERN_DEBUG "UDP%s: bad checksum. From %pI4:%u to %pI4:%u ulen %d\n",
proto == IPPROTO_UDPLITE ? "Lite" : "",
&saddr, ntohs(uh->source), &daddr, ntohs(uh->dest),
ulen);
drop:
UDP_INC_STATS_BH(net, UDP_MIB_INERRORS, proto == IPPROTO_UDPLITE);
kfree_skb(skb);
return 0;
}
int udp_rcv(struct sk_buff *skb)
{
return __udp4_lib_rcv(skb, &udp_table, IPPROTO_UDP);
}
void udp_destroy_sock(struct sock *sk)
{
bool slow = lock_sock_fast(sk);
udp_flush_pending_frames(sk);
unlock_sock_fast(sk, slow);
}
/*
* Socket option code for UDP
*/
int udp_lib_setsockopt(struct sock *sk, int level, int optname,
char __user *optval, unsigned int optlen,
int (*push_pending_frames)(struct sock *))
{
struct udp_sock *up = udp_sk(sk);
int val;
int err = 0;
int is_udplite = IS_UDPLITE(sk);
if (optlen < sizeof(int))
return -EINVAL;
if (get_user(val, (int __user *)optval))
return -EFAULT;
switch (optname) {
case UDP_CORK:
if (val != 0) {
up->corkflag = 1;
} else {
up->corkflag = 0;
lock_sock(sk);
(*push_pending_frames)(sk);
release_sock(sk);
}
break;
case UDP_ENCAP:
switch (val) {
case 0:
case UDP_ENCAP_ESPINUDP:
case UDP_ENCAP_ESPINUDP_NON_IKE:
up->encap_rcv = xfrm4_udp_encap_rcv;
/* FALLTHROUGH */
case UDP_ENCAP_L2TPINUDP:
up->encap_type = val;
udp_encap_enable();
break;
default:
err = -ENOPROTOOPT;
break;
}
break;
[NET]: Supporting UDP-Lite (RFC 3828) in Linux This is a revision of the previously submitted patch, which alters the way files are organized and compiled in the following manner: * UDP and UDP-Lite now use separate object files * source file dependencies resolved via header files net/ipv{4,6}/udp_impl.h * order of inclusion files in udp.c/udplite.c adapted accordingly [NET/IPv4]: Support for the UDP-Lite protocol (RFC 3828) This patch adds support for UDP-Lite to the IPv4 stack, provided as an extension to the existing UDPv4 code: * generic routines are all located in net/ipv4/udp.c * UDP-Lite specific routines are in net/ipv4/udplite.c * MIB/statistics support in /proc/net/snmp and /proc/net/udplite * shared API with extensions for partial checksum coverage [NET/IPv6]: Extension for UDP-Lite over IPv6 It extends the existing UDPv6 code base with support for UDP-Lite in the same manner as per UDPv4. In particular, * UDPv6 generic and shared code is in net/ipv6/udp.c * UDP-Litev6 specific extensions are in net/ipv6/udplite.c * MIB/statistics support in /proc/net/snmp6 and /proc/net/udplite6 * support for IPV6_ADDRFORM * aligned the coding style of protocol initialisation with af_inet6.c * made the error handling in udpv6_queue_rcv_skb consistent; to return `-1' on error on all error cases * consolidation of shared code [NET]: UDP-Lite Documentation and basic XFRM/Netfilter support The UDP-Lite patch further provides * API documentation for UDP-Lite * basic xfrm support * basic netfilter support for IPv4 and IPv6 (LOG target) Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-27 19:10:57 +00:00
/*
* UDP-Lite's partial checksum coverage (RFC 3828).
*/
/* The sender sets actual checksum coverage length via this option.
* The case coverage > packet length is handled by send module. */
case UDPLITE_SEND_CSCOV:
if (!is_udplite) /* Disable the option on UDP sockets */
[NET]: Supporting UDP-Lite (RFC 3828) in Linux This is a revision of the previously submitted patch, which alters the way files are organized and compiled in the following manner: * UDP and UDP-Lite now use separate object files * source file dependencies resolved via header files net/ipv{4,6}/udp_impl.h * order of inclusion files in udp.c/udplite.c adapted accordingly [NET/IPv4]: Support for the UDP-Lite protocol (RFC 3828) This patch adds support for UDP-Lite to the IPv4 stack, provided as an extension to the existing UDPv4 code: * generic routines are all located in net/ipv4/udp.c * UDP-Lite specific routines are in net/ipv4/udplite.c * MIB/statistics support in /proc/net/snmp and /proc/net/udplite * shared API with extensions for partial checksum coverage [NET/IPv6]: Extension for UDP-Lite over IPv6 It extends the existing UDPv6 code base with support for UDP-Lite in the same manner as per UDPv4. In particular, * UDPv6 generic and shared code is in net/ipv6/udp.c * UDP-Litev6 specific extensions are in net/ipv6/udplite.c * MIB/statistics support in /proc/net/snmp6 and /proc/net/udplite6 * support for IPV6_ADDRFORM * aligned the coding style of protocol initialisation with af_inet6.c * made the error handling in udpv6_queue_rcv_skb consistent; to return `-1' on error on all error cases * consolidation of shared code [NET]: UDP-Lite Documentation and basic XFRM/Netfilter support The UDP-Lite patch further provides * API documentation for UDP-Lite * basic xfrm support * basic netfilter support for IPv4 and IPv6 (LOG target) Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-27 19:10:57 +00:00
return -ENOPROTOOPT;
if (val != 0 && val < 8) /* Illegal coverage: use default (8) */
val = 8;
else if (val > USHRT_MAX)
val = USHRT_MAX;
