]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
17 years agoMFC: Move cpu_exit() earlier in exit1() to close a race between
jhb [Thu, 17 May 2007 13:52:37 +0000 (13:52 +0000)]
MFC: Move cpu_exit() earlier in exit1() to close a race between
SIGCHLD/kevent(2) notification and wait().

17 years agoHook lagg(4) up to the build.
thompsa [Thu, 17 May 2007 01:50:46 +0000 (01:50 +0000)]
Hook lagg(4) up to the build.

17 years agoMFC the lagg(4) driver which provides link aggregation, failover and fault
thompsa [Thu, 17 May 2007 01:49:41 +0000 (01:49 +0000)]
MFC the lagg(4) driver which provides link aggregation, failover and fault
tolerance.

17 years agoMFC r1.30
thompsa [Thu, 17 May 2007 01:46:42 +0000 (01:46 +0000)]
MFC r1.30
 Add Ethertype for 802.3ad LACP.

17 years agoMFC, the UNAME_ and OSVERSION environment over rides. This makes
ambrisko [Wed, 16 May 2007 21:39:44 +0000 (21:39 +0000)]
MFC, the UNAME_ and OSVERSION environment over rides.  This makes
easier to make a FreeBSD 6 chroot think it is that version.

17 years agoThis commit was manufactured by cvs2svn to create branch 'RELENG_6'.
cvs2svn [Wed, 16 May 2007 17:19:48 +0000 (17:19 +0000)]
This commit was manufactured by cvs2svn to create branch 'RELENG_6'.

17 years agoMFC:
mtm [Wed, 16 May 2007 10:15:19 +0000 (10:15 +0000)]
MFC:
----------------------------
revision 1.6
date: 2007/04/12 08:39:13;  author: ceri;  state: Exp;  lines: +2 -2
Bump .Dd for r1.5; fix grammatical problem.
----------------------------
revision 1.5
date: 2007/04/12 08:35:58;  author: mtm;  state: Exp;  lines: +18 -1
Specify the correct way to modify this file, and warn that the
user should not depend on the internal variables documented in
this man page.

17 years agoMFC:
mtm [Wed, 16 May 2007 10:10:30 +0000 (10:10 +0000)]
MFC:
----------------------------
revision 1.30
date: 2007/04/12 08:17:56;  author: mtm;  state: Exp;  lines: +2 -7
o Use the uidstart variable instead of uuid (bug #3)
o Actually save the variable to adduser.conf (bug #2)
o (bug #1 to be fixed in an upcomming commit to adduser.conf.5)

----------------------------
revision 1.29
date: 2007/04/12 07:38:26;  author: mtm;  state: Exp;  lines: +13 -12
Errm... I don't see how rev. 1.26 could have possibly worked or been tested.
Fix it for real.

Submitted by: Johnny Lee <johnny@bmtk.com>

17 years agoMFC: the new FTP_COMBINE_CWDS define allows libfetch to attempt to change
njl [Wed, 16 May 2007 07:04:02 +0000 (07:04 +0000)]
MFC: the new FTP_COMBINE_CWDS define allows libfetch to attempt to change
directories all at once, saving latency.

Tested by: des

17 years agoMFC: Makefile magic for the new snd_* manual pages.
joel [Tue, 15 May 2007 20:49:10 +0000 (20:49 +0000)]
MFC: Makefile magic for the new snd_* manual pages.

17 years agoMFC: manual pages for snd_hda, snd_envy24, snd_envy24ht and snd_spicds.
joel [Tue, 15 May 2007 18:40:22 +0000 (18:40 +0000)]
MFC: manual pages for snd_hda, snd_envy24, snd_envy24ht and snd_spicds.

17 years agoMFC rev. 1.53:
thomas [Tue, 15 May 2007 16:19:42 +0000 (16:19 +0000)]
MFC rev. 1.53:
(atapi_action, case XPT_SCSI_IO): Enable DMA only for READ and WRITE commands
 as some combinations of chipset, controller and target do not behave
 correctly when DMA is enabled for other commands.

PR: kern/103602

17 years agoMFC: revision 1.9 of rc.d/ip6fw and 1.15 of rc.d/ipfw
mtm [Tue, 15 May 2007 09:18:25 +0000 (09:18 +0000)]
MFC: revision 1.9 of rc.d/ip6fw and 1.15 of rc.d/ipfw

date: 2007/04/02 15:38:53;  author: mtm;  state: Exp;  lines: +1 -1
Instead of directly sourcing the firewall script, run it in a separate shell.
If the firewall script is sourced directly from the script, then any
exit statements in it will also terminate the rc.d script prematurely.

