- eliminate the fast/slow timeout lists for TCP and instead use a
callout entry for each timer.
- increase the TCP timer granularity to HZ
- implement "bad retransmit" recovery, as presented in
"On Estimating End-to-End Network Path Properties", by Allman and Paxson.
Submitted by: jlemon, wollmann
drop any segment arriving at a closed port.
tcp.blackhole=1 - only drop SYN without RST
tcp.blackhole=2 - drop everything without RST
tcp.blackhole=0 - always send RST - default behaviour
This confuses nmap -sF or -sX or -sN quite badly.
sysctl knobs.
With these knobs on, refused connection attempts are dropped
without sending a RST, or Port unreachable in the UDP case.
In the TCP case, sending of RST is inhibited iff the incoming
segment was a SYN.
Docs and rc.conf settings to follow.
This was missed in the 4.4-Lite2 merge.
Noticed by: Mohan Parthasarathy <Mohan.Parthasarathy@eng.Sun.COM> and
jayanth@loc201.tandem.com (vijayaraghavan_jayanth)
on the tcp-impl mailing list.
Add ICMP_BANDLIM option and 'net.inet.icmp.icmplim' sysctl. If option
is specified in kernel config, icmplim defaults to 100 pps. Setting it
to 0 will disable the feature. This feature limits ICMP error responses
for packets sent to bad tcp or udp ports, which does a lot to help the
machine handle network D.O.S. attacks.
The kernel will report packet rates that exceed the limit at a rate of
one kernel printf per second. There is one issue in regards to the
'tail end' of an attack... the kernel will not output the last report
until some unrelated and valid icmp error packet is return at some
point after the attack is over. This is a minor reporting issue only.
Any packet that can be matched by a ipfw rule can be redirected
transparently to another port or machine. Redirection to another port
mostly makes sense with tcp, where a session can be set up
between a proxy and an unsuspecting client. Redirection to another machine
requires that the other machine also be expecting to receive the forwarded
packets, as their headers will not have been modified.
/sbin/ipfw must be recompiled!!!
Reviewed by: Peter Wemm <peter@freebsd.org>
Submitted by: Chrisy Luke <chrisy@flix.net>
connect. This check was added as part of the defense against the "land"
attack, to prevent attacks which guess the ISS from going into ESTABLISHED.
The "src == dst" check will still prevent the single-homed case of the
"land" attack, and guessing ISS's should be hard anyway.
Submitted by: David Borman <dab@bsdi.com>
a hashed port list. In the new scheme, in_pcblookup() goes away and is
replaced by a new routine, in_pcblookup_local() for doing the local port
check. Note that this implementation is space inefficient in that the PCB
struct is now too large to fit into 128 bytes. I might deal with this in the
future by using the new zone allocator, but I wanted these changes to be
extensively tested in their current form first.
Also:
1) Fixed off-by-one errors in the port lookup loops in in_pcbbind().
2) Got rid of some unneeded rehashing. Adding a new routine, in_pcbinshash()
to do the initialial hash insertion.
3) Renamed in_pcblookuphash() to in_pcblookup_hash() for easier readability.
4) Added a new routine, in_pcbremlists() to remove the PCB from the various
hash lists.
5) Added/deleted comments where appropriate.
6) Removed unnecessary splnet() locking. In general, the PCB functions should
be called at splnet()...there are unfortunately a few exceptions, however.
7) Reorganized a few structs for better cache line behavior.
8) Killed my TCP_ACK_HACK kludge. It may come back in a different form in
the future, however.
These changes have been tested on wcarchive for more than a month. In tests
done here, connection establishment overhead is reduced by more than 50
times, thus getting rid of one of the major networking scalability problems.
Still to do: make tcp_fastimo/tcp_slowtimo scale well for systems with a
large number of connections. tcp_fastimo is easy; tcp_slowtimo is difficult.
WARNING: Anything that knows about inpcb and tcpcb structs will have to be
recompiled; at the very least, this includes netstat(1).
rev 1.66. This fix contains both belt and suspenders.
Belt: ignore packets where src == dst and srcport == dstport in TCPS_LISTEN.
These packets can only legitimately occur when connecting a socket to itself,
which doesn't go through TCPS_LISTEN (it goes CLOSED->SYN_SENT->SYN_RCVD->
ESTABLISHED). This prevents the "standard" "land" attack, although doesn't
prevent the multi-homed variation.
