]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
11 years agoPorts switched from dialog to dialog4ports some time ago.
Eitan Adler [Thu, 20 Jun 2013 21:16:46 +0000 (21:16 +0000)]
Ports switched from dialog to dialog4ports some time ago.

PR: docs/179785
Repored by: Kevin Oberman <rkoberman@gmail.com>
Submitted by: "Ilya A. Arkhipov" <rum1cro@yandex.ru>
MFC After: 3 days

11 years ago - Add a per-zone lock for zones without kegs.
Jeff Roberson [Thu, 20 Jun 2013 19:08:12 +0000 (19:08 +0000)]
 - Add a per-zone lock for zones without kegs.
 - Be more explicit about zone vs keg locking.  This functionally changes
   almost nothing.
 - Add a size parameter to uma_zcache_create() so we can size the buckets.
 - Pass the zone to bucket_alloc() so it can modify allocation flags
   as appropriate.
 - Fix a bug in zone_alloc_bucket() where I missed an address of operator
   in a failure case.  (Found by pho)

Sponsored by: EMC / Isilon Storage Division

11 years agoPull in r183984 from llvm trunk:
Dimitry Andric [Thu, 20 Jun 2013 18:25:10 +0000 (18:25 +0000)]
Pull in r183984 from llvm trunk:

  Make PrologEpilogInserter save/restore all callee saved registers in
  functions which call __builtin_unwind_init()

  __builtin_unwind_init() is an undocumented gcc intrinsic which has
  this effect, and is used in libgcc_eh.

  Goes part of the way toward fixing PR8541.

This obsoletes the ugly hack to libgcc's unwind code from r245272, and
should also work for other arches, so revert the hack too.

11 years agoDocument RA_RECURSED and RA_NOTRECURSED.
John Baldwin [Thu, 20 Jun 2013 17:26:25 +0000 (17:26 +0000)]
Document RA_RECURSED and RA_NOTRECURSED.

MFC after: 3 days

11 years agoeturn -1 when the specified backing store file is not found in the md
Hiroki Sato [Thu, 20 Jun 2013 17:01:02 +0000 (17:01 +0000)]
eturn -1 when the specified backing store file is not found in the md
device list.

MFC after: 1 week

11 years ago- Add "-f file" support to listing mode (-l). When a -f option is
Hiroki Sato [Thu, 20 Jun 2013 14:30:16 +0000 (14:30 +0000)]
- Add "-f file" support to listing mode (-l).  When a -f option is
  specified, only md(4) devices which have the specified file as backing
  store are displayed.
- Use MD_NAME instead of "md".
- Use _PATH_DEV instead of "/dev/".

MFC after: 1 week

11 years agoAllow immediate operand.
Konstantin Belousov [Thu, 20 Jun 2013 14:30:04 +0000 (14:30 +0000)]
Allow immediate operand.

Sponsored by: The FreeBSD Foundation

11 years agoUse corresponding macros to update statistics for AH, ESP, IPIP, IPCOMP,
Andrey V. Elsukov [Thu, 20 Jun 2013 11:44:16 +0000 (11:44 +0000)]
Use corresponding macros to update statistics for AH, ESP, IPIP, IPCOMP,
PFKEY.

MFC after: 2 weeks

11 years agoUse IPSECSTAT_INC() and IPSEC6STAT_INC() macros for ipsec statistics
Andrey V. Elsukov [Thu, 20 Jun 2013 09:55:53 +0000 (09:55 +0000)]
Use IPSECSTAT_INC() and IPSEC6STAT_INC() macros for ipsec statistics
accounting.

MFC after: 2 weeks

11 years agoFix a code typo in a case-statement match expression that prevented IPv6
Devin Teske [Thu, 20 Jun 2013 07:29:42 +0000 (07:29 +0000)]
Fix a code typo in a case-statement match expression that prevented IPv6
URLs with port designator from working properly (e.g. http://[::1]:80/).

11 years agoWhen the fall-back of a case-statement is the last thing executed in a
Devin Teske [Thu, 20 Jun 2013 05:51:44 +0000 (05:51 +0000)]
When the fall-back of a case-statement is the last thing executed in a
while-loop _and_ all prior matches in the same case-statement either break
or continue, we can safely break the fall-back out of the case-statement.
This should improve readability and allow for longer-lines by reducing the
level of indentation by-one for the fall-back case.
(a continuation of SVN r252019)

11 years agoWhen the fall-back of a case-statement is the last thing executed in a
Devin Teske [Thu, 20 Jun 2013 05:48:08 +0000 (05:48 +0000)]
When the fall-back of a case-statement is the last thing executed in a
while-loop _and_ all prior matches in the same case-statement either break
or continue, we can safely break the fall-back out of the case-statement.
This should improve readability and allow for longer-lines by reducing the
level of indentation by-one for the fall-back case.

