]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
17 years agoRev. 1.44 of this file didn't introduce a right solution,
yar [Tue, 25 Jul 2006 22:39:57 +0000 (22:39 +0000)]
Rev. 1.44 of this file didn't introduce a right solution,
but we don't seem to have one yet, so just add an XXX comment on
passing rux_runtime to bintime2timeval() wrongly.

Spotted by: gcc(1) (warning)

17 years agoFix 32-bit integer math on 64-bit processor. Just use int32_t(!) instead
jkim [Tue, 25 Jul 2006 22:20:05 +0000 (22:20 +0000)]
Fix 32-bit integer math on 64-bit processor.  Just use int32_t(!) instead
of incorrect and machine-dependent integer math.  Now we can encrypt a file
on an i386 and decrypt it on an amd64, and vice versa.

Submitted by: Andrew Heybey < ath at niksun dot com >

17 years agoMFC noted: minidumps.
bmah [Tue, 25 Jul 2006 21:48:28 +0000 (21:48 +0000)]
MFC noted: minidumps.

17 years agoNew release notes: acpi_thermal(4) temperature overrides, interface
bmah [Tue, 25 Jul 2006 21:30:58 +0000 (21:30 +0000)]
New release notes:  acpi_thermal(4) temperature overrides, interface
groups, mount(8) late/-l, FILE 4.17.

Modified release notes: Improved padlock(4) crypto support (+MFC),
OpenBSM 1.0alpha7.

17 years agoRemove redundant check committed by accident.
pjd [Tue, 25 Jul 2006 20:00:55 +0000 (20:00 +0000)]
Remove redundant check committed by accident.

17 years agoAvoid memory allocations when the given address is already 16 bytes aligned.
pjd [Tue, 25 Jul 2006 19:32:58 +0000 (19:32 +0000)]
Avoid memory allocations when the given address is already 16 bytes aligned.
Such an address can be used directly in padlock's AES.
This improves speed of geli(8) significantly:

# sysctl kern.geom.zero.clear=0
# geli onetime -s 4096 gzero
# dd if=/dev/gzero.eli of=/dev/null bs=1m count=1000

Before: 113MB/s
After: 203MB/s

BTW. If sector size is set to 128kB, I can read at 276MB/s :)

17 years agoModify PADLOCK_ALIGN() macro, so when the given address is already 16 bytes
pjd [Tue, 25 Jul 2006 19:06:54 +0000 (19:06 +0000)]
Modify PADLOCK_ALIGN() macro, so when the given address is already 16 bytes
aligned, it will be used directly, not 'address + 16'.

17 years agoStyle fixes.
pjd [Tue, 25 Jul 2006 19:04:26 +0000 (19:04 +0000)]
Style fixes.

17 years agoDe-uglify messages from the ipfw script.
yar [Tue, 25 Jul 2006 17:28:18 +0000 (17:28 +0000)]
De-uglify messages from the ipfw script.

17 years agoUse more rc.subr(8) features.
yar [Tue, 25 Jul 2006 17:25:44 +0000 (17:25 +0000)]
Use more rc.subr(8) features.
Skip useless work when ${natd_interface} is unset.
Double quote user-supplied vars unless there is a reason not to.

17 years agoAvoid extra runs of test(1) by using its built-in logical operations.
yar [Tue, 25 Jul 2006 17:20:22 +0000 (17:20 +0000)]
Avoid extra runs of test(1) by using its built-in logical operations.

17 years agoDon't load geom_md.ko if there is no need to.
pjd [Tue, 25 Jul 2006 17:19:00 +0000 (17:19 +0000)]
Don't load geom_md.ko if there is no need to.

17 years agoSync comments around _doit with the code.
yar [Tue, 25 Jul 2006 17:16:48 +0000 (17:16 +0000)]
Sync comments around _doit with the code.

