openbsd
14 years agoCorrect comment. scsi_sense_data is now 18 bytes, not 32.
krw [Tue, 13 Jul 2010 00:30:30 +0000 (00:30 +0000)]
Correct comment. scsi_sense_data is now 18 bytes, not 32.

14 years agoreplace rmd160 links with sha256. it's way more popular and more better, too.
tedu [Mon, 12 Jul 2010 23:16:55 +0000 (23:16 +0000)]
replace rmd160 links with sha256.  it's way more popular and more better, too.
too many oks to enumerate.
while here, add Xr to the relevant libc functions in each page.

14 years agoexpand %h to the hostname in ssh_config Hostname options. While this
djm [Mon, 12 Jul 2010 22:41:13 +0000 (22:41 +0000)]
expand %h to the hostname in ssh_config Hostname options. While this
sounds useless, it is actually handy for working with unqualified
hostnames:

Host *.*
Hostname %h
Host *
Hostname %h.example.org

"I like it" markus@

14 years agoMake ExitOnForwardFailure work with fork-after-authentication ("ssh -f")
djm [Mon, 12 Jul 2010 22:38:52 +0000 (22:38 +0000)]
Make ExitOnForwardFailure work with fork-after-authentication ("ssh -f")
for protocol 2. ok markus@

14 years agoUpdate to 1.4.14b.
nicm [Mon, 12 Jul 2010 18:03:38 +0000 (18:03 +0000)]
Update to 1.4.14b.

From their change log:
 o Fix memory-leak of signal handler array with kqueue. [backport]
 o Make evutil_make_socket_nonblocking() leave any other flags alone.
 o Adjusted fcntl() retval comparison on evutil_make_socket_nonblocking().
 o Re-add event_siglcb; some old code _was_ still using it. :(
 o Fix a free(NULL) in min_heap.h
 o Clean up properly when adding a signal handler fails.

Also a local change to use an int rather than a long for fcntl().

ok guenther deraadt

14 years agodocument gcc4; requested by brad
jmc [Mon, 12 Jul 2010 16:28:29 +0000 (16:28 +0000)]
document gcc4; requested by brad

14 years agoksh.1: add ((, )), ]], and `name' to the list of reserved words for compounds
jmc [Mon, 12 Jul 2010 16:26:45 +0000 (16:26 +0000)]
ksh.1: add ((, )), ]], and `name' to the list of reserved words for compounds
sh.1: add ]] and `name' to the list; additionally document the commands
`select' (requested by Ross L Richardson) and [[

ok millert sobrado

14 years agomissed one next -> return, as noticed by marco@
espie [Mon, 12 Jul 2010 15:54:48 +0000 (15:54 +0000)]
missed one next -> return, as noticed by marco@

14 years agoMerge duplicate log messages into one log_warn().
bluhm [Mon, 12 Jul 2010 14:35:13 +0000 (14:35 +0000)]
Merge duplicate log messages into one log_warn().
ok claudio@

14 years agoneed to copy f() parameters, otherwise strange bugs with sub expressions
espie [Mon, 12 Jul 2010 09:31:52 +0000 (09:31 +0000)]
need to copy f() parameters, otherwise strange bugs with sub expressions
may occur (bug in locality in @_ ?)

Add missing bagage to State.pm to be able to interact directly with
Makewhatis, later.

14 years agoAfter forking, clear the 'thread joining this one' pointer in the child,
guenther [Mon, 12 Jul 2010 03:52:52 +0000 (03:52 +0000)]
After forking, clear the 'thread joining this one' pointer in the child,
as the child only has the one thread remaining.

ok marc@

14 years agoremove unused variable and unused static prototype.
chl [Sun, 11 Jul 2010 23:16:42 +0000 (23:16 +0000)]
remove unused variable and unused static prototype.

ok nicm@

14 years agoReturn the command client return code with MSG_EXIT now that MSG_ERROR and
nicm [Sun, 11 Jul 2010 17:06:45 +0000 (17:06 +0000)]
Return the command client return code with MSG_EXIT now that MSG_ERROR and
MSG_PRINT are unused.