[NET]: Supporting UDP-Lite (RFC 3828) in Linux This is a revision of the previously submitted patch, which alters the way files are organized and compiled in the following manner: * UDP and UDP-Lite now use separate object files * source file dependencies resolved via header files net/ipv{4,6}/udp_impl.h * order of inclusion files in udp.c/udplite.c adapted accordingly [NET/IPv4]: Support for the UDP-Lite protocol (RFC 3828) This patch adds support for UDP-Lite to the IPv4 stack, provided as an extension to the existing UDPv4 code: * generic routines are all located in net/ipv4/udp.c * UDP-Lite specific routines are in net/ipv4/udplite.c * MIB/statistics support in /proc/net/snmp and /proc/net/udplite * shared API with extensions for partial checksum coverage [NET/IPv6]: Extension for UDP-Lite over IPv6 It extends the existing UDPv6 code base with support for UDP-Lite in the same manner as per UDPv4. In particular, * UDPv6 generic and shared code is in net/ipv6/udp.c * UDP-Litev6 specific extensions are in net/ipv6/udplite.c * MIB/statistics support in /proc/net/snmp6 and /proc/net/udplite6 * support for IPV6_ADDRFORM * aligned the coding style of protocol initialisation with af_inet6.c * made the error handling in udpv6_queue_rcv_skb consistent; to return `-1' on error on all error cases * consolidation of shared code [NET]: UDP-Lite Documentation and basic XFRM/Netfilter support The UDP-Lite patch further provides * API documentation for UDP-Lite * basic xfrm support * basic netfilter support for IPv4 and IPv6 (LOG target) Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-27 19:10:57 +00:00
up->pcslen = val;
up->pcflag |= UDPLITE_SEND_CC;
break;
/* The receiver specifies a minimum checksum coverage value. To make
* sense, this should be set to at least 8 (as done below). If zero is
[NET]: Supporting UDP-Lite (RFC 3828) in Linux This is a revision of the previously submitted patch, which alters the way files are organized and compiled in the following manner: * UDP and UDP-Lite now use separate object files * source file dependencies resolved via header files net/ipv{4,6}/udp_impl.h * order of inclusion files in udp.c/udplite.c adapted accordingly [NET/IPv4]: Support for the UDP-Lite protocol (RFC 3828) This patch adds support for UDP-Lite to the IPv4 stack, provided as an extension to the existing UDPv4 code: * generic routines are all located in net/ipv4/udp.c * UDP-Lite specific routines are in net/ipv4/udplite.c * MIB/statistics support in /proc/net/snmp and /proc/net/udplite * shared API with extensions for partial checksum coverage [NET/IPv6]: Extension for UDP-Lite over IPv6 It extends the existing UDPv6 code base with support for UDP-Lite in the same manner as per UDPv4. In particular, * UDPv6 generic and shared code is in net/ipv6/udp.c * UDP-Litev6 specific extensions are in net/ipv6/udplite.c * MIB/statistics support in /proc/net/snmp6 and /proc/net/udplite6 * support for IPV6_ADDRFORM * aligned the coding style of protocol initialisation with af_inet6.c * made the error handling in udpv6_queue_rcv_skb consistent; to return `-1' on error on all error cases * consolidation of shared code [NET]: UDP-Lite Documentation and basic XFRM/Netfilter support The UDP-Lite patch further provides * API documentation for UDP-Lite * basic xfrm support * basic netfilter support for IPv4 and IPv6 (LOG target) Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-27 19:10:57 +00:00
* used, this again means full checksum coverage. */
case UDPLITE_RECV_CSCOV:
if (!is_udplite) /* Disable the option on UDP sockets */
[NET]: Supporting UDP-Lite (RFC 3828) in Linux This is a revision of the previously submitted patch, which alters the way files are organized and compiled in the following manner: * UDP and UDP-Lite now use separate object files * source file dependencies resolved via header files net/ipv{4,6}/udp_impl.h * order of inclusion files in udp.c/udplite.c adapted accordingly [NET/IPv4]: Support for the UDP-Lite protocol (RFC 3828) This patch adds support for UDP-Lite to the IPv4 stack, provided as an extension to the existing UDPv4 code: * generic routines are all located in net/ipv4/udp.c * UDP-Lite specific routines are in net/ipv4/udplite.c * MIB/statistics support in /proc/net/snmp and /proc/net/udplite * shared API with extensions for partial checksum coverage [NET/IPv6]: Extension for UDP-Lite over IPv6 It extends the existing UDPv6 code base with support for UDP-Lite in the same manner as per UDPv4. In particular, * UDPv6 generic and shared code is in net/ipv6/udp.c * UDP-Litev6 specific extensions are in net/ipv6/udplite.c * MIB/statistics support in /proc/net/snmp6 and /proc/net/udplite6 * support for IPV6_ADDRFORM * aligned the coding style of protocol initialisation with af_inet6.c * made the error handling in udpv6_queue_rcv_skb consistent; to return `-1' on error on all error cases * consolidation of shared code [NET]: UDP-Lite Documentation and basic XFRM/Netfilter support The UDP-Lite patch further provides * API documentation for UDP-Lite * basic xfrm support * basic netfilter support for IPv4 and IPv6 (LOG target) Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-27 19:10:57 +00:00
return -ENOPROTOOPT;
if (val != 0 && val < 8) /* Avoid silly minimal values. */
val = 8;
else if (val > USHRT_MAX)
val = USHRT_MAX;