17 years agoWe notify of failures with err or warn.
yar [Tue, 25 Jul 2006 17:14:38 +0000 (17:14 +0000)]
We notify of failures with err or warn.
Messages should end in full stop unless
that can lead to confusion.

17 years agoAdd diagnostics to load_kld().
yar [Tue, 25 Jul 2006 17:10:35 +0000 (17:10 +0000)]
Add diagnostics to load_kld().

17 years agoNew release notes: snd_emu10kx(4), nfe(4), stge(4), enc(4) (+MFC),
bmah [Tue, 25 Jul 2006 16:01:46 +0000 (16:01 +0000)]
New release notes:  snd_emu10kx(4), nfe(4), stge(4), enc(4) (+MFC),
sysinstall(8) title (+MFC).

17 years agoIn udf_find_partmaps(), when we find a type 1 partition map, we have to
yar [Tue, 25 Jul 2006 14:15:50 +0000 (14:15 +0000)]
In udf_find_partmaps(), when we find a type 1 partition map, we have to
skip the actual type 1 length (6 bytes). With this change, it is now possible
to correctly spot the VAT partition map in certain discs.

Submitted by: Pedro Martelletto <pedro@ambientworks.net>

17 years agoo Add NetBSD 3.0.1 and DragonFlyBSD 1.6.0.
maxim [Tue, 25 Jul 2006 13:31:00 +0000 (13:31 +0000)]
o Add NetBSD 3.0.1 and DragonFlyBSD 1.6.0.

17 years ago1. Don't override underscore version of aio_suspend(), system(),
davidxu [Tue, 25 Jul 2006 12:50:05 +0000 (12:50 +0000)]
1. Don't override underscore version of aio_suspend(), system(),
   wait(), waitpid() and usleep(), they are internal versions and
   should not be cancellation points.
2. Make wait3() as a cancellation point.
3. Move raise() and pause() into file thr_sig.c.
4. Add functions _sigsuspend, _sigwait, _sigtimedwait and _sigwaitinfo,
   remove SIGCANCEL bit in wait-set for those functions, the signal is
   used internally to implement thread cancellation.

17 years agoSpecify correct argument range for tag/untag keywords.
oleg [Tue, 25 Jul 2006 11:07:31 +0000 (11:07 +0000)]
Specify correct argument range for tag/untag keywords.

Approved by: glebius (mentor)

17 years agosingular -> plural and a typo fix.
brueffer [Tue, 25 Jul 2006 05:20:06 +0000 (05:20 +0000)]
singular -> plural and a typo fix.

17 years agoAutogenerate hardware notes for stge(4).
brueffer [Tue, 25 Jul 2006 05:13:53 +0000 (05:13 +0000)]
Autogenerate hardware notes for stge(4).

17 years agoFix printing of negative decimal values in Kelvin to Celsius conversion.
njl [Tue, 25 Jul 2006 02:28:43 +0000 (02:28 +0000)]
Fix printing of negative decimal values in Kelvin to Celsius conversion.

MFC after: 3 days

17 years agoAdd support for overriding the values for _CRT, _HOT, and _PSV via sysctl.
njl [Tue, 25 Jul 2006 02:27:38 +0000 (02:27 +0000)]
Add support for overriding the values for _CRT, _HOT, and _PSV via sysctl.
Prevent casual modification by requiring hw.acpi.thermal.user_override to
be set first.  Fix printing of negative temperatures in the K->C conversion.
Document the remaining thermal sysctls.

MFC after: 3 days

17 years agoAdd uipc_sockbuf.c to standard files list; accidentally missed in earlier
rwatson [Tue, 25 Jul 2006 02:15:28 +0000 (02:15 +0000)]
Add uipc_sockbuf.c to standard files list; accidentally missed in earlier
commit.

Spotted by: tinderbox

17 years agoCleanup.
kevlo [Tue, 25 Jul 2006 01:08:41 +0000 (01:08 +0000)]
Cleanup.
Approved by: cognet

17 years agoAdd stge(4) to the list of drivers supported by GENERIC kernel.
yongari [Tue, 25 Jul 2006 01:06:32 +0000 (01:06 +0000)]
Add stge(4) to the list of drivers supported by GENERIC kernel.

