Remove <machine/atomic.h>; use <sys/atomic.h> instead.
Add <powerpc/cpuset.h> (for mpc85xx pmap).
Add some initial MP code for mpc85xx
Rework ipi code to be common across all ppcs
Change PPC to keep curlwp in %r13 while in the kernel.
Move astpending from cpu_info to mdlwp
Improve cpu_need_resched to be more MP friendly.

(matt)

Granularity is fine, but also adjust atf_tc_expect_fail() properly.

(jruoho)

Split device_t/softc.  Also rename some softc variables.  Tested on TT030.

XXX: old ugly config_console() hack should go away...

(tsutsui)

Umm, fix more split fallout.
While here use a common arg name (aux) as mentioned in autoconf(9).

(tsutsui)

bump symtab space

(christos)

assignment to sc_dev is in tcattach, we don't have sc_dev

(christos)

fix split fallout.

(christos)

Import OpenSSL 1.0.1 stable of 20110605:

this is sort of a sidegrade onto the release branch. Changes against the
last version imported:

*) Backport libcrypto audit of return value checking from HEAD, not
  all cases can be covered as some introduce binary incompatibilities.
  [Steve Henson]

*) Redirect RSA operations to FIPS module including keygen,
  encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
  [Steve Henson]

*) Add similar low level API blocking to ciphers.
  [Steve Henson]

*) Low level digest APIs are not approved in FIPS mode: any attempt
  to use these will cause a fatal error. Applications that *really* want
  to use them can use the private_* version instead.
  [Steve Henson]

*) Redirect cipher operations to FIPS module for FIPS builds.
  [Steve Henson]

*) Redirect digest operations to FIPS module for FIPS builds.
  [Steve Henson]

*) Update build system to add "fips" flag which will link in fipscanister.o
  for static and shared library builds embedding a signature if needed.
  [Steve Henson]

*) Output TLS supported curves in preference order instead of numerical
  order. This is currently hardcoded for the highest order curves first.
  This should be configurable so applications can judge speed vs strength.
  [Steve Henson]

*) Add protection against ECDSA timing attacks as mentioned in the paper
  by Billy Bob Brumley and Nicola Tuveri, see:

      http://eprint.iacr.org/2011/232.pdf

  [Billy Bob Brumley and Nicola Tuveri]

*) Add TLS v1.2 server support for client authentication.
  [Steve Henson]

*) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
  and enable MD5.
  [Steve Henson]

*) Functions FIPS_mode_set() and FIPS_mode() which call the underlying
  FIPS modules versions.
  [Steve Henson]

*) Add TLS v1.2 client side support for client authentication. Keep cache
  of handshake records longer as we don't know the hash algorithm to use
  until after the certificate request message is received.
  [Steve Henson]

*) Initial TLS v1.2 client support. Add a default signature algorithms
  extension including all the algorithms we support. Parse new signature
  format in client key exchange. Relax some ECC signing restrictions for
  TLS v1.2 as indicated in RFC5246.
  [Steve Henson]

*) Add server support for TLS v1.2 signature algorithms extension. Switch
  to new signature format when needed using client digest preference.
  All server ciphersuites should now work correctly in TLS v1.2. No client
  support yet and no support for client certificates.
  [Steve Henson]

*) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
  to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
  ciphersuites. At present only RSA key exchange ciphersuites work with
  TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
  SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
  and version checking.
  [Steve Henson]

*) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
  with this defined it will not be affected by any changes to ssl internal
  structures. Add several utility functions to allow openssl application
  to work with OPENSSL_NO_SSL_INTERN defined.
  [Steve Henson]

*) Add SRP support.
  [Tom Wu <tjw@cs.stanford.edu> and Ben Laurie]

*) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
  [Steve Henson]

*) Add EC_GFp_nistp224_method(), a 64-bit optimized implementation for
  elliptic curve NIST-P224 with constant-time single point multiplication on
  typical inputs.  EC_GROUP_new_by_curve_name() will automatically use this
  (while EC_GROUP_new_curve_GFp() currently won't and prefers the more
  flexible implementations).

