This gives significant performance improvements when many raw sockets used.
Benchmarks of mpd handeling 1000 simultaneous PPTP connections show up to 50%
performance boost. With higher number of connections benefit becomes even
bigger. PopTop snd others should also get some benefits.
- removing 'const' qualifier from an input parameter to conform to the type
required by rw_assert();
- using in_addr->s_addr to retrive 32 bits address value.
Observed by: tinderbox
information from rip_input() to rip_append(). Instead, pass the source
address for an IP datagram to rip_append() using a stack-allocated
sockaddr_in, similar to udp_input() and udp_append().
Prior to the move to rwlocks for inpcbinfo, this was not a problem, as
use of the global was synchronized using the ripcbinfo mutex, but with
read-locking there is the potential for a race during concurrent
receive.
This problem is not present in the IPv6 raw IP socket code, which
already used a stack variable for the address.
Spotted by: mav
MFC after: 1 week (before inpcbinfo rwlock changes)
completes the move to a fully parallel UDP transmit path by using
global read, rather than write, locking of inpcbinfo in further
semi-connected cases:
- Add macros to allow try-locking of inpcb and inpcbinfo.
- Always acquire an incpcb read lock in udp_output(), which stablizes the
local inpcb address and port bindings in order to determine what further
locking is required:
- If the inpcb is currently not bound (at all) and are implicitly
connecting, we require inpcbinfo and inpcb write locks, so drop the
read lock and re-acquire.
- If the inpcb is bound for at least one of the port or address, but an
explicit source or destination is requested, trylock the inpcbinfo
lock, and if that fails, drop the inpcb lock, lock the global lock,
and relock the inpcb lock.
- Otherwise, no further locking is required (common case).
- Update comments.
In practice, this means that the vast majority of consumers of UDP sockets
will not acquire any exclusive locks at the socket or UDP levels of the
network stack. This leads to a marked performance improvement in several
important workloads, including BIND, nsd, and memcached over UDP, as well
as significant improvements in pps microbenchmarks.
The plan is to MFC all of the rwlock changes to RELENG_7 once they have
settled for a weeks in the tree.
Tested by: ps, kris (older revision), bde
MFC after: 3 weeks
udp_output() so that argument validation occurs before jail processing.
Add additional comments explaining what's going on when we process
addresses and binding during udp_output().
MFC after: 3 weeks
2) Adds some __UserSpace__ on some of the common defines that
the user space code needs
3) Fixes a bug when we send up data to a user that failed. We
need to a) trim off the data chunk headers, if present, and
b) make sure the frag bit is communicated properly for the
msgs coming off the stream queues... i.e. we see if some
of the msg has been taken.
Obtained from: jeli contributed the VIMAGE changes on this pass Thanks Julain!
inpcb. When directly invoking udp_notify() from udp_ctlinput(), acquire
only a read lock; we may still see write locks in udp_notify() as the
in_pcbnotifyall() routine is shared with TCP and always uses a write lock
on the inpcb being notified.
MFC after: 1 month
some code paths, global or inpcb write locks are required, but for other
code paths, read locks or no locking at all are sufficient for the data
structures.
MFC after: 1 month
source or a specific destination address is requested as part of a send
on a UDP socket, read lock the inpcb rather than write lock it. This
will allow fully parallel transmit down to the IP layer when sending
simultaneously from multiple threads on a connected UDP socket.
Parallel transmit for more complex cases, such as when sendto(2) is
invoked with an address and there's already a local binding, will
follow.
MFC after: 1 month
possible to exhaust and garble stack with a packet that contains a couple
of hundreds nested encapsulation levels.
Submitted by: Ming Fu <fming@borderware.com>
Reviewed by: rwatson
PR: kern/85320
dispatched without Giant, and add NETISR_FORCEQUEUE, which allows specific
netisr handlers to always be dispatched via a queue (deferred). Mark the
usb and if_ppp netisr handlers as NETISR_FORCEQUEUE, and explicitly
acquire Giant in those handlers.
Previously, any netisr handler not marked NETISR_MPSAFE would necessarily
run deferred and with Giant acquired. This change removes Giant
scaffolding from the netisr infrastructure, but NETISR_FORCEQUEUE allows
non-MPSAFE handlers to continue to force deferred dispatch so as to avoid
lock order reversals between their acqusition of Giant and any calling
context.
It is likely we will be able to remove NETISR_FORCEQUEUE once
IFF_NEEDSGIANT is removed, as non-MPSAFE usb and if_ppp drivers will no
longer be supported.
Reviewed by: bz
MFC after: 1 month
X-MFC note: We can't remove NETISR_MPSAFE from stable/7 for KPI reasons,
but the rest can go back.
generating an RTM_MISS for every IP packet forwarded making user space
routing daemons unhappy.
PR: kern/123621, kern/124540, kern/122338
Reported by: Paul <paul gtcomm.net>, Mike Tancsa <mike sentex.net> on net@
Tested by: Paul and Mike
Reviewed by: andre
MFC after: 3 days
datagram-only protocols, such as UDP. This version removes use of
sblock(), which is not required due to an inability to interlace data
improperly with datagrams, as well as avoiding some of the larger loops
and state management that don't apply on datagram sockets.
This is experimental code, so hook it up only for UDPv4 for testing; if
there are problems we may need to revise it or turn it off by default,
but it offers *significant* performance improvements for threaded UDP
applications such as BIND9, nsd, and memcached using UDP.
Tested by: kris, ps
rather than write locking: while we need to maintain a valid reference
to the inpcb and fix its state, no protocol layer state is modified
during an IPv4 UDP receive -- there are only changes at the socket
layer, which is separately protected by socket locking.
While parallel concurrent receive on a single UDP socket is currently
relatively unusual, introducing read locking in the transmit path,
allowing concurrent receive and transmit, will significantly improve
performance for loads such as BIND, memcached, etc.
MFC after: 2 months
Tested by: gnn, kris, ps
in_ifaddrhashtbl in in_ifinit because error handler in in_control removes
entries only for AF_INET addresses. If in_ifinit is called for the cloned
inteface that has just been created its address family is not AF_INET and
therefor LIST_REMOVE is not called for respective LIST_INSERT_HEAD and
freed entries remain in in_ifaddrhashtbl and lead to memory corruption.
PR: kern/124384
This is needed for correct behavior when packets are lost or reordered.
PR: kern/123950
Reviewed by: andre@, silby@
Reported by: Yahoo!, Wang Jin
MFC after: 1 week
- only one functino to destroy an SCTP stack sctp_finish()
- Make it so this function also arranges for any threads
created by the image to do a kthread_exit()
- Vimage prep - these are major restructures to move
all global variables to be accessed via a macro or two.
The variables all go into a single structure.
- Asconf address addition tweaks (add_or_del Interfaces)
- Fix rwnd calcualtion to be more conservative.
- Support SACK_IMMEDIATE flag to skip delayed sack
by demand of peer.
- Comment updates in the sack mapping calculations
- Invarients panic added.
- Pre-support for UDP tunneling (we can do this on
MAC but will need added support from UDP to
get a "pipe" of UDP packets in.
- clear trace buffer sysctl added when local tracing on.
Note the majority of this huge patch is all the vimage prep stuff :-)
template, use an M_TEMP malloc(9) allocation rather than an mbuf
with mtod(9) and dtom(9). This eliminates the last use of
dtom(9) in TCP.
MFC after: 3 weeks
to reduce performance degradation under heavy outgoing scan/flood.
Scalability is now much more important then several kilobytes of RAM.
Remove unneded TCP-specific expiration handeling. Before this connected
TCP sessions could never expire. Now connected TCP sessions will expire
after 24hours of inactivity.
Simplify HouseKeeping() to avoid several mul/div-s per packet. Taking into
account increased LINK_TABLE_OUT_SIZE, precision is still much more then
required.
- to increase performance do not reallocate mbuf when possible,
- to support up to 16K packets (was 2K max) use mbuf cluster of proper size.
