freebsd-skq/sys/netinet/sctp_timer.c

1584 lines
47 KiB
C
Raw Normal View History

/*-
* Copyright (c) 2001-2007, by Cisco Systems, Inc. All rights reserved.
* Copyright (c) 2008-2012, by Randall Stewart. All rights reserved.
* Copyright (c) 2008-2012, by Michael Tuexen. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* a) Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
*
* b) Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the distribution.
*
* c) Neither the name of Cisco Systems, Inc. nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
* THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
* THE POSSIBILITY OF SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#define _IP_VHL
#include <netinet/sctp_os.h>
#include <netinet/sctp_pcb.h>
#ifdef INET6
#endif
#include <netinet/sctp_var.h>
#include <netinet/sctp_sysctl.h>
#include <netinet/sctp_timer.h>
#include <netinet/sctputil.h>
#include <netinet/sctp_output.h>
#include <netinet/sctp_header.h>
#include <netinet/sctp_indata.h>
#include <netinet/sctp_asconf.h>
#include <netinet/sctp_input.h>
#include <netinet/sctp.h>
#include <netinet/sctp_uio.h>
#if defined(INET) || defined(INET6)
#include <netinet/udp.h>
#endif
void
sctp_audit_retranmission_queue(struct sctp_association *asoc)
{
struct sctp_tmit_chunk *chk;
SCTPDBG(SCTP_DEBUG_TIMER4, "Audit invoked on send queue cnt:%d onqueue:%d\n",
asoc->sent_queue_retran_cnt,
asoc->sent_queue_cnt);
asoc->sent_queue_retran_cnt = 0;
asoc->sent_queue_cnt = 0;
TAILQ_FOREACH(chk, &asoc->sent_queue, sctp_next) {
if (chk->sent == SCTP_DATAGRAM_RESEND) {
sctp_ucount_incr(asoc->sent_queue_retran_cnt);
}
asoc->sent_queue_cnt++;
}
TAILQ_FOREACH(chk, &asoc->control_send_queue, sctp_next) {
if (chk->sent == SCTP_DATAGRAM_RESEND) {
sctp_ucount_incr(asoc->sent_queue_retran_cnt);
}
}
TAILQ_FOREACH(chk, &asoc->asconf_send_queue, sctp_next) {
if (chk->sent == SCTP_DATAGRAM_RESEND) {
sctp_ucount_incr(asoc->sent_queue_retran_cnt);
}
}
SCTPDBG(SCTP_DEBUG_TIMER4, "Audit completes retran:%d onqueue:%d\n",
asoc->sent_queue_retran_cnt,
asoc->sent_queue_cnt);
}
int
sctp_threshold_management(struct sctp_inpcb *inp, struct sctp_tcb *stcb,
struct sctp_nets *net, uint16_t threshold)
{
if (net) {
net->error_count++;
SCTPDBG(SCTP_DEBUG_TIMER4, "Error count for %p now %d thresh:%d\n",
(void *)net, net->error_count,
net->failure_threshold);
if (net->error_count > net->failure_threshold) {
/* We had a threshold failure */
if (net->dest_state & SCTP_ADDR_REACHABLE) {
net->dest_state &= ~SCTP_ADDR_REACHABLE;
net->dest_state &= ~SCTP_ADDR_REQ_PRIMARY;
net->dest_state &= ~SCTP_ADDR_PF;
sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_DOWN,
stcb, 0,
(void *)net, SCTP_SO_NOT_LOCKED);
}
} else if ((net->pf_threshold < net->failure_threshold) &&
(net->error_count > net->pf_threshold)) {
if (!(net->dest_state & SCTP_ADDR_PF)) {
net->dest_state |= SCTP_ADDR_PF;
net->last_active = sctp_get_tick_count();
sctp_send_hb(stcb, net, SCTP_SO_NOT_LOCKED);
sctp_timer_stop(SCTP_TIMER_TYPE_HEARTBEAT,
stcb->sctp_ep, stcb, net,
SCTP_FROM_SCTP_TIMER + SCTP_LOC_1);
sctp_timer_start(SCTP_TIMER_TYPE_HEARTBEAT, stcb->sctp_ep, stcb, net);
}
}
}
if (stcb == NULL)
return (0);
if (net) {
if ((net->dest_state & SCTP_ADDR_UNCONFIRMED) == 0) {
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_THRESHOLD_LOGGING) {
- Fix address add handling to clear cached routes and source addresses when peer acks the add in case the routing table changes. - Fix sctp_lower_sosend to send shutdown chunk for mbuf send case when sndlen = 0 and sinfoflag = SCTP_EOF - Fix sctp_lower_sosend for SCTP_ABORT mbuf send case with null data, So that it does not send the "null" data mbuf out and cause it to get freed twice. - Fix so auto-asconf sysctl actually effect the socket's asconf state. - Do not allow SCTP_AUTO_ASCONF option to be used on subset bound sockets. - Memset bug in sctp_output.c (arguments were reversed) submitted found and reported by Dave Jones (davej@codemonkey.org.uk). - PD-API point needs to be invoked >= not just > to conform to socket api draft this fixes sctp_indata.c in the two places need to be >=. - move M_NOTIFICATION to use M_PROTO5. - PEER_ADDR_PARAMS did not fail properly if you specify an address that is not in the association with a valid assoc_id. This meant you got or set the stcb level values instead of the destination you thought you were going to get/set. Now validate if the stcb is non-null and the net is NULL that the sa_family is set and the address is unspecified otherwise return an error. - The thread based iterator could crash if associations were freed at the exact time it was running. rework the worker thread to use the increment/decrement to prevent this and no longer use the markers that the timer based iterator uses. - Fix the memleak in sctp_add_addr_to_vrf() for the case when it is detected that ifa is already pointing to a ifn. - Fix it so that if someone is so insane that they drop the send window below the minimal add mark, they still can send. - Changed all state for associations to use mask safe macro. - During front states in association freeing in sctp_inpcbfree, we had a locking problem where locks were not in place where they should have been. - Free association calls were not testing the return value in sctp_inpcb_free() properly... others should be cast void returns where we don't care about the return value. - If a reference count is held on an assoc, even from the "force free" we should not do the actual free.. but instead let the timer free it. - When we enter sctp_input(), if the SCTP_ASOC_ABOUT_TO_BE_FREED flag is set, we must NOT process the packet but handle it like ootb. This is because while freeing an assoc we release the locks to get all the higher order locks so we can purge all the hash tables. This leaves a hole if a packet comes in just at that point. Now sctp_common_input_processing() will call the ootb code in such a case. - Change MBUF M_NOTIFICATION to use M_PROTO5 (per Sam L). This makes it so we don't have a conflict (I think this is a covertity change). We made this change AFTER some conversation and looking to make sure that M_PROTO5 does not have a problem between SCTP and the 802.11 stuff (which is the only other place its used). - Fixed lock order reversal and missing atomic protection around locked_tcb during association lookup and the 1-2-1 model. - Added debug to source address selection. - V6 output must always do checksum even for loopback. - Remove more locks around inp that are not needed for an atomically added/subtracted ref count. - slight optimization in the way we zero the array in sctp_sack_check() - It was possible to respond to a ABORT() with bad checksum with a PKT-DROP. This lead to a PKT-DROP/ABORT war. Add code to NOT send a PKT-DROP to any ABORT(). - Add an option for local logging (useful for macintosh or when you need better performing during debugging). Note no commands are here to get the log info, you must just use kgdb. - The timer code needs to be aware of if it needs to call sctp_sack_check() to slide the maps and adjust the cum-ack. This is because it may be out of sync cum-ack wise. - Added threshold managment logging. - If the user picked just the right size, that just filled the send window minus one mtu, we would enter a forever loop not copying and at the same time not blocking. Change from < to <= solves this. - Sysctl added to control the fragment interleave level which defaults to 1. - My rwnd control was not being used to control the rwnd properly (we did not add and subtract to it :-() this is now fixed so we handle small messages (1 byte etc) better to bring our rwnd down more slowly. Approved by: re@freebsd.org (Bruce Mah)
2007-08-24 00:53:53 +00:00
sctp_misc_ints(SCTP_THRESHOLD_INCR,
stcb->asoc.overall_error_count,
(stcb->asoc.overall_error_count + 1),
SCTP_FROM_SCTP_TIMER,
__LINE__);
}
stcb->asoc.overall_error_count++;
}
} else {
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_THRESHOLD_LOGGING) {
- Fix address add handling to clear cached routes and source addresses when peer acks the add in case the routing table changes. - Fix sctp_lower_sosend to send shutdown chunk for mbuf send case when sndlen = 0 and sinfoflag = SCTP_EOF - Fix sctp_lower_sosend for SCTP_ABORT mbuf send case with null data, So that it does not send the "null" data mbuf out and cause it to get freed twice. - Fix so auto-asconf sysctl actually effect the socket's asconf state. - Do not allow SCTP_AUTO_ASCONF option to be used on subset bound sockets. - Memset bug in sctp_output.c (arguments were reversed) submitted found and reported by Dave Jones (davej@codemonkey.org.uk). - PD-API point needs to be invoked >= not just > to conform to socket api draft this fixes sctp_indata.c in the two places need to be >=. - move M_NOTIFICATION to use M_PROTO5. - PEER_ADDR_PARAMS did not fail properly if you specify an address that is not in the association with a valid assoc_id. This meant you got or set the stcb level values instead of the destination you thought you were going to get/set. Now validate if the stcb is non-null and the net is NULL that the sa_family is set and the address is unspecified otherwise return an error. - The thread based iterator could crash if associations were freed at the exact time it was running. rework the worker thread to use the increment/decrement to prevent this and no longer use the markers that the timer based iterator uses. - Fix the memleak in sctp_add_addr_to_vrf() for the case when it is detected that ifa is already pointing to a ifn. - Fix it so that if someone is so insane that they drop the send window below the minimal add mark, they still can send. - Changed all state for associations to use mask safe macro. - During front states in association freeing in sctp_inpcbfree, we had a locking problem where locks were not in place where they should have been. - Free association calls were not testing the return value in sctp_inpcb_free() properly... others should be cast void returns where we don't care about the return value. - If a reference count is held on an assoc, even from the "force free" we should not do the actual free.. but instead let the timer free it. - When we enter sctp_input(), if the SCTP_ASOC_ABOUT_TO_BE_FREED flag is set, we must NOT process the packet but handle it like ootb. This is because while freeing an assoc we release the locks to get all the higher order locks so we can purge all the hash tables. This leaves a hole if a packet comes in just at that point. Now sctp_common_input_processing() will call the ootb code in such a case. - Change MBUF M_NOTIFICATION to use M_PROTO5 (per Sam L). This makes it so we don't have a conflict (I think this is a covertity change). We made this change AFTER some conversation and looking to make sure that M_PROTO5 does not have a problem between SCTP and the 802.11 stuff (which is the only other place its used). - Fixed lock order reversal and missing atomic protection around locked_tcb during association lookup and the 1-2-1 model. - Added debug to source address selection. - V6 output must always do checksum even for loopback. - Remove more locks around inp that are not needed for an atomically added/subtracted ref count. - slight optimization in the way we zero the array in sctp_sack_check() - It was possible to respond to a ABORT() with bad checksum with a PKT-DROP. This lead to a PKT-DROP/ABORT war. Add code to NOT send a PKT-DROP to any ABORT(). - Add an option for local logging (useful for macintosh or when you need better performing during debugging). Note no commands are here to get the log info, you must just use kgdb. - The timer code needs to be aware of if it needs to call sctp_sack_check() to slide the maps and adjust the cum-ack. This is because it may be out of sync cum-ack wise. - Added threshold managment logging. - If the user picked just the right size, that just filled the send window minus one mtu, we would enter a forever loop not copying and at the same time not blocking. Change from < to <= solves this. - Sysctl added to control the fragment interleave level which defaults to 1. - My rwnd control was not being used to control the rwnd properly (we did not add and subtract to it :-() this is now fixed so we handle small messages (1 byte etc) better to bring our rwnd down more slowly. Approved by: re@freebsd.org (Bruce Mah)
2007-08-24 00:53:53 +00:00
sctp_misc_ints(SCTP_THRESHOLD_INCR,
stcb->asoc.overall_error_count,
(stcb->asoc.overall_error_count + 1),
SCTP_FROM_SCTP_TIMER,
__LINE__);
}
stcb->asoc.overall_error_count++;
}
SCTPDBG(SCTP_DEBUG_TIMER4, "Overall error count for %p now %d thresh:%u state:%x\n",
(void *)&stcb->asoc, stcb->asoc.overall_error_count,
(uint32_t) threshold,
((net == NULL) ? (uint32_t) 0 : (uint32_t) net->dest_state));
/*
* We specifically do not do >= to give the assoc one more change
* before we fail it.