11 years agoRemove pedanticism and consolidate some logic.
Devin Teske [Thu, 20 Jun 2013 05:42:21 +0000 (05:42 +0000)]
Remove pedanticism and consolidate some logic.

11 years agoAdd debugging (for a case that shouldn't arise, but makes it more obvious
Devin Teske [Thu, 20 Jun 2013 05:40:11 +0000 (05:40 +0000)]
Add debugging (for a case that shouldn't arise, but makes it more obvious
if a menu addition is made in one plce but forgotten in another).

11 years ago- Add CIDR notation support like 192.168.1-2.10-16/24 to $ifconfig_IF_aliasN.
Hiroki Sato [Thu, 20 Jun 2013 02:29:49 +0000 (02:29 +0000)]
- Add CIDR notation support like 192.168.1-2.10-16/24 to $ifconfig_IF_aliasN.
  This is an extended version of ipv4_addr_IF which supports both IPv4 and
  IPv6, and multiple range specifications.  To avoid to generate too many
  addresses, the maximum number of the generated addresses is currently
  limited to 31.

- Add $ifconfig_IF_aliases, which accepts multiple IP aliases in a variable.

- ipv6_prefix_IF now supports !/64 prefix length.  In addition to the old
  64-bit format (2001:db8:1:1), a full 128-bit format like 2001:db8:1:1::/64
  is supported.

- Replace ifconfig command with $IFCONFIG_CMD variable to support
  a dry-run mode in the future.

- Remove IP aliases before removing all of IPv4 addresses when doing
  "rc.d/netif down".

- Add a DAD wait to network6_getladdr() because it is possible to fail to
  configure an EUI64 address when ipv6_prefix_IF is specified.

A summary of the supported ifconfig_* variables is as follows:

 # IPv4 configuration.
 ifconfig_em0="inet 192.168.0.1"
 # IPv6 configuration.
 ifconfig_em0_ipv6="inet6 2001:db8::1/64"
 # IPv4 address range spec.  Now deprecated.
 ipv4_addr_em0="10.2.1.1-10"
 # IPv6 alias.
 ifconfig_em0_alias0="inet6 2001:db8:5::1 prefixlen 70"
 # IPv4 alias.
 ifconfig_em0_alias1="inet 10.2.2.1/24"
 # IPv4 alias with range spec w/o AF keyword (backward compat).
 ifconfig_em0_alias2="10.3.1.1-10/32"
 # IPv6 alias with range spec.
 ifconfig_em0_alias3="inet6 2001:db8:20-2f::1/64"
 # ifconfig_IF_aliases is just like ifconfig_IF_aliasN.
 ifconfig_em0_aliases="inet 10.3.3.201-204/24 inet6 2001:db8:210-213::1/64 inet 10.1.1.1/24"
 # IPv6 alias (backward compat)
 ipv6_ifconfig_em0_alias0="inet6 2001:db8:f::1/64"
 # IPv6 alias w/o AF keyword (backward compat)
 ipv6_ifconfig_em0_alias1="2001:db8:f:1::1/64"
 # IPv6 prefix.
 ipv6_prefix_em0="2001:db8::/64"

Tested by: Kimmo Paasiala

11 years agoOnly enable svn on amd64/ia64/sparc64/i386.
Peter Wemm [Thu, 20 Jun 2013 02:26:32 +0000 (02:26 +0000)]
Only enable svn on amd64/ia64/sparc64/i386.

11 years agoTry to fix build of apr on FreeBSD/arm.
Tim Kientzle [Thu, 20 Jun 2013 02:04:03 +0000 (02:04 +0000)]
Try to fix build of apr on FreeBSD/arm.

11 years agoRename some prefixes in the Block Group Descriptor fields to ext4bgd_
Pedro F. Giffuni [Thu, 20 Jun 2013 00:00:33 +0000 (00:00 +0000)]
Rename some prefixes in the Block Group Descriptor fields to ext4bgd_

Change prefix to avoid confusion and denote that these fields
are generally only available starting with ext4.

MFC after: 3 days

11 years agoFix a mystery cut-n-paste corruption from the previous commit.
Scott Long [Wed, 19 Jun 2013 23:09:10 +0000 (23:09 +0000)]
Fix a mystery cut-n-paste corruption from the previous commit.