[NET]: Supporting UDP-Lite (RFC 3828) in Linux This is a revision of the previously submitted patch, which alters the way files are organized and compiled in the following manner: * UDP and UDP-Lite now use separate object files * source file dependencies resolved via header files net/ipv{4,6}/udp_impl.h * order of inclusion files in udp.c/udplite.c adapted accordingly [NET/IPv4]: Support for the UDP-Lite protocol (RFC 3828) This patch adds support for UDP-Lite to the IPv4 stack, provided as an extension to the existing UDPv4 code: * generic routines are all located in net/ipv4/udp.c * UDP-Lite specific routines are in net/ipv4/udplite.c * MIB/statistics support in /proc/net/snmp and /proc/net/udplite * shared API with extensions for partial checksum coverage [NET/IPv6]: Extension for UDP-Lite over IPv6 It extends the existing UDPv6 code base with support for UDP-Lite in the same manner as per UDPv4. In particular, * UDPv6 generic and shared code is in net/ipv6/udp.c * UDP-Litev6 specific extensions are in net/ipv6/udplite.c * MIB/statistics support in /proc/net/snmp6 and /proc/net/udplite6 * support for IPV6_ADDRFORM * aligned the coding style of protocol initialisation with af_inet6.c * made the error handling in udpv6_queue_rcv_skb consistent; to return `-1' on error on all error cases * consolidation of shared code [NET]: UDP-Lite Documentation and basic XFRM/Netfilter support The UDP-Lite patch further provides * API documentation for UDP-Lite * basic xfrm support * basic netfilter support for IPv4 and IPv6 (LOG target) Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-27 19:10:57 +00:00
up->pcrlen = val;
up->pcflag |= UDPLITE_RECV_CC;
break;
default:
err = -ENOPROTOOPT;
break;
}
return err;
}
EXPORT_SYMBOL(udp_lib_setsockopt);
int udp_setsockopt(struct sock *sk, int level, int optname,
char __user *optval, unsigned int optlen)
{
if (level == SOL_UDP || level == SOL_UDPLITE)
return udp_lib_setsockopt(sk, level, optname, optval, optlen,
udp_push_pending_frames);
return ip_setsockopt(sk, level, optname, optval, optlen);
}
#ifdef CONFIG_COMPAT
int compat_udp_setsockopt(struct sock *sk, int level, int optname,
char __user *optval, unsigned int optlen)
{
if (level == SOL_UDP || level == SOL_UDPLITE)
return udp_lib_setsockopt(sk, level, optname, optval, optlen,
udp_push_pending_frames);
return compat_ip_setsockopt(sk, level, optname, optval, optlen);
}
#endif
int udp_lib_getsockopt(struct sock *sk, int level, int optname,
char __user *optval, int __user *optlen)
{
struct udp_sock *up = udp_sk(sk);
int val, len;
if (get_user(len, optlen))
return -EFAULT;
len = min_t(unsigned int, len, sizeof(int));
if (len < 0)
return -EINVAL;
switch (optname) {
case UDP_CORK:
val = up->corkflag;
break;
case UDP_ENCAP:
val = up->encap_type;
break;
[NET]: Supporting UDP-Lite (RFC 3828) in Linux This is a revision of the previously submitted patch, which alters the way files are organized and compiled in the following manner: * UDP and UDP-Lite now use separate object files * source file dependencies resolved via header files net/ipv{4,6}/udp_impl.h * order of inclusion files in udp.c/udplite.c adapted accordingly [NET/IPv4]: Support for the UDP-Lite protocol (RFC 3828) This patch adds support for UDP-Lite to the IPv4 stack, provided as an extension to the existing UDPv4 code: * generic routines are all located in net/ipv4/udp.c * UDP-Lite specific routines are in net/ipv4/udplite.c * MIB/statistics support in /proc/net/snmp and /proc/net/udplite * shared API with extensions for partial checksum coverage [NET/IPv6]: Extension for UDP-Lite over IPv6 It extends the existing UDPv6 code base with support for UDP-Lite in the same manner as per UDPv4. In particular, * UDPv6 generic and shared code is in net/ipv6/udp.c * UDP-Litev6 specific extensions are in net/ipv6/udplite.c * MIB/statistics support in /proc/net/snmp6 and /proc/net/udplite6 * support for IPV6_ADDRFORM * aligned the coding style of protocol initialisation with af_inet6.c * made the error handling in udpv6_queue_rcv_skb consistent; to return `-1' on error on all error cases * consolidation of shared code [NET]: UDP-Lite Documentation and basic XFRM/Netfilter support The UDP-Lite patch further provides * API documentation for UDP-Lite * basic xfrm support * basic netfilter support for IPv4 and IPv6 (LOG target) Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-27 19:10:57 +00:00
/* The following two cannot be changed on UDP sockets, the return is
* always 0 (which corresponds to the full checksum coverage of UDP). */
case UDPLITE_SEND_CSCOV:
val = up->pcslen;
break;
case UDPLITE_RECV_CSCOV:
val = up->pcrlen;
break;
default:
return -ENOPROTOOPT;
}
if (put_user(len, optlen))
return -EFAULT;
if (copy_to_user(optval, &val, len))
return -EFAULT;
return 0;
}
EXPORT_SYMBOL(udp_lib_getsockopt);
int udp_getsockopt(struct sock *sk, int level, int optname,
char __user *optval, int __user *optlen)
{
if (level == SOL_UDP || level == SOL_UDPLITE)
return udp_lib_getsockopt(sk, level, optname, optval, optlen);
return ip_getsockopt(sk, level, optname, optval, optlen);
}
#ifdef CONFIG_COMPAT
int compat_udp_getsockopt(struct sock *sk, int level, int optname,
char __user *optval, int __user *optlen)
{
if (level == SOL_UDP || level == SOL_UDPLITE)
return udp_lib_getsockopt(sk, level, optname, optval, optlen);
return compat_ip_getsockopt(sk, level, optname, optval, optlen);
}
#endif
/**
* udp_poll - wait for a UDP event.
