]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
14 years agoUse more proper terms (from official documents) for AMD CPU's.
obrien [Wed, 10 Mar 2010 06:10:39 +0000 (06:10 +0000)]
Use more proper terms (from official documents) for AMD CPU's.

Reviewed by: imp

14 years agoAdd -i to the first post-install mergemaster example to make
dougb [Wed, 10 Mar 2010 05:44:57 +0000 (05:44 +0000)]
Add -i to the first post-install mergemaster example to make
it consistent with the other. [1]

Add a note about -U to the mergemaster footnote.

Submitted by: obrien [1]

14 years agoProvide workaround for the ages old bug affecting certain BCM5708S
sobomax [Wed, 10 Mar 2010 05:19:14 +0000 (05:19 +0000)]
Provide workaround for the ages old bug affecting certain BCM5708S
chip revision often found in the blades and resulting in interfaces
not sensing carrier signal. Looking at all problem reports it
appears that it only affects some very specific silicon revision
(ASIC (0x57081021); Rev (B2)) and version of the PHY that
supports 1000baseSX-FDX media only. Therefore, narrow the scope of
workaround to combination of that revision and media type. Given
that the first report on this issue is dated back to 2007, there is
not much hope that this issue will ever be properly resolved.

Among affected systems are IBM HS21, Intel SBXD132 and HP BL460c.

PR: 118238, 122551, 140970
MFC after: 1 month

14 years agoFix the build. The ssh-pkcs11-helper directory is empty, which is
dougb [Wed, 10 Mar 2010 02:17:57 +0000 (02:17 +0000)]
Fix the build. The ssh-pkcs11-helper directory is empty, which is
causing confusion.

14 years agoRevert to r197433, the more recent change doesn't work for the common
dougb [Wed, 10 Mar 2010 01:00:37 +0000 (01:00 +0000)]
Revert to r197433, the more recent change doesn't work for the common
case, and has not been well received.

14 years agoFix "Empty input line" mdoc warning.
sobomax [Wed, 10 Mar 2010 00:47:09 +0000 (00:47 +0000)]
Fix "Empty input line" mdoc warning.

Submitted by: Alexander Best

14 years agoIntegrate OpenBSD rev 1.5 of x86emu.c.
delphij [Tue, 9 Mar 2010 22:42:24 +0000 (22:42 +0000)]
Integrate OpenBSD rev 1.5 of x86emu.c.

14 years agoAdd SIZE_MAX overflow check
ache [Tue, 9 Mar 2010 21:06:17 +0000 (21:06 +0000)]
Add SIZE_MAX overflow check

14 years agoAdd SIZE_MAX overflow check
ache [Tue, 9 Mar 2010 21:06:01 +0000 (21:06 +0000)]
Add SIZE_MAX overflow check

14 years agoAdd warnings to <regexp.h> and <sys/timeb.h>.
ed [Tue, 9 Mar 2010 21:01:12 +0000 (21:01 +0000)]
Add warnings to <regexp.h> and <sys/timeb.h>.

These header files only provide functionality that can be used in
combination with libcompat. In order to prevent people from including
them without any actual use (which happens a lot with <sys/timeb.h>),
put a warning here to make people more aware.

This means we have to lower WARNS for libcompat, which is no big deal.

14 years agoRemove unneeded include of <sys/timeb.h>.
ed [Tue, 9 Mar 2010 20:58:15 +0000 (20:58 +0000)]
Remove unneeded include of <sys/timeb.h>.

This header file should not be included by anything.

14 years agouses KOBJMETHOD_END macro to indicate the end of method table.
weongyo [Tue, 9 Mar 2010 20:07:41 +0000 (20:07 +0000)]
uses KOBJMETHOD_END macro to indicate the end of method table.

Submitted by: yongari

14 years agoo uses bus accessor macros to read values from ivar so no more values
weongyo [Tue, 9 Mar 2010 19:58:00 +0000 (19:58 +0000)]
o uses bus accessor macros to read values from ivar so no more values
  are referenced directly from ivar pointer.  It's to do like what other
  buses do. [1]
o changes exported prototypes.  It doesn't use struct siba_* structures
  anymore that instead of it it uses only device_t.
o removes duplicate code and debug messages.
o style(9)

Pointed out by: imp [1]

14 years agoBetter TwinAx transceiver detection.
np [Tue, 9 Mar 2010 19:57:44 +0000 (19:57 +0000)]
Better TwinAx transceiver detection.