New clients should be compatible with old tmux servers but vice versa may print
an error.

14 years agobcrypt_gensalt is already declared in pwd.h
dhill [Sun, 11 Jul 2010 15:53:12 +0000 (15:53 +0000)]
bcrypt_gensalt is already declared in pwd.h

"sure" deraadt@

14 years agotypo
miod [Sun, 11 Jul 2010 14:09:10 +0000 (14:09 +0000)]
typo

14 years agomove compare into LibSpec, it's not so specific to Signature after all
espie [Sun, 11 Jul 2010 10:05:18 +0000 (10:05 +0000)]
move compare into LibSpec, it's not so specific to Signature after all

14 years agosort SEE ALSO;
jmc [Sun, 11 Jul 2010 07:37:30 +0000 (07:37 +0000)]
sort SEE ALSO;

14 years agodisplay updateset name alongside collisionreport (turns out it's not
espie [Sun, 11 Jul 2010 07:27:25 +0000 (07:27 +0000)]
display updateset name alongside collisionreport (turns out it's not
always obvious)

14 years agotie back to bsd.port.mk, to make REVISION and EPOCH more obvious.
espie [Sun, 11 Jul 2010 07:25:10 +0000 (07:25 +0000)]
tie back to bsd.port.mk, to make REVISION and EPOCH more obvious.

14 years agothinko (found by Daniel Dickman)
espie [Sun, 11 Jul 2010 07:19:05 +0000 (07:19 +0000)]
thinko (found by Daniel Dickman)

14 years agoFixup structures to be correct size
jordan [Sat, 10 Jul 2010 22:24:14 +0000 (22:24 +0000)]
Fixup structures to be correct size
ok claudio@

14 years agoA process on the zombie list can have a NULL p_pgrp if it sleeps when
guenther [Sat, 10 Jul 2010 21:29:37 +0000 (21:29 +0000)]
A process on the zombie list can have a NULL p_pgrp if it sleeps when
grabbing allproclk in proc_zap(); skip such processes in sysctl(KERN_PROC*)
and handle the NULL pointer in ddb's ps.

ok tedu@

14 years agoReduce the delays a bit in the miibus read/write routines.
naddy [Sat, 10 Jul 2010 21:00:34 +0000 (21:00 +0000)]
Reduce the delays a bit in the miibus read/write routines.

Although with aditional 20us delays added just before returning
from miibus read/writes as required by the hardware.

Based on the Linux r8169 driver; from Brad.

14 years agoAdd a bootstrap loader, this allow u-boot to auto-boot from the media
drahn [Sat, 10 Jul 2010 20:55:14 +0000 (20:55 +0000)]
Add a bootstrap loader, this allow u-boot to auto-boot from the media
without modifying the u-boot environment.

14 years agoRemove debug print left in on last commit.
drahn [Sat, 10 Jul 2010 20:47:10 +0000 (20:47 +0000)]
Remove debug print left in on last commit.

14 years agofields are incorrectly named in structure descriptions
gilles [Sat, 10 Jul 2010 20:40:34 +0000 (20:40 +0000)]
fields are incorrectly named in structure descriptions

"of course" deraadt@

14 years agoNo mention of compat_freebsd anymore.
miod [Sat, 10 Jul 2010 19:40:06 +0000 (19:40 +0000)]
No mention of compat_freebsd anymore.