PR: conf/78762

17 years agoMFC: 1.22
mtm [Tue, 15 May 2007 08:58:05 +0000 (08:58 +0000)]
MFC: 1.22

date: 2007/03/30 18:36:45;  author: mtm;  state: Exp;  lines: +1 -0
Some rc.d commands (such as stop|restart etc.) won't automagically work
if we don't explicitly set the name of the executable program.

PR: conf/104408

17 years agoMFC 1.4:
bruno [Tue, 15 May 2007 08:19:21 +0000 (08:19 +0000)]
MFC 1.4:
o introduce a flags 'errata' for HW bugs onto the softc.
o remove errata_a0 and introduce the corresponding flags into 'errata'.
o introduce a new errata for K8, namely some platform might set the
  PENDING_BIT but aren't able to unset it, also don't loop forever
  waiting PENDING_BIT being cleared.
o try to introduce a workaround for the PENDING_BIT stuck problem,
o support now half multipliers for K8.

Tested by: Abdullah Al-Marrie
Approved by: njl

17 years agoMFC Revision 1.123:
davidxu [Tue, 15 May 2007 08:10:15 +0000 (08:10 +0000)]
MFC Revision 1.123:
> Fix a TLS memory leak.

17 years agoMFC: 1.13
mtm [Tue, 15 May 2007 08:01:42 +0000 (08:01 +0000)]
MFC: 1.13

date: 2007/03/31 09:03:38;  author: mtm;  state: Exp;  lines: +4 -4
Make subroutine names more conformant with other scripts in rc.d.

17 years agoMFC: print GAS properly and man page updates
njl [Mon, 14 May 2007 21:32:10 +0000 (21:32 +0000)]
MFC: print GAS properly and man page updates

17 years agoMFC rev. 1.75: fill the list of icmp types; make its size depend
maxim [Mon, 14 May 2007 18:09:00 +0000 (18:09 +0000)]
MFC rev. 1.75: fill the list of icmp types; make its size depend
on ICMP_MAXTYPE.

17 years agoMFC revs. 1.30, 1.31: fix strtoul() error conditions check, kill EOLWS.
maxim [Mon, 14 May 2007 17:31:41 +0000 (17:31 +0000)]
MFC revs. 1.30, 1.31: fix strtoul() error conditions check, kill EOLWS.

17 years agoMFC nd6.c Rev.1.76-77 and nd6_nbr.c Rev.1.43-44.
suz [Mon, 14 May 2007 02:50:03 +0000 (02:50 +0000)]
MFC  nd6.c Rev.1.76-77 and nd6_nbr.c Rev.1.43-44.

Fixed a memory leak in unresolved ND queue processing.

17 years agoMFC: rev. 1.13
simokawa [Mon, 14 May 2007 00:58:23 +0000 (00:58 +0000)]
MFC: rev. 1.13
Fix a problem for multiple lun/initiators.

17 years agoMFC: [stage: 7/7]
ariff [Sun, 13 May 2007 21:13:11 +0000 (21:13 +0000)]
MFC: [stage: 7/7]

Add module loading options for snd_hda(4) and snd_envy24*/spicds(4).

17 years agoMFC: [stage: 6/7]
ariff [Sun, 13 May 2007 21:12:32 +0000 (21:12 +0000)]
MFC: [stage: 6/7]

Add notes and options for snd_hda(4) and snd_envy24*/spicds(4).

17 years agoMFC: [stage: 5/7]
ariff [Sun, 13 May 2007 21:11:40 +0000 (21:11 +0000)]
MFC: [stage: 5/7]

Connect both snd_hda(4) and snd_envy24*/spicds(4) to build process.

17 years agoMFC: [stage: 4/7]
ariff [Sun, 13 May 2007 21:09:24 +0000 (21:09 +0000)]
MFC: [stage: 4/7]

Allright you rightist.. here we go.

Add support for Intel High Definition Audio Controller.

This driver make a special guarantee that "playback" works
on majority hardwares with minimal or without specific vendor
quirk.

This driver is a product of collaborative effort made by:

  Stephane E. Potvin <sepotvin@videotron.ca>
       Andrea Bittau <a.bittau@cs.ucl.ac.uk>
       Wesley Morgan <morganw@chemikals.org>
      Daniel Eischen <deischen@FreeBSD.org>
     Maxime Guillaud <bsd-ports@mguillaud.net>
      Ariff Abdullah <ariff@FreeBSD.org>

....and various people from freebsd-multimedia@FreeBSD.org

Refer to snd_hda(4) for features, issues and various tricks.