This change depends on recent ng_nat and ip_fw_nat changes.
This fixes packet fragmentation handeling.
Pass really available buffer size to libalias instead of MCLBYTES constant.
MCLBYTES constant were used with believe that m_megapullup() always moves
date into a fresh cluster that sometimes may become not so.
some cases, add explicit inpcb locking rather than relying on the global
lock, as we dereference inp_socket, but also allowing us to drop the
global lock more quickly.
MFC after: 1 week
monitoring UDP connections using sysctls. In some cases, add
previously missing locking of inpcbs, as inp_socket is followed,
which also allows us to drop global locks more quickly.
MFC after: 1 week
ip6_savecontrol in preparation for udp_append() to no longer
need an WLOCK as we will no longer be modifying socket options.
Requested by: rwatson
Reviewed by: gnn
MFC after: 10 days
routines for those modules, rather than in the raw socket code. This
each privilege check to occur in exactly once place and avoids
duplicate checks across layers.
MFC after: 3 weeks
Sponsored by: nCircle Network Security, Inc.
Change so that we save off a type field for display and
NULL inp just for good measure.
- sctp_output.c - Fix it so in sending to the loopback we use the
src address of the inbound INIT. We don't want
to do this for non local addresses since otherwise
we might be ingressed filtered so we need to use
the best src address and list the address sent to.
Obtained from: time bug - Neil Wilson
MFC after: 1 week
- Adds some prepwork (Not all yet) for vimage in particular
support the delete the sctppcbinfo.xx structs. There is
still a leak in here if it were to be called plus we stil
need the regrouping (From Me and Michael Tuexen)
- Adds support for UDP tunneling. For BSD there is no
socket yet setup so its disabled, but major argument
changes are in here to emcompass the passing of the port
number (zero when you don't have a udp tunnel, the default
for BSD). Will add some hooks in UDP here shortly (discussed
with Robert) that will allow easy tunneling. (Mainly from
Peter Lei and Michael Tuexen with some BSD work from me :-D)
- Some ease for windows, evidently leave is reserved by their
compile move label leave: -> out:
MFC after: 1 week
- Bug in CA that does not get us incrementing the PBA properly which
made us more conservative.
- comment updated in sctp_input.c
- memsets added before we log
- added arg to hmac id's
MFC after: 2 weeks
This particular implementation is designed to be fully backwards compatible
and to be MFC-able to 7.x (and 6.x)
Currently the only protocol that can make use of the multiple tables is IPv4
Similar functionality exists in OpenBSD and Linux.
From my notes:
-----
One thing where FreeBSD has been falling behind, and which by chance I
have some time to work on is "policy based routing", which allows
different
packet streams to be routed by more than just the destination address.
Constraints:
------------
I want to make some form of this available in the 6.x tree
(and by extension 7.x) , but FreeBSD in general needs it so I might as
well do it in -current and back port the portions I need.
One of the ways that this can be done is to have the ability to
instantiate multiple kernel routing tables (which I will now
refer to as "Forwarding Information Bases" or "FIBs" for political
correctness reasons). Which FIB a particular packet uses to make
the next hop decision can be decided by a number of mechanisms.
The policies these mechanisms implement are the "Policies" referred
to in "Policy based routing".
One of the constraints I have if I try to back port this work to
6.x is that it must be implemented as a EXTENSION to the existing
ABIs in 6.x so that third party applications do not need to be
recompiled in timespan of the branch.
This first version will not have some of the bells and whistles that
will come with later versions. It will, for example, be limited to 16
tables in the first commit.
Implementation method, Compatible version. (part 1)
-------------------------------
For this reason I have implemented a "sufficient subset" of a
multiple routing table solution in Perforce, and back-ported it
to 6.x. (also in Perforce though not always caught up with what I
have done in -current/P4). The subset allows a number of FIBs
to be defined at compile time (8 is sufficient for my purposes in 6.x)
and implements the changes needed to allow IPV4 to use them. I have not
done the changes for ipv6 simply because I do not need it, and I do not
have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it.
Other protocol families are left untouched and should there be
users with proprietary protocol families, they should continue to work
and be oblivious to the existence of the extra FIBs.
To understand how this is done, one must know that the current FIB
code starts everything off with a single dimensional array of
pointers to FIB head structures (One per protocol family), each of
which in turn points to the trie of routes available to that family.
The basic change in the ABI compatible version of the change is to
extent that array to be a 2 dimensional array, so that
instead of protocol family X looking at rt_tables[X] for the
table it needs, it looks at rt_tables[Y][X] when for all
protocol families except ipv4 Y is always 0.
Code that is unaware of the change always just sees the first row
of the table, which of course looks just like the one dimensional
array that existed before.
The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign()
are all maintained, but refer only to the first row of the array,
so that existing callers in proprietary protocols can continue to
do the "right thing".
Some new entry points are added, for the exclusive use of ipv4 code
called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(),
which have an extra argument which refers the code to the correct row.
In addition, there are some new entry points (currently called
rtalloc_fib() and friends) that check the Address family being
looked up and call either rtalloc() (and friends) if the protocol
is not IPv4 forcing the action to row 0 or to the appropriate row
if it IS IPv4 (and that info is available). These are for calling
from code that is not specific to any particular protocol. The way
these are implemented would change in the non ABI preserving code
to be added later.
One feature of the first version of the code is that for ipv4,
the interface routes show up automatically on all the FIBs, so
that no matter what FIB you select you always have the basic
direct attached hosts available to you. (rtinit() does this
automatically).
You CAN delete an interface route from one FIB should you want
to but by default it's there. ARP information is also available
in each FIB. It's assumed that the same machine would have the
same MAC address, regardless of which FIB you are using to get
to it.
This brings us as to how the correct FIB is selected for an outgoing
IPV4 packet.
Firstly, all packets have a FIB associated with them. if nothing
has been done to change it, it will be FIB 0. The FIB is changed
in the following ways.
Packets fall into one of a number of classes.
1/ locally generated packets, coming from a socket/PCB.
Such packets select a FIB from a number associated with the
socket/PCB. This in turn is inherited from the process,
but can be changed by a socket option. The process in turn
inherits it on fork. I have written a utility call setfib
that acts a bit like nice..
setfib -3 ping target.example.com # will use fib 3 for ping.
It is an obvious extension to make it a property of a jail
but I have not done so. It can be achieved by combining the setfib and
jail commands.
2/ packets received on an interface for forwarding.
By default these packets would use table 0,
(or possibly a number settable in a sysctl(not yet)).
but prior to routing the firewall can inspect them (see below).
(possibly in the future you may be able to associate a FIB
with packets received on an interface.. An ifconfig arg, but not yet.)
3/ packets inspected by a packet classifier, which can arbitrarily
associate a fib with it on a packet by packet basis.
A fib assigned to a packet by a packet classifier
(such as ipfw) would over-ride a fib associated by
a more default source. (such as cases 1 or 2).
4/ a tcp listen socket associated with a fib will generate
accept sockets that are associated with that same fib.
5/ Packets generated in response to some other packet (e.g. reset
or icmp packets). These should use the FIB associated with the
packet being reponded to.
6/ Packets generated during encapsulation.
gif, tun and other tunnel interfaces will encapsulate using the FIB
that was in effect withthe proces that set up the tunnel.
thus setfib 1 ifconfig gif0 [tunnel instructions]
will set the fib for the tunnel to use to be fib 1.
Routing messages would be associated with their
process, and thus select one FIB or another.
messages from the kernel would be associated with the fib they
refer to and would only be received by a routing socket associated
with that fib. (not yet implemented)
In addition Netstat has been edited to be able to cope with the
fact that the array is now 2 dimensional. (It looks in system
memory using libkvm (!)). Old versions of netstat see only the first FIB.
In addition two sysctls are added to give:
a) the number of FIBs compiled in (active)
b) the default FIB of the calling process.