Originally submitted by: <Bruno dot Bittner at isilon dot com>
(This is a rewritten, corrected version of that patch)

MFC after:    1 week

14 years agoSince the interpreter slack mode is a tunable now, enable a local hack only
jkim [Tue, 9 Mar 2010 19:39:52 +0000 (19:39 +0000)]
Since the interpreter slack mode is a tunable now, enable a local hack only
when it is set.  Note the default behaviour does not change by this change.

14 years agoo bdeficize expand_number_int() function;
sobomax [Tue, 9 Mar 2010 19:31:08 +0000 (19:31 +0000)]
o bdeficize expand_number_int() function;

o revert most of the recent changes (int -> int64_t conversion) by using
this functon for parsing all options.

14 years agoDocument a new sysctl variable and a loader tunable.
jkim [Tue, 9 Mar 2010 19:19:07 +0000 (19:19 +0000)]
Document a new sysctl variable and a loader tunable.

14 years agoUpgrade to OpenSSH 5.4p1.
des [Tue, 9 Mar 2010 19:16:43 +0000 (19:16 +0000)]
Upgrade to OpenSSH 5.4p1.

MFC after: 1 month

14 years ago- Allow users to enable dumping Debug objects without ACPI debugger.
jkim [Tue, 9 Mar 2010 19:02:02 +0000 (19:02 +0000)]
- Allow users to enable dumping Debug objects without ACPI debugger.
Setting the new sysctl MIB "debug.acpi.enable_debug_objects" to a non-zero
value enables us to print Debug object when something is written to it.
- Allow users to disable interpreter slack mode.  Setting the new tunable
"debug.acpi.interpreter_slack" to zero disables some workarounds for common
BIOS mistakes and enables strict ACPI implementations by the specification.

14 years agoNow that the workaround for the AMD 10h CPUs is in place, re-enable machine
jhb [Tue, 9 Mar 2010 15:12:53 +0000 (15:12 +0000)]
Now that the workaround for the AMD 10h CPUs is in place, re-enable machine
checks by default on amd64.

Discussed with: alc

14 years agoChange secrorsize back to int, since that's the data type expected by the
sobomax [Tue, 9 Mar 2010 10:31:03 +0000 (10:31 +0000)]
Change secrorsize back to int, since that's the data type expected by the
ioctl(DIOCGSECTORSIZE). It creates issues on some architectures.

MFC after: 1 week
Reported by: Jayachandran C.

14 years ago- Fix the highlighting for non-terminals when the last week is not
edwin [Tue, 9 Mar 2010 06:43:35 +0000 (06:43 +0000)]
- Fix the highlighting for non-terminals when the last week is not
  7 days long.
- "-m <N> <YYYY>" now prints only the month, not the whole year.

14 years agoImplement AMD's recommended workaround for Erratum 383 on Family 10h
alc [Tue, 9 Mar 2010 03:30:31 +0000 (03:30 +0000)]
Implement AMD's recommended workaround for Erratum 383 on Family 10h
processors.  With this workaround, superpage promotion can be re-enabled
under virtualization.  Moreover, machine check exceptions can safely be
enabled when FreeBSD is running natively on Family 10h processors.

Most of the credit should go to Andriy Gapon for diagnosing the error and
working with Borislav Petkov at AMD to document it.  Andriy also reviewed
and tested my patches.

Discussed with: jhb
MFC after: 3 weeks

14 years agoRemove inclusion of <i386/include/psl.h>
marcel [Tue, 9 Mar 2010 02:08:02 +0000 (02:08 +0000)]
Remove inclusion of <i386/include/psl.h>
While here move inclusion of <sys/lock.h> in a better place.

14 years agoRemove support for SYS_RES_DRQ.
marcel [Tue, 9 Mar 2010 02:05:01 +0000 (02:05 +0000)]
Remove support for SYS_RES_DRQ.

14 years agoPlace interrupt handling in a critical section and remove double
nwhitehorn [Tue, 9 Mar 2010 02:00:53 +0000 (02:00 +0000)]
Place interrupt handling in a critical section and remove double
counting in incrementing the interrupt nesting level. This fixes a number
of bugs in which the interrupt thread could be preempted by an IPI,
indefinitely delaying acknowledgement of the interrupt to the PIC, causing
interrupt starvation and hangs.