P L E A S E  avoid (not that I forbid it) using send-pr, since many
problems can be solved (and was solved) quicker and easier through
various help channels, notably freebsd-multimedia or even better,
#freebsd-azalia @ irc.freenode.net.

Welcome To HDA.

Requested by: Everybody
Sponsored by: Defenxis Sdn. Bhd.

17 years agoMFC: [stage: 3/7]
ariff [Sun, 13 May 2007 21:03:46 +0000 (21:03 +0000)]
MFC: [stage: 3/7]

Envy24/HT and related dependency, spicds (as of current state).

From the original commit log (by netchild):

dd the envy24 driver as is to the tree. It's not connected to the build
yet. More commits to follow.

I got no response from the author, but since the driver is BSD licensed
I don't think he will complain. :-)

I got it from http://people.freebsd.org/~lofi/envy24.tar.gz

Written by: Katsurajima Naoto <raven@katsurajima.seya.yokohama.jp
            Konstantin Dimitrov <kosio.dimitrov@gmail.com>

17 years agoMFC: [stage: 2/7]
ariff [Sun, 13 May 2007 20:53:40 +0000 (20:53 +0000)]
MFC: [stage: 2/7]

Various fixups, especially for the upcomming High Definition Audio
commit.

1) sys/dev/sound/pcm/sound.h
   sys/dev/sound/pcm/channel.c
   * Be more specific: SD_F_SOFTVOL -> SD_F_SOFTPCMVOL
2) sys/dev/sound/pcm/mixer.[ch]
   * Implement
       mix_setparentchild()
       mix_setrealdev()
       mix_getparent()
       mix_getchild()
     The purpose of these functions is implement relative volume
     adjustment, such as to tie two or more mixer device into a
     single logical device. Usefull for the upcoming HDA driver
     and few AC97 codec where the master volume "vol" need to be
     implemented using this logical manner.
3) sys/dev/sound/pcm/ac97.c
   * ditto #1
   sys/dev/sound/pcm/usb/uaudio_pcm.c
   * ditto #1, hook "pcm" as logical child of "vol" if the real
     control does not exist.

17 years agoMFC: [stage: 1/7]
ariff [Sun, 13 May 2007 20:50:31 +0000 (20:50 +0000)]
MFC: [stage: 1/7]

buf_addr should be bus_addr_t rather than u_int32_t.

17 years ago- Remove the deprecated ``/stand'' hierarchy.
chinsan [Sun, 13 May 2007 12:22:54 +0000 (12:22 +0000)]
- Remove the deprecated ``/stand'' hierarchy.
  (MFC)

PR: docs/111137
Submitted by: Niclas Zeising
Approved by: delphij@ (mentor)

17 years agoMFC: v1.20
pav [Sun, 13 May 2007 11:50:32 +0000 (11:50 +0000)]
MFC: v1.20
  Document that uniq(1) limits input line length to LINE_MAX characters.

17 years agoThis commit was manufactured by cvs2svn to create branch 'RELENG_6'.
cvs2svn [Sun, 13 May 2007 09:33:36 +0000 (09:33 +0000)]
This commit was manufactured by cvs2svn to create branch 'RELENG_6'.

17 years agoMFC 1.10, 1.11
pav [Sat, 12 May 2007 23:10:08 +0000 (23:10 +0000)]
MFC 1.10, 1.11
  Revise the markup, sort sections, fix some grammar bugs.
  Fix the description of "media-type".

17 years agoMFC: r1.23
gabor [Sat, 12 May 2007 21:29:16 +0000 (21:29 +0000)]
MFC: r1.23

Approved by: keramida (mentor)

17 years agoMFC: Don't call bus_deactivate_resource() explicitly before calling
nyan [Sat, 12 May 2007 06:10:11 +0000 (06:10 +0000)]
MFC: Don't call bus_deactivate_resource() explicitly before calling
     bus_release_resource().

17 years agomfc: bad merge fix
mjacob [Fri, 11 May 2007 13:48:33 +0000 (13:48 +0000)]
mfc: bad merge fix

17 years agoFix pointy-hat problem with BUS_DMA_ROOTARG macro that caused problems for sparc64.
mjacob [Fri, 11 May 2007 06:29:23 +0000 (06:29 +0000)]
Fix pointy-hat problem with BUS_DMA_ROOTARG macro that caused problems for sparc64.

