value of the TCP_NODELAY socket option from the listener to the
newly connected connection. Agrees with how Linux & FreeBSD behave,
and goes more with the spirit of accept(2) creating a socket with
the same properties as the listener.
Analysis by Kevin Lahey. Closes PR 17616 by myself.
* Keep pointers to the first and last mbufs of the last record in the
socket buffer.
* Use the sb_lastrecord pointer in the sbappend*() family of functions
to avoid traversing the packet chain to find the last record.
* Add a new sbappend_stream() function for stream protocols which
guarantee that there will never be more than one record in the
socket buffer. This function uses the sb_mbtail pointer to perform
the data insertion. Make TCP use sbappend_stream().
On a profiling run, this makes sbappend of a TCP transmission using
a 1M socket buffer go from 50% of the time to .02% of the time.
Thanks to Bill Sommerfeld and YAMAMOTO Takashi for their debugging
assistance!
1. size_t is 64 bits, so use a u_32_t for iplused
2. microtime() and friends expect a struct timeval,
passing the first of two unsigned longs will not cut it.
as necessary:
* Implement a new mbuf utility routine, m_copyup(), is is like
m_pullup(), except that it always prepends and copies, rather
than only doing so if the desired length is larger than m->m_len.
m_copyup() also allows an offset into the destination mbuf, which
allows space for packet headers, in the forwarding case.
* Add *_HDR_ALIGNED_P() macros for IP, IPv6, ICMP, and IGMP. These
macros expand to 1 if __NO_STRICT_ALIGNMENT is defined, so that
architectures which do not have strict alignment constraints don't
pay for the test or visit the new align-if-needed path.
* Use the new macros to check if a header needs to be aligned, or to
assert that it already is, as appropriate.
Note: This code is still somewhat experimental. However, the new
code path won't be visited if individual device drivers continue
to guarantee that packets are delivered to layer 3 already properly
aligned (which are rules that are already in use).
- the destination is IPv4 multicast or 255.255.255.255, and
- outgoing interface is specified via socket option
this simplifies operation of routed
(no longer reqiure 224.0.0.0/4 to be set up)
behavior changes:
- two iocts used by ndp(8) are now obsolete (backward compat provided).
use sysctl path instead.
- lo0 does not get ::1 automatically. it will get ::1 when lo0 comes up.
benefit currently). Rework tcp_reass code to optimize the 4 most likely causes
of out-of-order packets: first OoO pkt, next OoO pkt in seq, OoO pkt is part
of new chuck of OoO packets, and the OoO pkt fills the first hole. Add evcnts
to instrument tcp_reass (enabled by the options TCP_REASS_COUNTERS). This is
part 1/2 of tcp_reass changes.
* Remove the code that allocates a cluster if the packet would
fit in one; it totally defeats doing references to M_EXT mbufs
in the socket buffer. This drastically reduces the number of
data copies in the tcp_output() path for applications which use
large writes. Kudos to Matt Thomas for pointing me in the right
direction.
the QNX licence seems to be allow both non-commercial and commercial
use actually.
According to Darren, the H.323 proxy code is buggy ATM, but is imported
here for reference anyway.
Configured by a new option "mssclamp" in NAT rules, like:
map pppoe0 192.168.1.0/24 -> 0/32 mssclamp 1452
This is based on work by Xiaodan Tang <xtang@qnx.com>.
deal with shortages of the VM maps where the backing pages are mapped
(usually kmem_map). Try to deal with this:
* Group all information about the backend allocator for a pool in a
separate structure. The pool references this structure, rather than
the individual fields.
* Change the pool_init() API accordingly, and adjust all callers.
* Link all pools using the same backend allocator on a list.
* The backend allocator is responsible for waiting for physical memory
to become available, but will still fail if it cannot callocate KVA
space for the pages. If this happens, carefully drain all pools using
the same backend allocator, so that some KVA space can be freed.
* Change pool_reclaim() to indicate if it actually succeeded in freeing
some pages, and use that information to make draining easier and more
efficient.
