2005-04-17 02:20:36 +04:00
/*
* 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 .
*
* Implementation of the Transmission Control Protocol ( TCP ) .
*
2005-05-06 03:16:16 +04:00
* Authors : Ross Biro
2005-04-17 02:20:36 +04:00
* Fred N . van Kempen , < waltje @ uWalt . NL . Mugnet . ORG >
* Mark Evans , < evansmp @ uhura . aston . ac . uk >
* Corey Minyard < wf - rch ! minyard @ relay . EU . net >
* Florian La Roche , < flla @ stud . uni - sb . de >
* Charles Hedrick , < hedrick @ klinzhai . rutgers . edu >
* Linus Torvalds , < torvalds @ cs . helsinki . fi >
* Alan Cox , < gw4pts @ gw4pts . ampr . org >
* Matthew Dillon , < dillon @ apollo . west . oic . com >
* Arnt Gulbrandsen , < agulbra @ nvg . unit . no >
* Jorge Cwik , < jorge @ laser . satlink . net >
*/
# include <linux/mm.h>
# include <linux/module.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 11:04:11 +03:00
# include <linux/slab.h>
2005-04-17 02:20:36 +04:00
# include <linux/sysctl.h>
# include <linux/workqueue.h>
# include <net/tcp.h>
# include <net/inet_common.h>
# include <net/xfrm.h>
2009-11-21 22:22:25 +03:00
int sysctl_tcp_syncookies __read_mostly = 1 ;
2008-02-08 08:49:26 +03:00
EXPORT_SYMBOL ( sysctl_tcp_syncookies ) ;
2006-09-23 01:15:41 +04:00
int sysctl_tcp_abort_on_overflow __read_mostly ;
2005-04-17 02:20:36 +04:00
2005-08-10 07:44:40 +04:00
struct inet_timewait_death_row tcp_death_row = {
. sysctl_max_tw_buckets = NR_FILE * 2 ,
. period = TCP_TIMEWAIT_LEN / INET_TWDR_TWKILL_SLOTS ,
2006-07-03 11:24:34 +04:00
. death_lock = __SPIN_LOCK_UNLOCKED ( tcp_death_row . death_lock ) ,
2005-08-10 07:44:40 +04:00
. hashinfo = & tcp_hashinfo ,
. tw_timer = TIMER_INITIALIZER ( inet_twdr_hangman , 0 ,
( unsigned long ) & tcp_death_row ) ,
. twkill_work = __WORK_INITIALIZER ( tcp_death_row . twkill_work ,
2006-11-22 17:55:48 +03:00
inet_twdr_twkill_work ) ,
2005-08-10 07:44:40 +04:00
/* Short-time timewait calendar */
. twcal_hand = - 1 ,
. twcal_timer = TIMER_INITIALIZER ( inet_twdr_twcal_tick , 0 ,
( unsigned long ) & tcp_death_row ) ,
} ;
EXPORT_SYMBOL_GPL ( tcp_death_row ) ;
2012-05-17 03:15:34 +04:00
static bool tcp_in_window ( u32 seq , u32 end_seq , u32 s_win , u32 e_win )
2005-04-17 02:20:36 +04:00
{
if ( seq = = s_win )
2012-05-17 03:15:34 +04:00
return true ;
2005-04-17 02:20:36 +04:00
if ( after ( end_seq , s_win ) & & before ( seq , e_win ) )
2012-05-17 03:15:34 +04:00
return true ;
2010-09-23 00:43:57 +04:00
return seq = = e_win & & seq = = end_seq ;
2005-04-17 02:20:36 +04:00
}
2007-02-09 17:24:47 +03:00
/*
2005-04-17 02:20:36 +04:00
* * Main purpose of TIME - WAIT state is to close connection gracefully ,
* when one of ends sits in LAST - ACK or CLOSING retransmitting FIN
* ( and , probably , tail of data ) and one or more our ACKs are lost .
* * What is TIME - WAIT timeout ? It is associated with maximal packet
* lifetime in the internet , which results in wrong conclusion , that
* it is set to catch " old duplicate segments " wandering out of their path .
* It is not quite correct . This timeout is calculated so that it exceeds
* maximal retransmission timeout enough to allow to lose one ( or more )
* segments sent by peer and our ACKs . This time may be calculated from RTO .
* * When TIME - WAIT socket receives RST , it means that another end
* finally closed and we are allowed to kill TIME - WAIT too .
* * Second purpose of TIME - WAIT is catching old duplicate segments .
* Well , certainly it is pure paranoia , but if we load TIME - WAIT
* with this semantics , we MUST NOT kill TIME - WAIT state with RSTs .
* * If we invented some more clever way to catch duplicates
* ( f . e . based on PAWS ) , we could truncate TIME - WAIT to several RTOs .
*
* The algorithm below is based on FORMAL INTERPRETATION of RFCs .
* When you compare it to RFCs , please , read section SEGMENT ARRIVES
* from the very beginning .
*
* NOTE . With recycling ( and later with fin - wait - 2 ) TW bucket
* is _not_ stateless . It means , that strictly speaking we must
* spinlock it . I do not want ! Well , probability of misbehaviour
* is ridiculously low and , seems , we could use some mb ( ) tricks
* to avoid misread sequence numbers , states etc . - - ANK
2012-09-19 18:46:06 +04:00
*
* We don ' t need to initialize tmp_out . sack_ok as we don ' t use the results
2005-04-17 02:20:36 +04:00
*/
enum tcp_tw_status
2005-08-10 07:09:30 +04:00
tcp_timewait_state_process ( struct inet_timewait_sock * tw , struct sk_buff * skb ,
const struct tcphdr * th )
2005-04-17 02:20:36 +04:00
{
struct tcp_options_received tmp_opt ;
2009-12-02 21:25:27 +03:00
struct tcp_timewait_sock * tcptw = tcp_twsk ( ( struct sock * ) tw ) ;
2012-05-17 03:15:34 +04:00
bool paws_reject = false ;
2005-04-17 02:20:36 +04:00
tcp: Revert per-route SACK/DSACK/TIMESTAMP changes.
It creates a regression, triggering badness for SYN_RECV
sockets, for example:
[19148.022102] Badness at net/ipv4/inet_connection_sock.c:293
[19148.022570] NIP: c02a0914 LR: c02a0904 CTR: 00000000
[19148.023035] REGS: eeecbd30 TRAP: 0700 Not tainted (2.6.32)
[19148.023496] MSR: 00029032 <EE,ME,CE,IR,DR> CR: 24002442 XER: 00000000
[19148.024012] TASK = eee9a820[1756] 'privoxy' THREAD: eeeca000
This is likely caused by the change in the 'estab' parameter
passed to tcp_parse_options() when invoked by the functions
in net/ipv4/tcp_minisocks.c
But even if that is fixed, the ->conn_request() changes made in
this patch series is fundamentally wrong. They try to use the
listening socket's 'dst' to probe the route settings. The
listening socket doesn't even have a route, and you can't
get the right route (the child request one) until much later
after we setup all of the state, and it must be done by hand.
