]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
10 years agoFix a few typos on the scm (control module) pin mux definitions.
Luiz Otavio O Souza [Mon, 9 Dec 2013 11:33:45 +0000 (11:33 +0000)]
Fix a few typos on the scm (control module) pin mux definitions.

Approved by: adrian (mentor)

10 years agoMake it easier to test build the USB code having the debug flags set
Hans Petter Selasky [Mon, 9 Dec 2013 07:26:55 +0000 (07:26 +0000)]
Make it easier to test build the USB code having the debug flags set
without having to build the complete kernel.

MFC after: 2 weeks

10 years agoAdd gpio config for usb1 in dts.
Ganbold Tsagaankhuu [Mon, 9 Dec 2013 07:15:46 +0000 (07:15 +0000)]
Add gpio config for usb1 in dts.
That way it gives power to other usb hub via gpio at boot time.

Reviewed by: stas@

10 years agoAdd gpio parse routines according to sys/boot/fdt/dts/bindings-gpio.txt.
Ganbold Tsagaankhuu [Mon, 9 Dec 2013 07:14:59 +0000 (07:14 +0000)]
Add gpio parse routines according to sys/boot/fdt/dts/bindings-gpio.txt.

Reviewed by: stas@

10 years agoReplace Sun RPC license with a 3-clause BSD license. This license change
Hiroki Sato [Mon, 9 Dec 2013 05:01:38 +0000 (05:01 +0000)]
Replace Sun RPC license with a 3-clause BSD license.  This license change
was approved in 2010 by Wim Coekaerts, Senior Vice President, Linux and
Virtualization at Oracle Corporation.

10 years agoReplace Sun RPC license with a 3-clause BSD license. This license change
Hiroki Sato [Mon, 9 Dec 2013 04:26:50 +0000 (04:26 +0000)]
Replace Sun RPC license with a 3-clause BSD license.  This license change
was approved in 2010 by Wim Coekaerts, Senior Vice President, Linux and
Virtualization at Oracle Corporation.

10 years agoAdd some initial support for the Intel 6235.
Adrian Chadd [Mon, 9 Dec 2013 03:40:02 +0000 (03:40 +0000)]
Add some initial support for the Intel 6235.

Tested:

* Intel 5100
* Intel 6235

Obtained from: mav, others

10 years agoForgot to utilize feature added in SVN r257782 to prevent truncating
Devin Teske [Mon, 9 Dec 2013 03:38:18 +0000 (03:38 +0000)]
Forgot to utilize feature added in SVN r257782 to prevent truncating
/tmp/bsdinstall_log each time we exec a module.

MFC after: 3 days

10 years agoChase down cryptodeflate.c change from r259109.
Alfred Perlstein [Mon, 9 Dec 2013 02:06:52 +0000 (02:06 +0000)]
Chase down cryptodeflate.c change from r259109.

10 years agoFix failed attempt to send pkg(8) stderr to /dev/null
Devin Teske [Mon, 9 Dec 2013 01:30:20 +0000 (01:30 +0000)]
Fix failed attempt to send pkg(8) stderr to /dev/null

MFC after: 3 days

10 years agoUse correct casts in gcc's emmintrin.h for the first arguments of the
Dimitry Andric [Sun, 8 Dec 2013 23:24:32 +0000 (23:24 +0000)]
Use correct casts in gcc's emmintrin.h for the first arguments of the
following builtin functions:

* __builtin_ia32_pslldi128() takes __v4si instead of __v8hi
* __builtin_ia32_psllqi128() takes __v2di instead of __v8hi
* __builtin_ia32_psradi128() takes __v4si instead of __v8hi

This should fix the following errors when building the LINT kernel with
gcc:

sys/crypto/aesni/aesni_wrap.c:191: error: incompatible type for argument 1 of
'__builtin_ia32_psradi128'
sys/crypto/aesni/aesni_wrap.c:195: error: incompatible type for argument 1 of
'__builtin_ia32_pslldi128'

MFC after: 3 days

10 years agoo Build syscons(9)'s splash support if both sc and splash are enabled.
Aleksandr Rybalko [Sun, 8 Dec 2013 22:49:12 +0000 (22:49 +0000)]
o Build syscons(9)'s splash support if both sc and splash are enabled.
o Include opt_splash.h for vt(9) to know when splash device is enabled.
o Build logo_freebsd.c only if splash and vt are enabled.
o Include opt_compat.h to know when we have to respect compatibility.

Sponsored by: The FreeBSD Foundation

10 years agoFix build with 'crypto' and 'zfs' in the same kernel. This was the path
Peter Wemm [Sun, 8 Dec 2013 21:59:46 +0000 (21:59 +0000)]
Fix build with 'crypto' and 'zfs' in the same kernel.  This was the path
of least pain I could find.

10 years agoWhen comparing device IDs, make sure that they have the same type
Alexander Motin [Sun, 8 Dec 2013 20:43:01 +0000 (20:43 +0000)]
When comparing device IDs, make sure that they have the same type
(like NAA assigned) and identify the same entity (like device or port).
Otherwise there can be false positives since at least some models of
Seagate disks use same IDs for the whole device and one of its ports.

