]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
17 years agoMFC revision 1.546
alc [Sun, 25 Mar 2007 22:01:24 +0000 (22:01 +0000)]
MFC revision 1.546
  Eliminate unnecessary invalidations of the entire TLB by pmap_remove().
  Specifically, on mappings with PG_G set pmap_remove() not only performs
  the necessary per-page invlpg invalidations but also performs an
  unnecessary invalidation of the entire set of non-PG_G entries.

17 years agoMFC: -j 1.95 -j 1.96
rik [Sun, 25 Mar 2007 21:33:38 +0000 (21:33 +0000)]
MFC: -j 1.95 -j 1.96
  rik         2007-03-18 23:28:53 UTC

  FreeBSD src repository

  Modified files:
    sys/net              if_bridge.c
  Log:
  Give a chance for packet to appear with a correct input interfaces
  in case of multiple interfaces with the same MAC in the same bridge.
  This commit do not solve the entire problem. Only case where packet
  arrived from such interface.

  PR:     kern/109815
  MFC after:      7 days
  Submitted by:   Eygene Ryabinkin and rik@
  Discussed with: bms@, thompsa@, yar@

  Revision  Changes    Path
  1.96      +50 -30    src/sys/net/if_bridge.c

17 years agoMFCrevision 1.560
alc [Sun, 25 Mar 2007 20:53:45 +0000 (20:53 +0000)]
MFCrevision 1.560
  Don't invalidate the TLB in pmap_qenter() unless the old mapping was
  valid.  Most often, it isn't.

17 years agoo ommitted -> omitted.
maxim [Sun, 25 Mar 2007 18:31:45 +0000 (18:31 +0000)]
o ommitted -> omitted.

PR: docs/110809
Submitted by: naddy

17 years agoMFC: add acpi_hpet to the build (pointed out as missing by kib@)
njl [Sat, 24 Mar 2007 16:00:44 +0000 (16:00 +0000)]
MFC: add acpi_hpet to the build (pointed out as missing by kib@)

17 years agoo Revert last: addr2ascii() is still there.
maxim [Fri, 23 Mar 2007 21:56:06 +0000 (21:56 +0000)]
o Revert last: addr2ascii() is still there.

Pointed out by: bmah

17 years agoMFC rev. 1.75: kill addr2ascii() man page.
maxim [Fri, 23 Mar 2007 21:37:32 +0000 (21:37 +0000)]
MFC rev. 1.75: kill addr2ascii() man page.

17 years agoo Kill EOL whitespaces. HEAD is OK already.
maxim [Fri, 23 Mar 2007 21:35:32 +0000 (21:35 +0000)]
o Kill EOL whitespaces.  HEAD is OK already.

17 years agoMFC rev. 1.20: add ENVIRONMENT section.
maxim [Fri, 23 Mar 2007 21:03:19 +0000 (21:03 +0000)]
MFC rev. 1.20: add ENVIRONMENT section.

17 years agoMFC: always return ok when getting number of units
njl [Fri, 23 Mar 2007 20:02:39 +0000 (20:02 +0000)]
MFC: always return ok when getting number of units

17 years agoMFC: validity checking of _TMP values
njl [Fri, 23 Mar 2007 19:48:31 +0000 (19:48 +0000)]
MFC: validity checking of _TMP values

17 years agoMFC: acpi_dock (docking station) and acpi_hpet (high-perf timer) support
njl [Fri, 23 Mar 2007 19:45:53 +0000 (19:45 +0000)]
MFC: acpi_dock (docking station) and acpi_hpet (high-perf timer) support

17 years agoMFC revision 1.2
delphij [Fri, 23 Mar 2007 01:55:20 +0000 (01:55 +0000)]
MFC revision 1.2
date: 2007/03/16 03:50:53;  author: delphij;  state: Exp;  lines: +9 -1
Mention a limitation that was inherted from RFC1952, making
it impossible to obtain correct file size from a file that
is larger than 4GB before compression.

PR: bin/110329

17 years agoMFC r1.64:
emaste [Fri, 23 Mar 2007 01:00:40 +0000 (01:00 +0000)]
MFC r1.64:
  KERN_PROC_ALL produces a kinfo_proc for each thread in a process, which
  caused fstat to produce duplicated output for threaded processes.  Instead
  use KERN_PROC_PROC to get just one kinfo_proc per process.