* @file - file struct
* @sock - socket
* @wait - poll table
*
* This is same as datagram poll, except for the special case of
* blocking sockets. If application is using a blocking fd
* and a packet with checksum error is in the queue;
* then it could get return from select indicating data available
* but then block when reading it. Add special case code
* to work around these arguably broken applications.
*/
unsigned int udp_poll(struct file *file, struct socket *sock, poll_table *wait)
{
unsigned int mask = datagram_poll(file, sock, wait);
struct sock *sk = sock->sk;
[NET]: Supporting UDP-Lite (RFC 3828) in Linux This is a revision of the previously submitted patch, which alters the way files are organized and compiled in the following manner: * UDP and UDP-Lite now use separate object files * source file dependencies resolved via header files net/ipv{4,6}/udp_impl.h * order of inclusion files in udp.c/udplite.c adapted accordingly [NET/IPv4]: Support for the UDP-Lite protocol (RFC 3828) This patch adds support for UDP-Lite to the IPv4 stack, provided as an extension to the existing UDPv4 code: * generic routines are all located in net/ipv4/udp.c * UDP-Lite specific routines are in net/ipv4/udplite.c * MIB/statistics support in /proc/net/snmp and /proc/net/udplite * shared API with extensions for partial checksum coverage [NET/IPv6]: Extension for UDP-Lite over IPv6 It extends the existing UDPv6 code base with support for UDP-Lite in the same manner as per UDPv4. In particular, * UDPv6 generic and shared code is in net/ipv6/udp.c * UDP-Litev6 specific extensions are in net/ipv6/udplite.c * MIB/statistics support in /proc/net/snmp6 and /proc/net/udplite6 * support for IPV6_ADDRFORM * aligned the coding style of protocol initialisation with af_inet6.c * made the error handling in udpv6_queue_rcv_skb consistent; to return `-1' on error on all error cases * consolidation of shared code [NET]: UDP-Lite Documentation and basic XFRM/Netfilter support The UDP-Lite patch further provides * API documentation for UDP-Lite * basic xfrm support * basic netfilter support for IPv4 and IPv6 (LOG target) Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-27 19:10:57 +00:00
/* Check for false positives due to checksum errors */
if ((mask & POLLRDNORM) && !(file->f_flags & O_NONBLOCK) &&
!(sk->sk_shutdown & RCV_SHUTDOWN) && !first_packet_length(sk))
mask &= ~(POLLIN | POLLRDNORM);
return mask;
}
EXPORT_SYMBOL(udp_poll);
struct proto udp_prot = {
.name = "UDP",
.owner = THIS_MODULE,
.close = udp_lib_close,
.connect = ip4_datagram_connect,
.disconnect = udp_disconnect,
.ioctl = udp_ioctl,
.destroy = udp_destroy_sock,
.setsockopt = udp_setsockopt,
.getsockopt = udp_getsockopt,
.sendmsg = udp_sendmsg,
.recvmsg = udp_recvmsg,
.sendpage = udp_sendpage,
.backlog_rcv = __udp_queue_rcv_skb,
.hash = udp_lib_hash,
.unhash = udp_lib_unhash,
.rehash = udp_v4_rehash,
.get_port = udp_v4_get_port,
.memory_allocated = &udp_memory_allocated,
.sysctl_mem = sysctl_udp_mem,
.sysctl_wmem = &sysctl_udp_wmem_min,
.sysctl_rmem = &sysctl_udp_rmem_min,
.obj_size = sizeof(struct udp_sock),
udp: RCU handling for Unicast packets. Goals are : 1) Optimizing handling of incoming Unicast UDP frames, so that no memory writes should happen in the fast path. Note: Multicasts and broadcasts still will need to take a lock, because doing a full lockless lookup in this case is difficult. 2) No expensive operations in the socket bind/unhash phases : - No expensive synchronize_rcu() calls. - No added rcu_head in socket structure, increasing memory needs, but more important, forcing us to use call_rcu() calls, that have the bad property of making sockets structure cold. (rcu grace period between socket freeing and its potential reuse make this socket being cold in CPU cache). David did a previous patch using call_rcu() and noticed a 20% impact on TCP connection rates. Quoting Cristopher Lameter : "Right. That results in cacheline cooldown. You'd want to recycle the object as they are cache hot on a per cpu basis. That is screwed up by the delayed regular rcu processing. We have seen multiple regressions due to cacheline cooldown. The only choice in cacheline hot sensitive areas is to deal with the complexity that comes with SLAB_DESTROY_BY_RCU or give up on RCU." - Because udp sockets are allocated from dedicated kmem_cache, use of SLAB_DESTROY_BY_RCU can help here. Theory of operation : --------------------- As the lookup is lockfree (using rcu_read_lock()/rcu_read_unlock()), special attention must be taken by readers and writers. Use of SLAB_DESTROY_BY_RCU is tricky too, because a socket can be freed, reused, inserted in a different chain or in worst case in the same chain while readers could do lookups in the same time. In order to avoid loops, a reader must check each socket found in a chain really belongs to the chain the reader was traversing. If it finds a mismatch, lookup must start again at the begining. This *restart* loop is the reason we had to use rdlock for the multicast case, because we dont want to send same message several times to the same socket. We use RCU only for fast path. Thus, /proc/net/udp still takes spinlocks. Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-29 09:11:14 +00:00
.slab_flags = SLAB_DESTROY_BY_RCU,
.h.udp_table = &udp_table,
#ifdef CONFIG_COMPAT
.compat_setsockopt = compat_udp_setsockopt,
.compat_getsockopt = compat_udp_getsockopt,
#endif
.clear_sk = sk_prot_clear_portaddr_nulls,
};
EXPORT_SYMBOL(udp_prot);
/* ------------------------------------------------------------------------ */
#ifdef CONFIG_PROC_FS
static struct sock *udp_get_first(struct seq_file *seq, int start)
{
struct sock *sk;
struct udp_iter_state *state = seq->private;
struct net *net = seq_file_net(seq);
for (state->bucket = start; state->bucket <= state->udp_table->mask;
++state->bucket) {
struct hlist_nulls_node *node;
struct udp_hslot *hslot = &state->udp_table->hash[state->bucket];
if (hlist_nulls_empty(&hslot->head))
continue;
spin_lock_bh(&hslot->lock);
sk_nulls_for_each(sk, node, &hslot->head) {
if (!net_eq(sock_net(sk), net))
continue;
if (sk->sk_family == state->family)
goto found;
}
spin_unlock_bh(&hslot->lock);
}
sk = NULL;
found:
return sk;
}
static struct sock *udp_get_next(struct seq_file *seq, struct sock *sk)
{
struct udp_iter_state *state = seq->private;
struct net *net = seq_file_net(seq);
do {
sk = sk_nulls_next(sk);
} while (sk && (!net_eq(sock_net(sk), net) || sk->sk_family != state->family));
if (!sk) {
if (state->bucket <= state->udp_table->mask)
udp: Wrong locking code in udp seq_file infrastructure Reading zero bytes from /proc/net/udp or other similar files which use the same seq_file udp infrastructure panics kernel in that way: ===================================== [ BUG: bad unlock balance detected! ] ------------------------------------- read/1985 is trying to release lock (&table->hash[i].lock) at: [<ffffffff81321d83>] udp_seq_stop+0x27/0x29 but there are no more locks to release! other info that might help us debug this: 1 lock held by read/1985: #0: (&p->lock){--..}, at: [<ffffffff810eefb6>] seq_read+0x38/0x348 stack backtrace: Pid: 1985, comm: read Not tainted 2.6.29-rc8 #9 Call Trace: [<ffffffff81321d83>] ? udp_seq_stop+0x27/0x29 [<ffffffff8106dab9>] print_unlock_inbalance_bug+0xd6/0xe1 [<ffffffff8106db62>] lock_release_non_nested+0x9e/0x1c6 [<ffffffff810ef030>] ? seq_read+0xb2/0x348 [<ffffffff8106bdba>] ? mark_held_locks+0x68/0x86 [<ffffffff81321d83>] ? udp_seq_stop+0x27/0x29 [<ffffffff8106dde7>] lock_release+0x15d/0x189 [<ffffffff8137163c>] _spin_unlock_bh+0x1e/0x34 [<ffffffff81321d83>] udp_seq_stop+0x27/0x29 [<ffffffff810ef239>] seq_read+0x2bb/0x348 [<ffffffff810eef7e>] ? seq_read+0x0/0x348 [<ffffffff8111aedd>] proc_reg_read+0x90/0xaf [<ffffffff810d878f>] vfs_read+0xa6/0x103 [<ffffffff8106bfac>] ? trace_hardirqs_on_caller+0x12f/0x153 [<ffffffff810d88a2>] sys_read+0x45/0x69 [<ffffffff8101123a>] system_call_fastpath+0x16/0x1b BUG: scheduling while atomic: read/1985/0xffffff00 INFO: lockdep is turned off. Modules linked in: cpufreq_ondemand acpi_cpufreq freq_table dm_multipath kvm ppdev snd_hda_codec_analog snd_hda_intel snd_hda_codec snd_hwdep snd_seq_dummy snd_seq_oss snd_seq_midi_event arc4 snd_s eq ecb thinkpad_acpi snd_seq_device iwl3945 hwmon sdhci_pci snd_pcm_oss sdhci rfkill mmc_core snd_mixer_oss i2c_i801 mac80211 yenta_socket ricoh_mmc i2c_core iTCO_wdt snd_pcm iTCO_vendor_support rs rc_nonstatic snd_timer snd lib80211 cfg80211 soundcore snd_page_alloc video parport_pc output parport e1000e [last unloaded: scsi_wait_scan] Pid: 1985, comm: read Not tainted 2.6.29-rc8 #9 Call Trace: [<ffffffff8106b456>] ? __debug_show_held_locks+0x1b/0x24 [<ffffffff81043660>] __schedule_bug+0x7e/0x83 [<ffffffff8136ede9>] schedule+0xce/0x838 [<ffffffff810d7972>] ? fsnotify_access+0x5f/0x67 [<ffffffff810112d0>] ? sysret_careful+0xb/0x37 [<ffffffff8106be9c>] ? trace_hardirqs_on_caller+0x1f/0x153 [<ffffffff8137127b>] ? trace_hardirqs_on_thunk+0x3a/0x3f [<ffffffff810112f6>] sysret_careful+0x31/0x37 read[1985]: segfault at 7fffc479bfe8 ip 0000003e7420a180 sp 00007fffc479bfa0 error 6 Kernel panic - not syncing: Aiee, killing interrupt handler! udp_seq_stop() tries to unlock not yet locked spinlock. The lock was lost during splitting global udp_hash_lock to subsequent spinlocks. Signed-off by: Vitaly Mayatskikh <v.mayatskih@gmail.com> Acked-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-23 22:22:33 +00:00