Submitted by: Brenden Fabeny

11 years agoMark geom_mirror as capable of unmapped i/o
Scott Long [Wed, 19 Jun 2013 21:52:32 +0000 (21:52 +0000)]
Mark geom_mirror as capable of unmapped i/o

Obtained from: Netflix
MFC after: 3 days

11 years agoUse PIM6STAT_INC() and MRT6STAT_INC() macros for IPv6 multicast
Andrey V. Elsukov [Wed, 19 Jun 2013 21:50:17 +0000 (21:50 +0000)]
Use PIM6STAT_INC() and MRT6STAT_INC() macros for IPv6 multicast
statistics accounting.

MFC after: 2 weeks

11 years agoUse RIP6STAT_INC() macro for raw ip6 statistics accounting.
Andrey V. Elsukov [Wed, 19 Jun 2013 20:48:34 +0000 (20:48 +0000)]
Use RIP6STAT_INC() macro for raw ip6 statistics accounting.

MFC after: 2 weeks

11 years agoFix a couple of typos that broke buildworld for me.
Don Lewis [Wed, 19 Jun 2013 19:44:57 +0000 (19:44 +0000)]
Fix a couple of typos that broke buildworld for me.

Reviewed by: peter

11 years agoStyle -- no ;; needed on fallback clause within case-statement.
Devin Teske [Wed, 19 Jun 2013 18:44:55 +0000 (18:44 +0000)]
Style -- no ;; needed on fallback clause within case-statement.

11 years agoWhitespace.
Devin Teske [Wed, 19 Jun 2013 18:32:18 +0000 (18:32 +0000)]
Whitespace.

11 years agoWhitespace and comments.
Devin Teske [Wed, 19 Jun 2013 18:13:58 +0000 (18:13 +0000)]
Whitespace and comments.

11 years agoAlphabetize reserved-word (resword) registration.
Devin Teske [Wed, 19 Jun 2013 17:14:59 +0000 (17:14 +0000)]
Alphabetize reserved-word (resword) registration.

11 years agoAlphabetize includes.
Devin Teske [Wed, 19 Jun 2013 17:13:16 +0000 (17:13 +0000)]
Alphabetize includes.

11 years agoUse ICMP6STAT_INC() macro for ICMPv6 errors accounting.
Andrey V. Elsukov [Wed, 19 Jun 2013 15:59:21 +0000 (15:59 +0000)]
Use ICMP6STAT_INC() macro for ICMPv6 errors accounting.

MFC after: 2 weeks

11 years agoFix an unfortunate typo with the compat shims
Scott Long [Wed, 19 Jun 2013 05:11:30 +0000 (05:11 +0000)]
Fix an unfortunate typo with the compat shims

Obtained from: Netflix
MFC after: now

11 years agoSome clarifications and updates for the comments, mostly retrieved
Konstantin Belousov [Wed, 19 Jun 2013 05:05:16 +0000 (05:05 +0000)]
Some clarifications and updates for the comments, mostly retrieved
from Bruce Evans.  Trim the trailing spaces.

MFC after: 1 week

11 years agoThe SUSv4tc1 requires that pthread_setcancelstate() shall be not a
Konstantin Belousov [Wed, 19 Jun 2013 04:47:41 +0000 (04:47 +0000)]
The SUSv4tc1 requires that pthread_setcancelstate() shall be not a
cancellation point.  When enabling the cancellation, only process the
pending cancellation for asynchronous mode.

Reported and reviewed by: Kohji Okuno <okuno.kohji@jp.panasonic.com>
Sponsored by: The FreeBSD Foundation
MFC after: 1 week

11 years agoWhen a previous call to sbsndptr() leaves sb->sb_sndptroff at the start of an
Lawrence Stewart [Wed, 19 Jun 2013 03:08:01 +0000 (03:08 +0000)]
When a previous call to sbsndptr() leaves sb->sb_sndptroff at the start of an
mbuf that was fully consumed by the previous call, the mbuf ptr returned by the
current call ends up being the previous mbuf in the sb chain to the one that
contains the data we want.

This does not cause any observable issues because the mbuf copy routines happily
walk the mbuf chain to get to the data at the moff offset, which in this case
means they effectively skip over the mbuf returned by sbsndptr().

We can't adjust sb->sb_sndptr during the previous call for this case because the
next mbuf in the chain may not exist yet. We therefore need to detect the
condition and make the adjustment during the current call.

Fix by detecting the special case of moff being at the start of the next mbuf in
the chain and adjust the required accounting variables accordingly.