17 years agoAdd stge(4) to the list of supported network interface.
yongari [Tue, 25 Jul 2006 01:03:55 +0000 (01:03 +0000)]
Add stge(4) to the list of supported network interface.

17 years agoAdd an entry for the stge(4) module.
yongari [Tue, 25 Jul 2006 01:02:22 +0000 (01:02 +0000)]
Add an entry for the stge(4) module.
While I'm here remove a stale wx(4) entry which was removed 4 years,
9 months ago.

17 years agobus_alloc_resource_any is actually defined in the
mjacob [Tue, 25 Jul 2006 01:01:09 +0000 (01:01 +0000)]
bus_alloc_resource_any is actually defined in the
RELENG_4 branch, so there's no need to have a compilation
difference here any more.

17 years agoWhen probing to attach the CAM functionality, check against
mjacob [Tue, 25 Jul 2006 00:59:54 +0000 (00:59 +0000)]
When probing to attach the CAM functionality, check against
desired role configuration instead of existing role. This gets
us out of the mess where we configured a role of NONE (or were
LAN only, for example), but didn't continue to attach the CAM
module (because we had neither initiator nor target role
set). Unfortunately, the code that rewrites NVRAM to match
actual to desired role only works if the CAM module attaches.

MFC after: 2 weeks

17 years agoAdd stge(4) to the list of device drivers use the miibus interface.
yongari [Tue, 25 Jul 2006 00:59:39 +0000 (00:59 +0000)]
Add stge(4) to the list of device drivers use the miibus interface.

17 years agoAdd stge(4) to the list of drivers supporting ALTQ.
yongari [Tue, 25 Jul 2006 00:58:00 +0000 (00:58 +0000)]
Add stge(4) to the list of drivers supporting ALTQ.

17 years agoAdd stge(4) to the list of devices supporting full VLAN processing
yongari [Tue, 25 Jul 2006 00:56:47 +0000 (00:56 +0000)]
Add stge(4) to the list of devices supporting full VLAN processing
in the hardware.

17 years agoAdd stge(4) to the list of devices supporting polling.
yongari [Tue, 25 Jul 2006 00:54:58 +0000 (00:54 +0000)]
Add stge(4) to the list of devices supporting polling.

17 years agoHook up state(4) to the build.
yongari [Tue, 25 Jul 2006 00:53:14 +0000 (00:53 +0000)]
Hook up state(4) to the build.

17 years agoAdd a man page for stge(4).
yongari [Tue, 25 Jul 2006 00:51:38 +0000 (00:51 +0000)]
Add a man page for stge(4).

17 years agoHook up stge(4) to the build.
yongari [Tue, 25 Jul 2006 00:45:55 +0000 (00:45 +0000)]
Hook up stge(4) to the build.

17 years agoAdd stge(4), a driver for Sundance/Tamarack TC9021 Gigabit Ethernet
yongari [Tue, 25 Jul 2006 00:37:09 +0000 (00:37 +0000)]
Add stge(4), a driver for Sundance/Tamarack TC9021 Gigabit Ethernet
controller ported from NetBSD. It supports the following Gigabit
Ethernet adapters.
o Antares Microsystems Gigabit Ethernet
o ASUS NX1101 Gigabit Ethernet
o D-Link DL-4000 Gigabit Ethernet
o IC Plus IP1000A Gigabit Ethernet
o Sundance ST-2021 Gigabit Ethernet
o Sundance ST-2023 Gigabit Ethernet
o Sundance TC9021 Gigabit Ethernet
o Tamarack TC9021 Gigabit Ethernet
The IP1000A Gigabit Ethernet is also found on some motherboards
(LOM) from ABIT.