*/
if (stcb->asoc.overall_error_count > threshold) {
/* Abort notification sends a ULP notify */
struct mbuf *op_err;
op_err = sctp_generate_cause(SCTP_CAUSE_PROTOCOL_VIOLATION,
2015-03-10 09:16:31 +00:00
"Association error counter exceeded");
inp->last_abort_code = SCTP_FROM_SCTP_TIMER + SCTP_LOC_2;
sctp_abort_an_association(inp, stcb, op_err, SCTP_SO_NOT_LOCKED);
return (1);
}
return (0);
}
/*
* sctp_find_alternate_net() returns a non-NULL pointer as long
* the argument net is non-NULL.
*/
struct sctp_nets *
sctp_find_alternate_net(struct sctp_tcb *stcb,
struct sctp_nets *net,
int mode)
{
/* Find and return an alternate network if possible */
struct sctp_nets *alt, *mnet, *min_errors_net = NULL, *max_cwnd_net = NULL;
int once;
/* JRS 5/14/07 - Initialize min_errors to an impossible value. */
int min_errors = -1;
uint32_t max_cwnd = 0;
if (stcb->asoc.numnets == 1) {
/* No others but net */
return (TAILQ_FIRST(&stcb->asoc.nets));
}
/*
* JRS 5/14/07 - If mode is set to 2, use the CMT PF find alternate
* net algorithm. This algorithm chooses the active destination (not
* in PF state) with the largest cwnd value. If all destinations are
* in PF state, unreachable, or unconfirmed, choose the desination
* that is in PF state with the lowest error count. In case of a
* tie, choose the destination that was most recently active.
*/
if (mode == 2) {
TAILQ_FOREACH(mnet, &stcb->asoc.nets, sctp_next) {
/*
* JRS 5/14/07 - If the destination is unreachable
* or unconfirmed, skip it.
*/
if (((mnet->dest_state & SCTP_ADDR_REACHABLE) != SCTP_ADDR_REACHABLE) ||
(mnet->dest_state & SCTP_ADDR_UNCONFIRMED)) {
continue;
}
/*
* JRS 5/14/07 - If the destination is reachable
* but in PF state, compare the error count of the
* destination to the minimum error count seen thus
* far. Store the destination with the lower error
* count. If the error counts are equal, store the
* destination that was most recently active.
*/
if (mnet->dest_state & SCTP_ADDR_PF) {
/*
* JRS 5/14/07 - If the destination under
* consideration is the current destination,
* work as if the error count is one higher.
* The actual error count will not be
* incremented until later in the t3
* handler.
*/
if (mnet == net) {
if (min_errors == -1) {
min_errors = mnet->error_count + 1;
min_errors_net = mnet;
} else if (mnet->error_count + 1 < min_errors) {
min_errors = mnet->error_count + 1;
min_errors_net = mnet;
} else if (mnet->error_count + 1 == min_errors
&& mnet->last_active > min_errors_net->last_active) {
min_errors_net = mnet;
min_errors = mnet->error_count + 1;
}
continue;
} else {
if (min_errors == -1) {
min_errors = mnet->error_count;
min_errors_net = mnet;
} else if (mnet->error_count < min_errors) {
min_errors = mnet->error_count;
min_errors_net = mnet;
} else if (mnet->error_count == min_errors
&& mnet->last_active > min_errors_net->last_active) {
min_errors_net = mnet;
min_errors = mnet->error_count;
}
continue;
}
}
/*
* JRS 5/14/07 - If the destination is reachable and
* not in PF state, compare the cwnd of the
* destination to the highest cwnd seen thus far.
* Store the destination with the higher cwnd value.
* If the cwnd values are equal, randomly choose one
* of the two destinations.
*/
if (max_cwnd < mnet->cwnd) {
max_cwnd_net = mnet;
max_cwnd = mnet->cwnd;
} else if (max_cwnd == mnet->cwnd) {
uint32_t rndval;
uint8_t this_random;
if (stcb->asoc.hb_random_idx > 3) {
rndval = sctp_select_initial_TSN(&stcb->sctp_ep->sctp_ep);
memcpy(stcb->asoc.hb_random_values, &rndval, sizeof(stcb->asoc.hb_random_values));
this_random = stcb->asoc.hb_random_values[0];
stcb->asoc.hb_random_idx++;
stcb->asoc.hb_ect_randombit = 0;
} else {
this_random = stcb->asoc.hb_random_values[stcb->asoc.hb_random_idx];
stcb->asoc.hb_random_idx++;
stcb->asoc.hb_ect_randombit = 0;
}
if (this_random % 2 == 1) {
max_cwnd_net = mnet;
max_cwnd = mnet->cwnd; /* Useless? */
}
}
}
if (max_cwnd_net == NULL) {
if (min_errors_net == NULL) {
return (net);
}
return (min_errors_net);
} else {
return (max_cwnd_net);
}
}
/*
* JRS 5/14/07 - If mode is set to 1, use the CMT policy for
* choosing an alternate net.
*/
else if (mode == 1) {
TAILQ_FOREACH(mnet, &stcb->asoc.nets, sctp_next) {
if (((mnet->dest_state & SCTP_ADDR_REACHABLE) != SCTP_ADDR_REACHABLE) ||
(mnet->dest_state & SCTP_ADDR_UNCONFIRMED)) {
/*
* will skip ones that are not-reachable or
* unconfirmed
*/
continue;
}
if (max_cwnd < mnet->cwnd) {
max_cwnd_net = mnet;
max_cwnd = mnet->cwnd;
} else if (max_cwnd == mnet->cwnd) {
uint32_t rndval;
uint8_t this_random;
if (stcb->asoc.hb_random_idx > 3) {
rndval = sctp_select_initial_TSN(&stcb->sctp_ep->sctp_ep);
memcpy(stcb->asoc.hb_random_values, &rndval,
sizeof(stcb->asoc.hb_random_values));
this_random = stcb->asoc.hb_random_values[0];
stcb->asoc.hb_random_idx = 0;
stcb->asoc.hb_ect_randombit = 0;
} else {
this_random = stcb->asoc.hb_random_values[stcb->asoc.hb_random_idx];
stcb->asoc.hb_random_idx++;
stcb->asoc.hb_ect_randombit = 0;
}
if (this_random % 2) {
max_cwnd_net = mnet;
max_cwnd = mnet->cwnd;
}
}
}
if (max_cwnd_net) {
return (max_cwnd_net);
}
}
mnet = net;
once = 0;
if (mnet == NULL) {
mnet = TAILQ_FIRST(&stcb->asoc.nets);
if (mnet == NULL) {
return (NULL);
}
}
for (;;) {
alt = TAILQ_NEXT(mnet, sctp_next);
if (alt == NULL) {
once++;
if (once > 1) {
break;
}
alt = TAILQ_FIRST(&stcb->asoc.nets);
if (alt == NULL) {
return (NULL);
}
}
if (alt->ro.ro_rt == NULL) {
if (alt->ro._s_addr) {
sctp_free_ifa(alt->ro._s_addr);
alt->ro._s_addr = NULL;
}
alt->src_addr_selected = 0;
}
if (((alt->dest_state & SCTP_ADDR_REACHABLE) == SCTP_ADDR_REACHABLE) &&
(alt->ro.ro_rt != NULL) &&
(!(alt->dest_state & SCTP_ADDR_UNCONFIRMED))) {
/* Found a reachable address */
break;
}
mnet = alt;
}
if (alt == NULL) {
/* Case where NO insv network exists (dormant state) */
/* we rotate destinations */
once = 0;
mnet = net;
for (;;) {
if (mnet == NULL) {
return (TAILQ_FIRST(&stcb->asoc.nets));
}
alt = TAILQ_NEXT(mnet, sctp_next);
if (alt == NULL) {
once++;
if (once > 1) {
break;
}
alt = TAILQ_FIRST(&stcb->asoc.nets);
if (alt == NULL) {
break;
}
}
if ((!(alt->dest_state & SCTP_ADDR_UNCONFIRMED)) &&
(alt != net)) {
/* Found an alternate address */
break;
}
mnet = alt;
}
}
if (alt == NULL) {
return (net);
}
return (alt);
}
static void
sctp_backoff_on_timeout(struct sctp_tcb *stcb,
struct sctp_nets *net,
int win_probe,
int num_marked, int num_abandoned)
{
if (net->RTO == 0) {
net->RTO = stcb->asoc.minrto;
}
net->RTO <<= 1;
if (net->RTO > stcb->asoc.maxrto) {
net->RTO = stcb->asoc.maxrto;
}
if ((win_probe == 0) && (num_marked || num_abandoned)) {
/* We don't apply penalty to window probe scenarios */
/* JRS - Use the congestion control given in the CC module */
stcb->asoc.cc_functions.sctp_cwnd_update_after_timeout(stcb, net);
}
}
#ifndef INVARIANTS
static void
sctp_recover_sent_list(struct sctp_tcb *stcb)
{
struct sctp_tmit_chunk *chk, *nchk;
struct sctp_association *asoc;
asoc = &stcb->asoc;
TAILQ_FOREACH_SAFE(chk, &asoc->sent_queue, sctp_next, nchk) {
if (SCTP_TSN_GE(asoc->last_acked_seq, chk->rec.data.TSN_seq)) {
SCTP_PRINTF("Found chk:%p tsn:%x <= last_acked_seq:%x\n",
(void *)chk, chk->rec.data.TSN_seq, asoc->last_acked_seq);
if (chk->sent != SCTP_DATAGRAM_NR_ACKED) {
if (asoc->strmout[chk->rec.data.stream_number].chunks_on_queues > 0) {
asoc->strmout[chk->rec.data.stream_number].chunks_on_queues--;
}
}
TAILQ_REMOVE(&asoc->sent_queue, chk, sctp_next);
if (PR_SCTP_ENABLED(chk->flags)) {
if (asoc->pr_sctp_cnt != 0)
asoc->pr_sctp_cnt--;
}
if (chk->data) {
/* sa_ignore NO_NULL_CHK */
sctp_free_bufspace(stcb, asoc, chk, 1);
sctp_m_freem(chk->data);
chk->data = NULL;
if (asoc->prsctp_supported && PR_SCTP_BUF_ENABLED(chk->flags)) {
asoc->sent_queue_cnt_removeable--;
}
}
asoc->sent_queue_cnt--;
sctp_free_a_chunk(stcb, chk, SCTP_SO_NOT_LOCKED);
}
}
SCTP_PRINTF("after recover order is as follows\n");
TAILQ_FOREACH(chk, &asoc->sent_queue, sctp_next) {
SCTP_PRINTF("chk:%p TSN:%x\n", (void *)chk, chk->rec.data.TSN_seq);
}
}
#endif
static int
sctp_mark_all_for_resend(struct sctp_tcb *stcb,
struct sctp_nets *net,
struct sctp_nets *alt,
int window_probe,
int *num_marked,
int *num_abandoned)
{
/*
* Mark all chunks (well not all) that were sent to *net for
* retransmission. Move them to alt for there destination as well...
* We only mark chunks that have been outstanding long enough to
* have received feed-back.
*/
struct sctp_tmit_chunk *chk, *nchk;
struct sctp_nets *lnets;
struct timeval now, min_wait, tv;
int cur_rto;
int cnt_abandoned;
int audit_tf, num_mk, fir;
unsigned int cnt_mk;
uint32_t orig_flight, orig_tf;
uint32_t tsnlast, tsnfirst;
int recovery_cnt = 0;
/* none in flight now */
audit_tf = 0;
fir = 0;
/*
* figure out how long a data chunk must be pending before we can
* mark it ..
