]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
10 years agogcc: Implement -Wmost for compatibility with clang.
pfg [Thu, 26 Dec 2013 18:09:16 +0000 (18:09 +0000)]
gcc: Implement -Wmost for compatibility with clang.

This is equivalent to -Wall -Wno-parentheses.

Obtained from: Apple GCC 4.2 - 5531
MFC after: 1 week

10 years agoUse feature_present(3) to determine whether to open an INET or an
bz [Thu, 26 Dec 2013 15:51:14 +0000 (15:51 +0000)]
Use feature_present(3) to determine whether to open an INET or an
INET6 socket when needed to allow pfctl to work on noinet and noinet6
kernels (and try to provide a fallback using AF_LINK as best effort).
Adjust the Makefile to also respect relevant src.conf(5) options
for compile time decisions on INET and INET6 support.

Reviewed by: glebius (no objections)
MFC after: 1 week

10 years agoFix AltGr, we should not only skip RAlt key release if enable_altgr is set, but
ray [Thu, 26 Dec 2013 14:25:37 +0000 (14:25 +0000)]
Fix AltGr, we should not only skip RAlt key release if enable_altgr is set, but
also process RAlt key press same way.

Sponsored by: The FreeBSD Foundation

10 years agoGenerate client sample code which compiles without warnings.
dfr [Thu, 26 Dec 2013 11:38:33 +0000 (11:38 +0000)]
Generate client sample code which compiles without warnings.
For 'rpcgen -a', generate a makefile where 'make clean all' works.

10 years agoFor libstand and sys/boot, split off gcc-only flags into CFLAGS.gcc.
dim [Thu, 26 Dec 2013 11:32:39 +0000 (11:32 +0000)]
For libstand and sys/boot, split off gcc-only flags into CFLAGS.gcc.

MFC after: 3 days
X-MFC-With: r259730

10 years agoFix "kptdir is itself virtual" error, caused by having the kptdir in PBVM.
marcel [Thu, 26 Dec 2013 07:10:54 +0000 (07:10 +0000)]
Fix "kptdir is itself virtual" error, caused by having the kptdir in PBVM.
While here improve errors by having them include addresses (either virtual
or physical).

10 years agoFor ia64, use pmap_remove_pages() and not pmap_remove(). The problem is
marcel [Thu, 26 Dec 2013 05:46:10 +0000 (05:46 +0000)]
For ia64, use pmap_remove_pages() and not pmap_remove(). The problem is
that we don't have a good way (yet) to iterate over the mapped pages by
virtual address and simply try each page within the range. Given that we
call pmap_remove() over the entire 2^63 bytes of address space, it takes
a while for pmap_remove to have tried all 2^50 pages.
By using pmap_remove_pages() we use the PV list to find all mappings.

Change derived from a patch by: alc

10 years agouse the correct netmap <-> nic slot mapping on the transmit ring for 'lem'.
luigi [Thu, 26 Dec 2013 05:22:38 +0000 (05:22 +0000)]
use the correct netmap <-> nic slot mapping on the transmit ring for 'lem'.
This bug would manifest only in netmap mode and on packets transmitted after
a NIC reset while netmap mode is active.

MFC after: 3 days

10 years agoDraft-ietf-tcpm-initcwnd-05 became RFC6928.
pluknet [Thu, 26 Dec 2013 04:24:08 +0000 (04:24 +0000)]
Draft-ietf-tcpm-initcwnd-05 became RFC6928.

MFC after: 1 week

10 years agoRemove some unneeded declarations which should have been removed in r257037.
markj [Thu, 26 Dec 2013 03:19:08 +0000 (03:19 +0000)]
Remove some unneeded declarations which should have been removed in r257037.

10 years agoIn sys/dev/drm/mach64_dma.c, remove static function mach64_set_dma_eol(),
dim [Thu, 26 Dec 2013 00:11:19 +0000 (00:11 +0000)]
In sys/dev/drm/mach64_dma.c, remove static function mach64_set_dma_eol(),
which has never been used, even by upstream, since its initial upstream
commit (see http://cgit.freedesktop.org/mesa/drm/commit/?id=873e1c4d )

MFC after: 3 days

10 years agoIn sys/dev/cxgb/ulp/tom/cxgb_cpl_io.c, remove static functions
dim [Wed, 25 Dec 2013 22:49:54 +0000 (22:49 +0000)]
In sys/dev/cxgb/ulp/tom/cxgb_cpl_io.c, remove static functions
mk_cpl_barrier_ulp(), mk_get_tcb_ulp() and mk_set_tcb_field_ulp(), which
are all unused since r237263.

