-Wold-style-definitions for i386 and amd64

(christos)

Kill of messy ifdefs, and use an ansi prototype.

(christos)

add tpm

(christos)

Fix support for negative keywords: the keyword is prefixed directly with
! without space, so that the splitting works as expected. Adjust entries
for bsdgrep.

(joerg)

Fix description of 'silent' in .MAKE.MODE
Also document 'env' in that context.

(sjg)

cleanup compilation (not TRT, but the expedient way)

(christos)

don't bother with ifdef's, just grab the biggest type you can.

(christos)

ld.elf_so is reentrant.

(joerg)

Welcome to 5.99.64 (struct exec_package changed).

(matt)

Add a hook for freeing an ep_emul_arg.  Add a wrapper routine
(exec_free_emul_arg) to call the hook and then clear the ep_emul_arg
and ep_emul_arg_free members in the exec_package.
Change users/accessors to use these routines.
Approved by releng.

(matt)

Use CPP __VA_ARGS__ for DPRINTF*

(matt)

Always allocate the kmem region.  Add UVMHIST support.  Approved by releng.

(matt)

cleanup last warnings, and remove local warnings overrides.

(christos)

Make sure perfused exit when the filesystem crashed, so that unmount
is done. Failure to do so caused deadlocks, with operation that
held a lock on the root vnode and got stuck in perfused forever.

Approved by releng.

(manu)

shutup debug.

(christos)

using 2 casts to fix an unsigned-signed comparison is a sign that this it
is done incorrectly. The RHS is guaranteed positive at this point so cast
it to unsigned to match the LHS.

(christos)

Clang compile fix from Joerg Sonneberger approved by releng.

(kardel)

Add missing <sys/rnd.h> include, ack: jdc@

(martin)

use a correct macro.
releng@ ok

(yamt)

use a correct macro.
releng@ ok

(yamt)

make the result stable.
releng@ ok

(yamt)

Only use MALLOC_DECLARE if it exists. Helps with the dirty _KERNEL
tricks fstat is using.

(joerg)

Really skip the current line only if the comment is at the start.

(joerg)

don't try to write in a const buffer.

(christos)

elf_args needs to be freed with kmem_free() since it is now allocated with
kmem_alloc(). Should fix running 32 bit elf binaries.

(christos)

PR/45764, PR/45914
Part 2:
Arrange so that the pointers that we free (ifp->if_afdata, dom->dom_ifqueues[i])
are set to NULL.
While I am here, add a continue.

(christos)

PR/45764, PR/45914
Part 1:
nd6_purge can be called after dom_ifdetach, and if_afdata[AF_INET6] is
going to be freed and point to garbage. Make sure we check for NULL, before
taking the pointer offset.
While I am here, add an M_ZERO.

(christos)

Fix missed inclusion from my previous commit.

(tls)

Add Mini2440 entry to sysinst kernel selection.

(nisimura)

Make sure to allow MEMORY_DISK as rootfs.

(nisimura)

Import dhcpcd-5.5.3

(roy)

Import dhcpcd-5.5.3 to fix a compile failure.

Status:

Vendor Tag: roy
Release Tags: dhcpcd-5-5-3

(roy)

Import dhcpcd-5.5.2 with the following changes:
* Stop checking IFF_RUNNING in RTM_IFINFO in an unknown link state
* RA expiry now tiggers off the first option if newer than the prefix
* Prefer a DHCP lease over an IPv4LL one

Status:

Vendor Tag: roy
Release Tags: dhcpcd-5-5-2

(roy)

We use avail_clusters_cnt now.

(matt)

Use "opt_sdmmc.h" for SDHC_DEBUG.

(nonaka)

Translate some new messages into german.

(phx)

Use sync(8) instead of echo(1) to avoid writing an RDB disklabel.
echo(1) is not part of the miniroot file system.

(phx)

Fixed writing of MBR disklabel. It was written by disklabel(8) -r directly
into sector 0, corrupting the MBR. Moved to sector 1.
Removed MBR_LABELSECTOR, which is no longer needed.

(phx)

call uvm_km_kmem_alloc with right arguments

fixes: kern/45913

(para)

Entropy-pool implementation move and cleanup.

1) Move core entropy-pool code and source/sink/sample management code
  to sys/kern from sys/dev.

2) Remove use of NRND as test for presence of entropy-pool code throughout
  source tree.

3) Remove use of RND_ENABLED in device drivers as microoptimization to
  avoid expensive operations on disabled entropy sources; make the
  rnd_add calls do this directly so all callers benefit.