Reported by: linimon
Reviewed by: marcel, jhb
MFC after: 1 week

14 years agoOne of the advantages of enabling ECMP (a.k.a RADIX_MPATH) is to
qingli [Tue, 9 Mar 2010 01:11:45 +0000 (01:11 +0000)]
One of the advantages of enabling ECMP (a.k.a RADIX_MPATH) is to
allow for connection load balancing across interfaces. Currently
the address alias handling method is colliding with the ECMP code.
For example, when two interfaces are configured on the same prefix,
only one prefix route is installed. So connection load balancing
among the available interfaces is not possible.

The other advantage of ECMP is for failover. The issue with the
current code, is that the interface link-state is not reflected
in the route entry. For example, if there are two interfaces on
the same prefix, the cable on one interface is unplugged, new and
existing connections should switch over to the other interface.
This is not done today and packets go into a black hole.

Also, there is a small bug in the kernel where deleting ECMP routes
in the userland will always return an error even though the command
is successfully executed.

MFC after: 5 days

14 years agoRemove the check for IFF_DRV_OACTIVE right before adding a port into lagg
delphij [Tue, 9 Mar 2010 00:52:16 +0000 (00:52 +0000)]
Remove the check for IFF_DRV_OACTIVE right before adding a port into lagg
interface.  The check itself seems to be coming from OpenBSD but does not
seem to be useful for our code.

Discussed with: thomasa
MFC after: 1 month

14 years agoBWCT boards can have 128MB SDRAM.
ticso [Tue, 9 Mar 2010 00:50:58 +0000 (00:50 +0000)]
BWCT boards can have 128MB SDRAM.

14 years agofix signed warning
ticso [Tue, 9 Mar 2010 00:48:06 +0000 (00:48 +0000)]
fix signed warning

14 years agoRewrite input processing to not exit with error on the first EILSEQ found
ache [Mon, 8 Mar 2010 22:27:46 +0000 (22:27 +0000)]
Rewrite input processing to not exit with error on the first EILSEQ found
in the input data but fallback to "binary comparison" instead.

POSIX says: "The input files shall be text files", nothing more,
so the text file with illegal sequence is valid input.
BTW, GNU sort does not fails on EILSEQ too.

14 years agoMFV of tzdata2010e:
edwin [Mon, 8 Mar 2010 21:17:10 +0000 (21:17 +0000)]
MFV of tzdata2010e:

- Adjust beginning / end of DST in Bangladesh (minimal impact)
- Fiji ends DST one month earlier to last Sunday of March
- Samoa changes
- Chile extends DST until 3 April this year.

14 years ago- Set missing flag when initiating a plex rebuild with the rebuildparity
lulf [Mon, 8 Mar 2010 21:16:28 +0000 (21:16 +0000)]
- Set missing flag when initiating a plex rebuild with the rebuildparity
  command.
- Check if plex is already syncing or rebuilding before initiating a parity
  rebuild or check.

14 years agoAdd file forgotten in r204879.
kib [Mon, 8 Mar 2010 20:45:52 +0000 (20:45 +0000)]
Add file forgotten in r204879.

14 years agoTeach procstat(1) to display some information about signal disposition
kib [Mon, 8 Mar 2010 20:44:22 +0000 (20:44 +0000)]
Teach procstat(1) to display some information about signal disposition
and pending/blocked status for signals.

Reviewed by: rwatson
MFC after: 2 weeks

14 years agoChange the way shutdown is handled for log file.
fabient [Mon, 8 Mar 2010 19:58:00 +0000 (19:58 +0000)]
Change the way shutdown is handled for log file.

pmc_flush_logfile is now non-blocking and just ask the kernel
to shutdown the file. From that point, no more data is
accepted by the log thread and when the last buffer is flushed
the file is closed.

This will remove a deadlock between pmcstat asking for
flush while it cannot flush the pipe itself.

MFC after: 3 days

14 years agoEnable ACPI module build on amd64. Although we strongly recommend building
jkim [Mon, 8 Mar 2010 19:40:31 +0000 (19:40 +0000)]
Enable ACPI module build on amd64.  Although we strongly recommend building
it into kernel, there is no need to prevent it from building at all.

14 years ago1) Reimplement (differently) unlimited line length restricted in prev.
ache [Mon, 8 Mar 2010 19:40:22 +0000 (19:40 +0000)]
1) Reimplement (differently) unlimited line length restricted in prev.
commit.

2) Honor missing the very last \n (if absent) on output.

