]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
14 years agoBump limits on PowerPC. This allows large executables like parts of LLVM
nwhitehorn [Wed, 2 Dec 2009 06:49:22 +0000 (06:49 +0000)]
Bump limits on PowerPC. This allows large executables like parts of LLVM
to function.

Reviewed by: grehan
Obtained from: NetBSD
MFC after: 2 weeks

14 years agoAdd missing ath_ar9* ath hal entries.
thompsa [Wed, 2 Dec 2009 00:38:11 +0000 (00:38 +0000)]
Add missing ath_ar9* ath hal entries.

14 years agoRemove unknown ath hal device entries.
thompsa [Wed, 2 Dec 2009 00:37:03 +0000 (00:37 +0000)]
Remove unknown ath hal device entries.

14 years agoAdd CAM_ATAIO_DMA ATA command flag to mark DMA protocol commands.
mav [Tue, 1 Dec 2009 23:01:29 +0000 (23:01 +0000)]
Add CAM_ATAIO_DMA ATA command flag to mark DMA protocol commands.
It is not needed for SATA controllers, but required for PATA.

14 years agoT3 firmware 7.8.0 for cxgb(4)
np [Tue, 1 Dec 2009 22:23:15 +0000 (22:23 +0000)]
T3 firmware 7.8.0 for cxgb(4)

Obtained from: Chelsio
MFC after: 3 days

14 years agoFix parenthesis typo -- copy full frame pointer for userland callchain,
emaste [Tue, 1 Dec 2009 21:54:53 +0000 (21:54 +0000)]
Fix parenthesis typo -- copy full frame pointer for userland callchain,
not just one byte.

Submitted by: Ryan Stone rysto32 at gmail dot com

14 years agoDon't allocate an input buffer for a TTY when the receiver is turned off.
ed [Tue, 1 Dec 2009 19:14:57 +0000 (19:14 +0000)]
Don't allocate an input buffer for a TTY when the receiver is turned off.

When the termios CREAD flag is not set, it makes little sense to
allocate an input buffer. Just set the size to 0 in this case to reduce
memory footprint.

Disallow CREAD to be disabled for pseudo-devices to prevent
foot-shooting.

14 years agoAdd media ioctl support and link notifications so that devd will attempt
gibbs [Tue, 1 Dec 2009 17:29:25 +0000 (17:29 +0000)]
Add media ioctl support and link notifications so that devd will attempt
to run dhclient on a netfront (xn) device that is setup for DHCP in
/etc/rc.conf.

PR: kern/136251 (fixed differently than the submitted patch)

14 years agoDon't try to bind to an anycast addeess. The KAME IPv6 stack doesn't
ume [Tue, 1 Dec 2009 16:07:50 +0000 (16:07 +0000)]
Don't try to bind to an anycast addeess.  The KAME IPv6 stack doesn't
allow bind to an anycast addeess.  It does away with an annoying
message.

Reviewed by: bz, roberto
MFC after: 2 weeks

14 years agoRemove -t from the manpage and usage.
jhb [Tue, 1 Dec 2009 15:18:25 +0000 (15:18 +0000)]
Remove -t from the manpage and usage.

14 years agoUnbreak user space after if_timer/if_watchdog removal in r199975.
bz [Tue, 1 Dec 2009 14:56:00 +0000 (14:56 +0000)]
Unbreak user space after if_timer/if_watchdog removal in r199975.

Tested by: glebius

14 years agoAdd an .Nm for strncat.
brueffer [Tue, 1 Dec 2009 07:28:56 +0000 (07:28 +0000)]
Add an .Nm for strncat.

PR: 141037
Submitted by: Jeremy Huddleston <jeremyhu@apple.com>
MFC after: 3 days

14 years agoTemporarily revert the previous change because the linker has been
green [Tue, 1 Dec 2009 06:42:47 +0000 (06:42 +0000)]
Temporarily revert the previous change because the linker has been
modified so that it will abort when the environment is bad.

14 years agoDescribe what setpgid(2) does when pgid=0. The text has been
keramida [Tue, 1 Dec 2009 06:12:31 +0000 (06:12 +0000)]
Describe what setpgid(2) does when pgid=0.  The text has been
copied from NetBSD's manpage, and it also matches the behavior
described by the Open Group's online copy of setpgid.2 at
http://www.opengroup.org/onlinepubs/009695399/functions/setpgid.html

Obtained from: NetBSD
Submitted by: Petros Barbayiannis <petrosbarbayiannis@yahoo.gr>
MFC after: 1 week

14 years agoDo not gratuitously fail *env(3) operations due to corrupt ('='-less)
green [Tue, 1 Dec 2009 05:04:31 +0000 (05:04 +0000)]
Do not gratuitously fail *env(3) operations due to corrupt ('='-less)
**environ entries.  This puts non-getenv(3) operations in line with
getenv(3) in that bad environ entries do not cause all operations to
fail.  There is still some inconsistency in that getenv(3) in the
absence of any environment-modifying operation does not emit corrupt
environ entry warnings.