* Get rid of PR_URGENT. There was only one use of it, and it could be
dealt with by the caller.
From art@openbsd.org.
This avoids kernel crashes when we don't handle nonsensial values
like 0 gracefully. Better check here once beforehand than having to
check for non meaningful values in time critical paths (like tcp_output).
Fixes PR 15709.
Don't copy ttl from the inner packet to the encapsulating packet. Make
the outer ttl sysctl'able. This should close PR 14269 from Jasper Wallace
(change partly from there) and it makes traceroute work over gre tunnels.
outgoing interface (ia == NULL after IFP_TO_IA).
historic behavior (up to revision 1.43) was to use 0.0.0.0 as source address,
but it seems like a mistake according to RFC1112/1122.
as config(8) will warn for value-less defparam options
- minor whitespace/formatting cleanup
- consolidate opt_tcp_recvspace.h and opt_tcp_sendspace.h into opt_tcp_space.h
as discussed in tech-net several weeks ago. It turned out that
KAME had already added this functionality to the IPv6 stack, so
I followed their example in adding the sysctl variables
net.inet.icmp.rediraccept and net.inet.icmp.redirtimeout.
Add capabilities bits that indicate an interface can only perform
in-bound TCPv4 or UDPv4 checksums. There is at least one Gig-E chip
for which this is true (Level One LXT-1001), and this is also the
case for the Intel i82559 10/100 Ethernet chips.
all open TCP connections in tcp_slowtimo() (which is called 2x
per second). It's fairly rare for TCP timers to actually fire,
so saving this list traversal is good, especially if you want
to scale to thousands of open connections.
and call it directly from tcp_slowtimo() (via a table) rather
than going through tcp_userreq().
This will allow us to call TCP timers directly from callouts,
in a future revision.
Instead of incrementing t_idle and t_rtt in tcp_slowtimo(), we now
take a timstamp (via tcp_now) and use subtraction to compute the
delta when we actually need it (using unsigned arithmetic so that
tcp_now wrapping is handled correctly).
Based on similar changes in FreeBSD.
provide a better support for multiple address with the same prefix better.
(like 10.0.0.1/8 and 10.0.0.2/8 on the same interface)
continuation of PR 13311.
remove irrelevant #if 0'ed segment for PR 10427.
- consider non-primary (2nd and beyond) IPv4 address as "local", and prevent
outgoing ARP.
- for routing entries generated by ARP, make sure to set rt->rt_ifa equal to
rt_key, to help IPv4 source address selection for traffic to myself.
PR 13311.
caveats/TODOs:
- interface routes ("connected routes" in cisco terminlogy) is tied with the
primary (1st) IPv4 address on the interface. should be fixed with updates
to rt_ifinit().
- source address selection for offlink locations. 1st address tend to be used
with the current code
(you can configure it right by setting rt->rt_ifa accordingly).
network interfaces. This works by pre-computing the pseudo-header
checksum and caching it, delaying the actual checksum to ip_output()
if the hardware cannot perform the sum for us. In-bound checksums
can either be fully-checked by hardware, or summed up for final
verification by software. This method was modeled after how this
is done in FreeBSD, although the code is significantly different in
most places.
We don't delay checksums for IPv6/TCP, but we do take advantage of the
cached pseudo-header checksum.
Note: hardware-assisted checksumming defaults to "off". It is
enabled with ifconfig(8). See the manual page for details.
Implement hardware-assisted checksumming on the DP83820 Gigabit Ethernet,
3c90xB/3c90xC 10/100 Ethernet, and Alteon Tigon/Tigon2 Gigabit Ethernet.
There is no place in the source where this bit could ever be set (or I'm
to blind to find it).
This fixes PR 12671.
If someone thinks this is the wrong solution, please make sure to (a) reopen
the PR and (b) explain to me how the tested bits would ever get set. I'll
be glad to then look further for the real cause (i.e. the flags not getting
set in the case described in the PR).
