Cleanup intr_info initialization.  Make it per SVR.

(matt)

Fix CLKF_INTR
Add debug evcnt.

(matt)

Add little endian bus_space_tags.
Note highest memory in cpu_softc.

(matt)

remove some unnecessary pointer typecasts
(one was wrong on BE systems, but was harmless here because the
result is effectively unused)

(drochner)

Add MPC8533 SVR.  Fix DBCR0 IAC bits

(matt)

Add USB_SNOOP1/2 related definitions

(matt)

Fix signed / unsigned issues. Refactor basename usage to use a local
copy and do it only once, not for each pattern. Remove late inline.

(joerg)

test rumpclient_vfork()

(pooka)

Support vfork.  Add rumpclient wrapper for daemon(3).

(pooka)

Add flags -v, -x, and -z for verbose, debug, and silent shutdowns,
respectively, by passing flags AB_VERBOSE, AB_DEBUG, and/or AB_SILENT to
reboot(2).

(dyoung)

Install "obsolete" files only when necessary.

(uebayasi)

Generate MAKEDEV only when really necessary.  Write dependency correctly.

(uebayasi)

no need to create a hardlinked name for the helper, just fake argv[]

(pooka)

To install 0-sized files, check if the destination already has one.
This avoids unnecessary updates (== timestamp changes in METALOG).

(uebayasi)

Don't use hijack here since exec is now easy in pure rump clients.

(pooka)

Push the fiddly tasks for exec and fork from rumphijack to rumpclient.
This makes it possible easily execute those operations also from
non-hijacked rump clients (plus fixes one memory leak in an error
branch).

(pooka)

Make the WMI modules depend on the parent mapper.

(jruoho)

Fix header, format and total count of system interrupt counters.

(nakayama)

Miscellaneous improvements from Garrett Cooper:
- Add to variables only once, instead of from within a loop.
- Use :tl instead of :M to match against strings for readability.
- Use CLEANFILES instead of a custom clean target.

Full release built successfully after this change.

(jmmv)

As suggested by Sverre Froyen in a follow-up to PR # 41698, change the
sensor names to real words (e.g. "temperature 1" instead of "TEMP_1").
No functional change; only consistent envstat(8) output with other sensors.

(jruoho)

Add (commented) aps(4) from i386 GENERIC.

(jruoho)

Change the preprocessor defines around the module(9)'s modcmd() calls such
that the modules show in modstat(8) even though these are "builtin".

(jruoho)

Use ioconf for the module declarations.

(jruoho)

Convert the homegrown module declarations to use ioconf.

(jruoho)

Convert the ioconf files to lower case for consistency. No functional change.

(jruoho)

Convert the ioconf files to lower case for consistency. No functional change.

(jruoho)

Add a skeleton driver for Asus EeePC acpiwmi(4) mappings.

(jruoho)

Comment.

(uebayasi)

RCSMETALOG: treat initial clean_METALOG nicely.

(uebayasi)

Add a hook to version METALOG using rcs(1).

(uebayasi)

- merge conflicts
- Fix UNUSED macro to not have "NULL EFFECT"
- Add /*CONSTCOND*/ to while (0) loops
- Change do while (1) loops to for (;;)
- remove stray continue from do while (0) loop.
- remove "" in comments that confuse lint
- fix strict aliases
- fix non ansi prototypes

(christos)

mention new bind

(christos)

bump for bind-9.8.0rc1

(christos)

Build certain file-systems and options(7) as module(7). 32 and 64 bits
support are still builtin, as well as FFS, NFS, TMPFS, and most COMPAT
code. Saves approx. 750kiB.

Reflect this in INSTALL kernel, where we have to support more file systems
statically.

See http://mail-index.netbsd.org/port-amd64/2011/02/13/msg001318.html

(jym)

Property name change:
  SASLC_PROP_SERVICENAME ("SERVICENAME")
to
  SASLC_PROP_SERVNAME ("SERVNAME")

Hopefully this will avoid confusion with SASLC_PROP_SERVICE ("SERVICE").
SERVNAME is also closer to the name used in the RFC2831 ("serv-name").
(Discussed with christos@.)

