svn commit: r252555 - in stable/9: sbin/ifconfig sys/amd64/conf sys/conf sys/contrib/rdma sys/contrib/rdma/krping sys/dev/cxgb/ulp/tom sys/i386/conf sys/modules sys/modules/rdma/krping sys/modules/...
Navdeep Parhar
np at FreeBSD.org
Wed Jul 3 09:25:34 UTC 2013
Author: np
Date: Wed Jul 3 09:25:29 2013
New Revision: 252555
URL: http://svnweb.freebsd.org/changeset/base/252555
Log:
MFC/backport core kernel and userspace parts of r237263 (TCP_OFFLOAD
rework). MFC r237563, r239511, r243603, r245915, r245916, r245919,
r245921, r245922, r245924, r245925, r245932, r245934 too.
Build tested with make universe.
r237263:
- Updated TOE support in the kernel.
...
r237563:
Fix clang warning when compiling iw_cxgb.
r239511:
Correctly handle the case where an inp has already been dropped by the time
the TOE driver reports that an active open failed. toe_connect_failed is
supposed to handle this but it should be provided the inpcb instead of the
tcpcb which may no longer be around.
r243603:
Make sure that tcp_timer_activate() correctly sees TCP_OFFLOAD (or not).
r245915:
Heed SO_NO_OFFLOAD.
r245916:
Teach toe_4tuple_check() to deal with IPv6 4-tuples too.
r245919:
Add TCP_OFFLOAD hook in syncache_respond for IPv6 too, just like the one
that exists for IPv4.
r245921:
There is no need to call into the TOE driver twice in pru_rcvd (tod_rcvd
and then tod_output right after that).
r245922:
Avoid NULL dereference in nd6_storelladdr when no mbuf is provided. It
is called this way from a couple of places in the OFED code. (toecore
calls it too but that's going to change shortly).
r245924:
Move lle_event to if_llatbl.h
lle_event replaced arp_update_event after the ARP rewrite and ended up
in if_ether.h simply because arp_update_event used to be there too.
IPv6 neighbor discovery is going to grow lle_event support and this is a
good time to move it to if_llatbl.h.
The two in-tree consumers of this event - OFED and toecore - are not
affected.
r245925:
Generate lle_event in the IPv6 neighbor discovery code too.
r245932:
Teach toe_l2_resolve to resolve IPv6 destinations too.
r245934:
Add checks for SO_NO_OFFLOAD in a couple of places that I missed earlier
in r245915.
Added:
stable/9/sys/modules/toecore/
- copied from r237263, head/sys/modules/toecore/
stable/9/sys/netinet/toecore.c
- copied, changed from r237263, head/sys/netinet/toecore.c
stable/9/sys/netinet/toecore.h
- copied, changed from r237263, head/sys/netinet/toecore.h
Deleted:
stable/9/sys/netinet/toedev.h
Modified:
stable/9/sbin/ifconfig/ifconfig.c
stable/9/sys/amd64/conf/GENERIC
stable/9/sys/conf/NOTES
stable/9/sys/conf/files
stable/9/sys/conf/options
stable/9/sys/contrib/rdma/krping/krping.c
stable/9/sys/contrib/rdma/krping/krping.h
stable/9/sys/contrib/rdma/krping/krping_dev.c
stable/9/sys/contrib/rdma/rdma_addr.c
stable/9/sys/contrib/rdma/rdma_cache.c
stable/9/sys/dev/cxgb/ulp/tom/cxgb_cpl_io.c
stable/9/sys/i386/conf/GENERIC
stable/9/sys/i386/conf/XEN
stable/9/sys/modules/Makefile
stable/9/sys/modules/rdma/krping/Makefile
stable/9/sys/net/if_llatbl.h
stable/9/sys/net/if_var.h
stable/9/sys/net/if_vlan.c
stable/9/sys/netinet/if_ether.c
stable/9/sys/netinet/if_ether.h
stable/9/sys/netinet/in.c
stable/9/sys/netinet/tcp_input.c
stable/9/sys/netinet/tcp_offload.c
stable/9/sys/netinet/tcp_offload.h
stable/9/sys/netinet/tcp_output.c
stable/9/sys/netinet/tcp_subr.c
stable/9/sys/netinet/tcp_syncache.c
stable/9/sys/netinet/tcp_syncache.h
stable/9/sys/netinet/tcp_timer.c
stable/9/sys/netinet/tcp_usrreq.c
stable/9/sys/netinet/tcp_var.h
stable/9/sys/netinet6/nd6.c
stable/9/sys/netinet6/nd6_nbr.c
stable/9/sys/ofed/drivers/infiniband/core/cma.c
stable/9/sys/ofed/drivers/infiniband/core/iwcm.c
stable/9/sys/ofed/include/linux/net.h
stable/9/sys/ofed/include/linux/workqueue.h
stable/9/sys/ofed/include/net/netevent.h
stable/9/sys/ofed/include/rdma/iw_cm.h
stable/9/usr.bin/netstat/inet.c
stable/9/usr.bin/sockstat/sockstat.c
Directory Properties:
stable/9/sbin/ifconfig/ (props changed)
stable/9/sys/ (props changed)
stable/9/sys/dev/ (props changed)
stable/9/sys/net/ (props changed)
stable/9/usr.bin/netstat/ (props changed)
stable/9/usr.bin/sockstat/ (props changed)
Modified: stable/9/sbin/ifconfig/ifconfig.c
==============================================================================
--- stable/9/sbin/ifconfig/ifconfig.c Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sbin/ifconfig/ifconfig.c Wed Jul 3 09:25:29 2013 (r252555)
@@ -910,7 +910,7 @@ unsetifdescr(const char *val, int value,
#define IFCAPBITS \
"\020\1RXCSUM\2TXCSUM\3NETCONS\4VLAN_MTU\5VLAN_HWTAGGING\6JUMBO_MTU\7POLLING" \
"\10VLAN_HWCSUM\11TSO4\12TSO6\13LRO\14WOL_UCAST\15WOL_MCAST\16WOL_MAGIC" \
-"\21VLAN_HWFILTER\23VLAN_HWTSO\24LINKSTATE\25NETMAP" \