17 years agoThis commit was manufactured by cvs2svn to create branch 'RELENG_6'.
cvs2svn [Thu, 22 Mar 2007 18:16:44 +0000 (18:16 +0000)]
This commit was manufactured by cvs2svn to create branch 'RELENG_6'.

17 years agoCarefully merge revs 1.123, 1.124, omitting the 1.120 change. This
glebius [Thu, 22 Mar 2007 10:55:13 +0000 (10:55 +0000)]
Carefully merge revs 1.123, 1.124, omitting the 1.120 change. This
should fix resets of the long living TCP connections with
EHOSTDOWN message.

PR: 100172
Submitted by: "Andrew - Supernews" <andrew supernews.net>

17 years agoMFC e1000phy.c rev. 1.17, 1.18 to RELENG_6
yongari [Thu, 22 Mar 2007 04:05:17 +0000 (04:05 +0000)]
MFC e1000phy.c rev. 1.17, 1.18 to RELENG_6

rev. 1.17
  o style(9) fixes

rev. 1.18
  o Add support code for newer Marvell PHYs.
  o Remove unused static global variable e1000phy_debug.
  o Take advantage of mii_phy_dev_probe().
  o Use MII_ANEGTICKS/MII_ANEGTICKS_GIGE instead of magic number 5.
  o Add IFM_NONE as e1000phy(4) supports it without issues.
  o Nuke magic PHY programming sequence in PHY reset and follow correct
    reset sequence. [1]
  o Make manual media selection work for all supported media types.
  o Don't set MIIF_NOISOLATE so e1000phy(4) can be used in
    configurations with multiple PHYs.
  o In 1000baseT, when setting the link manually, one side must be the
    master and the other the slave. If LINK0 is set, program the PHY
    to be a master, otherwise it's a slave.
  o When we lost a link, reset mii_ticks immediately so it correctly
    check number of seconds elapsed in autonegotiation phase.
  o Announce link loss right after it happens.
  o After kicking autonegotiation, report PHY status instead of
    returning immediatly.
  o When link state check is in progress, check auto negotiation
    completion bit only when auto negotiation is enbaled.
  o When PHY is resolved to a master, show it with IFM_FLAG2.

17 years agoMFC e1000phyreg.h, rev. 1.4 to RELENG_6
yongari [Thu, 22 Mar 2007 03:59:04 +0000 (03:59 +0000)]
MFC e1000phyreg.h, rev. 1.4 to RELENG_6
  Add additional Marvell PHY registers.

17 years agoMFC miidevs, rev. 1.37 to RELENG_6
yongari [Thu, 22 Mar 2007 03:57:00 +0000 (03:57 +0000)]
MFC miidevs, rev. 1.37 to RELENG_6
  Add a bunch of new Marvell PHY ids.

17 years agoMFC: rev 1.12
simokawa [Thu, 22 Mar 2007 02:47:23 +0000 (02:47 +0000)]
MFC: rev 1.12
Add a description about hw.firewire.hold_count.

17 years agoMFC: Sync bge(4) with -CURRENT (minus if_bge.c 1.185 and 1.186)
jkim [Wed, 21 Mar 2007 22:53:22 +0000 (22:53 +0000)]
MFC: Sync bge(4) with -CURRENT (minus if_bge.c 1.185 and 1.186)

- Support for BCM5754/5755/5786/5787.
- Support for IPMI/ASF mode (disabled by default)
- Fix lots of bugs including style(9) nits.

Tested by: net, stable

17 years agoFix previous commit due to libarchive 1.x vs libarchive 2.x differences.
cperciva [Wed, 21 Mar 2007 19:19:00 +0000 (19:19 +0000)]
Fix previous commit due to libarchive 1.x vs libarchive 2.x differences.

