Commit Graph

28 Commits

Author SHA1 Message Date
rwatson
6e29aea1db Implement a CPU-affine TCP and UDP connection lookup data structure,
struct inpcbgroup.  pcbgroups, or "connection groups", supplement the
existing inpcbinfo connection hash table, which when pcbgroups are
enabled, might now be thought of more usefully as a per-protocol
4-tuple reservation table.

Connections are assigned to connection groups base on a hash of their
4-tuple; wildcard sockets require special handling, and are members
of all connection groups.  During a connection lookup, a
per-connection group lock is employed rather than the global pcbinfo
lock.  By aligning connection groups with input path processing,
connection groups take on an effective CPU affinity, especially when
aligned with RSS work placement (see a forthcoming commit for
details).  This eliminates cache line migration associated with
global, protocol-layer data structures in steady state TCP and UDP
processing (with the exception of protocol-layer statistics; further
commit to follow).

Elements of this approach were inspired by Willman, Rixner, and Cox's
2006 USENIX paper, "An Evaluation of Network Stack Parallelization
Strategies in Modern Operating Systems".  However, there are also
significant differences: we maintain the inpcb lock, rather than using
the connection group lock for per-connection state.

Likewise, the focus of this implementation is alignment with NIC
packet distribution strategies such as RSS, rather than pure software
strategies.  Despite that focus, software distribution is supported
through the parallel netisr implementation, and works well in
configurations where the number of hardware threads is greater than
the number of NIC input queues, such as in the RMI XLR threaded MIPS
architecture.

Another important difference is the continued maintenance of existing
hash tables as "reservation tables" -- these are useful both to
distinguish the resource allocation aspect of protocol name management
and the more common-case lookup aspect.  In configurations where
connection tables are aligned with hardware hashes, it is desirable to
use the traditional lookup tables for loopback or encapsulated traffic
rather than take the expense of hardware hashes that are hard to
implement efficiently in software (such as RSS Toeplitz).

Connection group support is enabled by compiling "options PCBGROUP"
into your kernel configuration; for the time being, this is an
experimental feature, and hence is not enabled by default.

Subject to the limited MFCability of change dependencies in inpcb,
and its change to the inpcbinfo init function signature, this change
in principle could be merged to FreeBSD 8.x.

Reviewed by:    bz
Sponsored by:   Juniper Networks, Inc.
2011-06-06 12:55:02 +00:00
rwatson
e9eb5d3b9c Add _mbuf() variants of various inpcb-related interfaces, including lookup,
hash install, etc.  For now, these are arguments are unused, but as we add
RSS support, we will want to use hashes extracted from mbufs, rather than
manually calculated hashes of header fields, due to the expensive of the
software version of Toeplitz (and similar hashes).

Add notes that it would be nice to be able to pass mbufs into lookup
routines in pf(4), optimising firewall lookup in the same way, but the
code structure there doesn't facilitate that currently.

(In principle there is no reason this couldn't be MFCed -- the change
extends rather than modifies the KBI.  However, it won't be useful without
other previous possibly less MFCable changes.)

Reviewed by:    bz
Sponsored by:   Juniper Networks, Inc.
2011-06-04 16:33:06 +00:00
rwatson
fdfdadb612 Decompose the current single inpcbinfo lock into two locks:
- The existing ipi_lock continues to protect the global inpcb list and
  inpcb counter.  This lock is now relegated to a small number of
  allocation and free operations, and occasional operations that walk
  all connections (including, awkwardly, certain UDP multicast receive
  operations -- something to revisit).

- A new ipi_hash_lock protects the two inpcbinfo hash tables for
  looking up connections and bound sockets, manipulated using new
  INP_HASH_*() macros.  This lock, combined with inpcb locks, protects
  the 4-tuple address space.

Unlike the current ipi_lock, ipi_hash_lock follows the individual inpcb
connection locks, so may be acquired while manipulating a connection on
which a lock is already held, avoiding the need to acquire the inpcbinfo
lock preemptively when a binding change might later be required.  As a
result, however, lookup operations necessarily go through a reference
acquire while holding the lookup lock, later acquiring an inpcb lock --
if required.

