]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoRemoved -h option.
ru [Thu, 14 Jun 2001 14:21:11 +0000 (14:21 +0000)]
Removed -h option.

23 years agoReally delete the -h option.
ru [Thu, 14 Jun 2001 14:15:41 +0000 (14:15 +0000)]
Really delete the -h option.

CSRG revision 5.33 deleted -h option from getopt(), but not
from usage().  Revision 8.3 restored it in getopt().

23 years agovarious locking fixes, rework open logic and channel registration
cg [Thu, 14 Jun 2001 13:31:30 +0000 (13:31 +0000)]
various locking fixes, rework open logic and channel registration

PR:             kern/28084

23 years agoCommented out USERCONFIG and VISUAL_USERCONFIG options.
nyan [Thu, 14 Jun 2001 11:32:54 +0000 (11:32 +0000)]
Commented out USERCONFIG and VISUAL_USERCONFIG options.

23 years agoRemoved unused files.
nyan [Thu, 14 Jun 2001 11:15:00 +0000 (11:15 +0000)]
Removed unused files.

23 years agoCorrect typo.
nyan [Thu, 14 Jun 2001 11:09:11 +0000 (11:09 +0000)]
Correct typo.

23 years agoMoved the wd33c93 specific file to sys/dev/ic.
nyan [Thu, 14 Jun 2001 11:05:48 +0000 (11:05 +0000)]
Moved the wd33c93 specific file to sys/dev/ic.

23 years agoRemove bogus include.
bp [Thu, 14 Jun 2001 11:05:01 +0000 (11:05 +0000)]
Remove bogus include.

23 years agoCoda do not call vop_defaultop(), so add nesessary calls for VM objects.
bp [Thu, 14 Jun 2001 09:28:30 +0000 (09:28 +0000)]
Coda do not call vop_defaultop(), so add nesessary calls for VM objects.

Submitted by: Greg Troxel <gdt@ir.bbn.com>
MFC after: 2 days

23 years agoDECR() macro still depended on -traditional-cpp behaviour.
ru [Thu, 14 Jun 2001 08:33:59 +0000 (08:33 +0000)]
DECR() macro still depended on -traditional-cpp behaviour.

Noticed by: bde

23 years agoStyle nits:
ru [Thu, 14 Jun 2001 06:08:02 +0000 (06:08 +0000)]
Style nits:

GBINDIR -> ORIGBINDIR
GBINGRP -> ORIGBINGRP

Submitted by: bde

23 years agoThe 3Com 3c505 has not been supported since 3.0.
dd [Thu, 14 Jun 2001 05:29:10 +0000 (05:29 +0000)]
The 3Com 3c505 has not been supported since 3.0.

PR: 28085
Submitted by: Jon Hamilton <hamilton@pobox.com>
References: rev. 1.209 of sys/i386/conf/files.i386

23 years agoSilence warnings and minor style fixes. Mostly constify and don't
dd [Thu, 14 Jun 2001 04:55:26 +0000 (04:55 +0000)]
Silence warnings and minor style fixes.  Mostly constify and don't
assume that time_t is a long.  Clamp down with WARNS=2.

Not objected to by: -audit

23 years agoStart the correct copy of named (pass ${named_program} to ndc(8)).
dd [Thu, 14 Jun 2001 04:34:40 +0000 (04:34 +0000)]
Start the correct copy of named (pass ${named_program} to ndc(8)).

PR: 27051
Submitted by: Maxim Konovalov <maxim@macomnet.ru>

23 years agoPass -a to dmesg(8).
dd [Thu, 14 Jun 2001 04:30:46 +0000 (04:30 +0000)]
Pass -a to dmesg(8).

PR: 26870
Submitted by: Tomonobu AKIMOTO <akimoto@xephion.ne.jp>

23 years agoThis program uses ndc(8); it doesn't send a signal itself.
dd [Thu, 14 Jun 2001 04:28:45 +0000 (04:28 +0000)]
This program uses ndc(8); it doesn't send a signal itself.

PR: 28104
Submitted by: Maxim Konovalov <maxim@macomnet.ru>

23 years agoPart 2 of gensetdefs de-orbit burn. linker sets are now self contained
peter [Thu, 14 Jun 2001 01:35:23 +0000 (01:35 +0000)]
Part 2 of gensetdefs de-orbit burn.  linker sets are now self contained
for ELF, see src/sys/linker_set.h log.