This stuff really isn't ready, so the best thing to do is a
full revert. This reverts the following commits:
f55017a93f1a74d50244b1254b9a2bd7ac9bbf7d
022c3f7d82f0f1c68018696f2f027b87b9bb45c2
1aba721eba1d84a2defce45b950272cee1e6c72a
cda42ebd67ee5fdf09d7057b5a4584d36fe8a335
345cda2fd695534be5a4494f1b59da9daed33663
dc343475ed062e13fc260acccaab91d7d80fd5b2
05eaade2782fb0c90d3034fd7a7d5a16266182bb
6a2a2d6bf8581216e08be15fcb563cfd6c430e1e
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-12-16 07:56:42 +03:00
tmp_opt . saw_tstamp = 0 ;
2005-08-10 07:09:30 +04:00
if ( th - > doff > ( sizeof ( * th ) > > 2 ) & & tcptw - > tw_ts_recent_stamp ) {
2013-03-17 12:23:34 +04:00
tcp_parse_options ( skb , & tmp_opt , 0 , NULL ) ;
2005-04-17 02:20:36 +04:00
if ( tmp_opt . saw_tstamp ) {
2013-02-11 09:50:19 +04:00
tmp_opt . rcv_tsecr - = tcptw - > tw_ts_offset ;
2005-08-10 07:09:30 +04:00
tmp_opt . ts_recent = tcptw - > tw_ts_recent ;
tmp_opt . ts_recent_stamp = tcptw - > tw_ts_recent_stamp ;
2009-03-14 17:23:03 +03:00
paws_reject = tcp_paws_reject ( & tmp_opt , th - > rst ) ;
2005-04-17 02:20:36 +04:00
}
}
if ( tw - > tw_substate = = TCP_FIN_WAIT2 ) {
/* Just repeat all the checks of tcp_rcv_state_process() */
/* Out of window, send ACK */
if ( paws_reject | |
! tcp_in_window ( TCP_SKB_CB ( skb ) - > seq , TCP_SKB_CB ( skb ) - > end_seq ,
2005-08-10 07:09:30 +04:00
tcptw - > tw_rcv_nxt ,
tcptw - > tw_rcv_nxt + tcptw - > tw_rcv_wnd ) )
2005-04-17 02:20:36 +04:00
return TCP_TW_ACK ;
if ( th - > rst )
goto kill ;
2005-08-10 07:09:30 +04:00
if ( th - > syn & & ! before ( TCP_SKB_CB ( skb ) - > seq , tcptw - > tw_rcv_nxt ) )
2005-04-17 02:20:36 +04:00
goto kill_with_rst ;
/* Dup ACK? */
2009-06-25 02:29:31 +04:00
if ( ! th - > ack | |
! after ( TCP_SKB_CB ( skb ) - > end_seq , tcptw - > tw_rcv_nxt ) | |
2005-04-17 02:20:36 +04:00
TCP_SKB_CB ( skb ) - > end_seq = = TCP_SKB_CB ( skb ) - > seq ) {
2005-08-10 07:09:30 +04:00
inet_twsk_put ( tw ) ;
2005-04-17 02:20:36 +04:00
return TCP_TW_SUCCESS ;
}
/* New data or FIN. If new data arrive after half-duplex close,
* reset .
*/
if ( ! th - > fin | |
2005-08-10 07:09:30 +04:00
TCP_SKB_CB ( skb ) - > end_seq ! = tcptw - > tw_rcv_nxt + 1 ) {
2005-04-17 02:20:36 +04:00
kill_with_rst :
2005-08-10 07:44:40 +04:00
inet_twsk_deschedule ( tw , & tcp_death_row ) ;
2005-08-10 07:09:30 +04:00
inet_twsk_put ( tw ) ;
2005-04-17 02:20:36 +04:00
return TCP_TW_RST ;
}
/* FIN arrived, enter true time-wait state. */
2005-08-10 07:09:30 +04:00
tw - > tw_substate = TCP_TIME_WAIT ;
tcptw - > tw_rcv_nxt = TCP_SKB_CB ( skb ) - > end_seq ;
2005-04-17 02:20:36 +04:00
if ( tmp_opt . saw_tstamp ) {
2007-03-05 03:12:44 +03:00
tcptw - > tw_ts_recent_stamp = get_seconds ( ) ;
2005-08-10 07:09:30 +04:00
tcptw - > tw_ts_recent = tmp_opt . rcv_tsval ;
2005-04-17 02:20:36 +04:00
}
2010-12-02 05:09:13 +03:00
if ( tcp_death_row . sysctl_tw_recycle & &
tcptw - > tw_ts_recent_stamp & &
tcp_tw_remember_stamp ( tw ) )
2005-08-10 07:45:03 +04:00
inet_twsk_schedule ( tw , & tcp_death_row , tw - > tw_timeout ,
TCP_TIMEWAIT_LEN ) ;
2005-04-17 02:20:36 +04:00
else
2005-08-10 07:45:03 +04:00
inet_twsk_schedule ( tw , & tcp_death_row , TCP_TIMEWAIT_LEN ,
TCP_TIMEWAIT_LEN ) ;
2005-04-17 02:20:36 +04:00
return TCP_TW_ACK ;
}
/*
* Now real TIME - WAIT state .
*
* RFC 1122 :
* " When a connection is [...] on TIME-WAIT state [...]
* [ a TCP ] MAY accept a new SYN from the remote TCP to
* reopen the connection directly , if it :
2007-02-09 17:24:47 +03:00
*
2005-04-17 02:20:36 +04:00
* ( 1 ) assigns its initial sequence number for the new
* connection to be larger than the largest sequence
* number it used on the previous connection incarnation ,
* and
*
2007-02-09 17:24:47 +03:00
* ( 2 ) returns to TIME - WAIT state if the SYN turns out
2005-04-17 02:20:36 +04:00
* to be an old duplicate " .
*/
if ( ! paws_reject & &
2005-08-10 07:09:30 +04:00
( TCP_SKB_CB ( skb ) - > seq = = tcptw - > tw_rcv_nxt & &
2005-04-17 02:20:36 +04:00
( TCP_SKB_CB ( skb ) - > seq = = TCP_SKB_CB ( skb ) - > end_seq | | th - > rst ) ) ) {
/* In window segment, it may be only reset or bare ack. */
if ( th - > rst ) {
2005-11-11 04:13:47 +03:00
/* This is TIME_WAIT assassination, in two flavors.
2005-04-17 02:20:36 +04:00
* Oh well . . . nobody has a sufficient solution to this
* protocol bug yet .
*/
if ( sysctl_tcp_rfc1337 = = 0 ) {
kill :
2005-08-10 07:44:40 +04:00
inet_twsk_deschedule ( tw , & tcp_death_row ) ;
2005-08-10 07:09:30 +04:00
inet_twsk_put ( tw ) ;
2005-04-17 02:20:36 +04:00
return TCP_TW_SUCCESS ;
}
}
2005-08-10 07:45:03 +04:00
inet_twsk_schedule ( tw , & tcp_death_row , TCP_TIMEWAIT_LEN ,
TCP_TIMEWAIT_LEN ) ;
2005-04-17 02:20:36 +04:00
if ( tmp_opt . saw_tstamp ) {
2005-08-10 07:09:30 +04:00
tcptw - > tw_ts_recent = tmp_opt . rcv_tsval ;
2007-03-05 03:12:44 +03:00
tcptw - > tw_ts_recent_stamp = get_seconds ( ) ;
2005-04-17 02:20:36 +04:00
}
2005-08-10 07:09:30 +04:00
inet_twsk_put ( tw ) ;
2005-04-17 02:20:36 +04:00
return TCP_TW_SUCCESS ;
}
/* Out of window segment.
All the segments are ACKed immediately .
The only exception is new SYN . We accept it , if it is
not old duplicate and we are not in danger to be killed
by delayed old duplicates . RFC check is that it has
newer sequence number works at rates < 40 Mbit / sec .
However , if paws works , it is reliable AND even more ,
we even may relax silly seq space cutoff .
RED - PEN : we violate main RFC requirement , if this SYN will appear
old duplicate ( i . e . we receive RST in reply to SYN - ACK ) ,
we must return socket to time - wait state . It is not good ,
but not fatal yet .
*/
if ( th - > syn & & ! th - > rst & & ! th - > ack & & ! paws_reject & &
2005-08-10 07:09:30 +04:00
( after ( TCP_SKB_CB ( skb ) - > seq , tcptw - > tw_rcv_nxt ) | |
( tmp_opt . saw_tstamp & &
( s32 ) ( tcptw - > tw_ts_recent - tmp_opt . rcv_tsval ) < 0 ) ) ) {
u32 isn = tcptw - > tw_snd_nxt + 65535 + 2 ;
2005-04-17 02:20:36 +04:00
if ( isn = = 0 )
isn + + ;
2014-09-06 02:33:32 +04:00
TCP_SKB_CB ( skb ) - > tcp_tw_isn = isn ;
2005-04-17 02:20:36 +04:00
return TCP_TW_SYN ;
}
if ( paws_reject )
2008-07-17 07:31:16 +04:00
NET_INC_STATS_BH ( twsk_net ( tw ) , LINUX_MIB_PAWSESTABREJECTED ) ;
2005-04-17 02:20:36 +04:00
2007-03-09 07:45:19 +03:00
if ( ! th - > rst ) {
2005-04-17 02:20:36 +04:00
/* In this case we must reset the TIMEWAIT timer.