Early testing experience:
-------------------------
Basically our (IronPort's) appliance does this functionality already
using ipfw fwd but that method has some drawbacks.
For example,
It can't fully simulate a routing table because it can't influence the
socket's choice of local address when a connect() is done.
Testing during the generating of these changes has been
remarkably smooth so far. Multiple tables have co-existed
with no notable side effects, and packets have been routes
accordingly.
ipfw has grown 2 new keywords:
setfib N ip from anay to any
count ip from any to any fib N
In pf there seems to be a requirement to be able to give symbolic names to the
fibs but I do not have that capacity. I am not sure if it is required.
SCTP has interestingly enough built in support for this, called VRFs
in Cisco parlance. it will be interesting to see how that handles it
when it suddenly actually does something.
Where to next:
--------------------
After committing the ABI compatible version and MFCing it, I'd
like to proceed in a forward direction in -current. this will
result in some roto-tilling in the routing code.
Firstly: the current code's idea of having a separate tree per
protocol family, all of the same format, and pointed to by the
1 dimensional array is a bit silly. Especially when one considers that
there is code that makes assumptions about every protocol having the
same internal structures there. Some protocols don't WANT that
sort of structure. (for example the whole idea of a netmask is foreign
to appletalk). This needs to be made opaque to the external code.
My suggested first change is to add routing method pointers to the
'domain' structure, along with information pointing the data.
instead of having an array of pointers to uniform structures,
there would be an array pointing to the 'domain' structures
for each protocol address domain (protocol family),
and the methods this reached would be called. The methods would have
an argument that gives FIB number, but the protocol would be free
to ignore it.
When the ABI can be changed it raises the possibilty of the
addition of a fib entry into the "struct route". Currently,
the structure contains the sockaddr of the desination, and the resulting
fib entry. To make this work fully, one could add a fib number
so that given an address and a fib, one can find the third element, the
fib entry.
Interaction with the ARP layer/ LL layer would need to be
revisited as well. Qing Li has been working on this already.
This work was sponsored by Ironport Systems/Cisco
Reviewed by: several including rwatson, bz and mlair (parts each)
Obtained from: Ironport systems/Cisco
syncache that has an invalid SEQ instead of only doing it when we suceed
in mallocing space for the log message.
MFC after: 1 week
Reviewed by: sam, bz
receiving or transmitting.
With IPv6 raw sockets, read lock rather than write lock the inpcb when
receiving. Unfortunately, IPv6 source address selection appears to
require a write lock on the inpcb for the time being.
MFC after: 3 months
A lot of testing has shown that the problem people were seeing was due
to invalid padding after the end of option list option, which was corrected
in tcp_output.c rev. 1.146.
Thanks to: anders@, s3raphi, Matt Reimer
Thanks to: Doug Hardie and Randy Rose, John Mayer, Susan Guzzardi
Special thanks to: dwhite@ and BitGravity
Discussed with: silby
MFC after: 1 day
when reading credential data from sockets.
Teach pf to unlock the pcbinfo more quickly once it has acquired an
inpcb lock, as the inpcb lock is sufficient to protect the reference.
Assert locks, rather than read locks or write locks, on inpcbs in
subroutines--this is necessary as the inpcb may be passed down with a
write lock from the protocol, or may be passed down with a read lock
from the firewall lookup routine, and either is sufficient.
MFC after: 3 months
move most offload functionality from NIC to TOE
factor out all socket and inpcb direct access
factor out access to locking in incpb, pcbinfo, and sockbuf
explicitly select write locking for all use of the inpcb mutex.
Update some pcbinfo lock assertions to assert locked rather than
write-locked, although in practice almost all uses of the pcbinfo
rwlock main exclusive, and all instances of inpcb lock acquisition
are exclusive.
This change should introduce (ideally) little functional change.
However, it lays the groundwork for significantly increased
parallelism in the TCP/IP code.
MFC after: 3 months
Tested by: kris (superset of committered patch)
done by understandable macros.
Fix the bug that prevented the system from responding on interfaces with
link local addresses assigned.
PR: 120958
Submitted by: James Snow <snow at teardrop.org>
MFC after: 2 weeks
(ECMP) for both IPv4 and IPv6. Previously, multipath route insertion
is disallowed. For example,
route add -net 192.103.54.0/24 10.9.44.1
route add -net 192.103.54.0/24 10.9.44.2
The second route insertion will trigger an error message of
"add net 192.103.54.0/24: gateway 10.2.5.2: route already in table"
Multiple default routes can also be inserted. Here is the netstat
output:
default 10.2.5.1 UGS 0 3074 bge0 =>
default 10.2.5.2 UGS 0 0 bge0
When multipath routes exist, the "route delete" command requires
a specific gateway to be specified or else an error message would
be displayed. For example,
route delete default
would fail and trigger the following error message:
"route: writing to routing socket: No such process"
"delete net default: not in table"
On the other hand,
route delete default 10.2.5.2
would be successful: "delete net default: gateway 10.2.5.2"
One does not have to specify a gateway if there is only a single
route for a particular destination.
I need to perform more testings on address aliases and multiple
interfaces that have the same IP prefixes. This patch as it
stands today is not yet ready for prime time. Therefore, the ECMP
code fragments are fully guarded by the RADIX_MPATH macro.
Include the "options RADIX_MPATH" in the kernel configuration
to enable this feature.
Reviewed by: robert, sam, gnn, julian, kmacy
ICMP unreach, frag needed. Up to now we only looked at the
interface MTU. Make sure to only use the minimum of the two.
In case IPSEC is compiled in, loop the mtu through ip_ipsec_mtu()
to avoid any further conditional maths.
Without this, PMTU was broken in those cases when there was a
route with a lower MTU than the MTU of the outgoing interface.
PR: kern/122338
Tested by: Mark Cammidge mark peralex.com
Reviewed by: silence on net@
MFC after: 2 weeks
was changed in rev. 1.161 of tcp_var.h. All option now test for sufficient
space in TCP header before getting added.
Reported by: Mark Atkinson <atkin901-at-yahoo.com>
Tested by: Mark Atkinson <atkin901-at-yahoo.com>
MFC after: 1 week
Removed dead code that assumed that M_TRYWAIT can return NULL; it's not true
since the advent of MBUMA.
Reviewed by: arch
There are ongoing disputes as to whether we want to switch to directly using
UMA flags M_WAITOK/M_NOWAIT for mbuf(9) allocation.
In that case return an continue processing the packet without IPsec.
PR: 121384
MFC after: 5 days
Reported by: Cyrus Rahman (crahman gmail.com)
Tested by: Cyrus Rahman (crahman gmail.com) [slightly older version]
the NOPs used are 0x01.
While we could simply pad with EOLs (which are 0x00), rather use an
explicit 0x00 constant there to not confuse poeple with 'EOL padding'.
Put in a comment saying just that.
Problem discussed on: src-committers with andre, silby, dwhite as
follow up to the rev. 1.161 commit of tcp_var.h.
MFC after: 11 days
restrict the utilization of direct pointers to the content of
ip packet. These modifications are functionally nop()s thus
can be merged with no side effects.
IPPORT_EPHEMERALFIRST and IPPORT_EPHEMERALLAST with values
10000 and 65535 respectively.
The rationale behind is that it makes the attacker's life more
difficult if he/she wants to guess the ephemeral port range and
also lowers the probability of a port colision (described in
draft-ietf-tsvwg-port-randomization-01.txt).
While there, remove code duplication in in_pcbbind_setup().
Submitted by: Fernando Gont <fernando at gont.com.ar>
Approved by: njl (mentor)
Reviewed by: silby, bms
Discussed on: freebsd-net
- Move the assigment of the socket down before we first need it.
No need to do it at the beginning and then drop out the function
by one of the returns before using it 100 lines further down.
- Use t_maxopd which was assigned the "tcp_mssdflt" for the corrrect
AF already instead of another #ifdef ? : #endif block doing the same.