Unlike NetBSD stge(4) it does not require promiscuous mode operation
to revice packet and it supports all hardware features(TCP/UDP/IP
checksum offload, VLAN tag stripping/insertion features and JUMBO
frame) and polling(4).
Due to lack of hardware, hardwares that have TBI trantransceivers
were not tested at all.

Special thanks to wpaul who provided valauble datasheet for the
controller and helped to debug jumbo frame related issues. Whitout
his datasheet I would have spent many hours to debug this chip.

Tested on: i386, sparc64

17 years agoRevert previous commit. Spinlocks hold interrupts disabled, so
marcel [Tue, 25 Jul 2006 00:23:54 +0000 (00:23 +0000)]
Revert previous commit. Spinlocks hold interrupts disabled, so
preemption is not possible.

Pointed out by: jhb@

17 years agoConnect gentbi, ip1000phy to the build.
yongari [Tue, 25 Jul 2006 00:20:11 +0000 (00:20 +0000)]
Connect gentbi, ip1000phy to the build.

17 years agoAdd ip1000 PHY driver for IC Plus IP1000A integrated PHY.
yongari [Tue, 25 Jul 2006 00:16:09 +0000 (00:16 +0000)]
Add ip1000 PHY driver for IC Plus IP1000A integrated PHY.

17 years agoAdd IC Plus IP1000A integrated PHY id.
yongari [Tue, 25 Jul 2006 00:14:00 +0000 (00:14 +0000)]
Add IC Plus IP1000A integrated PHY id.

17 years agoAdd gentbi, a generic TBI(teb-bit interface) PHY driver ported
yongari [Tue, 25 Jul 2006 00:08:32 +0000 (00:08 +0000)]
Add gentbi, a generic TBI(teb-bit interface) PHY driver ported
from NetBSD.

17 years agoIf we have multiple interrupt resources, like for Z8530 clones on the
marcel [Mon, 24 Jul 2006 22:25:16 +0000 (22:25 +0000)]
If we have multiple interrupt resources, like for Z8530 clones on the
mac-io bus, we cannot setup FAST interrupt handlers. This because we
use spinlocks to protect the hardware and all interrupt resources are
assigned the same interrupt handler. When the interrupt handler is
invoked for interrupt X, it could be preempted for interrupt Y while
it was holding the lock (where X and Y are the interrupt resources
corresponding a single instance of this driver). This is a deadlock.
By only using a MPSAFE handler in that case we prevent preemption.

17 years agoOnly use -mno-apcs-frame if DDB is not in the kernel, as it prevent the
cognet [Mon, 24 Jul 2006 22:09:47 +0000 (22:09 +0000)]
Only use -mno-apcs-frame if DDB is not in the kernel, as it prevent the
backtraces from working.

MFC After: 3 days

17 years agoAdd a few more devices to the supported list. Specifically, the PHAROS and
imp [Mon, 24 Jul 2006 19:47:00 +0000 (19:47 +0000)]
Add a few more devices to the supported list.  Specifically, the PHAROS and
the silly dongly I just bought at Radio Shack.

17 years agoRemove non-socket buffer routines from uipc_sockbuf.c, and socket buffer
rwatson [Mon, 24 Jul 2006 16:21:31 +0000 (16:21 +0000)]
Remove non-socket buffer routines from uipc_sockbuf.c, and socket buffer
specific routines from uipc_socket2.c following repo-copy.  We might
rethink the location of one or two at some point, but the division was
relatively clean.  uipc_sockbuf.c is now the home of routines that
manipulate socket buffers.

17 years agoRather than calling mircotime() in catchpacket(), make catchpacket()
dwmalone [Mon, 24 Jul 2006 15:42:04 +0000 (15:42 +0000)]
Rather than calling mircotime() in catchpacket(), make catchpacket()
take a timeval indicating when the packet was captured. Move
microtime() to the calling functions and grab the timestamp as soon
as we know that we're going to call catchpacket at least once.