MFC after: 2 weeks

10 years agoEliminate a redundant parameter to vm_radix_replace().
Alan Cox [Sun, 8 Dec 2013 20:07:02 +0000 (20:07 +0000)]
Eliminate a redundant parameter to vm_radix_replace().

Improve the wording of the comment describing vm_radix_replace().

Reviewed by: attilio
MFC after: 6 weeks
Sponsored by: EMC / Isilon Storage Division

10 years agoSome improvements to the casperd manual page.
Pawel Jakub Dawidek [Sun, 8 Dec 2013 19:32:29 +0000 (19:32 +0000)]
Some improvements to the casperd manual page.

Submitted by: emaste

10 years agodrm/radeon: radeon_dp_i2c_aux_ch() must return 0 on FreeBSD
Jean-Sébastien Pédron [Sun, 8 Dec 2013 18:48:07 +0000 (18:48 +0000)]
drm/radeon: radeon_dp_i2c_aux_ch() must return 0 on FreeBSD

The code was unmodified compared to Linux and returned the amount of
received bytes from the i2c bus. This led to non-working i2c bus and
failure to eg. read monitor's EDID, if connected to DisplayPort.

MFC after: 3 days
Tested by: Mikaël Urankar <mikael.urankar@gmail.com>

10 years agocxgbe(4): save a copy of the RSS map for each port for the driver's use.
Navdeep Parhar [Sun, 8 Dec 2013 17:47:37 +0000 (17:47 +0000)]
cxgbe(4): save a copy of the RSS map for each port for the driver's use.

10 years agoRestore a vital comment nuked in r259016.
Marius Strobl [Sun, 8 Dec 2013 15:25:19 +0000 (15:25 +0000)]
Restore a vital comment nuked in r259016.

10 years agodrm/radeon: agp_info->ai_aperture_size is in bytes, not Mbytes
Jean-Sébastien Pédron [Sun, 8 Dec 2013 14:21:54 +0000 (14:21 +0000)]
drm/radeon: agp_info->ai_aperture_size is in bytes, not Mbytes

This fixes radeon_agp_init() and gtt_size is now correct. However, this
is not enough to make Radeon AGP cards work: ttm_agp_backend.c isn't
implemented yet.

Submitted by: tijl@

10 years agoPull in r196658 from upstream clang trunk:
Dimitry Andric [Sun, 8 Dec 2013 13:56:26 +0000 (13:56 +0000)]
Pull in r196658 from upstream clang trunk:

  CodeGen: Don't emit linkage on thunks that aren't emitted because they're
  vararg.

  This can happen when we're trying to emit a thunk with available_externally
  linkage with optimization enabled but bail because it doesn't make sense for
  vararg functions.

  [LLVM] PR18098.

This should fix clang "Broken module found, compilation aborted" errors when
building the qt4-based dvbcut port.

Reported by: se
MFC after: 3 days

10 years agoSimilar to r255816, fix the math for the DELAY() calculation. It was off
Luiz Otavio O Souza [Sun, 8 Dec 2013 13:46:27 +0000 (13:46 +0000)]
Similar to r255816, fix the math for the DELAY() calculation.  It was off
by a really small amount because of the higher timer resolution.

Approved by: adrian (mentor)
Verified on: BBB

10 years agoPartially revert r259080. bde@ pointed out that there are a lot more style bugs
Justin Hibbits [Sun, 8 Dec 2013 09:34:56 +0000 (09:34 +0000)]
Partially revert r259080.  bde@ pointed out that there are a lot more style bugs
going on in here than can be fixed, and I introduced some of my own.  Rather
than fix the whole host of them, back out my bugs.

Found by: bde
X-MFC with: r259080

10 years agoFix typos.
Hans Petter Selasky [Sun, 8 Dec 2013 06:52:22 +0000 (06:52 +0000)]
Fix typos.

Found by: remko

10 years agoRev 256256 had an undocumented side effect of breaking existing behavior
Peter Wemm [Sun, 8 Dec 2013 05:55:55 +0000 (05:55 +0000)]
Rev 256256 had an undocumented side effect of breaking existing behavior
for ipv6 jails.

Among the harmful side effects included putting a route to an entire /64
onto an interface even if you were in a smaller network - eg: /80.
This broke the freebsd.org cluster hosted at ISC which has /80 networks.

10 years agogcc: new fvisibility-ms-compat option
Pedro F. Giffuni [Sun, 8 Dec 2013 03:02:44 +0000 (03:02 +0000)]
gcc: new fvisibility-ms-compat option

Obtained from: gcc 4.3 (rev. 126088; GPLv2)
MFC after: 3 weeks

10 years agoDocument the noncontigwr NFS mount option.
Rick Macklem [Sun, 8 Dec 2013 00:59:04 +0000 (00:59 +0000)]
Document the noncontigwr NFS mount option.
This is a content change.