- Remove an unneeded (duplicate) assignment of mss to t_maxseg just before
we possibly change mss and re-do the assignment without using t_maxseg
in between.
Reviewed by: silby
No objections: net@ (silence)
MFC after: 5 days
the limit in bytes) hard coded into both the kernel and userland.
Make both these limits a sysctl, so it is easy to change the limit.
If the userland part of ipfw finds that the sysctls don't exist,
it will just fall back to the traditional limits.
(100 packets is quite a small limit these days. If you want to test
TCP at 100Mbps, 100 packets can only accommodate a DBP of 12ms.)
Note these sysctls in the man page and warn against increasing them
without thinking first.
MFC after: 3 weeks
the same order that FreeBSD 6 and before did. Doug
White and the other bloodhounds at ISC discovered that
while FreeBSD 7's ordering of options was more efficient,
it caused some cable modem routers to ignore the
SYN-ACKs ordered in this fashion.
The placement of sackOK after the timestamp option seems
to be the critical difference:
FreeBSD 6:
<mss 1460,nop,wscale 1,nop,nop,timestamp 3512155768 0,sackOK,eol>
FreeBSD 7.0:
<mss 1460,nop,wscale 3,sackOK,timestamp 1370692577 0>
FreeBSD 7.0 + this change:
<mss 1460,nop,wscale 3,nop,nop,timestamp 7371813 0,sackOK,eol>
MFC after: 1 week
obtained from OpenBSD with an algorithm suggested
by Amit Klein. The OpenBSD algorithm has a few
flaws; see Amit's paper for more information.
For a description of how this algorithm works,
please see the comments within the code.
Note that this commit does not yet enable random IP ID
generation by default. There are still some concerns
that doing so will adversely affect performance.
Reviewed by: rwatson
MFC After: 2 weeks
ipsec*_set_policy and do the privilege check only if needed.
Try to assimilate both ip*_ctloutput code blocks calling ipsec*_set_policy.
Reviewed by: rwatson
read socket buffers in shutdown() and close():
- Call socantrcvmore() before sblock() to dislodge any threads that
might be sleeping (potentially indefinitely) while holding sblock(),
such as a thread blocked in recv().
- Flag the sblock() call as non-interruptible so that a signal
delivered to the thread calling sorflush() doesn't cause sblock() to
fail. The sblock() is required to ensure that all other socket
consumer threads have, in fact, left, and do not enter, the socket
buffer until we're done flushin it.
To implement the latter, change the 'flags' argument to sblock() to
accept two flags, SBL_WAIT and SBL_NOINTR, rather than one M_WAITOK
flag. When SBL_NOINTR is set, it forces a non-interruptible sx
acquisition, regardless of the setting of the disposition of SB_NOINTR
on the socket buffer; without this change it would be possible for
another thread to clear SB_NOINTR between when the socket buffer mutex
is released and sblock() is invoked.
Reviewed by: bz, kmacy
Reported by: Jos Backus <jos at catnook dot com>
exposing them to all consumers of ip_fw.h. These structures are
used in both ipfw(8) and ipfw(4), but not part of the user<->kernel
interface for other applications to use, rather, shared
implementation.
MFC after: 3 days
Reported by: Paul Vixie <paul at vix dot com>
Introduce a new privilege allowing to set certain IP header options
(hop-by-hop, routing headers).
Leave a few comments to be addressed later.
Reviewed by: rwatson (older version, before addressing his comments)
while in principle a good idea, opened us up to a race inherrent to
the syncache's direct insertion of incoming TCP connections into the
"completed connection" listen queue, as it transpires that the socket
is inserted before the inpcb is fully filled in by syncache_expand().
The bug manifested with the occasional returning of 0.0.0.0:0 in the
address returned by the accept() system call, which occurred if accept
managed to execute tcp_usr_accept() before syncache_expand() had copied
the endpoint addresses into inpcb connection state.
Re-add tcbinfo locking around the address copyout, which has the effect
of delaying the copy until syncache_expand() has finished running, as
it is run while the tcbinfo lock is held. This is undesirable in that
it increases contention on tcbinfo further, but a more significant
change will be required to how the syncache inserts new sockets in
order to fix this and keep more granular locking here. In particular,
either more state needs to be passed into sonewconn() so that
pru_attach() can fill in the fields *before* the socket is inserted, or
the socket needs to be inserted in the incomplete connection queue
until it is actually ready to be used.
Reported by: glebius (and kris)
Tested by: glebius
drop the lock and then re-acquire it, revalidating TCP connection state
assumptions when we do so. This avoids a potential lock order reversal
(and potential deadlock, although none have been reported) due to the
inpcb lock being held over a page fault.
MFC after: 1 week
PR: 102752
Reviewed by: bz
Reported by: Václav Haisman <v dot haisman at sh dot cvut dot cz>
of two compares against 0. The negative effect of cache flushing
is probably more than the gain by not doing the two compares (the
value is almost certainly in register or at worst, cache).
Note that the uses of m_freem() are in error cases and m_freem()
handles NULL anyhow. So fast-path really isn't changed much at all.
free the MAC label on the inpcb before freeing the inpcb.
MFC after: 3 days
Submitted by: tanyong <tanyong at ercist dot iscas dot ac dot cn>,
zhouzhouyi
When system ticks are positive, for entries in the cache
bucket, syncache_timer() ran on every tick (doing nothing
useful) instead of the supposed 3, 6, 12, and 24 seconds
later (when it's time to retransmit SYN,ACK).
When ticks are negative, syncache_timer() was scheduled
for the too far future (up to ~25 days on systems with
HZ=1000), no SYN,ACK retransmits were attempted at all,
and syncache entries added in that period that correspond
to non-established connections stay there forever.
Only HEAD and RELENG_7 are affected.
Reviewed by: silby, kmacy (earlier version)
Submitted by: Maxim Dounin, ru
- Rename output routines tcp_gen_* -> tcp_output_*.
- Rename notification routines that turn in to no-ops in the absence of TOE
from tcp_gen_* -> tcp_offload_*.
- Fix some minor comment nits.
- Add a /* FALLTHROUGH */
Reviewed by: Sam Leffler, Robert Watson, and Mike Silbersack
- rename tcp_ofld.[ch] to tcp_offload.[ch]
- document usage and locking conventions of the functions in the
toe_usrreqs function vector
- document tcpcb, inpcb, and socket fields used by toe
- widen the listen interface into 2 functions
- rename DISABLE_TCP_OFFLOAD to TCP_OFFLOAD_DISABLE
- shrink conditional compilation to reduce the likelihood of bitrot
- replace sc->sc_toepcb checks in tcp_syncache.c with TOEPCB_ISSET
- make neccessary changes to release offload resources when a syncache
entry is removed before connection establishment
- disable checks for offloaded connection where insufficient information
is available
Reviewed by: silby
header, then don't try to pullup anything, because there is no next
header if we hit IPPROTO_NONE. Set ulp to a non-NULL value so the
search for an upper layer header terinates.
This is based on Pekka's diagnosis, but I chose a simpler fix.
PR: 115261
Submitted by: Pekka Savola <pekkas@netcore.fi>
Reviewed by: mlaier
MFC after: 2 weeks
the sent_queue. Sometimes I wonder why any code
ever works :-)
- Fix the pad of the last mbuf routine, It was working improperly
on non-4 byte aligned chunks which could cause memory overruns.
MFC after: 1 week
- Missing lock when sending data and moving it to the
outqueue.
- If a mbuf alloc fails during moving to outqueue the
reassembly of the old mbuf chain was incorrect.
- some_taken becomes a counter in sctputil.c instead of a set to 1.
- Fix a panic to be only under invarients and have a proper recovery.
- msg_flags needed to be set.to the value collected not or'd.
MFC after: 1 week
test incorrect.