14 years agoLooks like I forgot to add half of the copyright text when we switched to
joel [Mon, 8 Mar 2010 18:51:28 +0000 (18:51 +0000)]
Looks like I forgot to add half of the copyright text when we switched to
our preferred license.

14 years agoUpdate module Makefile for ACPICA 20100304.
jkim [Mon, 8 Mar 2010 18:40:22 +0000 (18:40 +0000)]
Update module Makefile for ACPICA 20100304.

Submitted by: Paul B Mahol (onemda at gmail dot com)

14 years agoPass the correct size to memset().
jh [Mon, 8 Mar 2010 16:23:32 +0000 (16:23 +0000)]
Pass the correct size to memset().

PR: bin/128094
Submitted by: Henning Petersen
MFC after: 1 week

14 years agoEnable tmpfs unconditionally on all platforms. No one I spoke to could
nwhitehorn [Mon, 8 Mar 2010 15:01:08 +0000 (15:01 +0000)]
Enable tmpfs unconditionally on all platforms. No one I spoke to could
remember why it was x86 only, and it works just as well on at least powerpc
as on x86.

14 years agoadd back DPADD (removed by mistake in a previous commit)
luigi [Mon, 8 Mar 2010 14:43:55 +0000 (14:43 +0000)]
add back DPADD (removed by mistake in a previous commit)

14 years agocosmetic changes and C++ compatibility
luigi [Mon, 8 Mar 2010 11:27:39 +0000 (11:27 +0000)]
cosmetic changes and C++ compatibility

14 years agodon't use C++ keywords as variable names
luigi [Mon, 8 Mar 2010 11:27:08 +0000 (11:27 +0000)]
don't use C++ keywords as variable names

14 years agoForgot to remove dead files
des [Mon, 8 Mar 2010 11:24:19 +0000 (11:24 +0000)]
Forgot to remove dead files

14 years agodo not report an error unnecessarily
luigi [Mon, 8 Mar 2010 11:22:47 +0000 (11:22 +0000)]
do not report an error unnecessarily

14 years agoVendor import of OpenSSH 5.4p1
des [Mon, 8 Mar 2010 11:19:52 +0000 (11:19 +0000)]
Vendor import of OpenSSH 5.4p1

14 years agoBump up the firmware_table from 30 to 50. bwn needs more than 30, it
imp [Sun, 7 Mar 2010 22:37:35 +0000 (22:37 +0000)]
Bump up the firmware_table from 30 to 50.  bwn needs more than 30, it
seems.

14 years ago- document the -3, -A and -B properly in Synopsis.
edwin [Sun, 7 Mar 2010 21:54:28 +0000 (21:54 +0000)]
- document the -3, -A and -B properly in Synopsis.
- add highlight of current date for non-terminals.
- fix -J option.
- code cleanup.

MFC after: 2 weeks

14 years agosh: Add some testcases for ${v=w}, ${v-w}, ${v+w}.
jilles [Sun, 7 Mar 2010 18:43:29 +0000 (18:43 +0000)]
sh: Add some testcases for ${v=w}, ${v-w}, ${v+w}.

These expansions, which were already in the Bourne shell, work correctly for
the most part. The testcases are only about the parts that already work
correctly.

14 years agoAs statfs.f_flags are uint64_t the local variables should be as well.
bz [Sun, 7 Mar 2010 16:45:18 +0000 (16:45 +0000)]
As statfs.f_flags are uint64_t the local variables should be as well.
We'll start noticing this with the next flag introduced as the lower
32bit are all used.
As this is old code we might need to do a full tree sweep one day, unless
changing our strategy to use a different `API' for getting/setting flags
along with the rest of the statfs data.

While here compare to 0 explicitly [1].

Suggested by: kib [1]
Reviewed by: kib
MFC after: 5 days

14 years agoDestroy TCP UMA zones (empty or not) upon network stack teardown
bz [Sun, 7 Mar 2010 15:58:44 +0000 (15:58 +0000)]
Destroy TCP UMA zones (empty or not) upon network stack teardown
to not leak them, otherwise making UMA/vmstat unhappy with every stoped vnet.
We will still leak pages (especially for zones marked NOFREE).

Reshuffle cleanup order in tcp_destroy() to get rid of what we can
easily free first.

Sponsored by: ISPsystem
Reviewed by: rwatson
MFC after: 5 days

14 years agoNot only flush the ipfw tables when unloading ipfw or tearing
bz [Sun, 7 Mar 2010 15:37:58 +0000 (15:37 +0000)]
Not only flush the ipfw tables when unloading ipfw or tearing
down a virtual netowrk stack, but also free the Radix Node Head.