MFC after: 3 weeks

10 years agoVendor import nvi-2.1.2-c80f493b038 a multikey mapping fix
Peter Wemm [Sun, 8 Dec 2013 00:08:03 +0000 (00:08 +0000)]
Vendor import nvi-2.1.2-c80f493b038 a multikey mapping fix

PR: bin/182463

10 years agoImport nvi2 as of c80f493, including:
Peter Wemm [Sun, 8 Dec 2013 00:04:25 +0000 (00:04 +0000)]
Import nvi2 as of c80f493, including:
Fix cleanup multikey mappgs (FreeBSD bin/182463)

10 years agoUse callout(9) to drive the vlapic timer instead of clocking it on each VM exit.
Neel Natu [Sat, 7 Dec 2013 23:11:12 +0000 (23:11 +0000)]
Use callout(9) to drive the vlapic timer instead of clocking it on each VM exit.

This decouples the guest's 'hz' from the host's 'hz' setting. For e.g. it is
now possible to have a guest run at 'hz=1000' while the host is at 'hz=100'.

Discussed with: grehan@
Tested by: Tycho Nightingale (tycho.nightingale@pluribusnetworks.com)

10 years agoFor software builds, the NFS client does many small
Rick Macklem [Sat, 7 Dec 2013 23:05:59 +0000 (23:05 +0000)]
For software builds, the NFS client does many small
synchronous (with FILE_SYNC) writes because non-contiguous
byte ranges in the same buffer cache block are being
written. This patch adds a new mount option "noncontigwr"
which allows the non-contiguous byte ranges to be combined,
with the dirty byte range becoming the superset of the bytes
that are dirty, if the file has not been file locked.
This reduces the number of writes significantly for software
builds. The only case where this change might break existing
applications is where an application is writing
non-overlapping byte ranges within the same buffer cache block
of a file from multiple clients concurrently.
Since such an application would normally do file locking on
the file, avoiding the byte range merge for files that have
been file locked should be sufficient for most (maybe all?) cases.

Submitted by: jhb (earlier version)
Reviewed by: kib
MFC after: 3 weeks

10 years agoFor WARNS <= 3, change the clang warning flag -Wno-conversion to
Dimitry Andric [Sat, 7 Dec 2013 22:30:07 +0000 (22:30 +0000)]
For WARNS <= 3, change the clang warning flag -Wno-conversion to
-Wno-enum-conversion.  In earlier clang versions (before 3.2), the
latter did not exist, and suppressing enum conversion warnings was
really the goal of this warning suppression flag.

This should enable the same kind of warning again as was fixed by
r259072 ("incompatible integer to pointer conversion passing 'Elf_Addr'
(aka 'unsigned int') to parameter of type 'void *'"), and which was only
emitted by gcc.

Noticed by: kib
MFC after: 3 days

10 years agoMake more unsigned ints signed.
Justin Hibbits [Sat, 7 Dec 2013 22:25:07 +0000 (22:25 +0000)]
Make more unsigned ints signed.

Found by: clang (powerpc64)
MFC after: 1 week

10 years agoIf a vcpu disables its local apic and then executes a 'HLT' then spin down the
Neel Natu [Sat, 7 Dec 2013 22:18:36 +0000 (22:18 +0000)]
If a vcpu disables its local apic and then executes a 'HLT' then spin down the
vcpu and destroy its thread context. Also modify the 'HLT' processing to ignore
pending interrupts in the IRR if interrupts have been disabled by the guest.
The interrupt cannot be injected into the guest in any case so resuming it
is futile.

With this change "halt" from a Linux guest works correctly.

Reviewed by: grehan@
Tested by: Tycho Nightingale (tycho.nightingale@pluribusnetworks.com)

10 years agoFix some integer signs. These unsigned integers should all be signed.
Justin Hibbits [Sat, 7 Dec 2013 19:55:34 +0000 (19:55 +0000)]
Fix some integer signs.  These unsigned integers should all be signed.

Found by: clang (powerpc64)

10 years agoAdd WITH_DVD to RELEASE_RMAKEFLAGS, otherwise it is not actually
Glen Barber [Sat, 7 Dec 2013 19:39:38 +0000 (19:39 +0000)]
Add WITH_DVD to RELEASE_RMAKEFLAGS, otherwise it is not actually
passed to 'make release'.

MFC after: 3 days
X-Before-RC2: yes
Sponsored by: The FreeBSD Foundation

10 years agoCast Elf_Addr to void * to match the free_aligned() argument type.
Konstantin Belousov [Sat, 7 Dec 2013 15:49:16 +0000 (15:49 +0000)]
Cast Elf_Addr to void * to match the free_aligned() argument type.

Found by: gcc
Sponsored by: The FreeBSD Foundation
MFC after: 6 days

10 years agoInclude dev/fb/fb_if.m in build always, without it kms modules complain about
Aleksandr Rybalko [Sat, 7 Dec 2013 15:24:43 +0000 (15:24 +0000)]
Include dev/fb/fb_if.m in build always, without it kms modules complain about
lack of fb_getinfo_desc symbol.