spin_unlock_bh(&state->udp_table->hash[state->bucket].lock);
return udp_get_first(seq, state->bucket + 1);
}
return sk;
}
static struct sock *udp_get_idx(struct seq_file *seq, loff_t pos)
{
struct sock *sk = udp_get_first(seq, 0);
if (sk)
while (pos && (sk = udp_get_next(seq, sk)) != NULL)
--pos;
return pos ? NULL : sk;
}
static void *udp_seq_start(struct seq_file *seq, loff_t *pos)
{
udp: Wrong locking code in udp seq_file infrastructure Reading zero bytes from /proc/net/udp or other similar files which use the same seq_file udp infrastructure panics kernel in that way: ===================================== [ BUG: bad unlock balance detected! ] ------------------------------------- read/1985 is trying to release lock (&table->hash[i].lock) at: [<ffffffff81321d83>] udp_seq_stop+0x27/0x29 but there are no more locks to release! other info that might help us debug this: 1 lock held by read/1985: #0: (&p->lock){--..}, at: [<ffffffff810eefb6>] seq_read+0x38/0x348 stack backtrace: Pid: 1985, comm: read Not tainted 2.6.29-rc8 #9 Call Trace: [<ffffffff81321d83>] ? udp_seq_stop+0x27/0x29 [<ffffffff8106dab9>] print_unlock_inbalance_bug+0xd6/0xe1 [<ffffffff8106db62>] lock_release_non_nested+0x9e/0x1c6 [<ffffffff810ef030>] ? seq_read+0xb2/0x348 [<ffffffff8106bdba>] ? mark_held_locks+0x68/0x86 [<ffffffff81321d83>] ? udp_seq_stop+0x27/0x29 [<ffffffff8106dde7>] lock_release+0x15d/0x189 [<ffffffff8137163c>] _spin_unlock_bh+0x1e/0x34 [<ffffffff81321d83>] udp_seq_stop+0x27/0x29 [<ffffffff810ef239>] seq_read+0x2bb/0x348 [<ffffffff810eef7e>] ? seq_read+0x0/0x348 [<ffffffff8111aedd>] proc_reg_read+0x90/0xaf [<ffffffff810d878f>] vfs_read+0xa6/0x103 [<ffffffff8106bfac>] ? trace_hardirqs_on_caller+0x12f/0x153 [<ffffffff810d88a2>] sys_read+0x45/0x69 [<ffffffff8101123a>] system_call_fastpath+0x16/0x1b BUG: scheduling while atomic: read/1985/0xffffff00 INFO: lockdep is turned off. Modules linked in: cpufreq_ondemand acpi_cpufreq freq_table dm_multipath kvm ppdev snd_hda_codec_analog snd_hda_intel snd_hda_codec snd_hwdep snd_seq_dummy snd_seq_oss snd_seq_midi_event arc4 snd_s eq ecb thinkpad_acpi snd_seq_device iwl3945 hwmon sdhci_pci snd_pcm_oss sdhci rfkill mmc_core snd_mixer_oss i2c_i801 mac80211 yenta_socket ricoh_mmc i2c_core iTCO_wdt snd_pcm iTCO_vendor_support rs rc_nonstatic snd_timer snd lib80211 cfg80211 soundcore snd_page_alloc video parport_pc output parport e1000e [last unloaded: scsi_wait_scan] Pid: 1985, comm: read Not tainted 2.6.29-rc8 #9 Call Trace: [<ffffffff8106b456>] ? __debug_show_held_locks+0x1b/0x24 [<ffffffff81043660>] __schedule_bug+0x7e/0x83 [<ffffffff8136ede9>] schedule+0xce/0x838 [<ffffffff810d7972>] ? fsnotify_access+0x5f/0x67 [<ffffffff810112d0>] ? sysret_careful+0xb/0x37 [<ffffffff8106be9c>] ? trace_hardirqs_on_caller+0x1f/0x153 [<ffffffff8137127b>] ? trace_hardirqs_on_thunk+0x3a/0x3f [<ffffffff810112f6>] sysret_careful+0x31/0x37 read[1985]: segfault at 7fffc479bfe8 ip 0000003e7420a180 sp 00007fffc479bfa0 error 6 Kernel panic - not syncing: Aiee, killing interrupt handler! udp_seq_stop() tries to unlock not yet locked spinlock. The lock was lost during splitting global udp_hash_lock to subsequent spinlocks. Signed-off by: Vitaly Mayatskikh <v.mayatskih@gmail.com> Acked-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-23 22:22:33 +00:00
struct udp_iter_state *state = seq->private;
state->bucket = MAX_UDP_PORTS;