23 years agoAs per comments in sys/linker_set.h:
peter [Thu, 14 Jun 2001 01:28:56 +0000 (01:28 +0000)]
As per comments in sys/linker_set.h:

BANG! BANG! BANG! BANG! BANG! BANG! CLICK! CLICK! CLICK! CLICK! CLICK!
<reload>
BANG! BANG! BANG! BANG! BANG! BANG! CLICK! CLICK! CLICK! CLICK! CLICK!

23 years agoNuke old gensetdefs based linker sets with extreme prejudice
peter [Thu, 14 Jun 2001 01:23:57 +0000 (01:23 +0000)]
Nuke old gensetdefs based linker sets with extreme prejudice

23 years agoFix libpam's linker set stuff to use the new API (unbreak world), and get
peter [Thu, 14 Jun 2001 01:13:30 +0000 (01:13 +0000)]
Fix libpam's linker set stuff to use the new API (unbreak world), and get
rid of gensetdefs from here as well.

23 years agoRespect REPLYTO in mailrc.
mikeh [Thu, 14 Jun 2001 01:08:30 +0000 (01:08 +0000)]
Respect REPLYTO in mailrc.

PR: bin/8322
Submitted by: Philipp Mergenthaler <philipp.mergenthaler@stud.uni-karlsruhe.de>
MFC after: 2 weeks

23 years agoBuild on the change in revision 1.98 by Tor.Egge@fast.no.
mckusick [Wed, 13 Jun 2001 23:13:13 +0000 (23:13 +0000)]
Build on the change in revision 1.98 by Tor.Egge@fast.no.
The symptom being treated in 1.98 was to avoid freeing a
pagedep dependency if there was still a newdirblk dependency
referencing it. That change is correct and no longer prints
a warning message when it occurs. The other part of revision
1.98 was to panic when a newdirblk dependency was encountered
during a file truncation. This fix removes that panic and
replaces it with code to find and delete the newdirblk
dependency so that the truncation can succeed.

23 years agoConvert IIJ copyrights to BSD copyrights.
brian [Wed, 13 Jun 2001 21:52:19 +0000 (21:52 +0000)]
Convert IIJ copyrights to BSD copyrights.

Approved by: Toshiharu OHNO <tohno@sirius.ocn.ne.jp>

23 years agoConvert to mdoc(7).
chris [Wed, 13 Jun 2001 21:52:07 +0000 (21:52 +0000)]
Convert to mdoc(7).

23 years agoFix usage bogon (-d doesn't take any arguments).
bmah [Wed, 13 Jun 2001 21:34:12 +0000 (21:34 +0000)]
Fix usage bogon (-d doesn't take any arguments).

MFC after: 3 days

23 years agoinitialize mp_ncpus in cpu_mp_probe() so that if the probe fails and
gallatin [Wed, 13 Jun 2001 20:40:24 +0000 (20:40 +0000)]
initialize mp_ncpus in cpu_mp_probe() so that if the probe fails and
cpu_mp_start() is never called, mp_ncpus will have a non-zero value.
This prevents systat from dying with an arithmatic exception caused
by a divide-by-zero error on UP alphas running a GENERIC kernel.

23 years agoAdd a -s option to pkg_version, which limits output to packages
bmah [Wed, 13 Jun 2001 20:38:46 +0000 (20:38 +0000)]
Add a -s option to pkg_version, which limits output to packages
matching a given string.

PR: 26114
Submitted by: edwin@mavetju.org (with modifications and manpage update)
MFC after: 3 days

23 years agoFix the build of C programs that link against libperl, and use the
tobez [Wed, 13 Jun 2001 20:37:03 +0000 (20:37 +0000)]
Fix the build of C programs that link against libperl, and use the
blessed way of doing this:

   cc -o interp interp.c `perl -MExtUtils::Embed -e ccopts -e ldopts`

In order for this to work, ldopts should contain -lcrypt.

PR: 21804
Reviewed by: markm

23 years agoThis is force commit intend to correct previous log.
ume [Wed, 13 Jun 2001 19:53:04 +0000 (19:53 +0000)]
This is force commit intend to correct previous log.
`(possible) remote kernel panic fix' was already fixed in 1.132.
I had some confusion during gathering log.