Reviewed by: andre
MFC after: 2 weeks

11 years ago - Persist the caller's flags in the bucket allocation flags so we don't
Jeff Roberson [Wed, 19 Jun 2013 02:30:32 +0000 (02:30 +0000)]
 - Persist the caller's flags in the bucket allocation flags so we don't
   lose a M_NOVM when we recurse into a bucket allocation.

Sponsored by: EMC / Isilon Storage Division

11 years agoBandaid: mips doesn't seem to have the full set of atomics builtins. I
Peter Wemm [Wed, 19 Jun 2013 02:16:04 +0000 (02:16 +0000)]
Bandaid: mips doesn't seem to have the full set of atomics builtins.  I
will investigate more.

11 years agoRemove extra whitespace lines.
Devin Teske [Wed, 19 Jun 2013 00:24:08 +0000 (00:24 +0000)]
Remove extra whitespace lines.

11 years agoChange a humongous if-statement at the end of f_install_zoneinfo_file() into
Devin Teske [Wed, 19 Jun 2013 00:13:54 +0000 (00:13 +0000)]
Change a humongous if-statement at the end of f_install_zoneinfo_file() into
an early return, allowing a huge chunk of code to be indented one-level less

11 years agoTake advantage of newly updated f_dialog_{yesno,noyes}() functions from
Devin Teske [Wed, 19 Jun 2013 00:09:21 +0000 (00:09 +0000)]
Take advantage of newly updated f_dialog_{yesno,noyes}() functions from
SVN r251977 (adding an $hline parameter).

11 years agoWhen I first wrote the timezone module, it was in sysutils/tzdialog and it
Devin Teske [Wed, 19 Jun 2013 00:08:29 +0000 (00:08 +0000)]
When I first wrote the timezone module, it was in sysutils/tzdialog and it
pre-dates bsdconfig. Update the code to take advantage of f_dialog_msgbox().

11 years agoChange the f_dialog_{yesno,noyes}() function-arguments in `dialog.subr' to
Devin Teske [Tue, 18 Jun 2013 23:48:05 +0000 (23:48 +0000)]
Change the f_dialog_{yesno,noyes}() function-arguments in `dialog.subr' to
accomodate an $hline value for overriding the default. This change does
not effect any current modules as it turns out that not one single usage of
either f_dialog_yesno() or f_dialog_noyes() relied on accepting more than a
first argument (read: all occurrences quoted the first parameter; so $* was
never depended upon).

This will allow some custom invocations of --yesno and --noyes to roll over
to these functions (for example, in `timezone/timezone').

11 years agoFix a gcc warning uncovered after r251745.
Sergey Kandaurov [Tue, 18 Jun 2013 23:31:09 +0000 (23:31 +0000)]
Fix a gcc warning uncovered after r251745.

Reported by: Sergey V. Dyatko
Reviewed by: neel

11 years agoPerform some code consolidation and replace nested case-statements with
Devin Teske [Tue, 18 Jun 2013 23:18:32 +0000 (23:18 +0000)]
Perform some code consolidation and replace nested case-statements with
more logical if-else statements for each menu selection.

11 years agoFix a regression introduced by r251967, resulting in:
Devin Teske [Tue, 18 Jun 2013 23:14:45 +0000 (23:14 +0000)]
Fix a regression introduced by r251967, resulting in:

/usr/libexec/bsdconfig/140.startup/startup: 130:
Syntax error: ";;" unexpected (expecting "fi")

11 years agoWhitespace.
Devin Teske [Tue, 18 Jun 2013 23:11:36 +0000 (23:11 +0000)]
Whitespace.

11 years agoPerform some code consolidation and add some additional error
Devin Teske [Tue, 18 Jun 2013 21:44:35 +0000 (21:44 +0000)]
Perform some code consolidation and add some additional error
checking/reporting. (similar to SVN revisions 251919 and 251928)

11 years agoTeach delete-old how to clean up after WITH/WITHOUT_SVN{LITE} permutations.
Peter Wemm [Tue, 18 Jun 2013 21:41:06 +0000 (21:41 +0000)]
Teach delete-old how to clean up after WITH/WITHOUT_SVN{LITE} permutations.

11 years agoComments.
Devin Teske [Tue, 18 Jun 2013 21:35:38 +0000 (21:35 +0000)]
Comments.

11 years agoAdd quad port probe support, this gives the admin proper information about the slot
Jack F Vogel [Tue, 18 Jun 2013 21:28:19 +0000 (21:28 +0000)]
Add quad port probe support, this gives the admin proper information about the slot
(which should be a PCIE Gen 3 slot for this adapter) by looking back thru the PCI
parent devices to the slot device.