+"\17TOE4\20TOE6\21VLAN_HWFILTER\23VLAN_HWTSO\24LINKSTATE\25NETMAP" \
"\26RXCSUM_IPV6\27TXCSUM_IPV6"
/*
@@ -1189,6 +1189,8 @@ static struct cmd basic_cmds[] = {
DEF_CMD("-tso4", -IFCAP_TSO4, setifcap),
DEF_CMD("tso", IFCAP_TSO, setifcap),
DEF_CMD("-tso", -IFCAP_TSO, setifcap),
+ DEF_CMD("toe", IFCAP_TOE, setifcap),
+ DEF_CMD("-toe", -IFCAP_TOE, setifcap),
DEF_CMD("lro", IFCAP_LRO, setifcap),
DEF_CMD("-lro", -IFCAP_LRO, setifcap),
DEF_CMD("wol", IFCAP_WOL, setifcap),
Modified: stable/9/sys/amd64/conf/GENERIC
==============================================================================
--- stable/9/sys/amd64/conf/GENERIC Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/amd64/conf/GENERIC Wed Jul 3 09:25:29 2013 (r252555)
@@ -28,6 +28,7 @@ options SCHED_ULE # ULE scheduler
options PREEMPTION # Enable kernel thread preemption
options INET # InterNETworking
options INET6 # IPv6 communications protocols
+options TCP_OFFLOAD # TCP offload
options SCTP # Stream Control Transmission Protocol
options FFS # Berkeley Fast Filesystem
options SOFTUPDATES # Enable FFS soft updates support
Modified: stable/9/sys/conf/NOTES
==============================================================================
--- stable/9/sys/conf/NOTES Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/conf/NOTES Wed Jul 3 09:25:29 2013 (r252555)
@@ -545,6 +545,8 @@ options INET6 #IPv6 communications pr
options ROUTETABLES=2 # max 16. 1 is back compatible.
+options TCP_OFFLOAD # TCP offload support.
+
# In order to enable IPSEC you MUST also add device crypto to
# your kernel configuration
options IPSEC #IP security (requires device crypto)
Modified: stable/9/sys/conf/files
==============================================================================
--- stable/9/sys/conf/files Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/conf/files Wed Jul 3 09:25:29 2013 (r252555)
@@ -2886,7 +2886,7 @@ netinet/tcp_hostcache.c optional inet |
netinet/tcp_input.c optional inet | inet6
netinet/tcp_lro.c optional inet | inet6
netinet/tcp_output.c optional inet | inet6
-netinet/tcp_offload.c optional inet | inet6
+netinet/tcp_offload.c optional tcp_offload inet | tcp_offload inet6
netinet/tcp_reass.c optional inet | inet6
netinet/tcp_sack.c optional inet | inet6
netinet/tcp_subr.c optional inet | inet6
Modified: stable/9/sys/conf/options
==============================================================================
--- stable/9/sys/conf/options Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/conf/options Wed Jul 3 09:25:29 2013 (r252555)
@@ -431,7 +431,7 @@ RADIX_MPATH opt_mpath.h
ROUTETABLES opt_route.h
SLIP_IFF_OPTS opt_slip.h
TCPDEBUG
-TCP_OFFLOAD_DISABLE opt_inet.h #Disable code to dispatch tcp offloading
+TCP_OFFLOAD opt_inet.h # Enable code to dispatch TCP offloading
TCP_SIGNATURE opt_inet.h
VLAN_ARRAY opt_vlan.h
XBONEHACK
Modified: stable/9/sys/contrib/rdma/krping/krping.c
==============================================================================
--- stable/9/sys/contrib/rdma/krping/krping.c Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/contrib/rdma/krping/krping.c Wed Jul 3 09:25:29 2013 (r252555)
@@ -41,7 +41,6 @@ __FBSDID("$FreeBSD$");
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/socket.h>
-#include <sys/module.h>
#include <sys/endian.h>
#include <sys/limits.h>
#include <sys/proc.h>
@@ -53,11 +52,13 @@ __FBSDID("$FreeBSD$");
#include <sys/queue.h>
#include <sys/taskqueue.h>
#include <sys/syslog.h>
+#include <netinet/in.h>
#include <vm/vm.h>
#include <vm/pmap.h>
-#include <contrib/rdma/rdma_cm.h>
+#include <linux/types.h>
+#include <rdma/rdma_cm.h>
#include "getopt.h"
#include "krping.h"
@@ -83,6 +84,7 @@ static const struct krping_option krping
{"bw", OPT_NOPARAM, 'B'},
{"tx-depth", OPT_INT, 't'},
{"poll", OPT_NOPARAM, 'P'},
+ {"memlimit", OPT_INT, 'm'},
{NULL, 0, 0}
};
@@ -254,10 +256,14 @@ static void krping_cq_event_handler(stru
ib_req_notify_cq(cb->cq, IB_CQ_NEXT_COMP);
while ((ret = ib_poll_cq(cb->cq, 1, &wc)) == 1) {
if (wc.status) {
- if (wc.status != IB_WC_WR_FLUSH_ERR)
- log(LOG_ERR, "cq completion failed status %d\n",
+ if (wc.status == IB_WC_WR_FLUSH_ERR) {
+ DEBUG_LOG("cq flushed\n");
+ continue;
+ } else {
+ log(LOG_CRIT, "cq completion failed status %d\n",
wc.status);
- goto error;
+ goto error;
+ }
}
switch (wc.opcode) {
@@ -432,8 +438,17 @@ static int krping_setup_buffers(struct k
}
}
- cb->rdma_buf = contigmalloc(cb->size, M_DEVBUF, M_WAITOK, 0, -1UL,
- PAGE_SIZE, 0);
+ /* RNIC adapters have a limit upto which it can register physical memory
+ * If DMA-MR memory mode is set then normally driver registers maximum
+ * supported memory. After that if contigmalloc allocates memory beyond the
+ * specified RNIC limit then Krping may not work.