23 years agoAdd description for the 'll' modifier for long long. In essence a copy
schweikh [Wed, 13 Jun 2001 19:05:12 +0000 (19:05 +0000)]
Add description for the 'll' modifier for long long. In essence a copy
from the 'l' description with s/long/long long/g.

PR: 27017
Submitted by: Guy Harris <guy@alum.mit.edu>
MFC after: 2 weeks

23 years agothe last argument to copyinstr is of t ype size_t, not u_int
mjacob [Wed, 13 Jun 2001 18:58:11 +0000 (18:58 +0000)]
the last argument to copyinstr is of t ype size_t, not u_int

23 years agofix recursive mtx_lock() call.
ume [Wed, 13 Jun 2001 18:53:53 +0000 (18:53 +0000)]
fix recursive mtx_lock() call.

Submitted by: mjacob

23 years agoAvoid the avm field joining the w field when avm is > 9.999.999 pages
schweikh [Wed, 13 Jun 2001 18:43:28 +0000 (18:43 +0000)]
Avoid the avm field joining the w field when avm is > 9.999.999 pages
which is slightly less than 4GB. To use a quote from someone who shall
remain nameless "No one will ever need more than 4 GB" :-) But FreeBSD
is prepared if we one day will.

Requested by: Eugene Aleynikov <eugenea@infospace.com>

23 years agoMake compilable. addlog(...) was replaced with log(-1, ...)
ume [Wed, 13 Jun 2001 17:30:55 +0000 (17:30 +0000)]
Make compilable.  addlog(...) was replaced with log(-1, ...)

Reported by: peter

23 years agoRemember to install diskcheckd.conf
phk [Wed, 13 Jun 2001 16:39:23 +0000 (16:39 +0000)]
Remember to install diskcheckd.conf

23 years agoGames ownership cleanup.
ru [Wed, 13 Jun 2001 15:26:31 +0000 (15:26 +0000)]
Games ownership cleanup.

Urged by: bde

23 years agoWARNS= is fully functional again.
ru [Wed, 13 Jun 2001 15:18:11 +0000 (15:18 +0000)]
WARNS= is fully functional again.

23 years agoAdded skeleton <complex.h> (aligned with the POSIX.1-200x), mostly
ru [Wed, 13 Jun 2001 15:16:30 +0000 (15:16 +0000)]
Added skeleton <complex.h> (aligned with the POSIX.1-200x), mostly
to fix the "-nostdinc WARNS=X" breakage caused by broken prototypes
for cabs() and cabsl() in <math.h>.

Reimplemented cabs() and cabsl() using new complex numbers types and
moved prototypes from <math.h> to <complex.h>.

23 years agoAxe -traditional-cpp that behaves badly with -nostdinc and WARNS=2.
ru [Wed, 13 Jun 2001 15:05:55 +0000 (15:05 +0000)]
Axe -traditional-cpp that behaves badly with -nostdinc and WARNS=2.

23 years agoRemoved wrong cast for fts_open()'s third argument.
ru [Wed, 13 Jun 2001 15:01:25 +0000 (15:01 +0000)]
Removed wrong cast for fts_open()'s third argument.

23 years agoComplete prototype for fts_compar.
ru [Wed, 13 Jun 2001 14:59:57 +0000 (14:59 +0000)]
Complete prototype for fts_compar.

23 years agoForced commit.
ru [Wed, 13 Jun 2001 13:25:10 +0000 (13:25 +0000)]
Forced commit.

Revision 1.50 stated that it fixed the -iface breakage introduced with
the latest KAME merge in revision 1.48.  Actually, revision 1.48 fixed
the bug in revision 1.12 which incorrectly tested the ifr_flags member
of the ifreq structure filled by ioctl(SIOCGIFCONF); ifr_flags is only
valid in the SIOCGIFFLAGS case.

But yes, we still want to be able to specify the interface name as the
gateway for non-P2P interfaces.

23 years agoFixed world breakage on systems where ntohl() doesn't return u_long
bde [Wed, 13 Jun 2001 11:56:00 +0000 (11:56 +0000)]
Fixed world breakage on systems where ntohl() doesn't return u_long
(e.g., on alphas, or even on i386's with a POSIX-200x-conformant
ntohl() (ntohl() returns uint32_t which is u_int on i386's)).