Submitted by: Henrik Nymann Jensen
Pointy hat to: cperciva

17 years agoMFC rev. 1.102-1.105
oleg [Wed, 21 Mar 2007 17:25:15 +0000 (17:25 +0000)]
MFC rev. 1.102-1.105

- style(9) cleanup
- Use separate thread for servicing dummynet(4).
  Utilize taskqueue(9) API.
- Convert
        net.inet.ip.dummynet.curr_time
        net.inet.ip.dummynet.searches
        net.inet.ip.dummynet.search_steps
  to SYSCTL_LONG nodes. It will prevent frequent wrap around on 64bit archs.
- Implement simple mechanics for dummynet(4) internal time correction.
  Under certain circumstances (system high load, dummynet lock contention, etc)
  dummynet's tick counter can be significantly slower than it should be.
  (I've observed up to 25% difference on one of my production servers).
  Since this counter used for packet scheduling, it's accuracy is vital for
  precise bandwidth limitation.

  Introduce new sysctl nodes:
  net.inet.ip.dummynet.
    tick_lost           - number of ticks coalesced by taskqueue thread.
    tick_adjustment     - number of time corrections done.
    tick_diff           - adjusted vs non-adjusted tick counter difference
    tick_delta          - last vs 'standard' tick differnece (usec).
    tick_delta_sum      - accumulated (and not corrected yet) time
                          difference (usec).
- Use non-recursive mutex. MTX_RECURSE is unnecessary since rev. 1.70
- Pay respect to net.isr.direct: use netisr_dispatch() instead of ip_input()
- purge_flow_set():
    Do not leak memory while purging queues which are not bound to pipe.

17 years agoMFC revision 1.13: Don't forget to increment the raw_position counter,
cperciva [Wed, 21 Mar 2007 15:26:02 +0000 (15:26 +0000)]
MFC revision 1.13: Don't forget to increment the raw_position counter,
even when operating in non-buffered mode.

17 years agoMFC revisions 1.5[6789]:
cperciva [Wed, 21 Mar 2007 15:15:57 +0000 (15:15 +0000)]
MFC revisions 1.5[6789]:
 * Remove unused pathlen argument from write_entry function.
 * Fix logic bug in deciding if do_chdir should be called.
 * Don't consider an lstat(2) failure to be an error.
 * Correct warning message when lstat(2) fails.

17 years agoMFC: power save fix:
sam [Wed, 21 Mar 2007 05:01:44 +0000 (05:01 +0000)]
MFC: power save fix:

When dispatching frames saved on the power save queue to a
station exiting power save mode prepend them to the driver's
send q instead of appending them.  This insures the packets
are not misordered wrt any packets already q'd for the station.

This corrects a problem noticed when using a VoIP phone talking
to an ath card in ap mode; the misordered packets caused noise.

Submitted by: "J.R. Oldroyd" <jr@opal.com>

17 years agoMFC 1.79: when starting up an ibss master use a random address for the bssid
sam [Wed, 21 Mar 2007 04:56:38 +0000 (04:56 +0000)]
MFC 1.79: when starting up an ibss master use a random address for the bssid

Obtained from: Atheros

17 years agoMFi386: revision 1.13.8.4.
nyan [Wed, 21 Mar 2007 03:41:25 +0000 (03:41 +0000)]
MFi386: revision 1.13.8.4.

    MFC: Fix the cdboot twiddle display.

17 years agoSync with HEAD.
bms [Wed, 21 Mar 2007 00:45:13 +0000 (00:45 +0000)]
Sync with HEAD.

17 years agoMFC rev 1.53:
bms [Tue, 20 Mar 2007 23:24:25 +0000 (23:24 +0000)]
MFC rev 1.53:
  Remove obsolete polling members from documentation for struct ifnet.

Submitted by: Aniruddha Bohra

17 years agoMFC: Don't free the buffer with the sysctl value before printing it.
brian [Tue, 20 Mar 2007 23:15:45 +0000 (23:15 +0000)]
MFC: Don't free the buffer with the sysctl value before printing it.

17 years agoMFC: r1.18: Allow overriding the HOME environment as per the man page.
will [Tue, 20 Mar 2007 20:22:18 +0000 (20:22 +0000)]
MFC: r1.18: Allow overriding the HOME environment as per the man page.