I also fixed another inconsistency in getenv(3) where updating the
global environ pointer would not be reflected in the return values.
It would have taken an intermediary setenv(3)/putenv(3)/unsetenv(3)
in order to see the change.

14 years agoFix local root vulnerability.
cperciva [Tue, 1 Dec 2009 02:57:06 +0000 (02:57 +0000)]
Fix local root vulnerability.

Security: Advisory will be coming soon.
X-MFC-After: 30 seconds

14 years agoRemove if_timer/if_watchdog now that they are no longer used. The space
jhb [Mon, 30 Nov 2009 21:25:57 +0000 (21:25 +0000)]
Remove if_timer/if_watchdog now that they are no longer used.  The space
used by if_timer is reserved for expanding if_index to an int in the
future.

Reviewed by: rwatson, brooks

14 years agoRemove extra parantheses from usb_ethernet.c and usb_serial.c lines.
jhb [Mon, 30 Nov 2009 21:03:44 +0000 (21:03 +0000)]
Remove extra parantheses from usb_ethernet.c and usb_serial.c lines.
config(8) doesn't parse parantheses and instead treated them as being
part of the device driver name (e.g. '(u3g' vs 'u3g').  While here, fix the
style of these long lines to match the wrapping used for other long lines
in this file.

Submitted by: Brett Glass
MFC after: 1 week

14 years agoUse switch out (SWO) instead of switch in (SWI) debug log mask in csw_out.
emaste [Mon, 30 Nov 2009 20:41:30 +0000 (20:41 +0000)]
Use switch out (SWO) instead of switch in (SWI) debug log mask in csw_out.

14 years agoFree pointer from getenv() when done with it.
emaste [Mon, 30 Nov 2009 18:26:46 +0000 (18:26 +0000)]
Free pointer from getenv() when done with it.

Submitted by: Phil Longstaff

14 years agoamdsbwd: new driver for AMD SB600/SB7xx watchdog timer
avg [Mon, 30 Nov 2009 11:44:03 +0000 (11:44 +0000)]
amdsbwd: new driver for AMD SB600/SB7xx watchdog timer

The hardware is compliant with WDRT specification, so I originally
considered including generic WDRT watchdog support, but decided
against it, because I couldn't find anyone to the code for me.
WDRT seems to be not very popular.
Besides, generic WDRT porbably requires a slightly different driver
approach.

Reviewed by: des, gavin, rpaulo
MFC after: 3 weeks

14 years agox86 cpu features: add MOVBE reporting and flag
avg [Mon, 30 Nov 2009 11:11:08 +0000 (11:11 +0000)]
x86 cpu features: add MOVBE reporting and flag

The check is glimpsed from Linux and OpenSolaris.
MOVBE instruction is found in Intel Atom processors.

14 years agoMerge Scott Long's latest blkfront now that the licensing issues are resolved
kmacy [Mon, 30 Nov 2009 04:32:34 +0000 (04:32 +0000)]
Merge Scott Long's latest blkfront now that the licensing issues are resolved

14 years agoUpdate license to reflect terms in xen 2.0 as of the time when the driver was ported
kmacy [Mon, 30 Nov 2009 04:20:43 +0000 (04:20 +0000)]
Update license to reflect terms in xen 2.0 as of the time when the driver was ported
to FreeBSD

14 years agoUpdate to BIND 9.6.1-P2. The vulnerability this is designed to fix is
dougb [Mon, 30 Nov 2009 03:38:34 +0000 (03:38 +0000)]
Update to BIND 9.6.1-P2. The vulnerability this is designed to fix is
related to DNSSEC validation on a resolving name server that allows
access to untrusted users. If your system does not fall into all 3 of
these categories you do not need to update immediately.

14 years agoDisable job control when running 'sh -i' in the testsuite.
jilles [Sun, 29 Nov 2009 22:58:10 +0000 (22:58 +0000)]
Disable job control when running 'sh -i' in the testsuite.

Job control tty manipulations sometimes cause the tests to stop (SIGTTOU
and the like) when run from the 'prove' tool.

14 years agoFix some cases where file descriptors from redirections leak to programs.
jilles [Sun, 29 Nov 2009 22:33:59 +0000 (22:33 +0000)]
Fix some cases where file descriptors from redirections leak to programs.