*
* If it is ACKless SYN it may be both old duplicate
* and new good SYN with random sequence number < rcv_nxt .
* Do not reschedule in the last case .
*/
if ( paws_reject | | th - > ack )
2005-08-10 07:45:03 +04:00
inet_twsk_schedule ( tw , & tcp_death_row , TCP_TIMEWAIT_LEN ,
TCP_TIMEWAIT_LEN ) ;
2005-04-17 02:20:36 +04:00
/* Send ACK. Note, we do not put the bucket,
* it will be released by caller .
*/
return TCP_TW_ACK ;
}
2005-08-10 07:09:30 +04:00
inet_twsk_put ( tw ) ;
2005-04-17 02:20:36 +04:00
return TCP_TW_SUCCESS ;
}
2010-07-10 01:22:10 +04:00
EXPORT_SYMBOL ( tcp_timewait_state_process ) ;
2005-04-17 02:20:36 +04:00
2007-02-09 17:24:47 +03:00
/*
2005-04-17 02:20:36 +04:00
* Move a socket to time - wait or dead fin - wait - 2 state .
2007-02-09 17:24:47 +03:00
*/
2005-04-17 02:20:36 +04:00
void tcp_time_wait ( struct sock * sk , int state , int timeo )
{
2005-08-10 07:09:30 +04:00
struct inet_timewait_sock * tw = NULL ;
2005-12-14 10:15:52 +03:00
const struct inet_connection_sock * icsk = inet_csk ( sk ) ;
2005-08-10 07:09:30 +04:00
const struct tcp_sock * tp = tcp_sk ( sk ) ;
2012-05-17 03:15:34 +04:00
bool recycle_ok = false ;
2005-04-17 02:20:36 +04:00
2012-07-10 14:27:56 +04:00
if ( tcp_death_row . sysctl_tw_recycle & & tp - > rx_opt . ts_recent_stamp )
2010-11-30 00:37:14 +03:00
recycle_ok = tcp_remember_stamp ( sk ) ;
2005-04-17 02:20:36 +04:00
2005-08-10 07:44:40 +04:00
if ( tcp_death_row . tw_count < tcp_death_row . sysctl_max_tw_buckets )
2005-08-10 07:09:59 +04:00
tw = inet_twsk_alloc ( sk , state ) ;
2005-04-17 02:20:36 +04:00
2005-08-10 07:09:30 +04:00
if ( tw ! = NULL ) {
struct tcp_timewait_sock * tcptw = tcp_twsk ( ( struct sock * ) tw ) ;
2005-08-10 07:10:42 +04:00
const int rto = ( icsk - > icsk_rto < < 2 ) - ( icsk - > icsk_rto > > 1 ) ;
2012-06-10 01:56:12 +04:00
struct inet_sock * inet = inet_sk ( sk ) ;
2005-08-10 07:09:30 +04:00
2012-06-10 01:56:12 +04:00
tw - > tw_transparent = inet - > transparent ;
2005-04-17 02:20:36 +04:00
tw - > tw_rcv_wscale = tp - > rx_opt . rcv_wscale ;
2005-08-10 07:09:30 +04:00
tcptw - > tw_rcv_nxt = tp - > rcv_nxt ;
tcptw - > tw_snd_nxt = tp - > snd_nxt ;
tcptw - > tw_rcv_wnd = tcp_receive_window ( tp ) ;
tcptw - > tw_ts_recent = tp - > rx_opt . ts_recent ;
tcptw - > tw_ts_recent_stamp = tp - > rx_opt . ts_recent_stamp ;
2013-02-11 09:50:17 +04:00
tcptw - > tw_ts_offset = tp - > tsoffset ;
2005-04-17 02:20:36 +04:00
2011-12-10 13:48:31 +04:00
# if IS_ENABLED(CONFIG_IPV6)
2005-04-17 02:20:36 +04:00
if ( tw - > tw_family = = PF_INET6 ) {
struct ipv6_pinfo * np = inet6_sk ( sk ) ;
ipv6: make lookups simpler and faster
TCP listener refactoring, part 4 :
To speed up inet lookups, we moved IPv4 addresses from inet to struct
sock_common
Now is time to do the same for IPv6, because it permits us to have fast
lookups for all kind of sockets, including upcoming SYN_RECV.
Getting IPv6 addresses in TCP lookups currently requires two extra cache
lines, plus a dereference (and memory stall).
inet6_sk(sk) does the dereference of inet_sk(__sk)->pinet6
This patch is way bigger than its IPv4 counter part, because for IPv4,
we could add aliases (inet_daddr, inet_rcv_saddr), while on IPv6,
it's not doable easily.
inet6_sk(sk)->daddr becomes sk->sk_v6_daddr
inet6_sk(sk)->rcv_saddr becomes sk->sk_v6_rcv_saddr
And timewait socket also have tw->tw_v6_daddr & tw->tw_v6_rcv_saddr
at the same offset.
We get rid of INET6_TW_MATCH() as INET6_MATCH() is now the generic
macro.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2013-10-04 02:42:29 +04:00
tw - > tw_v6_daddr = sk - > sk_v6_daddr ;
tw - > tw_v6_rcv_saddr = sk - > sk_v6_rcv_saddr ;
2011-10-27 08:44:35 +04:00
tw - > tw_tclass = np - > tclass ;
2014-01-16 20:21:22 +04:00
tw - > tw_flowlabel = np - > flow_label > > 12 ;
2014-06-27 19:36:16 +04:00
tw - > tw_ipv6only = sk - > sk_ipv6only ;
2005-08-10 07:09:59 +04:00
}
2005-04-17 02:20:36 +04:00
# endif
2006-11-15 06:07:45 +03:00
# ifdef CONFIG_TCP_MD5SIG
/*
* The timewait bucket does not have the key DB from the
* sock structure . We just make a quick copy of the
* md5 key being used ( if indeed we are using one )
* so the timewait ack generating code has the key .
*/
do {
struct tcp_md5sig_key * key ;
2012-01-31 09:18:33 +04:00
tcptw - > tw_md5_key = NULL ;
2006-11-15 06:07:45 +03:00
key = tp - > af_specific - > md5_lookup ( sk , sk ) ;
if ( key ! = NULL ) {
2012-01-31 09:18:33 +04:00
tcptw - > tw_md5_key = kmemdup ( key , sizeof ( * key ) , GFP_ATOMIC ) ;
2013-05-20 10:52:26 +04:00
if ( tcptw - > tw_md5_key & & ! tcp_alloc_md5sig_pool ( ) )
2006-11-15 06:07:45 +03:00
BUG ( ) ;
}
2007-03-09 07:45:19 +03:00
} while ( 0 ) ;
2006-11-15 06:07:45 +03:00
# endif
2005-04-17 02:20:36 +04:00
/* Linkage updates. */
2005-08-10 07:09:46 +04:00
__inet_twsk_hashdance ( tw , sk , & tcp_hashinfo ) ;
2005-04-17 02:20:36 +04:00
/* Get the TIME_WAIT timeout firing. */
if ( timeo < rto )
timeo = rto ;
if ( recycle_ok ) {
tw - > tw_timeout = rto ;
} else {
tw - > tw_timeout = TCP_TIMEWAIT_LEN ;
if ( state = = TCP_TIME_WAIT )
timeo = TCP_TIMEWAIT_LEN ;
}
2005-08-10 07:45:03 +04:00
inet_twsk_schedule ( tw , & tcp_death_row , timeo ,
TCP_TIMEWAIT_LEN ) ;
2005-08-10 07:09:30 +04:00
inet_twsk_put ( tw ) ;
2005-04-17 02:20:36 +04:00
} else {
/* Sorry, if we're out of memory, just CLOSE this
* socket up . We ' ve got bigger problems than
* non - graceful socket closings .