The fix above also corrects the bandwidth display to GT/s rather than the
incorrect Gb/s

Next, allow the use of ALTQ if you select the compile option IXGBE_LEGACY_TX.

Allow the use of 'unsupported' optic modules by a compile option as well.

Add a phy reset capability into the stop code, this is so a static configured
driver will still behave properly when taken down (not being able to unload it).

This revision synchronizes the shared code with Intel internal current code,
and note that it now includes DCB supporting code, this was necessitated by
some internal changes with the code, but it also will provide the opportunity
to develop this feature in the core driver down the road.

I have edited the README to get rid of some of the worse anachronisms in it
as well, its by no means as robust as I might wish at this point however.

Oh, I also have included some conditional stuff in the code so it will be
compatible in both the 9.X and 10 environments.

Performance has been a focus in recent changes and I believe this revision
driver will perform very well in most workloads.

MFC after: 2 weeks

11 years agoOn some generations of the Intel GPU, disabling of the VGA Display
Konstantin Belousov [Tue, 18 Jun 2013 20:19:09 +0000 (20:19 +0000)]
On some generations of the Intel GPU, disabling of the VGA Display
stops updating the vertical retrace indicator.  The text mouse
renderer in syscons is executing from the callout and spins waiting
for the start of next frame.  As result, after the X server finishes,
since the VGA cannot be turned on, but syscons does not know about
this, the clock swi spins forever.

Hack around the problem by disabling wait for the retrace if KMS is
activated.

Diagnosed and tested by: Michiel Boland <boland37@xs4all.nl>
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks

11 years agoSince the gem pagefault handler relocks the vm object lock, other
Konstantin Belousov [Tue, 18 Jun 2013 20:02:52 +0000 (20:02 +0000)]
Since the gem pagefault handler relocks the vm object lock, other
thread might fault on the same GTT offset meantime and instantiate the
mapping.  Recheck that the mgt device object still does not have a
page at the current offset after relocking, and return a possibly
installed page.

Reported by: Oleg Sidorkin <osidorkin@gmail.com>
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks

11 years agoRemove stray empty line.
Konstantin Belousov [Tue, 18 Jun 2013 19:56:52 +0000 (19:56 +0000)]
Remove stray empty line.

MFC after: 3 days

11 years agoFix use after free bug.
Simon J. Gerraty [Tue, 18 Jun 2013 19:35:51 +0000 (19:35 +0000)]
Fix use after free bug.
Parse_SetInput:
curFile->fname was using the buffer passed to it - which ReadMakefile frees.
This change makes the comment in ParseEOF about leaking curFile->fname true.

11 years agoMerge 1.8.0
Peter Wemm [Tue, 18 Jun 2013 16:36:21 +0000 (16:36 +0000)]
Merge 1.8.0

11 years agoImport svn-1.8.0 final
Peter Wemm [Tue, 18 Jun 2013 16:33:03 +0000 (16:33 +0000)]
Import svn-1.8.0 final

11 years agoTweak generated config to make it more likely to work on 32 bit systems.
Peter Wemm [Tue, 18 Jun 2013 16:05:33 +0000 (16:05 +0000)]
Tweak generated config to make it more likely to work on 32 bit systems.

Pointed out by: Trond.Endrestol@fagskolen.gjovik.no

11 years agoMore ext2fs header cleanups:
Pedro F. Giffuni [Tue, 18 Jun 2013 15:49:30 +0000 (15:49 +0000)]
More ext2fs header cleanups:

- Set MAXMNTLEN nearer to where it is used.
- Move EXT2_LINK_MAX to ext2_dir.h .

MFC after: 3 days

11 years agoPass proper memory type to free() in ata_ali_chipinit().
Alexander Motin [Tue, 18 Jun 2013 15:04:17 +0000 (15:04 +0000)]
Pass proper memory type to free() in ata_ali_chipinit().

Submitted by: Dmitry Luhtionov <dmitryluhtionov@gmail.com>
MFC after: 1 week

11 years agoFix bug in destructor for checker manager in DTC that caused segfaults on
David Chisnall [Tue, 18 Jun 2013 10:26:22 +0000 (10:26 +0000)]
Fix bug in destructor for checker manager in DTC that caused segfaults on
exit.

11 years agoComments and whitespace.
Devin Teske [Tue, 18 Jun 2013 09:54:09 +0000 (09:54 +0000)]
Comments and whitespace.

11 years agoComments.
Devin Teske [Tue, 18 Jun 2013 09:45:10 +0000 (09:45 +0000)]
Comments.