udp: Wrong locking code in udp seq_file infrastructure Reading zero bytes from /proc/net/udp or other similar files which use the same seq_file udp infrastructure panics kernel in that way: ===================================== [ BUG: bad unlock balance detected! ] ------------------------------------- read/1985 is trying to release lock (&table->hash[i].lock) at: [<ffffffff81321d83>] udp_seq_stop+0x27/0x29 but there are no more locks to release! other info that might help us debug this: 1 lock held by read/1985: #0: (&p->lock){--..}, at: [<ffffffff810eefb6>] seq_read+0x38/0x348 stack backtrace: Pid: 1985, comm: read Not tainted 2.6.29-rc8 #9 Call Trace: [<ffffffff81321d83>] ? udp_seq_stop+0x27/0x29 [<ffffffff8106dab9>] print_unlock_inbalance_bug+0xd6/0xe1 [<ffffffff8106db62>] lock_release_non_nested+0x9e/0x1c6 [<ffffffff810ef030>] ? seq_read+0xb2/0x348 [<ffffffff8106bdba>] ? mark_held_locks+0x68/0x86 [<ffffffff81321d83>] ? udp_seq_stop+0x27/0x29 [<ffffffff8106dde7>] lock_release+0x15d/0x189 [<ffffffff8137163c>] _spin_unlock_bh+0x1e/0x34 [<ffffffff81321d83>] udp_seq_stop+0x27/0x29 [<ffffffff810ef239>] seq_read+0x2bb/0x348 [<ffffffff810eef7e>] ? seq_read+0x0/0x348 [<ffffffff8111aedd>] proc_reg_read+0x90/0xaf [<ffffffff810d878f>] vfs_read+0xa6/0x103 [<ffffffff8106bfac>] ? trace_hardirqs_on_caller+0x12f/0x153 [<ffffffff810d88a2>] sys_read+0x45/0x69 [<ffffffff8101123a>] system_call_fastpath+0x16/0x1b BUG: scheduling while atomic: read/1985/0xffffff00 INFO: lockdep is turned off. Modules linked in: cpufreq_ondemand acpi_cpufreq freq_table dm_multipath kvm ppdev snd_hda_codec_analog snd_hda_intel snd_hda_codec snd_hwdep snd_seq_dummy snd_seq_oss snd_seq_midi_event arc4 snd_s eq ecb thinkpad_acpi snd_seq_device iwl3945 hwmon sdhci_pci snd_pcm_oss sdhci rfkill mmc_core snd_mixer_oss i2c_i801 mac80211 yenta_socket ricoh_mmc i2c_core iTCO_wdt snd_pcm iTCO_vendor_support rs rc_nonstatic snd_timer snd lib80211 cfg80211 soundcore snd_page_alloc video parport_pc output parport e1000e [last unloaded: scsi_wait_scan] Pid: 1985, comm: read Not tainted 2.6.29-rc8 #9 Call Trace: [<ffffffff8106b456>] ? __debug_show_held_locks+0x1b/0x24 [<ffffffff81043660>] __schedule_bug+0x7e/0x83 [<ffffffff8136ede9>] schedule+0xce/0x838 [<ffffffff810d7972>] ? fsnotify_access+0x5f/0x67 [<ffffffff810112d0>] ? sysret_careful+0xb/0x37 [<ffffffff8106be9c>] ? trace_hardirqs_on_caller+0x1f/0x153 [<ffffffff8137127b>] ? trace_hardirqs_on_thunk+0x3a/0x3f [<ffffffff810112f6>] sysret_careful+0x31/0x37 read[1985]: segfault at 7fffc479bfe8 ip 0000003e7420a180 sp 00007fffc479bfa0 error 6 Kernel panic - not syncing: Aiee, killing interrupt handler! udp_seq_stop() tries to unlock not yet locked spinlock. The lock was lost during splitting global udp_hash_lock to subsequent spinlocks. Signed-off by: Vitaly Mayatskikh <v.mayatskih@gmail.com> Acked-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-23 22:22:33 +00:00
return *pos ? udp_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
}
static void *udp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
struct sock *sk;
if (v == SEQ_START_TOKEN)
sk = udp_get_idx(seq, 0);
else
sk = udp_get_next(seq, v);
++*pos;
return sk;
}
static void udp_seq_stop(struct seq_file *seq, void *v)
{
struct udp_iter_state *state = seq->private;
if (state->bucket <= state->udp_table->mask)
spin_unlock_bh(&state->udp_table->hash[state->bucket].lock);
}
int udp_seq_open(struct inode *inode, struct file *file)
{
struct udp_seq_afinfo *afinfo = PDE(inode)->data;
struct udp_iter_state *s;
int err;
err = seq_open_net(inode, file, &afinfo->seq_ops,
sizeof(struct udp_iter_state));
if (err < 0)
return err;
s = ((struct seq_file *)file->private_data)->private;
s->family = afinfo->family;
s->udp_table = afinfo->udp_table;
return err;
}
EXPORT_SYMBOL(udp_seq_open);
/* ------------------------------------------------------------------------ */
int udp_proc_register(struct net *net, struct udp_seq_afinfo *afinfo)
{
struct proc_dir_entry *p;
int rc = 0;
afinfo->seq_ops.start = udp_seq_start;
afinfo->seq_ops.next = udp_seq_next;
afinfo->seq_ops.stop = udp_seq_stop;
p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
afinfo->seq_fops, afinfo);
if (!p)
rc = -ENOMEM;
return rc;
}
EXPORT_SYMBOL(udp_proc_register);
void udp_proc_unregister(struct net *net, struct udp_seq_afinfo *afinfo)
{
proc_net_remove(net, afinfo->name);
}
EXPORT_SYMBOL(udp_proc_unregister);
/* ------------------------------------------------------------------------ */
static void udp4_format_sock(struct sock *sp, struct seq_file *f,
int bucket, int *len)
{
struct inet_sock *inet = inet_sk(sp);
__be32 dest = inet->inet_daddr;
__be32 src = inet->inet_rcv_saddr;
__u16 destp = ntohs(inet->inet_dport);
__u16 srcp = ntohs(inet->inet_sport);
seq_printf(f, "%5d: %08X:%04X %08X:%04X"
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 12:17:35 +00:00