*/
(void)SCTP_GETTIME_TIMEVAL(&now);
/* get cur rto in micro-seconds */
cur_rto = (net->lastsa >> SCTP_RTT_SHIFT) + net->lastsv;
cur_rto *= 1000;
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_FR_LOGGING_ENABLE) {
sctp_log_fr(cur_rto,
stcb->asoc.peers_rwnd,
window_probe,
SCTP_FR_T3_MARK_TIME);
sctp_log_fr(net->flight_size, 0, 0, SCTP_FR_CWND_REPORT);
sctp_log_fr(net->flight_size, net->cwnd, stcb->asoc.total_flight, SCTP_FR_CWND_REPORT);
}
tv.tv_sec = cur_rto / 1000000;
tv.tv_usec = cur_rto % 1000000;
min_wait = now;
timevalsub(&min_wait, &tv);
if (min_wait.tv_sec < 0 || min_wait.tv_usec < 0) {
/*
* if we hit here, we don't have enough seconds on the clock
* to account for the RTO. We just let the lower seconds be
* the bounds and don't worry about it. This may mean we
* will mark a lot more than we should.
*/
min_wait.tv_sec = min_wait.tv_usec = 0;
}
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_FR_LOGGING_ENABLE) {
sctp_log_fr(cur_rto, now.tv_sec, now.tv_usec, SCTP_FR_T3_MARK_TIME);
sctp_log_fr(0, min_wait.tv_sec, min_wait.tv_usec, SCTP_FR_T3_MARK_TIME);
}
/*
* Our rwnd will be incorrect here since we are not adding back the
* cnt * mbuf but we will fix that down below.
*/
orig_flight = net->flight_size;
orig_tf = stcb->asoc.total_flight;
net->fast_retran_ip = 0;
/* Now on to each chunk */
cnt_abandoned = 0;
num_mk = cnt_mk = 0;
tsnfirst = tsnlast = 0;
#ifndef INVARIANTS
start_again:
#endif
TAILQ_FOREACH_SAFE(chk, &stcb->asoc.sent_queue, sctp_next, nchk) {
if (SCTP_TSN_GE(stcb->asoc.last_acked_seq, chk->rec.data.TSN_seq)) {
/* Strange case our list got out of order? */
2013-10-29 20:04:50 +00:00
SCTP_PRINTF("Our list is out of order? last_acked:%x chk:%x\n",
(unsigned int)stcb->asoc.last_acked_seq, (unsigned int)chk->rec.data.TSN_seq);
recovery_cnt++;
#ifdef INVARIANTS
panic("last acked >= chk on sent-Q");
#else
SCTP_PRINTF("Recover attempts a restart cnt:%d\n", recovery_cnt);
sctp_recover_sent_list(stcb);
if (recovery_cnt < 10) {
goto start_again;
} else {
SCTP_PRINTF("Recovery fails %d times??\n", recovery_cnt);
}
#endif
}
if ((chk->whoTo == net) && (chk->sent < SCTP_DATAGRAM_ACKED)) {
/*
* found one to mark: If it is less than
* DATAGRAM_ACKED it MUST not be a skipped or marked
* TSN but instead one that is either already set
* for retransmission OR one that needs
* retransmission.
*/
/* validate its been outstanding long enough */
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_FR_LOGGING_ENABLE) {
sctp_log_fr(chk->rec.data.TSN_seq,
chk->sent_rcv_time.tv_sec,
chk->sent_rcv_time.tv_usec,
SCTP_FR_T3_MARK_TIME);
}
if ((chk->sent_rcv_time.tv_sec > min_wait.tv_sec) && (window_probe == 0)) {
/*
* we have reached a chunk that was sent
* some seconds past our min.. forget it we
* will find no more to send.
*/
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_FR_LOGGING_ENABLE) {
sctp_log_fr(0,
chk->sent_rcv_time.tv_sec,
chk->sent_rcv_time.tv_usec,
SCTP_FR_T3_STOPPED);
}
continue;
} else if ((chk->sent_rcv_time.tv_sec == min_wait.tv_sec) &&
(window_probe == 0)) {
/*
* we must look at the micro seconds to
* know.
*/
if (chk->sent_rcv_time.tv_usec >= min_wait.tv_usec) {
/*
* ok it was sent after our boundary
* time.
*/
continue;
}
}
if (stcb->asoc.prsctp_supported && PR_SCTP_TTL_ENABLED(chk->flags)) {
/* Is it expired? */
if (timevalcmp(&now, &chk->rec.data.timetodrop, >)) {
/* Yes so drop it */
if (chk->data) {
(void)sctp_release_pr_sctp_chunk(stcb,
chk,
1,
SCTP_SO_NOT_LOCKED);
cnt_abandoned++;
}
continue;
}
}
if (stcb->asoc.prsctp_supported && PR_SCTP_RTX_ENABLED(chk->flags)) {
/* Has it been retransmitted tv_sec times? */
if (chk->snd_count > chk->rec.data.timetodrop.tv_sec) {
if (chk->data) {
(void)sctp_release_pr_sctp_chunk(stcb,
chk,
1,
SCTP_SO_NOT_LOCKED);
cnt_abandoned++;
}
continue;
}
}
if (chk->sent < SCTP_DATAGRAM_RESEND) {
sctp_ucount_incr(stcb->asoc.sent_queue_retran_cnt);
num_mk++;
if (fir == 0) {
fir = 1;
tsnfirst = chk->rec.data.TSN_seq;
}
tsnlast = chk->rec.data.TSN_seq;
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_FR_LOGGING_ENABLE) {
sctp_log_fr(chk->rec.data.TSN_seq, chk->snd_count,
0, SCTP_FR_T3_MARKED);
}
if (chk->rec.data.chunk_was_revoked) {
/* deflate the cwnd */
chk->whoTo->cwnd -= chk->book_size;
chk->rec.data.chunk_was_revoked = 0;
}
net->marked_retrans++;
stcb->asoc.marked_retrans++;
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_FLIGHT_LOGGING_ENABLE) {
sctp_misc_ints(SCTP_FLIGHT_LOG_DOWN_RSND_TO,
chk->whoTo->flight_size,
chk->book_size,
(uintptr_t) chk->whoTo,
chk->rec.data.TSN_seq);
}
sctp_flight_size_decrease(chk);
sctp_total_flight_decrease(stcb, chk);
stcb->asoc.peers_rwnd += chk->send_size;
stcb->asoc.peers_rwnd += SCTP_BASE_SYSCTL(sctp_peer_chunk_oh);
}
chk->sent = SCTP_DATAGRAM_RESEND;
SCTP_STAT_INCR(sctps_markedretrans);
/* reset the TSN for striking and other FR stuff */
chk->rec.data.doing_fast_retransmit = 0;
/* Clear any time so NO RTT is being done */
if (chk->do_rtt) {
if (chk->whoTo->rto_needed == 0) {
chk->whoTo->rto_needed = 1;
}
}
chk->do_rtt = 0;
if (alt != net) {
sctp_free_remote_addr(chk->whoTo);
chk->no_fr_allowed = 1;
chk->whoTo = alt;
atomic_add_int(&alt->ref_count, 1);
} else {
chk->no_fr_allowed = 0;
if (TAILQ_EMPTY(&stcb->asoc.send_queue)) {
chk->rec.data.fast_retran_tsn = stcb->asoc.sending_seq;
} else {
chk->rec.data.fast_retran_tsn = (TAILQ_FIRST(&stcb->asoc.send_queue))->rec.data.TSN_seq;
}
}
/*
* CMT: Do not allow FRs on retransmitted TSNs.
*/
if (stcb->asoc.sctp_cmt_on_off > 0) {
chk->no_fr_allowed = 1;
}
#ifdef THIS_SHOULD_NOT_BE_DONE
} else if (chk->sent == SCTP_DATAGRAM_ACKED) {
/* remember highest acked one */
could_be_sent = chk;
#endif
}
if (chk->sent == SCTP_DATAGRAM_RESEND) {
cnt_mk++;
}
}
if ((orig_flight - net->flight_size) != (orig_tf - stcb->asoc.total_flight)) {
/* we did not subtract the same things? */
audit_tf = 1;
}
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_FR_LOGGING_ENABLE) {
sctp_log_fr(tsnfirst, tsnlast, num_mk, SCTP_FR_T3_TIMEOUT);
}
#ifdef SCTP_DEBUG
if (num_mk) {
SCTPDBG(SCTP_DEBUG_TIMER1, "LAST TSN marked was %x\n",
tsnlast);
SCTPDBG(SCTP_DEBUG_TIMER1, "Num marked for retransmission was %d peer-rwd:%ld\n",
num_mk, (u_long)stcb->asoc.peers_rwnd);
SCTPDBG(SCTP_DEBUG_TIMER1, "LAST TSN marked was %x\n",
tsnlast);
SCTPDBG(SCTP_DEBUG_TIMER1, "Num marked for retransmission was %d peer-rwd:%d\n",
num_mk,
(int)stcb->asoc.peers_rwnd);
}
#endif
*num_marked = num_mk;
*num_abandoned = cnt_abandoned;
/*
* Now check for a ECN Echo that may be stranded And include the
* cnt_mk'd to have all resends in the control queue.
*/
TAILQ_FOREACH(chk, &stcb->asoc.control_send_queue, sctp_next) {
if (chk->sent == SCTP_DATAGRAM_RESEND) {
cnt_mk++;
}
if ((chk->whoTo == net) &&
(chk->rec.chunk_id.id == SCTP_ECN_ECHO)) {
sctp_free_remote_addr(chk->whoTo);
chk->whoTo = alt;
if (chk->sent != SCTP_DATAGRAM_RESEND) {
chk->sent = SCTP_DATAGRAM_RESEND;
sctp_ucount_incr(stcb->asoc.sent_queue_retran_cnt);
cnt_mk++;
}
atomic_add_int(&alt->ref_count, 1);
}
}
#ifdef THIS_SHOULD_NOT_BE_DONE
if ((stcb->asoc.sent_queue_retran_cnt == 0) && (could_be_sent)) {
/* fix it so we retransmit the highest acked anyway */
sctp_ucount_incr(stcb->asoc.sent_queue_retran_cnt);
cnt_mk++;
could_be_sent->sent = SCTP_DATAGRAM_RESEND;
}
#endif
if (stcb->asoc.sent_queue_retran_cnt != cnt_mk) {
#ifdef INVARIANTS
SCTP_PRINTF("Local Audit says there are %d for retran asoc cnt:%d we marked:%d this time\n",
cnt_mk, stcb->asoc.sent_queue_retran_cnt, num_mk);
#endif
#ifndef SCTP_AUDITING_ENABLED
stcb->asoc.sent_queue_retran_cnt = cnt_mk;
#endif
}
if (audit_tf) {
SCTPDBG(SCTP_DEBUG_TIMER4,
"Audit total flight due to negative value net:%p\n",
(void *)net);
stcb->asoc.total_flight = 0;
stcb->asoc.total_flight_count = 0;
/* Clear all networks flight size */
TAILQ_FOREACH(lnets, &stcb->asoc.nets, sctp_next) {
lnets->flight_size = 0;
SCTPDBG(SCTP_DEBUG_TIMER4,
"Net:%p c-f cwnd:%d ssthresh:%d\n",
(void *)lnets, lnets->cwnd, lnets->ssthresh);
}
TAILQ_FOREACH(chk, &stcb->asoc.sent_queue, sctp_next) {
if (chk->sent < SCTP_DATAGRAM_RESEND) {
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_FLIGHT_LOGGING_ENABLE) {
sctp_misc_ints(SCTP_FLIGHT_LOG_UP,
chk->whoTo->flight_size,
chk->book_size,
(uintptr_t) chk->whoTo,
chk->rec.data.TSN_seq);
}
sctp_flight_size_increase(chk);
sctp_total_flight_increase(stcb, chk);
}
}
}
/* We return 1 if we only have a window probe outstanding */
return (0);
}
int
sctp_t3rxt_timer(struct sctp_inpcb *inp,
struct sctp_tcb *stcb,
struct sctp_nets *net)
{
struct sctp_nets *alt;
int win_probe, num_mk, num_abandoned;
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_FR_LOGGING_ENABLE) {
sctp_log_fr(0, 0, 0, SCTP_FR_T3_TIMEOUT);
}
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) {
struct sctp_nets *lnet;
TAILQ_FOREACH(lnet, &stcb->asoc.nets, sctp_next) {
if (net == lnet) {
sctp_log_cwnd(stcb, lnet, 1, SCTP_CWND_LOG_FROM_T3);
} else {
sctp_log_cwnd(stcb, lnet, 0, SCTP_CWND_LOG_FROM_T3);
}
}
}
/* Find an alternate and mark those for retransmission */
if ((stcb->asoc.peers_rwnd == 0) &&
(stcb->asoc.total_flight < net->mtu)) {
SCTP_STAT_INCR(sctps_timowindowprobe);
win_probe = 1;
} else {
win_probe = 0;
}
if (win_probe == 0) {
/* We don't do normal threshold management on window probes */
if (sctp_threshold_management(inp, stcb, net,
stcb->asoc.max_send_times)) {
/* Association was destroyed */
return (1);
} else {
if (net != stcb->asoc.primary_destination) {
/* send a immediate HB if our RTO is stale */
struct timeval now;
unsigned int ms_goneby;
(void)SCTP_GETTIME_TIMEVAL(&now);
if (net->last_sent_time.tv_sec) {
ms_goneby = (now.tv_sec - net->last_sent_time.tv_sec) * 1000;
} else {
ms_goneby = 0;
}
if ((net->dest_state & SCTP_ADDR_PF) == 0) {
if ((ms_goneby > net->RTO) || (net->RTO == 0)) {
/*
* no recent feed back in an
* RTO or more, request a
* RTT update
*/
sctp_send_hb(stcb, net, SCTP_SO_NOT_LOCKED);
}
}
}
}
} else {
/*
* For a window probe we don't penalize the net's but only
* the association. This may fail it if SACKs are not coming
* back. If sack's are coming with rwnd locked at 0, we will
* continue to hold things waiting for rwnd to raise
*/
if (sctp_threshold_management(inp, stcb, NULL,
stcb->asoc.max_send_times)) {
/* Association was destroyed */
return (1);
}
}
if (stcb->asoc.sctp_cmt_on_off > 0) {
if (net->pf_threshold < net->failure_threshold) {
alt = sctp_find_alternate_net(stcb, net, 2);
} else {
/*
* CMT: Using RTX_SSTHRESH policy for CMT. If CMT is
* being used, then pick dest with largest ssthresh
* for any retransmission.