11 years agoPerform some code consolidation and add some additional error
Devin Teske [Tue, 18 Jun 2013 09:43:10 +0000 (09:43 +0000)]
Perform some code consolidation and add some additional error
checking/reporting. (similar to SVN revision 251919)

11 years agoWhitespace and comments.
Devin Teske [Tue, 18 Jun 2013 09:41:34 +0000 (09:41 +0000)]
Whitespace and comments.

11 years agoPerform some code consolidation and add some additional error
Devin Teske [Tue, 18 Jun 2013 09:28:49 +0000 (09:28 +0000)]
Perform some code consolidation and add some additional error
checking/reporting. (similar to SVN revisions 251905 and 251915)

11 years agoRemove unnecessary loops, perform some code consolidation, and add some
Devin Teske [Tue, 18 Jun 2013 09:19:59 +0000 (09:19 +0000)]
Remove unnecessary loops, perform some code consolidation, and add some
additional error checking/reporting. (same thing going on here as SVN
r251905 -- just this time for the mouse module instead of console)

11 years agoWhitespace.
Devin Teske [Tue, 18 Jun 2013 09:05:08 +0000 (09:05 +0000)]
Whitespace.

11 years agoUse newly enhanced f_dialog_msgbox() from SVN r251912.
Devin Teske [Tue, 18 Jun 2013 08:59:47 +0000 (08:59 +0000)]
Use newly enhanced f_dialog_msgbox() from SVN r251912.

11 years agoChange the f_dialog_msgbox() arguments in `dialog.subr' to accomodate an
Devin Teske [Tue, 18 Jun 2013 08:54:02 +0000 (08:54 +0000)]
Change the f_dialog_msgbox() arguments in `dialog.subr' to accomodate an
$hline argument for setting the --hline parameter value. This change does
not effect any current modules as it turns out that not one single usage of
f_dialog_msgbox() relied on accepting more than a first argument (read: all
occurrences quoted the first parameter; so $* was never depended upon).

This will allow some custom invocations of --msgbox to roll over to this
function (for example, in `mouse/disable').

11 years agoComments.
Devin Teske [Tue, 18 Jun 2013 08:40:16 +0000 (08:40 +0000)]
Comments.

11 years agoComments.
Devin Teske [Tue, 18 Jun 2013 08:35:44 +0000 (08:35 +0000)]
Comments.

11 years agoFix a typo in a comment.
Devin Teske [Tue, 18 Jun 2013 08:30:31 +0000 (08:30 +0000)]
Fix a typo in a comment.

11 years agoSwitch bsdconfig `Disk Management' from sade(8) to `bsdinstall partedit'.
Devin Teske [Tue, 18 Jun 2013 08:28:03 +0000 (08:28 +0000)]
Switch bsdconfig `Disk Management' from sade(8) to `bsdinstall partedit'.

11 years agoOops, in SVN r251905 I forgot that f_die takes the return code as the first
Devin Teske [Tue, 18 Jun 2013 08:22:51 +0000 (08:22 +0000)]
Oops, in SVN r251905 I forgot that f_die takes the return code as the first
argument (not the format).

11 years agoImprove the INDEX format. Whitespace improvements, format improvements, typo
Devin Teske [Tue, 18 Jun 2013 08:15:56 +0000 (08:15 +0000)]
Improve the INDEX format. Whitespace improvements, format improvements, typo
and grammatical fixes.

11 years agoRemove unnecessary loops, perform some code consolidation, and add some
Devin Teske [Tue, 18 Jun 2013 07:36:09 +0000 (07:36 +0000)]
Remove unnecessary loops, perform some code consolidation, and add some
additional error checking/reporting.

11 years agoWhitespace improvements.
Devin Teske [Tue, 18 Jun 2013 07:33:45 +0000 (07:33 +0000)]
Whitespace improvements.

11 years agoFix a bug that allowed a tracing process (e.g. gdb) to write
Dag-Erling Smørgrav [Tue, 18 Jun 2013 07:02:35 +0000 (07:02 +0000)]
Fix a bug that allowed a tracing process (e.g. gdb) to write
to a memory-mapped file in the traced process's address space
even if neither the traced process nor the tracing process had
write access to that file.

Security: CVE-2013-2171
Security: FreeBSD-SA-13:06.mmap
Approved by: so

11 years agoFix a KTR_BUSDMA format string.
Rui Paulo [Tue, 18 Jun 2013 06:55:58 +0000 (06:55 +0000)]
Fix a KTR_BUSDMA format string.