+ */
+ if (cb->use_dmamr && cb->memlimit)
+ cb->rdma_buf = contigmalloc(cb->size, M_DEVBUF, M_WAITOK, 0, cb->memlimit,
+ PAGE_SIZE, 0);
+ else
+ cb->rdma_buf = contigmalloc(cb->size, M_DEVBUF, M_WAITOK, 0, -1UL,
+ PAGE_SIZE, 0);
if (!cb->rdma_buf) {
log(LOG_ERR, "rdma_buf malloc failed\n");
@@ -458,8 +473,12 @@ static int krping_setup_buffers(struct k
}
if (!cb->server || cb->wlat || cb->rlat || cb->bw) {
- cb->start_buf = contigmalloc(cb->size, M_DEVBUF, M_WAITOK,
- 0, -1UL, PAGE_SIZE, 0);
+ if (cb->use_dmamr && cb->memlimit)
+ cb->start_buf = contigmalloc(cb->size, M_DEVBUF, M_WAITOK,
+ 0, cb->memlimit, PAGE_SIZE, 0);
+ else
+ cb->start_buf = contigmalloc(cb->size, M_DEVBUF, M_WAITOK,
+ 0, -1UL, PAGE_SIZE, 0);
if (!cb->start_buf) {
log(LOG_ERR, "start_buf malloc failed\n");
ret = ENOMEM;
@@ -1636,6 +1655,8 @@ int krping_doit(char *cmd)
cb->state = IDLE;
cb->size = 64;
cb->txdepth = RPING_SQ_DEPTH;
+ cb->use_dmamr = 1;
+ cb->memlimit = 0;
mtx_init(&cb->lock, "krping mtx", NULL, MTX_DUPOK|MTX_DEF);
while ((op = krping_getopt("krping", &cmd, krping_opts, NULL, &optarg,
@@ -1713,6 +1734,15 @@ int krping_doit(char *cmd)
case 'd':
debug++;
break;
+ case 'm':
+ cb->memlimit = optint;
+ if (cb->memlimit < 1) {
+ log(LOG_ERR, "Invalid memory limit %ju\n",
+ cb->memlimit);
+ ret = EINVAL;
+ } else
+ DEBUG_LOG(PFX "memory limit %d\n", (int)optint);
+ break;
default:
log(LOG_ERR, "unknown opt %s\n", optarg);
ret = EINVAL;
Modified: stable/9/sys/contrib/rdma/krping/krping.h
==============================================================================
--- stable/9/sys/contrib/rdma/krping/krping.h Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/contrib/rdma/krping/krping.h Wed Jul 3 09:25:29 2013 (r252555)
@@ -1,7 +1,7 @@
/*
* $FreeBSD$
*/
-#include <contrib/rdma/ib_verbs.h>
+#include <rdma/ib_verbs.h>
#include <netinet/in.h>
/*
@@ -92,6 +92,8 @@ struct krping_cb {
int count; /* ping count */
int size; /* ping data size */
int validate; /* validate ping data */
+ uint64_t memlimit; /* limit of the physical memory that
+ can be registered with dma_mr mode */
/* CM stuff */
struct rdma_cm_id *cm_id; /* connection on client side,*/
Modified: stable/9/sys/contrib/rdma/krping/krping_dev.c
==============================================================================
--- stable/9/sys/contrib/rdma/krping/krping_dev.c Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/contrib/rdma/krping/krping_dev.c Wed Jul 3 09:25:29 2013 (r252555)
@@ -14,7 +14,6 @@
__FBSDID("$FreeBSD$");
#include <sys/types.h>
-#include <sys/module.h>
#include <sys/systm.h> /* uprintf */
#include <sys/errno.h>
#include <sys/param.h> /* defines used in kernel.h */
@@ -51,6 +50,9 @@ typedef struct s_krping {
/* vars */
static struct cdev *krping_dev;
+#undef MODULE_VERSION
+#include <sys/module.h>
+
static int
krping_loader(struct module *m, int what, void *arg)
{
@@ -175,6 +177,4 @@ krping_write(struct cdev *dev, struct ui
return(err);
}
-MODULE_DEPEND(krping, rdma_core, 1, 1, 1);
-MODULE_DEPEND(krping, rdma_cma, 1, 1, 1);
DEV_MODULE(krping,krping_loader,NULL);
Modified: stable/9/sys/contrib/rdma/rdma_addr.c
==============================================================================
--- stable/9/sys/contrib/rdma/rdma_addr.c Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/contrib/rdma/rdma_addr.c Wed Jul 3 09:25:29 2013 (r252555)
@@ -117,7 +117,8 @@ int rdma_copy_addr(struct rdma_dev_addr
const unsigned char *dst_dev_addr)
{
dev_addr->dev_type = RDMA_NODE_RNIC;
- memcpy(dev_addr->src_dev_addr, IF_LLADDR(dev), MAX_ADDR_LEN);
+ memset(dev_addr->src_dev_addr, 0, MAX_ADDR_LEN);
+ memcpy(dev_addr->src_dev_addr, IF_LLADDR(dev), dev->if_addrlen);
memcpy(dev_addr->broadcast, dev->if_broadcastaddr, MAX_ADDR_LEN);
if (dst_dev_addr)
memcpy(dev_addr->dst_dev_addr, dst_dev_addr, MAX_ADDR_LEN);
@@ -207,7 +208,7 @@ static int addr_resolve_remote(struct so
goto put;
}
ret = arpresolve(iproute.ro_rt->rt_ifp, iproute.ro_rt, NULL,
- rt_key(iproute.ro_rt), dmac, &lle);
+ (struct sockaddr *)dst_in, dmac, &lle);
if (ret) {
goto put;
}
Modified: stable/9/sys/contrib/rdma/rdma_cache.c
==============================================================================
--- stable/9/sys/contrib/rdma/rdma_cache.c Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/contrib/rdma/rdma_cache.c Wed Jul 3 09:25:29 2013 (r252555)
@@ -132,7 +132,7 @@ int ib_find_cached_gid(struct ib_device
for (p = 0; p <= end_port(device) - start_port(device); ++p) {