14 years agosun4e bits
miod [Sat, 10 Jul 2010 19:38:36 +0000 (19:38 +0000)]
sun4e bits

14 years agosun4e (i.e. SPARCengine 1e) support. This platform is a mix between sun4 and
miod [Sat, 10 Jul 2010 19:32:20 +0000 (19:32 +0000)]
sun4e (i.e. SPARCengine 1e) support. This platform is a mix between sun4 and
sun4c, as it has a sun4c OpenPROM but a sun4 8KB pagesize. VME devices are
not supported yet.
ok deraadt@

14 years agoand cdboot must be linked with -N still, too. even more curious
deraadt [Sat, 10 Jul 2010 17:09:15 +0000 (17:09 +0000)]
and cdboot must be linked with -N still, too.  even more curious

14 years agoactually, pxeboot must remain linked with -N
deraadt [Sat, 10 Jul 2010 17:04:22 +0000 (17:04 +0000)]
actually, pxeboot must remain linked with -N

14 years agolink bootblocks with -Z instead of -N, to ensure that the file offset and
deraadt [Sat, 10 Jul 2010 16:11:43 +0000 (16:11 +0000)]
link bootblocks with -Z instead of -N, to ensure that the file offset and
addresses line up correctly.  Newer i386 gcc4 - for some reason -- is
prompting the linker with -N to not do this alignment which it used to do
(perhaps because the alignment of all .o's have dropped to very small
amounts??).  Using -Z wastes around a page of bootblock space which matters
even more on the small install media, so this definately needs revisiting.
Workaround from toby

14 years agoRetry requests when the btree is busy. Without this, clients will just hang
martinh [Sat, 10 Jul 2010 14:27:15 +0000 (14:27 +0000)]
Retry requests when the btree is busy. Without this, clients will just hang
waiting for a response if the btree was being reopened when the request was
received.

14 years agoWhen a file is created, initialize its ``busy cycles'' counter to zero.
ratchov [Sat, 10 Jul 2010 12:32:45 +0000 (12:32 +0000)]
When a file is created, initialize its ``busy cycles'' counter to zero.

14 years agoIn midi server mode, don't terminate the midi thru box if there are
ratchov [Sat, 10 Jul 2010 12:28:17 +0000 (12:28 +0000)]
In midi server mode, don't terminate the midi thru box if there are
no connections anymore.

14 years agofix parameter handling: don't try to open a ``default'' midi port
ratchov [Sat, 10 Jul 2010 11:56:28 +0000 (11:56 +0000)]
fix parameter handling: don't try to open a ``default'' midi port
if no files are given on the command line

14 years agotweak;
jmc [Sat, 10 Jul 2010 10:49:15 +0000 (10:49 +0000)]
tweak;

14 years agodocument EPOCH/REVISION
espie [Sat, 10 Jul 2010 09:12:02 +0000 (09:12 +0000)]
document EPOCH/REVISION

14 years agowhitespace cleanup from Brad, no change with diff -w
sthen [Sat, 10 Jul 2010 07:59:33 +0000 (07:59 +0000)]
whitespace cleanup from Brad, no change with diff -w

14 years agoVerify _ADR bdf is correct (some systems return 0xffff)
jordan [Sat, 10 Jul 2010 04:59:55 +0000 (04:59 +0000)]
Verify _ADR bdf is correct (some systems return 0xffff)
Works on Dell x9xx and Studio 1555

14 years agoEliminate two of the remaining three instances of drivers passing
matthew [Sat, 10 Jul 2010 03:06:51 +0000 (03:06 +0000)]
Eliminate two of the remaining three instances of drivers passing
their own buf to physio(9).

compiler tested by nick@ and miod@; ok miod@
general approval thib@, deraadt@

14 years agoEliminate some unnecessary #include lines.
matthew [Sat, 10 Jul 2010 02:56:16 +0000 (02:56 +0000)]
Eliminate some unnecessary #include lines.

"sure" marco@

14 years agoLimit SCIOCCOMMAND and ATAIOCCOMMAND requests to transferring MAXPHYS
matthew [Sat, 10 Jul 2010 02:52:38 +0000 (02:52 +0000)]
Limit SCIOCCOMMAND and ATAIOCCOMMAND requests to transferring MAXPHYS
bytes worth of data.  Also, pass M_CANFAIL to malloc(9) and handle
allocation failures by returning ENOMEM.