This means that we call microtime() once per matched packet, as
opposed to once per matched packet per bpf listener. It also means
that we return the same timestamp to all bpf listeners, rather than
slightly different ones.

It would be more accurate to call microtime() even earlier for all
packets, as you have to grab (1+#listener) locks before you can
determine if the packet will be logged. You could always grab a
timestamp before the locks, but microtime() can be costly, so this
didn't seem like a good idea.

(I guess most ethernet interfaces will have a bpf listener these
days because of dhclient. That means that we could be doing two bpf
locks on most packets going through the interface.)

PR: 71711

17 years agosoreceive_generic(), and sopoll_generic(). Add new functions sosend(),
rwatson [Mon, 24 Jul 2006 15:20:08 +0000 (15:20 +0000)]
soreceive_generic(), and sopoll_generic().  Add new functions sosend(),
soreceive(), and sopoll(), which are wrappers for pru_sosend,
pru_soreceive, and pru_sopoll, and are now used univerally by socket
consumers rather than either directly invoking the old so*() functions
or directly invoking the protocol switch method (about an even split
prior to this commit).

This completes an architectural change that was begun in 1996 to permit
protocols to provide substitute implementations, as now used by UDP.
Consumers now uniformly invoke sosend(), soreceive(), and sopoll() to
perform these operations on sockets -- in particular, distributed file
systems and socket system calls.

Architectural head nod: sam, gnn, wollman

17 years agoForced commit to recognize repo-copy of uipc_socket2.c to uipc_sockbuf.c,
rwatson [Mon, 24 Jul 2006 14:34:04 +0000 (14:34 +0000)]
Forced commit to recognize repo-copy of uipc_socket2.c to uipc_sockbuf.c,
which will hold socket buffer specific routines.

Thanks to: simon

17 years agoRemove a duplicated line.
davidxu [Mon, 24 Jul 2006 12:24:56 +0000 (12:24 +0000)]
Remove a duplicated line.

17 years agoFinally fix support for the newer MCP51/MCP55 nVidia chipsets.
sos [Mon, 24 Jul 2006 10:44:50 +0000 (10:44 +0000)]
Finally fix support for the newer MCP51/MCP55 nVidia chipsets.
The register layout has changed since the original NV4 - sigh.
Hotplug support has been fixed for all nVidia chipsets that supports it
(including the MCP51/55).

HW donated by: Kingsley College

17 years agoImprove mark-up: References to "start" as an rc.d method
yar [Mon, 24 Jul 2006 10:07:27 +0000 (10:07 +0000)]
Improve mark-up:  References to "start" as an rc.d method
should be tagged with ".Cm".  (Refs to "stop" are OK already.)

MFC after: 3 days

17 years agoo GC MT_FTABLE removed in rev. 1.192 mbuf.h.
maxim [Mon, 24 Jul 2006 07:32:23 +0000 (07:32 +0000)]
o GC MT_FTABLE removed in rev. 1.192 mbuf.h.
o Correct MT_HEADER define.

17 years agoNew release notes: iwi(4) update (+MFC), jexec(8) -u/-U (+MFC),
bmah [Mon, 24 Jul 2006 03:58:53 +0000 (03:58 +0000)]
New release notes:  iwi(4) update (+MFC), jexec(8) -u/-U (+MFC),
extensible printf(3) (+MFC), tail(1) -q (+MFC)

Updated release notes:  ath(4) HAL 0.9.17.2 (+MFC).

MFCs noted:  ath(4) et al. in GENERIC, resolver update, /media.

17 years agoRemove MT_FTABLE, as it's no longer used.
rwatson [Mon, 24 Jul 2006 01:49:57 +0000 (01:49 +0000)]
Remove MT_FTABLE, as it's no longer used.

Comment that many stats in mbstat are now not used, as libmemstat and
UMA stats are used.