Submitted by: dumbbell

Sponsored by: The FreeBSD Foundation

10 years agoRefactor out the scan id and scan vap as part of the scan work.
Adrian Chadd [Sat, 7 Dec 2013 08:32:15 +0000 (08:32 +0000)]
Refactor out the scan id and scan vap as part of the scan work.

Make the scan state optional - we'll obviously need a vap, but we now
won't require the scan state.  the only thing the scan state is needed
for is to check for the list of SSIDs to scan - which we can now
just plain ignore by passing in NULL as the scan state pointer.

Tested:

* Intel 5100 (STA)

10 years agoAdd a channel parameter to iwn_scan().
Adrian Chadd [Sat, 7 Dec 2013 08:25:24 +0000 (08:25 +0000)]
Add a channel parameter to iwn_scan().

This is in preparation for being able to use iwn_scan() to do an off
channel scan to reset the RF tuning.

It should be a no-op.

Tested:

* Intel 5100 (STA)

10 years agoRefactor out the scan channel to be assigned early on in iwn_scan()
Adrian Chadd [Sat, 7 Dec 2013 08:20:24 +0000 (08:20 +0000)]
Refactor out the scan channel to be assigned early on in iwn_scan()
rather than it all being a mess of 'c' and 'ic->ic_curchan'.

Tested:

* Intel 5100 (STA)

10 years agoBegin fleshing out some code to handle tracking PLCP error rates
Adrian Chadd [Sat, 7 Dec 2013 08:03:10 +0000 (08:03 +0000)]
Begin fleshing out some code to handle tracking PLCP error rates
in preparation for the scan based retune logic.

The linux iwlwifi driver does a rescan (onto a non-active channel)
to force an RF retune when the PLCP error rates exceed a certain threshold.

* Add code to track HT PLCP rate errors;
* Separate out the PLCP error count fetch and update so the delta
  can be used when checking for PLCP error rates;
* Implement the PLCP error logic from iwlwifi;
* For now, just print out whenever the error rate exceeds the
  threshold.

The actual scan based retune will take a bit more effort; the scan
command code right now assumes that a scan state is passed in.
This does need to change to be more flexible (both for this and
in preparation for scanning multiple channels at once.)

Tested:

* 5100 (STA mode)
* 2200 (STA mode)
* 2230 (STA mode)

10 years agoDocument -q and --quiet as discouraged compatibility option.
Xin LI [Sat, 7 Dec 2013 07:12:37 +0000 (07:12 +0000)]
Document -q and --quiet as discouraged compatibility option.

Suggested by: eadler
MFC after: 2 weeks

10 years agoAdd some PLCP thresholds from Linux iwlwifi driver in preparation for
Adrian Chadd [Sat, 7 Dec 2013 06:45:09 +0000 (06:45 +0000)]
Add some PLCP thresholds from Linux iwlwifi driver in preparation for
working on some RF tuning issues.

The linux iwlwifi driver has these thresholds which they use to see
if there are PLCP errors over a certain interval.  If they hit this,
they trigger a single-channel (different from active channels!)
scan to retune the RF front-end.

10 years agoRemove mention of the compatibility option 'q', which is
Xin LI [Sat, 7 Dec 2013 06:27:54 +0000 (06:27 +0000)]
Remove mention of the compatibility option 'q', which is
intentionally undocumented and its only purpose is that
we do not bail out when used as a drop-in replacement of
a different implementation.

PR: docs/184550
MFC after: 2 weeks

10 years agoRemove "Warning: file \"%s\" unreadable". It was introduced with revision
Marcel Moolenaar [Sat, 7 Dec 2013 02:20:22 +0000 (02:20 +0000)]
Remove "Warning: file \"%s\" unreadable". It was introduced with revision
44599 and turned less than useful ever since fortunes-o.dat got removed.
Running "fortunes -a" is enough to always see it for fortunes-o.dat.
Since we compile with debugging enabled, the removal is easily compensated
for by running "fortune -D -D".

10 years agoFix r203922 for "fortune -o": we free tpath when was_malloc is true
Marcel Moolenaar [Sat, 7 Dec 2013 01:35:18 +0000 (01:35 +0000)]
Fix r203922 for "fortune -o": we free tpath when was_malloc is true
but don't define tpath everytime we set was_malloc to true. We end
up calling free() multiple times for the same pointer, resulting in:
<jemalloc>: jemalloc_arena.c:355: Failed assertion: ...
When we set path to offensive, we need to set tpath as well.