Prevents user space from causing a kernel panic by passing
ridiculously large datalen values.

ok dlg@, krw@

14 years agorewrite the polling codepath in mpii.
dlg [Fri, 9 Jul 2010 22:33:21 +0000 (22:33 +0000)]
rewrite the polling codepath in mpii.

it basically has the poll code swap the ccb completion path out for
something that clears the flag the poll code is looping on. it just
runs the interrupt handler to try and get that ccb back off the
hardware.

the reply queue is now protected by a semaphore, meaning that if
one cpu is already pulling stuff off the chip, another cpu may try
to also pull stuff off, but will return immediately cos the first
cpu will end up doing the work it wants anyway.

i did it this way so the ccb completion paths wouldnt be run while
a mutex was held, and to avoid reordering of ccb completions and
updates to the reply queue consumer that could be caused by allowing
concurrent runs of the interrupt handler to step on each other.

there is a big change here in that polling doesnt time out anymore.
it keeps trying until it succeeds, or it keeps trying. i think it
is dangerous to complete a ccb until th echip has returned it.

the only safe alternative is to do a full reset of the chip, but
that is a massive amount of work.

reviewed by mikeb
ok krw@
testing by me on real and logical disks

14 years agosync
deraadt [Fri, 9 Jul 2010 22:27:36 +0000 (22:27 +0000)]
sync

14 years agoneed sysctl.h to compile with GPROF; Luis Henriques
deraadt [Fri, 9 Jul 2010 20:30:48 +0000 (20:30 +0000)]
need sysctl.h to compile with GPROF; Luis Henriques

14 years agoMake cvs and rcs consistent in what characters they'll accept in
naddy [Fri, 9 Jul 2010 20:07:05 +0000 (20:07 +0000)]
Make cvs and rcs consistent in what characters they'll accept in
symbol/tag names: require the first character to be a letter and
allow any printable character (except the RCS specials $,.:;@)
afterwards.  This matches GNU cvs.

ok xsa@ zinovik@ stsp@

14 years agoWe need to preserve %eax, %ecx and %edx since GCC 4.x might use these
kettenis [Fri, 9 Jul 2010 18:44:10 +0000 (18:44 +0000)]
We need to preserve %eax, %ecx and %edx since GCC 4.x might use these
registers in the prologue to realign the stack.

ok marco@, deraadt@, guenther@

14 years agowhen opencvs is started with trace option `-t' it must print function it
zinovik [Fri, 9 Jul 2010 18:42:14 +0000 (18:42 +0000)]
when opencvs is started with trace option `-t' it must print function it
enters, even if it was executed simultaneously with `-n' (no exec option).
cvs_unedit_local() does not conform that behaviour, so lets make it behave
like all other functions.

ok xsa@, stsp@

14 years agoun-terminated string found when comparing to i386 boot code. These
deraadt [Fri, 9 Jul 2010 17:36:38 +0000 (17:36 +0000)]
un-terminated string found when comparing to i386 boot code.  These
should be 100% identical.
ok toby

14 years agoFix cross build problem with cpp, marex pointed out that it wasn't committed.
drahn [Fri, 9 Jul 2010 17:36:08 +0000 (17:36 +0000)]
Fix cross build problem with cpp, marex pointed out that it wasn't committed.

14 years agoAdd support for using IPsec in multiple rdomains.
reyk [Fri, 9 Jul 2010 16:58:06 +0000 (16:58 +0000)]
Add support for using IPsec in multiple rdomains.

This allows to run isakmpd/iked/ipsecctl in multiple rdomains
independently (with "route exec"); the kernel will pickup the rdomain
from the process context of the pfkey socket and load the flows and
SAs into the matching rdomain encap routing table.  The network stack
also needs to pass the rdomain to the ipsec stack to lookup the
correct rdomain that belongs to an interface/mbuf/... You can now run
individual IPsec configs per rdomain or create IPsec VPNs between
multiple rdomains on the same machine ;).  Note that a primary enc(4)
in addition to enc0 interface is required per rdomain, eg. enc1 rdomain 1.