Fixed related bugs and bogons while I'm here:
- ntohl() was "fixed" for printing in 1 place by casting to
  "(unsigned int )".  This breaks the value on systems where u_int
  is smaller than uint32_t, and has 2 style bugs.
- spell u_int consistently (never use "unsigned").
- break K&R support some more (don't cast malloc()'s arg to a wrong
  type...).

23 years agoUse getdtablesize() instead of OPEN_MAX.
des [Wed, 13 Jun 2001 11:55:40 +0000 (11:55 +0000)]
Use getdtablesize() instead of OPEN_MAX.

Submitted by: bde

23 years agoWith this commit, I hereby pronounce gensetdefs past its use-by date.
peter [Wed, 13 Jun 2001 10:58:39 +0000 (10:58 +0000)]
With this commit, I hereby pronounce gensetdefs past its use-by date.

Replace the a.out emulation of 'struct linker_set' with something
a little more flexible.  <sys/linker_set.h> now provides macros for
accessing elements and completely hides the implementation.

The linker_set.h macros have been on the back burner in various
forms since 1998 and has ideas and code from Mike Smith (SET_FOREACH()),
John Polstra (ELF clue) and myself (cleaned up API and the conversion
of the rest of the kernel to use it).

The macros declare a strongly typed set.  They return elements with the
type that you declare the set with, rather than a generic void *.

For ELF, we use the magic ld symbols (__start_<setname> and
__stop_<setname>).  Thanks to Richard Henderson <rth@redhat.com> for the
trick about how to force ld to provide them for kld's.

For a.out, we use the old linker_set struct.

NOTE: the item lists are no longer null terminated.  This is why
the code impact is high in certain areas.

The runtime linker has a new method to find the linker set
boundaries depending on which backend format is in use.

linker sets are still module/kld unfriendly and should never be used
for anything that may be modular one day.

Reviewed by: eivind

23 years agoAdd terminal type configuration to the Options screen. It allows selecting
olgeni [Wed, 13 Jun 2001 10:25:09 +0000 (10:25 +0000)]
Add terminal type configuration to the Options screen. It allows selecting
a monochrome display after booting into sysinstall, if you have any trouble
with the default color scheme.

Approved by: jkh
MFC after: 2 weeks

23 years agoAdd '-I mask' cmdline flag to ignore/skip files and subdirectories
roam [Wed, 13 Jun 2001 06:38:58 +0000 (06:38 +0000)]
Add '-I mask' cmdline flag to ignore/skip files and subdirectories
matching a specified shell-glob mask.

Reviewed by: no serious objections on -arch and -audit over
the last few months
MFC after: 1 month

23 years agoAdd warning that some people are having problems with pccard modems in
imp [Wed, 13 Jun 2001 06:36:28 +0000 (06:36 +0000)]
Add warning that some people are having problems with pccard modems in
current, reasons unknown.

Add note about softupdates being unstable since the middle of May.

23 years agoPlug two memory leaks: call login_close() after login_getclass(), and
dd [Wed, 13 Jun 2001 05:49:37 +0000 (05:49 +0000)]
Plug two memory leaks: call login_close() after login_getclass(), and
use free_entry() instead of free() to free a struct _entry.

PR: 28108
Submitted by: Mark Peek <mark@whistle.com>

23 years agoPatch up a blunder I made a few days ago. nmbcnt was being initialized
peter [Wed, 13 Jun 2001 00:36:41 +0000 (00:36 +0000)]
Patch up a blunder I made a few days ago.  nmbcnt was being initialized
too late.

Noted by:      bmilekic
Pointy-hat to: peter

23 years agoMove the definition of epsvall out of #ifdef VIRTUAL_HOSTING so that
dd [Wed, 13 Jun 2001 00:06:42 +0000 (00:06 +0000)]
Move the definition of epsvall out of #ifdef VIRTUAL_HOSTING so that
the latter is not required for ftpd to compile.