*/
alt = sctp_find_alternate_net(stcb, net, 1);
/*
* CUCv2: If a different dest is picked for the
* retransmission, then new (rtx-)pseudo_cumack
* needs to be tracked for orig dest. Let CUCv2
* track new (rtx-) pseudo-cumack always.
*/
net->find_pseudo_cumack = 1;
net->find_rtx_pseudo_cumack = 1;
}
} else {
alt = sctp_find_alternate_net(stcb, net, 0);
}
num_mk = 0;
num_abandoned = 0;
(void)sctp_mark_all_for_resend(stcb, net, alt, win_probe,
&num_mk, &num_abandoned);
/* FR Loss recovery just ended with the T3. */
stcb->asoc.fast_retran_loss_recovery = 0;
/* CMT FR loss recovery ended with the T3 */
net->fast_retran_loss_recovery = 0;
if ((stcb->asoc.cc_functions.sctp_cwnd_new_transmission_begins) &&
(net->flight_size == 0)) {
(*stcb->asoc.cc_functions.sctp_cwnd_new_transmission_begins) (stcb, net);
}
/*
* setup the sat loss recovery that prevents satellite cwnd advance.
*/
stcb->asoc.sat_t3_loss_recovery = 1;
stcb->asoc.sat_t3_recovery_tsn = stcb->asoc.sending_seq;
/* Backoff the timer and cwnd */
sctp_backoff_on_timeout(stcb, net, win_probe, num_mk, num_abandoned);
if ((!(net->dest_state & SCTP_ADDR_REACHABLE)) ||
(net->dest_state & SCTP_ADDR_PF)) {
/* Move all pending over too */
sctp_move_chunks_from_net(stcb, net);
/*
* Get the address that failed, to force a new src address
* selecton and a route allocation.
*/
if (net->ro._s_addr) {
sctp_free_ifa(net->ro._s_addr);
net->ro._s_addr = NULL;
}
net->src_addr_selected = 0;
/* Force a route allocation too */
if (net->ro.ro_rt) {
RTFREE(net->ro.ro_rt);
net->ro.ro_rt = NULL;
}
/* Was it our primary? */
if ((stcb->asoc.primary_destination == net) && (alt != net)) {
/*
* Yes, note it as such and find an alternate note:
* this means HB code must use this to resent the
* primary if it goes active AND if someone does a
* change-primary then this flag must be cleared
* from any net structures.
*/
if (stcb->asoc.alternate) {
sctp_free_remote_addr(stcb->asoc.alternate);
}
stcb->asoc.alternate = alt;
atomic_add_int(&stcb->asoc.alternate->ref_count, 1);
}
}
/*
* Special case for cookie-echo'ed case, we don't do output but must
* await the COOKIE-ACK before retransmission
*/
if (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_COOKIE_ECHOED) {
/*
* Here we just reset the timer and start again since we
* have not established the asoc
*/
sctp_timer_start(SCTP_TIMER_TYPE_SEND, inp, stcb, net);
return (0);
}
if (stcb->asoc.prsctp_supported) {
struct sctp_tmit_chunk *lchk;
lchk = sctp_try_advance_peer_ack_point(stcb, &stcb->asoc);
/* C3. See if we need to send a Fwd-TSN */
if (SCTP_TSN_GT(stcb->asoc.advanced_peer_ack_point, stcb->asoc.last_acked_seq)) {
send_forward_tsn(stcb, &stcb->asoc);
if (lchk) {
/* Assure a timer is up */
sctp_timer_start(SCTP_TIMER_TYPE_SEND, stcb->sctp_ep, stcb, lchk->whoTo);
}
}
}
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) {
sctp_log_cwnd(stcb, net, net->cwnd, SCTP_CWND_LOG_FROM_RTX);
}
return (0);
}
int
sctp_t1init_timer(struct sctp_inpcb *inp,
struct sctp_tcb *stcb,
struct sctp_nets *net)
{
/* bump the thresholds */
if (stcb->asoc.delayed_connection) {
/*
* special hook for delayed connection. The library did NOT
* complete the rest of its sends.
*/
stcb->asoc.delayed_connection = 0;
sctp_send_initiate(inp, stcb, SCTP_SO_NOT_LOCKED);
return (0);
}
if (SCTP_GET_STATE((&stcb->asoc)) != SCTP_STATE_COOKIE_WAIT) {
return (0);
}
if (sctp_threshold_management(inp, stcb, net,
stcb->asoc.max_init_times)) {
/* Association was destroyed */
return (1);
}
stcb->asoc.dropped_special_cnt = 0;
sctp_backoff_on_timeout(stcb, stcb->asoc.primary_destination, 1, 0, 0);
if (stcb->asoc.initial_init_rto_max < net->RTO) {
net->RTO = stcb->asoc.initial_init_rto_max;
}
if (stcb->asoc.numnets > 1) {
/* If we have more than one addr use it */
struct sctp_nets *alt;
alt = sctp_find_alternate_net(stcb, stcb->asoc.primary_destination, 0);
if (alt != stcb->asoc.primary_destination) {
sctp_move_chunks_from_net(stcb, stcb->asoc.primary_destination);
stcb->asoc.primary_destination = alt;
}
}
/* Send out a new init */
sctp_send_initiate(inp, stcb, SCTP_SO_NOT_LOCKED);
return (0);
}
/*
* For cookie and asconf we actually need to find and mark for resend, then
* increment the resend counter (after all the threshold management stuff of
* course).
*/
int
sctp_cookie_timer(struct sctp_inpcb *inp,
struct sctp_tcb *stcb,
struct sctp_nets *net SCTP_UNUSED)
{
struct sctp_nets *alt;
struct sctp_tmit_chunk *cookie;
/* first before all else we must find the cookie */
TAILQ_FOREACH(cookie, &stcb->asoc.control_send_queue, sctp_next) {
if (cookie->rec.chunk_id.id == SCTP_COOKIE_ECHO) {
break;
}
}
if (cookie == NULL) {
if (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_COOKIE_ECHOED) {
/* FOOBAR! */
struct mbuf *op_err;
op_err = sctp_generate_cause(SCTP_CAUSE_PROTOCOL_VIOLATION,
"Cookie timer expired, but no cookie");
inp->last_abort_code = SCTP_FROM_SCTP_TIMER + SCTP_LOC_3;
sctp_abort_an_association(inp, stcb, op_err, SCTP_SO_NOT_LOCKED);
} else {
#ifdef INVARIANTS
panic("Cookie timer expires in wrong state?");
#else
SCTP_PRINTF("Strange in state %d not cookie-echoed yet c-e timer expires?\n", SCTP_GET_STATE(&stcb->asoc));
return (0);
#endif
}
return (0);
}
/* Ok we found the cookie, threshold management next */
if (sctp_threshold_management(inp, stcb, cookie->whoTo,
stcb->asoc.max_init_times)) {
/* Assoc is over */
return (1);
}
/*
* cleared theshold management now lets backoff the address & select
* an alternate
*/
stcb->asoc.dropped_special_cnt = 0;
sctp_backoff_on_timeout(stcb, cookie->whoTo, 1, 0, 0);
alt = sctp_find_alternate_net(stcb, cookie->whoTo, 0);
if (alt != cookie->whoTo) {
sctp_free_remote_addr(cookie->whoTo);
cookie->whoTo = alt;
atomic_add_int(&alt->ref_count, 1);
}
/* Now mark the retran info */
if (cookie->sent != SCTP_DATAGRAM_RESEND) {
sctp_ucount_incr(stcb->asoc.sent_queue_retran_cnt);
}
cookie->sent = SCTP_DATAGRAM_RESEND;
/*
* Now call the output routine to kick out the cookie again, Note we
* don't mark any chunks for retran so that FR will need to kick in
* to move these (or a send timer).