Change the hash parameters to keep that collision-less after the above
name change.

While here, go back to using .Sh in the manpage for unknown section
headers as the PostScript output from .Ss is slightly different.
(Discussed with wiz@.)

(christos)

Replace usr.bin/grep with the BSD grep implementation from FreeBSD.

(joerg)

Move default definition of "PKGROOT" inside the "HAVE_CYRUS_SASL" block
as it shouldn't be used anywhere else.

(tron)

Reset the firmware waking vector on resume. Adapted from FreeBSD and Linux.
No functional change intended.

(jruoho)

from ftp.isc.org

Status:

Vendor Tag: ISC
Release Tags: bind9-8-0rc1

(christos)

implement pmap_mmap_flags() and teach PowerPC's bus_space_mmap() to actually
use BUS_SPACE_MAP_PREFETCHABLE which, now that /dev/pci* knows how to use it,
helps improve X performance

(macallan)

From Anon Ymous:
1) Fix a memory leak in cipher_context_create().
2) Fix a goof in the construction of the digest-uri.
3) Allow SASLC_PROP_SERVICENAME to be a hostname qualified comma
delimited list of service names to select from and update the manpage
to reflect this.
4) Make libsaslc.3 pass mdoclint(1).

(christos)

Fix wrong assertion logic.

(jruoho)

Fix a BE64 bug not fetching argc correctly

(matt)

implement COMPAT_NETBSD32 quotactl syscalls (both new and COMPAT_50)

(bouyer)

Fix and add comments.

(jruoho)

fix snafu

(pooka)

Implement COMPAT_50 quotactl(2)

(bouyer)

Fix non-fortify build.

(christos)

From Anon Ymous:

1) libsaslc is an SASL client only.
2) dovecot is an SASL server only.
3) cyrus-sasl is both a client and a server.
4) postfix allows us to have multiple SASL servers and clients.
5) The SASL server to use at runtime is determined by the setting of
  "smtpd_sasl_type" in main.cf (note that is smtpd_ not smtp_).  If
  that is not set, then it defaults to the value of
  DEF_SERVER_SASL_TYPE at build time, which if not set, defaults to
  "cyrus".  See postfix/dist/src/global/mail_params.h.
6) The SASL client to use at runtime is determined by the setting of
  "smtp_sasl_type" in main.cf.  If that is not set, then it defaults
  to the value of DEF_CLIENT_SASL_TYPE at build time, which if not
  set, defaults to "cyrus".  See postfix/dist/src/global/mail_params.h.
7) If MKCRYPTO is "no", libsaslc will not link as it requires the
  crypto libraries, so libsaslc cannot be enabled (as it was before)
  without crypto.
8) I have made the definition of DEF_CLIENT_SASL_TYPE conditional on
  MKCRYPTO due to (7).  Without crypto it will default to cyrus.
9) HAVE_CYRUS_SASL is _never_ defined during a normal build and _never_
  should be!  It is there for the convenience of users who wish to
  install cyrus-sasl and rebuild postfix with it.  It is also very
  useful for testing if it is suspected that something might be wrong
  with libsaslc.  PLEASE DO NOT REMOVE IT!

(christos)

Cleanup server socket before sending back "ok" for halt.  Fixes
some occasionally-seen race conditions in tests which do server
restarts (e.g. the raidframe tests).

(pooka)

remove unnecessary debug printf

(pooka)

test FD_CLOEXEC

(pooka)

Support FD_CLOEXEC in rump kernels.

(pooka)

another day, another tp

(pooka)

add some exec() tests

(pooka)

map the framebuffer and agp with BUS_SPACE_MAP_PREFETCHABLE

(jmcneill)

dup() is now implemented using fcntl()

(pooka)

Properly implement fcntl commands: F_DUPFD, F_CLOSEM, F_MAXFD

(pooka)

retire azalia from GENERIC

(jmcneill)

augment comment

(pooka)

give memfault_sun*() some entry points that both gdb and ddb will find.

