Commit Graph

448 Commits

Author SHA1 Message Date
tls 31d0752b99 change IN_IFADDR_HASH_SIZE to 509, which actually uses no more space than 293 due to rounding up to nearest power of two in hashinit. 1998-07-29 05:18:54 +00:00
pk 84840da908 in_pcballoc(): we can't afford to wait for memory. 1998-07-23 08:24:33 +00:00
mycroft cca4e566a9 Implement a better fix for the `gratuitous FIN' problem, as
mentioned on tcp-impl but with a bit more commentary.
1998-07-21 10:46:00 +00:00
thorpej 3a9ed00799 Document that we are more conservative after doing MTU discovery than the
suggestion in draft-floyd-incr-init-win-03.  Rather than scaling cwnd back
by the ratio of new segment size to old segment size, we perform a slow start
using the Initial Window, computed with the new segment size.
1998-07-17 23:09:58 +00:00
thorpej 0f909866c0 Clarify that we're using the Loss Window when we receive a source quench. 1998-07-17 23:02:38 +00:00
thorpej fa20f24cd9 Add a comment wrt. a current issue w/ CWM. 1998-07-17 23:00:02 +00:00
thorpej a3f4316cba Clarify that we are using the Loss Window if a retransmission occurred
during the three-way handshake.
1998-07-17 22:58:56 +00:00
thorpej 830879a809 Comment where the Restart Window is computed, and in the non-CWM case,
make sure it never _increases_ cwnd.
1998-07-17 22:52:01 +00:00
thorpej 1c4ff0a086 Comment where we use the Loss Window. 1998-07-17 22:18:49 +00:00
sommerfe 69b1b4758d Fix PR5559: if fast-forwarding, DF set, and packet too large, send ICMP error.. 1998-07-17 00:35:23 +00:00
sommerfe 534520d815 Fix PR5508: ipfil cut-through forwarding causes panic 1998-07-17 00:28:00 +00:00
tls deac3540de Put original hash function back. It wastes a little bit of space, but is much more even -- think of the case of a web service provider, some of whose customers end up getting 'inferior service' because they're on addresses that happen to be out at the end of a hash chain. With webservers with thousands of addresses, this is a real issue. If the wasted space is a big deal, we could pick a prime number that's slightly _less_ than a power of two... 1998-07-16 06:45:09 +00:00
thorpej 389da54091 Garbage collect `imp' and `hy'. We don't have the rest of the code, and
it's not like anyone is ever going to be using either of them.
1998-07-15 17:39:20 +00:00
veego 97ab1bd53b Resolve conflicts from the import. 1998-07-12 15:23:59 +00:00
mycroft 3a64270ca6 Back out the change from TCP/IP vol 2, in revision 1.7, which removed TH_FIN
from the output flags for CLOSING state.  There is no harm in retransmitting
the FIN, and this change has unexpected side effects that break simultaneous
close behaviour.
1998-07-09 05:49:56 +00:00
sommerfe 065cac9798 Delete bogus (void) cast of m_freem (which is already a void function..) 1998-07-07 00:04:59 +00:00
jonathan b37021c1a1 defopt NATM. 1998-07-05 22:48:05 +00:00
jonathan 9bf2ba0928 Garbage-collect ``needs-flag'' from attributes ether, fddi, arc:
NETHER, NFDDI, NARC are  not used anywhere. Remove #include "ether.h",
   which had no effect.
Removes clash with "options NATM" for native-ATM network protocol stack.
1998-07-05 22:29:51 +00:00
jonathan 011f2bda08 defopt NS, NSIP. 1998-07-05 06:49:00 +00:00
jonathan 5c0c5dd0b4 defopt ISO TPIP. 1998-07-05 04:37:35 +00:00
jonathan f2a2327e0a defopt EON. 1998-07-05 01:06:49 +00:00
jonathan 3751946b97 defopt INET, NETATALK. 1998-07-05 00:51:04 +00:00
jonathan 466e784ee1 defopt DDB. 1998-07-04 22:18:13 +00:00
thorpej 8cfe8959a6 Fix TCPS_HAVERCVDFIN() to actually catch all TCP states in which a FIN
has been received (CLOSE_WAIT, CLOSING, LAST_ACK, and TIME_WAIT).

From David Borman <dab@bsdi.com>.
1998-07-03 05:39:56 +00:00
is 0ca02c68a7 Thinko in last fix: we have to actually check each address for a copy on
our ifp, else we might fail for some strange configurations.
1998-07-02 14:00:39 +00:00
is d8b8a41918 The rewrite of if_arp.c to work with the hashed interface address lists
(1.44) missed a test for the right interface, making some machines answer
to some bogus arp requests (like for WHO-HAS 127.0.0.1).

The quick patch in 1.46-1.47 does not work for so-called "unnumbered"
interfaces, that is, (point-to-point) interfaces that share their local
address with another (e.g., the Ethernet) interface.