17 years agoMFC:
bms [Mon, 19 Mar 2007 22:57:58 +0000 (22:57 +0000)]
MFC:
  In regular forwarding path, reject packets destined for 169.254.0.0/16
  link-local addresses. See RFC 3927 section 2.7.

  Revision  Changes    Path
  1.96      +1 -1      src/sys/netinet/in.c
  1.324     +6 -0      src/sys/netinet/ip_input.c

17 years agoMFC:
brueffer [Mon, 19 Mar 2007 21:42:39 +0000 (21:42 +0000)]
MFC:

- Add cxgb(4) entry, remove stale wx(4) entry
- (gigabit|fast) ethernet -> (Gigabit|Fast) Ethernet

Approved by: rwatson (mentor)

17 years agoMFC: Default output to stdout as the man page suggests.
brian [Mon, 19 Mar 2007 19:16:49 +0000 (19:16 +0000)]
MFC: Default output to stdout as the man page suggests.
     Only go through our cylinder group and inode info when we need to.

17 years agoMFC: Reindent main()
brian [Mon, 19 Mar 2007 19:15:42 +0000 (19:15 +0000)]
MFC: Reindent main()

17 years agoMFC: Account for di_blocks allocations when IN_SPACECOUNTED is set in an
brian [Mon, 19 Mar 2007 18:56:40 +0000 (18:56 +0000)]
MFC: Account for di_blocks allocations when IN_SPACECOUNTED is set in an
     inode's i_flag.  This fixes an eventual df/du discrepency resulting
     from fs_pendingblocks being reduced to less than zero.

17 years agoMFC:
brueffer [Mon, 19 Mar 2007 16:56:48 +0000 (16:56 +0000)]
MFC:

my(4) doesn't need miibus(4).

Approved by: rwatson (mentor)

17 years agoMFC: Various changes to sync this manpage to HEAD.
bmah [Mon, 19 Mar 2007 16:22:56 +0000 (16:22 +0000)]
MFC:  Various changes to sync this manpage to HEAD.

cxgb.4 1.4 -> 1.7

Discussed with: brueffer

17 years agoMFC: npe(4).
bmah [Mon, 19 Mar 2007 15:26:24 +0000 (15:26 +0000)]
MFC:  npe(4).

New release note:  freebsd-update(8) SMP kernel update fix.

17 years agoMFC if_ste.c rev 1.98, if_stereg.h rev 1.22 to RELENG_6.
yongari [Mon, 19 Mar 2007 06:58:47 +0000 (06:58 +0000)]
MFC if_ste.c rev 1.98, if_stereg.h rev 1.22 to RELENG_6.
  Add new ST201 PCI id.

17 years agoMFC: Change mtx's to use the formulated name as type so witness does not
sam [Mon, 19 Mar 2007 05:34:30 +0000 (05:34 +0000)]
MFC: Change mtx's to use the formulated name as type so witness does not
     complain on nested tx q lock acquisitions when processing the cab q.

17 years agoMFC 1.163: Kick tx after processing rx'd frames
sam [Mon, 19 Mar 2007 05:32:04 +0000 (05:32 +0000)]
MFC 1.163: Kick tx after processing rx'd frames

Submitted by: "J.R. Oldroyd" <jr@opal.com>

17 years agoMFC 1.46: correct conversions between TU and ms/ticks
sam [Mon, 19 Mar 2007 05:27:18 +0000 (05:27 +0000)]
MFC 1.46: correct conversions between TU and ms/ticks

Submitted by: sephe

17 years agoMFC: rev 1.86 and 1.87
simokawa [Mon, 19 Mar 2007 05:13:21 +0000 (05:13 +0000)]
MFC: rev 1.86 and 1.87
Support MAXPHYS up to 512KB

17 years agoMFC: rev 1.57
simokawa [Mon, 19 Mar 2007 05:11:07 +0000 (05:11 +0000)]
MFC: rev 1.57
Don't mess with PCIM_CMD_SERRESPEN and PCIM_CMD_PERRESPEN.
This will fix 'NMI RAM parity error' while booting on some machines.