- Redirecting fds that were not open before kept two copies of the
  redirected file.
    sh -c '{ :; } 7>/dev/null; fstat -p $$; true'
    (both fd 7 and 10 remained open)
- File descriptors used to restore things after redirection were not
  set close-on-exec, instead they were explicitly closed before executing
  a program normally and before executing a shell procedure. The latter
  must remain but the former is replaced by close-on-exec.
    sh -c 'exec 7</; { exec fstat -p $$; } 7>/dev/null; true'
    (fd 10 remained open)

The examples above are simpler than the testsuite because I do not want to
use fstat or procstat in the testsuite.

14 years agoCorrect typo.
bz [Sun, 29 Nov 2009 21:03:54 +0000 (21:03 +0000)]
Correct typo.

Reported by: gabor
MFC after: 5 days

14 years agoAdd atp(4) to powerpc GENERIC. Most late-generation Apple PowerPC laptops
nwhitehorn [Sun, 29 Nov 2009 20:51:23 +0000 (20:51 +0000)]
Add atp(4) to powerpc GENERIC. Most late-generation Apple PowerPC laptops
have trackpads that do not work at all without this driver.

14 years agoEarly-generation touchpads do not send periodic calibration frames for
nwhitehorn [Sun, 29 Nov 2009 20:48:19 +0000 (20:48 +0000)]
Early-generation touchpads do not send periodic calibration frames for
baseline subtraction, and are very temperature sensitive, so would slowly
drift out of a calibrated state when under load. Escape this by taking
the last frame before we decide that the pad is idle as a finger-free
baseline.

Tested on: iBook G4

14 years agoEnable IPcomp by default.
bz [Sun, 29 Nov 2009 20:47:43 +0000 (20:47 +0000)]
Enable IPcomp by default.

PR: kern/123587
MFC after: 5 days

14 years agoAdd more statistics variables for IPcomp.
bz [Sun, 29 Nov 2009 20:37:30 +0000 (20:37 +0000)]
Add more statistics variables for IPcomp.

Try to version the struct in a backward compatible way.
People asked for the versioning of the stats structs in general before.

MFC after: 5 days

14 years agoMove the sysctl related fields to the end of the structure and
marcel [Sun, 29 Nov 2009 20:17:50 +0000 (20:17 +0000)]
Move the sysctl related fields to the end of the structure and
make them conditional upon _KERNEL. libkvm includes <sys/pcpu.h>
and <sys/sysctl.h> does not expose the structure definitions to
userland.

14 years agoIn case the compression result is the same size as the orignal version,
bz [Sun, 29 Nov 2009 17:53:57 +0000 (17:53 +0000)]
In case the compression result is the same size as the orignal version,
the compression was useless as well.  Make sure to not update the data
and return, else we would waste resources when decompressing.

This also avoids the copyback() changing data other consumers like
xform_ipcomp.c would have ignored because of no win and sent out without
noting that compression was used, resulting in invalid packets at the
receiver.

MFC after: 5 days

14 years agoAssimilate very similar input and output code paths
bz [Sun, 29 Nov 2009 17:47:49 +0000 (17:47 +0000)]
Assimilate very similar input and output code paths
(no real functional change).

MFC after: 5 days

14 years agoAdd SDT iter probes forgotten in r199885.
bz [Sun, 29 Nov 2009 17:46:40 +0000 (17:46 +0000)]
Add SDT iter probes forgotten in r199885.

MFC after: 5 days

14 years agoOnly add the IPcomp header if crypto reported success and we have a lower
bz [Sun, 29 Nov 2009 10:53:34 +0000 (10:53 +0000)]
Only add the IPcomp header if crypto reported success and we have a lower
payload size.  Before we had always added the header, no matter if we
actually send out compressed data or not.

With this, after the opencrypto/deflate changes, IPcomp starts to work
apart from edge cases.  Leave it disabled by default until those are
fixed as well.

PR: kern/123587
MFC after: 5 days

14 years agoDecompose <sys/termios.h>.
ed [Sat, 28 Nov 2009 23:50:48 +0000 (23:50 +0000)]
Decompose <sys/termios.h>.

The <sys/termios.h> header file is hardlinked to <termios.h>. It
contains both the structures and the flag definitions, but also the C
library interface that's implemented by the C library.

This header file has the typical problem of including too many random
things and being badly ordered. Instead of trying to fix this, decompose
it into two header files:

- <sys/_termios.h>, which contains struct termios and the flags.
- <termios.h>, which includes <sys/_termios.h> and contains the C
  library interface.

This means userspace has to include <termios.h> for struct termios,
while kernelspace code has to include <sys/tty.h>. Also add a
<sys/termios.h>, which prints a warning message before including
<termios.h>. I am aware that there are some applications that use this
header file as well.

14 years agoRemove whitespace.
bz [Sat, 28 Nov 2009 21:42:39 +0000 (21:42 +0000)]
Remove whitespace.