  The implementation requires support for the nonstandard type __uint128_t,
  and so is disabled by default.  To include this in your build of OpenSSL,
  use -DEC_NISTP224_64_GCC_128 on the Configure (or config) command line,
  and run "make depend" (or "make update").
  [Emilia K<E4>sper <emilia.kasper@esat.kuleuven.be> (Google)]

*) Permit abbreviated handshakes when renegotiating using the function
  SSL_renegotiate_abbreviated().
  [Robin Seggelmann <seggelmann@fh-muenster.de>]

*) Add call to ENGINE_register_all_complete() to
  ENGINE_load_builtin_engines(), so some implementations get used
  automatically instead of needing explicit application support.
  [Steve Henson]

*) Add support for TLS key exporter as described in RFC5705.
  [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson]

Status:

Vendor Tag: OPENSSL
Release Tags: openssl-1-0-1-stable-SNAP-20110605

(spz)

Revert previous and make frach be 0xc0000000 for NAN's generated by strtold()
as suggested by tron.

(christos)

Create empty temporary rumphdr.types file if missing; to avoid error
messages for compat syscalls files regen.

(njoly)

Do not blindly cast things (obviously makes the test fail).

(jruoho)

Remove the dirent(3) check entirely.

(jruoho)

Use 'curlwp' here insteald of the 'l' - which isn't passsed as an arg
to the #define, but will always be curlwp.

(dsl)

Don't directly call sys_sync() from random bits of code, instead
  add do_sys_sync() that takes an 'lwp' (for l_cred) as an argument.
Explicitly pass &lwp0 rather than NULL and expecting sys_sync to
  substitute some random lwp.

(dsl)

Delete some splurious commas and very unexpected trailing space

(dsl)

Do not blindly skip test code.

Tests are not supposed to skip whole parts of code to later report a success.
Instead, they need to report a 'skipped' result so that it is clear to the
user that some part of the tests were not run.

To do this, add proper calls to atf_tc_skip where some pieces of code are
skipped.  Also, make the strtod/strtof/strtold inf and nan tests more granular
so that the *ld versions can be skipped altogether when there is no support
for them.  As a result of this, the atf_tc_expect_fail becomes accurate; it
could have hidden bugs in strtod and strtof before.

(jmmv)

Split device_t/softc. No crash on TT030.

(tsutsui)

- use CFATTACH_DECL_NEW() (no softc)
- use device_t, cfdata_t etc.

(tsutsui)

Brett Blymn -> Brett Lymn

(jnemeth)

more malloc style.

(christos)

- sprinkle const
- malloc style

(christos)

- use c99 to avoid extra ifdefs and tidy up the code
- require that undeflow returns 0 and ERANGE

(christos)

Only do the long double tests if __HAVE_LONG_DOUBLE is defined

(matt)

Fix problem with overflowing constant definition

t_strtol.c:95: warning: overflow in implicit constant conversion

(haad)

Initial import of hash64 functionality (libhash64(3), and hash64(1))
into othersrc.

This library and program use two 32-bit hash algorithms combined to
create one 64-bit hash value.  The thinking behind this is that, while
the individual algorithms are subject to collisions in the same way
that any hash routine is, a hash value created by a combination of two
hash algorithms is more difficult to exploit.

The hash algorithms gathered here are the ones aggregated by Simon Burge
in

ftp://ftp.netbsd.org/pub/NetBSD/misc/simonb/hashtest.tar.gz

and relate to this discussion from 2001(!):

http://mail-index.netbsd.org/tech-perform/2001/11/29/0002.html

Specifically, the algorithms in use are:

perlhash/perl
mousehash/mouse
dumbhash/dumb
dumbmulhash/dumbmul
crchash/crc
perlxorhash/perlxor
pythonhash/python
bernsteinhash/bernstein
honeymanhash/honeyman
pjwhash/pjw
torekhash/torek
byacchash/byacc
tclhash/tcl
gawkhash/gawk
nemhash/nem

and

size

(the last of which is a simple number of characters in the data that has
been hashed). Whilst "size" is not technically a hash value, it is useful
for a number of use cases.

A number of hashing algorithms from the original discussion have been
left out of this implementation, as they require more than 32-bits of
context to calculate the resultant hash, and one of the goals of
hash64(3) is to keep the implementation small and simple.

The API of hash64(3) has been designed to model the API of sha1(3) and
friends, with the exception that a single unsigned 64-bit integer is
returned to indicate the hash value, rather than a raw/cooked character
buffer.

The internals of hash64(3) have been designed to allow easy addition of
further 32-bit hash algorithms. As an aside, perlhash() is used internally
to invoke the desired hashing functions.