PR: kern/95077

17 years agoThe driver would hang during attach on RELENG_6 - by "fixing" the issue
kmacy [Mon, 19 Mar 2007 04:50:15 +0000 (04:50 +0000)]
The driver would hang during attach on RELENG_6 - by "fixing" the issue
by adding debug print statements I've come to the conclusion that too little
time is being allowed to pass between register updates. Increasing the DELAY
interval by a constant interval eliminates this issue on my hardware.

17 years agoNew errata: EN-07:05.freebsd-update.
bmah [Sun, 18 Mar 2007 19:42:24 +0000 (19:42 +0000)]
New errata:  EN-07:05.freebsd-update.

17 years agoMFC rev 1.9: quiet down the errors when all fs are read-only
njl [Sun, 18 Mar 2007 05:34:47 +0000 (05:34 +0000)]
MFC rev 1.9: quiet down the errors when all fs are read-only

17 years agomove inline function before use so that -O works
kmacy [Sat, 17 Mar 2007 05:25:28 +0000 (05:25 +0000)]
move inline function before use so that -O works

17 years agoMFC: r1.67: Make error messages regarding -t and -l consistent.
will [Sat, 17 Mar 2007 02:10:14 +0000 (02:10 +0000)]
MFC: r1.67: Make error messages regarding -t and -l consistent.

17 years agoMFC: r1.316, accept valid RST packet when delayed ack is in effect.
qingli [Fri, 16 Mar 2007 22:04:25 +0000 (22:04 +0000)]
MFC: r1.316, accept valid RST packet when delayed ack is in effect.

17 years agoMFC rev. 1.16: fix prototypes.
maxim [Fri, 16 Mar 2007 21:35:13 +0000 (21:35 +0000)]
MFC rev. 1.16: fix prototypes.

PR: docs/110012

17 years agoMFC rev. 1.71, 1.74:
yar [Fri, 16 Mar 2007 15:34:09 +0000 (15:34 +0000)]
MFC rev. 1.71, 1.74:

Prepend ${_chroot} to pathnames where appropriate,
and do that properly.

17 years agoMFC:
ariff [Fri, 16 Mar 2007 13:55:34 +0000 (13:55 +0000)]
MFC:

Add AC97 inverted external amplifier quirk for ASUS A6R laptop.

PR: kern/110244

17 years agoMFC:
brueffer [Thu, 15 Mar 2007 13:46:08 +0000 (13:46 +0000)]
MFC:

New release notes: vge(4) altq support, cxgb(4) added.

17 years agoMFC:
brueffer [Thu, 15 Mar 2007 09:17:01 +0000 (09:17 +0000)]
MFC:

Autogenerate the hardware list for cxgb(4).

17 years agoMFC fixes to regexp(3) from OpenBSD, this includes:
delphij [Thu, 15 Mar 2007 05:51:24 +0000 (05:51 +0000)]
MFC fixes to regexp(3) from OpenBSD, this includes:
engine.c: 1.19+1.20
grot/tests: 1.3

17 years agohook cxgbtool into buildworld
kmacy [Thu, 15 Mar 2007 04:05:33 +0000 (04:05 +0000)]
hook cxgbtool into buildworld
update include names

17 years agoAdd cxgb man page to build
kmacy [Thu, 15 Mar 2007 03:52:23 +0000 (03:52 +0000)]
Add cxgb man page to build

17 years agoadd cxgb management tool cxgbtool
kmacy [Thu, 15 Mar 2007 03:21:32 +0000 (03:21 +0000)]
add cxgb management tool cxgbtool

17 years agohook cxgb into LINT build
kmacy [Thu, 15 Mar 2007 03:18:59 +0000 (03:18 +0000)]
hook cxgb into LINT build

17 years agohook cxgb into modules build
kmacy [Thu, 15 Mar 2007 03:13:49 +0000 (03:13 +0000)]
hook cxgb into modules build

17 years agoMFC Chelsio T3 10 Gigabit Ethernet support
kmacy [Thu, 15 Mar 2007 03:06:32 +0000 (03:06 +0000)]
MFC Chelsio T3 10 Gigabit Ethernet support

Don't hook into build just

17 years agoMFC: 1.163
rodrigc [Thu, 15 Mar 2007 00:11:56 +0000 (00:11 +0000)]
MFC: 1.163
  Add "force" to ext2_ops, to match what was in the old mount_ext2fs binary.