A new function in_pcblookup() looks up connections, and accepts flags
indicating how to return the inpcb.  Due to lock order changes, callers
no longer need acquire locks before performing a lookup: the lookup
routine will acquire the ipi_hash_lock as needed.  In the future, it will
also be able to use alternative lookup and locking strategies
transparently to callers, such as pcbgroup lookup.  New lookup flags are,
supplementing the existing INPLOOKUP_WILDCARD flag:

  INPLOOKUP_RLOCKPCB - Acquire a read lock on the returned inpcb
  INPLOOKUP_WLOCKPCB - Acquire a write lock on the returned inpcb

Callers must pass exactly one of these flags (for the time being).

Some notes:

- All protocols are updated to work within the new regime; especially,
  TCP, UDPv4, and UDPv6.  pcbinfo ipi_lock acquisitions are largely
  eliminated, and global hash lock hold times are dramatically reduced
  compared to previous locking.
- The TCP syncache still relies on the pcbinfo lock, something that we
  may want to revisit.
- Support for reverting to the FreeBSD 7.x locking strategy in TCP input
  is no longer available -- hash lookup locks are now held only very
  briefly during inpcb lookup, rather than for potentially extended
  periods.  However, the pcbinfo ipi_lock will still be acquired if a
  connection state might change such that a connection is added or
  removed.
- Raw IP sockets continue to use the pcbinfo ipi_lock for protection,
  due to maintaining their own hash tables.
- The interface in6_pcblookup_hash_locked() is maintained, which allows
  callers to acquire hash locks and perform one or more lookups atomically
  with 4-tuple allocation: this is required only for TCPv6, as there is no
  in6_pcbconnect_setup(), which there should be.
- UDPv6 locking remains significantly more conservative than UDPv4
  locking, which relates to source address selection.  This needs
  attention, as it likely significantly reduces parallelism in this code
  for multithreaded socket use (such as in BIND).
- In the UDPv4 and UDPv6 multicast cases, we need to revisit locking
  somewhat, as they relied on ipi_lock to stablise 4-tuple matches, which
  is no longer sufficient.  A second check once the inpcb lock is held
  should do the trick, keeping the general case from requiring the inpcb
  lock for every inpcb visited.
- This work reminds us that we need to revisit locking of the v4/v6 flags,
  which may be accessed lock-free both before and after this change.
- Right now, a single lock name is used for the pcbhash lock -- this is
  undesirable, and probably another argument is required to take care of
  this (or a char array name field in the pcbinfo?).

This is not an MFC candidate for 8.x due to its impact on lookup and
locking semantics.  It's possible some of these issues could be worked
around with compatibility wrappers, if necessary.

Reviewed by:    bz
Sponsored by:   Juniper Networks, Inc.
2011-05-30 09:43:55 +00:00
bz
a8839212d2 Make callers to in6_selectsrc() and in6_pcbladdr() pass in memory
to save the selected source address rather than returning an
unreferenced copy to a pointer that might long be gone by the
time we use the pointer for anything meaningful.

Asked for by:	rwatson
Reviewed by:	rwatson
2009-06-23 22:08:55 +00:00
bz
864141180e Merge in6_pcbfree() into in_pcbfree() which after the previous
IPsec change in r185366 only differed in two additonal IPv6 lines.
Rather than splattering conditional code everywhere add the v6
check centrally at this single place.

Reviewed by:	rwatson (as part of a larger changset)
MFC after:	6 weeks (*)
(*) possibly need to leave a stub wrapper in 7 to keep the symbol.
2008-11-27 12:04:35 +00:00
bz
b089198d31 Remove in6_pcbdetach() as it is exactly the same function
as in_pcbdetach() and we don't need the code twice.

Reviewed by:	rwatson
MFC after:	6 weeks (*)
(*) possibly need to leave a stub wrapper in 7 to keep the symbol.
2008-11-26 20:52:26 +00:00
bz
362cb79214 Pass the ucred along into in{,6}_pcblookup_local for upcoming
prison checks.