4) Fix bug in recent rnd_add_data()/rnd_add_uint32() changes that might
  have lead to slight entropy overestimation for some sources.

5) Add new source types for environmental sensors, power sensors, VM
  system events, and skew between clocks, with a sample implementation
  for each.

ok releng to go in before the branch due to the difficulty of later
pullup (widespread #ifdef removal and moved files).  Tested with release
builds on amd64 and evbarm and live testing on amd64.

(tls)

use FOREACH_SAFE.

(christos)

Compiler warning on a for loop that does not terminate correctly

(darrenr)

- bringing kmeminit_nkmempages back and revert pmaps that called this early
- use nkmempages to scale the kmem_arena
- reducing diff to pre kmem/vmem change
  (NKMEMPAGES_MAX_DEFAULT will need adjusting on some archs)

(para)

Handle another case the new gcc_assert(0) caught.

(matt)

For fast lwp_getprivate, just do the chmk $SYS_lwp_getprivate inline.

(matt)

In the syscall vector, deal with SYS_lwp_getprivate quickly.  XXX what about
non-NetBSD syscall tables.

(matt)

fill in timestamps in outgoing data buffers

(drochner)

Always include <m68k/cpu.h> (i.e. outside #if defined(_KERNEL))
for crash(8) as amiga and sun3.  There are _KERNEL protections in it.

Briefly tested by "build.sh -m news68k -U build".
Further possible botch will be fixed later.

(tsutsui)

Don't enable DB_MACHINE_COMMANDS on crash(8) (i.e. !defined(_KERNEL) case).
No easy way to switch sun3 and sun3x at run time for now.

Briefly tested by "build.sh -U -m sun3 build".

(tsutsui)

Build INSTALL_C700 ramdisk kernel and install into releasedir.

(tsutsui)

Select a proper kernel set (GENERIC or C700) per running INSTALL kernel name.

(tsutsui)

Add INSTALL kernel config for SL-C700.

(tsutsui)

Try to shrink INSTALL kernel a bit.
Loading INSTALL kernel on SL-C1000 running Zaurus Linux still sometimes
fails (due to its size), but at least it works on maintenance kernel.

(tsutsui)

On a kernel segfault, print the registers from the trapframe.

(matt)

Fix the initial sizing of the system page table / address space.

(matt)

Deal with instruction with no arguments.

(matt)

Fix a bug in movmemsi in which couldn't handle sym(reg) expressions properly.

(matt)

delete virtual screens on detach - this allows to hot-unplug
a udl@usb monitor without crash

(drochner)

tell the compiler that the i387 runs in double-rounding mode, so it
doesn't need to issue memory store-read sequences to kill excess
precision. makes code smaller and faster, depending on optimization
flags
(as tests on Linux have shown, the compiler doesn't even succeed in
avoiding excess precision)

(drochner)

spl_intr() is not defined anymore so remove from the header file

(reinoud)

Regen

(reinoud)

Add cdefs.h to usermode's genheaders.sh

(reinoud)

Welcome to 5.1.2!

(snj)

add a bunch of register definitions
not used yet but I better commit them now than have he tell me that I forgot
them when pm2fb.c starts using them

(macallan)

When creating a very small filesystem, use well-known small segment,
block and fragment sizes by default instead of the ordinary 1M/8k/1k
default for larger filesystems.

(perseant)

If invoked with -D and creating an image file, ensure that the image
file is filled to the given size.  This prevents false failures
with "lfs: ifile read: 22" in ATF tests.

(perseant)

Shut down the cleaner earlier if the filesystem is unmounted.

(perseant)

Make this compile on vax (uninitialized use warning).

(matt)

Make this compile on vax (uninitialized use warning).

(matt)

Clean up display of initial hardware provided limits.  No functional change.

(pgoyette)

Disable PTRACE_FORK for vforked() children, because the parent is waiting
and will not receive the SIGTRAP in time.

(christos)

Fix XLP case where we weren't copying mem_clusters to avail_clusters.

(matt)

allocate uareas and buffers from kernel_map again
add code to drain pools if kmem_arena runs out of space

(para)

Add more FLAGS

(matt)

Add XLP specific CGM mode for SDHC_CLOCK_CTL.

(matt)

cleanup printf fmt differences between i386/amd64

(kardel)

Add some more FreeScale ESDHC defintions.

(matt)

Use "opt_sdmmc.h" for SDMMC_DEBUG, etc.

(matt)

Add ext2/msdos support, allow mount from udf, add flashctl.

(matt)

Put ld0/ld1 on the ramdisk (for sdmmc).