10 years agoPerformance and debugging enhancements:
Devin Teske [Sat, 7 Dec 2013 00:31:01 +0000 (00:31 +0000)]
Performance and debugging enhancements:
+ Remove UNAME_P=$(...) from startup/misc -- already supplied by common.subr
+ Use f_getvar instead of $(eval echo \$$var) -- f_getvar is sub-shell free
+ Add `-e' and `-k var' options to f_eval_catch -- increasing use-cases
+ Use f_eval_catch to display errors on failure -- reducing duplicated code
+ Use f_eval_catch when we need output from a command -- improving debugging
+ Optimize f_isinter of strings.subr for performance -- now sub-shell free
+ Improve error checking on pidfiles -- using f_eval_catch and f_isinteger
+ Use $var_to_set arg of f_ifconfig_{inet,netmask} -- eliminate sub-shells
+ Use f_sprintf instead of $(printf ...) -- consolidate sub-shells
+ Use $var_to_set arg of f_route_get_default -- eliminate sub-shells
+ Add f_count to replace $(set -- ...;echo $#) -- eliminate sub-shells
+ Add f_count_ifs to replace $(IFS=x;set -- ...;echo $#) -- no sub-shells
+ Replace var="$var${var:+ }..." in loops with var="$var ..." with a follow-
  up var="${var# }" to trim leading whitespace -- optimize loops
+ Use $var_to_set arg of f_resolv_conf_nameservers -- eliminate sub-shells
+ Comments for the f_eval_catch function
+ Remove a duplicate `local ... desc ...' in f_device_get_all of device.subr
+ Use $var_to_set arg of f_device_capacity -- eliminate sub-shells
+ Whitespace fixes in f_dialog_init of dialog.subr
+ Optimize f_inet_atoi of media/tcpip.subr for performance -- sub-shell free
+ In several cases, send stderr to /dev/null -- clean up runtime execution
+ Change f_err of common.subr to go to program stderr not terminal stderr,
  allowing redirection of output from functions that use f_err
+ Disable debugging when using f_getvar to get variable argument to
  f_startup_rcconf_map_expand of startup/rcconf.subr
+ Use f_replace_all instead of $(echo ... | tr | sed) -- performance
+ Add a $var_to_set option to f_index_{file,menusel_{command,keyword}} of
  common.subr -- centralize sub-shells

10 years agoPull in r196590 from upstream clang trunk (by rdivacky):
Dimitry Andric [Sat, 7 Dec 2013 00:05:45 +0000 (00:05 +0000)]
Pull in r196590 from upstream clang trunk (by rdivacky):

  Move the body of GCCInstallationDetector ctor into an init() function
  and call it from its only user. The linux toolchain. This saves quite
  a lot of directory searching on other platforms.

See http://docs.freebsd.org/cgi/mid.cgi?51E6FAF5.3080802 for the
original discussion.  With this fix, the search for gcc installations is
completely eliminated on FreeBSD.

Reported by: Kurt Lidl <lidl@pix.net>
MFC after: 3 days

10 years agoExpose spa_asize_inflation.
Xin LI [Fri, 6 Dec 2013 23:49:16 +0000 (23:49 +0000)]
Expose spa_asize_inflation.

X-MFC-With: r258632

10 years agoFix td_frame flags for i386.
Aleksandr Rybalko [Fri, 6 Dec 2013 23:12:21 +0000 (23:12 +0000)]
Fix td_frame flags for i386.

Submitted by: jilles

Sponsored by: The FreeBSD Foundation

10 years agoTwo new cxgbetool subcommands to set up scheduler classes and to bind
Navdeep Parhar [Fri, 6 Dec 2013 23:02:57 +0000 (23:02 +0000)]
Two new cxgbetool subcommands to set up scheduler classes and to bind
them to NIC queues.

Obtained from: Chelsio

10 years agosh: Split set -x output into a separate function.
Jilles Tjoelker [Fri, 6 Dec 2013 22:24:37 +0000 (22:24 +0000)]
sh: Split set -x output into a separate function.

10 years agoFix compilation when the "USB_DEBUG" option is set.
Hans Petter Selasky [Fri, 6 Dec 2013 22:13:51 +0000 (22:13 +0000)]
Fix compilation when the "USB_DEBUG" option is set.

10 years agoDisallow optimizations which potentially remove boundary checks
Konstantin Belousov [Fri, 6 Dec 2013 21:44:13 +0000 (21:44 +0000)]
Disallow optimizations which potentially remove boundary checks
for signed values due to a compiler authors considering integer
overflow as impossible.

The change follows suit of other projects taking the same measure.

Sponsored by: The FreeBSD Foundation
MFC after: 1 week

10 years agoFor variant II static TLS, properly align tls segments. Pre-calculate
Konstantin Belousov [Fri, 6 Dec 2013 21:39:45 +0000 (21:39 +0000)]
For variant II static TLS, properly align tls segments.  Pre-calculate
the max required alignment for the static tls segments, and honor it
when carving the pieces for next module, from the static space.  Use
aligned allocator to get properly-aligned dynamic blocks.

Reported by: dt71@gmx.com
Reviewed by: kan
Sponsored by: The FreeBSD Foundation
MFC after: 1 week

10 years agoBuild an allocator for the aligned memory on top of the rtld-private
Konstantin Belousov [Fri, 6 Dec 2013 21:30:31 +0000 (21:30 +0000)]
Build an allocator for the aligned memory on top of the rtld-private
malloc.