Reviewed by:	rwatson
2008-07-10 13:31:11 +00:00
bz
4b9bb0069f For consistency take lport as u_short in in{,6}_pcblookup_local.
All callers either pass in an u_short or u_int16_t.

Reviewed by:	rwatson
2008-07-10 13:23:22 +00:00
obrien
0d684d927b Clean up VCS Ids. 2007-12-10 16:03:40 +00:00
rwatson
47d37a80be Reduce network stack oddness: implement .pru_sockaddr and .pru_peeraddr
protocol entry points using functions named proto_getsockaddr and
proto_getpeeraddr rather than proto_setsockaddr and proto_setpeeraddr.
While it's true that sockaddrs are allocated and set, the net effect is
to retrieve (get) the socket address or peer address from a socket, not
set it, so align names to that intent.
2007-05-11 10:20:51 +00:00
rwatson
00366b63e3 Modify in6_pcbpurgeif0() to accept a pcbinfo structure rather than a pcb
list head structure; this improves congruence to IPv4, and also allows
in6_pcbpurgeif0() to lock the pcbinfo.  Modify in6_pcbpurgeif0() to lock
the pcbinfo before iterating the pcb list, use queue(9)'s LIST_FOREACH()
for the iteration, and to lock individual inpcb's while manipulating
them.

MFC after:	3 months
2006-04-23 15:06:16 +00:00
rwatson
71cc03392b Break out in_pcbdetach() into two functions:
- in_pcbdetach(), which removes the link between an inpcb and its
  socket.

- in_pcbfree(), which frees a detached pcb.

Unlike the previous in_pcbdetach(), neither of these functions will
attempt to conditionally free the socket, as they are responsible only
for managing in_pcb memory.  Mirror these changes into in6_pcbdetach()
by breaking it into in6_pcbdetach() and in6_pcbfree().

While here, eliminate undesired checks for NULL inpcb pointers in
sockets, as we will now have as an invariant that sockets will always
have valid so_pcb pointers.

MFC after:	3 months
2006-04-01 16:04:42 +00:00
imp
2b54eeafae /* -> /*- for license, minor formatting changes, separate for KAME 2005-01-07 02:30:35 +00:00
rwatson
11a2e8ce3c Pass pcbinfo structures to in6_pcbnotify() rather than pcbhead
structures, allowing in6_pcbnotify() to lock the pcbinfo and each
inpcb that it notifies of ICMPv6 events.  This prevents inpcb
assertions from firing when IPv6 generates and delievers event
notifications for inpcbs.

Reported by:	kuriyama
Tested by:	kuriyama
2004-08-06 03:45:45 +00:00
imp
b49b7fe799 Remove advertising clause from University of California Regent's
license, per letter dated July 22, 1999 and email from Peter Wemm,
Alan Cox and Robert Watson.

Approved by: core, peter, alc, rwatson
2004-04-07 20:46:16 +00:00
pjd
49554d1bd8 Reduce 'td' argument to 'cred' (struct ucred) argument in those functions:
- in_pcbbind(),
	- in_pcbbind_setup(),
	- in_pcbconnect(),
	- in_pcbconnect_setup(),
	- in6_pcbbind(),
	- in6_pcbconnect(),
	- in6_pcbsetport().
"It should simplify/clarify things a great deal." --rwatson

Requested by:	rwatson
Reviewed by:	rwatson, ume
2004-03-27 21:05:46 +00:00
pjd
bfbc341729 Remove unused prototype.
Reviewed by:	ume
2004-03-27 20:38:02 +00:00
ume
f35565e63f supported IPV6_RECVPATHMTU socket option.
Obtained from:	KAME
2004-02-13 14:50:01 +00:00
ume
41930d5085 source address selection part of RFC3484.
TODO: since there is scope issue to be solved, multicast and
link-local address are treated as special for workaround for
now.