(matt)

Build the INSTALL kernels for the P2020DS and P2020RDB

(matt)

Use kmem instead of malloc.

(matt)

Allow ofwboot to read files from FAT(dosfs).

(matt)

Add tmpfs

(matt)

Make sure the walnut system_tail_extra is very last thing.

(matt)

use WARNS=0 until const issue is resolved

(kardel)

initialize min_dist

(kardel)

build libopts with WARNS=0 for now

(kardel)

back to WARNS=4 (constify and signedness corrections)

(kardel)

amd64/Xen doesn't require special treatment for pmap_is_curpmap(),
since cpu_load_pmap() ensures that the linear map is in place for the
kernel. This emulates normal shared kernel mappings, except for the
recursive mapping of the PDP_BASE, which will point to the per-cpu
pdir, which will be a copy of the pmap_kernel()->pm_pdir; instead of
the user pmap->pm_pdir.

(cherry)

Set list updates for libquota bump.

(dholland)

Bump libquota.so major version (to 1)

(dholland)

libquota no longer requires libprop. Also remove some other no-longer-used
.PATH'd-in bits from repquota.

(dholland)

Remove old (5.99.48 - 5.99.62) libquota interface.

(dholland)

Simplify elaborate calls to quota_check_limit().

(dholland)

PR kern/45907
#ifdef USE_INET6 guards are missing around IPv6 code

(darrenr)

Set HDAUDIO_PCI_TCSEL to TC0 like Linux does.
Fixes stalled audio on Lenovo X60 with coreboot firmware.

Additionally, move snooping adjustment to be executed on resume too,
as Linux does.

(jakllsch)

old style def

(christos)

ansify

(christos)

Postfix 2.8.8 and 2.9.0 has released.

(taca)

Remove trailing whitespace. Bump date for previous.

(wiz)

Document when recv(2) returns 0

(manu)

bin/45906: regenerate ntpd/ntp_keyword.h to accommodate our changes

(kardel)

Don't refer to ipf_log_soft_destroy(), ipf_log_soft_create(),
ipf_log_soft_init(), and ipf_log_soft_fini() unless IPFILTER_LOG is
defined, since ipf_log.c won't be built unless that flag is defined,
ref. sys/netinet/files.ipfilter.

(he)

Bump SYMTAB_SPACE.

(he)

Don't forward-declare ipf_dolog unless IPFILTER_LOG is defined,
since its implementation is under that ifdef.

(he)

Use kmem instead of malloc.  Remove unneeded <sys/malloc.h> includes.

(matt)

Need runtime checks for mach mmu command.

(matt)

zlib-1.2.6 out.

(wiz)

Restore missing word, remove trailing whitespace.

(wiz)

Bump date for previous.

(wiz)

document ntp 4.2.6p5 import

(kardel)

Document added timecounter support to the CSI Hurdler Centronics
Parallel Interface cpi(4) driver.

(hauke)

add libotps for autoopts headers to CPPFLAGS

(kardel)

NTP 4.2.6p5 (Harlan Stenn <stenn@ntp.org>, 2011/12/24)

Focus: Bug fixes

Severity: Medium

This is a recommended upgrade.

This release updates sys_rootdisp and sys_jitter calculations to match the
RFC specification, fixes a potential IPv6 address matching error for the
"nic" and "interface" configuration directives, suppresses the creation of
extraneous ephemeral associations for certain broadcastclient and
multicastclient configurations, cleans up some ntpq display issues, and
includes improvements to orphan mode, minor bugs fixes and code clean-ups.

New features / changes in this release:

ntpd

* Updated "nic" and "interface" IPv6 address handling to prevent
  mismatches with localhost [::1] and wildcard [::] which resulted from
  using the address/prefix format (e.g. fe80::/64)
* Fix orphan mode stratum incorrectly counting to infinity
* Orphan parent selection metric updated to includes missing ntohl()
* Non-printable stratum 16 refid no longer sent to ntp
* Duplicate ephemeral associations suppressed for broadcastclient and
  multicastclient without broadcastdelay
* Exclude undetermined sys_refid from use in loopback TEST12
* Exclude MODE_SERVER responses from KoD rate limiting
* Include root delay in clock_update() sys_rootdisp calculations
* get_systime() updated to exclude sys_residual offset (which only
  affected bits "below" sys_tick, the precision threshold)
* sys.peer jitter weighting corrected in sys_jitter calculation

ntpq

* -n option extended to include the billboard "server" column
* IPv6 addresses in the local column truncated to prevent overruns

---
NTP 4.2.6p4 (Harlan Stenn <stenn@ntp.org>, 2011/09/22)

Focus: Bug fixes and portability improvements

Severity: Medium

This is a recommended upgrade.