17 years agoRemove MT_FTABLE printout in systat, as MT_FTABLE mbufs are no longer
rwatson [Mon, 24 Jul 2006 01:49:24 +0000 (01:49 +0000)]
Remove MT_FTABLE printout in systat, as MT_FTABLE mbufs are no longer
used in the kernel.

17 years agoGarbage collect #if 0'd MT_ mbuf types, as they are no longer used, and
rwatson [Mon, 24 Jul 2006 01:14:05 +0000 (01:14 +0000)]
Garbage collect #if 0'd MT_ mbuf types, as they are no longer used, and
there are no plans to re-introduce them.

17 years agoTweak so_gencnt comment: it was once last, but that is no longer the
rwatson [Mon, 24 Jul 2006 01:05:36 +0000 (01:05 +0000)]
Tweak so_gencnt comment: it was once last, but that is no longer the
case.

17 years agoTweak comment for so_head: it is a pointer to the listen socket, rather
rwatson [Mon, 24 Jul 2006 01:02:07 +0000 (01:02 +0000)]
Tweak comment for so_head: it is a pointer to the listen socket, rather
than the accept socket.

17 years agoFix a spelling error in a comment.
rwatson [Mon, 24 Jul 2006 00:33:24 +0000 (00:33 +0000)]
Fix a spelling error in a comment.

Found with: mckusick's code walkthrough DVDs

17 years agoRemove duplicate 'or'.
rwatson [Sun, 23 Jul 2006 21:01:09 +0000 (21:01 +0000)]
Remove duplicate 'or'.

Submitted by: ru

17 years agoUpdate various uipc_socket.c comments, and reformat others.
rwatson [Sun, 23 Jul 2006 20:36:04 +0000 (20:36 +0000)]
Update various uipc_socket.c comments, and reformat others.

17 years agoAdd additional comments to the top of the UNIX domain socket implementation
rwatson [Sun, 23 Jul 2006 20:06:45 +0000 (20:06 +0000)]
Add additional comments to the top of the UNIX domain socket implementation
providing some high level pointers regarding the implementation.

17 years agoRemove old kern.malloc sysctl, which generated a text representation of
rwatson [Sun, 23 Jul 2006 19:55:41 +0000 (19:55 +0000)]
Remove old kern.malloc sysctl, which generated a text representation of
the kernel malloc(9) state for vmstat -m.  libmemstat is now used to
generate a machine-readable version which is converged by vmstat -m
into a human-readable version.

Not for MFC.

17 years agoExpand comments for malloc(9) to better describe the design and
rwatson [Sun, 23 Jul 2006 19:51:39 +0000 (19:51 +0000)]
Expand comments for malloc(9) to better describe the design and
statistics / memory types model.

17 years agoUpdate and reformat comments for POSIX.1e ACL utility routines.
rwatson [Sun, 23 Jul 2006 19:35:10 +0000 (19:35 +0000)]
Update and reformat comments for POSIX.1e ACL utility routines.

17 years agoComment extended attribute name space constants.
rwatson [Sun, 23 Jul 2006 19:26:54 +0000 (19:26 +0000)]
Comment extended attribute name space constants.

17 years agoImprove comments for label data structure.
rwatson [Sun, 23 Jul 2006 19:26:32 +0000 (19:26 +0000)]
Improve comments for label data structure.

17 years agoAdd missing ;.
nork [Sun, 23 Jul 2006 15:53:12 +0000 (15:53 +0000)]
Add missing ;.

Pointy hat to: myself
Approved by: deischen (implicitly)

17 years agodo not overload the port number on to the return value of
ume [Sun, 23 Jul 2006 15:31:03 +0000 (15:31 +0000)]
do not overload the port number on to the return value of
str2number().  this could result in an unexpected code path.

Obtained from: KAME
MFC after: 1 week

17 years agoAdd missing syscalls.
nork [Sun, 23 Jul 2006 13:08:54 +0000 (13:08 +0000)]
Add missing syscalls.