MFC after: 3 days

10 years agoIn sys/dev/cxgb/common/cxgb_mc5.c, remove static function
dim [Wed, 25 Dec 2013 22:45:33 +0000 (22:45 +0000)]
In sys/dev/cxgb/common/cxgb_mc5.c, remove static function
dbgi_wr_addr3(), which is unused since r167514.

MFC after: 3 days

10 years agoIn sys/vm/vm_pageout.c, since vm_pageout_worker() takes a void * as
dim [Wed, 25 Dec 2013 22:32:34 +0000 (22:32 +0000)]
In sys/vm/vm_pageout.c, since vm_pageout_worker() takes a void * as
argument, cast the incoming 0 argument to void *, to silence a warning
from clang 3.4 ("expression which evaluates to zero treated as a null
pointer constant of type 'void *' [-Wnon-literal-null-conversion]").

MFC after: 3 days

10 years agoIn sys/kern/vfs_mountroot.c, remove static function parse_isspace(),
dim [Wed, 25 Dec 2013 22:14:42 +0000 (22:14 +0000)]
In sys/kern/vfs_mountroot.c, remove static function parse_isspace(),
which is unused since r214006.

MFC after: 3 days

10 years agoPull in r183971 from upstream llvm trunk:
dim [Wed, 25 Dec 2013 20:58:02 +0000 (20:58 +0000)]
Pull in r183971 from upstream llvm trunk:

  X86: cvtpi2ps is just an SSE instruction with MMX operands. It has no AVX
  equivalent.

  Give it the right register format so we can also emit it when AVX is enabled.

This should fix a "Cannot select: intrinsic %llvm.x86.sse.cvtpi2ps" fatal error
in clang while building the gnuradio port for amd64.

Reported by: db
MFC after: 3 days

10 years agoAdd more (IPv6) related Internet Protocols:
bz [Wed, 25 Dec 2013 20:26:49 +0000 (20:26 +0000)]
Add more (IPv6) related Internet Protocols:
- Host Identity Protocol (RFC5201)
- Shim6 Protocol (RFC5533)
- 2x experimentation and testing (RFC3692, RFC4727)

This does not indicate interest to implement/support these protocols,
but they are part of the "IPv6 Extension Header Types" [1] based on RFC7045
and might thus be needed by filtering and next header parsing
implementations.

References: [1] http://www.iana.org/assignments/ipv6-parameters
Obtained from: http://www.iana.org/assignments/protocol-numbers
MFC after: 1 week

10 years agoBite the bullet and start removing the first compile time warnings
bz [Wed, 25 Dec 2013 20:15:48 +0000 (20:15 +0000)]
Bite the bullet and start removing the first compile time warnings
by removing unsued file local functions and then unused callees.
A lot more warnings to resolve but someone had to break the ice.

MFC after: 10 days
X-Comment: I am not the new maintainer; chime in, it's ours.

10 years agoAs constantly reported during kernel compilation, m_buflen is unsigned so
bz [Wed, 25 Dec 2013 20:10:17 +0000 (20:10 +0000)]
As constantly reported during kernel compilation, m_buflen is unsigned so
can never be < 0.  Remove the expression, which can never be true.

MFC after: 1 week

10 years agoCorrect warnings comparing unsigned variables < 0 constantly reported
bz [Wed, 25 Dec 2013 20:08:44 +0000 (20:08 +0000)]
Correct warnings comparing unsigned variables < 0 constantly reported
while building kernels.  All instances removed are indeed unsigned so
the expressions could not be true.

MFC after: 1 week

10 years agoFix a typo.
pluknet [Wed, 25 Dec 2013 19:38:16 +0000 (19:38 +0000)]
Fix a typo.

10 years agoUse statndard (syscons) way to disable bell.
ray [Wed, 25 Dec 2013 19:07:14 +0000 (19:07 +0000)]
Use statndard (syscons) way to disable bell.

Testesd by: markj

Sponsored by: The FreeBSD Foundation

10 years agoPut the release objdir inside the chroot.
jmmv [Wed, 25 Dec 2013 18:25:58 +0000 (18:25 +0000)]
Put the release objdir inside the chroot.

When building the system from outside the chroot as part of the release
process, ensure the objdir is within the chroot so that the whole output
of the release.sh script is self-contained in a single directory.

Use CHROOTDIR/tmp/obj instead of CHROOTDIR/usr/obj to minimize possible
interferences with the output of the build itself.

Reviewed by: gjb
Approved by: rpaulo (mentor)

10 years agoIn sys/dev/sym/sym_hipd.c, remove static functions sym_que_first(),
dim [Wed, 25 Dec 2013 17:28:18 +0000 (17:28 +0000)]
In sys/dev/sym/sym_hipd.c, remove static functions sym_que_first(),
sym_que_last() and sym_remque_tail(), which are all unused since r53790.