cache = device->cache.gid_cache[p];
for (i = 0; i < cache->table_len; ++i) {
- if (!memcmp(gid, &cache->table[i], 6)) { /* XXX */
+ if (!memcmp(gid, &cache->table[i], sizeof *gid)) {
*port_num = p + start_port(device);
if (index)
*index = i;
Modified: stable/9/sys/dev/cxgb/ulp/tom/cxgb_cpl_io.c
==============================================================================
--- stable/9/sys/dev/cxgb/ulp/tom/cxgb_cpl_io.c Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/dev/cxgb/ulp/tom/cxgb_cpl_io.c Wed Jul 3 09:25:29 2013 (r252555)
@@ -880,10 +880,10 @@ act_open_rpl_status_to_errno(int status)
case CPL_ERR_CONN_TIMEDOUT:
return (ETIMEDOUT);
case CPL_ERR_TCAM_FULL:
- return (ENOMEM);
+ return (EAGAIN);
case CPL_ERR_CONN_EXIST:
log(LOG_ERR, "ACTIVE_OPEN_RPL: 4-tuple in use\n");
- return (EADDRINUSE);
+ return (EAGAIN);
default:
return (EIO);
}
@@ -912,8 +912,7 @@ do_act_open_rpl(struct sge_qset *qs, str
unsigned int atid = G_TID(ntohl(rpl->atid));
struct toepcb *toep = lookup_atid(&td->tid_maps, atid);
struct inpcb *inp = toep->tp_inp;
- struct tcpcb *tp = intotcpcb(inp);
- int s = rpl->status;
+ int s = rpl->status, rc;
CTR3(KTR_CXGB, "%s: atid %u, status %u ", __func__, atid, s);
@@ -923,17 +922,14 @@ do_act_open_rpl(struct sge_qset *qs, str
if (act_open_has_tid(s))
queue_tid_release(tod, GET_TID(rpl));
- if (s == CPL_ERR_TCAM_FULL || s == CPL_ERR_CONN_EXIST) {
- INP_WLOCK(inp);
- toe_connect_failed(tod, tp, EAGAIN);
- toepcb_release(toep); /* unlocks inp */
- } else {
+ rc = act_open_rpl_status_to_errno(s);
+ if (rc != EAGAIN)
INP_INFO_WLOCK(&V_tcbinfo);
- INP_WLOCK(inp);
- toe_connect_failed(tod, tp, act_open_rpl_status_to_errno(s));
- toepcb_release(toep); /* unlocks inp */
+ INP_WLOCK(inp);
+ toe_connect_failed(tod, inp, rc);
+ toepcb_release(toep); /* unlocks inp */
+ if (rc != EAGAIN)
INP_INFO_WUNLOCK(&V_tcbinfo);
- }
m_freem(m);
return (0);
Modified: stable/9/sys/i386/conf/GENERIC
==============================================================================
--- stable/9/sys/i386/conf/GENERIC Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/i386/conf/GENERIC Wed Jul 3 09:25:29 2013 (r252555)
@@ -30,6 +30,7 @@ options SCHED_ULE # ULE scheduler
options PREEMPTION # Enable kernel thread preemption
options INET # InterNETworking
options INET6 # IPv6 communications protocols
+options TCP_OFFLOAD # TCP offload
options SCTP # Stream Control Transmission Protocol
options FFS # Berkeley Fast Filesystem
options SOFTUPDATES # Enable FFS soft updates support
Modified: stable/9/sys/i386/conf/XEN
==============================================================================
--- stable/9/sys/i386/conf/XEN Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/i386/conf/XEN Wed Jul 3 09:25:29 2013 (r252555)
@@ -7,7 +7,7 @@ cpu I686_CPU
ident XEN
makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols
-makeoptions WITHOUT_MODULES="aha ahb amd cxgb dpt drm drm2 hptmv ida malo mps mwl nve sound sym trm xfs"
+makeoptions WITHOUT_MODULES="aha ahb amd cxgb dpt drm drm2 hptmv ida malo mps mwl nve rdma sound sym trm xfs"
options SCHED_ULE # ULE scheduler
options PREEMPTION # Enable kernel thread preemption
Modified: stable/9/sys/modules/Makefile
==============================================================================
--- stable/9/sys/modules/Makefile Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/modules/Makefile Wed Jul 3 09:25:29 2013 (r252555)
@@ -313,6 +313,7 @@ SUBDIR= \
${_ti} \
tl \
tmpfs \
+ ${_toecore} \
${_tpm} \
trm \
${_twa} \
@@ -395,6 +396,7 @@ _if_gre= if_gre
.if (${MK_INET_SUPPORT} != "no" || ${MK_INET6_SUPPORT} != "no") || \
defined(ALL_MODULES)
_if_carp= if_carp
+_toecore= toecore
.endif
.if ${MK_IPFILTER} != "no" || defined(ALL_MODULES)
Modified: stable/9/sys/modules/rdma/krping/Makefile
==============================================================================
--- stable/9/sys/modules/rdma/krping/Makefile Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/modules/rdma/krping/Makefile Wed Jul 3 09:25:29 2013 (r252555)
@@ -6,5 +6,7 @@ RDMA= ${.CURDIR}/../../../contrib/rdma/k
KMOD= krping
SRCS= krping.c krping_dev.c getopt.c
SRCS+= bus_if.h device_if.h opt_sched.h pci_if.h pcib_if.h
+SRCS+= vnode_if.h
+CFLAGS+= -I${.CURDIR}/../../../ofed/include
.include <bsd.kmod.mk>
Modified: stable/9/sys/net/if_llatbl.h
==============================================================================
--- stable/9/sys/net/if_llatbl.h Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/net/if_llatbl.h Wed Jul 3 09:25:29 2013 (r252555)