Example of its use:

% hash64 -a perl,mouse Makefile
HASH64 (Makefile) = 562f50c7f2030d74
% hash64 Makefile
HASH64 (Makefile) = 562f50c7f2030d74
% hash64 < Makefile
562f50c7f2030d74
% hash64 -a mouse,perl Makefile
HASH64 (Makefile) = f2030d74562f50c7
% hash64 -l
perlhash/perl
mousehash/mouse
dumbhash/dumb
dumbmulhash/dumbmul
crchash/crc
perlxorhash/perlxor
pythonhash/python
bernsteinhash/bernstein
honeymanhash/honeyman
pjwhash/pjw
torekhash/torek
byacchash/byacc
tclhash/tcl
gawkhash/gawk
nemhash/nem
size
% hash64 -a pjw,size Makefile Makefile
HASH64 (Makefile) = 09aa1a5a0000009c
HASH64 (Makefile) = 09aa1a5a0000009c
% wc -c Makefile
    156 Makefile
%

This whole idea can and will be extended to combining other message
digests to produce larger, but much more collision-resistant digests
and hashes, including better protection against various attacks
against digest algorithms such as pre-imaging.

Status:

Vendor Tag: CROOKS
Release Tags: hash64-base

(agc)

For completenesses sake, note, in 2011, /etc/floppytab.

(jruoho)

Change some printfs to aprint_debug(9).

(jruoho)

Make sure the kernel is compiled with -msdata=none since it uses %r2 and %r13
for other uses.

(matt)

Until PR bin/44837 is fixed, use atf_tc_fail("anticipated error did not
occur") even if the bug did not trigger.

(jruoho)

Do not fail if readdir(3) fails, probably due fts(3).

(jruoho)

Make NAN tests consistent with other platforms and FreeBSD.

(christos)

Add some tests for strtol(3).

(jruoho)

remove string placed by error(1)

(christos)

Since there is no longer a value-avg property, remove the code that
extracts it.

(pgoyette)

Update to remove value_avg from list of properties.

(pgoyette)

Remove ability to the value_avg since it no longer exists

(pgoyette)

Update to use the new name for the sensor's private data.

XXX This driver still uses the sensor's value_min for an unrelated
XXX purpose, but changing that will have to wait for another day.

(pgoyette)

Since nothing actually records or maintains an average sensor value,
remove value_avg from the prop_dict.  We can't completely remove it
because there's one driver that uses this field for its own private
purposes, so for now we just rename the member.

(pgoyette)

Add -ffloat-store add test lib/45020 also with __isnanl().

(jruoho)

Add more strings to the "infinity test".

(jruoho)

Verify PR lib/45020. Fails at least on amd64.

(jruoho)

Add some tests for stat(2) (or tests that use stat(2)).

(jruoho)

fix a couple typos

(dholland)

Handle implicit library dependency for static builds:
- libheimbase is required by libkrb5
- libwind is required by libhx509

(tsutsui)

Document new options in the proplist

(pgoyette)

Enable creation of value_{max,min,avg} entries via the proplist, as well
as setting the ENVSYS_FPERCENT flag

(pgoyette)

Split device_t/softc.  pmax is tested on GXemul.
Compile test only for alpha and vax.

XXX: alpha/tc/tcasic.c calls tcattach() via config_found(9), but
XXX: pmax/tc/tcbus.c and vax/vsa/tc_vsbus.c calls tcattach() directly.
XXX: It looks horrible...

(tsutsui)

Split device_t/softc. pmax is tested on GXemul, compile test only for alpha.

(tsutsui)

Split device_t/softc. Tested on GXemul.

(tsutsui)

- split device_t/softc
- make sure that *(dtdv_handler)() registered via dt_establish_handler()
  takes softc, not device_t as other interrupt handlers

No crash during device attach on GXemul.

(tsutsui)

Split device_t/softc.  No crash during device attach on GXemul.

(tsutsui)

Split device_t/softc of am7930.  No crash on TME and GXemul.

(tsutsui)

ApplyModifiers: if we parse a variable in the modifier list,
do not consider it a nested modifier set, unless it is followed by
endc, a ':' or end of string.

(sjg)

Don't sleep until memory becomes available.

Use kmem_zalloc() instead of kmem_alloc() + bzero().

During initialization, try to get all of the memory we need for the
vestigial time-wait structures before we set any of the structures up,
and if any single allocation fails, release all of the memory.

This should help low-memory hosts.  A much better fix postpones
allocating any memory until vtw is enabled through the sysctl.

(dyoung)

Import bind 9.7.3-P1

Status:

Vendor Tag: ISC
Release Tags: bind-9-7-3-P1

(spz)

Add locking to account improbable but possible race between the notify
handler and sysmon_envsys(9)'s refresh routine.