Sponsored by: ISPsystem
Reviewed by: julian
MFC after: 5 days

14 years agosh: Add various testcases for here documents.
jilles [Sun, 7 Mar 2010 15:08:42 +0000 (15:08 +0000)]
sh: Add various testcases for here documents.

They are mainly about expansions in here documents but because all the
testcases are in $() command substitution, we also test that $() command
substitution is recursively parsed (or very close to it).

14 years agoRemove old documents when they are not compressed.
antoine [Sun, 7 Mar 2010 14:55:43 +0000 (14:55 +0000)]
Remove old documents when they are not compressed.
(NO_INFOCOMPRESS, NO_DOCCOMPRESS or NO_MANCOMPRESS)

MFC after: 1 month

14 years agoLocking the tcbinfo structure should not be necessary in tcp_timer_delack(),
rwatson [Sun, 7 Mar 2010 14:23:44 +0000 (14:23 +0000)]
Locking the tcbinfo structure should not be necessary in tcp_timer_delack(),
so don't.

MFC after:      1 week
Reviewed by:    bz
Sponsored by:   Juniper Networks

14 years agoAdd comment in tcp_discardcb() talking about how we don't, but should,
rwatson [Sun, 7 Mar 2010 14:13:59 +0000 (14:13 +0000)]
Add comment in tcp_discardcb() talking about how we don't, but should,
address TCP races relating to not calling tcp_drain() on stopped callouts.

Discussed with: bz

14 years agoMake udp_set_kernel_tunneling() less forgiving when its invariants are
rwatson [Sun, 7 Mar 2010 10:47:47 +0000 (10:47 +0000)]
Make udp_set_kernel_tunneling() less forgiving when its invariants are
violated: so_pcb can never be NULL for a valid UDP socket, and it is
always SOCK_DGRAM.  Use sotoinpcb() as the rest of the UDP code does.

MFC after: 1 week
Reviewed by: bz
Sponsored by: Juniper Networks

14 years agoMake /proc/self/fd `work'.
ed [Sun, 7 Mar 2010 10:43:45 +0000 (10:43 +0000)]
Make /proc/self/fd `work'.

On Linux, /proc/<pid>/fd is comparable to fdescfs, where it allows you
to inspect the file descriptors used by each process. Glibc's ttyname()
works by performing a readlink() on these nodes, since all nodes in this
directory are symlinks.

It is a bit hard to implement this in linprocfs right now, so I am not
going to bother. Add a way to make ttyname(3) work, by adding a
/proc/<pid>/fd symlink, which points to /dev/fd only if the calling
process matches. When fdescfs is mounted, this will cause the
readlink() in ttyname() to fail, causing it to fall back on manually
finding a matching node in /dev.

Discussed on: emulation@

14 years agoLook for "compile" to decide if this is run as part of the kernel build.
obrien [Sun, 7 Mar 2010 10:08:00 +0000 (10:08 +0000)]
Look for "compile" to decide if this is run as part of the kernel build.
The assumption of "${ARCH}/compile/FOO" is much harder to change in our
build (so assume one hasn't), then assuming every kernel is rooted at "sys/".

14 years agoImprove the contents of termcap.small.
ed [Sun, 7 Mar 2010 08:54:06 +0000 (08:54 +0000)]
Improve the contents of termcap.small.

- Remove dosansi, pc and pc3. I suspect nobody ever needs these.
- Add vt100, screen and xterm-color.

This file is now probably more than sufficient in most cases, even for
common use outside single user mode, where people just use the console
driver, a graphical terminal emulator and a terminal multiplexer.

14 years agoInstall termcap.small.
ed [Sun, 7 Mar 2010 08:30:21 +0000 (08:30 +0000)]
Install termcap.small.

Right now we have a termcap.small in the tree, but we don't install it.
If we do install this file by default, it is more likely for
applications to work in single user mode.

I am not entirely happy with the contents of this file. In my opinion we
should remove the `dosansi', `pc' and `pc3' entries and replace them
with `vt100' in case someone uses a serial console.

The file does already have entries for `cons25' and `xterm', which is
used most often.

Requested by: brucec

14 years agoSwitch to our preferred license text.
joel [Sun, 7 Mar 2010 07:59:05 +0000 (07:59 +0000)]
Switch to our preferred license text.