MFC after: 6 days

14 years agoDirectly send data uncompressed if the packet payload size is lower than
bz [Sat, 28 Nov 2009 21:40:57 +0000 (21:40 +0000)]
Directly send data uncompressed if the packet payload size is lower than
the compression algorithm threshold.

MFC after: 6 days

14 years agoChange memory managment from a fixed size array to a list.
bz [Sat, 28 Nov 2009 21:08:19 +0000 (21:08 +0000)]
Change memory managment from a fixed size array to a list.
This is needed to avoid running into out of buffer situations
where we cannot alloc a new buffer because we hit the array size
limit (ZBUF).
Use a combined allocation for the struct and the actual data buffer
to not increase the number of malloc calls. [1]

Defer initialization of zbuf until we actually need it.

Make sure the output buffer will be large enough in all cases.

Details discussed with: kib [1]
Reviewed by: kib [1]
MFC after: 6 days

14 years agoCorrect a typo.
bz [Sat, 28 Nov 2009 21:01:26 +0000 (21:01 +0000)]
Correct a typo.

MFC after: 6 days

14 years agoEliminate teh use of MAXCPU in static arrays of interrupt counters by
marcel [Sat, 28 Nov 2009 21:01:15 +0000 (21:01 +0000)]
Eliminate teh use of MAXCPU in static arrays of interrupt counters by
adding statistics counters to the PCPU structure. Export the counters
through sysctl by giving each PCPU structure its own sysctl context.

While here, fix cnt.v_intr by not just having it count clock interrupts,
but every interrupt and add more counters for each interrupt source.

14 years agoAdd support for interpreting taps on ADB touchpads as a button click.
nwhitehorn [Sat, 28 Nov 2009 17:48:25 +0000 (17:48 +0000)]
Add support for interpreting taps on ADB touchpads as a button click.

Submitted by: Andreas Tobler <andreast-list at fgznet dot ch>
MFC after: 2 weeks

14 years agoZ_PARTIAL_FLUSH is marked deprecated. Z_SYNC_FLUSH is the suggested
bz [Sat, 28 Nov 2009 17:44:57 +0000 (17:44 +0000)]
Z_PARTIAL_FLUSH is marked deprecated. Z_SYNC_FLUSH is the suggested
replacement but only use it for inflate. For deflate use Z_FINISH
as Z_SYNC_FLUSH adds a trailing marker in some cases that inflate(),
despite the comment in zlib, does npt seem to cope well with, resulting
in errors when uncompressing exactly fills the outbut buffer without
a Z_STREAM_END and a successive call returns an error.

MFC after: 6 days

14 years agoAdd a CPU features framework on PowerPC and simplify CPU setup a little
nwhitehorn [Sat, 28 Nov 2009 17:33:19 +0000 (17:33 +0000)]
Add a CPU features framework on PowerPC and simplify CPU setup a little
more. This provides three new sysctls to user space:
hw.cpu_features - A bitmask of available CPU features
hw.floatingpoint - Whether or not there is hardware FP support
hw.altivec - Whether or not Altivec is available

PR: powerpc/139154
MFC after: 10 days

14 years agoAdd SDT probes for opencrypto:deflate:deflate_gobal:*.
bz [Sat, 28 Nov 2009 17:20:41 +0000 (17:20 +0000)]
Add SDT probes for opencrypto:deflate:deflate_gobal:*.
They are not nice but they were helpful.

MFC after: 6 days

14 years agoDefine an SDT provider for "opencrypto".
bz [Sat, 28 Nov 2009 16:54:18 +0000 (16:54 +0000)]
Define an SDT provider for "opencrypto".

MFC after: 6 days

14 years agoAdd SDT_PROBE[1-5] in the same way we have SDT_PROBE_DEFINE[1-5] to
bz [Sat, 28 Nov 2009 16:47:42 +0000 (16:47 +0000)]
Add SDT_PROBE[1-5] in the same way we have SDT_PROBE_DEFINE[1-5] to
avoid having to add all the unused trailing arguments as zeros.

MFC after: 6 days

14 years agoInclude <sys/tty.h> instead of <sys/termios.h>.
ed [Sat, 28 Nov 2009 16:30:06 +0000 (16:30 +0000)]
Include <sys/tty.h> instead of <sys/termios.h>.

Right now <sys/termios.h> includes <sys/ttycom.h>, which provides the
TTY ioctls to the svr4 code. We need both struct termios and the ioctls,
so include <sys/tty.h> for now.

14 years agoInclude <sys/ttydefaults.h>, instead of doing it through <sys/termios.h>.
ed [Sat, 28 Nov 2009 16:25:55 +0000 (16:25 +0000)]
Include <sys/ttydefaults.h>, instead of doing it through <sys/termios.h>.

I want to prevent the header polution of <sys/termios.h> eventually.