(reported by: grendel@heorot.stanford.edu (Ted U)
- Don't cast malloc/realloc/calloc return values because they hide LP64 bugs.
- Don't destroy the whole array when realloc fails
- Use calloc in all cases (malloc was used inconsistently).
- Avoid duplicating code.
Reviewed by: ross
"lots of fragmented packets" DoS attack.
the current default value is derived from ipv6 counterpart, which is
a magical value "200". it should be enough for normal systems, not sure
if it is enough when you take hundreds of thousands of tcp connections on
your system. if you have proposal for a better value with concrete reasons,
let me know.
ISS attacks (which we already fend off quite well).
1. First-cut implementation of RFC1948, Steve Bellovin's cryptographic
hash method of generating TCP ISS values. Note, this code is experimental
and disabled by default (experimental enough that I don't export the
variable via sysctl yet, either). There are a couple of issues I'd
like to discuss with Steve, so this code should only be used by people
who really know what they're doing.
2. Per a recent thread on Bugtraq, it's possible to determine a system's
uptime by snooping the RFC1323 TCP timestamp options sent by a host; in
4.4BSD, timestamps are created by incrementing the tcp_now variable
at 2 Hz; there's even a company out there that uses this to determine
web server uptime. According to Newsham's paper "The Problem With
Random Increments", while NetBSD's TCP ISS generation method is much
better than the "random increment" method used by FreeBSD and OpenBSD,
it is still theoretically possible to mount an attack against NetBSD's
method if the attacker knows how many times the tcp_iss_seq variable
has been incremented. By not leaking uptime information, we can make
that much harder to determine. So, we avoid the leak by giving each
TCP connection a timebase of 0.
normal operation (/var can get filled up by flodding bogus packets).
sysctl net.inet6.icmp6.nd6_debug will turn on diagnostic messages.
(#define ND6_DEBUG will turn it on by default)
improve stats in ND6 code.
lots of synchronziation with kame (including comments and cometic ones).
- let ipfilter look at wire-format packet only (not the decapsulated ones),
so that VPN setting can work with NAT/ipfilter settings.
sync with kame.
TODO: use header history for stricter inbound validation
since we do not have feedback mechanism from path MTU to tunnel MTU
(not sure if we should), and inner packet source will not get informed of
outer PMTUD (we shouldn't do this), 1.15 behavior can lead us to
blackhole behavior.
configurable behavior (as suggested in RFC2401 6.1) would be nice to have,
however, reusing net.inet.ipsec.dfbit would be hairy.
have sys/net/route.c:rtrequest1(), which takes rt_addrinfo * as the argument.
pass rt_addrinfo all the way down to rtrequest, and ifa->ifa_rtrequest.
3rd arg of ifa->ifa_rtrequest is now rt_addrinfo * instead of sockaddr *
(almost noone is using it anyways).
benefit: the follwoing command now works. previously we need two route(8)
invocations, "add" then "change".
# route add -inet6 default ::1 -ifp gif0
remove unsafe typecast in rtrequest(), from rtentry * to sockaddr *. it was
introduced by 4.3BSD-reno and never corrected.
XXX is eon_rtrequest() change correct regarding to 3rd arg?
eon_rtrequest() and rtrequest() were incorrect since 4.3BSD-reno,
so i do not have correct answer in the source code.
someone with more clue about netiso-over-ip, please help.
the link level name for the interface (ifp->if_sadl) is allocated
before ifp->if_addrlen is initialized, which could lead to allocating
too little space for the link level address.
Do this by splitting allocation of the link level name out of
if_attach() and into if_alloc_sadl(), which is normally called
by functions like ether_ifattach(). Network interfaces which
don't have a link-specific attach routine must call if_alloc_sadl()
themselves (example: gif).
Link level names are freed by if_free_sadl(), which can be called
from e.g. ether_ifdetach(). Drivers never need call if_free_sadl()
themselves as if_detach() will do it if it is not already done.