*/
2010-12-08 23:16:33 +03:00
NET_INC_STATS_BH ( sock_net ( sk ) , LINUX_MIB_TCPTIMEWAITOVERFLOW ) ;
2005-04-17 02:20:36 +04:00
}
tcp_update_metrics ( sk ) ;
tcp_done ( sk ) ;
}
2006-11-15 06:07:45 +03:00
void tcp_twsk_destructor ( struct sock * sk )
{
2012-07-10 14:27:56 +04:00
# ifdef CONFIG_TCP_MD5SIG
2006-11-15 06:53:22 +03:00
struct tcp_timewait_sock * twsk = tcp_twsk ( sk ) ;
2012-06-10 01:56:12 +04:00
2013-05-20 10:52:26 +04:00
if ( twsk - > tw_md5_key )
2012-01-31 09:18:33 +04:00
kfree_rcu ( twsk - > tw_md5_key , rcu ) ;
2006-11-15 06:07:45 +03:00
# endif
}
EXPORT_SYMBOL_GPL ( tcp_twsk_destructor ) ;
2014-05-12 07:22:11 +04:00
void tcp_openreq_init_rwin ( struct request_sock * req ,
struct sock * sk , struct dst_entry * dst )
{
struct inet_request_sock * ireq = inet_rsk ( req ) ;
struct tcp_sock * tp = tcp_sk ( sk ) ;
__u8 rcv_wscale ;
int mss = dst_metric_advmss ( dst ) ;
if ( tp - > rx_opt . user_mss & & tp - > rx_opt . user_mss < mss )
mss = tp - > rx_opt . user_mss ;
/* Set this up on the first call only */
req - > window_clamp = tp - > window_clamp ? : dst_metric ( dst , RTAX_WINDOW ) ;
/* limit the window selection if the user enforce a smaller rx buffer */
if ( sk - > sk_userlocks & SOCK_RCVBUF_LOCK & &
( req - > window_clamp > tcp_full_space ( sk ) | | req - > window_clamp = = 0 ) )
req - > window_clamp = tcp_full_space ( sk ) ;
/* tcp_full_space because it is guaranteed to be the first packet */
tcp_select_initial_window ( tcp_full_space ( sk ) ,
mss - ( ireq - > tstamp_ok ? TCPOLEN_TSTAMP_ALIGNED : 0 ) ,
& req - > rcv_wnd ,
& req - > window_clamp ,
ireq - > wscale_ok ,
& rcv_wscale ,
dst_metric ( dst , RTAX_INITRWND ) ) ;
ireq - > rcv_wscale = rcv_wscale ;
}
EXPORT_SYMBOL ( tcp_openreq_init_rwin ) ;
2014-09-29 15:08:30 +04:00
static void tcp_ecn_openreq_child ( struct tcp_sock * tp ,
const struct request_sock * req )
2007-05-27 13:04:16 +04:00
{
tp - > ecn_flags = inet_rsk ( req ) - > ecn_ok ? TCP_ECN_OK : 0 ;
}
net: tcp: add per route congestion control
This work adds the possibility to define a per route/destination
congestion control algorithm. Generally, this opens up the possibility
for a machine with different links to enforce specific congestion
control algorithms with optimal strategies for each of them based
on their network characteristics, even transparently for a single
application listening on all links.
For our specific use case, this additionally facilitates deployment
of DCTCP, for example, applications can easily serve internal
traffic/dsts in DCTCP and external one with CUBIC. Other scenarios
would also allow for utilizing e.g. long living, low priority
background flows for certain destinations/routes while still being
able for normal traffic to utilize the default congestion control
algorithm. We also thought about a per netns setting (where different
defaults are possible), but given its actually a link specific
property, we argue that a per route/destination setting is the most
natural and flexible.
The administrator can utilize this through ip-route(8) by appending
"congctl [lock] <name>", where <name> denotes the name of a
congestion control algorithm and the optional lock parameter allows
to enforce the given algorithm so that applications in user space
would not be allowed to overwrite that algorithm for that destination.
The dst metric lookups are being done when a dst entry is already
available in order to avoid a costly lookup and still before the
algorithms are being initialized, thus overhead is very low when the
feature is not being used. While the client side would need to drop
the current reference on the module, on server side this can actually
even be avoided as we just got a flat-copied socket clone.
Joint work with Florian Westphal.
Suggested-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-01-06 01:57:48 +03:00
void tcp_ca_openreq_child ( struct sock * sk , const struct dst_entry * dst )
{
struct inet_connection_sock * icsk = inet_csk ( sk ) ;
u32 ca_key = dst_metric ( dst , RTAX_CC_ALGO ) ;
bool ca_got_dst = false ;
if ( ca_key ! = TCP_CA_UNSPEC ) {
const struct tcp_congestion_ops * ca ;
rcu_read_lock ( ) ;
ca = tcp_ca_find_key ( ca_key ) ;
if ( likely ( ca & & try_module_get ( ca - > owner ) ) ) {
icsk - > icsk_ca_dst_locked = tcp_ca_dst_locked ( dst ) ;
icsk - > icsk_ca_ops = ca ;
ca_got_dst = true ;
}
rcu_read_unlock ( ) ;
}
if ( ! ca_got_dst & & ! try_module_get ( icsk - > icsk_ca_ops - > owner ) )
tcp_assign_congestion_control ( sk ) ;
tcp_set_ca_state ( sk , TCP_CA_Open ) ;
}
EXPORT_SYMBOL_GPL ( tcp_ca_openreq_child ) ;
2005-04-17 02:20:36 +04:00
/* This is not only more efficient than what we used to do, it eliminates
* a lot of code duplication between IPv4 / IPv6 SYN recv processing . - DaveM
*
* Actually , we could lots of memory writes here . tp of listening
* socket contains all necessary default parameters .
*/
2005-06-19 09:47:21 +04:00
struct sock * tcp_create_openreq_child ( struct sock * sk , struct request_sock * req , struct sk_buff * skb )
2005-04-17 02:20:36 +04:00
{
2011-11-09 02:07:07 +04:00
struct sock * newsk = inet_csk_clone_lock ( sk , req , GFP_ATOMIC ) ;
2005-04-17 02:20:36 +04:00
2005-08-10 07:10:12 +04:00
if ( newsk ! = NULL ) {
2005-08-10 07:11:24 +04:00
const struct inet_request_sock * ireq = inet_rsk ( req ) ;
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 09:46:52 +04:00
struct tcp_request_sock * treq = tcp_rsk ( req ) ;
2007-02-28 22:05:56 +03:00
struct inet_connection_sock * newicsk = inet_csk ( newsk ) ;
TCPCT part 1d: define TCP cookie option, extend existing struct's
Data structures are carefully composed to require minimal additions.
For example, the struct tcp_options_received cookie_plus variable fits
between existing 16-bit and 8-bit variables, requiring no additional
space (taking alignment into consideration). There are no additions to
tcp_request_sock, and only 1 pointer in tcp_sock.
This is a significantly revised implementation of an earlier (year-old)
patch that no longer applies cleanly, with permission of the original
author (Adam Langley):
http://thread.gmane.org/gmane.linux.network/102586
The principle difference is using a TCP option to carry the cookie nonce,
instead of a user configured offset in the data. This is more flexible and
less subject to user configuration error. Such a cookie option has been
suggested for many years, and is also useful without SYN data, allowing
several related concepts to use the same extension option.
"Re: SYN floods (was: does history repeat itself?)", September 9, 1996.
http://www.merit.net/mail.archives/nanog/1996-09/msg00235.html
"Re: what a new TCP header might look like", May 12, 1998.
ftp://ftp.isi.edu/end2end/end2end-interest-1998.mail
These functions will also be used in subsequent patches that implement
additional features.
Requires:
TCPCT part 1a: add request_values parameter for sending SYNACK
TCPCT part 1b: generate Responder Cookie secret
TCPCT part 1c: sysctl_tcp_cookie_size, socket option TCP_COOKIE_TRANSACTIONS
Signed-off-by: William.Allen.Simpson@gmail.com
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-12-02 21:17:05 +03:00
struct tcp_sock * newtp = tcp_sk ( newsk ) ;
2005-04-17 02:20:36 +04:00
/* Now setup tcp_sock */
newtp - > pred_flags = 0 ;
TCPCT part 1d: define TCP cookie option, extend existing struct's
Data structures are carefully composed to require minimal additions.