(jruoho)

Sync (mostly, not fully) uvm_km(9) and uvm_map(9) man pages with reality.

(rmind)

Remove unnecessary (due SME_DISABLE_REFRESH) locking.

(jruoho)

Revert maxdmap/maxsmap constification, as it causes problems on some
sparc models.  Reported by tsutsui@.

(rmind)

Defer scheduling vtw_tick() and setting the vtw hooks until
vtw_control() is called.  In this way, vtw_tick() will be re-scheduled
repeatedly while vtw is in use.

Pay tcp_vtw_was_enabled no attention in vtw_earlyinit(), since it's
always going to be 0 during initialization.

(dyoung)

- remove simple_lock(9) and use atomic_ops(3) to account pmap reference count,
  perrequest from rmind@
- while here no need to export struct pmap on sun2/sun3/sun3x

No particular problem for a week on hp300 and sun3x kernels with disabled
tcp_vtw which has been broken for a month on low memory machines.

(tsutsui)

Keep printf() happy - cast the max name length to type int

(pgoyette)

Split device_t/softc of i82586.  No crash on TME emulating sun2.

(tsutsui)

fix output format for long module name.

(nonaka)

INCLUDEDIR and LIBEXECDIR only need to be defined for the
pcc front end

(plunky)

Initial import of xen MP sources, with kernel and userspace tests.
- this is a source priview.
- boots to single user.
- spurious interrupt and pmap related panics are normal

(cherry)

fix crash by NULL pointer dereferencing I get everytime
right after this message:

bge0: discarding oversize frame (len=-4)

(cegger)

Flush CPU caches before entering S1, as noted in the specifications.

(jruoho)

sterilise the environment before running configure

(plunky)

Sort errors.

(wiz)

Sort SEE ALSO.

(wiz)

Restore $NetBSD$

(matt)

Restore $NetBSD$

(matt)

Restore $NetBSD$

(matt)

Restore $NetBSD$

(matt)

-CFATTACH_DECL(foo, sizeof(struct device),
+CFATTACH_DECL_NEW(foo, 0,

(matt)

device_t, cfdata_t, device_private, device_xname cleanup
CFATTACH_DECL_NEW for struct device only devices
bump SYMTAB_SPACE for EB7500ATX

(matt)

Missed needed use of device_private

(matt)

CFATTACH_DECL -> CFATTACH_DECL_NEW for 'struct device' devices.
change to device_t, cfdata_t.  Use device_private()

(matt)

Add some fundamental checks for memset(3). Prompted by a recent nasty
bug in the Google Android libc.

(jruoho)

Note STANDARDS and use more markup.

(jruoho)

Test also setdomainname(3).

(jruoho)

correct the section number of membar_producer.  noted by wiz@

(yamt)

forgot _NEW again.

(christos)

CFATTACH_DECL_NEW

(christos)

CFATTACH_DECL_NEW

(christos)

CFATTACH_DECL_NEW

(christos)

forgot _NEW

(christos)

forgot the _NEW part

(christos)

CFATTACH_DECL_NEW

(christos)

CFATTACH_DECL_NEW

(christos)

CFATTACH_DECL_NEW

(christos)

Rework page free lists to be sorted by color first rather than free_list.
Kept per color PGFL_* counter in each page free list.
Minor cleanups.

(matt)

more CFATTACH_DECL_NEW fixes

(christos)

CFATTACH_DECL_NEW, device_t, cfdata_t

(matt)

Covert to CFATTACH_DECL_NEW, device_t, cfdata_t

(matt)

Fix some xrefs and formatting, use more markup.

(wiz)

Add note about cherry-xenmp branch

(cherry)

Fix type in comment
(before I replace the 'l' with 'curlwp')

(dsl)

Add C stubs to allow prints early in the boot code.

(dsl)

Put all the 'CFLAGS+= -Dxxx' together.

(dsl)

Replace the trapframe fields for segment registers with two uint16_t
fields - one of which is an explicit pad.

(dsl)

Fix simple typo which made stripe target not working very at all. Issue
found by mhitch@.

(haad)

Add PPCBoot version from DS-207 and CS-407e.

(phx)

Fix xref, remove unnecessary space.

(wiz)

Initial import of the tiger reference sources to othersrc - this is
not their ultimate destination, but I'd like to check that tiger works
optimally on all our supported platforms before it gets moved to the
final destination.

http://www.cs.technion.ac.il/~biham/Reports/Tiger/tiger-src.tar.gz

Accordingly, these sources include the original testtiger program, and
a tiger(1) driver program, as well as a libtiger.