This release includes build infrastructure updates, code
clean-ups, minor bug fixes, fixes for a number of minor
ref-clock issues, and documentation revisions.

Portability improvements affect AIX, HP-UX, Linux, OS X and 64-bit time_t.

New features / changes in this release:

Build system

* Fix checking for struct rtattr
* Update config.guess and config.sub for AIX
* Upgrade required version of autogen and libopts for building
  from our source code repository

ntpd

* Back-ported several fixes for Coverity warnings from ntp-dev
* Fix a rare boundary condition in UNLINK_EXPR_SLIST()
* Allow "logconfig =allall" configuration directive
* Bind tentative IPv6 addresses on Linux
* Correct WWVB/Spectracom driver to timestamp CR instead of LF
* Improved tally bit handling to prevent incorrect ntpq peer status reports
* Exclude the Undisciplined Local Clock and ACTS drivers from the initial
  candidate list unless they are designated a "prefer peer"
* Prevent the consideration of Undisciplined Local Clock or ACTS drivers for
  selection during the 'tos orphanwait' period
* Prefer an Orphan Mode Parent over the Undisciplined Local Clock or ACTS
  drivers
* Improved support of the Parse Refclock trusttime flag in Meinberg mode
* Back-port utility routines from ntp-dev: mprintf(), emalloc_zero()
* Added the NTPD_TICKADJ_PPM environment variable for specifying baseline
  clock slew on Microsoft Windows
* Code cleanup in libntpq

ntpdc

* Fix timerstats reporting

ntpdate

* Reduce time required to set clock
* Allow a timeout greater than 2 seconds

sntp

* Backward incompatible command-line option change:
  -l/--filelog changed -l/--logfile (to be consistent with ntpd)

Documentation

* Update html2man. Fix some tags in the .html files
* Distribute ntp-wait.html

---
NTP 4.2.6p3 (Harlan Stenn <stenn@ntp.org>, 2011/01/03)

Focus: Bug fixes and portability improvements

Severity: Medium

This is a recommended upgrade.

This release includes build infrastructure updates, code
clean-ups, minor bug fixes, fixes for a number of minor
ref-clock issues, and documentation revisions.

Portability improvements in this release affect AIX, Atari FreeMiNT,
FreeBSD4, Linux and Microsoft Windows.

New features / changes in this release:

Build system
* Use lsb_release to get information about Linux distributions.
* 'test' is in /usr/bin (instead of /bin) on some systems.
* Basic sanity checks for the ChangeLog file.
* Source certain build files with ./filename for systems without . in PATH.
* IRIX portability fix.
* Use a single copy of the "libopts" code.
* autogen/libopts upgrade.
* configure.ac m4 quoting cleanup.

ntpd
* Do not bind to IN6_IFF_ANYCAST addresses.
* Log the reason for exiting under Windows.
* Multicast fixes for Windows.
* Interpolation fixes for Windows.
* IPv4 and IPv6 Multicast fixes.
* Manycast solicitation fixes and general repairs.
* JJY refclock cleanup.
* NMEA refclock improvements.
* Oncore debug message cleanup.
* Palisade refclock now builds under Linux.
* Give RAWDCF more baud rates.
* Support Truetime Satellite clocks under Windows.
* Support Arbiter 1093C Satellite clocks under Windows.
* Make sure that the "filegen" configuration command defaults to "enable".
* Range-check the status codes (plus other cleanup) in the RIPE-NCC driver.
* Prohibit 'includefile' directive in remote configuration command.
* Fix 'nic' interface bindings.
* Fix the way we link with openssl if openssl is installed in the base
  system.

ntp-keygen
* Fix -V coredump.
* OpenSSL version display cleanup.

ntpdc
* Many counters should be treated as unsigned.

ntpdate
* Do not ignore replies with equal receive and transmit timestamps.

ntpq
* libntpq warning cleanup.

ntpsnmpd
* Correct SNMP type for "precision" and "resolution".
* Update the MIB from the draft version to RFC-5907.

sntp
* Display timezone offset when showing time for sntp in the local
  timezone.
* Pay proper attention to RATE KoD packets.
* Fix a miscalculation of the offset.
* Properly parse empty lines in the key file.
* Logging cleanup.
* Use tv_usec correctly in set_time().
* Documentation cleanup.