14 years agoRevert r199830 for now. Too many ports dlopen() libraries linked with
kib [Sat, 28 Nov 2009 14:34:28 +0000 (14:34 +0000)]
Revert r199830 for now. Too many ports dlopen() libraries linked with
libthr, but forgot to link main binary with it.

14 years agoAllow to load not-openable dso when tracing. This fixes ldd on such dso or
kib [Sat, 28 Nov 2009 14:29:32 +0000 (14:29 +0000)]
Allow to load not-openable dso when tracing. This fixes ldd on such dso or
dso linked to non-openable object.
Remove '\n' at the end of error message.
End comments with dot.

MFC after: 3 weeks (together with r199829)

14 years agoRemove unneeded inclusion of <sys/termios.h>.
ed [Sat, 28 Nov 2009 13:20:29 +0000 (13:20 +0000)]
Remove unneeded inclusion of <sys/termios.h>.

14 years agoProvide a set of sysctls and tunables to disable device node creation
trasz [Sat, 28 Nov 2009 11:57:43 +0000 (11:57 +0000)]
Provide a set of sysctls and tunables to disable device node creation
for specific "kinds" of disk labels - for example, GPT UUIDs.  Reason
for this is that sometimes, other GEOM classes attach to these device
nodes instead of the proper ones - e.g. they attach to /dev/gptid/XXX
instead of /dev/ada0p2, which is annoying.

Reviewed by: pjd (earlier version)
MFC after: 1 month

14 years agoUse <termios.h> instead of <sys/termios.h>.
ed [Sat, 28 Nov 2009 11:57:25 +0000 (11:57 +0000)]
Use <termios.h> instead of <sys/termios.h>.

<sys/termios.h> only works on FreeBSD by accident.

14 years ago- correct xref sections
danger [Sat, 28 Nov 2009 11:27:37 +0000 (11:27 +0000)]
- correct xref sections

PR: docs/140940
Submitted by: Bruce Cran <bruce@cran.org.uk>
MFC after: 1 week

14 years agoRemove unneeded includes of <sys/termios.h>.
ed [Sat, 28 Nov 2009 11:13:50 +0000 (11:13 +0000)]
Remove unneeded includes of <sys/termios.h>.

14 years ago- correctly render the provided example
danger [Sat, 28 Nov 2009 11:05:22 +0000 (11:05 +0000)]
- correctly render the provided example

PR: docs/140962
Submitted by: mharo

14 years agoProperly synchronize the previous change.
alc [Sat, 28 Nov 2009 00:50:09 +0000 (00:50 +0000)]
Properly synchronize the previous change.

14 years agoSupport the new VM_PROT_COPY option on wired pages. The effect of which
alc [Fri, 27 Nov 2009 22:08:29 +0000 (22:08 +0000)]
Support the new VM_PROT_COPY option on wired pages.  The effect of which
is that a debugger can now set a breakpoint in a program that uses mlock(2)
on its text segment or mlockall(2) on its entire address space.

14 years agoSimplify the invocation of vm_fault(). Specifically, eliminate the flag
alc [Fri, 27 Nov 2009 20:24:11 +0000 (20:24 +0000)]
Simplify the invocation of vm_fault().  Specifically, eliminate the flag
VM_FAULT_DIRTY.  The information provided by this flag can be trivially
inferred by vm_fault().

Discussed with: kib

14 years agounifdef: fix invalid array access when nesting limit exceeded
fanf [Fri, 27 Nov 2009 17:53:49 +0000 (17:53 +0000)]
unifdef: fix invalid array access when nesting limit exceeded

If the number of nested #if blocks exceeds 64, nest() increments
the nesting depth and then reports an error.  The message includes
the line number for the start of the current #if block, which is
read from past the end of the relevant array.

Avoid the out-of-bounds read by reporting the error and exiting
before the nesting depth has a chance to increase.

Submitted by: Jonathan Nieder <jrnieder@gmail.com>

14 years agoUse the default stack size for the iterator thread.
tuexen [Fri, 27 Nov 2009 17:25:19 +0000 (17:25 +0000)]
Use the default stack size for the iterator thread.
This fixes a crash reported by Irene Ruengeler.

Approved by: rrs (mentor)
MFC after: 1 month

14 years agoRemove spinconsole from pc98, some parts seem to be missed and it's
sobomax [Fri, 27 Nov 2009 13:19:06 +0000 (13:19 +0000)]
Remove spinconsole from pc98, some parts seem to be missed and it's
too late (early?) to figure out what exactly.

Reported by: TAKAHASHI Yoshihiro

14 years agoFix the cross-reference to sctp_opt_info in the text - section 3, not 2.
roam [Fri, 27 Nov 2009 13:08:25 +0000 (13:08 +0000)]
Fix the cross-reference to sctp_opt_info in the text - section 3, not 2.