For example, the struct tcp_options_received cookie_plus variable fits
between existing 16-bit and 8-bit variables, requiring no additional
space (taking alignment into consideration). There are no additions to
tcp_request_sock, and only 1 pointer in tcp_sock.
This is a significantly revised implementation of an earlier (year-old)
patch that no longer applies cleanly, with permission of the original
author (Adam Langley):
http://thread.gmane.org/gmane.linux.network/102586
The principle difference is using a TCP option to carry the cookie nonce,
instead of a user configured offset in the data. This is more flexible and
less subject to user configuration error. Such a cookie option has been
suggested for many years, and is also useful without SYN data, allowing
several related concepts to use the same extension option.
"Re: SYN floods (was: does history repeat itself?)", September 9, 1996.
http://www.merit.net/mail.archives/nanog/1996-09/msg00235.html
"Re: what a new TCP header might look like", May 12, 1998.
ftp://ftp.isi.edu/end2end/end2end-interest-1998.mail
These functions will also be used in subsequent patches that implement
additional features.
Requires:
TCPCT part 1a: add request_values parameter for sending SYNACK
TCPCT part 1b: generate Responder Cookie secret
TCPCT part 1c: sysctl_tcp_cookie_size, socket option TCP_COOKIE_TRANSACTIONS
Signed-off-by: William.Allen.Simpson@gmail.com
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-12-02 21:17:05 +03:00
newtp - > rcv_wup = newtp - > copied_seq =
newtp - > rcv_nxt = treq - > rcv_isn + 1 ;
newtp - > snd_sml = newtp - > snd_una =
2013-03-17 12:23:34 +04:00
newtp - > snd_nxt = newtp - > snd_up = treq - > snt_isn + 1 ;
2005-04-17 02:20:36 +04:00
tcp_prequeue_init ( newtp ) ;
tcp: TCP Small Queues
This introduce TSQ (TCP Small Queues)
TSQ goal is to reduce number of TCP packets in xmit queues (qdisc &
device queues), to reduce RTT and cwnd bias, part of the bufferbloat
problem.
sk->sk_wmem_alloc not allowed to grow above a given limit,
allowing no more than ~128KB [1] per tcp socket in qdisc/dev layers at a
given time.
TSO packets are sized/capped to half the limit, so that we have two
TSO packets in flight, allowing better bandwidth use.
As a side effect, setting the limit to 40000 automatically reduces the
standard gso max limit (65536) to 40000/2 : It can help to reduce
latencies of high prio packets, having smaller TSO packets.
This means we divert sock_wfree() to a tcp_wfree() handler, to
queue/send following frames when skb_orphan() [2] is called for the
already queued skbs.
Results on my dev machines (tg3/ixgbe nics) are really impressive,
using standard pfifo_fast, and with or without TSO/GSO.
Without reduction of nominal bandwidth, we have reduction of buffering
per bulk sender :
< 1ms on Gbit (instead of 50ms with TSO)
< 8ms on 100Mbit (instead of 132 ms)
I no longer have 4 MBytes backlogged in qdisc by a single netperf
session, and both side socket autotuning no longer use 4 Mbytes.
As skb destructor cannot restart xmit itself ( as qdisc lock might be
taken at this point ), we delegate the work to a tasklet. We use one
tasklest per cpu for performance reasons.
If tasklet finds a socket owned by the user, it sets TSQ_OWNED flag.
This flag is tested in a new protocol method called from release_sock(),
to eventually send new segments.
[1] New /proc/sys/net/ipv4/tcp_limit_output_bytes tunable
[2] skb_orphan() is usually called at TX completion time,
but some drivers call it in their start_xmit() handler.
These drivers should at least use BQL, or else a single TCP
session can still fill the whole NIC TX ring, since TSQ will
have no effect.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Dave Taht <dave.taht@bufferbloat.net>
Cc: Tom Herbert <therbert@google.com>
Cc: Matt Mathis <mattmathis@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-07-11 09:50:31 +04:00
INIT_LIST_HEAD ( & newtp - > tsq_node ) ;
2005-04-17 02:20:36 +04:00
2009-03-03 09:42:02 +03:00
tcp_init_wl ( newtp , treq - > rcv_isn ) ;
2005-04-17 02:20:36 +04:00
2014-02-27 02:02:48 +04:00
newtp - > srtt_us = 0 ;
newtp - > mdev_us = jiffies_to_usecs ( TCP_TIMEOUT_INIT ) ;
2005-08-10 07:10:42 +04:00
newicsk - > icsk_rto = TCP_TIMEOUT_INIT ;
2005-04-17 02:20:36 +04:00
newtp - > packets_out = 0 ;
newtp - > retrans_out = 0 ;
newtp - > sacked_out = 0 ;
newtp - > fackets_out = 0 ;
2009-09-15 12:30:10 +04:00
newtp - > snd_ssthresh = TCP_INFINITE_SSTHRESH ;
2012-05-02 17:30:03 +04:00
tcp_enable_early_retrans ( newtp ) ;
2013-03-11 14:00:44 +04:00
newtp - > tlp_high_seq = 0 ;
2013-07-23 03:20:45 +04:00
newtp - > lsndtime = treq - > snt_synack ;
newtp - > total_retrans = req - > num_retrans ;
2005-04-17 02:20:36 +04:00
/* So many TCP implementations out there (incorrectly) count the
* initial SYN frame in their delayed - ACK and congestion control
* algorithms that we must have the following bandaid to talk
* efficiently to them . - DaveM
*/
2011-06-08 15:08:38 +04:00
newtp - > snd_cwnd = TCP_INIT_CWND ;
2005-04-17 02:20:36 +04:00
newtp - > snd_cwnd_cnt = 0 ;
tcp_init_xmit_timers ( newsk ) ;
2014-01-06 21:36:12 +04:00
__skb_queue_head_init ( & newtp - > out_of_order_queue ) ;
2013-03-17 12:23:34 +04:00
newtp - > write_seq = newtp - > pushed_seq = treq - > snt_isn + 1 ;
2005-04-17 02:20:36 +04:00
newtp - > rx_opt . saw_tstamp = 0 ;
newtp - > rx_opt . dsack = 0 ;
newtp - > rx_opt . num_sacks = 0 ;
2009-02-28 07:44:38 +03:00
2005-04-17 02:20:36 +04:00
newtp - > urg_data = 0 ;
if ( sock_flag ( newsk , SOCK_KEEPOPEN ) )
2005-08-10 07:10:42 +04:00
inet_csk_reset_keepalive_timer ( newsk ,
keepalive_time_when ( newtp ) ) ;
2005-04-17 02:20:36 +04:00
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 09:46:52 +04:00
newtp - > rx_opt . tstamp_ok = ireq - > tstamp_ok ;
2007-03-09 07:45:19 +03:00
if ( ( newtp - > rx_opt . sack_ok = ireq - > sack_ok ) ! = 0 ) {
2005-04-17 02:20:36 +04:00
if ( sysctl_tcp_fack )
2007-08-09 16:14:46 +04:00
tcp_enable_fack ( newtp ) ;
2005-04-17 02:20:36 +04:00
}
newtp - > window_clamp = req - > window_clamp ;
newtp - > rcv_ssthresh = req - > rcv_wnd ;
newtp - > rcv_wnd = req - > rcv_wnd ;
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 09:46:52 +04:00
newtp - > rx_opt . wscale_ok = ireq - > wscale_ok ;
2005-04-17 02:20:36 +04:00
if ( newtp - > rx_opt . wscale_ok ) {
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 09:46:52 +04:00
newtp - > rx_opt . snd_wscale = ireq - > snd_wscale ;