Approved by: jedgar

14 years agoImplement the idea of parallel-only-at-start-time in a cleaner, more
dougb [Sun, 7 Mar 2010 04:26:21 +0000 (04:26 +0000)]
Implement the idea of parallel-only-at-start-time in a cleaner, more
rc.d'ish way.

Not objected to by: netchild

14 years agoRemove vestiges of old %-format which prevents build on amd64
ache [Sat, 6 Mar 2010 22:38:38 +0000 (22:38 +0000)]
Remove vestiges of old %-format which prevents build on amd64

14 years agoRemove unnecessary locking of divcbinfo lock from div_output(): this has not
rwatson [Sat, 6 Mar 2010 22:04:45 +0000 (22:04 +0000)]
Remove unnecessary locking of divcbinfo lock from div_output(): this has not
been required since FreeBSD 7.0 when the so_pcb pointer leading to inp was
guaranteed to be stable when a valid socket reference is held (as it is in
the output path).

MFC after: 1 week
Reviewed by: bz
Sponsored by: Juniper Networks

14 years agoAdd a comment to tcp_usr_accept() to indicate why it is we acquire the
rwatson [Sat, 6 Mar 2010 21:38:31 +0000 (21:38 +0000)]
Add a comment to tcp_usr_accept() to indicate why it is we acquire the
tcbinfo lock there: r175612, which re-added it, masked a race between
sonewconn(2) and accept(2) that could allow an incompletely initialized
address on a newly-created socket on a listen queue to be exposed.  Full
details can be found in that commit message.

MFC after: 1 week
Sponsored by: Juniper Networks

14 years agoIntroduce a function rn_detachhead() that will free the
bz [Sat, 6 Mar 2010 21:27:26 +0000 (21:27 +0000)]
Introduce a function rn_detachhead() that will free the
radix table root nodes.  This is only needed (and available)
in the virtualization case to free the resources when tearing
down a virtual network stack.

Sponsored by: ISPsystem
Reviewed by: julian, zec
MFC after: 5 days

14 years agoDestroy UDP UMA zones (empty or not) upon network stack teardown
bz [Sat, 6 Mar 2010 21:24:32 +0000 (21:24 +0000)]
Destroy UDP UMA zones (empty or not) upon network stack teardown
to not leak them making the VM subsystem unhappy with every stoped vnet(*).
We will still leak pages (especially as zones are marked NOFREE).

(*) This will also keep vmstat -z more usable.

Sponsored by: ISPsystem
MFC after: 5 days

14 years agoWrap use of rw_try_upgrade() on pcbinfo with macro INP_INFO_TRY_UPGRADE()
rwatson [Sat, 6 Mar 2010 21:24:11 +0000 (21:24 +0000)]
Wrap use of rw_try_upgrade() on pcbinfo with macro INP_INFO_TRY_UPGRADE()
to match other pcbinfo locking macros.

MFC after: 1 week

14 years agoRework reference counting in case we queue into the netisr,
bz [Sat, 6 Mar 2010 21:22:28 +0000 (21:22 +0000)]
Rework reference counting in case we queue into the netisr,
or overflow the netisr queue and fall back to the interface
queue so that we can garuantee that the ifnet pointer stays
valid.   Formerly we ended up with reference counts <= 0 in
case the netisr had returned ENOBUFS.  The idea is to track
any packet in the netisr queue and only change the refount
on edge operations for the fallback interface queue. This
also avoids problems in case the if_snd.ifq_len lies to us.

Also rework refount assertions to make sure they trigger if
we go below 1. Formerly a negative refence count did not
trigger the assert as the refcount variable is u_int.

Sponsored by: ISPsystem
MFC after: 5 days

14 years agoRemove racy assertion.
pjd [Sat, 6 Mar 2010 20:03:26 +0000 (20:03 +0000)]
Remove racy assertion.

Reported by: Attila Nagy <bra@fsn.hu>
Obtained from: OpenSolaris, Bug ID 6827260
MFC after: 1 week

14 years ago1) Rewrite input processing to not exit with error on the first EILSEQ found
ache [Sat, 6 Mar 2010 19:21:57 +0000 (19:21 +0000)]
1) Rewrite input processing to not exit with error on the first EILSEQ found
in the input data but fallback to "binary equal" check instead.

POSIX says: "The input file shall be a text file", nothing more,
so the text file with illegal sequence is valid input.
BTW, GNU sort does not fails on EILSEQ too.