*/
return (0);
}
int
sctp_strreset_timer(struct sctp_inpcb *inp, struct sctp_tcb *stcb,
struct sctp_nets *net)
{
struct sctp_nets *alt;
struct sctp_tmit_chunk *strrst = NULL, *chk = NULL;
if (stcb->asoc.stream_reset_outstanding == 0) {
return (0);
}
/* find the existing STRRESET, we use the seq number we sent out on */
(void)sctp_find_stream_reset(stcb, stcb->asoc.str_reset_seq_out, &strrst);
if (strrst == NULL) {
return (0);
}
/* do threshold management */
if (sctp_threshold_management(inp, stcb, strrst->whoTo,
stcb->asoc.max_send_times)) {
/* Assoc is over */
return (1);
}
/*
* cleared theshold management now lets backoff the address & select
* an alternate
*/
sctp_backoff_on_timeout(stcb, strrst->whoTo, 1, 0, 0);
alt = sctp_find_alternate_net(stcb, strrst->whoTo, 0);
sctp_free_remote_addr(strrst->whoTo);
strrst->whoTo = alt;
atomic_add_int(&alt->ref_count, 1);
/* See if a ECN Echo is also stranded */
TAILQ_FOREACH(chk, &stcb->asoc.control_send_queue, sctp_next) {
if ((chk->whoTo == net) &&
(chk->rec.chunk_id.id == SCTP_ECN_ECHO)) {
sctp_free_remote_addr(chk->whoTo);
if (chk->sent != SCTP_DATAGRAM_RESEND) {
chk->sent = SCTP_DATAGRAM_RESEND;
sctp_ucount_incr(stcb->asoc.sent_queue_retran_cnt);
}
chk->whoTo = alt;
atomic_add_int(&alt->ref_count, 1);
}
}
if (!(net->dest_state & SCTP_ADDR_REACHABLE)) {
/*
* If the address went un-reachable, we need to move to
* alternates for ALL chk's in queue
*/
sctp_move_chunks_from_net(stcb, net);
}
/* mark the retran info */
if (strrst->sent != SCTP_DATAGRAM_RESEND)
sctp_ucount_incr(stcb->asoc.sent_queue_retran_cnt);
strrst->sent = SCTP_DATAGRAM_RESEND;
/* restart the timer */
sctp_timer_start(SCTP_TIMER_TYPE_STRRESET, inp, stcb, strrst->whoTo);
return (0);
}
int
sctp_asconf_timer(struct sctp_inpcb *inp, struct sctp_tcb *stcb,
struct sctp_nets *net)
{
struct sctp_nets *alt;
struct sctp_tmit_chunk *asconf, *chk;
- take out a needless panic under invariants for sctp_output.c - Fix addrs's error checking of sctp_sendx(3) when addrcnt is less than SCTP_SMALL_IOVEC_SIZE - re-add back inpcb_bind local address check bypass capability - Fix it so sctp_opt_info is independant of assoc_id postion. - Fix cookie life set to use MSEC_TO_TICKS() macro. - asconf changes o More comment changes/clarifications related to the old local address "not" list which is now an explicit restricted list. o Rename some functions for clarity: - sctp_add/del_local_addr_assoc to xxx_local_addr_restricted() - asconf related iterator functions to sctp_asconf_iterator_xxx() o Fix bug when the same address is deleted and added (and removed from the asconf queue) where the ifa is "freed" twice refcount wise, possibly freeing it completely. o Fix bug in output where the first ASCONF would not go out after the last address is changed (e.g. only goes out when retransmitted). o Fix bug where multiple ASCONFs can be bundled in the same packet with the and with the same serial numbers. o Fix asconf stcb iterator to not send ASCONF until after all work queue entries have been processed. o Change behavior so that when the last address is deleted (auto asconf on a bound all endpoint) no action is taken until an address is added; at that time, an ASCONF add+delete is sent (if the assoc is still up). o Fix local address counting so that address scoping is taken into account. o #ifdef SCTP_TIMER_BASED_ASCONF the old timer triggered sending of ASCONF (after an RTO). The default now is to send ASCONF immediately (except for the case of changing/deleting the last usable address). Approved by: re(ken smith)@freebsd.org
2007-07-24 20:06:02 +00:00
/* is this a first send, or a retransmission? */
if (TAILQ_EMPTY(&stcb->asoc.asconf_send_queue)) {
/* compose a new ASCONF chunk and send it */
sctp_send_asconf(stcb, net, SCTP_ADDR_NOT_LOCKED);
} else {
- take out a needless panic under invariants for sctp_output.c - Fix addrs's error checking of sctp_sendx(3) when addrcnt is less than SCTP_SMALL_IOVEC_SIZE - re-add back inpcb_bind local address check bypass capability - Fix it so sctp_opt_info is independant of assoc_id postion. - Fix cookie life set to use MSEC_TO_TICKS() macro. - asconf changes o More comment changes/clarifications related to the old local address "not" list which is now an explicit restricted list. o Rename some functions for clarity: - sctp_add/del_local_addr_assoc to xxx_local_addr_restricted() - asconf related iterator functions to sctp_asconf_iterator_xxx() o Fix bug when the same address is deleted and added (and removed from the asconf queue) where the ifa is "freed" twice refcount wise, possibly freeing it completely. o Fix bug in output where the first ASCONF would not go out after the last address is changed (e.g. only goes out when retransmitted). o Fix bug where multiple ASCONFs can be bundled in the same packet with the and with the same serial numbers. o Fix asconf stcb iterator to not send ASCONF until after all work queue entries have been processed. o Change behavior so that when the last address is deleted (auto asconf on a bound all endpoint) no action is taken until an address is added; at that time, an ASCONF add+delete is sent (if the assoc is still up). o Fix local address counting so that address scoping is taken into account. o #ifdef SCTP_TIMER_BASED_ASCONF the old timer triggered sending of ASCONF (after an RTO). The default now is to send ASCONF immediately (except for the case of changing/deleting the last usable address). Approved by: re(ken smith)@freebsd.org
2007-07-24 20:06:02 +00:00
/*
* Retransmission of the existing ASCONF is needed
*/
/* find the existing ASCONF */
asconf = TAILQ_FIRST(&stcb->asoc.asconf_send_queue);
if (asconf == NULL) {
return (0);
}
/* do threshold management */
if (sctp_threshold_management(inp, stcb, asconf->whoTo,
stcb->asoc.max_send_times)) {
/* Assoc is over */
return (1);
}
if (asconf->snd_count > stcb->asoc.max_send_times) {
/*
- take out a needless panic under invariants for sctp_output.c - Fix addrs's error checking of sctp_sendx(3) when addrcnt is less than SCTP_SMALL_IOVEC_SIZE - re-add back inpcb_bind local address check bypass capability - Fix it so sctp_opt_info is independant of assoc_id postion. - Fix cookie life set to use MSEC_TO_TICKS() macro. - asconf changes o More comment changes/clarifications related to the old local address "not" list which is now an explicit restricted list. o Rename some functions for clarity: - sctp_add/del_local_addr_assoc to xxx_local_addr_restricted() - asconf related iterator functions to sctp_asconf_iterator_xxx() o Fix bug when the same address is deleted and added (and removed from the asconf queue) where the ifa is "freed" twice refcount wise, possibly freeing it completely. o Fix bug in output where the first ASCONF would not go out after the last address is changed (e.g. only goes out when retransmitted). o Fix bug where multiple ASCONFs can be bundled in the same packet with the and with the same serial numbers. o Fix asconf stcb iterator to not send ASCONF until after all work queue entries have been processed. o Change behavior so that when the last address is deleted (auto asconf on a bound all endpoint) no action is taken until an address is added; at that time, an ASCONF add+delete is sent (if the assoc is still up). o Fix local address counting so that address scoping is taken into account. o #ifdef SCTP_TIMER_BASED_ASCONF the old timer triggered sending of ASCONF (after an RTO). The default now is to send ASCONF immediately (except for the case of changing/deleting the last usable address). Approved by: re(ken smith)@freebsd.org
2007-07-24 20:06:02 +00:00
* Something is rotten: our peer is not responding
* to ASCONFs but apparently is to other chunks.
* i.e. it is not properly handling the chunk type
* upper bits. Mark this peer as ASCONF incapable
* and cleanup.
*/
SCTPDBG(SCTP_DEBUG_TIMER1, "asconf_timer: Peer has not responded to our repeated ASCONFs\n");
sctp_asconf_cleanup(stcb, net);
return (0);
}
/*
- take out a needless panic under invariants for sctp_output.c - Fix addrs's error checking of sctp_sendx(3) when addrcnt is less than SCTP_SMALL_IOVEC_SIZE - re-add back inpcb_bind local address check bypass capability - Fix it so sctp_opt_info is independant of assoc_id postion. - Fix cookie life set to use MSEC_TO_TICKS() macro. - asconf changes o More comment changes/clarifications related to the old local address "not" list which is now an explicit restricted list. o Rename some functions for clarity: - sctp_add/del_local_addr_assoc to xxx_local_addr_restricted() - asconf related iterator functions to sctp_asconf_iterator_xxx() o Fix bug when the same address is deleted and added (and removed from the asconf queue) where the ifa is "freed" twice refcount wise, possibly freeing it completely. o Fix bug in output where the first ASCONF would not go out after the last address is changed (e.g. only goes out when retransmitted). o Fix bug where multiple ASCONFs can be bundled in the same packet with the and with the same serial numbers. o Fix asconf stcb iterator to not send ASCONF until after all work queue entries have been processed. o Change behavior so that when the last address is deleted (auto asconf on a bound all endpoint) no action is taken until an address is added; at that time, an ASCONF add+delete is sent (if the assoc is still up). o Fix local address counting so that address scoping is taken into account. o #ifdef SCTP_TIMER_BASED_ASCONF the old timer triggered sending of ASCONF (after an RTO). The default now is to send ASCONF immediately (except for the case of changing/deleting the last usable address). Approved by: re(ken smith)@freebsd.org
2007-07-24 20:06:02 +00:00
* cleared threshold management, so now backoff the net and
* select an alternate
*/
sctp_backoff_on_timeout(stcb, asconf->whoTo, 1, 0, 0);
alt = sctp_find_alternate_net(stcb, asconf->whoTo, 0);
if (asconf->whoTo != alt) {
sctp_free_remote_addr(asconf->whoTo);
asconf->whoTo = alt;
atomic_add_int(&alt->ref_count, 1);
}
- take out a needless panic under invariants for sctp_output.c - Fix addrs's error checking of sctp_sendx(3) when addrcnt is less than SCTP_SMALL_IOVEC_SIZE - re-add back inpcb_bind local address check bypass capability - Fix it so sctp_opt_info is independant of assoc_id postion. - Fix cookie life set to use MSEC_TO_TICKS() macro. - asconf changes o More comment changes/clarifications related to the old local address "not" list which is now an explicit restricted list. o Rename some functions for clarity: - sctp_add/del_local_addr_assoc to xxx_local_addr_restricted() - asconf related iterator functions to sctp_asconf_iterator_xxx() o Fix bug when the same address is deleted and added (and removed from the asconf queue) where the ifa is "freed" twice refcount wise, possibly freeing it completely. o Fix bug in output where the first ASCONF would not go out after the last address is changed (e.g. only goes out when retransmitted). o Fix bug where multiple ASCONFs can be bundled in the same packet with the and with the same serial numbers. o Fix asconf stcb iterator to not send ASCONF until after all work queue entries have been processed. o Change behavior so that when the last address is deleted (auto asconf on a bound all endpoint) no action is taken until an address is added; at that time, an ASCONF add+delete is sent (if the assoc is still up). o Fix local address counting so that address scoping is taken into account. o #ifdef SCTP_TIMER_BASED_ASCONF the old timer triggered sending of ASCONF (after an RTO). The default now is to send ASCONF immediately (except for the case of changing/deleting the last usable address). Approved by: re(ken smith)@freebsd.org
2007-07-24 20:06:02 +00:00
/* See if an ECN Echo is also stranded */
TAILQ_FOREACH(chk, &stcb->asoc.control_send_queue, sctp_next) {
if ((chk->whoTo == net) &&
(chk->rec.chunk_id.id == SCTP_ECN_ECHO)) {
sctp_free_remote_addr(chk->whoTo);
chk->whoTo = alt;
if (chk->sent != SCTP_DATAGRAM_RESEND) {
chk->sent = SCTP_DATAGRAM_RESEND;
sctp_ucount_incr(stcb->asoc.sent_queue_retran_cnt);
}
atomic_add_int(&alt->ref_count, 1);
}
}
TAILQ_FOREACH(chk, &stcb->asoc.asconf_send_queue, sctp_next) {
if (chk->whoTo != alt) {
sctp_free_remote_addr(chk->whoTo);
chk->whoTo = alt;
atomic_add_int(&alt->ref_count, 1);
}
if (asconf->sent != SCTP_DATAGRAM_RESEND && chk->sent != SCTP_DATAGRAM_UNSENT)
sctp_ucount_incr(stcb->asoc.sent_queue_retran_cnt);
chk->sent = SCTP_DATAGRAM_RESEND;
}
if (!(net->dest_state & SCTP_ADDR_REACHABLE)) {
/*
* If the address went un-reachable, we need to move
