Eliminate few references to ltsleep.

(rmind)

- Replace shmmap_entry_pool with kmem(9), as pool is not worth.
- Sprinkle __cacheline_aligned and __read_mostly.

(rmind)

Sprinkle __cacheline_aligned and __read_mostly, also const-ify.

(rmind)

Sprinkle __cacheline_aligned and __read_mostly, make some functions static.

(rmind)

Sprinkle __cacheline_aligned and __read_mostly.

(rmind)

remove dup code.

(christos)

Create sd0, wd1 and cd0 devices, to allow mounting installation media via
USB, a second disk or a CDROM drive.

(phx)

Fix a comment.

(matt)

Update for new mips_vector_init and savectx definitions.

(matt)

Remove #error which were there for debugging

(matt)

new mdocml-1.11.2 out.

(wiz)

mention new file

(christos)

bump libmagic

(christos)

resolve conflicts

(christos)

regen.

(msaitoh)

add entries for 82579LM and 82579V

(msaitoh)

from ftp.astron.com.
- many security related fixes
- no MAXPATHLEN limits
- fixed missing text specification on ascii magic
- new ``pascal'' style string formats
- whitespace comparison fix
- more magic

Status:

Vendor Tag: CHRISTOS
Release Tags: FILE5_07

(christos)

Make intiobase and intiolimit unsigned for consistency.

(tsutsui)

Update LLVM and Clang snapshot

(joerg)

Mont FUSE filesystem with proprer source and fstype so that df and mount
display something that makes sense

(manu)

Sort options. Remove unnecessary quotes and Bk/Ek.

(wiz)

Remove trailing whitespace; consistency; add comma in SEE ALSO; sort options.

(wiz)

Remove trailing whitespace.

(wiz)

Fix spl0().  Our IPL_NONE is not 0.

(kiyohara)

remove a sparc-specific comment.

(mrg)

add a test with a sector size of 1024000 bytes

(agc)

first steps towards fujitsu SPARC64 support:

- use cputyp when writing the kernel core

(mrg)

first steps towards fujitsu SPARC64 support:

- SPARC64 systems have interrupts with upto 14 bits set, but we only look
  at the first 11 bits.  trim non IPI interrupts to 11 bits

(mrg)

first steps towards fujitsu SPARC64 support:

- use CPU_IS_USIIIi()
- only apply spitfire erratum #54 on CPU_IS_SPITFIRE() systems (this faults
  on SPARC64 cpus)
- sun4us and sun4v systems don't need dcache_flush_page or
  blast_dcache ops

(mrg)

first steps towards fujitsu SPARC64 support:

- use INTVEC() for the interrupt number, avoids overflow in the (normally)
  11 bit interrupt number (from openbsd)

(mrg)

first steps towards fujitsu SPARC64 support:

- sun4us and sun4v systems don't need dcache_flush_page or
  blast_icache ops

(mrg)

first steps towards fujitsu SPARC64 support:

- re-add cputyp, default to sun4u
- set cputyp for sun4us or sun4v systems (from openbsd)

(mrg)

catch up in dynamic buffer size changes in driver program.

add some tests for larger files, one with large sector sizes

(agc)

first steps towards fujitsu SPARC64 support:

- add IMPL_* defines for fujitsu SPARC64 processors (mostly from openbsd)
- rename CPU_IS_JALAPENO() as CPU_IS_USIIIi()
- add GETVER_CPU_MANUF()
- add CPU_IS_SPITFIRE() and CPU_IS_SPARC64_V_UP()

(mrg)

first steps towards fujitsu SPARC64 support:

- add CPU_SUN4US and CPU_SUN4V defines
- re-introduce "cputyp" and use it for sun4u/sun4us/sun4v

(mrg)

keep the documentation in sync with the library

(agc)

clear up space after we've finished using it

(agc)

move c1 and c3 data arrays into the circa struct, and allocate them
dynamically once at initialisation time.

add a separate function to free up this space at finalisation time.

this allows use of dynamic sector sizes, if desired

(agc)

use err(3) and warn(3) for error and warning printing, tiny knf.

(christos)

Fix two rather obviousbuffer overflows, one of them hitting all the time.
Inspired by corresponding FreeBSD patch.

(joerg)

when looking for a device's node we want to start with the child nodes of the
root bus, not its peers

(macallan)

Suppress whitespace at EOL to fix lib/librumphijack/t_tcpip.