2) Speedup input processing a bit in complex cases like skipping fields,
chars or ignore case.

3) Enforce the implied LINE_MAX limit (from POSIX definition of "text file"
and POSIX uniq(1) description).

14 years agosh: Make sure to popredir() even if a function caused an error.
jilles [Sat, 6 Mar 2010 17:31:09 +0000 (17:31 +0000)]
sh: Make sure to popredir() even if a function caused an error.

14 years agosh: Make sure to popredir() even if a special builtin caused an error.
jilles [Sat, 6 Mar 2010 17:09:22 +0000 (17:09 +0000)]
sh: Make sure to popredir() even if a special builtin caused an error.

14 years agosh: Improve the command builtin:
jilles [Sat, 6 Mar 2010 16:57:53 +0000 (16:57 +0000)]
sh: Improve the command builtin:
* avoid unnecessary fork
* allow executing builtins via command
* executing a special builtin via command removes its special properties

Obtained from: NetBSD (parts)

14 years agoSwitch to our preferred license text.
joel [Sat, 6 Mar 2010 05:57:24 +0000 (05:57 +0000)]
Switch to our preferred license text.

Approved by: imp

14 years agoCheck for device faults and for failures to set DRQ when expected, rather
jmallett [Sat, 6 Mar 2010 05:49:15 +0000 (05:49 +0000)]
Check for device faults and for failures to set DRQ when expected, rather
than spinning forever.  This fixes booting with CF ejected.

NB: I've made the driver pretty chatty about errors in case there's hardware
    that operates differently to mine, so we can easily track down any issues.

Reviewed by: imp
Sponsored by: Packet Forensics

14 years agoo) Consistently use MIPS_KSEGn_TO_PHYS instead of MIPS_{,UN}CACHED_TO_PHYS etc.
jmallett [Sat, 6 Mar 2010 05:45:49 +0000 (05:45 +0000)]
o) Consistently use MIPS_KSEGn_TO_PHYS instead of MIPS_{,UN}CACHED_TO_PHYS etc.
   Get rid of the macros that spell KSEG0 CACHED and KSEG1 UNCACHED.
o) Get rid of some nearby duplicated and unused macros.

Reviewed by: imp

14 years agoBug fixed:
fabient [Fri, 5 Mar 2010 23:08:19 +0000 (23:08 +0000)]
Bug fixed:
- no display on serial terminal in top mode.
- display alignment for continuation string.
- correct invalid value used for display limit.

MFC after: 3 days

14 years agoHarmonize Octeon ldscripts, reducing gratuitous diffs and using KERNLOADADDR in
jmallett [Fri, 5 Mar 2010 23:03:20 +0000 (23:03 +0000)]
Harmonize Octeon ldscripts, reducing gratuitous diffs and using KERNLOADADDR in
all of them to set the load address.

Set svn:keywords.

Sponsored by: Packet Forensics

14 years agoo) Simplify the implementation of bus read/write functions, and eliminate some
jmallett [Fri, 5 Mar 2010 22:48:34 +0000 (22:48 +0000)]
o) Simplify the implementation of bus read/write functions, and eliminate some
   redundant implementations.
o) Use ABI, not ISA, to determine address length.
o) Disable and restore interrupts around any operation that uses all 64 bits of
   a register.  In kernels using the O32 ABI, the upper 32 bits of those
   registers is likely to be corrupted by an interrupt.

Sponsored by: Packet Forensics

14 years agoProperly detect a type of real board that claims to have a 0.0 revision.
jmallett [Fri, 5 Mar 2010 22:46:11 +0000 (22:46 +0000)]
Properly detect a type of real board that claims to have a 0.0 revision.
This fixes at least memory detection on that board.

Sponsored by: Packet Forensics

14 years agoDo not mask off the low byte of the chipid, it makes some of the case
jmallett [Fri, 5 Mar 2010 22:44:49 +0000 (22:44 +0000)]
Do not mask off the low byte of the chipid, it makes some of the case
statements unreachable and seems to be wrong.  Fixes detection of the number
of ports available on some models.

Sponsored by: Packet Forensics

14 years agoMerge ACPICA 20100304.
jkim [Fri, 5 Mar 2010 21:39:16 +0000 (21:39 +0000)]
Merge ACPICA 20100304.

14 years agoTweak the linker spec a smidge.
imp [Fri, 5 Mar 2010 21:25:20 +0000 (21:25 +0000)]
Tweak the linker spec a smidge.
Correct a typo.