- take out a needless panic under invariants for sctp_output.c - Fix addrs's error checking of sctp_sendx(3) when addrcnt is less than SCTP_SMALL_IOVEC_SIZE - re-add back inpcb_bind local address check bypass capability - Fix it so sctp_opt_info is independant of assoc_id postion. - Fix cookie life set to use MSEC_TO_TICKS() macro. - asconf changes o More comment changes/clarifications related to the old local address "not" list which is now an explicit restricted list. o Rename some functions for clarity: - sctp_add/del_local_addr_assoc to xxx_local_addr_restricted() - asconf related iterator functions to sctp_asconf_iterator_xxx() o Fix bug when the same address is deleted and added (and removed from the asconf queue) where the ifa is "freed" twice refcount wise, possibly freeing it completely. o Fix bug in output where the first ASCONF would not go out after the last address is changed (e.g. only goes out when retransmitted). o Fix bug where multiple ASCONFs can be bundled in the same packet with the and with the same serial numbers. o Fix asconf stcb iterator to not send ASCONF until after all work queue entries have been processed. o Change behavior so that when the last address is deleted (auto asconf on a bound all endpoint) no action is taken until an address is added; at that time, an ASCONF add+delete is sent (if the assoc is still up). o Fix local address counting so that address scoping is taken into account. o #ifdef SCTP_TIMER_BASED_ASCONF the old timer triggered sending of ASCONF (after an RTO). The default now is to send ASCONF immediately (except for the case of changing/deleting the last usable address). Approved by: re(ken smith)@freebsd.org
2007-07-24 20:06:02 +00:00
* to the alternate for ALL chunks in queue
*/
sctp_move_chunks_from_net(stcb, net);
}
/* mark the retran info */
if (asconf->sent != SCTP_DATAGRAM_RESEND)
sctp_ucount_incr(stcb->asoc.sent_queue_retran_cnt);
asconf->sent = SCTP_DATAGRAM_RESEND;
/* send another ASCONF if any and we can do */
sctp_send_asconf(stcb, alt, SCTP_ADDR_NOT_LOCKED);
}
return (0);
}
- send call has a reference to uio->uio_resid in the recent send code, but uio may be NULL on sendfile calls. Change to use sndlen variable. - EMSGSIZE is not being returned in non-blocking mode and needs a small tweak to look if the msg would ever fit when returning EWOULDBLOCK. - FWD-TSN has a bug in stream processing which could cause a panic. This is a follow on to the codenomicon fix. - PDAPI level 1 and 2 do not work unless the reader gets his returned buffer full. Fix so we can break out when at level 1 or 2. - Fix fast-handoff features to copy across properly on accepted sockets - Fix sctp_peeloff() system call when no true system call exists to screen arguments for errors. In cases where a real system call exists the system call itself does this. - Fix raddr leak in recent add-ip code change for bundled asconfs (even when non-bundled asconfs are received) - Make sure ipi_addr lock is held when walking global addr list. Need to change this lock type to a rwlock(). - Add don't wake flag on both input and output when the socket is closing. - When deleting an address verify the interface is correct before allowing the delete to process. This protects panda and unnumbered. - Clean up old sysctl stuff and get rid of the old Open/Net BSD structures. - Add a function to watch the ranges in the sysctl sets. - When appending in the reassembly queue, validate that the assoc has not gone to about to be freed. If so (in the middle) abort out. Note this especially effects MAC I think due to the lock/unlock they do (or with LOCK testing in place). - Netstat patch to get rid of warnings. - Make sure that no data gets queued to inactive/unconfirmed destinations. This especially effect CMT but also makes a impact on regular SCTP as well. - During init collision when we detect seq number out of sync we need to treat it like Case C and discard the cookie (no invarient needed here). - Atomic access to the random store. - When we declare a vtag good, we need to shove it into the time wait hash to prevent further use. When the tag is put into the assoc hash, we need to remove it from the twait hash (where it will surely be). This prevents duplicate tag assignments. - Move decr-ref count to better protect sysctl out of data. - ltrace error corrections in sctp6_usrreq.c - Add hook for interface up/down to be sent to us. - Make sysctl() exported structures independent of processor architecture. - Fix route and src addr cache clearing for delete address case. - Make sure address marked SCTP_DEL_IP_ADDRESS is never selected as src addr. - in icmp handling fixed so we actually look at the icmp codes to figure out what to do. - Modified mobility code. Reception of DELETE IP ADDRESS for a primary destination and SET PRIMARY for a new primary destination is used for retransmission trigger to the new primary destination. Also, in this case, destination of chunks in send_queue are changed to the new primary destination. - Fix so that we disallow sending by mbuf to ever have EEOR mode set upon it. Approved by: re@freebsd.org (B Mah)
2007-09-08 17:48:46 +00:00
/* Mobility adaptation */
void
- send call has a reference to uio->uio_resid in the recent send code, but uio may be NULL on sendfile calls. Change to use sndlen variable. - EMSGSIZE is not being returned in non-blocking mode and needs a small tweak to look if the msg would ever fit when returning EWOULDBLOCK. - FWD-TSN has a bug in stream processing which could cause a panic. This is a follow on to the codenomicon fix. - PDAPI level 1 and 2 do not work unless the reader gets his returned buffer full. Fix so we can break out when at level 1 or 2. - Fix fast-handoff features to copy across properly on accepted sockets - Fix sctp_peeloff() system call when no true system call exists to screen arguments for errors. In cases where a real system call exists the system call itself does this. - Fix raddr leak in recent add-ip code change for bundled asconfs (even when non-bundled asconfs are received) - Make sure ipi_addr lock is held when walking global addr list. Need to change this lock type to a rwlock(). - Add don't wake flag on both input and output when the socket is closing. - When deleting an address verify the interface is correct before allowing the delete to process. This protects panda and unnumbered. - Clean up old sysctl stuff and get rid of the old Open/Net BSD structures. - Add a function to watch the ranges in the sysctl sets. - When appending in the reassembly queue, validate that the assoc has not gone to about to be freed. If so (in the middle) abort out. Note this especially effects MAC I think due to the lock/unlock they do (or with LOCK testing in place). - Netstat patch to get rid of warnings. - Make sure that no data gets queued to inactive/unconfirmed destinations. This especially effect CMT but also makes a impact on regular SCTP as well. - During init collision when we detect seq number out of sync we need to treat it like Case C and discard the cookie (no invarient needed here). - Atomic access to the random store. - When we declare a vtag good, we need to shove it into the time wait hash to prevent further use. When the tag is put into the assoc hash, we need to remove it from the twait hash (where it will surely be). This prevents duplicate tag assignments. - Move decr-ref count to better protect sysctl out of data. - ltrace error corrections in sctp6_usrreq.c - Add hook for interface up/down to be sent to us. - Make sysctl() exported structures independent of processor architecture. - Fix route and src addr cache clearing for delete address case. - Make sure address marked SCTP_DEL_IP_ADDRESS is never selected as src addr. - in icmp handling fixed so we actually look at the icmp codes to figure out what to do. - Modified mobility code. Reception of DELETE IP ADDRESS for a primary destination and SET PRIMARY for a new primary destination is used for retransmission trigger to the new primary destination. Also, in this case, destination of chunks in send_queue are changed to the new primary destination. - Fix so that we disallow sending by mbuf to ever have EEOR mode set upon it. Approved by: re@freebsd.org (B Mah)
2007-09-08 17:48:46 +00:00
sctp_delete_prim_timer(struct sctp_inpcb *inp, struct sctp_tcb *stcb,
struct sctp_nets *net SCTP_UNUSED)
- send call has a reference to uio->uio_resid in the recent send code, but uio may be NULL on sendfile calls. Change to use sndlen variable. - EMSGSIZE is not being returned in non-blocking mode and needs a small tweak to look if the msg would ever fit when returning EWOULDBLOCK. - FWD-TSN has a bug in stream processing which could cause a panic. This is a follow on to the codenomicon fix. - PDAPI level 1 and 2 do not work unless the reader gets his returned buffer full. Fix so we can break out when at level 1 or 2. - Fix fast-handoff features to copy across properly on accepted sockets - Fix sctp_peeloff() system call when no true system call exists to screen arguments for errors. In cases where a real system call exists the system call itself does this. - Fix raddr leak in recent add-ip code change for bundled asconfs (even when non-bundled asconfs are received) - Make sure ipi_addr lock is held when walking global addr list. Need to change this lock type to a rwlock(). - Add don't wake flag on both input and output when the socket is closing. - When deleting an address verify the interface is correct before allowing the delete to process. This protects panda and unnumbered. - Clean up old sysctl stuff and get rid of the old Open/Net BSD structures. - Add a function to watch the ranges in the sysctl sets. - When appending in the reassembly queue, validate that the assoc has not gone to about to be freed. If so (in the middle) abort out. Note this especially effects MAC I think due to the lock/unlock they do (or with LOCK testing in place). - Netstat patch to get rid of warnings. - Make sure that no data gets queued to inactive/unconfirmed destinations. This especially effect CMT but also makes a impact on regular SCTP as well. - During init collision when we detect seq number out of sync we need to treat it like Case C and discard the cookie (no invarient needed here). - Atomic access to the random store. - When we declare a vtag good, we need to shove it into the time wait hash to prevent further use. When the tag is put into the assoc hash, we need to remove it from the twait hash (where it will surely be). This prevents duplicate tag assignments. - Move decr-ref count to better protect sysctl out of data. - ltrace error corrections in sctp6_usrreq.c - Add hook for interface up/down to be sent to us. - Make sysctl() exported structures independent of processor architecture. - Fix route and src addr cache clearing for delete address case. - Make sure address marked SCTP_DEL_IP_ADDRESS is never selected as src addr. - in icmp handling fixed so we actually look at the icmp codes to figure out what to do. - Modified mobility code. Reception of DELETE IP ADDRESS for a primary destination and SET PRIMARY for a new primary destination is used for retransmission trigger to the new primary destination. Also, in this case, destination of chunks in send_queue are changed to the new primary destination. - Fix so that we disallow sending by mbuf to ever have EEOR mode set upon it. Approved by: re@freebsd.org (B Mah)
2007-09-08 17:48:46 +00:00
{
if (stcb->asoc.deleted_primary == NULL) {
SCTPDBG(SCTP_DEBUG_ASCONF1, "delete_prim_timer: deleted_primary is not stored...\n");
sctp_mobility_feature_off(inp, SCTP_MOBILITY_PRIM_DELETED);
return;