Obtained from:	KAME
2003-11-04 20:22:33 +00:00
truckman
7199888e8f Create new functions in_sockaddr(), in6_sockaddr(), and
in6_v4mapsin6_sockaddr() which allocate the appropriate sockaddr_in*
structure and initialize it with the address and port information passed
as arguments.  Use calls to these new functions to replace code that is
replicated multiple times in in_setsockaddr(), in_setpeeraddr(),
in6_setsockaddr(), in6_setpeeraddr(), in6_mapped_sockaddr(), and
in6_mapped_peeraddr().  Inline COMMON_END in tcp_usr_accept() so that
we can call in_sockaddr() with temporary copies of the address and port
after the PCB is unlocked.

Fix the lock violation in tcp6_usr_accept() (caused by calling MALLOC()
inside in6_mapped_peeraddr() while the PCB is locked) by changing
the implementation of tcp6_usr_accept() to match tcp_usr_accept().

Reviewed by:	suz
2002-08-21 11:57:12 +00:00
hsu
abda76de0b Notify functions can destroy the pcb, so they have to return an
indication of whether this happenned so the calling function
knows whether or not to unlock the pcb.

Submitted by:	Jennifer Yang (yangjihui@yahoo.com)
Bug reported by:  Sid Carter (sidcarter@symonds.net)
2002-06-14 08:35:21 +00:00
alfred
9da4687481 Fix warnings caused by discarding const.
Hairy Eyeball At: peter
2002-02-27 02:44:45 +00:00
julian
5596676e6c KSE Milestone 2
Note ALL MODULES MUST BE RECOMPILED
make the kernel aware that there are smaller units of scheduling than the
process. (but only allow one thread per process at this time).
This is functionally equivalent to teh previousl -current except
that there is a thread associated with each process.

Sorry john! (your next MFC will be a doosie!)

Reviewed by: peter@freebsd.org, dillon@freebsd.org

X-MFC after:    ha ha ha ha
2001-09-12 08:38:13 +00:00
ume
215c0c107e When running aplication joined multicast address,
removing network card, and kill aplication.
imo_membership[].inm_ifp refer interface pointer
after removing interface.
When kill aplication, release socket,and imo_membership.
imo_membership use already not exist interface pointer.
Then, kernel panic.

PR:		29345
Submitted by:	Inoue Yuichi <inoue@nd.net.fujitsu.co.jp>
Obtained from:	KAME
MFC after:	3 days
2001-08-04 17:10:14 +00:00
ume
832f8d2249 Sync with recent KAME.
This work was based on kame-20010528-freebsd43-snap.tgz and some
critical problem after the snap was out were fixed.
There are many many changes since last KAME merge.

TODO:
  - The definitions of SADB_* in sys/net/pfkeyv2.h are still different
    from RFC2407/IANA assignment because of binary compatibility
    issue.  It should be fixed under 5-CURRENT.
  - ip6po_m member of struct ip6_pktopts is no longer used.  But, it
    is still there because of binary compatibility issue.  It should
    be removed under 5-CURRENT.

Reviewed by:	itojun
Obtained from:	KAME
MFC after:	3 weeks
2001-06-11 12:39:29 +00:00
itojun
5f4e854de1 sync with kame tree as of july00. tons of bug fixes/improvements.
API changes:
- additional IPv6 ioctls
- IPsec PF_KEY API was changed, it is mandatory to upgrade setkey(8).
  (also syntax change)
2000-07-04 16:35:15 +00:00
peter
15b9bcb121 Change #ifdef KERNEL to #ifdef _KERNEL in the public headers. "KERNEL"
is an application space macro and the applications are supposed to be free
to use it as they please (but cannot).  This is consistant with the other
BSD's who made this change quite some time ago.  More commits to come.
1999-12-29 04:46:21 +00:00
shin
cad2014b27 KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP
for IPv6 yet)

With this patch, you can assigne IPv6 addr automatically, and can reply to
IPv6 ping.

Reviewed by: freebsd-arch, cvs-committers
Obtained from: KAME project
1999-11-22 02:45:11 +00:00