  Reported by:    Ivan Voras <ivoras fer hr>

17 years agoOnly enter the debugger on a Fatal op if this is a debug build of the
njl [Wed, 14 Mar 2007 19:50:07 +0000 (19:50 +0000)]
Only enter the debugger on a Fatal op if this is a debug build of the
acpi module.  Also clean up print of args a little.

17 years agoThis commit was manufactured by cvs2svn to create branch 'RELENG_6'.
cvs2svn [Wed, 14 Mar 2007 18:20:37 +0000 (18:20 +0000)]
This commit was manufactured by cvs2svn to create branch 'RELENG_6'.

17 years agoMFC:
bms [Wed, 14 Mar 2007 16:44:31 +0000 (16:44 +0000)]
MFC:
  AF_LINK support for getnameinfo().

Obtained from: NetBSD

17 years agoMark addr2ascii() and ascii2addr() clearly as being deprecated.
bms [Wed, 14 Mar 2007 16:31:16 +0000 (16:31 +0000)]
Mark addr2ascii() and ascii2addr() clearly as being deprecated.

17 years agoMFC:
yar [Wed, 14 Mar 2007 14:19:42 +0000 (14:19 +0000)]
MFC:

  As suggested more than once in the lists, drop -M from flags to mfs
  for /tmp and /var.  This makes the memory discs swap-backed instead
  of malloc-backed.  A swap-backed memory disc should not be worse
  than a malloc-backed one in any scenario because it will start
  touching swap only when needed.  OTOH, a malloc-backed disc can
  starve limited kernel resources and evenually crash the system.

  Reflect the change in the rc.conf(5) manpage.  Also stop telling
  lies there about softupdates: it does not waste disc space, it
  just can delay its freeing.

src/etc/defaults/rc.conf 1.306
src/share/man/man5/rc.conf.5 1.317

17 years agoMFC rev 1.4:
bms [Wed, 14 Mar 2007 14:10:22 +0000 (14:10 +0000)]
MFC rev 1.4:
  Update host-mode multicast group information output.
   Display IPv4 and IPv6 memberships separately.
    Obey the MK_INET6_SUPPORT flag.
   Display link-layer memberships.
    Use addr2ascii() to correctly print non-IEEE 802 sockaddr_dl instances.
   Eliminate redundant switch..case blocks.
   Update copyright.
   Misc style changes.

17 years agoMFC rev 1.203:
bms [Tue, 13 Mar 2007 22:12:23 +0000 (22:12 +0000)]
MFC rev 1.203:
  Fix IP_SENDSRCADDR semantics.

   * To use this option with a UDP socket, it must be bound to a local port,
     and INADDR_ANY, to disallow possible collisions with existing udp inpcbs
     bound to the same port on other interfaces at send time.

   * If the socket is bound to INADDR_ANY, specifying IP_SENDSRCADDR with
     INADDR_ANY will be rejected as it is ambiguous.

   * If the socket is bound to an address other than INADDR_ANY, specifying
     IP_SENDSRCADDR with INADDR_ANY will be disallowed by in_pcbbind_setup().

Reviewed by: silence on -net
Tested with: src/tools/regression/netinet/ipbroadcast

17 years agoMFC rev 1.159:
bms [Tue, 13 Mar 2007 21:59:20 +0000 (21:59 +0000)]
MFC rev 1.159:
  Comply with RFC 3927, by forcing ARP replies which contain a source
  address within the link-local IPv4 prefix 169.254.0.0/16, to be
  broadcast at link layer.

Reviewed by: fenner

17 years agoMFC rev 1.38:
bms [Tue, 13 Mar 2007 21:56:18 +0000 (21:56 +0000)]
MFC rev 1.38:
  Document SO_ACCEPTCONN.

Submitted by: Vlad GALU (with changes)

17 years agoMFC mount_nfs.c rev. 1.72, mount_nfs.8 rev. 1.47:
thomas [Tue, 13 Mar 2007 20:28:24 +0000 (20:28 +0000)]
MFC mount_nfs.c rev. 1.72, mount_nfs.8 rev. 1.47:
Add "fg" option as antonym to "bg"; add "hard" option as antonym to "soft".
This is for better compatibility with other environments (Linux, Solaris,
HP-UX, AIX and Tru64 support these options).