- send call has a reference to uio->uio_resid in the recent send code, but uio may be NULL on sendfile calls. Change to use sndlen variable. - EMSGSIZE is not being returned in non-blocking mode and needs a small tweak to look if the msg would ever fit when returning EWOULDBLOCK. - FWD-TSN has a bug in stream processing which could cause a panic. This is a follow on to the codenomicon fix. - PDAPI level 1 and 2 do not work unless the reader gets his returned buffer full. Fix so we can break out when at level 1 or 2. - Fix fast-handoff features to copy across properly on accepted sockets - Fix sctp_peeloff() system call when no true system call exists to screen arguments for errors. In cases where a real system call exists the system call itself does this. - Fix raddr leak in recent add-ip code change for bundled asconfs (even when non-bundled asconfs are received) - Make sure ipi_addr lock is held when walking global addr list. Need to change this lock type to a rwlock(). - Add don't wake flag on both input and output when the socket is closing. - When deleting an address verify the interface is correct before allowing the delete to process. This protects panda and unnumbered. - Clean up old sysctl stuff and get rid of the old Open/Net BSD structures. - Add a function to watch the ranges in the sysctl sets. - When appending in the reassembly queue, validate that the assoc has not gone to about to be freed. If so (in the middle) abort out. Note this especially effects MAC I think due to the lock/unlock they do (or with LOCK testing in place). - Netstat patch to get rid of warnings. - Make sure that no data gets queued to inactive/unconfirmed destinations. This especially effect CMT but also makes a impact on regular SCTP as well. - During init collision when we detect seq number out of sync we need to treat it like Case C and discard the cookie (no invarient needed here). - Atomic access to the random store. - When we declare a vtag good, we need to shove it into the time wait hash to prevent further use. When the tag is put into the assoc hash, we need to remove it from the twait hash (where it will surely be). This prevents duplicate tag assignments. - Move decr-ref count to better protect sysctl out of data. - ltrace error corrections in sctp6_usrreq.c - Add hook for interface up/down to be sent to us. - Make sysctl() exported structures independent of processor architecture. - Fix route and src addr cache clearing for delete address case. - Make sure address marked SCTP_DEL_IP_ADDRESS is never selected as src addr. - in icmp handling fixed so we actually look at the icmp codes to figure out what to do. - Modified mobility code. Reception of DELETE IP ADDRESS for a primary destination and SET PRIMARY for a new primary destination is used for retransmission trigger to the new primary destination. Also, in this case, destination of chunks in send_queue are changed to the new primary destination. - Fix so that we disallow sending by mbuf to ever have EEOR mode set upon it. Approved by: re@freebsd.org (B Mah)
2007-09-08 17:48:46 +00:00
}
SCTPDBG(SCTP_DEBUG_ASCONF1, "delete_prim_timer: finished to keep deleted primary ");
SCTPDBG_ADDR(SCTP_DEBUG_ASCONF1, &stcb->asoc.deleted_primary->ro._l_addr.sa);
sctp_free_remote_addr(stcb->asoc.deleted_primary);
stcb->asoc.deleted_primary = NULL;
sctp_mobility_feature_off(inp, SCTP_MOBILITY_PRIM_DELETED);
return;
- send call has a reference to uio->uio_resid in the recent send code, but uio may be NULL on sendfile calls. Change to use sndlen variable. - EMSGSIZE is not being returned in non-blocking mode and needs a small tweak to look if the msg would ever fit when returning EWOULDBLOCK. - FWD-TSN has a bug in stream processing which could cause a panic. This is a follow on to the codenomicon fix. - PDAPI level 1 and 2 do not work unless the reader gets his returned buffer full. Fix so we can break out when at level 1 or 2. - Fix fast-handoff features to copy across properly on accepted sockets - Fix sctp_peeloff() system call when no true system call exists to screen arguments for errors. In cases where a real system call exists the system call itself does this. - Fix raddr leak in recent add-ip code change for bundled asconfs (even when non-bundled asconfs are received) - Make sure ipi_addr lock is held when walking global addr list. Need to change this lock type to a rwlock(). - Add don't wake flag on both input and output when the socket is closing. - When deleting an address verify the interface is correct before allowing the delete to process. This protects panda and unnumbered. - Clean up old sysctl stuff and get rid of the old Open/Net BSD structures. - Add a function to watch the ranges in the sysctl sets. - When appending in the reassembly queue, validate that the assoc has not gone to about to be freed. If so (in the middle) abort out. Note this especially effects MAC I think due to the lock/unlock they do (or with LOCK testing in place). - Netstat patch to get rid of warnings. - Make sure that no data gets queued to inactive/unconfirmed destinations. This especially effect CMT but also makes a impact on regular SCTP as well. - During init collision when we detect seq number out of sync we need to treat it like Case C and discard the cookie (no invarient needed here). - Atomic access to the random store. - When we declare a vtag good, we need to shove it into the time wait hash to prevent further use. When the tag is put into the assoc hash, we need to remove it from the twait hash (where it will surely be). This prevents duplicate tag assignments. - Move decr-ref count to better protect sysctl out of data. - ltrace error corrections in sctp6_usrreq.c - Add hook for interface up/down to be sent to us. - Make sysctl() exported structures independent of processor architecture. - Fix route and src addr cache clearing for delete address case. - Make sure address marked SCTP_DEL_IP_ADDRESS is never selected as src addr. - in icmp handling fixed so we actually look at the icmp codes to figure out what to do. - Modified mobility code. Reception of DELETE IP ADDRESS for a primary destination and SET PRIMARY for a new primary destination is used for retransmission trigger to the new primary destination. Also, in this case, destination of chunks in send_queue are changed to the new primary destination. - Fix so that we disallow sending by mbuf to ever have EEOR mode set upon it. Approved by: re@freebsd.org (B Mah)
2007-09-08 17:48:46 +00:00
}
/*
* For the shutdown and shutdown-ack, we do not keep one around on the
* control queue. This means we must generate a new one and call the general
* chunk output routine, AFTER having done threshold management.
* It is assumed that net is non-NULL.
*/
int
sctp_shutdown_timer(struct sctp_inpcb *inp, struct sctp_tcb *stcb,
struct sctp_nets *net)
{
struct sctp_nets *alt;
/* first threshold managment */
if (sctp_threshold_management(inp, stcb, net, stcb->asoc.max_send_times)) {
/* Assoc is over */
return (1);
}
sctp_backoff_on_timeout(stcb, net, 1, 0, 0);
/* second select an alternative */
alt = sctp_find_alternate_net(stcb, net, 0);
/* third generate a shutdown into the queue for out net */
sctp_send_shutdown(stcb, alt);
/* fourth restart timer */
sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWN, inp, stcb, alt);
return (0);
}
int
sctp_shutdownack_timer(struct sctp_inpcb *inp, struct sctp_tcb *stcb,
struct sctp_nets *net)
{
struct sctp_nets *alt;
/* first threshold managment */
if (sctp_threshold_management(inp, stcb, net, stcb->asoc.max_send_times)) {
/* Assoc is over */
return (1);
}
sctp_backoff_on_timeout(stcb, net, 1, 0, 0);
/* second select an alternative */
alt = sctp_find_alternate_net(stcb, net, 0);
/* third generate a shutdown into the queue for out net */
sctp_send_shutdown_ack(stcb, alt);
/* fourth restart timer */
sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWNACK, inp, stcb, alt);
return (0);
}
static void
sctp_audit_stream_queues_for_size(struct sctp_inpcb *inp,
struct sctp_tcb *stcb)
{
struct sctp_stream_queue_pending *sp;
unsigned int i, chks_in_queue = 0;
int being_filled = 0;
/*
* This function is ONLY called when the send/sent queues are empty.
*/
if ((stcb == NULL) || (inp == NULL))
return;
if (stcb->asoc.sent_queue_retran_cnt) {
SCTP_PRINTF("Hmm, sent_queue_retran_cnt is non-zero %d\n",
stcb->asoc.sent_queue_retran_cnt);
stcb->asoc.sent_queue_retran_cnt = 0;
}
if (stcb->asoc.ss_functions.sctp_ss_is_empty(stcb, &stcb->asoc)) {
/* No stream scheduler information, initialize scheduler */
stcb->asoc.ss_functions.sctp_ss_init(stcb, &stcb->asoc, 0);
if (!stcb->asoc.ss_functions.sctp_ss_is_empty(stcb, &stcb->asoc)) {
/* yep, we lost a stream or two */
SCTP_PRINTF("Found additional streams NOT managed by scheduler, corrected\n");
} else {
/* no streams lost */
stcb->asoc.total_output_queue_size = 0;
}
}
/* Check to see if some data queued, if so report it */
for (i = 0; i < stcb->asoc.streamoutcnt; i++) {
if (!TAILQ_EMPTY(&stcb->asoc.strmout[i].outqueue)) {
TAILQ_FOREACH(sp, &stcb->asoc.strmout[i].outqueue, next) {
if (sp->msg_is_complete)
being_filled++;
chks_in_queue++;
}
}
}
if (chks_in_queue != stcb->asoc.stream_queue_cnt) {
SCTP_PRINTF("Hmm, stream queue cnt at %d I counted %d in stream out wheel\n",
stcb->asoc.stream_queue_cnt, chks_in_queue);
}
if (chks_in_queue) {
/* call the output queue function */
sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_T3, SCTP_SO_NOT_LOCKED);
if ((TAILQ_EMPTY(&stcb->asoc.send_queue)) &&
(TAILQ_EMPTY(&stcb->asoc.sent_queue))) {
/*
* Probably should go in and make it go back through
* and add fragments allowed
*/
if (being_filled == 0) {
SCTP_PRINTF("Still nothing moved %d chunks are stuck\n",
chks_in_queue);
}
}
} else {
SCTP_PRINTF("Found no chunks on any queue tot:%lu\n",
(u_long)stcb->asoc.total_output_queue_size);
stcb->asoc.total_output_queue_size = 0;
}
}
int
sctp_heartbeat_timer(struct sctp_inpcb *inp, struct sctp_tcb *stcb,
struct sctp_nets *net)
{
uint8_t net_was_pf;
if (net->dest_state & SCTP_ADDR_PF) {
net_was_pf = 1;
} else {
net_was_pf = 0;
}
if (net->hb_responded == 0) {
if (net->ro._s_addr) {
/*
* Invalidate the src address if we did not get a
* response last time.
*/
sctp_free_ifa(net->ro._s_addr);
net->ro._s_addr = NULL;
net->src_addr_selected = 0;
}
sctp_backoff_on_timeout(stcb, net, 1, 0, 0);
if (sctp_threshold_management(inp, stcb, net, stcb->asoc.max_send_times)) {
/* Assoc is over */
return (1);
}
}
/* Zero PBA, if it needs it */
if (net->partial_bytes_acked) {
net->partial_bytes_acked = 0;
}
if ((stcb->asoc.total_output_queue_size > 0) &&
(TAILQ_EMPTY(&stcb->asoc.send_queue)) &&
(TAILQ_EMPTY(&stcb->asoc.sent_queue))) {
sctp_audit_stream_queues_for_size(inp, stcb);
}
if (!(net->dest_state & SCTP_ADDR_NOHB) &&
!((net_was_pf == 0) && (net->dest_state & SCTP_ADDR_PF))) {
/*
* when move to PF during threshold mangement, a HB has been
* queued in that routine
*/
uint32_t ms_gone_by;
if ((net->last_sent_time.tv_sec > 0) ||
(net->last_sent_time.tv_usec > 0)) {
struct timeval diff;
SCTP_GETTIME_TIMEVAL(&diff);
timevalsub(&diff, &net->last_sent_time);
ms_gone_by = (uint32_t) (diff.tv_sec * 1000) +
(uint32_t) (diff.tv_usec / 1000);
} else {
ms_gone_by = 0xffffffff;
}
if ((ms_gone_by >= net->heart_beat_delay) ||
(net->dest_state & SCTP_ADDR_PF)) {
sctp_send_hb(stcb, net, SCTP_SO_NOT_LOCKED);
}
}
return (0);
}
void
sctp_pathmtu_timer(struct sctp_inpcb *inp,
struct sctp_tcb *stcb,
struct sctp_nets *net)
{
uint32_t next_mtu, mtu;
next_mtu = sctp_get_next_mtu(net->mtu);
if ((next_mtu > net->mtu) && (net->port == 0)) {
if ((net->src_addr_selected == 0) ||
(net->ro._s_addr == NULL) ||
(net->ro._s_addr->localifa_flags & SCTP_BEING_DELETED)) {
if ((net->ro._s_addr != NULL) && (net->ro._s_addr->localifa_flags & SCTP_BEING_DELETED)) {
sctp_free_ifa(net->ro._s_addr);
net->ro._s_addr = NULL;
net->src_addr_selected = 0;
} else if (net->ro._s_addr == NULL) {
#if defined(INET6) && defined(SCTP_EMBEDDED_V6_SCOPE)
if (net->ro._l_addr.sa.sa_family == AF_INET6) {
struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)&net->ro._l_addr;
/* KAME hack: embed scopeid */
2009-09-17 15:11:12 +00:00
(void)sa6_embedscope(sin6, MODULE_GLOBAL(ip6_use_defzone));
}
#endif
net->ro._s_addr = sctp_source_address_selection(inp,
stcb,
(sctp_route_t *) & net->ro,
net, 0, stcb->asoc.vrf_id);
#if defined(INET6) && defined(SCTP_EMBEDDED_V6_SCOPE)
if (net->ro._l_addr.sa.sa_family == AF_INET6) {
struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)&net->ro._l_addr;
(void)sa6_recoverscope(sin6);
}
#endif /* INET6 */
}
if (net->ro._s_addr)
net->src_addr_selected = 1;
}
if (net->ro._s_addr) {
mtu = SCTP_GATHER_MTU_FROM_ROUTE(net->ro._s_addr, &net->ro._s_addr.sa, net->ro.ro_rt);
#if defined(INET) || defined(INET6)
if (net->port) {
mtu -= sizeof(struct udphdr);
}
#endif
if (mtu > next_mtu) {
net->mtu = next_mtu;
}
}
}
/* restart the timer */
sctp_timer_start(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net);
}
void
sctp_autoclose_timer(struct sctp_inpcb *inp,
struct sctp_tcb *stcb,
struct sctp_nets *net)
{
struct timeval tn, *tim_touse;
struct sctp_association *asoc;
int ticks_gone_by;
(void)SCTP_GETTIME_TIMEVAL(&tn);
if (stcb->asoc.sctp_autoclose_ticks &&
sctp_is_feature_on(inp, SCTP_PCB_FLAGS_AUTOCLOSE)) {
/* Auto close is on */
asoc = &stcb->asoc;
/* pick the time to use */
if (asoc->time_last_rcvd.tv_sec >
asoc->time_last_sent.tv_sec) {
tim_touse = &asoc->time_last_rcvd;
} else {
tim_touse = &asoc->time_last_sent;
}
/* Now has long enough transpired to autoclose? */
ticks_gone_by = SEC_TO_TICKS(tn.tv_sec - tim_touse->tv_sec);
if ((ticks_gone_by > 0) &&
(ticks_gone_by >= (int)asoc->sctp_autoclose_ticks)) {
/*
* autoclose time has hit, call the output routine,
* which should do nothing just to be SURE we don't
* have hanging data. We can then safely check the
* queues and know that we are clear to send
* shutdown
*/
sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_AUTOCLOSE_TMR, SCTP_SO_NOT_LOCKED);
/* Are we clean? */
if (TAILQ_EMPTY(&asoc->send_queue) &&
TAILQ_EMPTY(&asoc->sent_queue)) {
/*
* there is nothing queued to send, so I'm
* done...