PR: 140938
Submitted by: Bruce Cran <bruce@cran.org.uk>
MFC after: 2 weeks

14 years agoProperly use the envp argument in execvPe().
ed [Fri, 27 Nov 2009 13:05:14 +0000 (13:05 +0000)]
Properly use the envp argument in execvPe().

execvPe() is called by _execvpe(), which we added to implement
posix_spawnp(). We just took execvP() and added the envp argument.
Unfortunately we forgot to change the implementation to use envp over
environ.

This fixes the following piece of code:

| char * const arg[2] = { "env", NULL };
| char * const env[2] = { "FOO=BAR", NULL };
| posix_spawnp(NULL, "/usr/bin/env", NULL, NULL, arg, env);

MFC after: 2 weeks

14 years agoo HEAD is FreeBSD 9-current now.
maxim [Fri, 27 Nov 2009 07:55:39 +0000 (07:55 +0000)]
o HEAD is FreeBSD 9-current now.

Submitted by: Alex Kozlov

14 years agoo FreeBSD 8.0 added.
maxim [Fri, 27 Nov 2009 06:23:47 +0000 (06:23 +0000)]
o FreeBSD 8.0 added.

14 years agoWhitespace-only: another instance of identation with spaces.
sobomax [Fri, 27 Nov 2009 04:00:52 +0000 (04:00 +0000)]
Whitespace-only: another instance of identation with spaces.

14 years agoWhitespace on: use tabs for identation consistently.
sobomax [Fri, 27 Nov 2009 03:58:21 +0000 (03:58 +0000)]
Whitespace on: use tabs for identation consistently.

14 years agoAdd new loader console type: "spinconsole". This console selects the
sobomax [Fri, 27 Nov 2009 03:55:42 +0000 (03:55 +0000)]
Add new loader console type: "spinconsole". This console selects the
video console which doesn't take any input from keyboard and hides
all output replacing it with ``spinning'' character (useful for
embedded products and custom installations).

Sponsored by: Sippy Software, Inc.

14 years agoi386 has not (yet) any DEV_ATPIC conditional than axe it out from Xen
attilio [Fri, 27 Nov 2009 01:02:17 +0000 (01:02 +0000)]
i386 has not (yet) any DEV_ATPIC conditional than axe it out from Xen
version.

No objections by: kmacy

14 years agoAdd two Cirrus Logic codec IDs.
mav [Thu, 26 Nov 2009 20:25:57 +0000 (20:25 +0000)]
Add two Cirrus Logic codec IDs.
Add GPIO setting quirk for Apple MacBookPro5,5.

Submitted by: ed

14 years agoReset path name back to original correctly in fts_build() when
jh [Thu, 26 Nov 2009 19:11:44 +0000 (19:11 +0000)]
Reset path name back to original correctly in fts_build() when
FTS_NOCHDIR option is used. fts_build() could strip a trailing slash
from path name in post-order visit if a path pointing to an empty
directory was given for fts_open().

PR: bin/133907, kern/134513
Reviewed by: das
Approved by: trasz (mentor)
MFC after: 1 month

14 years agoClarify that the value of the fts_info field is different in post-order.
jh [Thu, 26 Nov 2009 19:09:10 +0000 (19:09 +0000)]
Clarify that the value of the fts_info field is different in post-order.

Discussed with: das
Approved by: trasz (mentor)
MFC after: 1 week

14 years agounifdefall: optimise the loop that builds the unifdef command.
fanf [Thu, 26 Nov 2009 19:08:33 +0000 (19:08 +0000)]
unifdefall: optimise the loop that builds the unifdef command.

The old code used a shell loop to convert each controlling macro
definition into a command-line argument, reading the macro definitions
file each time. The new code converts the list of controlling macros
into a sed script which can run through the list of macro definitions
in one go.

Add some explanatory comments, since the code is quite meta.

Use {} instead of () for redirecting a group of commands.

Submitted by: Jonathan Nieder <jrnieder@gmail.com>

14 years agoLibthr cannot be dynamically loaded into the running process.
kib [Thu, 26 Nov 2009 14:01:14 +0000 (14:01 +0000)]
Libthr cannot be dynamically loaded into the running process.
Mark it with -z nodlopen for now.

Discussed with: jhb, kan
MFC after: 3 weeks

14 years agoImplement rtld part of the support for -z nodlopen (see ld(1)).
kib [Thu, 26 Nov 2009 13:57:20 +0000 (13:57 +0000)]
Implement rtld part of the support for -z nodlopen (see ld(1)).

Reviewed by: kan
MFC after: 3 weeks

14 years agoFlag controlling origin expansion in DT_FLAGS is DF_ORIGIN, not DF_1_ORIGIN.
kib [Thu, 26 Nov 2009 13:55:49 +0000 (13:55 +0000)]
Flag controlling origin expansion in DT_FLAGS is DF_ORIGIN, not DF_1_ORIGIN.