MFC after: 3 days

10 years agoClean up manual pages after BIND removal.
pluknet [Wed, 25 Dec 2013 17:28:08 +0000 (17:28 +0000)]
Clean up manual pages after BIND removal.

MFC after: 1 week

10 years agoFor vmcore, calculate time relative to device creation upon time_uptime.
pluknet [Wed, 25 Dec 2013 17:11:49 +0000 (17:11 +0000)]
For vmcore, calculate time relative to device creation upon time_uptime.
Previously it used a clock from live kernel.

MFC after: 2 weeks

10 years agoSlightly simplify expiration logic introduced in r254337.
mav [Wed, 25 Dec 2013 16:58:42 +0000 (16:58 +0000)]
Slightly simplify expiration logic introduced in r254337.

 - Do not update the histogram for items we are any way deleting from cache.
 - Do not update the histogram if nfsrc_tcphighwater is not set.
 - Remove some extra math operations.

10 years agoIn sys/kern/subr_witness.c, remove static function
dim [Wed, 25 Dec 2013 16:58:14 +0000 (16:58 +0000)]
In sys/kern/subr_witness.c, remove static function
witness_lock_order_key_empty(), which is unused since r181695.

MFC after: 3 days

10 years agoIn sys/kern/sched_ule.c, remove static function sched_both(), which is
dim [Wed, 25 Dec 2013 16:25:54 +0000 (16:25 +0000)]
In sys/kern/sched_ule.c, remove static function sched_both(), which is
unused since r232207.

MFC after: 3 days

10 years agosh: Simplify code related to PPID variable.
jilles [Wed, 25 Dec 2013 16:14:02 +0000 (16:14 +0000)]
sh: Simplify code related to PPID variable.

10 years agogcc: small enhancements for the arm support.
pfg [Wed, 25 Dec 2013 16:01:48 +0000 (16:01 +0000)]
gcc: small enhancements for the arm support.

Very small updates: fixes GCC-PR target/31152

Tested by building the cross-compiler.

Obtained from: gcc 4.3 (rev. r118461, 125973: GPLv2)
MFC after: 2 weeks

10 years agoThe compile time constant limit on number of swap devices was removed in 5.2.
pluknet [Wed, 25 Dec 2013 16:01:29 +0000 (16:01 +0000)]
The compile time constant limit on number of swap devices was removed in 5.2.
As such, remove the EINVAL error saying so.  Currently the vm.nswapdev sysctl
just represents the number of added swap devices.

MFC after: 1 week

10 years agoDo not truncate the ``command'' column in ``ps'' output.
pluknet [Wed, 25 Dec 2013 15:23:01 +0000 (15:23 +0000)]
Do not truncate the ``command'' column in ``ps'' output.

Reviewed by: jhb
MFC after: 1 week

10 years agoIn sys/dev/mwl/if_mwl.c, put the static RD4() function under #ifdef
dim [Wed, 25 Dec 2013 14:48:27 +0000 (14:48 +0000)]
In sys/dev/mwl/if_mwl.c, put the static RD4() function under #ifdef
MWL_DEBUG guards, since it only used in DPRINTF statements.

MFC after: 3 days

10 years agoDelay copying of resolv.conf into the chroot until /etc exists.
jmmv [Wed, 25 Dec 2013 13:57:23 +0000 (13:57 +0000)]
Delay copying of resolv.conf into the chroot until /etc exists.

With an unpopulated chroot dir, release building was failing because
the script was attempting to copy resolv.conf into a non-existent /etc
directory of the chroot.  Fix this by copying the file only after the
installworld has happened, which will create this directory.

Reviewed by: gjb
Approved by: rpaulo (mentor)

10 years agoIt'll be okay to use LibAliasDetachHandlers() here, relying
glebius [Wed, 25 Dec 2013 09:43:51 +0000 (09:43 +0000)]
It'll be okay to use LibAliasDetachHandlers() here, relying
on the fact that all handlers come from modules' bss and are
followed by NODIR handler.

10 years agovlapic code restructuring to make it easy to support hardware-assist for APIC
neel [Wed, 25 Dec 2013 06:46:31 +0000 (06:46 +0000)]
vlapic code restructuring to make it easy to support hardware-assist for APIC
emulation.

The vlapic initialization and cleanup is done via processor specific vmm_ops.
This will allow the VT-x/SVM modules to layer any hardware-assist for APIC
emulation or virtual interrupt delivery on top of the vlapic device model.

Add a parameter to 'vcpu_notify_event()' to distinguish between vlapic
interrupts versus other events (e.g. NMI). This provides an opportunity to
use hardware-assists like Posted Interrupts (VT-x) or doorbell MSR (SVM)
to deliver an interrupt to a guest without causing a VM-exit.