Reviewed by: deischen
Approved by: deischen

17 years agoAlign IPv6 socket locking with IPv4 locking: lock socket buffer explicitly
rwatson [Sun, 23 Jul 2006 12:24:22 +0000 (12:24 +0000)]
Align IPv6 socket locking with IPv4 locking: lock socket buffer explicitly
and use _locked variants to avoid extra lock and unlock operations.

Reviewed by: gnn
MFC after: 1 week

17 years agoAdd two new unpcb flags, UNP_BINDING and UNP_CONNECTING, which will be
rwatson [Sun, 23 Jul 2006 12:01:14 +0000 (12:01 +0000)]
Add two new unpcb flags, UNP_BINDING and UNP_CONNECTING, which will be
used to mark UNIX domain sockets as being in the process of binding or
connecting.  Use these to prevent simultaneous bind or connect
operations by multiple threads or processes on the same socket at the
same time, which closes race conditions present in the UNIX domain
socket implementation since inception.

17 years agoMerge unp_bind() into uipc_bind(), as it is called only from uipc_bind().
rwatson [Sun, 23 Jul 2006 11:02:12 +0000 (11:02 +0000)]
Merge unp_bind() into uipc_bind(), as it is called only from uipc_bind().

17 years agoSince unp_attach() and unp_detach() are now called only from uipc_attach()
rwatson [Sun, 23 Jul 2006 10:25:28 +0000 (10:25 +0000)]
Since unp_attach() and unp_detach() are now called only from uipc_attach()
and uipc_detach(), merge them into their calling functions.

17 years agoMove various UNIX socket global variables and sysctls from the middle of
rwatson [Sun, 23 Jul 2006 10:19:04 +0000 (10:19 +0000)]
Move various UNIX socket global variables and sysctls from the middle of
the file to the top.

17 years agostop use of mutex lock in ICMP lookup.
ume [Sun, 23 Jul 2006 07:07:49 +0000 (07:07 +0000)]
stop use of mutex lock in ICMP lookup.

MFC after: 1 week

17 years agoExport the number of object bypasses and collapses through sysctl.
alc [Sat, 22 Jul 2006 22:31:57 +0000 (22:31 +0000)]
Export the number of object bypasses and collapses through sysctl.

17 years agoIn uipc_send() and uipc_rcvd(), store unp->unp_conn pointer in unp2
rwatson [Sat, 22 Jul 2006 18:41:42 +0000 (18:41 +0000)]
In uipc_send() and uipc_rcvd(), store unp->unp_conn pointer in unp2
while working with the second unpcb to make the code more clear.

17 years agoRe-wrap and other minor formatting and punctuation fixes for UNIX domain
rwatson [Sat, 22 Jul 2006 17:24:55 +0000 (17:24 +0000)]
Re-wrap and other minor formatting and punctuation fixes for UNIX domain
socket comments.

17 years agoImplement support for HMAC/SHA1 and HMAC/SHA256 acceleration found in
pjd [Sat, 22 Jul 2006 16:18:47 +0000 (16:18 +0000)]
Implement support for HMAC/SHA1 and HMAC/SHA256 acceleration found in
new VIA CPUs.
For older CPUs HMAC/SHA1 and HMAC/SHA256 (and others) will still be done
in software.

Move symmetric cryptography (currently only AES-CBC 128/192/256) to
padlock_cipher.c file. Move HMAC cryptography to padlock_hash.c file.

Hardware from: Centaur Technologies

17 years agoAdd initial platform support.
obrien [Sat, 22 Jul 2006 15:27:18 +0000 (15:27 +0000)]
Add initial platform support.

Submitted by: cognet

17 years agoAdd ARM platform Makefile for building disassembler.
obrien [Sat, 22 Jul 2006 14:59:50 +0000 (14:59 +0000)]
Add ARM platform Makefile for building disassembler.
(rev 1.1 was an accidental commit)

17 years agoAdd ARM platform Makefile for building BFD. We include the New ARM ABI only.
obrien [Sat, 22 Jul 2006 14:55:55 +0000 (14:55 +0000)]
Add ARM platform Makefile for building BFD.  We include the New ARM ABI only.
(rev 1.1 was an accidental commit)

17 years agoBe consistent in quoting.
obrien [Sat, 22 Jul 2006 14:37:17 +0000 (14:37 +0000)]
Be consistent in quoting.