Reviewed by: kan
Sponsored by: The FreeBSD Foundation
MFC after: 1 week

10 years agoDo not force to run atexit handlers, which text comes from a dso
Konstantin Belousov [Fri, 6 Dec 2013 21:26:57 +0000 (21:26 +0000)]
Do not force to run atexit handlers, which text comes from a dso
owning the handle passed to __cxa_finalize() but which are registered
by other dso, when the process is inside exit(3).

Running them makes the destruction order wrong, and there is hope that
such destructors would not call dlclose(3), since it is pointless at
this stage of the process existence.

The change effectively disables the r211706 after the exit(3) is
called.

Reported and tested by: Michael Gmelin <freebsd@grem.de>
Analyzed by: dim
Sponsored by: The FreeBSD Foundation
MFC after: 1 week

10 years agomdoc: remove EOL whitespace.
Joel Dahl [Fri, 6 Dec 2013 21:22:33 +0000 (21:22 +0000)]
mdoc: remove EOL whitespace.

10 years agoAlthough not strictly required to boot a 64MB board, bump
Warner Losh [Fri, 6 Dec 2013 18:41:16 +0000 (18:41 +0000)]
Although not strictly required to boot a 64MB board, bump
vm_max_virtual_address to be KERNVIRTADDR + 256MB. This allows some
future shock protection since the KVA requirements have gone up since
the unmapped changes have gone in, as well as preventing us from
overlapping with the hardware devices, which we map at 0xd0000000,
which we'd hit with anything more than 85MB...

MFC after: 3 days

10 years agoFix the pin value reading on AM335x. Because of the inverted logic it was
Luiz Otavio O Souza [Fri, 6 Dec 2013 18:09:10 +0000 (18:09 +0000)]
Fix the pin value reading on AM335x.  Because of the inverted logic it was
always returning '0' for all the reads, even for the outputs.  It is now
known to work with gpioiic(4) and gpioled(4).

Approved by: adrian (mentor)
Tested on: BBB

10 years agoMove the GPIOBUS_SET_PINFLAGS(..., ..., pin, GPIO_PIN_OUTPUT) to led(4)
Luiz Otavio O Souza [Fri, 6 Dec 2013 17:56:20 +0000 (17:56 +0000)]
Move the GPIOBUS_SET_PINFLAGS(..., ..., pin, GPIO_PIN_OUTPUT) to led(4)
control callback function.  This makes gpioled(4) works even if the pin
is accidentally set to an input.

Approved by: adrian (mentor)

10 years agoRemove unnecessary includes and an unused softc variable. While here apply
Luiz Otavio O Souza [Fri, 6 Dec 2013 17:49:34 +0000 (17:49 +0000)]
Remove unnecessary includes and an unused softc variable.  While here apply
two minor style(9) fixes.

Approved by: adrian (mentor)

10 years agoMake the sysctl node read-only.
Luiz Otavio O Souza [Fri, 6 Dec 2013 17:45:14 +0000 (17:45 +0000)]
Make the sysctl node read-only.

Approved by: adrian (mentor)

10 years agoRegen.
Kevin Lo [Fri, 6 Dec 2013 15:27:42 +0000 (15:27 +0000)]
Regen.

10 years agoAdd support for the MediaTek/Ralink RT5572 chipset.
Kevin Lo [Fri, 6 Dec 2013 15:26:39 +0000 (15:26 +0000)]
Add support for the MediaTek/Ralink RT5572 chipset.
Committed over the TP-LINK TL-WDN3200 (RT5572) on amd64 with WPA.

While here, add my copyright.

10 years agoReplace the magic numbers with something more readable.
Kevin Lo [Fri, 6 Dec 2013 15:17:28 +0000 (15:17 +0000)]
Replace the magic numbers with something more readable.

10 years agoCosmetic changes.
Kevin Lo [Fri, 6 Dec 2013 15:15:58 +0000 (15:15 +0000)]
Cosmetic changes.

10 years agoRemove daft KASSERT.
Kevin Lo [Fri, 6 Dec 2013 15:14:18 +0000 (15:14 +0000)]
Remove daft KASSERT.

10 years agoImprove the XHCI command timeout recovery handling code.
Hans Petter Selasky [Fri, 6 Dec 2013 08:42:41 +0000 (08:42 +0000)]
Improve the XHCI command timeout recovery handling code.

MFC after: 1 week

10 years agogcc: Include types in error message for build_binary_op.
Pedro F. Giffuni [Fri, 6 Dec 2013 02:19:58 +0000 (02:19 +0000)]
gcc: Include types in error message for build_binary_op.

Mostly cosmetic change, again to reduce differences with Apple's gcc.