While here, add the ability to pass an AF_LINK address to
SIOCSIFADDR in ether_ioctl() (this is what caused me to notice
the problem that the above fixes).
by interface index. if the interface address specified is in 0.0.0.0/8
it will be considered as interface index in network byteorder.
getsockopt(IP_MULTICAST_IF) preserves old behavior if
setsockopt(IP_MULTICAST_IF) was done with interface address, and
returns interface index if setsockopt(IP_MULTICAST_IF) was done with
interface index (again using the form in 0.0.0.0/8).
Suggested by Dave Thaler, based on RIPv2 MIB spec (RFC1724 section 3.3).
http://mail-index.netbsd.org/tech-net/2001/01/13/0003.html
"key" and a "dlt", use a "type" (PFIL_TYPE_{AF,IFNET} for now) and
a val/ptr appropriate for that type. This allows for more future
flexibility with the pfil_hook mechanism.
mandatory for IPv6 (so we can't just validate by using connected pcb - we need
to allow traffic from unconnected pcb to do pmtud).
- if the traffic is validated by xx_ctlinput, allow up to "hiwat" pmtud
route entries.
- if the traffic was not validated by xx_ctlinput, allow up to "lowat" pmtud
route entries (there's upper limit, so bad guys cannot blow up our routing
table).
sync with kame
XXX need to think again about default hiwat/lowat value.
XXX victim selection to help starvation case
- All packets are passed to PFIL_HOOKS as they come off the wire, i.e.
fields in protocol headers in network order, etc.
- Allow for multiple hooks to be registered, using a "key" and a "dlt".
The "dlt" is a BPF data link type, indicating what type of header is
present.
- INET and INET6 register with key == AF_INET or AF_INET6, and
dlt == DLT_RAW.
- PFIL_HOOKS now take an argument for the filter hook, and mbuf **,
an ifnet *, and a direction (PFIL_IN or PFIL_OUT), thus making them
less IP (really, IP Filter) centric.
Maintain compatibility with IP Filter by adding wrapper functions for
IP Filter.
Current status:
Only OHCI chip is supported (fwohci).
ping (IPv4) works with Sony's implementation (SmartConnect) on Win98.
sometimes works but not stable.
Not implemented yet:
IRM (Isochronous Resource Manager) functionality.
Link layer fragmentation.
Topology map.
More to do:
clean ups
MCAP
charactor device part
dhcp
There is no entry in GENERIC config file yet.
Follow sys/dev/ieee1394/IMPLEMENTATION to enable if_fw.
TODO: udp6, and sendto consideration. as pmtud is mandatory for IPv6,
it is rather important for us to support those cases.
TODO: more testing
TODO: kame sync
entering rtentry's for hosts we're not actually communicating
with.
Do this by invoking the ctlinput for the protocol, which is
responsible for validating the ICMP message:
* TCP -- Lookup the connection based on the address/port
pairs in the ICMP message.
* AH/ESP -- Lookup the SA based on the SPI in the ICMP message.
If validation succeeds, ctlinput is responsible for calling
icmp_mtudisc(). icmp_mtudisc() then invokes callbacks registered
by protocols (such as TCP) which want to take some sort of special
action when a path's MTU changes. For TCP, this is where we now
refresh cached routes and re-enter slow-start.
As a side-effect, this fixes the problem where TCP would not be
notified when a path's MTU changed if AH/ESP were being used.
XXX Note, this is only a fix for the IPv4 case. For the IPv6
XXX case, we need to wait for the KAME folks.
Reviewed by sommerfeld@netbsd.org and itojun@netbsd.org.
ip_output(). This flag, if set, causes ip_output() to set
DF in the IP header if the MTU in the route is not locked.
This allows a bunch of redundant code, which I was never
really all that happy about adding in the first place, to
be eliminated.
Inspired by a similar change made by provos@openbsd.org when
he integrated NetBSD's Path MTU Discovery code into OpenBSD.
packets. PR 11082.