The sources have been extended to generate tiger2 hashes as well as
plain tiger.

The test programs return the same results as given in the wikipedia
page for both tiger and tiger2 on little-endian hosts - what needs to
happen now is to check the same results on big-endian and vaxual
hosts.

http://en.wikipedia.org/wiki/Tiger_(cryptography)

In passing, the original alpha optimisations have been extended to be
compiled in for LP64 hosts.  This results in a 2x speedup over the
non-LP64 code, on an amd64 VM.

The original CPP definitions for endianness have been replaced by a
runtime mechanism which should be optimised by the compiler.

Status:

Vendor Tag: ANDERSON-BIHAM
Release Tags: tiger-base

(agc)

Split device_t/softc of all ad1848 variants properly.  Compile test only.

Please check all attribute users in all files.* on device_t/softc split.

XXX: not sure if gus(4) needs device_t in ad1848_isa_softc

(tsutsui)

fix compile failure.

(nonaka)

fix compile failure.

(nonaka)

Cleanup.

(jruoho)

Few naive tests for sethostname(3).

(jruoho)

A simple test for mincore(2).

(jruoho)

Use .Ss and -offset indent.

(jruoho)

Note USE_FORT.

(jruoho)

some steps to make multiple instances work:
- don't bail if we're not the console. If we get the right paramters there is no
  reason not to work
- move wsdisplay_accessops into the softc
- call wsdisplay_cnattach only if we're the console

(macallan)

compatibility note

(yamt)

mention memory barriers necessary for non-interlocked adaptive mutex release.

(yamt)

split mainbus

(christos)

split auxio, ebus, psycho, sab.

(christos)

split audiocs

(christos)

also look for "FJSV,su" as found on PRIMEPOWER machines.
while here, copy some code from openbsd that should allow this to work
on the Fujitsu SPARC Enterprise Mx000 systems.

(mrg)

convert stackdump() to look up symbols rather than dumping raw %pc values.

(mrg)

Don't refer to extern tcbtable here, it is unused.

(dyoung)

Bump date for previous.

(wiz)

kern/42030 - tracking of file descriptors by ktrace/kdump

(alnsn)

kern/42030 - tracking of file descriptors by ktrace/kdump

(alnsn)

Describe the arguments to ctor and dtor.

(dyoung)

Add new test "syscall/t_poll".

(tron)

Add a regression test for poll(2) based on the test for pollts(2).

(tron)

pollts(2) is supposed to return -1 in particular and not a random
non-zero value in case of an error.

(tron)

allow display devices to work if they have been set up by OF even of they're
not the firmware console. While there get rid of some more code carried over
from macppc that doesn't work on sparc64 and consistently use node vs.
console_node.

(macallan)

Fix race conditions between write and getattr/setattr, which lead to
inconsitencies between kernel and filesystem idea of file size during
writes with IO_APPEND.

At mine, this resulted in a configure script producing config.status
with ": clr\n" lines stripped (not 100% reproductible, but always this
specific string. That is of little interest except for my own future
reference).

When a write is in progress, getattr/setattr get/set the maximum size
among kernel idea (grown by write) and filesystem idea (not yet grown).

(manu)

include config.h

(hans)

Recognise LINKS definitions in Makefiles, and generate the appropriate
set list entries, so that "make setlist" in src/usr.bin/cksum, for
example, DTRT:

% pwd
/usr/src/usr.bin/cksum
% make setlist

[src/distrib/sets/lists/base/mi]
./usr/bin/cksum        base-cksum-bin
./usr/bin/sum          base-cksum-bin
./usr/bin/md2          base-cksum-bin
./usr/bin/md4          base-cksum-bin
./usr/bin/md5          base-cksum-bin
./usr/bin/sha1          base-cksum-bin
./usr/bin/rmd160                base-cksum-bin

...

(agc)

Update LLVM/Clang to r132395. This brings the usual set of improvements
and support for dependency tracking in tblgen. Use this to make updates
more robust.

(joerg)

Regen for Chipsbank vendor id.

(njoly)

Add Chipsbank vendor id.

(njoly)

Remove unused header.

(tsutsui)

- remove extern from function declarations
- some KNF

(tsutsui)

Fix harmless merge botch.

(tsutsui)

Fix a typo.

(jruoho)

Indent the two lists with the same width.

(jruoho)