PR: bin/109924

17 years agoMerge Peter's fix from rev 1.14 back to RELENG_6. This line of code
joerg [Mon, 12 Mar 2007 21:41:01 +0000 (21:41 +0000)]
Merge Peter's fix from rev 1.14 back to RELENG_6.  This line of code
appears to have been accidentally committed in rev 1.11, and was never
meant to go into any official code.

17 years agoMFC: Don't reject file descriptors higher than FD_SETSIZE when using
jhb [Mon, 12 Mar 2007 19:51:57 +0000 (19:51 +0000)]
MFC: Don't reject file descriptors higher than FD_SETSIZE when using
kevent(2).

17 years agoMFC: Add a simple device driver to "eat" any I/O APICs that show up as PCI
jhb [Mon, 12 Mar 2007 19:50:29 +0000 (19:50 +0000)]
MFC: Add a simple device driver to "eat" any I/O APICs that show up as PCI
devices.

17 years agoMFC: Don't block on the socket zone limit during the socket()
ru [Mon, 12 Mar 2007 12:13:53 +0000 (12:13 +0000)]
MFC: Don't block on the socket zone limit during the socket()
syscall which can lock up a system otherwise; instead, return
ENOBUFS as documented, which matches the FreeBSD 4.x behavior.

17 years agoMerge rev 1.16 and 1.18 to fix I/O problems under high load. Thanks to
scottl [Mon, 12 Mar 2007 07:43:07 +0000 (07:43 +0000)]
Merge rev 1.16 and 1.18 to fix I/O problems under high load.  Thanks to
Areca for working on this.

Submitted by: Erich Chen

17 years agoUse 2K Login firmware for 2300s.
mjacob [Mon, 12 Mar 2007 05:28:29 +0000 (05:28 +0000)]
Use 2K Login firmware for 2300s.

17 years agoMFC (revision 1.31)
ariff [Mon, 12 Mar 2007 02:03:25 +0000 (02:03 +0000)]
MFC (revision 1.31)

Fix long standing multi playback/recording issues, caused by
excessive interrupt clock timer reset, screwing interrupt generation
for already active channels. Track moving DMA pointer and call buffer
interrupt on each blocksize boundary.

PR: kern/109791

17 years agoMFC 1.32: Support entries up to 4G.
kientzle [Sun, 11 Mar 2007 20:19:45 +0000 (20:19 +0000)]
MFC 1.32: Support entries up to 4G.

17 years agoMFC 1.22: Re-enable archiving of ACLs
kientzle [Sun, 11 Mar 2007 19:46:56 +0000 (19:46 +0000)]
MFC 1.22:  Re-enable archiving of ACLs

17 years agoMFC: Don't hard-code the nfs root socket as SOCK_DGRAM. This is currently
kris [Sun, 11 Mar 2007 19:44:52 +0000 (19:44 +0000)]
MFC: Don't hard-code the nfs root socket as SOCK_DGRAM.  This is currently
a NOP in 6.x but this may change if further code is merged from 7.0.

17 years agomdmfs.c: MFC revs 1.29-1.31
matteo [Sun, 11 Mar 2007 18:24:22 +0000 (18:24 +0000)]
mdmfs.c: MFC revs 1.29-1.31
mdmfs.8: MFC revs 1.27-1.29

PR: bin/66763 103501 kern/109863

17 years agoMFC src/lib/libc/Makefile (1.69) src/lib/libc/rpc/clnt_dg.c (1.19):
simon [Sun, 11 Mar 2007 10:15:49 +0000 (10:15 +0000)]
MFC src/lib/libc/Makefile (1.69) src/lib/libc/rpc/clnt_dg.c (1.19):

Disable RPC exponential back-off for FreeBSD.org systems (IE. hidden
behind _FREEFALL_CONFIG).  This is done mainly to make NIS even more
resistant to packet loss.