Test by some people, mostly on existing "rdomain 0" setups.  Was in
snaps for some days and people didn't complain.

ok claudio@ naddy@

14 years agoAdd the rtableid to struct rttimer and therefor make it available to
claudio [Fri, 9 Jul 2010 15:44:20 +0000 (15:44 +0000)]
Add the rtableid to struct rttimer and therefor  make it available to
the callback functions. This fixes a problem where dynamic routes in
different tables would not get deleted because the callback was doing
the remove on the wrong table.
OK henning@

14 years agoospfd is a AF_INET only routing daemon so limit the routing socket to that
claudio [Fri, 9 Jul 2010 15:38:28 +0000 (15:38 +0000)]
ospfd is a AF_INET only routing daemon so limit the routing socket to that
af. This is possible since this filtering no longer blocks RTM_IFINFO.
OK henning@

14 years agoWhen protocol filtering is used on the rtsocket filter only messages that
claudio [Fri, 9 Jul 2010 15:36:54 +0000 (15:36 +0000)]
When protocol filtering is used on the rtsocket filter only messages that
actually have a protocol/address family set. Messages like RTM_IFINFO or
RTM_IFANNOUNCE are family independent and should not be filtered but
it is possible to use a ROUTE_MSGFILTER to filter these messages.
This allows to reduce the messages sent to AF_INET or AF_INET6 only daemons.
OK henning@, deraadt@

14 years agoinstead of saying we're using the all the states in the table when
dlg [Fri, 9 Jul 2010 13:09:34 +0000 (13:09 +0000)]
instead of saying we're using the all the states in the table when
sending a bulk update, only say we're using the states referenced
by the next and last pointers into the table. this means the pf
timeout thread can keep trimming states during a bulk update.

14 years agofix pkg_mklocatedb
espie [Fri, 9 Jul 2010 12:42:43 +0000 (12:42 +0000)]
fix pkg_mklocatedb

14 years agoDo not try to originate an intra-area-prefix-lsa if an interface
bluhm [Fri, 9 Jul 2010 12:39:46 +0000 (12:39 +0000)]
Do not try to originate an intra-area-prefix-lsa if an interface
address changes that does not belong to any area.  This fixes an
ospf6d crash.  To not search for the area in orig_intra_lsa_net(),
just pass the area as function parameter.
ok claudio@

14 years agoinetname is only called when !nflag. There is no need to again check for
dhill [Fri, 9 Jul 2010 12:27:09 +0000 (12:27 +0000)]
inetname is only called when !nflag.  There is no need to again check for
!nflag within inetname().

from Alexandr Shadchin

ok claudio@

14 years agoClear the noise
jordan [Fri, 9 Jul 2010 12:27:02 +0000 (12:27 +0000)]
Clear the noise

14 years agouse macro defines for magical numbers, instead hardcoded values
zinovik [Fri, 9 Jul 2010 12:17:34 +0000 (12:17 +0000)]
use macro defines for magical numbers, instead hardcoded values

ok krw@

14 years ago-#if 1 || defined(PFSYNC_DEBUG)
dlg [Fri, 9 Jul 2010 12:07:21 +0000 (12:07 +0000)]
-#if 1 || defined(PFSYNC_DEBUG)
+#if defined(PFSYNC_DEBUG)

14 years agothe current code doesnt detect when its filled a bulk packet so it
dlg [Fri, 9 Jul 2010 11:16:45 +0000 (11:16 +0000)]
the current code doesnt detect when its filled a bulk packet so it
keeps on building them, which means that it floods the tx ring on
the pfsync interface rather than actually sending many packets.

this change correctly calculates when we've filled a bulk update
packet.

many thanks to david@ for making me go chase this. ive been wondering
why the state count on my firewalls has had such a discrepency for
a long time now.