*/
- Copyright updates (aka 2007) - ZONE get now also take a type cast so it does the cast like mtod does. - New macro SCTP_LIST_EMPTY, which in bsd is just LIST_EMPTY - Removal of const in some of the static hmac functions (not needed) - Store length changes to allow for new fields in auth - Auth code updated to current draft (this should be the RFC version we think). - use uint8_t instead of u_char in LOOPBACK address comparison - Some u_int32_t converted to uint32_t (in crc code) - A bug was found in the mib counts for ordered/unordered count, this was fixed (was referencing a freed mbuf). - SCTP_ASOCLOG_OF_TSNS added (code will probably disappear after my testing completes. It allows us to keep a small log on each assoc of the last 40 TSN's in/out and stream assignment. It is NOT in options and so is only good for private builds. - Some CMT changes in prep for Jana fixing his problem with reneging when CMT is enabled (Concurrent Multipath Transfer = CMT). - Some missing mib stats added. - Correction to number of open assoc's count in mib - Correction to os_bsd.h to get right sha2 macros - Add of special AUTH_04 flags so you can compile the code with the old format (in case the peer does not yet support the latest auth code). - Nonce sum was incorrectly being set in when ecn_nonce was NOT on. - LOR in listen with implicit bind found and fixed. - Moved away from using mbuf's for socket options to using just data pointers. The mbufs were used to harmonize NetBSD code since both Net and Open used this method. We have decided to move away from that and more conform to FreeBSD style (which makes more sense). - Very very nasty bug found in some of my "debug" code. The cookie_how collision case tracking had an endless loop in it if you got a second retransmission of a cookie collision case. This would lock up a CPU .. ugly.. - auth function goes to using size_t instead of int which conforms to socketapi better - Found the nasty bug that happens after 9 days of testing.. you get the data chunk, deliver it and due to the reference to a ch-> that every now and then has been deleted (depending on the postion in the mbuf) you have an invalid ch->ch.flags.. and thus you don't advance the stream sequence number.. so you block the stream permanently. The fix is to make local variables of these guys and set them up before you have any chance of trimming the mbuf. - style fix in sctp_util.h, not sure how this got bad maybe in the last patch? (aka it may not be in the real source). - Found interesting bug when using the extended snd/rcv info where we would get an error on receiving with this. Thats because it was NOT padded to the same size as the snd_rcv info. We increase (add the pad) so the two structs are the same size in sctp_uio.h - In sctp_usrreq.c one of the most common things we did for socket options was to cast the pointer and validate the size. This as been macro-ized to help make the code more readable. - in sctputil.c two things, the socketapi class found a missing flag type (the next msg is a notification) and a missing scope recovery was also fixed. Reviewed by: gnn
2007-02-12 23:24:31 +00:00
if (SCTP_GET_STATE(asoc) != SCTP_STATE_SHUTDOWN_SENT) {
/* only send SHUTDOWN 1st time thru */
struct sctp_nets *netp;
- Copyright updates (aka 2007) - ZONE get now also take a type cast so it does the cast like mtod does. - New macro SCTP_LIST_EMPTY, which in bsd is just LIST_EMPTY - Removal of const in some of the static hmac functions (not needed) - Store length changes to allow for new fields in auth - Auth code updated to current draft (this should be the RFC version we think). - use uint8_t instead of u_char in LOOPBACK address comparison - Some u_int32_t converted to uint32_t (in crc code) - A bug was found in the mib counts for ordered/unordered count, this was fixed (was referencing a freed mbuf). - SCTP_ASOCLOG_OF_TSNS added (code will probably disappear after my testing completes. It allows us to keep a small log on each assoc of the last 40 TSN's in/out and stream assignment. It is NOT in options and so is only good for private builds. - Some CMT changes in prep for Jana fixing his problem with reneging when CMT is enabled (Concurrent Multipath Transfer = CMT). - Some missing mib stats added. - Correction to number of open assoc's count in mib - Correction to os_bsd.h to get right sha2 macros - Add of special AUTH_04 flags so you can compile the code with the old format (in case the peer does not yet support the latest auth code). - Nonce sum was incorrectly being set in when ecn_nonce was NOT on. - LOR in listen with implicit bind found and fixed. - Moved away from using mbuf's for socket options to using just data pointers. The mbufs were used to harmonize NetBSD code since both Net and Open used this method. We have decided to move away from that and more conform to FreeBSD style (which makes more sense). - Very very nasty bug found in some of my "debug" code. The cookie_how collision case tracking had an endless loop in it if you got a second retransmission of a cookie collision case. This would lock up a CPU .. ugly.. - auth function goes to using size_t instead of int which conforms to socketapi better - Found the nasty bug that happens after 9 days of testing.. you get the data chunk, deliver it and due to the reference to a ch-> that every now and then has been deleted (depending on the postion in the mbuf) you have an invalid ch->ch.flags.. and thus you don't advance the stream sequence number.. so you block the stream permanently. The fix is to make local variables of these guys and set them up before you have any chance of trimming the mbuf. - style fix in sctp_util.h, not sure how this got bad maybe in the last patch? (aka it may not be in the real source). - Found interesting bug when using the extended snd/rcv info where we would get an error on receiving with this. Thats because it was NOT padded to the same size as the snd_rcv info. We increase (add the pad) so the two structs are the same size in sctp_uio.h - In sctp_usrreq.c one of the most common things we did for socket options was to cast the pointer and validate the size. This as been macro-ized to help make the code more readable. - in sctputil.c two things, the socketapi class found a missing flag type (the next msg is a notification) and a missing scope recovery was also fixed. Reviewed by: gnn
2007-02-12 23:24:31 +00:00
if ((SCTP_GET_STATE(asoc) == SCTP_STATE_OPEN) ||
(SCTP_GET_STATE(asoc) == SCTP_STATE_SHUTDOWN_RECEIVED)) {
SCTP_STAT_DECR_GAUGE32(sctps_currestab);
}
- Fix address add handling to clear cached routes and source addresses when peer acks the add in case the routing table changes. - Fix sctp_lower_sosend to send shutdown chunk for mbuf send case when sndlen = 0 and sinfoflag = SCTP_EOF - Fix sctp_lower_sosend for SCTP_ABORT mbuf send case with null data, So that it does not send the "null" data mbuf out and cause it to get freed twice. - Fix so auto-asconf sysctl actually effect the socket's asconf state. - Do not allow SCTP_AUTO_ASCONF option to be used on subset bound sockets. - Memset bug in sctp_output.c (arguments were reversed) submitted found and reported by Dave Jones (davej@codemonkey.org.uk). - PD-API point needs to be invoked >= not just > to conform to socket api draft this fixes sctp_indata.c in the two places need to be >=. - move M_NOTIFICATION to use M_PROTO5. - PEER_ADDR_PARAMS did not fail properly if you specify an address that is not in the association with a valid assoc_id. This meant you got or set the stcb level values instead of the destination you thought you were going to get/set. Now validate if the stcb is non-null and the net is NULL that the sa_family is set and the address is unspecified otherwise return an error. - The thread based iterator could crash if associations were freed at the exact time it was running. rework the worker thread to use the increment/decrement to prevent this and no longer use the markers that the timer based iterator uses. - Fix the memleak in sctp_add_addr_to_vrf() for the case when it is detected that ifa is already pointing to a ifn. - Fix it so that if someone is so insane that they drop the send window below the minimal add mark, they still can send. - Changed all state for associations to use mask safe macro. - During front states in association freeing in sctp_inpcbfree, we had a locking problem where locks were not in place where they should have been. - Free association calls were not testing the return value in sctp_inpcb_free() properly... others should be cast void returns where we don't care about the return value. - If a reference count is held on an assoc, even from the "force free" we should not do the actual free.. but instead let the timer free it. - When we enter sctp_input(), if the SCTP_ASOC_ABOUT_TO_BE_FREED flag is set, we must NOT process the packet but handle it like ootb. This is because while freeing an assoc we release the locks to get all the higher order locks so we can purge all the hash tables. This leaves a hole if a packet comes in just at that point. Now sctp_common_input_processing() will call the ootb code in such a case. - Change MBUF M_NOTIFICATION to use M_PROTO5 (per Sam L). This makes it so we don't have a conflict (I think this is a covertity change). We made this change AFTER some conversation and looking to make sure that M_PROTO5 does not have a problem between SCTP and the 802.11 stuff (which is the only other place its used). - Fixed lock order reversal and missing atomic protection around locked_tcb during association lookup and the 1-2-1 model. - Added debug to source address selection. - V6 output must always do checksum even for loopback. - Remove more locks around inp that are not needed for an atomically added/subtracted ref count. - slight optimization in the way we zero the array in sctp_sack_check() - It was possible to respond to a ABORT() with bad checksum with a PKT-DROP. This lead to a PKT-DROP/ABORT war. Add code to NOT send a PKT-DROP to any ABORT(). - Add an option for local logging (useful for macintosh or when you need better performing during debugging). Note no commands are here to get the log info, you must just use kgdb. - The timer code needs to be aware of if it needs to call sctp_sack_check() to slide the maps and adjust the cum-ack. This is because it may be out of sync cum-ack wise. - Added threshold managment logging. - If the user picked just the right size, that just filled the send window minus one mtu, we would enter a forever loop not copying and at the same time not blocking. Change from < to <= solves this. - Sysctl added to control the fragment interleave level which defaults to 1. - My rwnd control was not being used to control the rwnd properly (we did not add and subtract to it :-() this is now fixed so we handle small messages (1 byte etc) better to bring our rwnd down more slowly. Approved by: re@freebsd.org (Bruce Mah)
2007-08-24 00:53:53 +00:00
SCTP_SET_STATE(asoc, SCTP_STATE_SHUTDOWN_SENT);
SCTP_CLEAR_SUBSTATE(asoc, SCTP_STATE_SHUTDOWN_PENDING);
sctp_stop_timers_for_shutdown(stcb);
if (stcb->asoc.alternate) {
netp = stcb->asoc.alternate;
} else {
netp = stcb->asoc.primary_destination;
}
sctp_send_shutdown(stcb, netp);
sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWN,
stcb->sctp_ep, stcb,
netp);
sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWNGUARD,
stcb->sctp_ep, stcb,
netp);
}
}
} else {
/*
* No auto close at this time, reset t-o to check
* later
*/
int tmp;
/* fool the timer startup to use the time left */
tmp = asoc->sctp_autoclose_ticks;
asoc->sctp_autoclose_ticks -= ticks_gone_by;
sctp_timer_start(SCTP_TIMER_TYPE_AUTOCLOSE, inp, stcb,
net);
/* restore the real tick value */
asoc->sctp_autoclose_ticks = tmp;
}
}
}