---
NTP 4.2.6p2 (Harlan Stenn <stenn@ntp.org>, 2010/07/08)

Focus: Bug fixes and portability improvements

Severity: Medium

This is a recommended upgrade.

This release includes build infrastructure updates, code
clean-ups, minor bug fixes, fixes for a number of minor
ref-clock issues, improved KOD handling, OpenSSL related
updates and documentation revisions.

Portability improvements in this release affect Irix, Linux,
Mac OS, Microsoft Windows, OpenBSD and QNX6

New features / changes in this release:

ntpd
* Range syntax for the trustedkey configuration directive
* Unified IPv4 and IPv6 restrict lists

ntpdate
* Rate limiting and KOD handling

ntpsnmpd
* default connection to net-snmpd via a unix-domain socket
* command-line 'socket name' option

ntpq / ntpdc
* support for the "passwd ..." syntax
* key-type specific password prompts

sntp
* MD5 authentication of an ntpd
* Broadcast and crypto
* OpenSSL support

---
NTP 4.2.6p1 (Harlan Stenn <stenn@ntp.org>, 2010/04/09)

Focus: Bug fixes, portability fixes, and documentation improvements

Severity: Medium

This is a recommended upgrade.

(kardel)

merge import ntp-4-2-6p5

(kardel)

merge import ntp-4-2-6p5

(kardel)

Fix stupid bug in cursor_getn - it's supposed to return the number of
values retrieved, but it was returning 0 on success. Fortunately nothing
was using it yet.

(dholland)

Add missing RCSID().

(dholland)

Note quota changes in the form of the addition of libquota.

(dholland)

5.99.63: new quotactl syscall, bump required for rump and due to vfs-level
adjustments.

(dholland)

Be consistent about whether idtype and objtype codes are signed or
unsigned. They are signed. (While unsigned might have been a better
choice, it doesn't really matter and the majority of preexisting uses
were signed. And consistency is good.)

(dholland)

Improve the names of some members of struct quotactl_args. These are
effectively function parameter names, but since they need to be
described with the same names in the man page the choices do matter.
Some.

(dholland)

Regen syscalls with proper id info.

(dholland)

Update compat_netbsd32 for new quotactl.

(dholland)

Split out a do_sys_quotactl for compat_netbsd32.

(dholland)

Change the syscall API for quotas over to the new non-proplib one.

  - struct vfs_quotactl_args -> struct quotactl_args
  - add sys/stdint.h to sys/quotactl.h for clean userland build
  - install sys/quotactl.h in /usr/include
  - update set lists for same
  - add new marshalling code in libquota
  - add new unmarshalling code in vfs_syscalls.c
  - discard proplib interpreter code in vfs_quotactl.c
  - add dispatching code for the 14 quotactl ops in vfs_quotactl.c
  - mark the proplib quotactl syscall obsolete
  - add a new syscall number for the new quotactl syscall
  - change the name of the syscall to __quotactl()
  - remove the decl of the old quotactl from quota/quotaprop.h
  - add a decl of the new quotactl to sys/quotactl.h
  - update the libc build
  - update ktruss
  - remove proplib marshalling code from libquota
  - update copy of syscall table in gdb ppc sources
  - hack rumphijack to accomodate new quotactl name (as I recall,
    pooka wanted such a name change to simplify something, but I
    don't really see what/how)

This change appears to require a kernel version bump for rumpish
reasons.

(dholland)

Enable XBSEN and HIGHBAT for OEA 7455 and related CPUs.
The BAT entries now have a resolution of 8MB.  (Adjacent entries are merged
up to a total of 2GB per entry).

(matt)

Add QUOTACTL_IDTYPESTAT and QUOTACTL_OBJTYPESTAT for retrieving info
about idtypes and objtypes. This avoids compiling in the names of
the id and object types.

I overlooked this last week because the proplib syscall interface has
no way to convey this information.

Renumber the operation codes again (since we still can) to insert
the new operations into the list in a semantically sensible place.

Requires kernel version bump.

(dholland)

Reimplement repquota -x to print in tabular form instead of XML.

(dholland)

Fix problems in cursor iteration that came to light when iterating one
value at a time, instead of in bulk. Yeah, repquota should do bulk get,
but it doesn't yet.

(dholland)

Remove unwanted decl of the old quotactl syscall.

(dholland)

add MINI2440_INSTALL image to the distribution.

(nisimura)

When using socket loaning, make sure the KVA used for the loan has the same
color as the UVA being loaned.

(matt)

Use right UVM_xxx_COLORMATCH flag (even both use the same value).

(matt)