17 years agoMerge uipc_usrreq.c:1.193 from HEAD to RELENG_6:
rwatson [Thu, 10 May 2007 20:00:35 +0000 (20:00 +0000)]
Merge uipc_usrreq.c:1.193 from HEAD to RELENG_6:

  Change unp_mtx to supporting recursion, and do not drop the unp_mtx over
  sonewconn() in unp_connect().  This avoids a race that occurs due to
  v_socket being an uncounted reference, as the lock was being released in
  order to call sonewconn(), which otherwise recurses into the UNIX domain
  socket code via pru_attach, as well as holding the lock over a sleeping
  memory allocation in uipc_attach().  Switch to a non-sleeping memory
  allocation during UNIX domain socket attach.

  This fix non-ideal in that it requires enabling recursion, but is a much
  smaller change than moving to using true references for v_socket.  The
  reported panic occurs in unp_connect() following the return of
  sonewconn().

  Update copyright year.

  Panic reported by:      jhb

Tested by: jhb

17 years agoMFC: We need four disks these days, so reflect this.
delphij [Thu, 10 May 2007 13:20:45 +0000 (13:20 +0000)]
MFC: We need four disks these days, so reflect this.

Pointed out by: phk
PR: docs/112571

17 years agoMFC rev. 1.259:
kib [Thu, 10 May 2007 09:18:07 +0000 (09:18 +0000)]
MFC rev. 1.259:
Allow the dounmount() to proceed even for doomed coveredvp.

17 years agoMFC rev 1.42
gnn [Thu, 10 May 2007 07:12:52 +0000 (07:12 +0000)]
MFC rev 1.42

Reduce the default number of header options that the IPv6 protocol
stack will process from 50 to 15.  As this is a sysctl variable it
can be tuned up or down at the user/administrator's whim.

17 years agoMFC rev 1.5: Add PCI Ids.
scottl [Wed, 9 May 2007 17:32:10 +0000 (17:32 +0000)]
MFC rev 1.5: Add PCI Ids.

17 years agoMFC if_mskreg.h rev. 1.4 to RELENG_6.
yongari [Wed, 9 May 2007 06:36:41 +0000 (06:36 +0000)]
MFC if_mskreg.h rev. 1.4 to RELENG_6.
  Work around GMAC hardware hang bug.
  It seems that valid pause frames(Tx flow control) cause GMAC to hang
  such that it resulted in watchdog timeout. As a work around don't
  flush Rx MAC FIFO if we've received pause frames.

17 years agoMFC if_msk.c rev. 1.15 to RELENG_6.
yongari [Wed, 9 May 2007 06:34:29 +0000 (06:34 +0000)]
MFC if_msk.c rev. 1.15 to RELENG_6.
  Disable TSO support.
  Under certain circumtances, if TSO is active, Yukon II generates
  corrupted IP packets. All corrupted IP packets I noticed were the the
  last segmented packet in a TSO request. The corrupted packet resulted
  in retransmission of the damaged packet which in turn decreased network
  performance dramatically.
  Unfortunately it seems that there is no way to workaround this bug
  as TSO is completely handled in hardware. Disable TSO until we find a
  working workaround or a new silicon revision that doesn't have this
  hardware bug.

17 years agoMFC if_msk.c rev. 1.14 to RELENG_6.
yongari [Wed, 9 May 2007 06:32:19 +0000 (06:32 +0000)]
MFC if_msk.c rev. 1.14 to RELENG_6.
  Fix TCP header size calculation logic which is used for setting
  TSO MTU.

17 years agoMFC if_stge.c rev. 1.9 to RELENG_6.
yongari [Wed, 9 May 2007 06:26:45 +0000 (06:26 +0000)]
MFC if_stge.c rev. 1.9 to RELENG_6.
  If we've got watchdog timeouts try to get more packets going after
  resetting the hardware.

17 years agoMFC if_stge.c rev. 1.8, if_stgereg.h rev. 1.3 to RELENG_6.
yongari [Wed, 9 May 2007 06:25:01 +0000 (06:25 +0000)]
MFC if_stge.c rev. 1.8, if_stgereg.h rev. 1.3 to RELENG_6.
  Honor link up/down state in stge_start().
  While I'm here move MAC control settings to stge_link_task, a task
  queue which handles link state and duplex/flow controls.