23 years agoIncrease field width by one for 'fre' to accomodate today's memory sizes.
schweikh [Tue, 12 Jun 2001 19:47:40 +0000 (19:47 +0000)]
Increase field width by one for 'fre' to accomodate today's memory sizes.
PR: 25267
Submitted by: Eugene Aleynikov <eugene@infospace.com>
Reviewed by: joerg
MFC after: 2 weeks

23 years agoFix about 90-100 warnings one gets when trying to compile lpr&friends
gad [Tue, 12 Jun 2001 16:38:20 +0000 (16:38 +0000)]
Fix about 90-100 warnings one gets when trying to compile lpr&friends
with BDECFLAGS on, mainly by adding 'const' to parameters in a number
of routine declarations.  While I'm at it, ANSI-fy all of the routine
declarations.  The resulting object code is exactly the same after
this update as before it, with the exception of one unavoidable
change to lpd.o on freebsd/alpha.

Also added $FreeBSD$ line to lpc/extern.h lpc/lpc.h lptest/lptest.c

Reviewed by: /sbin/md5, and no feedback from freebsd-audit

23 years agoAllow -u to work with UID's also.
obrien [Tue, 12 Jun 2001 15:58:31 +0000 (15:58 +0000)]
Allow -u to work with UID's also.

Submitted by: Seth Kingsley <seth.kingsley@windriver.com>

23 years agoFixed printing of netatalk statistics WRT the -s flag.
ru [Tue, 12 Jun 2001 14:00:39 +0000 (14:00 +0000)]
Fixed printing of netatalk statistics WRT the -s flag.

PR: bin/6994
Submitted by: Zahemszky Gabor <zgabor@zg.CoDe.hu>
MFC after: 1 week

23 years agoAllow route(8) to create "proxy only" published ARP entries.
ru [Tue, 12 Jun 2001 13:31:53 +0000 (13:31 +0000)]
Allow route(8) to create "proxy only" published ARP entries.

PR: bin/12357
Submitted by: Craig Leres <leres@ee.lbl.gov>

23 years agoFixed the -iface breakage introduced with the latest KAME merge
ru [Tue, 12 Jun 2001 13:23:43 +0000 (13:23 +0000)]
Fixed the -iface breakage introduced with the latest KAME merge
in revision 1.48.  It is pretty valid and often feasible to use
a non-point-to-point interface as the gateway.  One might, for
example, use this to route some hosts through an ARP on a local
interface, without having to assign an additional IP address:

Script started on Tue Jun 12 16:16:09 2001
# ifconfig rl0 inet
rl0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
inet 192.168.4.115 netmask 0xffffff00 broadcast 192.168.4.255
# netstat -arn -finet | grep -w rl0
192.168.4          link#1             UC          3        0     rl0 =>
192.168.4.65       0:d0:b7:16:9c:c6   UHLW        1        0     rl0   1197
# route add -net 192.168.100 -iface rl0
add net 192.168.100: gateway rl0
# ping 192.168.100.1
PING 192.168.100.1 (192.168.100.1): 56 data bytes
64 bytes from 192.168.100.1: icmp_seq=0 ttl=255 time=0.551 ms
64 bytes from 192.168.100.1: icmp_seq=1 ttl=255 time=0.268 ms
^C
--- 192.168.100.1 ping statistics ---
2 packets transmitted, 2 packets received, 0% packet loss
round-trip min/avg/max/stddev = 0.268/0.410/0.551/0.142 ms
# netstat -arn -finet | grep -w rl0
192.168.4          link#1             UC          3        0     rl0 =>
192.168.4.65       0:d0:b7:16:9c:c6   UHLW        1        0     rl0   1165
192.168.100        link#1             UCSc        1        0     rl0 =>
192.168.100.1      0:d0:b7:16:9c:c6   UHLW        1        4     rl0   1192

Script done on Tue Jun 12 16:17:12 2001

23 years agoFreeBSD already avoided namespace pollution (rev.1.45).
ume [Tue, 12 Jun 2001 11:12:23 +0000 (11:12 +0000)]
FreeBSD already avoided namespace pollution (rev.1.45).

Submitted by: bde

23 years agoHints overhaul:
peter [Tue, 12 Jun 2001 09:40:04 +0000 (09:40 +0000)]
Hints overhaul:
- Replace some very poorly thought out API hacks that should have been
  fixed a long while ago.
- Provide some much more flexible search functions (resource_find_*())
- Use strings for storage instead of an outgrowth of the rather
  inconvenient temporary ioconf table from config().  We already had a
  fallback to using strings before malloc/vm was running anyway.