(mrg)

Fix partition ids in the previous disklabel(8)'s COMPAT_386BSD_MBRPART entry.

(tsutsui)

Note hp700 HIL support.

(tsutsui)

- make local functions static
- use aprint_normal(9) and variants
- misc KNF

(tsutsui)

- move nmi_sun4m intr counting into nmi_soft/nmi_hard.
- request the appropriate stack space for nmi_sun4m, in particular,
  make sure we have space for %g2...%g5.  now entering ddb via eg,
  serial break no longer causes cpu1 to fault.

(mrg)

support HANDSHAKE_EXEC

(pooka)

- make local functions static
- remove parentheses from return
- use aprint_error(9)

(tsutsui)

Add an "exec" callback for the proxy code.  The client can now
notify the rump kernel of an exec having taken place.

(pooka)

Add MI HIL (keyboard/mouse) support from OpenBSD.
Tested on 735/125 and ok'ed by skrll@.

(tsutsui)

update cpuid bits

(cegger)

sprinkle some kpreempt_{dis,en}able() in various strategic points
we will need when we get to actually enabling kernel preemption.

(mrg)

remove unused CPUFLG_READY and CPU_NOTREADY() macro.

(mrg)

add AGP v3 support to agp_generic_enable from FreeBSD

(jmcneill)

print AGP mode (v3/v2compat/v2) with aprint_debug instead of aprint_normal

(jmcneill)

Bug fix: in a game with 26 planes, the last one to be allocated wouldn't
be allocated if it was the only eligible one.
From Jonathan David Amery via Debian Bug report 214626.

(is)

don't muck with the LVDS_BLON bit

(macallan)

port-i386/44563: boot loader prompt doesn't read keyboard input on a MacBook1,1

check for keystroke (int 16h, AH=01h) before getting keystroke (int 16h,
AH=00h), fixes bootloader keyboard input with Apple firmware, from Taylor R
Campbell

(jmcneill)

missing lvm keyword.

(njoly)

MSFT assigned the copyright to these files to TNF, so fix license.
Also, make rcsid placement match man.template.

(pooka)

fix nic name in diagnostic.  from sandrof

(pooka)

When reading quota1 entries, initialize default quota to unlimited,
for POLA when using -x.
When displaying quotas, do not skip users which have a 0 usage but are
overquota.
When -v is used display all existing quota entries, including those with a
0 usage.

(bouyer)

With quota2 semantic, we may be overquota even if usage is 0.
Print quota in default output in this case.

(bouyer)

The rpc.rquotad protocol uses the legacy definitions and integer types.
Convert from/to dqblk using the quota1_subr.c routines to have the
necessery convertions done.

(bouyer)

explicitly set BLOCKSIZE so that we don't get fooled again

(pooka)

-debugging crud

(pooka)

Fix fallout from ticket #1524 (patch applied twice to new files),
pointed out by veego@

(bouyer)

Fix small negligence mistake.

(jruoho)

one more botched statistics counter (could increment semi-random locations)

(drochner)

ACPICA 20110211 is out.

(Mostly bug fixes; for once, no big API changes.)

(jruoho)

AISE, AYF, BAK, BOC, BTFT, CPC, DIAFYO, EDS, FITB, FYEO, GLWT, IAE, ICUR,
IDTS, IMBO, INPO, IUD, KMA, KWIM, MAD, NFA, NIMBY, NUNP, ONNA, ONNTA,
OOSOOM, PFI, PPP, ROI, SAR, SITD, TCB, TIC, TINWIS, TM, TMK, TTM, TWAIN,
WAFWOT, WDYMBT, WELC, WWJD, YAOTM, YWSYLS.

(jruoho)

Add -D_KMEMUSER to CPPFLAGS

(matt)

Add NetLogic Fast Messaging instructions

(matt)

Add libsaslc entries

(matt)

remove a decidedly illogical statement

(pooka)

fix output bytecount statcounter

(drochner)

Suppress a superfluous whitespace after "time=..." in the output.