17 years agoMFC if_stge.c rev. 1.7, if_stgereg.h rev. 1.2 to RELENG_6.
yongari [Wed, 9 May 2007 06:22:14 +0000 (06:22 +0000)]
MFC if_stge.c rev. 1.7, if_stgereg.h rev. 1.2 to RELENG_6.
  Use our own timer for watchdog instead of if_watchdog/if_timer
  interface.

17 years agoMFC: Add some more examples. It always takes me a long time to find the S
brian [Tue, 8 May 2007 22:45:45 +0000 (22:45 +0000)]
MFC: Add some more examples.  It always takes me a long time to find the S
     modifier (-t is ignored without it).

17 years agoRe-MFC 1.112 and this time include the fix for a bug in it (MFC of 1.162)
julian [Tue, 8 May 2007 20:24:38 +0000 (20:24 +0000)]
Re-MFC 1.112 and this time include the fix for a bug in it (MFC of 1.162)
This changes to using a monotomic time referrence for all time based
operations.
Changing the time should no longer accellerate ipfw operations.

17 years agoMFC 1.93 1.94.
julian [Tue, 8 May 2007 20:03:01 +0000 (20:03 +0000)]
MFC 1.93 1.94.
allow table entries to store/return an IP address as a value.

17 years agoMFC to synchronize with -current. The bugs fixed are relatively few and
mjacob [Tue, 8 May 2007 17:47:08 +0000 (17:47 +0000)]
MFC to synchronize with -current. The bugs fixed are relatively few and
hopefully the bugs introduved are even fewer.

17 years agoMFC: 1.189
jkim [Tue, 8 May 2007 16:18:21 +0000 (16:18 +0000)]
MFC: 1.189

Reflect MFC of pci_find_extcap().

17 years agoMerge uipc_usrreq.c:1.196 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 11:50:57 +0000 (11:50 +0000)]
Merge uipc_usrreq.c:1.196 from HEAD to RELENG_6:

  Add an additional MAC check to the UNIX domain socket connect path:
  check that the subject has read/write access to the vnode using the
  vnode MAC check.

  Submitted by:   Spencer Minear <spencer_minear at securecomputing dot com>
  Obtained from:  TrustedBSD Project

17 years agoMerge uipc_usrreq.c:1.187 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 11:47:19 +0000 (11:47 +0000)]
Merge uipc_usrreq.c:1.187 from HEAD to RELENG_6:

  Minor white space tweaks.

17 years agoMerge uipc_usrreq.c:1.186, uipc_proto.c:1.27, domain.h:1.22, files:1.1137
rwatson [Tue, 8 May 2007 11:45:12 +0000 (11:45 +0000)]
Merge uipc_usrreq.c:1.186, uipc_proto.c:1.27, domain.h:1.22, files:1.1137
from HEAD to RELENG_6:

  Move definition of UNIX domain socket protosw and domain entries from
  uipc_proto.c to uipc_usrreq.c, making localdomain static.  Remove
  uipc_proto.c as it's no longer used.  With this change, UNIX domain
  sockets are entirely encapsulated in uipc_usrreq.c.

17 years agoMerge uipc_proto.c:1.26 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 11:07:24 +0000 (11:07 +0000)]
Merge uipc_proto.c:1.26 from HEAD to RELENG_6:

  Remove UNIX domain socket raw socket support.  This feature is documented
  as being undocumented in Stevens, and was broken in 1997 during network
  stack infrastructure work.  It is the one remaining (and incorrect)
  direct protocol reference to raw_usrreq.pru_attach; this is incorrect
  because the raw socket code assumes that raw_uattach is called only after
  the protocol has allocated a PCB.

17 years agoMerge uipc_usrreq.c:1.184 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 10:22:46 +0000 (10:22 +0000)]
Merge uipc_usrreq.c:1.184 from HEAD to RELENG_6:

  Remove now unneeded ENOTCONN clause from SOCK_DGRAM side of uipc_send():
  we have to check it regardless of the target address, so don't check it
  twice.

17 years agoMerge uipc_usrreq.c:1.179,1.180 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 10:20:22 +0000 (10:20 +0000)]
Merge uipc_usrreq.c:1.179,1.180 from HEAD to RELENG_6:

  Add additional comments to the top of the UNIX domain socket implementation
  providing some high level pointers regarding the implementation.

  Remove duplicate 'or'.

  Submitted by:   ru

17 years agoMerge unpcb.h:1.20 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 10:16:57 +0000 (10:16 +0000)]
Merge unpcb.h:1.20 from HEAD to RELENG_6:

  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 uipc_usrreq.c:1.178,1.192 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 10:16:23 +0000 (10:16 +0000)]
Merge uipc_usrreq.c:1.178,1.192 from HEAD to RELENG_6:

  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.