@@ -205,4 +205,14 @@ lla_lookup(struct lltable *llt, u_int fl
}
int lla_rt_output(struct rt_msghdr *, struct rt_addrinfo *);
+
+#include <sys/eventhandler.h>
+enum {
+ LLENTRY_RESOLVED,
+ LLENTRY_TIMEDOUT,
+ LLENTRY_DELETED,
+ LLENTRY_EXPIRED,
+};
+typedef void (*lle_event_fn)(void *, struct llentry *, int);
+EVENTHANDLER_DECLARE(lle_event, lle_event_fn);
#endif /* _NET_IF_LLATBL_H_ */
Modified: stable/9/sys/net/if_var.h
==============================================================================
--- stable/9/sys/net/if_var.h Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/net/if_var.h Wed Jul 3 09:25:29 2013 (r252555)
@@ -755,6 +755,8 @@ drbr_inuse(struct ifnet *ifp, struct buf
#define IF_MINMTU 72
#define IF_MAXMTU 65535
+#define TOEDEV(ifp) ((ifp)->if_llsoftc)
+
#endif /* _KERNEL */
/*
Modified: stable/9/sys/net/if_vlan.c
==============================================================================
--- stable/9/sys/net/if_vlan.c Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/net/if_vlan.c Wed Jul 3 09:25:29 2013 (r252555)
@@ -748,8 +748,8 @@ vlan_modevent(module_t mod, int type, vo
vlan_trunk_cap_p = NULL;
vlan_trunkdev_p = NULL;
vlan_tag_p = NULL;
- vlan_cookie_p = vlan_cookie;
- vlan_setcookie_p = vlan_setcookie;
+ vlan_cookie_p = NULL;
+ vlan_setcookie_p = NULL;
vlan_devat_p = NULL;
VLAN_LOCK_DESTROY();
if (bootverbose)
@@ -1510,6 +1510,22 @@ vlan_capabilities(struct ifvlan *ifv)
ifp->if_capenable &= ~(p->if_capenable & IFCAP_TSO);
ifp->if_hwassist &= ~(p->if_hwassist & CSUM_TSO);
}
+
+ /*
+ * If the parent interface can offload TCP connections over VLANs then
+ * propagate its TOE capability to the VLAN interface.
+ *
+ * All TOE drivers in the tree today can deal with VLANs. If this
+ * changes then IFCAP_VLAN_TOE should be promoted to a full capability
+ * with its own bit.
+ */
+#define IFCAP_VLAN_TOE IFCAP_TOE
+ if (p->if_capabilities & IFCAP_VLAN_TOE)
+ ifp->if_capabilities |= p->if_capabilities & IFCAP_TOE;
+ if (p->if_capenable & IFCAP_VLAN_TOE) {
+ TOEDEV(ifp) = TOEDEV(p);
+ ifp->if_capenable |= p->if_capenable & IFCAP_TOE;
+ }
}
static void
Modified: stable/9/sys/netinet/if_ether.c
==============================================================================
--- stable/9/sys/netinet/if_ether.c Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/netinet/if_ether.c Wed Jul 3 09:25:29 2013 (r252555)
@@ -178,6 +178,16 @@ arptimer(void *arg)
ifp = lle->lle_tbl->llt_ifp;
CURVNET_SET(ifp->if_vnet);
+ if ((lle->la_flags & LLE_DELETED) == 0) {
+ int evt;
+
+ if (lle->la_flags & LLE_VALID)
+ evt = LLENTRY_EXPIRED;
+ else
+ evt = LLENTRY_TIMEDOUT;
+ EVENTHANDLER_INVOKE(lle_event, lle, evt);
+ }
+
callout_stop(&lle->la_timer);
/* XXX: LOR avoidance. We still have ref on lle. */
@@ -730,7 +740,7 @@ match:
(void)memcpy(&la->ll_addr, ar_sha(ah), ifp->if_addrlen);
la->la_flags |= LLE_VALID;
- EVENTHANDLER_INVOKE(arp_update_event, la);
+ EVENTHANDLER_INVOKE(lle_event, la, LLENTRY_RESOLVED);
if (!(la->la_flags & LLE_STATIC)) {
int canceled;
Modified: stable/9/sys/netinet/if_ether.h
==============================================================================
--- stable/9/sys/netinet/if_ether.h Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/netinet/if_ether.h Wed Jul 3 09:25:29 2013 (r252555)
@@ -117,11 +117,6 @@ int arpresolve(struct ifnet *ifp, struct
struct llentry **lle);
void arp_ifinit(struct ifnet *, struct ifaddr *);
void arp_ifinit2(struct ifnet *, struct ifaddr *, u_char *);
-
-#include <sys/eventhandler.h>
-typedef void (*llevent_arp_update_fn)(void *, struct llentry *);
-EVENTHANDLER_DECLARE(arp_update_event, llevent_arp_update_fn);
-
#endif
#endif
Modified: stable/9/sys/netinet/in.c
==============================================================================
--- stable/9/sys/netinet/in.c Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/netinet/in.c Wed Jul 3 09:25:29 2013 (r252555)
@@ -1527,7 +1527,7 @@ in_lltable_lookup(struct lltable *llt, u
if (!(lle->la_flags & LLE_IFADDR) || (flags & LLE_IFADDR)) {
LLE_WLOCK(lle);
lle->la_flags |= LLE_DELETED;
- EVENTHANDLER_INVOKE(arp_update_event, lle);
+ EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_DELETED);
#ifdef DIAGNOSTIC
log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
#endif
Modified: stable/9/sys/netinet/tcp_input.c
==============================================================================