This is a short-term workaround. whenever new ipfilter comes out with
proper non-IPv4 support, we should migrate to the new ipfilter.
each in_ifaddr and delete it when an address is purged.
- Don't simply try to delete a multicast address record listed in the
ia_multiaddrs. It results a dangling pointer. Let who holds a
reference to it to delete it.
- when all the interface address is removed from an interface, and there's
multicast groups still left joined, keep it in kludge table.
- when an interface address is added again, recover multicast groups from
kludge table.
this will avoid problem with dangling in_ifaddr on pcmcia card removal,
due to the link from multicast group info (in_multi).
the code is basically from sys/netinet6/in6.c (jinmei@kame).
pointed out by: Shiva Shenoy <shiva_s@yahoo.com>
Without this, if a v6 address is placed before a v4 address in if_addrlist,
a PRU_PURGEIF request for v6 tcp protocol purges also v4 addresses and,
as a result, if_detach fails to request PRU_PURGEIF for v4 protocols
other than tcp.
this avoids too aggressive memory usage on heavy load web server, for example.
From: Kevin Lahey <kml@dotrocket.com>
release and reallocate t_template, if t_template->m_len changes.
(this happens if we connect to IPv4 mapped destination and then IPv6
destination, on a single AF_INET6 socket)
KAME 1.26 -> 1.28
rev 1.35 of ip_nat.c checks if packets are too short.
For ICMP packets, this packet length checking double counts
the length of an IP header contained in ICMP messages.
So, unless ICMP packets are long enough (such as echo-reply),
packets are mistakingly considered too short and are dropped.
multiple addresses from same prefix, onto single interface. PR 10427.
more info:
- 4.4BSD did not check return code from in_ifinit() at all.
4.4BSD does not support multiple address from same prefix.
- past KAME change passed in{,6}_ifinit() to upwards, toward ifconfig(8).
the behavior is filed as PR 10427.
- the commit inhibits EEXIST from rtinit(), hence partially recovers old
4.4BSD behavior.
- the right thing to happen is to properly support multiple address assignment
from the same prefix. KAME tree has more extensive change, however, it needs
much more time to get stabilized (rtentry refcnt change can cause serious
issue, we really need to bake it before bring it to netbsd)
fails, return EIO instead.
- iplioctl(): If performing a NAT operation, and IP Filter is not
yet initialized (e.g. by `ipf -E'), enable it implicitly before
doing the NAT operation.
basis. default: 100pps
set default value for net.inet.tcp.rstratelimit to 0 (disabled),
NOTE: it does not work right for smaller-than-1/hz interval. maybe we should
nuke it, or make it impossible to set smaller-than-1/hz value.
unspecified address (::) to mean "unbounded" or "unconnected",
and can be confused by packets from outside.
use of :: as source is not documented well in IPv6 specification.
not sure if it presents a real threat. the worst case scenario is a DoS
against TCP listening socket:
- outsider transmit TCP SYN with :: as IPv6 source
- receiving side creates TCP control block with:
local address = my addres
remote address = :: (meaning "unconnected")
state = SYN_RCVD
note that SYN ACK will not be sent due to ip6_output() filter.
this stays until it timeouts.
- the TCP control block prevents listening TCP control block from
being contacted (DoS).
udp6/raw6 socket may have similar problem, but as they are connectionless,
it may too much to filter it out.
reports a size smaller than the udp header; defends against bogosity
detected by Assar Westerlund.
This patch and the previous ip_icmp.c change were the joint work of
assar, itojun, and myself.
- allow it to work when icmpreturndatabytes is sufficiently large that the
icmp error message doesn't fit in a header mbuf.
- defend against mbuf chains shorter than their contained ip->ip_len.
- add protection mechanism against ND cache corruption due to bad NUD hints.
- more stats
- icmp6 pps limitation. TOOD: should implement ppsratecheck(9).
in ip_mroute.c for duplicate tunnel entries, too. Well, what
really needs to happen is that the mrouting code needs to be
changed to work w/ `gif' tunnels... but...