- Fix the initial buf calculation to be more friendly, calc is the same
but we use different variable to make it easier amongst the different
code versions.
MFC after: 1 week
sending, once the locks are all unlocked to
do the copy's in, its possible that other
events could then raise the number of bytes
outstanding pushing it so not all the message
would fit. This would then cause us to send
only part of the message. This fix makes it
so we keep a "reserved" amount that can be
kept in mind when making calculations to send.
- rcv msg args with a NULL/NULL for to/tolen will return an error incorrectly
for the 1-2-1 model.
- We were not doing 0 len return correctly and not setting cantrcv more
correctly. Previouly we "fixed" this area by taking out the socantrcv
since we then could not get the data out. The correct rix is to still
flag the socket but alow a by-pass route to continue to read until
all data is consumed.
MFC after: 1 week
Before this fix, FreeBSD would negotiate SACK on outgoing
connections, but would always fail to negotiate it on incoming
connections.
Discovered by: James Healy and Lawrence Stewart
Submitted by: James Healy and Lawrence Stewart
MFC after: 3 days
1. A packet comes in that is to be forwarded
2. The destination of the packet is rewritten by some firewall code
3. The next link's MTU is too small
4. The packet has the DF bit set
Then the current code is such that instead of setting the next
link's MTU in the ICMP error, ip_next_mtu() is called and a guess
is sent as to which MTU is supposed to be tried next. This is because
in this case ip_forward() is called with srcrt set to 1. In that
case the ia pointer remains NULL but it is needed to get the MTU
of the interface the packet is to be sent out from.
Thus, we always set ia to the outgoing interface.
MFC after: 2 weeks
in the tcp header. With relevant parts of the tcp header changing after
the 'signature' was computed, the signature becomes invalid.
Reviewed by: tools/regression/netinet/tcpconnect
MFC after: 3 days
Tested by: Nick Hilliard (see net@)
the inpcb when there's an inpcb without associated timewait state, and
not unlocking when the inpcb has been freed. This avoids a kernel panic
when tcpdrop(8) is run on a socket in the TIMEWAIT state.
MFC after: 3 days
Reported by: Rako <rako29 at gmail dot com>
The lookup hurts a bit for connections but had been there anyway
if IPSEC was compiled in. So moving the lookup up a bit gives us
TSO support at not extra cost.
PR: kern/115586
Tested by: gallatin
Discussed with: kmacy
MFC after: 2 months
timestamps in the initial SYN packet actually use them in the rest of the
connection. Unfortunately, during the 7.0 testing cycle users have already
found network devices that violate this constraint.
RFC 1323 states 'and may send a TSopt in other segments' rather than
'and MUST send', so we must allow it.
Discovered by: Rob Zietlow
Tracked down by: Kip Macy
PR: bin/118005
If it is set to zero value (default) dummynet module will try to emulate
real link as close as possible (bandwidth & latency): packet will not leave
pipe faster than it should be on real link with given bandwidth.
(This is original behaviour of dummynet which was altered in previous commit)
If it is set to non-zero value only bandwidth is enforced: packet's latency
can be lower comparing to real link with given bandwidth.
- Document recently introduced dummynet(4) sysctl variables.
Requested by: luigi, julian
MFC after: 3 month
2) Alter packet flow inside dummynet: allow certain packets to bypass
dummynet scheduler. Benefits are:
- lower latency: if packet flow does not exceed pipe bandwidth, packets
will not be (up to tick) delayed (due to dummynet's scheduler granularity).
- lower overhead: if packet avoids dummynet scheduler it shouldn't reenter ip
stack later. Such packets can be fastforwarded.
- recursion (which can lead to kernel stack exhaution) eliminated. This fix
long existed panic, which can be triggered this way:
kldload dummynet
sysctl net.inet.ip.fw.one_pass=0
ipfw pipe 1 config bw 0
for i in `jot 30`; do ipfw add 1 pipe 1 icmp from any to any; done
ping -c 1 localhost
3) Three new sysctl nodes are added:
net.inet.ip.dummynet.io_pkt - packets passed to dummynet
net.inet.ip.dummynet.io_pkt_fast - packets avoided dummynet scheduler
net.inet.ip.dummynet.io_pkt_drop - packets dropped by dummynet
P.S. Above comments are true only for layer 3 packets. Layer 2 packet flow
is not changed yet.
MFC after: 3 month
- Select a tag gains ability to optionally save new tags
off in the timewait system.
- When looking up associations do not give back a stcb that
is in the about-to-be-freed state, and instead continue
looking for other candiates.
- New function to query to see if value is in time-wait.
- Timewait had a time comparison error that caused very
few vtags to actually stay in time-wait.
- When setting tags in time-wait, we now use the time
requested NOT a fixed constant value.
- sstat now gets the proper associd when we do the query.
- When we process an association, we expect the tag chosen
(if we have one from a cookie) to be in time-wait. Before
we would NOT allow the assoc up by checking if its good.
In theory this should have caused almost all assoc not
to come up except for the time-comparison bug above (this
bug was hidden by the time comparison bug :-D).
- Don't save tags for nonce values in the time-wait cache
since these are used only during cookie collisions and do
not matter if they are unique or not.
MFC after: 1 week
Framework by moving from mac_mbuf_create_netlayer() to more specific
entry points for specific network services:
- mac_netinet_firewall_reply() to be used when replying to in-bound TCP
segments in pf and ipfw (etc).
- Rename mac_netinet_icmp_reply() to mac_netinet_icmp_replyinplace() and
add mac_netinet_icmp_reply(), reflecting that in some cases we overwrite
a label in place, but in others we apply the label to a new mbuf.
Obtained from: TrustedBSD Project
in the TrustedBSD MAC Framework:
- Add mac_atalk.c and add explicit entry point mac_netatalk_aarp_send()
for AARP packet labeling, rather than using a generic link layer
entry point.
- Add mac_inet6.c and add explicit entry point mac_netinet6_nd6_send()
for ND6 packet labeling, rather than using a generic link layer entry
point.
- Add expliict entry point mac_netinet_arp_send() for ARP packet
labeling, and mac_netinet_igmp_send() for IGMP packet labeling,
rather than using a generic link layer entry point.
- Remove previous genering link layer entry point,
mac_mbuf_create_linklayer() as it is no longer used.
- Add implementations of new entry points to various policies, largely
by replicating the existing link layer entry point for them; remove
old link layer entry point implementation.
- Make MAC_IFNET_LOCK(), MAC_IFNET_UNLOCK(), and mac_ifnet_mtx global
to the MAC Framework rather than static to mac_net.c as it is now
needed outside of mac_net.c.
Obtained from: TrustedBSD Project
we move towards netinet as a pseudo-object for the MAC Framework.
Rename 'mac_create_mbuf_linklayer' to 'mac_mbuf_create_linklayer' to
reflect general object-first ordering preference.
Sponsored by: SPARTA (original patches against Mac OS X)
Obtained from: TrustedBSD Project, Apple Computer
from Mac OS X Leopard--rationalize naming for entry points to
the following general forms:
mac_<object>_<method/action>
mac_<object>_check_<method/action>
The previous naming scheme was inconsistent and mostly
reversed from the new scheme. Also, make object types more
consistent and remove spaces from object types that contain
multiple parts ("posix_sem" -> "posixsem") to make mechanical
parsing easier. Introduce a new "netinet" object type for
certain IPv4/IPv6-related methods. Also simplify, slightly,
some entry point names.
All MAC policy modules will need to be recompiled, and modules
not updates as part of this commit will need to be modified to
conform to the new KPI.
Sponsored by: SPARTA (original patches against Mac OS X)
Obtained from: TrustedBSD Project, Apple Computer
to kproc_xxx as they actually make whole processes.
Thos makes way for us to add REAL kthread_create() and friends
that actually make theads. it turns out that most of these
calls actually end up being moved back to the thread version
when it's added. but we need to make this cosmetic change first.