Get rid of lapic_pending_intr() and lapic_intr_accepted() and use the
vlapic_xxx() counterparts directly.

Associate an 'Apic Page' with each vcpu and reference it from the 'vlapic'.
The 'Apic Page' is intended to be referenced from the Intel VMCS as the
'virtual APIC page' or from the AMD VMCB as the 'vAPIC backing page'.

10 years agoHarvest one no longer used constant string.
mjacob [Wed, 25 Dec 2013 04:51:56 +0000 (04:51 +0000)]
Harvest one no longer used constant string.

Remove another and place it into play in the
normally ifdef protected zone it would be used
int.

Noticed by: dim

10 years agoCleanup alias module handler register/unregister.
glebius [Wed, 25 Dec 2013 03:24:20 +0000 (03:24 +0000)]
Cleanup alias module handler register/unregister.

- Remove locking, since all module(9) events are running under &Giant.
- Use TAILQ for protocol handlers and fix a bug which led to
  infinite cycle. Bug found in VirtualBox [1]
- Simplify code everywhere.
- Fix documentation.

[1]  https://www.virtualbox.org/pipermail/vbox-dev/2013-November/011936.html

PR: 183792 [1]
Submitted by: Valery Ushakov <uwe NetBSD.org> [1]
Sponsored by: Nginx, Inc.

10 years agoKill space at eols.
glebius [Wed, 25 Dec 2013 02:06:57 +0000 (02:06 +0000)]
Kill space at eols.

10 years agoRemove from kernel the "dll" code.
glebius [Wed, 25 Dec 2013 01:58:19 +0000 (01:58 +0000)]
Remove from kernel the "dll" code.

10 years agoWhitespace cleanup.
glebius [Wed, 25 Dec 2013 01:52:55 +0000 (01:52 +0000)]
Whitespace cleanup.

10 years agoWe have in base iconv.
glebius [Wed, 25 Dec 2013 01:20:14 +0000 (01:20 +0000)]
We have in base iconv.

PR: 185135
Submitted by: Lars Engels <lars 0x20.net>

10 years agoThe NFSv4 server would call VOP_SETATTR() with a shared locked vnode
rmacklem [Wed, 25 Dec 2013 01:03:14 +0000 (01:03 +0000)]
The NFSv4 server would call VOP_SETATTR() with a shared locked vnode
when a Getattr for a file is done by a client other than the one that
holds the file's delegation. This would only happen when delegations
are enabled and the problem is fixed by this patch.

MFC after: 1 week

10 years agoMFV r258384:
delphij [Wed, 25 Dec 2013 00:39:04 +0000 (00:39 +0000)]
MFV r258384:

2583 Add -p (parsable) option to zfs list

illumos/illumos-gate@43d68d68c1ce08fb35026bebfb141af422e7082e

MFC after: 2 weeks

10 years agosh: Initialize OPTIND=1 even if it came from the environment.
jilles [Tue, 24 Dec 2013 22:38:24 +0000 (22:38 +0000)]
sh: Initialize OPTIND=1 even if it came from the environment.

10 years agoAn intermittent problem with NFSv4 exporting of ZFS snapshots was
rmacklem [Tue, 24 Dec 2013 22:24:17 +0000 (22:24 +0000)]
An intermittent problem with NFSv4 exporting of ZFS snapshots was
reported to the freebsd-fs mailing list. I believe the problem was
caused by the Readdir operation using VFS_VGET() for a snapshot file entry
instead of VOP_LOOKUP(). This would not occur for NFSv3, since it
will do a VFS_VGET() of "." which fails with ENOTSUPP at the beginning
of the directory, whereas NFSv4 does not check "." or "..". This
patch adds a call to VFS_VGET() for the directory being read to check
for ENOTSUPP.
I also observed that the mount_on_fileid and fsid attributes were
not correct at the snapshot's auto mountpoints when looking at packet
traces for the Readdir. This patch fixes the attributes by doing a check
for different v_mount structure, even if the vnode v_mountedhere is not
set.

Reported by: jas@cse.yorku.ca
Tested by: jas@cse.yorku.ca
Reviewed by: asomers
MFC after: 1 week

10 years agosh: Remove an unused variable.
jilles [Tue, 24 Dec 2013 22:04:44 +0000 (22:04 +0000)]
sh: Remove an unused variable.

10 years agoMove a static const variable to the #if 0 part where it is only used.
dim [Tue, 24 Dec 2013 20:57:26 +0000 (20:57 +0000)]
Move a static const variable to the #if 0 part where it is only used.
(Note the #if 0 part has been inactive since the initial commit,
r177633, so maybe it should be removed altogether).