11 years agoMerge the commit template patch.
Peter Wemm [Tue, 18 Jun 2013 04:57:36 +0000 (04:57 +0000)]
Merge the commit template patch.

11 years agoMerge the 3-way merge marker tweak.
Peter Wemm [Tue, 18 Jun 2013 04:56:11 +0000 (04:56 +0000)]
Merge the 3-way merge marker tweak.

11 years agoRefine UMA bucket allocation to reduce space consumption and improve
Jeff Roberson [Tue, 18 Jun 2013 04:50:20 +0000 (04:50 +0000)]
Refine UMA bucket allocation to reduce space consumption and improve
performance.

 - Always free to the alloc bucket if there is space.  This gives LIFO
   allocation order to improve hot-cache performance.  This also allows
   for zones with a single bucket per-cpu rather than a pair if the entire
   working set fits in one bucket.
 - Enable per-cpu caches of buckets.  To prevent recursive bucket
   allocation one bucket zone still has per-cpu caches disabled.
 - Pick the initial bucket size based on a table driven maximum size
   per-bucket rather than the number of items per-page.  This gives
   more sane initial sizes.
 - Only grow the bucket size when we face contention on the zone lock, this
   causes bucket sizes to grow more slowly.
 - Adjust the number of items per-bucket to account for the header space.
   This packs the buckets more efficiently per-page while making them
   not quite powers of two.
 - Eliminate the per-zone free bucket list.  Always return buckets back
   to the bucket zone.  This ensures that as zones grow into larger
   bucket sizes they eventually discard the smaller sizes.  It persists
   fewer buckets in the system.  The locking is slightly trickier.
 - Only switch buckets in zalloc, not zfree, this eliminates pathological
   cases where we ping-pong between two buckets.
 - Ensure that the thread that fills a new bucket gets to allocate from
   it to give a better upper bound on allocation time.

Sponsored by: EMC / Isilon Storage Division

11 years agoAdd new FOREACH_FROM variants of the queue(3) FOREACH macros which can
Lawrence Stewart [Tue, 18 Jun 2013 02:57:56 +0000 (02:57 +0000)]
Add new FOREACH_FROM variants of the queue(3) FOREACH macros which can
optionally start the traversal from a previously found element by passing the
element in as "var". Passing a NULL "var" retains the same semantics as the
regular FOREACH macros.

Kudos to phk for suggesting the "FROM" suffix instead of my original proposal.

Reviewed by: jhb (previous version), rpaulo
MFC after: 1 week

11 years agoIntroduce svnlite so that we can check out our source code again.
Peter Wemm [Tue, 18 Jun 2013 02:53:45 +0000 (02:53 +0000)]
Introduce svnlite so that we can check out our source code again.

This is actually a fully functional build except:
* All internal shared libraries are static linked to make sure there
  is no interference with ports (and to reduce build time).
* It does not have the python/perl/etc plugin or API support.
* By default, it installs as "svnlite" rather than "svn".
* If WITH_SVN added in make.conf, you get "svn".
* If WITHOUT_SVNLITE is in make.conf, this is completely disabled.

To be absolutely clear, this is not intended for any use other than
checking out freebsd source and committing, like we once did with cvs.

It should be usable for small scale local repositories that don't
need the python/perl plugin architecture.

11 years agoAllow $ntpdate_config to be NULL. Due to a lack of surrounding quotes, when
Devin Teske [Tue, 18 Jun 2013 02:37:15 +0000 (02:37 +0000)]
Allow $ntpdate_config to be NULL. Due to a lack of surrounding quotes, when
ntpdate_config was set to NULL the conditional would (counter to prevailing
logic) succeed -- leading to awk attempting to redirect from a NULL pathname
standard-in. While we're here, make the script consistant with itself by
removing the {curlies} around ntpdate_config (they are unnecessary).

11 years agoImport sqlite-3071700
Peter Wemm [Tue, 18 Jun 2013 02:19:57 +0000 (02:19 +0000)]
Import sqlite-3071700

11 years agoImport trimmed svn-1.8.0-rc3
Peter Wemm [Tue, 18 Jun 2013 02:07:41 +0000 (02:07 +0000)]
Import trimmed svn-1.8.0-rc3

11 years agoImport serf-1.2.1
Peter Wemm [Tue, 18 Jun 2013 02:00:50 +0000 (02:00 +0000)]
Import serf-1.2.1

11 years agoImport apache apr-util 1.4.1
Peter Wemm [Tue, 18 Jun 2013 01:59:55 +0000 (01:59 +0000)]
Import apache apr-util 1.4.1

11 years agoImport apache apr-1.4.6
Peter Wemm [Tue, 18 Jun 2013 01:59:18 +0000 (01:59 +0000)]
Import apache apr-1.4.6

11 years agoBe sure to actually decrement the "count" parameter for each processed
Mark Johnston [Mon, 17 Jun 2013 22:59:47 +0000 (22:59 +0000)]
Be sure to actually decrement the "count" parameter for each processed
descriptor so that we return when the threshold has been reached.