--- stable/9/sys/netinet/tcp_input.c Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/netinet/tcp_input.c Wed Jul 3 09:25:29 2013 (r252555)
@@ -105,6 +105,9 @@ __FBSDID("$FreeBSD$");
#ifdef TCPDEBUG
#include <netinet/tcp_debug.h>
#endif /* TCPDEBUG */
+#ifdef TCP_OFFLOAD
+#include <netinet/tcp_offload.h>
+#endif
#ifdef IPSEC
#include <netipsec/ipsec.h>
@@ -995,6 +998,14 @@ relocked:
goto dropwithreset;
}
+#ifdef TCP_OFFLOAD
+ if (tp->t_flags & TF_TOE) {
+ tcp_offload_input(tp, m);
+ m = NULL; /* consumed by the TOE driver */
+ goto dropunlock;
+ }
+#endif
+
/*
* We've identified a valid inpcb, but it could be that we need an
* inpcbinfo write lock but don't hold it. In this case, attempt to
Modified: stable/9/sys/netinet/tcp_offload.c
==============================================================================
--- stable/9/sys/netinet/tcp_offload.c Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/netinet/tcp_offload.c Wed Jul 3 09:25:29 2013 (r252555)
@@ -1,145 +1,176 @@
/*-
- * Copyright (c) 2007, Chelsio Inc.
+ * Copyright (c) 2012 Chelsio Communications, Inc.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * 2. 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.
*
- * 1. Redistributions of source code must retain the above copyright notice,
- * this list of conditions and the following disclaimer.
- *
- * 2. Neither the name of the Chelsio Corporation 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
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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.
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR 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("$FreeBSD$");
+#include "opt_inet.h"
+
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/types.h>
-#include <sys/malloc.h>
-#include <sys/kernel.h>
-#include <sys/sysctl.h>
#include <sys/mbuf.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
-
+#include <sys/sockopt.h>
#include <net/if.h>
-#include <net/if_types.h>
-#include <net/if_var.h>
#include <net/route.h>
-#include <net/vnet.h>
-
#include <netinet/in.h>
-#include <netinet/in_systm.h>
#include <netinet/in_pcb.h>
#include <netinet/tcp.h>
#include <netinet/tcp_var.h>
#include <netinet/tcp_offload.h>
-#include <netinet/toedev.h>
+#define TCPOUTFLAGS
+#include <netinet/tcp_fsm.h>
+#include <netinet/toecore.h>
-uint32_t toedev_registration_count;
+int registered_toedevs;
+/*
+ * Provide an opportunity for a TOE driver to offload.
+ */
int
tcp_offload_connect(struct socket *so, struct sockaddr *nam)
{
struct ifnet *ifp;
- struct toedev *tdev;
+ struct toedev *tod;
struct rtentry *rt;
- int error;
+ int error = EOPNOTSUPP;
+
+ INP_WLOCK_ASSERT(sotoinpcb(so));
+ KASSERT(nam->sa_family == AF_INET || nam->sa_family == AF_INET6,
+ ("%s: called with sa_family %d", __func__, nam->sa_family));
- if (toedev_registration_count == 0)
- return (EINVAL);
-
- /*
- * Look up the route used for the connection to
- * determine if it uses an interface capable of
- * offloading the connection.
- */
- rt = rtalloc1(nam, 0 /*report*/, 0 /*ignflags*/);
- if (rt)
+ if (registered_toedevs == 0)
+ return (error);
+
+ rt = rtalloc1(nam, 0, 0);
+ if (rt)
RT_UNLOCK(rt);
- else
+ else
return (EHOSTUNREACH);
ifp = rt->rt_ifp;
- if ((ifp->if_capenable & IFCAP_TOE) == 0) {
- error = EINVAL;
- goto fail;
- }
-
- tdev = TOEDEV(ifp);
- if (tdev == NULL) {
- error = EPERM;
- goto fail;
- }
-
- if (tdev->tod_can_offload(tdev, so) == 0) {
- error = EPERM;
- goto fail;
- }
-
- return (tdev->tod_connect(tdev, so, rt, nam));
-fail:
+
+ if (nam->sa_family == AF_INET && !(ifp->if_capenable & IFCAP_TOE4))
+ goto done;
+ if (nam->sa_family == AF_INET6 && !(ifp->if_capenable & IFCAP_TOE6))
+ goto done;
+
+ tod = TOEDEV(ifp);
+ if (tod != NULL)
+ error = tod->tod_connect(tod, so, rt, nam);
+done:
RTFREE(rt);
return (error);
}
+void
+tcp_offload_listen_start(struct tcpcb *tp)
+{
-/*
- * This file contains code as a short-term staging area before it is moved in
- * to sys/netinet/tcp_offload.c
- */
+ INP_WLOCK_ASSERT(tp->t_inpcb);
+
+ EVENTHANDLER_INVOKE(tcp_offload_listen_start, tp);
+}
void
-tcp_offload_twstart(struct tcpcb *tp)
+tcp_offload_listen_stop(struct tcpcb *tp)
{
- INP_INFO_WLOCK(&V_tcbinfo);
- INP_WLOCK(tp->t_inpcb);
- tcp_twstart(tp);
- INP_INFO_WUNLOCK(&V_tcbinfo);
+ INP_WLOCK_ASSERT(tp->t_inpcb);
+
+ EVENTHANDLER_INVOKE(tcp_offload_listen_stop, tp);
}
-struct tcpcb *
-tcp_offload_close(struct tcpcb *tp)
+void
+tcp_offload_input(struct tcpcb *tp, struct mbuf *m)
{