newtp - > rx_opt . rcv_wscale = ireq - > rcv_wscale ;
2005-04-17 02:20:36 +04:00
} else {
newtp - > rx_opt . snd_wscale = newtp - > rx_opt . rcv_wscale = 0 ;
newtp - > window_clamp = min ( newtp - > window_clamp , 65535U ) ;
}
2007-04-11 08:04:22 +04:00
newtp - > snd_wnd = ( ntohs ( tcp_hdr ( skb ) - > window ) < <
newtp - > rx_opt . snd_wscale ) ;
2005-04-17 02:20:36 +04:00
newtp - > max_window = newtp - > snd_wnd ;
if ( newtp - > rx_opt . tstamp_ok ) {
newtp - > rx_opt . ts_recent = req - > ts_recent ;
2007-03-05 03:12:44 +03:00
newtp - > rx_opt . ts_recent_stamp = get_seconds ( ) ;
2005-04-17 02:20:36 +04:00
newtp - > tcp_header_len = sizeof ( struct tcphdr ) + TCPOLEN_TSTAMP_ALIGNED ;
} else {
newtp - > rx_opt . ts_recent_stamp = 0 ;
newtp - > tcp_header_len = sizeof ( struct tcphdr ) ;
}
2013-02-11 09:50:17 +04:00
newtp - > tsoffset = 0 ;
2006-11-15 06:07:45 +03:00
# ifdef CONFIG_TCP_MD5SIG
newtp - > md5sig_info = NULL ; /*XXX*/
if ( newtp - > af_specific - > md5_lookup ( sk , newsk ) )
newtp - > tcp_header_len + = TCPOLEN_MD5SIG_ALIGNED ;
# endif
2009-11-10 12:51:18 +03:00
if ( skb - > len > = TCP_MSS_DEFAULT + newtp - > tcp_header_len )
2005-08-10 07:10:42 +04:00
newicsk - > icsk_ack . last_seg_size = skb - > len - newtp - > tcp_header_len ;
2005-04-17 02:20:36 +04:00
newtp - > rx_opt . mss_clamp = req - > mss ;
2014-09-29 15:08:30 +04:00
tcp_ecn_openreq_child ( newtp , req ) ;
2012-08-31 16:29:12 +04:00
newtp - > fastopen_rsk = NULL ;
2012-10-19 19:14:44 +04:00
newtp - > syn_data_acked = 0 ;
2005-04-17 02:20:36 +04:00
2008-07-17 07:22:25 +04:00
TCP_INC_STATS_BH ( sock_net ( sk ) , TCP_MIB_PASSIVEOPENS ) ;
2005-04-17 02:20:36 +04:00
}
return newsk ;
}
2010-07-10 01:22:10 +04:00
EXPORT_SYMBOL ( tcp_create_openreq_child ) ;
2005-04-17 02:20:36 +04:00
2007-02-09 17:24:47 +03:00
/*
2012-08-31 16:29:12 +04:00
* Process an incoming packet for SYN_RECV sockets represented as a
* request_sock . Normally sk is the listener socket but for TFO it
* points to the child socket .
*
* XXX ( TFO ) - The current impl contains a special check for ack
* validation and inside tcp_v4_reqsk_send_ack ( ) . Can we do better ?
2012-09-19 18:46:06 +04:00
*
* We don ' t need to initialize tmp_opt . sack_ok as we don ' t use the results
2005-04-17 02:20:36 +04:00
*/
2008-11-03 11:24:34 +03:00
struct sock * tcp_check_req ( struct sock * sk , struct sk_buff * skb ,
2005-06-19 09:47:21 +04:00
struct request_sock * req ,
2012-08-31 16:29:12 +04:00
struct request_sock * * prev ,
bool fastopen )
2005-04-17 02:20:36 +04:00
{
2009-12-02 21:25:27 +03:00
struct tcp_options_received tmp_opt ;
struct sock * child ;
2007-04-11 08:04:22 +04:00
const struct tcphdr * th = tcp_hdr ( skb ) ;
2006-11-15 07:51:49 +03:00
__be32 flg = tcp_flag_word ( th ) & ( TCP_FLAG_RST | TCP_FLAG_SYN | TCP_FLAG_ACK ) ;
2012-05-17 03:15:34 +04:00
bool paws_reject = false ;
2005-04-17 02:20:36 +04:00
2012-08-31 16:29:12 +04:00
BUG_ON ( fastopen = = ( sk - > sk_state = = TCP_LISTEN ) ) ;
tcp: Revert per-route SACK/DSACK/TIMESTAMP changes.
It creates a regression, triggering badness for SYN_RECV
sockets, for example:
[19148.022102] Badness at net/ipv4/inet_connection_sock.c:293
[19148.022570] NIP: c02a0914 LR: c02a0904 CTR: 00000000
[19148.023035] REGS: eeecbd30 TRAP: 0700 Not tainted (2.6.32)
[19148.023496] MSR: 00029032 <EE,ME,CE,IR,DR> CR: 24002442 XER: 00000000
[19148.024012] TASK = eee9a820[1756] 'privoxy' THREAD: eeeca000
This is likely caused by the change in the 'estab' parameter
passed to tcp_parse_options() when invoked by the functions
in net/ipv4/tcp_minisocks.c
But even if that is fixed, the ->conn_request() changes made in
this patch series is fundamentally wrong. They try to use the
listening socket's 'dst' to probe the route settings. The
listening socket doesn't even have a route, and you can't
get the right route (the child request one) until much later
after we setup all of the state, and it must be done by hand.
This stuff really isn't ready, so the best thing to do is a
full revert. This reverts the following commits:
f55017a93f1a74d50244b1254b9a2bd7ac9bbf7d
022c3f7d82f0f1c68018696f2f027b87b9bb45c2
1aba721eba1d84a2defce45b950272cee1e6c72a
cda42ebd67ee5fdf09d7057b5a4584d36fe8a335
345cda2fd695534be5a4494f1b59da9daed33663
dc343475ed062e13fc260acccaab91d7d80fd5b2
05eaade2782fb0c90d3034fd7a7d5a16266182bb
6a2a2d6bf8581216e08be15fcb563cfd6c430e1e
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-12-16 07:56:42 +03:00
tmp_opt . saw_tstamp = 0 ;
if ( th - > doff > ( sizeof ( struct tcphdr ) > > 2 ) ) {
2013-03-17 12:23:34 +04:00
tcp_parse_options ( skb , & tmp_opt , 0 , NULL ) ;
2005-04-17 02:20:36 +04:00
if ( tmp_opt . saw_tstamp ) {
tmp_opt . ts_recent = req - > ts_recent ;
/* We do not store true stamp, but it is not required,
* it can be estimated ( approximately )
* from another data .
*/
2012-10-28 03:16:46 +04:00
tmp_opt . ts_recent_stamp = get_seconds ( ) - ( ( TCP_TIMEOUT_INIT / HZ ) < < req - > num_timeout ) ;
2009-03-14 17:23:03 +03:00
paws_reject = tcp_paws_reject ( & tmp_opt , th - > rst ) ;
2005-04-17 02:20:36 +04:00
}
}
/* Check for pure retransmitted SYN. */
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 09:46:52 +04:00
if ( TCP_SKB_CB ( skb ) - > seq = = tcp_rsk ( req ) - > rcv_isn & &
2005-04-17 02:20:36 +04:00
flg = = TCP_FLAG_SYN & &
! paws_reject ) {
/*
* RFC793 draws ( Incorrectly ! It was fixed in RFC1122 )
* this case on figure 6 and figure 8 , but formal
* protocol description says NOTHING .
* To be more exact , it says that we should send ACK ,
* because this segment ( at least , if it has no data )
* is out of window .
*
* CONCLUSION : RFC793 ( even with RFC1122 ) DOES NOT
* describe SYN - RECV state . All the description
* is wrong , we cannot believe to it and should
* rely only on common sense and implementation
* experience .
*
* Enforce " SYN-ACK " according to figure 8 , figure 6
* of RFC793 , fixed by RFC1122 .
2012-08-31 16:29:12 +04:00
*
* Note that even if there is new data in the SYN packet
* they will be thrown away too .
2013-04-29 12:44:51 +04:00
*
* Reset timer after retransmitting SYNACK , similar to
* the idea of fast retransmit in recovery .
2005-04-17 02:20:36 +04:00
*/
2013-04-29 12:44:51 +04:00
if ( ! inet_rtx_syn_ack ( sk , req ) )
req - > expires = min ( TCP_TIMEOUT_INIT < < req - > num_timeout ,
TCP_RTO_MAX ) + jiffies ;
2005-04-17 02:20:36 +04:00
return NULL ;
}
/* Further reproduces section "SEGMENT ARRIVES"
for state SYN - RECEIVED of RFC793 .