MFC after: 3 days

10 years agoRemove some unused static const strings under sys/rpc, which have never
dim [Tue, 24 Dec 2013 20:55:22 +0000 (20:55 +0000)]
Remove some unused static const strings under sys/rpc, which have never
been used since the initial commit (r177633).

MFC after: 3 days

10 years agogcc: Add ability to generate DWARF pubtypes section if
pfg [Tue, 24 Dec 2013 20:42:48 +0000 (20:42 +0000)]
gcc:  Add ability to generate DWARF pubtypes section if
DEBUG_PUBTYPES_SECTION is defined.

Obtained from: gcc 4.3 (rev. 118826; GPLv2)

MFC after: 2 weeks

10 years agoIn sys/netinet6/in6_mcast.c, in6m_is_ifp_detached() is only used
dim [Tue, 24 Dec 2013 20:30:13 +0000 (20:30 +0000)]
In sys/netinet6/in6_mcast.c, in6m_is_ifp_detached() is only used
whenever KTR is defined, so put it between #ifdef KTR guards.  This
avoids a warning about a unused function if KTR is not enabled.

MFC after: 3 days

10 years agoIn sys/netinet/in_mcast.c, inm_is_ifp_detached() is only used whenever
dim [Tue, 24 Dec 2013 20:25:18 +0000 (20:25 +0000)]
In sys/netinet/in_mcast.c, inm_is_ifp_detached() is only used whenever
KTR is defined, so put it between #ifdef KTR guards.  This avoids a
warning about a unused function if KTR is not enabled.

MFC after:  3 days

10 years agoAdd another HP iLO serial (console) port, found on Itanium servers.
marcel [Tue, 24 Dec 2013 19:58:27 +0000 (19:58 +0000)]
Add another HP iLO serial (console) port, found on Itanium servers.

Derived from information provided by: felip.nadal@hob.de

10 years agoRemove another unused static const variable num_chip_names, from
dim [Tue, 24 Dec 2013 18:52:36 +0000 (18:52 +0000)]
Remove another unused static const variable num_chip_names, from
aic7xxx.c this time.

Noticed by: pluknet
MFC after: 3 days

10 years agoFix copy-pasting of CJK fullwidth characters.
ed [Tue, 24 Dec 2013 18:42:26 +0000 (18:42 +0000)]
Fix copy-pasting of CJK fullwidth characters.

They are stored as two separate characters in the vtbuf, so copy-pasting
will cause them to be passed to terminal_input_char() twice. Extend
terminal_input_char() to explicitly discard characters with TF_CJK_RIGHT
set. This causes only the left part to generate input.

10 years agoFix a bug introduced at r259632, triggering infinite loop in some cases.
mav [Tue, 24 Dec 2013 17:28:27 +0000 (17:28 +0000)]
Fix a bug introduced at r259632, triggering infinite loop in some cases.

10 years agoRemove unused static const variable num_chip_names from aic79xx.c.
dim [Tue, 24 Dec 2013 16:46:06 +0000 (16:46 +0000)]
Remove unused static const variable num_chip_names from aic79xx.c.

Reviewed by: gibbs
MFC after: 3 days

10 years agoSupport soft power-off via the ACPI S5 state for bhyve guests.
jhb [Tue, 24 Dec 2013 16:14:19 +0000 (16:14 +0000)]
Support soft power-off via the ACPI S5 state for bhyve guests.
- Implement the PM1_EVT and PM1_CTL registers required by ACPI.
  The PM1_EVT register is mostly a dummy as bhyve doesn't support any
  of the hardware-initiated events.  The only bit of PM1_CNT that is
  implemented are the sleep request bits (SPL_EN and SLP_TYP) which
  request a graceful power off for S5.  In particular, for S5, bhyve
  exits with a non-zero value which terminates the loop in vmrun.sh.
- Emulate the Reset Control register at I/O port 0xcf9 and advertise
  it as the reset register via ACPI.
- Advertise an _S5 package.
- Extend the in/out interface to allow an in/out handler to request
  that the hypervisor trigger a reset or power-off.
- While here, note that all vCPUs in a guest support C1 ("hlt").

Reviewed by: neel (earlier version)

10 years agoRemove unused static function adwccbstatus() from adw(4).
dim [Tue, 24 Dec 2013 15:56:16 +0000 (15:56 +0000)]
Remove unused static function adwccbstatus() from adw(4).

Reviewed by: gibbs
MFC after: 3 days

10 years agoFix i386 build.
jhb [Tue, 24 Dec 2013 14:48:52 +0000 (14:48 +0000)]
Fix i386 build.

Pointy hat to: jhb

10 years agoFix incorrect markup introduced in r259813.
delphij [Tue, 24 Dec 2013 07:27:55 +0000 (07:27 +0000)]
Fix incorrect markup introduced in r259813.