Suspenders: send a RST in response to a SYN/ACK in SYN_RECEIVED state.
The only packets we should get in SYN_RECEIVED are
1. A retransmitted SYN, or
2. An ack of our SYN/ACK.
The "land" attack depends on us accepting our own SYN/ACK as an ACK;
in SYN_RECEIVED state; this should prevent all "land" attacks.
We also move up the sequence number check for the ACK in SYN_RECEIVED.
This neither helps nor hurts with respect to the "land" attack, but
puts more of the validation checking in one spot.
PR: kern/5103
RST's being ignored, keeping a connection around until it times out, and
thus has the opposite effect of what was intended (which is to make the
system more robust to DoS attacks).
socket addresses in mbufs. (Socket buffers are the one exception.) A number
of kernel APIs needed to get fixed in order to make this happen. Also,
fix three protocol families which kept PCBs in mbufs to not malloc them
instead. Delete some old compatibility cruft while we're at it, and add
some new routines in the in_cksum family.
be dropped when it has an unusual traffic pattern. For full details
as well as a test case that demonstrates the failure, see the
referenced PR.
Under certain circumstances involving the persist state, it is
possible for the receive side's tp->rcv_nxt to advance beyond its
tp->rcv_adv. This causes (tp->rcv_adv - tp->rcv_nxt) to become
negative. However, in the code affected by this fix, that difference
was interpreted as an unsigned number by max(). Since it was
negative, it was taken as a huge unsigned number. The effect was
to cause the receiver to believe that its receive window had negative
size, thereby rejecting all received segments including ACKs. As
the test case shows, this led to fruitless retransmissions and
eventually to a dropped connection. Even connections using the
loopback interface could be dropped. The fix substitutes the signed
imax() for the unsigned max() function.
PR: closes kern/3998
Reviewed by: davidg, fenner, wollman
This commit includes the following changes:
1) Old-style (pr_usrreq()) protocols are no longer supported, the compatibility
glue for them is deleted, and the kernel will panic on boot if any are compiled
in.
2) Certain protocol entry points are modified to take a process structure,
so they they can easily tell whether or not it is possible to sleep, and
also to access credentials.
3) SS_PRIV is no more, and with it goes the SO_PRIVSTATE setsockopt()
call. Protocols should use the process pointer they are now passed.
4) The PF_LOCAL and PF_ROUTE families have been updated to use the new
style, as has the `raw' skeleton family.
5) PF_LOCAL sockets now obey the process's umask when creating a socket
in the filesystem.
As a result, LINT is now broken. I'm hoping that some enterprising hacker
with a bit more time will either make the broken bits work (should be
easy for netipx) or dike them out.
This will make a number of things easier in the future, as well as (finally!)
avoiding the Id-smashing problem which has plagued developers for so long.
Boy, I'm glad we're not using sup anymore. This update would have been
insane otherwise.
(yes I had tested the hell out of this).
I've also temporarily disabled the code so that it behaves as it previously
did (tail drop's the syns) pending discussion with fenner about some socket
state flags that I don't fully understand.
Submitted by: fenner
callers of it to take advantage of this. This reduces new connection
request overhead in the face of a large number of PCBs in the system.
Thanks to David Filo <filo@yahoo.com> for suggesting this and providing
a sample implementation (which wasn't used, but showed that it could be
done).
Reviewed by: wollman
drop the oldest entry in the queue.
There was a fair bit of discussion as to whether or not the
proper action is to drop a random entry in the queue. It's
my conclusion that a random drop is better than a head drop,
however profiling this section of code (done by John Capo)
shows that a head-drop results in a significant performance
increase.
There are scenarios where a random drop is more appropriate.
If I find one in reality, I'll add the random drop code under
a conditional.
Obtained from: discussions and code done by Vernon Schryver (vjs@sgi.com).
time, in seconds, that state for non-established TCP sessions stays about)
a sysctl modifyable variable.
[part 1 of two commits, I just realized I can't play with the indices as
I was typing this commit message.]
to "keepidle". this should not occur unless the connection has
been established via the 3-way handshake which requires an ACK
Submitted by: jmb
Obtained from: problem discussed in Stevens vol. 3