  Set UNP_CONNECTING when committing to moving ahead in unp_connect().
  This logic was lost when merging the remainder of these changes in
  1.178.

17 years agoMerge uipc_usrreq.c:1.177 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 10:01:51 +0000 (10:01 +0000)]
Merge uipc_usrreq.c:1.177 from HEAD to RELENG_6:

  Merge unp_bind() into uipc_bind(), as it is called only from uipc_bind().

17 years agoMerge uipc_usrreq.c:1.175 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 08:41:13 +0000 (08:41 +0000)]
Merge uipc_usrreq.c:1.175 from HEAD to RELENG_6:

  Move various UNIX socket global variables and sysctls from the middle of
  the file to the top.

17 years agoMerge uipc_usrreq.c:1.174 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 08:37:27 +0000 (08:37 +0000)]
Merge uipc_usrreq.c:1.174 from HEAD to RELENG_6:

  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 agoMerge uipc_usrreq.c:1.171 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 08:32:29 +0000 (08:32 +0000)]
Merge uipc_usrreq.c:1.171 from HEAD to RELENG_6:

  Reduce periods of simultaneous acquisition of various socket buffer
  locks and the unplock during uipc_rcvd() and uipc_send() by caching
  certain values from one structure while its locks are held, and
  applying them to a second structure while its locks are held.  If
  done carefully, this should be correct, and will reduce the amount
  of work done with the global unp lock held.

  Tested by:      kris (earlier version)

17 years agoMerge uipc_usrreq.c:1.173 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 08:18:55 +0000 (08:18 +0000)]
Merge uipc_usrreq.c:1.173 from HEAD to RELENG_6:

  Re-wrap and other minor formatting and punctuation fixes for UNIX domain
  socket comments.

17 years agoMerge uipc_usrreq.c:1.169 from HEAD to RELENG_6:
rwatson [Tue, 8 May 2007 08:10:37 +0000 (08:10 +0000)]
Merge uipc_usrreq.c:1.169 from HEAD to RELENG_6:

  Remove unused (and ifdef'd) unp_abort() and unp_drain().

17 years agoRFC:
grog [Tue, 8 May 2007 02:31:56 +0000 (02:31 +0000)]
RFC:

Add -A flag to adjust existing time stamps.
Print name by which program was started in usage() message.

17 years agoMFC:
simokawa [Mon, 7 May 2007 08:00:21 +0000 (08:00 +0000)]
MFC:
- Simplify tlabel handling.
- Remove unused fw_asybusy().
- Add a sysctl knob to disable cycle master mode and add some comments.
- Fix broken userland API for async packets.
- Fix typo in recv spd.
- Update state in fw_xferq_dorain() after removed from the send queue.
- Remove unnecessary 'goto err;".
- Improve asynchronous packet receive process.

17 years agoMFC: Free tlabel in fw_xfer_done().
simokawa [Mon, 7 May 2007 06:35:23 +0000 (06:35 +0000)]
MFC: Free tlabel in fw_xfer_done().

17 years agoMFC:
simokawa [Mon, 7 May 2007 06:20:49 +0000 (06:20 +0000)]
MFC:
- Detect cycle lost.
- Less verbose debug messages.
- Remove xfer->retry_req.
- Remove retry_count.
- Replace xfer->act.hand with xfer->hand.

17 years agoMFC: Configuration ROM length should be unsigned.
simokawa [Mon, 7 May 2007 06:01:25 +0000 (06:01 +0000)]
MFC: Configuration ROM length should be unsigned.

17 years agoMFC:
simokawa [Mon, 7 May 2007 05:42:45 +0000 (05:42 +0000)]
MFC:
Initialize configuration ROM before a bus reset.

17 years agoMFC a change to pmap_copy() that was overlooked because revision 1.583
alc [Sun, 6 May 2007 18:36:57 +0000 (18:36 +0000)]
MFC a change to pmap_copy() that was overlooked because revision 1.583
was MFCed before revision 1.540.

17 years agoMFC Revision 1.531 of pmap.c:
alc [Sun, 6 May 2007 05:23:19 +0000 (05:23 +0000)]
MFC Revision 1.531 of pmap.c:
  Avoid unnecessary differences between the amd64 and i386
  format strings.

MFC Revision 1.577 of pmap.c:
  Use a different bitmask for superpages' base address so that it
  doesn't conflict with the PG_PDE_PAT bit.