Obtained from: gcc 4.3 (rev. 125239; GPLv2)
MFC after: 3 weeks

10 years agoAdd VT(9) font tools.
Aleksandr Rybalko [Thu, 5 Dec 2013 22:58:05 +0000 (22:58 +0000)]
Add VT(9) font tools.
Reviewed by: nwhitehorn
MFC_to_10_after: re approval

Sponsored by: The FreeBSD Foundation

10 years agoInstall teken.h for userland.
Aleksandr Rybalko [Thu, 5 Dec 2013 22:56:37 +0000 (22:56 +0000)]
Install teken.h for userland.
Part of VT(9) project merge.
Reviewed by: nwhitehorn
MFC_to_10_after: re approval

Sponsored by: The FreeBSD Foundation

10 years agotest: Avoid looking up again the type of a known binary operator.
Jilles Tjoelker [Thu, 5 Dec 2013 22:53:32 +0000 (22:53 +0000)]
test: Avoid looking up again the type of a known binary operator.

10 years agoMerge VT(9) project (a.k.a. newcons).
Aleksandr Rybalko [Thu, 5 Dec 2013 22:38:53 +0000 (22:38 +0000)]
Merge VT(9) project (a.k.a. newcons).

Reviewed by: nwhitehorn
MFC_to_10_after: re approval

Sponsored by: The FreeBSD Foundation

10 years agoFix a typo.
John Baldwin [Thu, 5 Dec 2013 21:58:02 +0000 (21:58 +0000)]
Fix a typo.

10 years agoThere is no sysctl with the MIB { CTL_KERN, KERN_MAXID }.
John Baldwin [Thu, 5 Dec 2013 21:55:10 +0000 (21:55 +0000)]
There is no sysctl with the MIB { CTL_KERN, KERN_MAXID }.

MFC after: 2 weeks

10 years agoFix the processor table entry structure to use a fixed-width type for
John Baldwin [Thu, 5 Dec 2013 21:51:54 +0000 (21:51 +0000)]
Fix the processor table entry structure to use a fixed-width type for
32-bit fields so it is the correct size on amd64.  Remove a workaround
for the broken structure from bhyve(8).

MFC after: 1 week

10 years agoFix debug printfs in FPU_EMU to compile on powerpc64 and enable it for
John Baldwin [Thu, 5 Dec 2013 21:49:14 +0000 (21:49 +0000)]
Fix debug printfs in FPU_EMU to compile on powerpc64 and enable it for
powerpc64.  This fixes the LINT64 kernel config.

Approved by: nwhitehorn (the idea, not the actual patch)

10 years agoDo some cosmetic fixes.
Andreas Tobler [Thu, 5 Dec 2013 21:35:52 +0000 (21:35 +0000)]
Do some cosmetic fixes.

10 years agoIncrease PHYS_AVAIL_SZ because on pSeries machines we can have many logical
Andreas Tobler [Thu, 5 Dec 2013 21:34:33 +0000 (21:34 +0000)]
Increase PHYS_AVAIL_SZ because on pSeries machines we can have many logical
regions which represent the total amount of memory. The size of these regions
is not the physical size of the chip but it is a logical one and it is given
by the OpenFirmware, it is selectable at boot time and varies between 16MB and
256MB in my case. There is an 'automatic' option which would select the size as
64MB in case you have around 16GB of RAM.
To make sure we can allocate RAM with the automatic option bump this value
of PHYS_AVAIL_SZ to 256.

10 years agoIn case we have many continous regions without gaps we have to make soure that
Andreas Tobler [Thu, 5 Dec 2013 21:25:56 +0000 (21:25 +0000)]
In case we have many continous regions without gaps we have to make soure that
we collapse them into one region. The previous version missed to go back one
step and do it again.

10 years agogcc: Add -flax-vector-conversions
Pedro F. Giffuni [Thu, 5 Dec 2013 21:22:51 +0000 (21:22 +0000)]
gcc: Add -flax-vector-conversions

Obtained from: gcc 4.3 (rev. 120572, 120688; GPLv2)

10 years agoInitialize modesetting sysctls in radeonkms.
Robert Millan [Thu, 5 Dec 2013 20:23:32 +0000 (20:23 +0000)]
Initialize modesetting sysctls in radeonkms.

This is intended for MFC if re@ permits.

Reviewed by: kib, dumbbell
Tested by: Steven Chamberlain <steven@pyro.eu.org>
MFC after: 3 days

10 years agoFix external compiler warning about write-only assigned variable.
Hans Petter Selasky [Thu, 5 Dec 2013 07:18:06 +0000 (07:18 +0000)]
Fix external compiler warning about write-only assigned variable.

10 years agoMake panic_reboot_wait_time static.
Colin Percival [Thu, 5 Dec 2013 03:01:41 +0000 (03:01 +0000)]
Make panic_reboot_wait_time static.

Submitted by: jhb

10 years agoAdd -F to flush output after each write. With this, I can
Tom Rhodes [Thu, 5 Dec 2013 01:44:24 +0000 (01:44 +0000)]
Add -F to flush output after each write.  With this, I can
set up a pipe and allow a jr user to watch what I'm doing
by running 'script -F pipefile' on it.

While here, spell out the month in the .Dd tag like other
manual pages.