14 years agosending of bulk updates has been very broken since r1.124.
dlg [Fri, 9 Jul 2010 09:01:32 +0000 (09:01 +0000)]
sending of bulk updates has been very broken since r1.124.
pfsync_bulk_start sets up a bulk transfer by doing this:

sc->sc_bulk_next = TAILQ_FIRST(&state_list);
sc->sc_bulk_last = sc->sc_bulk_next;

and then calls pfsync_bulk_update which loops over state_list like
this:

        st = sc->sc_bulk_next;
        while (st != sc->sc_bulk_last) {

basically bulk updates never got sent.

this diff handles jjs problem by not attempting to send a bulk
update when the state table is empty and fixed the loop to break
at the right times.

reported by david@ who's initial tests are positive.
reviewed by mcbride@
tested in production at work.

14 years agoModified to make sure that pipex_mppe_ouput() keeps the packet 32bit
yasuoka [Fri, 9 Jul 2010 08:36:31 +0000 (08:36 +0000)]
Modified to make sure that pipex_mppe_ouput() keeps the packet 32bit
aligned.

OK claudio@

14 years agosync from mirrors.dat
sthen [Fri, 9 Jul 2010 08:28:03 +0000 (08:28 +0000)]
sync from mirrors.dat

14 years agodon't print directly, use an interface that's similar enough to pkg_add
espie [Fri, 9 Jul 2010 08:12:49 +0000 (08:12 +0000)]
don't print directly, use an interface that's similar enough to pkg_add
print framework so that we'll be able to integrate with it better.
ok millert@

14 years agokill redundant docheck() function; ok deraadt@
otto [Fri, 9 Jul 2010 08:06:37 +0000 (08:06 +0000)]
kill redundant docheck() function; ok deraadt@

14 years ago3 more JEDEC JC-42.4 sensors are the market
deraadt [Fri, 9 Jul 2010 08:05:45 +0000 (08:05 +0000)]
3 more JEDEC JC-42.4 sensors are the market

14 years agosync
deraadt [Fri, 9 Jul 2010 07:58:49 +0000 (07:58 +0000)]
sync

14 years agogcc4 is a bloated pig, so a few things must fall off the floppy
deraadt [Fri, 9 Jul 2010 07:50:33 +0000 (07:50 +0000)]
gcc4 is a bloated pig, so a few things must fall off the floppy
install media until we can shrink something else.  decisions on
what has to go made by jsg and I.  (subject to change..)

14 years agoUpdate for some of the newer chipsets supported by bge(4). From Brad.
sthen [Fri, 9 Jul 2010 07:10:55 +0000 (07:10 +0000)]
Update for some of the newer chipsets supported by bge(4).  From Brad.

14 years agoAvoid crashes by using correct types for block numbers, which can grow
otto [Fri, 9 Jul 2010 06:41:17 +0000 (06:41 +0000)]
Avoid crashes by using correct types for block numbers, which can grow
large on very large filesystems; reported by Benny Lofgren; partly
from FreeBSD. ok deraadt@ beck@ millert@

14 years agogood enough sync
deraadt [Fri, 9 Jul 2010 03:05:21 +0000 (03:05 +0000)]
good enough sync

14 years agonew sentence, new line;
jmc [Fri, 9 Jul 2010 01:24:45 +0000 (01:24 +0000)]
new sentence, new line;

14 years ago- Setup proper mbuf pool watermarks for BCM5717 / BCM57765 chipsets.
sthen [Fri, 9 Jul 2010 00:04:42 +0000 (00:04 +0000)]
- Setup proper mbuf pool watermarks for BCM5717 / BCM57765 chipsets.
- Disable initiation of multiple DMA reads for BCM5717 chipset
  to workaround errata.
- Performance tweak for BCM5785 chipset.
- Correct the return ring count used for BCM5717 / BCM57765 chipsets.
- Fix fiber media detection for BCM5717 chipsets as they use a MII-based
  fiber PHY.