17 years agoMFC Revision 1.538:
alc [Sun, 6 May 2007 00:34:43 +0000 (00:34 +0000)]
MFC Revision 1.538:
  Remove stale KSE code.

MFC Revision 1.551:
  In general, bits in the page directory entry (PDE) and the page table
  entry (PTE) have the same meaning.  The exception to this rule is the
  eighth bit (0x080).  It is the PS bit in a PDE and the PAT bit in a
  PTE.  This change avoids the possibility that pmap_enter() confuses a
  PAT bit with a PS bit, avoiding a panic().

  Eliminate a diagnostic printf() from the i386 pmap_enter() that serves
  no current purpose.

MFC Revision 1.559:
  Eliminate a comment that became stale after revision 1.540.

MFC Revision 1.575:
  Finish the PG_NX support at the pmap level.

MFC Revision 1.582:
  Eliminate the misuse of PG_FRAME to truncate a virtual address
  to a virtual page boundary.

17 years agoMFC
alc [Sat, 5 May 2007 21:13:16 +0000 (21:13 +0000)]
MFC
  Acquiring smp_ipi_mtx on every call to pmap_invalidate_*() is wasteful.
  For example, during a buildworld more than half of the calls do not
  generate an IPI because the only TLB entry invalidated is on the calling
  processor.  This revision pushes down the acquisition and release of
  smp_ipi_mtx into smp_tlb_shootdown() and smp_targeted_tlb_shootdown() and
  instead uses sched_pin() and sched_unpin() in pmap_invalidate_*() so that
  thread migration doesn't lead to a missed TLB invalidation.

17 years agoMFC Revision 1.530
alc [Sat, 5 May 2007 18:58:22 +0000 (18:58 +0000)]
MFC Revision 1.530
  Replace diagnostic printf()s by assertions.  Use consistent style
  for similar assertions.

MFC Revision 1.540
  Introduce pmap_try_insert_pv_entry().

  Use pmap_try_insert_pv_entry() in pmap_copy() instead of
  pmap_insert_entry().

  Eliminate the explicit low-memory checks in pmap_copy().

MFC Revision 1.544
  Retire pmap_track_modified().

MFC Revision 1.555
  Introduce the function pmap_enter_object().

MFC Revision 1.558 (in part)
  Change pmap_enter_quick_locked() to fail rather than wait if it is
  unable to allocate a page table page.  Similarly, change
  pmap_enter_quick_locked() to call pmap_try_insert_pv_entry() rather
  than pmap_insert_entry().

  Add an assertion that the object containing m_start is locked in
  pmap_enter_object().  Remove a similar assertion from
  pmap_enter_quick_locked() because that function no longer accesses
  the containing object.

  Remove a stale comment.

17 years agoMFC: enable i810 (including i915, i945, etc) on amd64
njl [Fri, 4 May 2007 22:04:41 +0000 (22:04 +0000)]
MFC: enable i810 (including i915, i945, etc) on amd64

17 years agoMFC 1.61:
le [Fri, 4 May 2007 17:36:37 +0000 (17:36 +0000)]
MFC 1.61:

   Accept passwords which contain whitespace.

   PR:     bin/53434

17 years agoMFC all changes regarding new option -M.
le [Fri, 4 May 2007 17:33:04 +0000 (17:33 +0000)]
MFC all changes regarding new option -M.

17 years agoMFC 1.8:
le [Fri, 4 May 2007 17:22:26 +0000 (17:22 +0000)]
MFC 1.8:

    Do the right thing with symlinks in the skeleton directory.

    PR:     bin/63659

17 years agoMFC 1.28:
le [Fri, 4 May 2007 17:17:24 +0000 (17:17 +0000)]
MFC 1.28:

   Check if the new user already exists right after entering the
   username instead of watching the final call to pw(8) fail.

17 years agoMFC: s/jail id/jail ID/ in top
rafan [Fri, 4 May 2007 15:45:13 +0000 (15:45 +0000)]
MFC: s/jail id/jail ID/ in top

  - s/jail id/jail ID/, acronyms should be in uppercase in general. Also,
    it is written this way in jail(8).

Suggested by: brueffer
Approved by: delphij (mentor, implicit)

17 years ago- Add top '-j' switch
rafan [Fri, 4 May 2007 15:25:05 +0000 (15:25 +0000)]
- Add top '-j' switch

Reviewed by: brueffer
Approved by: delphij (mentor, implicit), brueffer

17 years agoMFC rev. 1.35 - remove MSG_PEEK from the description of the send(2)
roam [Fri, 4 May 2007 09:06:59 +0000 (09:06 +0000)]
MFC rev. 1.35 - remove MSG_PEEK from the description of the send(2)
syscall; it is only relevant for incoming data.