14 years agoMake little endian compiles produce little endian binaries on mips.
imp [Fri, 5 Mar 2010 21:24:41 +0000 (21:24 +0000)]
Make little endian compiles produce little endian binaries on mips.

Submitted by: neel@

14 years agoImport ACPICA 20100304.
jkim [Fri, 5 Mar 2010 19:58:45 +0000 (19:58 +0000)]
Import ACPICA 20100304.

14 years agoRemove stale path reference.
raj [Fri, 5 Mar 2010 19:51:02 +0000 (19:51 +0000)]
Remove stale path reference.

14 years agoProvide correct TCLK value for Kirkwood A1 silicon revision.
raj [Fri, 5 Mar 2010 19:45:45 +0000 (19:45 +0000)]
Provide correct TCLK value for Kirkwood A1 silicon revision.

While there improve SOC ID output accordingly.

Obtained from: Semihalf
MFC after: 1 week

14 years agoplug a memory leak on pipe's reconfiguration
luigi [Fri, 5 Mar 2010 17:53:28 +0000 (17:53 +0000)]
plug a memory leak on pipe's reconfiguration

14 years agoAdd support for o32 (when it isn't the default), n32, n64, and o64 to
imp [Fri, 5 Mar 2010 16:56:08 +0000 (16:56 +0000)]
Add support for o32 (when it isn't the default), n32, n64, and o64 to
the linker spec.  Provide the ability to have a default ABI that's
different than o32 (again, for all 4).

Submitted by: C. Jayachandran (JC) with tweaks for o64/o32 by me

14 years ago- Use errx(3) instead of err(3) when checking if snprintf(3) succeeded.
jh [Fri, 5 Mar 2010 15:23:01 +0000 (15:23 +0000)]
- Use errx(3) instead of err(3) when checking if snprintf(3) succeeded.
  snprintf(3) doesn't set errno in the tested cases.
- If the same argument reference (for example %1) was specified more than
  once, the command didn't necessarily fit to the final command buffer. Fix
  this using a dynamic sbuf buffer. Add a few regression tests for the case.

PR: bin/95079
No objections: freebsd-hackers

14 years agoRedirect stdin from /dev/null when starting a jail:
netchild [Fri, 5 Mar 2010 14:34:33 +0000 (14:34 +0000)]
Redirect stdin from /dev/null when starting a jail:
  At least in RELENG_7 this fixes some start problems for some programs
  from the ports. It is also more correct, as a jail shall not expect
  input (interactivity) from the jail-host.

Revert the current behavior of starting jails in the background and
make it optional only for the start of jails (jail_parallell_start=YES
in rc.conf):
 - The stop can not be done in the background, the system needs to wait
   until everything is stopped correctly before it can reboot or power
   down.
 - The start should not be done in parallel by default, this not only
   breaks POLA for people comming from RELENG_x, it may also break a
   dependency chain with other scripts in the jail-host, which need to
   do some stuff after the jails are up and running (e.g. hardlinking
   a mysql socket from one jail into another one).

Discussed on: freebsd-jails@

14 years agomore documentation on new dummynet features.
luigi [Fri, 5 Mar 2010 14:13:58 +0000 (14:13 +0000)]
more documentation on new dummynet features.

14 years agoUse default WARNS setting (of 6) for lib/csu.
uqs [Fri, 5 Mar 2010 13:29:05 +0000 (13:29 +0000)]
Use default WARNS setting (of 6) for lib/csu.

PR: bin/140089
Reviewed by: jmallett
Approved by: ed (co-mentor)

14 years agoProperly declare non-extern functions in crt1
uqs [Fri, 5 Mar 2010 13:28:05 +0000 (13:28 +0000)]
Properly declare non-extern functions in crt1

Also move the declarations after __progname consistently to
make the distinction clearer.

Reviewed by: jmallett
Approved by: ed (co-mentor)

14 years agoUpdate the list of the process flags. Note that the lists of pending
kib [Fri, 5 Mar 2010 13:13:35 +0000 (13:13 +0000)]
Update the list of the process flags. Note that the lists of pending
signals for process and its threads are distinct.

Reviewed by: jilles
MFC after: 2 weeks

14 years agofix a memory leak when deleting RED queues
luigi [Fri, 5 Mar 2010 12:58:19 +0000 (12:58 +0000)]
fix a memory leak when deleting RED queues