the INP_INFO lock from tcp_usr_accept. As the PR/patch states
this was following the advice already in the code.
See the PR below for a full disucssion of this change and its
measured effects.
PR: 183659
Submitted by: Julian Charbon
Reviewed by: jhb
of a syncache connection, copy it into the inp_flowid field.
Without this, an incoming TCP connection won't have an inp_flowid marked
until some data comes in, and this means that things like the per-CPU
TCP timer option will choose a different CPU for the timer work.
(It also means that if one grabbed the flowid via an ioctl from userland,
it won't be available until some data has been received.)
Sponsored by: Netflix, Inc.
to this event, adding if_var.h to files that do need it. Also, include
all includes that now are included due to implicit pollution via if_var.h
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
dynamic translation so that their arguments match the definitions for
these providers in Solaris and illumos. Thus, existing scripts for these
providers should work unmodified on FreeBSD.
Tested by: gnn, hiren
MFC after: 1 month
information into the ISN (initial sequence number) without the additional
use of timestamp bits and switching to the very fast and cryptographically
strong SipHash-2-4 MAC hash algorithm to protect the SYN cookie against
forgeries.
The purpose of SYN cookies is to encode all necessary session state in
the 32 bits of our initial sequence number to avoid storing any information
locally in memory. This is especially important when under heavy spoofed
SYN attacks where we would either run out of memory or the syncache would
fill with bogus connection attempts swamping out legitimate connections.
The original SYN cookies method only stored an indexed MSS values in the
cookie. This isn't sufficient anymore and breaks down in the presence of
WSCALE information which is only exchanged during SYN and SYN-ACK. If we
can't keep track of it then we may severely underestimate the available
send or receive window. This is compounded with large windows whose size
information on the TCP segment header is even lower numerically. A number
of years back SYN cookies were extended to store the additional state in
the TCP timestamp fields, if available on a connection. While timestamps
are common among the BSD, Linux and other *nix systems Windows never enabled
them by default and thus are not present for the vast majority of clients
seen on the Internet.
The common parameters used on TCP sessions have changed quite a bit since
SYN cookies very invented some 17 years ago. Today we have a lot more
bandwidth available making the use window scaling almost mandatory. Also
SACK has become standard making recovering from packet loss much more
efficient.
This change moves all necessary information into the ISS removing the need
for timestamps. Both the MSS (16 bits) and send WSCALE (4 bits) are stored
in 3 bit indexed form together with a single bit for SACK. While this is
significantly less than the original range, it is sufficient to encode all
common values with minimal rounding.
The MSS depends on the MTU of the path and with the dominance of ethernet
the main value seen is around 1460 bytes. Encapsulations for DSL lines
and some other overheads reduce it by a few more bytes for many connections
seen. Rounding down to the next lower value in some cases isn't a problem
as we send only slightly more packets for the same amount of data.
The send WSCALE index is bit more tricky as rounding down under-estimates
the available send space available towards the remote host, however a small
number values dominate and are carefully selected again.
The receive WSCALE isn't encoded at all but recalculated based on the local
receive socket buffer size when a valid SYN cookie returns. A listen socket
buffer size is unlikely to change while active.
The index values for MSS and WSCALE are selected for minimal rounding errors
based on large traffic surveys. These values have to be periodically
validated against newer traffic surveys adjusting the arrays tcp_sc_msstab[]
and tcp_sc_wstab[] if necessary.
In addition the hash MAC to protect the SYN cookies is changed from MD5
to SipHash-2-4, a much faster and cryptographically secure algorithm.
Reviewed by: dwmalone
Tested by: Fabian Keil <fk@fabiankeil.de>
from an unprotected u_int that reports garbage on SMP to a function
based sysctl obtaining the current value from UMA.
Also read back the actual cache_limit after page size rounding by UMA.
PR: kern/165879
MFC after: 2 weeks
reduce the initial CWND to one segment. This reduction got lost
some time ago due to a change in initialization ordering.
Additionally in tcp_timer_rexmt() avoid entering fast recovery when
we're still in TCPS_SYN_SENT state.
MFC after: 2 weeks
in network byte order. Any host byte order processing is
done in local variables and host byte order values are
never[1] written to a packet.
After this change a packet processed by the stack isn't
modified at all[2] except for TTL.
After this change a network stack hacker doesn't need to
scratch his head trying to figure out what is the byte order
at the given place in the stack.
[1] One exception still remains. The raw sockets convert host
byte order before pass a packet to an application. Probably
this would remain for ages for compatibility.
[2] The ip_input() still subtructs header len from ip->ip_len,
but this is planned to be fixed soon.