Pointy hat to: delphij
X-MFC-after: r259813

10 years agoMFV r258374:
delphij [Tue, 24 Dec 2013 07:14:25 +0000 (07:14 +0000)]
MFV r258374:

4171 clean up spa_feature_*() interfaces

4172 implement extensible_dataset feature for use by other zpool
features

illumos/illumos-gate@2acef22db7808606888f8f92715629ff3ba555b9

MFC after: 2 weeks

10 years agoSome devices have a dual mode such as Tenda W326U and ZyXEL NWD2705,
kevlo [Tue, 24 Dec 2013 07:02:05 +0000 (07:02 +0000)]
Some devices have a dual mode such as Tenda W326U and ZyXEL NWD2705,
add the product id and use a standard scsi eject.

Tested on the ZyXEL NWD2705 wlan dongle.

10 years agoMFV r258373:
delphij [Tue, 24 Dec 2013 06:56:17 +0000 (06:56 +0000)]
MFV r258373:

4168 ztest assertion failure in dbuf_undirty

4169 verbatim import causes zdb to segfa
4170 zhack leaves pool in ACTIVE state

illumos/illumos-gate@7fdd916c474ea52896c671bbe7b56ba34a1ca132

MFC after: 2 weeks

10 years agoAdd pointer to wiki for Hyper-V information.
rodrigc [Tue, 24 Dec 2013 05:10:41 +0000 (05:10 +0000)]
Add pointer to wiki for Hyper-V information.

Wiki content for Hyper-V in FreeBSD 10 was significantly updated by
Abhishek Gupta <abgupta@microsoft.com>.

10 years agoThe NFSv4 client was passing both the p and cred arguments to
rmacklem [Tue, 24 Dec 2013 00:48:39 +0000 (00:48 +0000)]
The NFSv4 client was passing both the p and cred arguments to
nfsv4_fillattr() as NULLs for the Getattr callback. This caused
nfsv4_fillattr() to not fill in the Change attribute for the reply.
I believe this was a violation of the RFC, but had little effect on
server behaviour. This patch passes a non-NULL p argument to fix this.

MFC after: 1 week

10 years agoRemove references to SUP_UPDATE and CVS_UPDATE.
gjb [Mon, 23 Dec 2013 21:58:03 +0000 (21:58 +0000)]
Remove references to SUP_UPDATE and CVS_UPDATE.
Include base svn when evaluating if svn(1) exists.

MFC after: 3-days
X-MFC-to: stable/10 only
Sponsored by: The FreeBSD Foundation

10 years agoUpdate FreeBSD version entities to reflect head/ is 11.0-CURRENT.
gjb [Mon, 23 Dec 2013 20:39:54 +0000 (20:39 +0000)]
Update FreeBSD version entities to reflect head/ is 11.0-CURRENT.

Sponsored by: The FreeBSD Foundation

10 years agoAdd a resume hook for bhyve that runs a function on all CPUs during
jhb [Mon, 23 Dec 2013 19:48:22 +0000 (19:48 +0000)]
Add a resume hook for bhyve that runs a function on all CPUs during
resume.  For Intel CPUs, invoke vmxon for CPUs that were in VMX mode
at the time of suspend.

Reviewed by: neel

10 years agoMention jemalloc upgrade to 3.4.0
rodrigc [Mon, 23 Dec 2013 19:48:02 +0000 (19:48 +0000)]
Mention jemalloc upgrade to 3.4.0

10 years agoext2fs: make the hashing algorithm match the linux code.
pfg [Mon, 23 Dec 2013 19:47:34 +0000 (19:47 +0000)]
ext2fs: make the hashing algorithm match the linux code.

There appears to be a hash function compatibility issue.
The code is currently disabled but fix it nevertheless.

PR: kern/183230
MFC after: 3 days

10 years agoExtend the support for local interrupts on the local APIC:
jhb [Mon, 23 Dec 2013 19:29:07 +0000 (19:29 +0000)]
Extend the support for local interrupts on the local APIC:
- Add a generic routine to trigger an LVT interrupt that supports both
  fixed and NMI delivery modes.
- Add an ioctl and bhyvectl command to trigger local interrupts inside a
  guest.  In particular, a global NMI similar to that raised by SERR# or
  PERR# can be simulated by asserting LINT1 on all vCPUs.
- Extend the LVT table in the vCPU local APIC to support CMCI.
- Flesh out the local APIC error reporting a bit to cache errors and
  report them via ESR when ESR is written to.  Add support for asserting
  the error LVT when an error occurs.  Raise illegal vector errors when
  attempting to signal an invalid vector for an interrupt or when sending
  an IPI.
- Ignore writes to reserved bits in LVT entries.
- Export table entries the MADT and MP Table advertising the stock x86
  config of LINT0 set to ExtInt and LINT1 wired to NMI.