We add a macro to in_var.h, to step (in the current implementation) through
the hash chain and fine more entries with the same address, and use that
in if_arp.c to find one which belongs to our interface.
1998-07-02 11:39:56 +00:00
tls b0d2c08b6b Fix buglet where we might respond to arp on wrong interface. 1998-06-25 20:47:48 +00:00
cgd 651b44e211 Rework the way kernel include files are installed. In the new method,
as with user-land programs, include files are installed by each directory
in the tree that has includes to install.  (This allows more flexibility
as to what gets installed, makes 'partial installs' easier, and gives us
more options as to which machines' includes get installed at any given
time.)  The old SYS_INCLUDES={symlinks,copies} behaviours are _both_
still supported, though at least one bug in the 'symlinks' case is
fixed by this change.  Include files can't be build before installation,
so directories that have includes as targets (e.g. dev/pci) have to move
those targets into a different Makefile.
1998-06-12 23:22:30 +00:00
sommerfe a90d5cd12e Truncate mbufs to the correct length before forwarding; fixes pr5560 1998-06-10 00:47:57 +00:00
thorpej b22946827d Add a comment explaining why we do _not_ ACK data that might accompany
a SYN (avoidance of a DoS attack).
1998-06-02 18:33:02 +00:00
thorpej c296923d2f Loss window MUST be one segment, per draft-floyd-incr-init-win-03. 1998-06-02 17:22:26 +00:00
thorpej edc01ec330 In addition to the IP flow hash table, put the flows on a list. The table
is used for fast lookup, the list for traversal of all flows.  Also, use
PRT timers.
1998-06-02 15:48:03 +00:00
thorpej 837a8317b5 Eek, we were wasting almost half of the in_ifaddr hash space by modulo'ing
with IN_IFADDR_HASH_SIZE.  Instead, AND with the hash mask computed by
hashinit().
1998-06-01 00:50:07 +00:00
thorpej 08b5a4ecb8 Protect the ipflow_reap() call with splsoftnet. 1998-06-01 00:39:37 +00:00
cgd dd8ed56342 Another demonstration that when you're converting variables from 'long's
to fixed 32-bit integers, you have to exercise care.
1998-05-31 19:39:13 +00:00
veego 6250554a65 Fix compiler warnings: Add missing ()'s. 1998-05-29 20:28:17 +00:00
veego 630030601c Fix some compiler warnings: Missing prototype and ()'s. 1998-05-29 20:27:18 +00:00
veego a4c89e3e2e Resolve conflicts from the import of IPFilter 3.2.7. 1998-05-29 20:24:36 +00:00
matt c0a1601f95 Change arp so its console log messages print out IP addresses in
dotted quad format instead of hex.
1998-05-29 15:34:24 +00:00
thorpej f555f6d93f Fix OBOB in IP timestamp option processing, as noted in FreeBSD PR 6738,
from Jennifer Dawn Meyers <jdm@enteract.com>.
1998-05-24 20:14:53 +00:00
matt f070ddb8ed Move the ppcb pointer towards the front of the structure so that it and the
pcb chain pointers can possibly be in the same cache line.
1998-05-18 17:10:37 +00:00
matt 1b2b1d801b Fix two bugs. 1998-05-18 17:08:56 +00:00
veego 82423e3d01 Resolve conflicts 1998-05-17 16:50:15 +00:00
kml dd5ed34b88 Changed initialization of peermss to ensure that it didn't have
the TCP and IP options lengths removed from it -- the IP options can
change over the course of a connection...
1998-05-12 21:45:51 +00:00
thorpej c5fc2e9acd Back out previous. This problem was already fixed in a different way. 1998-05-11 23:13:40 +00:00
matt 53b04a8d3c Let usr.sbin/tcpdump build again. 1998-05-11 23:09:35 +00:00
thorpej 49573284f5 Make sure a timer is marked "disarmed" once it has expired. 1998-05-11 20:52:18 +00:00
thorpej 5596fe2614 Nuke TUBA per my note to tech-net; there's no reason to keep it around. 1998-05-11 19:57:23 +00:00
kml 1216b9a560 Change comments on tcp_mss_to_advertise to match actual arguments 1998-05-07 22:30:23 +00:00
thorpej ce3d776874 Rework the syn cache code somewhat:
- Don't use home-grown queue manipulation.  Use <sys/queue.h> instead.  The
  data structures are a little larger, but we are otherwise wasting the
  memory chunk anyway (we're already a 64-byte malloc bucket).
- Fix a bug in the cache-is-full case: if the oldest element removed from
  the first non-empty bucket was the only element in the bucket, the
  bucket wouldn't be removed from the bucket cache, causing queue corruption
  later.
- Optimize the syn cache timers by using PRT timers rather than home-grown
  decrement-and-propagate timers.

This code is now a fair bit smaller, and significantly easier to read
and understand.
1998-05-07 01:37:27 +00:00