Reviewed by: kan
MFC after: 3 days

14 years agoImplement sighold, sigignore, sigpause, sigrelse, sigset functions
kib [Thu, 26 Nov 2009 13:49:37 +0000 (13:49 +0000)]
Implement sighold, sigignore, sigpause, sigrelse, sigset functions
from SUSv4 XSI. Note that the functions are obsoleted, and only
provided to ease porting from System V-like systems. Since sigpause
already exists in compat with different interface, XSI sigpause is
named xsi_sigpause.

Reviewed by: davidxu
MFC after: 3 weeks

14 years agosigset() is the name of function specified by SUSv4.
kib [Thu, 26 Nov 2009 13:41:15 +0000 (13:41 +0000)]
sigset() is the name of function specified by SUSv4.
Replace it to avoid conflict.

MFC after: 3 weeks

14 years agoAdd IBM ServeRAID-MR10i to the hardware list.
brueffer [Thu, 26 Nov 2009 13:25:07 +0000 (13:25 +0000)]
Add IBM ServeRAID-MR10i to the hardware list.

Submitted by: pluknet <pluknet@gmail.com>
MFC after: 3 days

14 years agoDrop USB mass storage devices support from ata(4). It is out of the build as
mav [Thu, 26 Nov 2009 12:41:43 +0000 (12:41 +0000)]
Drop USB mass storage devices support from ata(4). It is out of the build as
long as I remember, and completely superseded by better maintained umass(4).
It's main idea was to optionally avoid CAM dependency for such devices, but
with move ATA to CAM, it is not actual any more.

No objections: hselasky@, thompsa@, arch@

14 years agoMFp4:
mav [Thu, 26 Nov 2009 08:49:46 +0000 (08:49 +0000)]
MFp4:
Improve ATA mode/SATA revision control.

14 years agoReplace VM_PROT_OVERRIDE_WRITE by VM_PROT_COPY. VM_PROT_OVERRIDE_WRITE has
alc [Thu, 26 Nov 2009 05:16:07 +0000 (05:16 +0000)]
Replace VM_PROT_OVERRIDE_WRITE by VM_PROT_COPY.  VM_PROT_OVERRIDE_WRITE has
represented a write access that is allowed to override write protection.
Until now, VM_PROT_OVERRIDE_WRITE has been used to write breakpoints into
text pages.  Text pages are not just write protected but they are also
copy-on-write.  VM_PROT_OVERRIDE_WRITE overrides the write protection on the
text page and triggers the replication of the page so that the breakpoint
will be written to a private copy.  However, here is where things become
confused.  It is the debugger, not the process being debugged that requires
write access to the copied page.  Nonetheless, the copied page is being
mapped into the process with write access enabled.  In other words, once the
debugger sets a breakpoint within a text page, the program can write to its
private copy of that text page.  Whereas prior to setting the breakpoint, a
SIGSEGV would have occurred upon a write access.  VM_PROT_COPY addresses
this problem.  The combination of VM_PROT_READ and VM_PROT_COPY forces the
replication of a copy-on-write page even though the access is only for read.
Moreover, the replicated page is only mapped into the process with read
access, and not write access.

Reviewed by: kib
MFC after: 4 weeks

14 years ago- New style of jail(8) usage requires "-c" argument to create a jail.
kuriyama [Thu, 26 Nov 2009 03:26:59 +0000 (03:26 +0000)]
- New style of jail(8) usage requires "-c" argument to create a jail.

Reviewed by: jamie

14 years agoRemove debugging remnants from unifdefall.
fanf [Thu, 26 Nov 2009 02:14:08 +0000 (02:14 +0000)]
Remove debugging remnants from unifdefall.

Submitted by: Jonathan Nieder <jrnieder@gmail.com>

14 years agoRemove overuse of exclamation marks in kernel printfs, there mere fact a
thompsa [Thu, 26 Nov 2009 00:43:17 +0000 (00:43 +0000)]
Remove overuse of exclamation marks in kernel printfs, there mere fact a
message has been printed is enough to get someones attention. Also remove the
line number for DPRINTF/DPRINTFN, it already prints the funtion name and a
unique message.

14 years agoFix a performance bug in factor(6).
fanf [Thu, 26 Nov 2009 00:38:13 +0000 (00:38 +0000)]
Fix a performance bug in factor(6).

Check if large factor is prime before applying Pollard's algorithm;
fixes "factor 2147483647111311".  Increase base if p-1 algorithm
reaches 1; fixes "factor 99999999999991".

Testcases from David A Bagley <bagleyd@tux.org>.
Fixes from Joseph Myers <jsm@NetBSD.org>.
Problem rediscovered by an attempt to factor my phone number.