Reviewed by: luigi, Maxim Dounin <mdounin mdounin.ru>
Tested by: ray, Olivier Cochard-Labbe <olivier cochard.me>
- Stateful TCP offload drivers for Terminator 3 and 4 (T3 and T4) ASICs.
These are available as t3_tom and t4_tom modules that augment cxgb(4)
and cxgbe(4) respectively. The cxgb/cxgbe drivers continue to work as
usual with or without these extra features.
- iWARP driver for Terminator 3 ASIC (kernel verbs). T4 iWARP in the
works and will follow soon.
Build-tested with make universe.
30s overview
============
What interfaces support TCP offload? Look for TOE4 and/or TOE6 in the
capabilities of an interface:
# ifconfig -m | grep TOE
Enable/disable TCP offload on an interface (just like any other ifnet
capability):
# ifconfig cxgbe0 toe
# ifconfig cxgbe0 -toe
Which connections are offloaded? Look for toe4 and/or toe6 in the
output of netstat and sockstat:
# netstat -np tcp | grep toe
# sockstat -46c | grep toe
Reviewed by: bz, gnn
Sponsored by: Chelsio communications.
MFC after: ~3 months (after 9.1, and after ensuring MFC is feasible)
headers for TSO but also for generic checksum offloading. Ideally we
would only have one common function shared amongst all drivers, and
perhaps when updating them for IPv6 we should introduce that.
Eventually we should provide the meta information along with mbufs to
avoid (re-)parsing entirely.
To not break IPv6 (checksums and offload) and to be able to MFC the
changes without risking to hurt 3rd party drivers, duplicate the v4
framework, as other OSes have done as well.
Introduce interface capability flags for TX/RX checksum offload with
IPv6, to allow independent toggling (where possible). Add CSUM_*_IPV6
flags for UDP/TCP over IPv6, and reserve further for SCTP, and IPv6
fragmentation. Define CSUM_DELAY_DATA_IPV6 as we do for legacy IP and
add an alias for CSUM_DATA_VALID_IPV6.
This pretty much brings IPv6 handling in line with IPv4.
TSO is still handled in a different way and not via if_hwassist.
Update ifconfig to allow (un)setting of the new capability flags.
Update loopback to announce the new capabilities and if_hwassist flags.
Individual driver updates will have to follow, as will SCTP.
Reported by: gallatin, dim, ..
Reviewed by: gallatin (glanced at?)
MFC after: 3 days
X-MFC with: r235961,235959,235958
Add code to handle pre-checked TCP checksums as indicated by mbuf
flags to save the entire computation for validation if not needed.
In the IPv6 TCP output path only compute the pseudo-header checksum,
set the checksum offset in the mbuf field along the appropriate flag
as done in IPv4.
In tcp_respond() just initialize the IPv6 payload length to 0 as
ip6_output() will properly set it.
Sponsored by: The FreeBSD Foundation
Sponsored by: iXsystems
Reviewed by: gnn (as part of the whole)
MFC After: 3 days
hz >> 1000 and thus getting outside the timestamp clock frequenceny of
1ms < x < 1s per tick as mandated by RFC1323, leading to connection
resets on idle connections.
Always use a granularity of 1ms using getmicrouptime() making all but
relevant callouts independent of hz.
Use getmicrouptime(), not getmicrotime() as the latter may make a jump
possibly breaking TCP nfsroot mounts having our timestamps move forward
for more than 24.8 days in a second without having been idle for that
long.
PR: kern/61404
Reviewed by: jhb, mav, rrs
Discussed with: silby, lstewart
Sponsored by: Sandvine Incorporated (originally in 2011)
MFC after: 6 weeks
TCP_KEEPCNT, that allow to control initial timeout, idle time, idle
re-send interval and idle send count on a per-socket basis.
Reviewed by: andre, bz, lstewart
The SYSCTL_NODE macro defines a list that stores all child-elements of
that node. If there's no SYSCTL_DECL macro anywhere else, there's no
reason why it shouldn't be static.
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.
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.
- 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.
Add some comments at #endifs given more nestedness. To make the compiler
happy, some default initializations were added in accordance with the style
on the files.
Reviewed by: gnn
Sponsored by: The FreeBSD Foundation
Sponsored by: iXsystems
MFC after: 4 days
As long as this is a costy function, even when compiled in (along with
the option TCP_SIGNATURE), it can be disabled via the
net.inet.tcp.signature_verify_input sysctl.
Sponsored by: Sandvine Incorporated
Reviewed by: emaste, bz
MFC after: 2 weeks
DPCPU_DEFINE and VNET_DEFINE macros, as these cause problems for various
people working on the affected files. A better long-term solution is
still being considered. This reversal may give some modules empty
set_pcpu or set_vnet sections, but these are harmless.