Reviewed by: neel (earlier version)

10 years agoo Add virtual terminal mmap request handler.
ray [Mon, 23 Dec 2013 18:09:10 +0000 (18:09 +0000)]
o Add virtual terminal mmap request handler.
o Forward termianl framebuffer ioctl to fbd.
o Forward terminal mmap request to fbd.
o Move inclusion of sys/conf.h to vt.h.

Sponsored by: The FreeBSD Foundation

10 years agoUpdate whitespace to match mkkfont output
emaste [Mon, 23 Dec 2013 17:49:12 +0000 (17:49 +0000)]
Update whitespace to match mkkfont output

Sponsored by: The FreeBSD Foundation

10 years agoFix yet another typo
bapt [Mon, 23 Dec 2013 17:47:04 +0000 (17:47 +0000)]
Fix yet another typo

Reported by: kib

10 years agoFix typos
bapt [Mon, 23 Dec 2013 17:26:56 +0000 (17:26 +0000)]
Fix typos

Reported by: bryanv

10 years agoDo not print a question in non-interractive mode but explain why boostrapping
bapt [Mon, 23 Dec 2013 17:16:44 +0000 (17:16 +0000)]
Do not print a question in non-interractive mode but explain why boostrapping
has failed and how to allow it to happen

Reported by: jmmv

10 years agoOutput all {normal,bold} {left,right} maps
emaste [Mon, 23 Dec 2013 17:14:11 +0000 (17:14 +0000)]
Output all {normal,bold} {left,right} maps

This change was missed in r259680.

Sponsored by: The FreeBSD Foundation

10 years agoThe NFSv4.1 client didn't return NFSv4.1 specific error codes
rmacklem [Mon, 23 Dec 2013 15:16:53 +0000 (15:16 +0000)]
The NFSv4.1 client didn't return NFSv4.1 specific error codes
for the Getattr and Recall callbacks. This patch fixes it.
Since the NFSv4.1 specific error codes would only happen for
abnormal circumstances, this patch has little effect, in practice.

MFC after: 1 week

10 years agomdoc: nuke whitespace.
joel [Mon, 23 Dec 2013 15:00:15 +0000 (15:00 +0000)]
mdoc: nuke whitespace.

10 years agoTry and fix the dependency/bootstrap issues in kerberos5
uqs [Mon, 23 Dec 2013 14:23:17 +0000 (14:23 +0000)]
Try and fix the dependency/bootstrap issues in kerberos5

libkafs5 needs a header from libkrb5, it includes this from
${.OBJDIR}/mumble, this used to work fine as long as you happen to have
a krb_err.h in your base system, this doesn't work for bootstrapping or
using a cross-compiler with a different sysroot. This is just a
best-effort bandaid, sufficient parallelism can still break it.

Fix a SRCS override that dropped krb5_err.h.

Discussed with: stas

10 years agoUpdates for various userland tools.
rodrigc [Mon, 23 Dec 2013 11:03:37 +0000 (11:03 +0000)]
Updates for various userland tools.

10 years agoMention that sysinstall is gone, replaced by bsdinstall and bsdconfig.
rodrigc [Mon, 23 Dec 2013 08:57:49 +0000 (08:57 +0000)]
Mention that sysinstall is gone, replaced by bsdinstall and bsdconfig.
Mention that bsdinstall supports installing ZFS on root.

10 years agoFix RPC server threads file handle affinity to work better with ZFS.
mav [Mon, 23 Dec 2013 08:43:16 +0000 (08:43 +0000)]
Fix RPC server threads file handle affinity to work better with ZFS.

  Instead of taking 8 specific bytes of file handle to identify file during
RPC thread affitinity handling, use trivial hash of the full file handle.
ZFS's struct zfid_short does not have padding field after the length field,
as result, originally picked 8 bytes are loosing lower 16 bits of object ID,
causing many false matches and unneeded requests affinity to same thread.
  This fix substantially improves NFS server latency and scalability in SPEC
NFS benchmark by more flexible use of multiple NFS threads.

Sponsored by: iXsystems, Inc.

10 years agoMention RDRAND and Xen PVHVM. Add a revision number for 4TB RAM support on amd64.
rodrigc [Mon, 23 Dec 2013 08:22:08 +0000 (08:22 +0000)]
Mention RDRAND and Xen PVHVM.  Add a revision number for 4TB RAM support on amd64.

10 years agoAdd notes for ARM.
rodrigc [Mon, 23 Dec 2013 07:58:46 +0000 (07:58 +0000)]
Add notes for ARM.