This is not enabled by default for "normal" FreeBSD since it might cause
the server providing the RPC service to be hit heavily with RPC traffic
in case of problems.  freefall.FreeBSD.org and hub.FreeBSD.org have been
running with a patch similar to this for a couple of weeks.

MFC after: 1 week
Discussed with: peter

17 years agoMFC: 1.38: Fix interface output being stuck under low memory conditions.
ru [Sun, 11 Mar 2007 10:11:16 +0000 (10:11 +0000)]
MFC: 1.38: Fix interface output being stuck under low memory conditions.

17 years agoMFC 1.50:
simon [Sun, 11 Mar 2007 09:59:51 +0000 (09:59 +0000)]
MFC 1.50:

- Bump _yplib_timeout limit from 10 to 20 seconds to better handle
  packet loss when talking to a NIS server.
- Set 1 second retry timeout to further realistically handle UDP
  packet loss for yp_next packet bursts.  If the packet hasn't come
  back within 1 second its rather unlikely to come back at all.  There
  is still back-off mechanism in RPC so if there is another reason
  than packet loss for the lack of response within 1 second, the NIS
  server will not be totally bombarded with requests.

This reduces the risk of NIS failing with:

yp_next: clnt_call: RPC: Timed out

considerably.  This is mainly a problem if you have larger NIS maps
(like at FreeBSD.org) since enumerations of the lists will cause a UDP
packet bursts where a few packets being lost once in a while do
happen.

Discussed with: peter
Problem mainly diagnosed by:    peter

17 years agofeedback from RELENG_5 port
mjacob [Sun, 11 Mar 2007 01:59:44 +0000 (01:59 +0000)]
feedback from RELENG_5 port

17 years agoMFC header/license cleanup. No effect on code.
mjacob [Sun, 11 Mar 2007 01:51:20 +0000 (01:51 +0000)]
MFC header/license cleanup. No effect on code.

17 years agoMFC from revision 1.165
mckusick [Sun, 11 Mar 2007 00:56:17 +0000 (00:56 +0000)]
MFC from revision 1.165

17 years agoMFC rev. 1.100:
yar [Sat, 10 Mar 2007 14:29:31 +0000 (14:29 +0000)]
MFC rev. 1.100:

Typo fix: preceed -> precede.

17 years agoMFC rev. 1.9:
yar [Sat, 10 Mar 2007 14:20:34 +0000 (14:20 +0000)]
MFC rev. 1.9:

Fix a typo in a warning message.

17 years agoMFC rev. 1.9-1.11:
yar [Sat, 10 Mar 2007 14:17:27 +0000 (14:17 +0000)]
MFC rev. 1.9-1.11:

Better handling for unset or null $hostname, which can appear
in a system just installed, or configured via DHCP.

17 years agoMFC rev. 1.26:
yar [Sat, 10 Mar 2007 14:07:01 +0000 (14:07 +0000)]
MFC rev. 1.26:

pkill(1) and pgrep(1) have been moved to /bin so that they are
available to rc.d scripts early in the boot sequence.

17 years agoMFC rev. 1.10:
yar [Sat, 10 Mar 2007 14:04:01 +0000 (14:04 +0000)]
MFC rev. 1.10:

Improve rc.d conformance:
- don't play a needless trick with prestart, just use start method;
- provide no-op stop method so that we don't get bogus "abi not running" error.

17 years agoMFC rev. 1.18:
yar [Sat, 10 Mar 2007 14:00:51 +0000 (14:00 +0000)]
MFC rev. 1.18:

Syscons cannot be stopped, so provide a no-op stop method.
The default stop method from rc.subr isn't suited for this
case and produces a bogus warning: "syscons not running".

17 years agoMFC rev. 1.73: Allow for interpreted scripts using env(1)
yar [Sat, 10 Mar 2007 13:53:04 +0000 (13:53 +0000)]
MFC rev. 1.73: Allow for interpreted scripts using env(1)
in their shebang lines.

PR: bin/100287

17 years agoMFC: set the antenna switch when fixing the tx antenna using the
sam [Sat, 10 Mar 2007 00:53:09 +0000 (00:53 +0000)]
MFC: set the antenna switch when fixing the tx antenna using the
     dev.ath.X.txantenna sysctl