(dyoung)

Remove shebang lines from test-script sources; avoids duplicate with
the build process which adds one too.

(njoly)

Typo in error message.

(njoly)

convert the main raidPtr mutex to a kmutex, and add a couple of cv's to
cover the old sleep/wakeup points for adding_hot_spare and waitForReconCond.
convert all remaining simple_lock's to kmutexes (they're not used or compiled
right now... even with all options enabled) and remove the support for them.

this leaves just a pair of tsleep()/wakeup() calls using old scheduling APIs.

(mrg)

Now that GENERIC32_IPX3x builds again, fix some pci_attach_args
constification that I'd missed.

(dyoung)

Separate the IP and UDP code paths inside libsa, so I can add HTTP booting
later.

(zoltan)

Make ${DBSYM} failure fatal.

(jakllsch)

use getmicrouptime(9) rather than microtime(9) for TIME_WAIT duration
calculation, because this doesn't get confused by system time changes,
and uses less CPU cycles
reviewed by dyoung

(drochner)

Use sysconf(_SC_PAGESIZE) instead of PAGE_SIZE.

(jakllsch)

Defer fpu_probe() call until curlwp and related variables are initialized.
Fixes pre-consinit KASSERT() in trap() caused by fnop instruction in
fpu_probe() on machines without FPU, including XM6i.

(tsutsui)

Commit the patch from
<http://mail-index.netbsd.org/current-users/2010/09/12/msg014289.html>,
fixing a "panic: pool 'pfrktable' is IPL_NONE, but called from
interrupt context" that occurred on NetBSD/sparc.

(hauke)

Heimdal has been upgraded to 1.5pre1.

(elric)

Postfix has been updated to version 2.8.3.

(tron)

Import Postfix 2.8.3:
Postfix releases 2.8.3, 2.7.4, 2.6.10 and 2.5.13 are available. These contain
a fix for CVE-2011-1720 which affects Postfix SMTP server configurations that
use Cyrus SASL authentication. Besides full releases, patches are available
for Postfix 1.1 and later.

This defect was introduced with the Postfix SASL patch, and is present in all
Postfix versions where the command "postconf mail_release_date" reports a
value of 20000314 (March 14, 2000) or greater.

Note: CVE-2011-1720 does not affect Postfix SMTP servers that use Dovecot
SASL authentication. It also does not affect the common Postfix SMTP server
configurations that use only Cyrus SASL mechanisms PLAIN and LOGIN.

More details will be available at http://www.postfix.org/CVE-2011-1720.html.

Status:

Vendor Tag: VENEMA
Release Tags: PFIX-2-8-3

(tron)

Adjust.

(jruoho)

Do also a small functional verification that user did not succeed in writing
garbage values, even if sysctl(8) reported a failure.

(jruoho)

convert regionBufferPool.mutex/cond and parityBufferPool.mutex/cond
to kmutex/cv.

(mrg)

convert parityLogPool.mutex to a kmutex.

(mrg)

convert regionInfo[regionID].mutex to a kmutex.

(mrg)

convert reintMutex to a kmutex.

(mrg)

convert parityLogDiskQueue.mutex to a kmutex/cv.

XXX: some splbio() usage to clean up here

(mrg)

Don't use a static buffer for ifaddrs (and not check for overflowing it!)
(speciallly if you are setuid!)

(christos)

sync from matt-nb5-mips64 branch

(cliff)

simplify rmixl_com_initmap() a bit

(cliff)

Constify pci_attach_args.

(dyoung)

Constify pci_attach_args.

(dyoung)

Add to tags some assembly files in common/.

(dyoung)

Constify pci_attach_args.

(dyoung)

According to their home page, N of IANA is plural.

(enami)

Use the same conditional check from <fenv.h> to fix an immediate build error
pointed out by cliff@. (This does not however sound like a sane thing to
mandate from a C99 header.)

(jruoho)

Add require.progs unifdef, to avoid failure if comp set is not
installed.

(njoly)

Stop abuse of PCI_FLAGS_IO_ENABLED and PCI_FLAGS_MEM_ENABLED.  The flags
specifically tell a driver that bridges upstream forward PCI I/O (or
memory) transactions to the device."  Only bus drivers have any business
modifying these.

(dyoung)

Use .Ss.

(jruoho)

Verify that strtod(3) honors the current rounding mode set by fesetround(3).
Passes on amd64. Prompted by a bug in the GNU C library.