A few other incidental fixes: correct a couple of factually incorrect
comments; use ident string macros; move from 4-clause to 3-clause
BSD licence (University of California copyright).

Obtained from: NetBSD

14 years agoDisable interrupts after doing early takeover of the usb controller in case usb
thompsa [Wed, 25 Nov 2009 20:50:43 +0000 (20:50 +0000)]
Disable interrupts after doing early takeover of the usb controller in case usb
isnt actually compiled in (or kldloaded) as the controller could cause spurious
interrupts.

Tested by: Florian Smeets

14 years agoUpdate unifdef to my upstream version 1.188
fanf [Wed, 25 Nov 2009 20:23:18 +0000 (20:23 +0000)]
Update unifdef to my upstream version 1.188

Main highlights:

(A) The new -B option compresses blank lines around a deleted section
    so that blank lines around "paragraphs" of code don't get doubled.

(B) Lenient evaluation of && and || so that #if expressions can be
    evaluated even when some of their sub-expressions cannot be.

(C) The evaluator can now handle macros with arguments.

(D) Portability fixes, especially for unifdefall.

Contributions from:
Ben Hutchings at Solarflare Communications (A and B)
Anders H Kaseorg <andersk@mit.edu> (A and C)
Jonathan Nieder <jrnieder@gmail.com> (D)

Obtained from:  http://dotat.at/prog/unifdef/

14 years agoexit if "make distribition" fails
julian [Wed, 25 Nov 2009 19:00:30 +0000 (19:00 +0000)]
exit if "make distribition" fails

Submitted by: patrick tracanelli
MFC after: 1 week

14 years agoFix typo which inversed the logic which in turn disabled MSI.
yongari [Wed, 25 Nov 2009 17:51:14 +0000 (17:51 +0000)]
Fix typo which inversed the logic which in turn disabled MSI.

Pointy hat to:  yongari

14 years agoMake sure one shot MSI is enabled.
yongari [Wed, 25 Nov 2009 17:30:38 +0000 (17:30 +0000)]
Make sure one shot MSI is enabled.

Submitted by: marius

14 years agoBe nice, don't use the f-word.
trasz [Wed, 25 Nov 2009 16:36:07 +0000 (16:36 +0000)]
Be nice, don't use the f-word.

14 years agoChange gcore in order to get rid of the procfs accesses and use FreeBSD's
attilio [Wed, 25 Nov 2009 15:23:14 +0000 (15:23 +0000)]
Change gcore in order to get rid of the procfs accesses and use FreeBSD's
specific sysctls and ptrace interfaces.
This change switches a bit gcore POLA that is summarized here:
- now gcore can recognize threads within the process and handle dumps
  on thread-scope
- the process to be analyzed will be stopped during its gcore run
- gcore may not work with processes which are actively being analyzed
  by gdb or truss
- the ptrace interface may cause syscalls to return EINTR, thus
  interferring with signals handling within the process

Side note: <janitor task> the interface can be further lifted in order to
get rid of the very last procfs interfaces remnants and made more
suitable for copying with sysctl/ptrace interface </janitor task>.

Obtained from: Sandvine Incorporated
Reviewed by: emaste, rwatson
Sponsored by: Sandvine Incorporated
MFC: 1 month

14 years agoAvoid sshd, cron, syslogd and inetd to be killed under high-pressure swap
attilio [Wed, 25 Nov 2009 15:12:24 +0000 (15:12 +0000)]
Avoid sshd, cron, syslogd and inetd to be killed under high-pressure swap
environments.
Please note that this can't be done while such processes run in jails.

Note: in future it would be interesting to find a way to do that
selectively for any desired proccess (choosen by user himself), probabilly
via a ptrace interface or whatever.

Obtained from: Sandvine Incorporated
Reviewed by: emaste, arch@
Sponsored by: Sandvine Incorporated
MFC: 1 month

14 years agoAdd the possibility to show informations about dropped packets on the
attilio [Wed, 25 Nov 2009 15:02:32 +0000 (15:02 +0000)]
Add the possibility to show informations about dropped packets on the
input path when showing informations about the interfaces.

Obtained from: Sandvine Incorporated
Reviewed by: emaste
Sponsored by: Sandvine Incorporated
MFC: 2 weeks

14 years agoIn tac_get_av_value() empty attributes should be handled like 0-length
attilio [Wed, 25 Nov 2009 14:59:28 +0000 (14:59 +0000)]
In tac_get_av_value() empty attributes should be handled like 0-length
strings rather than unset strings.
Fix the present wrong behaviour.

Obtained from: Sandvine Incorporated
Reviewed by: emaste
Sponsored by: Sandvine Incorporated
MFC: 1 week