(uebayasi)

if DIAGNOSTIC, check the size of the transaction in wapbl_end().
Hopefully this will point us to the place which generaed the large
transaction, before an asynchronous panic() in wabl_end()

(bouyer)

Update quota in the same WAPBL transaction as we write mode=0 back to inode.

(bouyer)

slap this with summa dat +files

(pooka)

Test that hijacked ssh/sshd work.

Copypastes jmmv's sshd magic from fs/psshfs.
(dunno if it's worth sharing the code, or even what the
best practice for doing so would be)

(pooka)

A bunch of changes which essentially make sshd work with a hijacked
rump tcp/ip stack:

* sshd likes to fork and then re-exec itself
  ==> trap execve() and augment the env with the current parameters
      essential to a rump kernel (kernel communication fd, information
      about dup2'd file descriptors)

* sshd likes to play lots of games with pipes, socketpairs and dup{,2}()
  ==> make sure we do not close essential rump client descriptors:
      dup() them to a safe place, except for F_CLOSEM where we
      simply leave them alone.  also, partially solved by the above,
      make sure the process's set of rump kernel descriptors persists
      over exec()

* sshd likes to chdir() before exec
  ==> for unix-style rump_sp(7) sockets save the full path on the
      initial exec and use it afterwards.  thread the path through
      the environment in execve()

(pooka)

ibmhawk_set: No need to set states ENVSYS_SWARNxxx, sysmon will do it.

(hannken)

change locking order, to make sure the cpu is at splsoftnet()
before the softnet_lock (adaptive) mutex is acquired, from
Wolfgang Stukenbrock, should fix a recursive lock panic

(drochner)

- remove the mangled lines of the previous commit.
- withdraw DS407 description for now as pieces of info are conflicting
  and remain uncertain.

(nisimura)

fix the loop sentinel.

(christos)

add Thecus N3200 to the research list.  Along with N1200 they use
MPC8347 single PCI SoC.

(nisimura)

get rid of unused variable/assignments

(kefren)

Provide a macro for pmap_mmap_flags() since we define PMAP_EXCLUDE_DECLS
here.

(he)

call hardclock() the same way on cpu0 in MP and !MP cases.

(mrg)

use schedintr() (not schedintr_4m()) on MP or single CPU configurations.

(mrg)

Fix a typo, add a serial comma.

(wiz)

add EPIC definition for Netronics NH230/NH231 and the brothers.

(nisimura)

Initial implementation of ibmhawk(4) driver for sensors behind the IBM Hawk
on-board Integrated Systems Management Processor found on some eServers.

Tested on an IBM eServer x335.

(hannken)

- add preliminary Netronics NH230 and NH231 supports.
- assign PCI configuration register 0x3c with IDSEL value by PCI framework,
  eliminating most of pcifixup() logic in simple NASes.

(nisimura)

add a "mach xcall" command to dump info about each cpus xpmsg{} values.

while here, move the prototype for cpu_debug_dump() to cpuvar.h and
avoid the copy in db_interface.c.

(mrg)

A number greater than or equal to 1.0 is a positive number.

(enami)

Put space after comma and around binary operator.

(enami)

Don't forget to set errno. Pointed out by yamt.

(christos)

Mark unused args.

(sjg)

When making a target that involves creating itermediate dirs,
realpath() will fail until after those dirs exist.
Try to ensure the same meta file name is used regardless.

(sjg)

Add a test checking that an unlinked file cleaned by the log replay keeps
quotas up to date.

(bouyer)

Added a note that only the Rev.B is supported for D-Link DSM-G600.

(phx)

uvideo works, so include it

(phx)

Install "etc/gpio.conf" if it is missing.

(tron)

Add "npf" to the list of startup scripts that get checked.