23 years agoRestore the code wrongly nuked by previous commit.
ume [Tue, 12 Jun 2001 08:52:54 +0000 (08:52 +0000)]
Restore the code wrongly nuked by previous commit.

Following changed was made by previous commit:

  - IPV6CP supporting in kernel level ppp from NetBSD.

Submitted by: y.shirasaki@ntt.com

23 years ago-Wunused cleanup
peter [Tue, 12 Jun 2001 07:01:27 +0000 (07:01 +0000)]
-Wunused cleanup

23 years agoMove the -I../../../include or -I/usr/include to the last entry on the
peter [Tue, 12 Jun 2001 06:06:18 +0000 (06:06 +0000)]
Move the -I../../../include or -I/usr/include to the last entry on the
cc arguments.  Otherwise ipfilter's bogus #include lines will compile
reference /usr/include/netinet/ip_frag.h etc.

23 years agoANSI C prohibits text after #endif.
dd [Tue, 12 Jun 2001 03:44:35 +0000 (03:44 +0000)]
ANSI C prohibits text after #endif.

Reviewed by: md5(1)

23 years agoEnable Kerberos 5 support in sshd again.
green [Tue, 12 Jun 2001 03:43:47 +0000 (03:43 +0000)]
Enable Kerberos 5 support in sshd again.

23 years agoTurn on the extra 'const' settings for various curses functions. While
peter [Tue, 12 Jun 2001 01:14:02 +0000 (01:14 +0000)]
Turn on the extra 'const' settings for various curses functions.  While
this is not strictly compliant with XSI curses, it enables us to pass
const strings to many more functions that are actually const safe than
before.  This should be harmless.

Requested by:  lots of folks

23 years agoProperly quote recursive pkg_add operations so that filenames with
jkh [Mon, 11 Jun 2001 23:27:42 +0000 (23:27 +0000)]
Properly quote recursive pkg_add operations so that filenames with
spaces or special characters in them won't barf.

23 years agoMake the non-threaded stub for pthread_sigmask() a no-op instead
iedowse [Mon, 11 Jun 2001 23:18:22 +0000 (23:18 +0000)]
Make the non-threaded stub for pthread_sigmask() a no-op instead
of calling sigprocmask(). This matches the behaviour of thr_sigsetmask()
on Solaris; _pthread_sigmask_stub was added purely for compatibility
with Solaris (for TI-RPC), so it might as well do the same thing.

This fixes the problem where client RPC calls ignored all signals
for the complete duration of the RPC. This behaviour is currently
necessary in the threaded case due to locking issues, but was never
intended to occur in non-threaded programs.

Reviewed by: deischen

23 years agoSay one thing, do the other... nextpid -> lastpid
des [Mon, 11 Jun 2001 23:00:35 +0000 (23:00 +0000)]
Say one thing, do the other... nextpid -> lastpid

23 years agoLinprocfs builds on Alpha now.
des [Mon, 11 Jun 2001 21:57:18 +0000 (21:57 +0000)]
Linprocfs builds on Alpha now.

23 years agoImplement proc/cpuinfo for the Alpha (thanks to gallatin).
des [Mon, 11 Jun 2001 21:55:40 +0000 (21:55 +0000)]
Implement proc/cpuinfo for the Alpha (thanks to gallatin).
Implement proc/pid/cmdline.

23 years agoRename nextpid to lastpid and externalize it.
des [Mon, 11 Jun 2001 21:54:19 +0000 (21:54 +0000)]
Rename nextpid to lastpid and externalize it.

23 years agoRecognize and support new output which pkg_version(1) might produce.
tobez [Mon, 11 Jun 2001 21:31:50 +0000 (21:31 +0000)]
Recognize and support new output which pkg_version(1) might produce.

PR:             27707
Approved by:    bmah, markm

23 years agoCheck for the existence of the Makefile in the port skeleton directory
tobez [Mon, 11 Jun 2001 21:31:04 +0000 (21:31 +0000)]
Check for the existence of the Makefile in the port skeleton directory
before running make.  If the package origin points to a non-existent or
stale port, report this package as orphaned, instead of producing more
general `unknown in index' message.