I'd LOVE to do this rename in 7.0 so that we can eventually MFC the
new kthread_xxx() calls.
us to scale up to sb_max, aka kern.ipc.maxsockbuf.
We do this because there are broken firewalls that will corrupt the window
scale option, leading to the other endpoint believing that our advertised
window is unscaled. At scale factors larger than 5 the unscaled window will
drop below 1500 bytes, leading to serious problems when traversing these
broken firewalls.
With the default maxsockbuf of 256K, a scale factor of 3 will be chosen by
this algorithm. Those who choose a larger maxsockbuf should watch out
for the compatiblity problems mentioned above.
Reviewed by: andre
- fix a bug during cookie collision that prevented an
association from coming up in a specific restart case.
- Fix it so the shutdown-pending flag gets removed (this is
more for correctness then needed) when we enter shutdown-sent
or shutdown-ack-sent states.
- Fix a bug that caused the receiver to sometimes NOT send
a SACK when a duplicate TSN arrived. Without this fix
it was possible for the association to fall down if the
- Deleted primary destination is also stored when SCTP_MOBILITY_BASE.
(Previously, it is stored when only SCTP_MOBILITY_FASTHANDOFF)
- Fix a locking issue where we might call send_initiate_ack() and
incorrectly state the lock held/not held. Also fix it so that
when we release the lock the inp cannot be deleted on us.
- Add the debug option that can cause the stack to panic instead
of aborting an assoc. This does not and should never show up
in options but is useful for debugging unexpected aborts.
- Add cumack_log sent to track sending cumack information for
the debug case where we are running a special log per assoc.
- Added extra () aroudn sctp_sbspace macro to avoid compile warnings.
MFC after: 1 week
TCP: [X.X.X.X]:X to [X.X.X.X]:X tcpflags 0x18<PUSH,ACK>; tcp_do_segment: FIN_WAIT_2: Received data after socket was closed, sending RST and removing tcpcb
So that it also includes how many bytes of data were received. It now looks
like this:
TCP: [X.X.X.X]:X to [X.X.X.X]:X tcpflags 0x18<PUSH,ACK>; tcp_do_segment: FIN_WAIT_2: Received X bytes of data after socket was closed, sending RST and removing tcpcb
Approved by: re (gnn)
problems with the syncache, it produces a lot of console noise and has led
to quite a few false positive bug reports. It can be selectively
re-enabled when debugging specific problems by frobbing the same sysctl.
Discussed with: silby
Approved by: re (gnn)
retransmittion by handover event (fast mobility code)
- Fixed problem of mobility code which is caused by remaining
parameters in the deleted primary destination.
- Add a missing lock. When a peer sends an INIT, and while we
are processing it to send an INIT-ACK the socket is closed,
we did not hold a lock to keep the socket from going away.
Add protection for this case.
- Fix so that arwnd is alway uses the minimal rwnd if the user
has set the socket buffer smaller. Found this when the test
org decided to see what happens when you set in a rwnd of 10
bytes (which is not allowed per RFC .. 4k is minimum).
- Fixes so a cookie-echo ootb will NOT cause an abort to
be sent. This was happening in a MPI collision case.
- Examined all panics and unless there was no recovery, moved
any that were not already to INVARANTS.
Approved by: re@freebsd.org (gnn)
- Reintegrate the ANSI C function declaration change
from tcp_timer.c rev 1.92
- Reorganize the tcpcb structure so that it has a single
pointer to the "tcp_timer" structure which contains all
of the tcp timer callouts. This change means that when
the single tcp timer change is reintegrated, tcpcb will
not change in size, and therefore the ABI between
netstat and the kernel will not change.
Neither of these changes should have any functional
impact.
Reviewed by: bmah, rrs
Approved by: re (bmah)
route and once they are done with it, call rtfree(). rtfree() should
only be used when we are certain we hold the last reference to the
route. This bug results in console messages like the following:
rtfree: 0xc40f7000 has 1 refs
This patch switches the rtfree() to use RTFREE_LOCKED() instead,
which should handle the reference counting on the route better.
Approved by: re@ (gnn)
Reviewed by: bms
Reported by: many via net@ and current@
Tested by: many
last interface should own the address, but the current code
fumbles the handoff. This fixes that.
- move address related debugs to PCB4 and add additional ones to
help in debugging address problems.
Approved by: re@freebsd.org (K Smith)
also involves macro changes to have a RLOCK and a WLOCK
and placing the correct version within the code.
- The INP-INFO lock is changed to a rwlock.
- When sctp_shutdown() is called on Mac OS X, the socket lock is held.
So call sctp_chunk_output with SCTP_SO_LOCKED and
not SCTP_SO_NOT_LOCKED.
- Add SCTP_IPI_ADDR_[RW]LOCK and SCTP_IPI_ADDR_[RW]UNLOCK for Mac OS X.
- u_int64_t -> uint64_t
- add missing addr unlock for error return path
Approved by: re@freebsd.org (K Smith)
- Fix panic from mutex unlock on freed lock when ASCONF-ACK
aborts an assoc
- Fix panic from addr lock recursion when ASCONFs are queued
in the front states
- ASCONFs "queued" in the front states should really be
bundled after the COOKIE-ACK, not in front of it
- Fix issue with addresses deleted in the front states from
being sent with ASCONF(DELETE)-- replaced
sctp_asconf_queue_add_sa() with delete specific function
- Comment change in sctp.h the drafts are now RFC's
Approved by: re@freebsd.org (B Mah)
incorrect and should be OFF letting IP fragment
large cookie-echos.
- Rename sysctl variable logging to log_level.
- Fix description of sysctl variable stats.
- Add sysctl variable log to make sctp_log readable via sysctl
mechanism (this is by compile switch and targets non KTR platforms or
when someone wants to do performance wise tracing).
- Removed debug code
Approved by: re@freebsd.org (B Mah)
stream (using EEOR mode). Changed to EINVAL (in sctp_output.c)
- Static analysis comments added
- fix in mobility code to return a value (static analysis found).
- sctp6_notify function made visible instead of
static (this is needed for Panda).
Approved by: re@freebsd.org (B Mah)
code comes from.
- Fix a LOR on Mac OS X: Do not hold an stcb lock when
calling soisconnected for a socket which has the
SS_INCOMP bit set on so_state.
- fix a comment to be non c++ style.
Approved by: re@freebsd.org (B Mah)
jumping to dropunlock to avoid a panic. While here move the calls to
ipsec4_in_reject() and ipsec6_in_reject() so they are after we obtain
the lock on inp.
Original patch to avoid panic: pjd
Review of locking adjustments: gnn, sam
Approved by: re (rwatson)
- Resort includes a bit.
- Correct typos and wording problems in comments.
- Rename udpcksum to udp_cksum to be consistent with other UDP-related
configuration variables.
- Remove indirection of udp_notify through local notify variable in
udp_ctlinput(), which is presumably due to copying and pasting from TCP,
where multiple notify routines exist.
Approved by: re (kensmith)
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)
additional flags to many function calls. The flags only
get used in BSD when we compile with lock testing. These
flags allow apple to escape the "giant" lock it holds on
the socket and have more fine-grained locking in the NKE.
It also allows us to test (with witness) the locking used
by apple via a compile switch (manually applied).
Approved by: re@freebsd.org(B Mah)
TCP timers as a single timer, but retain the API changes necessary to
reintroduce this change. This will back out the source of at least two
reported problems: lock leaks in certain timer edge cases, and TCP timers
continuing to fire after a connection has closed (a bug previously fixed and
then reintroduced with the timer rewrite).
In a follow-up commit, some minor restylings and comment changes performed
after the TCP timer rewrite will be reapplied, and a further change to allow
the TCP timer rewrite to be added back without disturbing the ABI. The new
design is believed to be a good thing, but the outstanding issues are
leading to significant stability/correctness problems that are holding
up 7.0.
This patch was generated by silby, but is being committed by proxy due to
poor network connectivity for silby this week.