It is broken , however , it does not work only
when SYNs are crossed .
You would think that SYN crossing is impossible here , since
we should have a SYN_SENT socket ( from connect ( ) ) on our end ,
but this is not true if the crossed SYNs were sent to both
ends by a malicious third party . We must defend against this ,
and to do that we first verify the ACK ( as per RFC793 , page
36 ) and reset if it is invalid . Is this a true full defense ?
To convince ourselves , let us consider a way in which the ACK
test can still pass in this ' malicious crossed SYNs ' case .
Malicious sender sends identical SYNs ( and thus identical sequence
numbers ) to both A and B :
A : gets SYN , seq = 7
B : gets SYN , seq = 7
By our good fortune , both A and B select the same initial
send sequence number of seven : - )
A : sends SYN | ACK , seq = 7 , ack_seq = 8
B : sends SYN | ACK , seq = 7 , ack_seq = 8
So we are now A eating this SYN | ACK , ACK test passes . So
does sequence test , SYN is truncated , and thus we consider
it a bare ACK .
tcp: Revert 'process defer accept as established' changes.
This reverts two changesets, ec3c0982a2dd1e671bad8e9d26c28dcba0039d87
("[TCP]: TCP_DEFER_ACCEPT updates - process as established") and
the follow-on bug fix 9ae27e0adbf471c7a6b80102e38e1d5a346b3b38
("tcp: Fix slab corruption with ipv6 and tcp6fuzz").
This change causes several problems, first reported by Ingo Molnar
as a distcc-over-loopback regression where connections were getting
stuck.
Ilpo Järvinen first spotted the locking problems. The new function
added by this code, tcp_defer_accept_check(), only has the
child socket locked, yet it is modifying state of the parent
listening socket.
Fixing that is non-trivial at best, because we can't simply just grab
the parent listening socket lock at this point, because it would
create an ABBA deadlock. The normal ordering is parent listening
socket --> child socket, but this code path would require the
reverse lock ordering.
Next is a problem noticed by Vitaliy Gusev, he noted:
----------------------------------------
>--- a/net/ipv4/tcp_timer.c
>+++ b/net/ipv4/tcp_timer.c
>@@ -481,6 +481,11 @@ static void tcp_keepalive_timer (unsigned long data)
> goto death;
> }
>
>+ if (tp->defer_tcp_accept.request && sk->sk_state == TCP_ESTABLISHED) {
>+ tcp_send_active_reset(sk, GFP_ATOMIC);
>+ goto death;
Here socket sk is not attached to listening socket's request queue. tcp_done()
will not call inet_csk_destroy_sock() (and tcp_v4_destroy_sock() which should
release this sk) as socket is not DEAD. Therefore socket sk will be lost for
freeing.
----------------------------------------
Finally, Alexey Kuznetsov argues that there might not even be any
real value or advantage to these new semantics even if we fix all
of the bugs:
----------------------------------------
Hiding from accept() sockets with only out-of-order data only
is the only thing which is impossible with old approach. Is this really
so valuable? My opinion: no, this is nothing but a new loophole
to consume memory without control.
----------------------------------------
So revert this thing for now.
Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-13 03:31:35 +04:00
If icsk - > icsk_accept_queue . rskq_defer_accept , we silently drop this
bare ACK . Otherwise , we create an established connection . Both
ends ( listening sockets ) accept the new incoming connection and try
to talk to each other . 8 - )
2005-04-17 02:20:36 +04:00
Note : This case is both harmless , and rare . Possibility is about the
same as us discovering intelligent life on another plant tomorrow .
But generally , we should ( RFC lies ! ) to accept ACK
from SYNACK both here and in tcp_rcv_state_process ( ) .
tcp_rcv_state_process ( ) does not , hence , we do not too .
Note that the case is absolutely generic :
we cannot optimize anything here without
violating protocol . All the checks must be made
before attempt to create socket .
*/
/* RFC793 page 36: "If the connection is in any non-synchronized state ...
* and the incoming segment acknowledges something not yet
2005-11-11 04:13:47 +03:00
* sent ( the segment carries an unacceptable ACK ) . . .
2005-04-17 02:20:36 +04:00
* a reset is sent . "
*
2012-08-31 16:29:12 +04:00
* Invalid ACK : reset will be sent by listening socket .
* Note that the ACK validity check for a Fast Open socket is done
* elsewhere and is checked directly against the child socket rather
* than req because user data may have been sent out .
2005-04-17 02:20:36 +04:00
*/
2012-08-31 16:29:12 +04:00
if ( ( flg & TCP_FLAG_ACK ) & & ! fastopen & &
TCPCT part 1d: define TCP cookie option, extend existing struct's
Data structures are carefully composed to require minimal additions.
For example, the struct tcp_options_received cookie_plus variable fits
between existing 16-bit and 8-bit variables, requiring no additional
space (taking alignment into consideration). There are no additions to
tcp_request_sock, and only 1 pointer in tcp_sock.
This is a significantly revised implementation of an earlier (year-old)
patch that no longer applies cleanly, with permission of the original
author (Adam Langley):
http://thread.gmane.org/gmane.linux.network/102586
The principle difference is using a TCP option to carry the cookie nonce,
instead of a user configured offset in the data. This is more flexible and
less subject to user configuration error. Such a cookie option has been
suggested for many years, and is also useful without SYN data, allowing
several related concepts to use the same extension option.
"Re: SYN floods (was: does history repeat itself?)", September 9, 1996.
http://www.merit.net/mail.archives/nanog/1996-09/msg00235.html
"Re: what a new TCP header might look like", May 12, 1998.
ftp://ftp.isi.edu/end2end/end2end-interest-1998.mail
These functions will also be used in subsequent patches that implement
additional features.
Requires:
TCPCT part 1a: add request_values parameter for sending SYNACK
TCPCT part 1b: generate Responder Cookie secret
TCPCT part 1c: sysctl_tcp_cookie_size, socket option TCP_COOKIE_TRANSACTIONS
Signed-off-by: William.Allen.Simpson@gmail.com
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-12-02 21:17:05 +03:00
( TCP_SKB_CB ( skb ) - > ack_seq ! =
2013-03-17 12:23:34 +04:00
tcp_rsk ( req ) - > snt_isn + 1 ) )
2005-04-17 02:20:36 +04:00
return sk ;
/* Also, it would be not so bad idea to check rcv_tsecr, which
* is essentially ACK extension and too early or too late values
* should cause reset in unsynchronized states .
*/
/* RFC793: "first check sequence number". */
if ( paws_reject | | ! tcp_in_window ( TCP_SKB_CB ( skb ) - > seq , TCP_SKB_CB ( skb ) - > end_seq ,
2012-08-31 16:29:12 +04:00
tcp_rsk ( req ) - > rcv_nxt , tcp_rsk ( req ) - > rcv_nxt + req - > rcv_wnd ) ) {
2005-04-17 02:20:36 +04:00
/* Out of window: send ACK and drop. */
if ( ! ( flg & TCP_FLAG_RST ) )
2008-08-07 10:50:04 +04:00
req - > rsk_ops - > send_ack ( sk , skb , req ) ;
2005-04-17 02:20:36 +04:00
if ( paws_reject )
2008-07-17 07:31:16 +04:00
NET_INC_STATS_BH ( sock_net ( sk ) , LINUX_MIB_PAWSESTABREJECTED ) ;
2005-04-17 02:20:36 +04:00
return NULL ;
}
/* In sequence, PAWS is OK. */
2012-08-31 16:29:12 +04:00
if ( tmp_opt . saw_tstamp & & ! after ( TCP_SKB_CB ( skb ) - > seq , tcp_rsk ( req ) - > rcv_nxt ) )
2008-08-08 07:27:45 +04:00
req - > ts_recent = tmp_opt . rcv_tsval ;
2005-04-17 02:20:36 +04:00
2008-08-08 07:27:45 +04:00
if ( TCP_SKB_CB ( skb ) - > seq = = tcp_rsk ( req ) - > rcv_isn ) {
/* Truncate SYN, it is out of window starting
at tcp_rsk ( req ) - > rcv_isn + 1. */
flg & = ~ TCP_FLAG_SYN ;
}
2005-04-17 02:20:36 +04:00
2008-08-08 07:27:45 +04:00
/* RFC793: "second check the RST bit" and
* " fourth, check the SYN bit "
*/
if ( flg & ( TCP_FLAG_RST | TCP_FLAG_SYN ) ) {
TCP_INC_STATS_BH ( sock_net ( sk ) , TCP_MIB_ATTEMPTFAILS ) ;
goto embryonic_reset ;
}
2005-04-17 02:20:36 +04:00
2008-08-08 07:27:45 +04:00
/* ACK sequence verified above, just make sure ACK is
* set . If ACK not set , just silently drop the packet .