PR:             27707
Submitted by:   myself, roamer
Approved by:    bmah, markm

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 21:17:59 +0000 (21:17 +0000)]
This is force commit to mention about previous commit.

- add a pointer to struct mauxtag.  two integer was too restrictive
- have m_aux_{add,find}2.
- make sure to return non-cluster on m_pulldown().  this is safer
  (but of course less performant) when we have non-loopback L2 code
  which throws the mbuf back to input path, like L2 bridging or some
  multicast handling code.

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 21:04:12 +0000 (21:04 +0000)]
This is force commit to mention about previous commit.

- move m_aux freeing code into MFREE().

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 20:57:55 +0000 (20:57 +0000)]
This is force commit to mention about previous commit.

- avoid namespace pollution by CMSG_ALIGN().

23 years agoThis is force commit to mention about previous commit (1.50).
ume [Mon, 11 Jun 2001 20:40:49 +0000 (20:40 +0000)]
This is force commit to mention about previous commit (1.50).

- show mtu on netstat -rl

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 20:36:24 +0000 (20:36 +0000)]
This is force commit to mention about previous commit.

- print ipstat.ips_badaddr

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 20:34:19 +0000 (20:34 +0000)]
This is force commit to mention about previous commit.

- add SIOC[GS]LIFPHYADDR

23 years agoNuke wrongly introduced experimental stuff during KAME merge.
ume [Mon, 11 Jun 2001 20:14:45 +0000 (20:14 +0000)]
Nuke wrongly introduced experimental stuff during KAME merge.

Following changed was made by previous commit:

  - add a pointer to struct mauxtag.  two integer was too restrictive.
  - add m_aux_{add,find}2.
  - make sure to nuke mbuf pointed to m_aux.

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 19:48:18 +0000 (19:48 +0000)]
This is force commit to mention about previous commit.

- pass error from ipsec_setsocket() all the way up.

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 19:34:19 +0000 (19:34 +0000)]
This is force commit to mention about previous commit.

- supported the IPV6_V6ONLY socket option.
- prohibit AF_INET6 sockets from sending IPv4 packet (using a mapped address)
- prohibit AF_INET6 sockets from connecting to IPv4 hosts
  (using a mapped address)

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 19:27:05 +0000 (19:27 +0000)]
This is force commit to mention about previous commit.

- do not assume that the ro_dst member of route_in6{} is sockaddr_in6.
- repair IPsec header size prediction.
- validate mbuf chain better in {tcp,udp}6_ctlinput.
- loosened validation inner packets of icmp6 errors as much as possible.
- scope-awareness.
- be friendly with pfctlinput2.
- simplified address scope handling in the ctlinput function.
- type change of in6_pcbnotify.
- pass error from ipsec_setsocket() all the way up.

23 years agoCleanup the tabbing
dillon [Mon, 11 Jun 2001 19:17:05 +0000 (19:17 +0000)]
Cleanup the tabbing

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 19:16:06 +0000 (19:16 +0000)]
This is force commit to mention about previous commit.

- make sure to check MGET failure during IPsec m_aux operations.

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 19:03:42 +0000 (19:03 +0000)]
This is force commit to mention about previous commit.

- added comments about why in6p_inputopts should not be copied in
  tcp6_input().
- call ip6_copypktopts() in order to copy in6p_outputopts
  from a listening PCB to a corresponding accepting one.
- be proactive about unspecified IPv6 source address.  pcb layer
  uses unspecified address (::) to mean "unbounded" or
  "unconnected", and can be confused by packets from outside.
- made consist between tcp and udp in using mappedaddr
- setsockopt(BIND_IPV6ONLY) now works
- deprecated address consideration on TCP SYN.
- get rid of M_ANYCAST6

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 18:46:51 +0000 (18:46 +0000)]
This is force commit to mention about previous commit.

- pass error from ipsec_setsocket() all the way up.
- check ipsec policy (mis)match on raw ip{4,6} socket.

23 years agoBlah, I cut out a tad too much in the previous commit. (thanks again, Jake!)
des [Mon, 11 Jun 2001 18:43:32 +0000 (18:43 +0000)]
Blah, I cut out a tad too much in the previous commit. (thanks again, Jake!)