17 years ago- MFC up to 1.106:
stas [Thu, 3 May 2007 22:39:24 +0000 (22:39 +0000)]
- MFC up to 1.106:
  * add hpiod and hpssd (1.106)
  * add amanda control tcp port (1.105)
  * sync xmmp service names with IANA (1.104).

17 years agoRegen after addition of 32bit versons of getcontext, setcontext and
sobomax [Thu, 3 May 2007 11:18:44 +0000 (11:18 +0000)]
Regen after addition of 32bit versons of getcontext, setcontext and
swapcontext.

17 years agoMFC: Implement 32 bit getcontext/setcontext/swapcontext on amd64. Still
sobomax [Thu, 3 May 2007 11:13:58 +0000 (11:13 +0000)]
MFC: Implement 32 bit getcontext/setcontext/swapcontext on amd64. Still
stubs for ia64 to keep it compiling.  These are used by 32 bit apps such
as gdb.

17 years agoMFC rev. 1.32:
yar [Thu, 3 May 2007 09:56:51 +0000 (09:56 +0000)]
MFC rev. 1.32:

  Fix a typo: argv -> argvp.

  Rationale:
  We are interested in the current (last) element of the argv array
  there, not in its first element. [...]

  Also add a comment that the argv array ends up null-terminated in
  any case (it's due to the design of the for loop) as an answer to
  a possible question why the whole argv isn't zero-filled.

17 years agoMFC rev. 1.52:
thomas [Thu, 3 May 2007 09:38:54 +0000 (09:38 +0000)]
MFC rev. 1.52:
(atapi_cb): Fix test for the presence of sense data. An incorrect condition
 was being tested, which would result in a system hang in some configurations.

PR: kern/112119

17 years agoArgh, fix a mis-merge.
jhb [Wed, 2 May 2007 20:16:29 +0000 (20:16 +0000)]
Argh, fix a mis-merge.

Reported by: kib

17 years agoMFC: Honor the BUS_DMA_NOCACHE flag to bus_dmamem_alloc() on amd64 and
jhb [Wed, 2 May 2007 18:48:18 +0000 (18:48 +0000)]
MFC: Honor the BUS_DMA_NOCACHE flag to bus_dmamem_alloc() on amd64 and
i386 by mapping the pages as UC (uncacheable) using pmap_change_attr().

Requested by: ariff

17 years agoMFC: Initial PAT support including the following:
jhb [Wed, 2 May 2007 18:42:47 +0000 (18:42 +0000)]
MFC: Initial PAT support including the following:
- New pmap_mapdev_attr() function for amd64 and i386.
- pmap_mapdev() on i386 and amd64 uses UC now rather than WB.
- New pmap_mapbios()/pmap_unmapbios() functions to map firmware tables.
- New pmap_change_attr() function for amd64 and i386.
- Bump __FreeBSD_version.

17 years agoMFC: Add various constants for the PAT MSR and the PAT PTE and PDE flags
jhb [Wed, 2 May 2007 16:16:57 +0000 (16:16 +0000)]
MFC: Add various constants for the PAT MSR and the PAT PTE and PDE flags
and initialize the PAT MSR during boot.

17 years agoMFC: Add 'pmap_invalidate_cache()'.
jhb [Wed, 2 May 2007 15:40:15 +0000 (15:40 +0000)]
MFC: Add 'pmap_invalidate_cache()'.

17 years agoMFC: Comment fix in 1.110.
jhb [Wed, 2 May 2007 15:27:24 +0000 (15:27 +0000)]
MFC: Comment fix in 1.110.

17 years agoMFC: Comment fixes in 1.106 and 1.107.
jhb [Wed, 2 May 2007 15:22:20 +0000 (15:22 +0000)]
MFC: Comment fixes in 1.106 and 1.107.

17 years agoMFC: Various fixes to NFS DirectIO support.
jhb [Wed, 2 May 2007 15:15:51 +0000 (15:15 +0000)]
MFC: Various fixes to NFS DirectIO support.

17 years agoMFC rev. 1.35-1.41: Clean up and update the HARDWARE section.
yar [Wed, 2 May 2007 10:26:32 +0000 (10:26 +0000)]
MFC rev. 1.35-1.41: Clean up and update the HARDWARE section.