2012-08-31 16:29:12 +04:00
*
* XXX ( TFO ) - if we ever allow " data after SYN " , the
* following check needs to be removed .
2008-08-08 07:27:45 +04:00
*/
if ( ! ( flg & TCP_FLAG_ACK ) )
return NULL ;
tcp: Revert 'process defer accept as established' changes.
This reverts two changesets, ec3c0982a2dd1e671bad8e9d26c28dcba0039d87
("[TCP]: TCP_DEFER_ACCEPT updates - process as established") and
the follow-on bug fix 9ae27e0adbf471c7a6b80102e38e1d5a346b3b38
("tcp: Fix slab corruption with ipv6 and tcp6fuzz").
This change causes several problems, first reported by Ingo Molnar
as a distcc-over-loopback regression where connections were getting
stuck.
Ilpo Järvinen first spotted the locking problems. The new function
added by this code, tcp_defer_accept_check(), only has the
child socket locked, yet it is modifying state of the parent
listening socket.
Fixing that is non-trivial at best, because we can't simply just grab
the parent listening socket lock at this point, because it would
create an ABBA deadlock. The normal ordering is parent listening
socket --> child socket, but this code path would require the
reverse lock ordering.
Next is a problem noticed by Vitaliy Gusev, he noted:
----------------------------------------
>--- a/net/ipv4/tcp_timer.c
>+++ b/net/ipv4/tcp_timer.c
>@@ -481,6 +481,11 @@ static void tcp_keepalive_timer (unsigned long data)
> goto death;
> }
>
>+ if (tp->defer_tcp_accept.request && sk->sk_state == TCP_ESTABLISHED) {
>+ tcp_send_active_reset(sk, GFP_ATOMIC);
>+ goto death;
Here socket sk is not attached to listening socket's request queue. tcp_done()
will not call inet_csk_destroy_sock() (and tcp_v4_destroy_sock() which should
release this sk) as socket is not DEAD. Therefore socket sk will be lost for
freeing.
----------------------------------------
Finally, Alexey Kuznetsov argues that there might not even be any
real value or advantage to these new semantics even if we fix all
of the bugs:
----------------------------------------
Hiding from accept() sockets with only out-of-order data only
is the only thing which is impossible with old approach. Is this really
so valuable? My opinion: no, this is nothing but a new loophole
to consume memory without control.
----------------------------------------
So revert this thing for now.
Signed-off-by: David S. Miller <davem@davemloft.net>
2008-06-13 03:31:35 +04:00
2012-08-31 16:29:12 +04:00
/* For Fast Open no more processing is needed (sk is the
* child socket ) .
*/
if ( fastopen )
return sk ;
2009-10-19 14:01:56 +04:00
/* While TCP_DEFER_ACCEPT is active, drop bare ACK. */
2012-10-28 03:16:46 +04:00
if ( req - > num_timeout < inet_csk ( sk ) - > icsk_accept_queue . rskq_defer_accept & &
2008-08-08 07:27:45 +04:00
TCP_SKB_CB ( skb ) - > end_seq = = tcp_rsk ( req ) - > rcv_isn + 1 ) {
inet_rsk ( req ) - > acked = 1 ;
2010-03-19 08:37:18 +03:00
NET_INC_STATS_BH ( sock_net ( sk ) , LINUX_MIB_TCPDEFERACCEPTDROP ) ;
2008-08-08 07:27:45 +04:00
return NULL ;
}
/* OK, ACK is valid, create big socket and
* feed this segment to it . It will repeat all
* the tests . THIS SEGMENT MUST MOVE SOCKET TO
* ESTABLISHED STATE . If it will be dropped after
* socket is created , wait for troubles .
*/
child = inet_csk ( sk ) - > icsk_af_ops - > syn_recv_sock ( sk , skb , req , NULL ) ;
if ( child = = NULL )
goto listen_overflow ;
2005-04-17 02:20:36 +04:00
2008-08-08 07:27:45 +04:00
inet_csk_reqsk_queue_unlink ( sk , req , prev ) ;
inet_csk_reqsk_queue_removed ( sk , req ) ;
2005-04-17 02:20:36 +04:00
2008-08-08 07:27:45 +04:00
inet_csk_reqsk_queue_add ( sk , req , child ) ;
return child ;
2005-04-17 02:20:36 +04:00
2008-08-08 07:27:45 +04:00
listen_overflow :
if ( ! sysctl_tcp_abort_on_overflow ) {
inet_rsk ( req ) - > acked = 1 ;
return NULL ;
}
2005-04-17 02:20:36 +04:00
2008-08-08 07:27:45 +04:00
embryonic_reset :
2012-08-31 16:29:12 +04:00
if ( ! ( flg & TCP_FLAG_RST ) ) {
/* Received a bad SYN pkt - for TFO We try not to reset
* the local connection unless it ' s really necessary to
* avoid becoming vulnerable to outside attack aiming at
* resetting legit local connections .
*/
2008-08-08 07:27:45 +04:00
req - > rsk_ops - > send_reset ( sk , skb ) ;
2012-08-31 16:29:12 +04:00
} else if ( fastopen ) { /* received a valid RST pkt */
reqsk_fastopen_remove ( sk , req , true ) ;
tcp_reset ( sk ) ;
}
if ( ! fastopen ) {
inet_csk_reqsk_queue_drop ( sk , req , prev ) ;
NET_INC_STATS_BH ( sock_net ( sk ) , LINUX_MIB_EMBRYONICRSTS ) ;
}
2008-08-08 07:27:45 +04:00
return NULL ;
2005-04-17 02:20:36 +04:00
}
2010-07-10 01:22:10 +04:00
EXPORT_SYMBOL ( tcp_check_req ) ;
2005-04-17 02:20:36 +04:00
/*
* Queue segment on the new socket if the new socket is active ,
* otherwise we just shortcircuit this and continue with
* the new socket .
2012-08-31 16:29:12 +04:00
*
* For the vast majority of cases child - > sk_state will be TCP_SYN_RECV
* when entering . But other states are possible due to a race condition
* where after __inet_lookup_established ( ) fails but before the listener
* locked is obtained , other packets cause the same connection to
* be created .
2005-04-17 02:20:36 +04:00
*/
int tcp_child_process ( struct sock * parent , struct sock * child ,
struct sk_buff * skb )
{
int ret = 0 ;
int state = child - > sk_state ;
if ( ! sock_owned_by_user ( child ) ) {
2007-04-11 08:04:22 +04:00
ret = tcp_rcv_state_process ( child , skb , tcp_hdr ( skb ) ,
skb - > len ) ;
2005-04-17 02:20:36 +04:00
/* Wakeup parent, send SIGIO */
if ( state = = TCP_SYN_RECV & & child - > sk_state ! = state )
2014-04-12 00:15:36 +04:00
parent - > sk_data_ready ( parent ) ;
2005-04-17 02:20:36 +04:00
} else {
/* Alas, it is possible again, because we do lookup
* in main socket hash table and lock on listening
* socket does not protect us more .
*/
2010-03-04 21:01:47 +03:00
__sk_add_backlog ( child , skb ) ;
2005-04-17 02:20:36 +04:00
}
bh_unlock_sock ( child ) ;
sock_put ( child ) ;
return ret ;
}
EXPORT_SYMBOL ( tcp_child_process ) ;