From Brad, based on info gleaned from the Linux tg3 driver.
Should only affect the new chipsets, but tested for regressions
anyway (on BCM5701, BCM5703, BCM5704C, BCM5705M Alt, BCM5722, BCM5751,
BCM5751M, BCM5780) by Brad, weerd@, myself and Denis Doroshenko.

14 years agodont count requeued io as outstanding io. there is a 1:1 mapping
dlg [Thu, 8 Jul 2010 23:18:55 +0000 (23:18 +0000)]
dont count requeued io as outstanding io. there is a 1:1 mapping
between calls to bufq_enqueue and bufq_done calls, but bufq_requeue
can be called multiple times on an io in between the enqueue and
done. if you keep incrementing outstanding you'll never stop sleeping
for suspend.

bufq_requeue can be called via interrupts, so we cannot sleep
there.

the problems fixed by this diff were never hit cos the adapters
people are testing/running with do not cause bufq_requeue to be
called.

thanks to thib@ for helping me understand the code better
ok thib@ deraadt@ kettenis@ tedu@

14 years agosync
deraadt [Thu, 8 Jul 2010 22:58:08 +0000 (22:58 +0000)]
sync

14 years agoFirst shot at documenting new www/mozilla MODULE, with feedback from jmc@.
landry [Thu, 8 Jul 2010 21:14:57 +0000 (21:14 +0000)]
First shot at documenting new www/mozilla MODULE, with feedback from jmc@.
ok jmc@.

14 years agoUpdate sudo to version 1.7.2p8
millert [Thu, 8 Jul 2010 21:11:31 +0000 (21:11 +0000)]
Update sudo to version 1.7.2p8

14 years agoAdd mapping for ACPI device to PCI bdf (match autoconf tree)
jordan [Thu, 8 Jul 2010 20:56:31 +0000 (20:56 +0000)]
Add mapping for ACPI device to PCI bdf (match autoconf tree)
Simplify resource parsing function to use buffer argument
Convert namespace linked lists to use queue macros
ok marco@, deraadt@

14 years agoSwitch hppa, i386 and powerpc to gcc4. Bump libc major to handle ABI corner
kettenis [Thu, 8 Jul 2010 20:25:40 +0000 (20:25 +0000)]
Switch hppa, i386 and powerpc to gcc4.  Bump libc major to handle ABI corner
cases fixed in gcc4.

14 years agoLet traceroute parse extended ICMP messages as defined by RFC 4884.
claudio [Thu, 8 Jul 2010 20:23:03 +0000 (20:23 +0000)]
Let traceroute parse extended ICMP messages as defined by RFC 4884.
Currently only the MPLS label can be shown (RFC 4950). The extended
information is only printed if either -x or -v was used.
Initialy based on a port from NetBSD done by dhill@ but mostly rewritten now.
OK deraadt@, dhill@

14 years agoAdd definitions for ICMP extended headers available for some ICMP messages
claudio [Thu, 8 Jul 2010 20:20:11 +0000 (20:20 +0000)]
Add definitions for ICMP extended headers available for some ICMP messages
like time exceeded messages. This will allow to add additional information
to those messges like the incomming MPLS label.
OK deraadt@, dhill@

14 years agouse config_activate_children to activate children that might need it
deraadt [Thu, 8 Jul 2010 20:18:32 +0000 (20:18 +0000)]
use config_activate_children to activate children that might need it

14 years agouse config_activate_children to get down to the isa bus activation code
deraadt [Thu, 8 Jul 2010 20:17:54 +0000 (20:17 +0000)]
use config_activate_children to get down to the isa bus activation code

14 years agoDevices which don't have read or write functionality should not return
deraadt [Thu, 8 Jul 2010 20:15:01 +0000 (20:15 +0000)]
Devices which don't have read or write functionality should not return
enodev to poll, because this returns an errno of 19 in revents.  Oops.
Use seltrue where needed, and use a new selfalse function for those which
don't know if the next op will be non-blocking
Mostly discussed with guenther and miod