Changes reverted:
------------------------------------------------------------------------
r215318 | dim | 2010-11-14 21:40:55 +0100 (Sun, 14 Nov 2010) | 4 lines
Instead of unconditionally emitting .globl's for the __start_set_xxx and
__stop_set_xxx symbols, only emit them when the set_vnet or set_pcpu
sections are actually defined.
------------------------------------------------------------------------
r215317 | dim | 2010-11-14 21:38:11 +0100 (Sun, 14 Nov 2010) | 3 lines
Apply the STATIC_VNET_DEFINE and STATIC_DPCPU_DEFINE macros throughout
the tree.
------------------------------------------------------------------------
r215316 | dim | 2010-11-14 21:23:02 +0100 (Sun, 14 Nov 2010) | 2 lines
Add macros to define static instances of VNET_DEFINE and DPCPU_DEFINE.
report when a new socket couldn't be created because one of
in_pcbinshash(), in6_pcbconnect() or in_pcbconnect() failed.
Logging is conditional on net.inet.tcp.log_debug being enabled.
MFC after: 1 week
was limited to one segment under the faulty assumption of a retransmit.
Due to this the opportunity to initialize the increased congestion window
according to RFC3390 was missed.
Support for RFC3465 introduced in r187289 uncovered the bug as the ACK
to SYN/ACK no longer caused snd_cwnd increase by MSS (actually, this
increase shouldn't happen as it's explicitly forbidden by RFC3390, but
it's another issue). Snd_cwnd remains really small (1*MSS + 1) and this
causes really bad interaction with delayed acks on other side.
The variable name sc_rxmits is a bit misleading as it counts all transmits,
not just retransmits.
Submitted by: Maxim Dounin <mdounin-at-mdounin-dot-ru>
MFC after: 10 days
"Whitspace" churn after the VIMAGE/VNET whirls.
Remove the need for some "init" functions within the network
stack, like pim6_init(), icmp_init() or significantly shorten
others like ip6_init() and nd6_init(), using static initialization
again where possible and formerly missed.
Move (most) variables back to the place they used to be before the
container structs and VIMAGE_GLOABLS (before r185088) and try to
reduce the diff to stable/7 and earlier as good as possible,
to help out-of-tree consumers to update from 6.x or 7.x to 8 or 9.
This also removes some header file pollution for putatively
static global variables.
Revert VIMAGE specific changes in ipfilter::ip_auth.c, that are
no longer needed.
Reviewed by: jhb
Discussed with: rwatson
Sponsored by: The FreeBSD Foundation
Sponsored by: CK Software GmbH
MFC after: 6 days
vnet.h, we now use jails (rather than vimages) as the abstraction
for virtualization management, and what remained was specific to
virtual network stacks. Minor cleanups are done in the process,
and comments updated to reflect these changes.
Reviewed by: bz
Approved by: re (vimage blanket)
(DPCPU), as suggested by Peter Wemm, and implement a new per-virtual
network stack memory allocator. Modify vnet to use the allocator
instead of monolithic global container structures (vinet, ...). This
change solves many binary compatibility problems associated with
VIMAGE, and restores ELF symbols for virtualized global variables.
Each virtualized global variable exists as a "reference copy", and also
once per virtual network stack. Virtualized global variables are
tagged at compile-time, placing the in a special linker set, which is
loaded into a contiguous region of kernel memory. Virtualized global
variables in the base kernel are linked as normal, but those in modules
are copied and relocated to a reserved portion of the kernel's vnet
region with the help of a the kernel linker.
Virtualized global variables exist in per-vnet memory set up when the
network stack instance is created, and are initialized statically from
the reference copy. Run-time access occurs via an accessor macro, which
converts from the current vnet and requested symbol to a per-vnet
address. When "options VIMAGE" is not compiled into the kernel, normal
global ELF symbols will be used instead and indirection is avoided.
This change restores static initialization for network stack global
variables, restores support for non-global symbols and types, eliminates
the need for many subsystem constructors, eliminates large per-subsystem
structures that caused many binary compatibility issues both for
monitoring applications (netstat) and kernel modules, removes the
per-function INIT_VNET_*() macros throughout the stack, eliminates the
need for vnet_symmap ksym(2) munging, and eliminates duplicate
definitions of virtualized globals under VIMAGE_GLOBALS.
Bump __FreeBSD_version and update UPDATING.
Portions submitted by: bz
Reviewed by: bz, zec
Discussed with: gnn, jamie, jeff, jhb, julian, sam
Suggested by: peter
Approved by: re (kensmith)
the TCP syncache. This returns struct tcpopt to being private within the TCP
implementation, thus allowing it to be modified without ABI concerns.