+ struct toedev *tod = tp->tod;
- INP_INFO_WLOCK(&V_tcbinfo);
- INP_WLOCK(tp->t_inpcb);
- tp = tcp_close(tp);
- INP_INFO_WUNLOCK(&V_tcbinfo);
- if (tp)
- INP_WUNLOCK(tp->t_inpcb);
+ KASSERT(tod != NULL, ("%s: tp->tod is NULL, tp %p", __func__, tp));
+ INP_WLOCK_ASSERT(tp->t_inpcb);
- return (tp);
+ tod->tod_input(tod, tp, m);
}
-struct tcpcb *
-tcp_offload_drop(struct tcpcb *tp, int error)
+int
+tcp_offload_output(struct tcpcb *tp)
{
+ struct toedev *tod = tp->tod;
+ int error, flags;
+
+ KASSERT(tod != NULL, ("%s: tp->tod is NULL, tp %p", __func__, tp));
+ INP_WLOCK_ASSERT(tp->t_inpcb);
- INP_INFO_WLOCK(&V_tcbinfo);
- INP_WLOCK(tp->t_inpcb);
- tp = tcp_drop(tp, error);
- INP_INFO_WUNLOCK(&V_tcbinfo);
- if (tp)
- INP_WUNLOCK(tp->t_inpcb);
+ flags = tcp_outflags[tp->t_state];
- return (tp);
+ if (flags & TH_RST) {
+ /* XXX: avoid repeated calls like we do for FIN */
+ error = tod->tod_send_rst(tod, tp);
+ } else if ((flags & TH_FIN || tp->t_flags & TF_NEEDFIN) &&
+ (tp->t_flags & TF_SENTFIN) == 0) {
+ error = tod->tod_send_fin(tod, tp);
+ if (error == 0)
+ tp->t_flags |= TF_SENTFIN;
+ } else
+ error = tod->tod_output(tod, tp);
+
+ return (error);
+}
+
+void
+tcp_offload_rcvd(struct tcpcb *tp)
+{
+ struct toedev *tod = tp->tod;
+
+ KASSERT(tod != NULL, ("%s: tp->tod is NULL, tp %p", __func__, tp));
+ INP_WLOCK_ASSERT(tp->t_inpcb);
+
+ tod->tod_rcvd(tod, tp);
}
+void
+tcp_offload_ctloutput(struct tcpcb *tp, int sopt_dir, int sopt_name)
+{
+ struct toedev *tod = tp->tod;
+
+ KASSERT(tod != NULL, ("%s: tp->tod is NULL, tp %p", __func__, tp));
+ INP_WLOCK_ASSERT(tp->t_inpcb);
+
+ tod->tod_ctloutput(tod, tp, sopt_dir, sopt_name);
+}
+
+void
+tcp_offload_detach(struct tcpcb *tp)
+{
+ struct toedev *tod = tp->tod;
+
+ KASSERT(tod != NULL, ("%s: tp->tod is NULL, tp %p", __func__, tp));
+ INP_WLOCK_ASSERT(tp->t_inpcb);
+
+ tod->tod_pcb_detach(tod, tp);
+}
Modified: stable/9/sys/netinet/tcp_offload.h
==============================================================================
--- stable/9/sys/netinet/tcp_offload.h Wed Jul 3 08:39:18 2013 (r252554)
+++ stable/9/sys/netinet/tcp_offload.h Wed Jul 3 09:25:29 2013 (r252555)
@@ -1,30 +1,30 @@
/*-
- * Copyright (c) 2007, Chelsio Inc.
+ * Copyright (c) 2012 Chelsio Communications, Inc.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * 2. 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.
*
- * 1. Redistributions of source code must retain the above copyright notice,
- * this list of conditions and the following disclaimer.
- *
- * 2. Neither the name of the Chelsio Corporation 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
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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.
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR 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.
*
* $FreeBSD$
+ *
*/
#ifndef _NETINET_TCP_OFFLOAD_H_
@@ -34,321 +34,15 @@
#error "no user-serviceable parts inside"
#endif
-/*
- * A driver publishes that it provides offload services
- * by setting IFCAP_TOE in the ifnet. The offload connect
- * will bypass any further work if the interface that a
- * connection would use does not support TCP offload.
- *
- * The TOE API assumes that the tcp offload engine can offload the
- * the entire connection from set up to teardown, with some provision
- * being made to allowing the software stack to handle time wait. If
- * the device does not meet these criteria, it is the driver's responsibility
- * to overload the functions that it needs to in tcp_usrreqs and make
- * its own calls to tcp_output if it needs to do so.
- *
- * There is currently no provision for the device advertising the congestion
- * control algorithms it supports as there is currently no API for querying
- * an operating system for the protocols that it has loaded. This is a desirable
- * future extension.
- *
- *
- *
- * It is assumed that individuals deploying TOE will want connections
- * to be offloaded without software changes so all connections on an
- * interface providing TOE are offloaded unless the SO_NO_OFFLOAD
- * flag is set on the socket.
- *
- *
- * The toe_usrreqs structure constitutes the TOE driver's
- * interface to the TCP stack for functionality that doesn't
- * interact directly with userspace. If one wants to provide
- * (optional) functionality to do zero-copy to/from
- * userspace one still needs to override soreceive/sosend
- * with functions that fault in and pin the user buffers.