Reviewed by: yongari
MFC after: 1 week

11 years agoAdd missing dependency to linux${SFX}_genassym.c
Eitan Adler [Mon, 17 Jun 2013 21:30:46 +0000 (21:30 +0000)]
Add missing dependency to linux${SFX}_genassym.c

Submitted by: nox
MFC After: 3 days

11 years agoRestore "all rights reserved" (spelled correctly). This was actually part of the...
Eitan Adler [Mon, 17 Jun 2013 20:27:20 +0000 (20:27 +0000)]
Restore "all rights reserved" (spelled correctly).  This was actually part of the standard text of the license which I did not realize prior.

Approved by: bushman

11 years agoFix header guards.
Eitan Adler [Mon, 17 Jun 2013 20:15:39 +0000 (20:15 +0000)]
Fix header guards.

This was ready about the same time as r251862 so just make one final cleanup

Submitted by: dt71@gmx.com

11 years agoClean up -Wheader-guard warnings.
Sergey Kandaurov [Mon, 17 Jun 2013 20:11:04 +0000 (20:11 +0000)]
Clean up -Wheader-guard warnings.

Submitted by: <dt71@gmx.com>
MFC after: 3 days
X-MFC with: r251848

11 years agoReturn ENETDOWN instead of ENOENT when all lagg(4) links are
Xin LI [Mon, 17 Jun 2013 19:31:03 +0000 (19:31 +0000)]
Return ENETDOWN instead of ENOENT when all lagg(4) links are
inactive when upper layer tries to transmit packet.  This
gives better feedback and meaningful errors for applications.

MFC after: 2 weeks
Reviewed by: thompsa

11 years agoFold in frame-unwind patch
Ed Maste [Mon, 17 Jun 2013 18:34:34 +0000 (18:34 +0000)]
Fold in frame-unwind patch

After moving to svn there's no need to avoid pulling files off a vendor
branch.

11 years agoAdd a checker to dtc, based on a feature request from rwatson / brooks.
David Chisnall [Mon, 17 Jun 2013 15:34:22 +0000 (15:34 +0000)]
Add a checker to dtc, based on a feature request from rwatson / brooks.
This checks that every node that has children specifies their register sizes.
This is not enabled by default, as the default sizes are sometimes required
(including by some DTS in the tree), but can help when writing new device
trees so that you can check that you actually meant the defaults.

11 years agoRename a parameter in sys/time.h so that you don't get warnings for things
David Chisnall [Mon, 17 Jun 2013 15:30:47 +0000 (15:30 +0000)]
Rename a parameter in sys/time.h so that you don't get warnings for things
like libdialog that include both this header and math.h.

11 years agoFix bindings of keys when in the partition editor. By adding the usual input binding...
Baptiste Daroussin [Mon, 17 Jun 2013 15:16:14 +0000 (15:16 +0000)]
Fix bindings of keys when in the partition editor. By adding the usual input binding to the "partlist" sub window.
This is a workaround, as for unknown yet reason the keys binded on the Partition Edition window are the one from partlist instead of the one from standard "formfield"

Reported by: alfred, nwhitehorn

11 years agoImport change e4ac6417c7504e1c55ec556ce908974c04e29e3c from upstream wpa:
Sergey Kandaurov [Mon, 17 Jun 2013 14:46:54 +0000 (14:46 +0000)]
Import change e4ac6417c7504e1c55ec556ce908974c04e29e3c from upstream wpa:

  From: Guy Eilam <guy@wizery.com>
  Date: Mon, 21 Feb 2011 20:44:46 +0000 (+0200)
  Subject: utils: Corrected a typo in header's name definition

  utils: Corrected a typo in header's name definition

  Corrected a typo in the BASE64_H definition that
  might cause the header file to be included more than once.

Signed-off-by: Guy Eilam <guy@wizery.com>
Submitted by: <dt71@gmx.com>
MFC after: 3 days

11 years agolower the WARNS to 1 again until I have more time to figure out the problems with...
Baptiste Daroussin [Mon, 17 Jun 2013 13:02:39 +0000 (13:02 +0000)]
lower the WARNS to 1 again until I have more time to figure out the problems with WARNS=4

Reported by: gavin