14 years agoremove some unused netinet6 code
jsg [Thu, 8 Jul 2010 19:42:46 +0000 (19:42 +0000)]
remove some unused netinet6 code
ok stsp@ henning@ claudio@

14 years agoUse correct alignment for scrub max-mss. Based on a diff from deraadt.
sthen [Thu, 8 Jul 2010 19:30:16 +0000 (19:30 +0000)]
Use correct alignment for scrub max-mss. Based on a diff from deraadt.
"that looks about right even though the offset calculation is pure
horror" claudio@, ok deraadt@

14 years agoFor DVACT_SUSPEND and DVACT_RESUME we should always return success.
deraadt [Thu, 8 Jul 2010 19:29:25 +0000 (19:29 +0000)]
For DVACT_SUSPEND and DVACT_RESUME we should always return success.

14 years agomake the buffer big enough to hold a 1500 byte packet. Workaround (mtu
ckuethe [Thu, 8 Jul 2010 18:22:01 +0000 (18:22 +0000)]
make the buffer big enough to hold a 1500 byte packet. Workaround (mtu
decrease) by me, fix from armani@ (who seems to be away right now). ok mk@

14 years agouse more natural English in messages; ok espie@
sthen [Thu, 8 Jul 2010 10:16:13 +0000 (10:16 +0000)]
use more natural English in messages; ok espie@

14 years agoRevert part of previous.
thib [Thu, 8 Jul 2010 09:46:50 +0000 (09:46 +0000)]
Revert part of previous.

The splvm protection is needed after all, as we are walking the list
of registered crypto drivers and doing that unprotected is unwise.

Pointed out by kettenis@

14 years agoUpdate after kroute.c change. As a goody print the route priority.
claudio [Thu, 8 Jul 2010 09:42:32 +0000 (09:42 +0000)]
Update after kroute.c change. As a goody print the route priority.
OK michele@

14 years agoKill more code in kroute.c that is unneeded in ldpd. ldpd is a bit strange
claudio [Thu, 8 Jul 2010 09:41:05 +0000 (09:41 +0000)]
Kill more code in kroute.c that is unneeded in ldpd. ldpd is a bit strange
since it does not care that much about reachability of routes. The idea is
to have diverse LSP in the kernel and the kernel should then decide which
path should be used.
OK michele@

14 years agominor tweak from Brad; add the word 'chips' after a list of devices
sthen [Thu, 8 Jul 2010 09:19:11 +0000 (09:19 +0000)]
minor tweak from Brad; add the word 'chips' after a list of devices
used in a sentence

14 years agoRemove references to kvm.
lum [Thu, 8 Jul 2010 08:47:07 +0000 (08:47 +0000)]
Remove references to kvm.
ok claudio@

14 years agopipex didn't work on output. Fixed following problems:
yasuoka [Thu, 8 Jul 2010 08:40:29 +0000 (08:40 +0000)]
pipex didn't work on output.  Fixed following problems:
 - pipex failed to lookup the radix tree because address and netmask
   were not initialized.
 - pipex used wrong place as a ip header because it didn't adjust
   32bit address family header that are added at tun_output.

14 years agobetter op buf handling. Instead of mallocing a new buffer for every
thib [Thu, 8 Jul 2010 08:15:18 +0000 (08:15 +0000)]
better op buf handling. Instead of mallocing a new buffer for every
encdec operation, keep a PAGE_SIZE (or not, depending if we fail during
setup) buf around; if the crd_len is larger then the current buf, re allocate
a new one that fits the data, and keep it around.

14 years agoMove pool initialization to init_crypto and zap the crypto_pool_initialized
thib [Thu, 8 Jul 2010 08:12:48 +0000 (08:12 +0000)]
Move pool initialization to init_crypto and zap the crypto_pool_initialized
variable. This way we don't have to check if the pool are initialized every
time we do a crypto_getreq().

Move splvm lower as it isnt need all through crypto_newsession().

tiny KNF nit.

From mikeb

OK deraadt@