Approved by: re (kensmith)
Submitted by: silby
Tested by: rwatson, kris
Problems reported by: peter, kris, others
import. The PF mbuf-tagging support routines changed to link the
allocated tags into the provided mbuf themselves, so the left-over
m_tag_prepend() was trying to add a bogus (usually NULL) tag.
Reviewed by: mlaier
Approved by: re
the last message on the send stream was "null" but still
there, a state we allow, we could get hung and not clean
it up and wait for the shutdown guard timer to clear the
association without a graceful close. Fix this so that
that we properly clean up.
- Added support for Multiple ASCONF per new RFC. We only
(so far) accept input of these and cannot yet generate
a multi-asconf.
- Sysctl'd support for experimental Fast Handover feature. Always
disabled unless sysctl or socket option changes to enable.
- Error case in add-ip where the peer supports AUTH and ADD-IP
but does NOT require AUTH of ASCONF/ASCONF-ACK. We need to
ABORT in this case.
- According to the Kyoto summit of socket api developers
(Solaris, Linux, BSD). We need to have:
o non-eeor mode messages be atomic - Fixed
o Allow implicit setup of an assoc in 1-2-1 model if
using the sctp_**() send calls - Fixed
o Get rid of HAVE_XXX declarations - Done
o add a sctp_pr_policy in hole in sndrcvinfo structure - Done
o add a PR_SCTP_POLICY_VALID type flag - yet to-do in a future patch!
- Optimize sctp6 calls to reuse code in sctp_usrreq. Also optimize
when we close sending out the data and disabling Nagle.
- Change key concatenation order to match the auth RFC
- When sending OOTB shutdown_complete always do csum.
- Don't send PKT-DROP to a PKT-DROP
- For abort chunks just always checksums same for
shutdown-complete.
- inpcb_free front state had a bug where in queue
data could wedge an assoc. We need to just abandon
ones in front states (free_assoc).
- If a peer sends us a 64k abort, we would try to
assemble a response packet which may be larger than
64k. This then would be dropped by IP. Instead make
a "minimum" size for us 64k-2k (we want at least
2k for our initack). If we receive such an init
discard it early without all the processing.
- When we peel off we must increment the tcb ref count
to keep it from being freed from underneath us.
- handling fwd-tsn had bugs that caused memory overwrites
when given faulty data, fixed so can't happen and we
also stop at the first bad stream no.
- Fixed so comm-up generates the adaption indication.
- peeloff did not get the hmac params copied.
- fix it so we lock the addr list when doing src-addr selection
(in future we need to use a multi-reader/one writer lock here)
- During lowlevel output, we could end up with a _l_addr set
to null if the iterator is calling the output routine. This
means we would possibly crash when we gather the MTU info.
Fix so we only do the gather where we have a src address
cached.
- we need to be sure to set abort flag on conn state when
we receive an abort.
- peeloff could leak a socket. Moved code so the close will
find the socket if the peeloff fails (uipc_syscalls.c)
Approved by: re@freebsd.org(Ken Smith)
pack a set number correctly.
Submitted by: oleg
o Plug a memory leak.
Submitted by: oleg and Andrey V. Elsukov
Approved by: re (kensmith)
MFC after: 1 week
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)
- Remove unneeded WLOCK/UNLOCK of inp for getting TCB lock.
- Fix panic that may occur when freeing an assoc that has partial
delivery in progress (may dereference null socket pointer when
queuing partial delivery aborted notification)
- Some spacing and comment fixes.
- Fix address add handling to clear cached routes and source addresses
when peer acks the add in case the routing table changes.
Approved by: re@freebsd.org (Bruce Mah)
projected_offset against isn_offset to account for
wrap around.
Reviewed by: gnn, kmacy, silby
Submitted by: yusheng.huang@bluecoat.com
Approved by: re
MFC: 3 days
the use of divert sockets to dead locks. A number of LORs have been reported
between divert and a number of other network subsystems including: IPSEC, Pfil,
multicast, ipfw and others. Other dead locks could occur because of recursive
entry into the IP stack. This change should take care of most if not all of
these issues.
A summary of the changes follow:
- We disallow multicast operations on divert sockets. It really doesn't make
semantic sense to allow this, since typically you would set multicast
parameters on multicast end points.
NOTE: As a part of this change, we actually dis-allow multicast options on
any socket that IS a divert socket OR IS NOT a SOCK_RAW or SOCK_DGRAM family
- We check to see if there are any socket options that have been specified on
the socket, and if there was (which is very un-common and also probably
doesnt make sense to support) we duplicate the mbuf carrying the options.
- We then drop the INP/INFO locks over the call to ip_output(). It should be
noted that since we no longer support multicast operations on divert sockets
and we have duplicated any socket options, we no longer need the reference
to the pcb to be coherent.
- Finally, we replaced the call to ip_input() to use netisr queuing. This
should remove the recursive entry into the IP stack from divert.
By dropping the locks over the call to ip_output() we eliminate all the lock
ordering issues above. By switching over to netisr on the inbound path,
we can no longer recursively enter the ip_input() code via divert.
I have tested this change by using the following command:
ipfwpcap -r 8000 - | tcpdump -r - -nn -v
This should exercise the input and re-injection (outbound) path, which is
very similar to the work load performed by natd(8). Additionally, I have
run some ospf daemons which have a heavy reliance on raw sockets and
multicast.
Approved by: re@ (kensmith)
MFC after: 1 month
LOR: 163
LOR: 181
LOR: 202
LOR: 203
Discussed with: julian, andre et al (on freebsd-net)
In collaboration with: bms [1], rwatson [2]
[1] bms helped out with the multicast decisions
[2] rwatson submitted the original netisr patches and came up with some
of the original ideas on how to combat this issue.
for bakeoff.. using the next sequential ones)
- In cookie processing 1-2-1, we did not increment the stcb
refcnt before releasing the tcb lock. We need to do this
to keep the tcb from being freed by a abort or ?? unlikely
but worth doing. Also get rid of unneed INP_WLOCK.
- extra receive info included the rcvinfo which killed the
padding/alignment. We now redefine all the fields properly
so they both align properly both to 128 bytes.
- A peeled off socket would not close without an error due to
its misguided idea that sctp_disconnect() was not supported
on it. This fixes it so it goes through the proper path.
- When an assoc was being deleted after abort (via a timer) a
small race condition exists where we might take a packet for
the old assoc (since we are waiting for a cleanup timer). This
state especially happens in mac. We now add a state in the asoc
so these can properly handle the packet as OOTB.
Approved by: re@freebsd.org(Ken Smith)
previously conditionally acquired Giant based on debug.mpsafenet. As that
has now been removed, they are no longer required. Removing them
significantly simplifies error-handling in the socket layer, eliminated
quite a bit of unwinding of locking in error cases.
While here clean up the now unneeded opt_net.h, which previously was used
for the NET_WITH_GIANT kernel option. Clean up some related gotos for
consistency.
Reviewed by: bz, csjp
Tested by: kris
Approved by: re (kensmith)
Also rename the related functions in a similar way.
There are no functional changes.
For a packet coming in with IPsec tunnel mode, the default is
to only call into the firewall with the "outer" IP header and
payload.
With this option turned on, in addition to the "outer" parts,
the "inner" IP header and payload are passed to the
firewall too when going through ip_input() the second time.
The option was never only related to a gif(4) tunnel within
an IPsec tunnel and thus the name was very misleading.
Discussed at: BSDCan 2007
Best new name suggested by: rwatson
Reviewed by: rwatson
Approved by: re (bmah)
be in ticks "for algorithm stability" when originally committed, it turns
out that it has a significant impact in timing out connections. When we
changed HZ from 100 to 1000, this had a big effect on reducing the time
before dropping connections.
To demonstrate, boot with kern.hz=100. ssh to a box on local ethernet
and establish a reliable round-trip-time (ie: type a few commands).