17 years agoremove obsolete comments.
ume [Sat, 22 Jul 2006 14:36:56 +0000 (14:36 +0000)]
remove obsolete comments.

MFC after: 3 days

17 years agoCreate a genscripts driver based on the PowerPC one.
obrien [Sat, 22 Jul 2006 14:36:15 +0000 (14:36 +0000)]
Create a genscripts driver based on the PowerPC one.

17 years agoThe platform name is ARM not StrongARM.
obrien [Sat, 22 Jul 2006 14:12:02 +0000 (14:12 +0000)]
The platform name is ARM not StrongARM.

Submitted by: cognet

17 years agoTell when a .include is processed with '-dd'.
obrien [Sat, 22 Jul 2006 14:00:31 +0000 (14:00 +0000)]
Tell when a .include is processed with '-dd'.

17 years agoCorrect few bzero()s.
pjd [Sat, 22 Jul 2006 13:14:11 +0000 (13:14 +0000)]
Correct few bzero()s.

MFC after: 3 days

17 years agoNote the overhaul of asf(8).
yar [Sat, 22 Jul 2006 12:52:51 +0000 (12:52 +0000)]
Note the overhaul of asf(8).

17 years agoDon't forget to initialize crp_olen field, which is used to calculate
pjd [Sat, 22 Jul 2006 10:05:55 +0000 (10:05 +0000)]
Don't forget to initialize crp_olen field, which is used to calculate
bio_completed value.

17 years agoSet ses_ictx and ses_octx to NULL after freeing them, so we won't free
pjd [Sat, 22 Jul 2006 10:04:47 +0000 (10:04 +0000)]
Set ses_ictx and ses_octx to NULL after freeing them, so we won't free
them twice.
This is possible for example in situation when session is used in
authentication context, then freed and then used in encryption context
and freed - in encryption context ses_ictx and ses_octx are not touched
at newsession time, but padlock_freesession could still try to free them
when they are not NULL.

17 years agoFix build breakage from previous commit which confused key_abort and key_close.
gnn [Sat, 22 Jul 2006 09:18:02 +0000 (09:18 +0000)]
Fix build breakage from previous commit which confused key_abort and key_close.

17 years agoThe month name in .Dd should be spelled in full.
yar [Sat, 22 Jul 2006 07:23:46 +0000 (07:23 +0000)]
The month name in .Dd should be spelled in full.

Pointed out by: ru

17 years agoThe KAME project ceased work on IPv6 and IPSec in March of 2006.
gnn [Sat, 22 Jul 2006 02:32:32 +0000 (02:32 +0000)]
The KAME project ceased work on IPv6 and IPSec in March of 2006.
Remove the README file which warns against cosmetic or local only
changes.  FreeBSD committers should now feel free to work on the
IPv6 and IPSec code without fetters.  The KAME mailing lists still
exist and it is always a good idea to ask questions about this code
on the snap-users@kame.net mailing list.

Reviewed by: rwatson, brooks

17 years agoRetire debug.mpsafevm. None of the architectures supported in CVS require
alc [Fri, 21 Jul 2006 23:22:49 +0000 (23:22 +0000)]
Retire debug.mpsafevm.  None of the architectures supported in CVS require
it any longer.

17 years agoRemove mention of the `W' flag, which has been turned to no-op by the
sobomax [Fri, 21 Jul 2006 22:13:06 +0000 (22:13 +0000)]
Remove mention of the `W' flag, which has been turned to no-op by the
neworder change. Keep the option in a config file parser, to not
violate POLA.

MFC after: 2 weeks