The patch breaks the ABI. Bump __FreeBSD_version to 800103 accordingly. The cxgb
driver is the only TOE consumer affected by this change, and needs to be
recompiled along with the kernel.
Suggested by: rwatson
Reviewed by: rwatson, kmacy
Approved by: re (kensmith), kensmith (mentor temporarily unavailable)
Vnet modules and protocol domains may now register destructor
functions to clean up and release per-module state. The destructor
mechanisms can be triggered by invoking "vimage -d", or a future
equivalent command which will be provided via the new jail framework.
While this patch introduces numerous placeholder destructor functions,
many of those are currently incomplete, thus leaking memory or (even
worse) failing to stop all running timers. Many of such issues are
already known and will be incrementaly fixed over the next weeks in
smaller incremental commits.
Apart from introducing new fields in structs ifnet, domain, protosw
and vnet_net, which requires the kernel and modules to be rebuilt, this
change should have no impact on nooptions VIMAGE builds, since vnet
destructors can only be called in VIMAGE kernels. Moreover,
destructor functions should be in general compiled in only in
options VIMAGE builds, except for kernel modules which can be safely
kldunloaded at run time.
Bump __FreeBSD_version to 800097.
Reviewed by: bz, julian
Approved by: rwatson, kib (re), julian (mentor)
and used in a large number of files, but also because an increasing number
of incorrect uses of MAC calls were sneaking in due to copy-and-paste of
MAC-aware code without the associated opt_mac.h include.
Discussed with: pjd
count of the number of registered policies.
Rather than unconditionally locking sockets before passing them into MAC,
lock them in the MAC entry points only if mac_policy_count is non-zero.
This avoids locking overhead for a number of socket system calls when no
policies are registered, eliminating measurable overhead for the MAC
Framework for the socket subsystem when there are no active policies.
Possibly socket locks should be acquired by policies if they are required
for socket labels, which would further avoid locking overhead when there
are policies but they don't require labeling of sockets, or possibly
don't even implement socket controls.
Obtained from: TrustedBSD Project
active network stack instance. Turning on options VIMAGE at compile
time yields the following changes relative to default kernel build:
1) V_ accessor macros for virtualized variables resolve to structure
fields via base pointers, instead of being resolved as fields in global
structs or plain global variables. As an example, V_ifnet becomes:
options VIMAGE: ((struct vnet_net *) vnet_net)->_ifnet
default build: vnet_net_0._ifnet
options VIMAGE_GLOBALS: ifnet
2) INIT_VNET_* macros will declare and set up base pointers to be used
by V_ accessor macros, instead of resolving to whitespace:
INIT_VNET_NET(ifp->if_vnet); becomes
struct vnet_net *vnet_net = (ifp->if_vnet)->mod_data[VNET_MOD_NET];
3) Memory for vnet modules registered via vnet_mod_register() is now
allocated at run time in sys/kern/kern_vimage.c, instead of per vnet
module structs being declared as globals. If required, vnet modules
can now request the framework to provide them with allocated bzeroed
memory by filling in the vmi_size field in their vmi_modinfo structures.
4) structs socket, ifnet, inpcbinfo, tcpcb and syncache_head are
extended to hold a pointer to the parent vnet. options VIMAGE builds
will fill in those fields as required.
5) curvnet is introduced as a new global variable in options VIMAGE
builds, always pointing to the default and only struct vnet.
6) struct sysctl_oid has been extended with additional two fields to
store major and minor virtualization module identifiers, oid_v_subs and
oid_v_mod. SYSCTL_V_* family of macros will fill in those fields
accordingly, and store the offset in the appropriate vnet container
struct in oid_arg1.
In sysctl handlers dealing with virtualized sysctls, the
SYSCTL_RESOLVE_V_ARG1() macro will compute the address of the target
variable and make it available in arg1 variable for further processing.
Unused fields in structs vnet_inet, vnet_inet6 and vnet_ipfw have
been deleted.
Reviewed by: bz, rwatson
Approved by: julian (mentor)
TCPSTAT_INC(), rather than directly manipulating the fields across the
kernel. This will make it easier to change the implementation of
these statistics, such as using per-CPU versions of the data structures.
MFC after: 3 days
had been the only flag with random usage patterns.
Switch inc_flags to be used as a real bit field by using
INC_ISIPV6 with bitops to check for the 'isipv6' condition.
While here fix a place or two where in case of v4 inc_flags
were not properly initialized before.[1]
Found by: rwatson during review [1]
Discussed with: rwatson
Reviewed by: rwatson
MFC after: 4 weeks