- *
- * + tu_send
- * - tells the driver that new data may have been added to the
- * socket's send buffer - the driver should not fail if the
- * buffer is in fact unchanged
- * - the driver is responsible for providing credits (bytes in the send window)
- * back to the socket by calling sbdrop() as segments are acknowledged.
- * - The driver expects the inpcb lock to be held - the driver is expected
- * not to drop the lock. Hence the driver is not allowed to acquire the
- * pcbinfo lock during this call.
- *
- * + tu_rcvd
- * - returns credits to the driver and triggers window updates
- * to the peer (a credit as used here is a byte in the peer's receive window)
- * - the driver is expected to determine how many bytes have been
- * consumed and credit that back to the card so that it can grow
- * the window again by maintaining its own state between invocations.
- * - In principle this could be used to shrink the window as well as
- * grow the window, although it is not used for that now.
- * - this function needs to correctly handle being called any number of
- * times without any bytes being consumed from the receive buffer.
- * - The driver expects the inpcb lock to be held - the driver is expected
- * not to drop the lock. Hence the driver is not allowed to acquire the
- * pcbinfo lock during this call.
- *
- * + tu_disconnect
- * - tells the driver to send FIN to peer
- * - driver is expected to send the remaining data and then do a clean half close
- * - disconnect implies at least half-close so only send, reset, and detach
- * are legal
- * - the driver is expected to handle transition through the shutdown
- * state machine and allow the stack to support SO_LINGER.
- * - The driver expects the inpcb lock to be held - the driver is expected
- * not to drop the lock. Hence the driver is not allowed to acquire the
- * pcbinfo lock during this call.
- *
- * + tu_reset
- * - closes the connection and sends a RST to peer
- * - driver is expectd to trigger an RST and detach the toepcb
- * - no further calls are legal after reset
- * - The driver expects the inpcb lock to be held - the driver is expected
- * not to drop the lock. Hence the driver is not allowed to acquire the
- * pcbinfo lock during this call.
- *
- * The following fields in the tcpcb are expected to be referenced by the driver:
- * + iss
- * + rcv_nxt
- * + rcv_wnd
- * + snd_isn
- * + snd_max
- * + snd_nxt
- * + snd_una
- * + t_flags
- * + t_inpcb
- * + t_maxseg
- * + t_toe
- *
- * The following fields in the inpcb are expected to be referenced by the driver:
- * + inp_lport
- * + inp_fport
- * + inp_laddr
- * + inp_fport
- * + inp_socket
- * + inp_ip_tos
- *
- * The following fields in the socket are expected to be referenced by the
- * driver:
- * + so_comp
- * + so_error
- * + so_linger
- * + so_options
- * + so_rcv
- * + so_snd
- * + so_state
- * + so_timeo
- *
- * These functions all return 0 on success and can return the following errors
- * as appropriate:
- * + EPERM:
- * + ENOBUFS: memory allocation failed
- * + EMSGSIZE: MTU changed during the call
- * + EHOSTDOWN:
- * + EHOSTUNREACH:
- * + ENETDOWN:
- * * ENETUNREACH: the peer is no longer reachable
- *
- * + tu_detach
- * - tells driver that the socket is going away so disconnect
- * the toepcb and free appropriate resources
- * - allows the driver to cleanly handle the case of connection state
- * outliving the socket
- * - no further calls are legal after detach
- * - the driver is expected to provide its own synchronization between
- * detach and receiving new data.
- *
- * + tu_syncache_event
- * - even if it is not actually needed, the driver is expected to
- * call syncache_add for the initial SYN and then syncache_expand
- * for the SYN,ACK
- * - tells driver that a connection either has not been added or has
- * been dropped from the syncache
- * - the driver is expected to maintain state that lives outside the
- * software stack so the syncache needs to be able to notify the
- * toe driver that the software stack is not going to create a connection
- * for a received SYN
- * - The driver is responsible for any synchronization required between
- * the syncache dropping an entry and the driver processing the SYN,ACK.
- *
- */
-struct toe_usrreqs {
- int (*tu_send)(struct tcpcb *tp);
- int (*tu_rcvd)(struct tcpcb *tp);
- int (*tu_disconnect)(struct tcpcb *tp);
- int (*tu_reset)(struct tcpcb *tp);
- void (*tu_detach)(struct tcpcb *tp);
- void (*tu_syncache_event)(int event, void *toep);
-};
-
-/*
- * Proxy for struct tcpopt between TOE drivers and TCP functions.
- */
-struct toeopt {
- u_int64_t to_flags; /* see tcpopt in tcp_var.h */
- u_int16_t to_mss; /* maximum segment size */
- u_int8_t to_wscale; /* window scaling */
-
- u_int8_t _pad1; /* explicit pad for 64bit alignment */
- u_int32_t _pad2; /* explicit pad for 64bit alignment */
- u_int64_t _pad3[4]; /* TBD */
-};
-
-#define TOE_SC_ENTRY_PRESENT 1 /* 4-tuple already present */
-#define TOE_SC_DROP 2 /* connection was timed out */
-
-/*
- * Because listen is a one-to-many relationship (a socket can be listening
- * on all interfaces on a machine some of which may be using different TCP
- * offload devices), listen uses a publish/subscribe mechanism. The TCP
- * offload driver registers a listen notification function with the stack.
- * When a listen socket is created all TCP offload devices are notified
- * so that they can do the appropriate set up to offload connections on the
- * port to which the socket is bound. When the listen socket is closed,
- * the offload devices are notified so that they will stop listening on that
- * port and free any associated resources as well as sending RSTs on any
- * connections in the SYN_RCVD state.
- *
- */
-
-typedef void (*tcp_offload_listen_start_fn)(void *, struct tcpcb *);
-typedef void (*tcp_offload_listen_stop_fn)(void *, struct tcpcb *);
*** DIFF OUTPUT TRUNCATED AT 1000 LINES ***
More information about the svn-src-all
mailing list