10 years agoFix text for loading Hyper-V drivers via loader.conf.
rodrigc [Mon, 23 Dec 2013 07:35:34 +0000 (07:35 +0000)]
Fix text for loading Hyper-V drivers via loader.conf.

Noticed by: hrs

10 years agoFix linewrapping behaviour for CJK fullwidth characters.
ed [Mon, 23 Dec 2013 05:47:27 +0000 (05:47 +0000)]
Fix linewrapping behaviour for CJK fullwidth characters.

Instead of only wrapping when in the 'wrapped state', also force
wrapping when the character to be rendered does not fit on the line
anymore.

Tested by: lwhsu

10 years agoGet rid of register keyword usage in gperf, it is totally obsolete for
dim [Mon, 23 Dec 2013 00:02:18 +0000 (00:02 +0000)]
Get rid of register keyword usage in gperf, it is totally obsolete for
C++, and this allows gperf to be compiled for C++11 without a warning
about it.

MFC after: 3 days

10 years agoAdd PPS support to the am335x timer driver. This uses the timer hardware's
ian [Sun, 22 Dec 2013 23:03:29 +0000 (23:03 +0000)]
Add PPS support to the am335x timer driver.  This uses the timer hardware's
capture mode together with the timecounter's PPS polling feature to get
very accurate PPS capture without any interrupt processing (or latency).

Hardware timers 4 through 7 have associated capture-trigger input pins.
When the PPS support is compiled in the code automatically chooses the
first timer it finds that has the capture-trigger pin set to input mode
(this is configured via the fdt data).

10 years agoAdd updates for Raspberry PI support, unmapped VMIO, netmap(4), nvme(4).
rodrigc [Sun, 22 Dec 2013 22:56:45 +0000 (22:56 +0000)]
Add updates for Raspberry PI support, unmapped VMIO, netmap(4), nvme(4).

10 years agoA variety of cleanups...
ian [Sun, 22 Dec 2013 21:44:32 +0000 (21:44 +0000)]
A variety of cleanups...
 - Use named constants for register bits, instead of mystery numebrs
   scattered around in the code.
 - Use inline functions for bus space read/write, instead of macros
   that rely on global variables.
 - Move the timecounter struct into the softc instead of treating it
   as a global variable.  Backlink from it to the softc.
 - This leaves a pointer to the softc as the only static/global variable
   and it's now used only by DELAY().

10 years agoMap out all the timer-related registers, and define named constants for
ian [Sun, 22 Dec 2013 21:35:18 +0000 (21:35 +0000)]
Map out all the timer-related registers, and define named constants for
the bits within the registers.

10 years agoIn usr.bin/sort/radixsort.c, pop_ls_mt() is only referenced if
dim [Sun, 22 Dec 2013 20:46:31 +0000 (20:46 +0000)]
In usr.bin/sort/radixsort.c, pop_ls_mt() is only referenced if
SORT_THREADS is defined, so make the whole function conditional, instead
of just the pthread calls in it.

MFC after: 3 days

10 years agoShorten the DMTIMER_ prefix used for register names to DMT_. This is in
ian [Sun, 22 Dec 2013 20:40:56 +0000 (20:40 +0000)]
Shorten the DMTIMER_ prefix used for register names to DMT_.  This is in
preparation for adding more complete register defintions, some of which
have fairly long names.

10 years agoAdd a parameter to 'vcpu_set_state()' to enforce that the vcpu is in the IDLE
neel [Sun, 22 Dec 2013 20:29:59 +0000 (20:29 +0000)]
Add a parameter to 'vcpu_set_state()' to enforce that the vcpu is in the IDLE
state before the requested state transition. This guarantees that there is
exactly one ioctl() operating on a vcpu at any point in time and prevents
unintended state transitions.

More details available here:
http://lists.freebsd.org/pipermail/freebsd-virtualization/2013-December/001825.html

Reviewed by: grehan
Reported by: Markiyan Kushnir (markiyan.kushnir at gmail.com)
MFC after: 3 days

10 years agoFix incorrect header guard define in sys/netpfil/pf/pf.h, which snuck in
dim [Sun, 22 Dec 2013 19:47:22 +0000 (19:47 +0000)]
Fix incorrect header guard define in sys/netpfil/pf/pf.h, which snuck in
in r257186.  Found by clang 3.4.

10 years agoGroup virtualization items in one section.
rodrigc [Sun, 22 Dec 2013 19:38:51 +0000 (19:38 +0000)]
Group virtualization items in one section.

10 years agoAdd a note for Microsoft Hyper-V.
rodrigc [Sun, 22 Dec 2013 19:33:15 +0000 (19:33 +0000)]
Add a note for Microsoft Hyper-V.