Then unplug the ethernet and press a key. Time how long it takes to
drop the connection.
The old behavior (with hz=100) caused the connection to typically drop
between 90 and 110 seconds of getting no response.
Now boot with kern.hz=1000 (default). The same test causes the ssh session
to drop after just 9-10 seconds. This is a big deal on a wifi connection.
With kern.hz=1000, change sysctl net.inet.tcp.rexmit_min from 3 to 30.
Note how it behaves the same as when HZ was 100. Also, note that when
booting with hz=100, net.inet.tcp.rexmit_min *used* to be 30.
This commit changes TCPTV_MIN to be scaled with hz. rexmit_min should
always be about 30. If you set hz to Really Slow(TM), there is a safety
feature to prevent a value of 0 being used.
This may be revised in the future, but for the time being, it restores the
old, pre-hz=1000 behavior, which is significantly less annoying.
As a workaround, to avoid rebooting or rebuilding a kernel, you can run
"sysctl net.inet.tcp.rexmit_min=30" and add "net.inet.tcp.rexmit_min=30"
to /etc/sysctl.conf. This is safe to run from 6.0 onwards.
Approved by: re (rwatson)
Reviewed by: andre, silby
<netinet/tcp_fsm.h> is included into any compilation unit that needs
tcpstates[]. Also remove incorrect extern declarations and TCPDEBUG
conditionals. This allows kernels both with and without TCPDEBUG to
build, and unbreaks the tinderbox.
Approved by: re (rwatson)
sys.net.inet.tcp.log_debug = 1
It defaults to enabled for the moment and is to be turned off for
the next release like other diagnostics from development branches.
It is important to note that sysctl sys.net.inet.tcp.log_in_vain
uses the same logging function as log_debug. Enabling of the former
also causes the latter to engage, but not vice versa.
Use consistent terminology in tcp log messages:
"ignored" means a segment contains invalid flags/information and
is dropped without changing state or issuing a reply.
"rejected" means a segments contains invalid flags/information but
is causing a reply (usually RST) and may cause a state change.
Approved by: re (rwatson)
SYNCACHE_TIMEOUT to new function syncache_timeout().
o Fix inverted timeout callout engagement logic to actually
enable the timer for the bucket row. Before SYN|ACK was
not retransmitted.
o Simplify SYN|ACK retransmit timeout backoff calculation.
o Improve logging of retransmit and timeout events.
o Reset timeout when duplicate SYN arrives.
o Add comments.
o Rearrange SYN cookie statistics counting.
Bug found by: silby
Submitted by: silby (different version)
Approved by: re (rwatson)
syncache_rst().
o Fix tests for flag combinations of RST and SYN, ACK, FIN. Before
a RST for a connection in syncache did not properly free the entry.
o Add more detailed logging.
Approved by: re (rwatson)
- 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
- remove duplicate #include <sys/priv.h> that is not under
#ifdef FreeBSD version to allow compile on 6.1
- static analysis changes per the cisco SA tool including:
o some SA_IGNORE comments
o some checks for NULL before unlock.
o type corrections int -> size_t
- Fix it so sctp_alloc_asoc takes a thread/proc argument. Without this
we pass a NULL in to bind on implicit assoc setup and crash :-(
Approved by: re@freebsd.org(Ken Smith)
UDPv4 features to UDPv6:
- Add MAC checks on delivery and MAC labeling on transmit.
- Check for (and reject) datagrams with destination port 0.
- For multicast delivery, check the source port only if the socket being
considered as a destination has been connected.
- Implement UDP blackholing based on net.inet.udp.blackhole.
- Add a new ICMPv6 unreachable reply rate limiting category for failed
delivery attempts and implement rate limiting for UDPv6 (submitted by
bz).
Approved by: re (kensmith)
Reviewed by: bz
- use proper tick gathering macro instead of ticks directly.
- Placed reasonable boundaries on sets that a user can do
that are converted to ticks from ms.
- Fix CMT_PF to always check to be sure CMT is on.
- Fix ticks use of CMT_PF.
- put back code to allow asconfs to be queued while INITs are in flight
and before the assoc is established.
- During window probes, an ack'd packet might be left with the window
probe mark on it causing it to be retransmitted. Change so that
the flight decrease macro clears the window_probe mark.
- Additional logging flight size/reading and ASOC LOG. This
is only enabled if you manually insert things into opt_sctp.h
since its a set of debug code only.
- Found an interesting SMP race in the way data was appended which
could cause a reader to lose a part of a message, had to
reorder when we marked the message was complete to after
the data was appended.
- bug in ADD-IP for the subset bound socket case when the peer has only
one address
- fix ASCONF implicit success/error handling case
- proper support of jails in Freebsd 6>
- copy out the timeval for the 64 bit sparc world on cookie-echo
alignment error crashes without this).
Approved by: re(Ken Smith)
- CMT_PF states added (w/sysctl to turn the PF version on)
- sctp_input.c had a missing incr of cookie case when the
auth was bad. This meant a free was called without an
increment to refcnt, added increment like rest of code.
- There was a case, unlikely, when the scope of the destination
changed (this is a TSNH case). In that case, it would not free
the alloc'ed asoc (in sctp_input.c).
- When listed addresses found a colliding cookie/Init, then
the collided upon tcb was not unlocked in sctp_pcb.c
- Add error checking on arguments of sctp_sendx(3) to prevent it from
referencing a NULL pointer.
- Fix an error return of sctp_sendx(3), it was returing
ENOMEM not -1.
- Get assoc id was changed to use the sanctified socket api
method for getting a assoc id (PEER_ADDR_INFO instead of
PEER_ADDR_PARAMS).
- Fix it so a peeled off socket will get a proper error return
if it trys to send to a different address then it is connected to.
- Fix so that select_a_stream can avoid an endless loop that
could hang a caller.
- time_entered (state set time) was not being set in all cases
to the time we went established.
Approved by: re(ken smith)
- Move udp_sendspace and udp_recvspace global variables and associated
sysctls to the top of the file where most other such things are present.
- Rename static variable 'blackhole' to 'udp_blackhole' and unstaticize
so that we can add blackhole support for UDPv6 using the same MIB
variable.
- Move udp_append() above udp_input() to match the function order in
udp6_usrreq.c.
Approved by: re (kensmith)
With the in_mcast.c code, if an interface for an IPv4 multicast join was
not specified, and a route did not exist for the specified group in the
unicast forwarding tables, the join would be rejected with the error
EADDRNOTAVAIL.
This change restores the old behaviour whereby if no interface is specified,
and no route exists for the group destination, the IPv4 address list is
walked to find a non-loopback, multicast-capable interface to satisfy
the join request.
This should resolve problems with starting multicast services during
system boot or when a default forwarding entry does not exist.
Approved by: re (rwatson)
free to be consistent with other error handling, and release socket buffer
lock before freeing mbufs and statistics updates rather than after.
Approved by: re (kensmith)
call the sctp_free_remote_address() function.
- Assure that when we allocate a chunk the whoTo is NULL,
also when we free it and place it into the cache we NULL
it (that way the consolidation code will always work).
- Fix a small race, when a empty data holder is left on the stream
out queue, and both sides do a shutdown, the empty data holder
would prevent us from sending a SHUTDOWN-ACK and at the same time we
never would cleanup the empty holder (since nothing was ever in queue).
We now add a utility function that a) cleans up empty holders and
b) properly determines if there are still pending data chunks on
the stream out wheel.
Approved by: re@freebsd.org (Ken Smith)
This commit includes only the kernel files, the rest of the files
will follow in a second commit.
Reviewed by: bz
Approved by: re
Supported by: Secure Computing
- Fix fwd-tsn to use proper accessor so it does not overrun mbufs
- Fix stream reset error reporting to actually work (it has always been
broken if the peer rejects a stream reset)
- Some 64 bit friendly changes
Approved by: re(bmah@freebsd.org)