(tron)

Remove a couple '#ifdef linux's to let the driver use optimized memcpy
functions on NetBSD for video copy. Gives nearly 2x performance on my
test system (previously "kernel" memcpy was always used):

CHROME(0): Benchmarking video copy.  Less time is better.
CHROME(0): Timed  libc YUV420 copy... 3000272. Throughput: 315.6 MiB/s.
CHROME(0): Timed kernel YUV420 copy... 3024946. Throughput: 313.0 MiB/s.
CHROME(0): Timed    SSE YUV420 copy... 1591352. Throughput: 595.0 MiB/s.
CHROME(0): Timed    MMX YUV420 copy... 2566279. Throughput: 369.0 MiB/s.
CHROME(0): Ditching 3DNow! YUV420 copy. Not supported by CPU.
CHROME(0): Timed  MMX2 YUV420 copy... 1696662. Throughput: 558.1 MiB/s.
CHROME(0): Using SSE YUV42X copy for video.

(jmcneill)

switch back to polling rirb, this will have to be revisited

(jmcneill)

don't pass a NULL ci to xc_unicast

(jmcneill)

Notify that the file system has to be mounted first before modules can
be loaded from it.

ok pgoyette@

(jym)

Turn on detachment for ixpide(4).

(jakllsch)

Explain "Cannot dlopen non-loadable /usr/lib/libpthread.so.1".
Bump date.

(wiz)

Add options PXA2X0_LCD_WRITETHROUGH that improves console speed.
Tested on SL-C1000.

(tsutsui)

Typo.

(uebayasi)

Don't enable the powerfail interrupt on Netra X1 since it may hang.

Discussed on port-sparc64.

(nakayama)

The ServerWorks CSB5 smbus controller often times out and hangs until reboot.

Increase the delay between setting the command register and reading the
status register.  Reset the smbus controller if it has timed out.

Tested on an IBM eServer x335.

(hannken)

Determine BAR type with pci_mapreg_probe() instead of pci_mapreg_type(),
to skip unimplemented BARs.
Do not probe non-BAR registers (like subsystem id). This fixes potential
panics in pci_mem_find().

(phx)

I'm now maintaining aund separately from NetBSD, so it's no longer
appropriate for it to live in othersrc.

(bjh21)

fix typo, from Randolf Richardson in PR 44559

(dholland)

use MMC_CSD_CSDVER_EXT_CSD.

(nonaka)

- Don't switch MMC high-speed timing, if host controller isn't supported.
- Only check EXT_CSD STRUCTURE version when CSD version is 3.
- initialize width at sdmmc_function_alloc().

(nonaka)

Remove one of trailing whitespace which adds unnecessary difference
against netbsd-5 branch.

(enami)

fix typo (clean up the rx list on failure setting it up, not the tx list)
from openbsd -r1.75, mentioned by Alexander Nasonov in PR 42667

(dholland)

fix typos.

(christos)

Make all mechanisms optional, so we can compile in only the ones we want.

(christos)

support FFUARTCONSOLE.

(nonaka)

change the variable to USE_LIB_SASL that can be set to saslc or cyrus

(christos)

Fix some typos.

(matt)

Comment out FFS and NFS, they are part of GENERIC.

(jym)

Compile FFS and NFS statically (e.g. not modular) for GENERIC. These
file-systems can be critical for mountroot; as kernel cannot have access
to module(7)s without having / mounted first... yes, you see the point.

See http://mail-index.netbsd.org/port-i386/2011/02/11/msg002323.html

XXX Hopefully, /stand does not reside in a separate partition :o

(jym)

Add missing PUFFS (commented out).

(jym)

Missing MPLS options (commented out).

(jym)

Sync ramdisks content across i386 and amd64. Biggest change is for amd64,
where configuration for crunchgen(1) is moved from ramdisk/ to common/.

For i386, only mbrlabel(8) and various LFS related binaries are added. Yes,
LFS. It was added to amd64 ramdisk 3 years ago, I believe it's for a good
reason... ?

(jym)

Sync the ramdisks dot.profile file between amd64 and i386.

(jym)

don't hard-code /usr/pkg

(christos)

change the saslc define to USE_ and test against "no" instead of just defined.

(christos)