10 years agoEnable Hyper-V support in i386 GENERIC.
Xin LI [Thu, 5 Dec 2013 00:56:50 +0000 (00:56 +0000)]
Enable Hyper-V support in i386 GENERIC.

MFC after: 2 weeks

10 years agoEnsure WITH_DVD is not empty to prevent 'WITH_DVD= ' from evaluating
Glen Barber [Thu, 5 Dec 2013 00:56:03 +0000 (00:56 +0000)]
Ensure WITH_DVD is not empty to prevent 'WITH_DVD= ' from evaluating
to true.

Submitted by: hrs
MFC after: 3 days
X-MFC-to-10: immediate
Sponsored by: The FreeBSD Foundation

10 years agoSupport Hyper-V on i386:
Xin LI [Thu, 5 Dec 2013 00:54:38 +0000 (00:54 +0000)]
Support Hyper-V on i386:

 - Add 'hyperv' module into build;
 - Allow building Hyper-V support as part of the kernel;
 - Hook Hyper-V build into NOTES.

This is intended for MFC if re@ permits.

MFC after: 3 days

10 years agogcc: On rs6000 update sp_offset depending only on size.
Pedro F. Giffuni [Wed, 4 Dec 2013 21:17:39 +0000 (21:17 +0000)]
gcc: On rs6000 update sp_offset depending only on size.

This fixes a nasty bug introduced in r258651.

Reported and tested by: Justin Hibbits
Obtained from: gcc pre-4.3 (rev. 125116; GPLv2)
MFC after: 2 weeks

10 years agoClean up paragraph remnants.
Sergey Kandaurov [Wed, 4 Dec 2013 20:46:25 +0000 (20:46 +0000)]
Clean up paragraph remnants.

10 years agoApply vendor improvements to oce(4) driver:
Xin LI [Wed, 4 Dec 2013 20:24:18 +0000 (20:24 +0000)]
Apply vendor improvements to oce(4) driver:

 - Add support to 40Gbps devices;
 - Add support to control adaptive interrupt coalescing (AIC)
   via sysctl;
 - Improve support of BE3 devices;

Many thanks to Emulex for their continued support of FreeBSD.

Submitted by: Venkata Duvvuru <VenkatKumar.Duvvuru Emulex Com>
MFC after: 3 days

10 years agoroute(1): Pull static data to the top of the file.
Eitan Adler [Wed, 4 Dec 2013 20:15:53 +0000 (20:15 +0000)]
route(1): Pull static data to the top of the file.

This is a pre-requisisite to some upcoming changes.

Submitted by: Sebastian Huber <sebastian.huber@embedded-brains.de>
Discussed on: -hackers

10 years agoroute(1): Pull static buffer out of the function and into function scope.
Eitan Adler [Wed, 4 Dec 2013 20:13:29 +0000 (20:13 +0000)]
route(1): Pull static buffer out of the function and into function scope.

This will make it easier to link as a library.

Submitted by: Sebastian Huber <sebastian.huber@embedded-brains.de>
Discussed on: -hackers

10 years agoroute(1): Pull static buffer out of the function and into function scope.
Eitan Adler [Wed, 4 Dec 2013 20:08:57 +0000 (20:08 +0000)]
route(1): Pull static buffer out of the function and into function scope.

This will make it easier to link as a library.

Submitted by: Sebastian Huber <sebastian.huber@embedded-brains.de> (older version)
Discussed on: -hackers

10 years agoroute(1): Pull static variable out of the function and into function scope.
Eitan Adler [Wed, 4 Dec 2013 20:07:34 +0000 (20:07 +0000)]
route(1): Pull static variable out of the function and into function scope.

This will make it easier to link as a library.

Submitted by: Sebastian Huber <sebastian.huber@embedded-brains.de> (older version)
Discussed on: -hackers

10 years agoroute(1): Pull static buffer out of the function and into function scope.
Eitan Adler [Wed, 4 Dec 2013 20:03:55 +0000 (20:03 +0000)]
route(1): Pull static buffer out of the function and into function scope.

This will make it easier to link as a library.

Submitted by: Sebastian Huber <sebastian.huber@embedded-brains.de> (older version)
Discussed on: -hackers

10 years agoAdd missing bits from the vendor's 2005-05-04 change to
David E. O'Brien [Wed, 4 Dec 2013 19:15:56 +0000 (19:15 +0000)]
Add missing bits from the vendor's 2005-05-04 change to
contrib/byacc/makefile.in ("add YYPATCH here so it can be tested by
applications") so that applications have a hope of detecting newer
FreeBSD YACC output from an older one.

Submitted by: Juniper Networks

10 years agodrm: Read PCIER_LINK_CAP/PCIER_LINK_CAP2 from the PCI bridge
Jean-Sébastien Pédron [Wed, 4 Dec 2013 19:04:56 +0000 (19:04 +0000)]
drm: Read PCIER_LINK_CAP/PCIER_LINK_CAP2 from the PCI bridge

Before this fix, capabilities were read from vgapci and were incorrect.