" %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d%n",
bucket, src, srcp, dest, destp, sp->sk_state,
sk_wmem_alloc_get(sp),
sk_rmem_alloc_get(sp),
0, 0L, 0,
from_kuid_munged(seq_user_ns(f), sock_i_uid(sp)),
0, sock_i_ino(sp),
atomic_read(&sp->sk_refcnt), sp,
atomic_read(&sp->sk_drops), len);
}
int udp4_seq_show(struct seq_file *seq, void *v)
{
if (v == SEQ_START_TOKEN)
seq_printf(seq, "%-127s\n",
" sl local_address rem_address st tx_queue "
"rx_queue tr tm->when retrnsmt uid timeout "
"inode ref pointer drops");
else {
struct udp_iter_state *state = seq->private;
int len;
udp4_format_sock(v, seq, state->bucket, &len);
seq_printf(seq, "%*s\n", 127 - len, "");
}
return 0;
}
static const struct file_operations udp_afinfo_seq_fops = {
.owner = THIS_MODULE,
.open = udp_seq_open,
.read = seq_read,
.llseek = seq_lseek,
.release = seq_release_net
};
/* ------------------------------------------------------------------------ */
static struct udp_seq_afinfo udp4_seq_afinfo = {
.name = "udp",
.family = AF_INET,
.udp_table = &udp_table,
.seq_fops = &udp_afinfo_seq_fops,
.seq_ops = {
.show = udp4_seq_show,
},
};
static int __net_init udp4_proc_init_net(struct net *net)
{
return udp_proc_register(net, &udp4_seq_afinfo);
}
static void __net_exit udp4_proc_exit_net(struct net *net)
{
udp_proc_unregister(net, &udp4_seq_afinfo);
}
static struct pernet_operations udp4_net_ops = {
.init = udp4_proc_init_net,
.exit = udp4_proc_exit_net,
};
int __init udp4_proc_init(void)
{
return register_pernet_subsys(&udp4_net_ops);
}
void udp4_proc_exit(void)
{
unregister_pernet_subsys(&udp4_net_ops);
}
#endif /* CONFIG_PROC_FS */
static __initdata unsigned long uhash_entries;
static int __init set_uhash_entries(char *str)
{
ssize_t ret;
if (!str)
return 0;
ret = kstrtoul(str, 0, &uhash_entries);
if (ret)
return 0;
if (uhash_entries && uhash_entries < UDP_HTABLE_SIZE_MIN)
uhash_entries = UDP_HTABLE_SIZE_MIN;
return 1;
}
__setup("uhash_entries=", set_uhash_entries);
void __init udp_table_init(struct udp_table *table, const char *name)
{
unsigned int i;
table->hash = alloc_large_system_hash(name,
2 * sizeof(struct udp_hslot),
uhash_entries,
21, /* one slot per 2 MB */
0,
&table->log,
&table->mask,
UDP_HTABLE_SIZE_MIN,
64 * 1024);
table->hash2 = table->hash + (table->mask + 1);
for (i = 0; i <= table->mask; i++) {
INIT_HLIST_NULLS_HEAD(&table->hash[i].head, i);
table->hash[i].count = 0;
spin_lock_init(&table->hash[i].lock);
}
for (i = 0; i <= table->mask; i++) {
INIT_HLIST_NULLS_HEAD(&table->hash2[i].head, i);
table->hash2[i].count = 0;
spin_lock_init(&table->hash2[i].lock);
}
}
void __init udp_init(void)
{
unsigned long limit;
udp_table_init(&udp_table, "UDP");
limit = nr_free_buffer_pages() / 8;
limit = max(limit, 128UL);
sysctl_udp_mem[0] = limit / 4 * 3;
sysctl_udp_mem[1] = limit;
sysctl_udp_mem[2] = sysctl_udp_mem[0] * 2;
sysctl_udp_rmem_min = SK_MEM_QUANTUM;
sysctl_udp_wmem_min = SK_MEM_QUANTUM;
}
int udp4_ufo_send_check(struct sk_buff *skb)
{
const struct iphdr *iph;
struct udphdr *uh;
if (!pskb_may_pull(skb, sizeof(*uh)))
return -EINVAL;
iph = ip_hdr(skb);
uh = udp_hdr(skb);
uh->check = ~csum_tcpudp_magic(iph->saddr, iph->daddr, skb->len,
IPPROTO_UDP, 0);
skb->csum_start = skb_transport_header(skb) - skb->head;
skb->csum_offset = offsetof(struct udphdr, check);
skb->ip_summed = CHECKSUM_PARTIAL;
return 0;
}
struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb,
netdev_features_t features)
{
struct sk_buff *segs = ERR_PTR(-EINVAL);
unsigned int mss;
int offset;
__wsum csum;
mss = skb_shinfo(skb)->gso_size;
if (unlikely(skb->len <= mss))
goto out;
if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
/* Packet is from an untrusted source, reset gso_segs. */
int type = skb_shinfo(skb)->gso_type;
if (unlikely(type & ~(SKB_GSO_UDP | SKB_GSO_DODGY) ||
!(type & (SKB_GSO_UDP))))
goto out;
skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
segs = NULL;
goto out;
}
/* Do software UFO. Complete and fill in the UDP checksum as HW cannot
* do checksum of UDP packets sent as multiple IP fragments.
*/
offset = skb_checksum_start_offset(skb);
csum = skb_checksum(skb, offset, skb->len - offset, 0);
offset += skb->csum_offset;
*(__sum16 *)(skb->data + offset) = csum_fold(csum);
skb->ip_summed = CHECKSUM_NONE;
/* Fragment the skb. IP headers of the fragments are updated in
* inet_gso_segment()
*/
segs = skb_segment(skb, features);
out:
return segs;
}