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 18:41:57 +0000 (18:41 +0000)]
This is force commit to mention about previous commit.

- use 0/8 to specify interface index on multicast get/setsockopt
- introduce ipstat.ips_badaddr

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 18:38:11 +0000 (18:38 +0000)]
This is force commit to mention about previous commit.

- use 0/8 to specify interface index on multicast get/setsockopt
- make sure to nuke m->m_aux pointer for ipsec, on if_output.
- pass error from ipsec_setsocket() all the way up.
- move ipsec output processing before filtering section.

23 years agocopyin(9) doesn't return ENAMETOOLONG. (thanks, Jake!)
des [Mon, 11 Jun 2001 18:36:18 +0000 (18:36 +0000)]
copyin(9) doesn't return ENAMETOOLONG. (thanks, Jake!)

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 18:21:31 +0000 (18:21 +0000)]
This is force commit to mention about previous commit.

- (possible) remote kernel panic fix - out of bounds access on
  ill-formed ipopt.
- strict boundary check on ipopt.
- make sure to enforce inbound IPsec policy on all final header.
- add missing ipcomp entry from ipprotosw.
- 127/8 must not appear on wire - RFC1122.
  this is rather important as we use weak host model, so outsider
  can abuse 127.0.0.1 from outside.
- introduce ipstat.ips_badaddr
- use ipsec_gethist() to prevent packet filters from looking at
  decapulated packets.
- remove duplicate 127.0.0.0/8 checking.

23 years agoUse struct device * rather than device_t for (*pci_intr_route args), so
gallatin [Mon, 11 Jun 2001 18:19:44 +0000 (18:19 +0000)]
Use struct device * rather than device_t for (*pci_intr_route args), so
as to match other prototypes in the file.

23 years agoFix small bogon.
jlemon [Mon, 11 Jun 2001 18:15:31 +0000 (18:15 +0000)]
Fix small bogon.

Submitted by: Anton Berezin <tobez@tobez.org>

23 years ago- Restore -nostdinc that got lost in rev.1.105; we don't
ru [Mon, 11 Jun 2001 18:09:08 +0000 (18:09 +0000)]
- Restore -nostdinc that got lost in rev.1.105; we don't
  want host headers during `buildworld'.

- During `buildworld', install headers in a "copy" mode
  until we decide what to do with the (currently broken)
  SHARED=symlinks.

- Temporarily run `buildworld' with -DNO_WERROR, which
  effectively disabled the -Werror bit of recently added
  WARNS=X feature.  This is required because adding the
  -nostdinc bit back revealed bugs in some header files
  that were hiding after not using -nostdinc.
  It is unclear currently how exactly (and why) -nostdinc
  affects gcc(1) warnings.

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 18:00:56 +0000 (18:00 +0000)]
This is force commit to mention about previous commit.

- allow oversized routing messages.

23 years agoThis is force commit to mention about previous commit.
ume [Mon, 11 Jun 2001 17:52:02 +0000 (17:52 +0000)]
This is force commit to mention about previous commit.

- use getifaddrs() instead of directly executing SIOCGIFCONF
- added a sockaddr_storage member to sockunion{} for safety
- scoped address printing

23 years agoBackout removal of -I${.CURDIR}/../../../sys in previous revision.
ru [Mon, 11 Jun 2001 17:48:15 +0000 (17:48 +0000)]
Backout removal of -I${.CURDIR}/../../../sys in previous revision.
It is needed to pick up (missing in SHARED=copy case) netatm/uni/
headers.

The right fix would be to install missing netatm/uni headers.

23 years agoBackout previous change (removal of -I${.CURDIR}/../../sys/netinet).
ru [Mon, 11 Jun 2001 17:41:58 +0000 (17:41 +0000)]
Backout previous change (removal of -I${.CURDIR}/../../sys/netinet).
This is needed to pick up the right headers.  Wrong headers from
src/contrib/ipfilter are used otherwise.

The right fix would be to fix contrib/ipfilter C sources to pick up
headers from <sys/netinet>.

Noticed by: peter

23 years agoMention about -rl behavior change. Now, MTU is printed.
ume [Mon, 11 Jun 2001 17:34:25 +0000 (17:34 +0000)]
Mention about -rl behavior change.  Now, MTU is printed.