(jruoho)

Move isrinit() call from early next68k_init() to cpu_configure(9)
since interrupt service routines would want evcnt(9).

(tsutsui)

Move isrinit() call from early news68k_init() to cpu_configure(9)
since interrupt service routines would want evcnt(9).

(tsutsui)

Move intr_init() call from early hp300_init() to cpu_configure(9)
since interrupt functions would want evcnt(9).

(tsutsui)

Few naive tests for pause(3).

(jruoho)

Add also a simplistic stress unit test.

(jruoho)

Fix prologue macros order.

(njoly)

Small typo in macro (Xd -> Xr).

(njoly)

Xref div(3) and remainder(3).

(jruoho)

Remove trailing '\'.

(jruoho)

Fix filesystem root leaked lock when using UFS_EXTATTR_AUTOSTART.
This way, statvfs(2) calls obtained by df(1) or umount(8) will no
longer sleep forever in the kernel to acquire the lock.

(manu)

convert RF_CommonLogData_s/RF_ReconMap_s mutex to a kmutex/cv.

(mrg)

Fix comment/string typos.

(jruoho)

Fix typo.

(jruoho)

add license

set binary directory explicitly

(agc)

Occasional MAN definitions can hold multiple pages (see src/bin/pax for one
example).

Change the parsing of the contents of MAN definitions so that both single
and multiple definitions of man pages can be recognised.

(agc)

Xref ilogb(3).

(jruoho)

Split out the ilogb(3) family out of ieee(3) and properly document it.

(jruoho)

print "raidN: detached" when unconfiguring a raid device.

(mrg)

Use ptrdiff_t to hold pointer difference to avoid coredump on LP64 system.

(enami)

AS additions from Zafer Aydogan to print the AS path. Minor cleanups to compile
without WARNS=1 hackery.

(christos)

yes, more C99 please (back out previous change).

(matt)

Make rump compile things with -std=gnu99 like the kernel and modules.

(matt)

Ticket 1429

(bouyer)

Pull up following revision(s) (requested by taca in ticket #1429):
lib/libc/resolv/res_init.c: revisions 1.11, 1.12 via patch
Don't check for ext being NULL to destroy the res_state. Instead check
the option bits like we did before.
Only try free-ing a version of res if it is not supposed to be pre-inited
like before. Fixes issues with programs passing random uninitialized data
as res_state.

(bouyer)

rearrange variable usage to kill __UNCONST
reviewed by sjg

(drochner)

Add one missing cleanup routine declaration.

(jruoho)

Add a test case for PR kern/44946. This tests that common first level sysctl
nodes (ddb, hw, machdep, etc.) are not writable by a normal user.

(jruoho)

Attempt to fix build after perfuse addition (there is no MKPERFUSE)

(manu)

derive syspkg category (CATEGORY) from LIB or PROG if one of them is defined

don't try to be smart with the gensetlist invocation, just call it when
needed.

(agc)

Add a ".debug" extension to all compat debug libraries - as pointed out by
Nicholas Joly on tech-userlevel.

Many thanks!

(agc)

Mention MKPERFUSE keyword here, too. I have forgot to commit this file before.

(haad)

Add MKPERFUSE flag to those who are enabled by default.

(haad)

Postfix 2.8.3 has released.

(taca)

Small typo (perfuse.debug -> perfused.debug).

(njoly)

On a second thought, require that close(2) fails for a descriptor already
closed with closefrom(3). For some odd reason (yet to be determined), this
check fails on 5.99.51 (amd64).

(jruoho)

Remove superfluous close(2) pointed out by martin@.

(jruoho)

Add one more naive unit test.

(jruoho)

Alter the sync-on-green and composite-sync settings:

  if the monitor doesn't support composite sync, enable sync-on-green

  then, if the monitor supports separate sync, disable composite vsync

(Prompted by macallan@.)

(jdc)

Xref raise_default_signal(3).

(jruoho)

Enable the build of perfused and libperfuse

(manu)

Correc the "enable pedestal" register definition.

(jdc)

Few basic tests for exit(3).

(jruoho)

List the ATF_TC_() thins in the order of appearance.

(jruoho)

Few basic tests for strerror(3).

(jruoho)

Few basic tests for closefrom(3).

(jruoho)

update (unused) ND option identifiers and corresponding comments

(spz)

Revert previous commit.  Locking the snapshot vnode while the file system
is suspended extends the suspension until the vnode gets unlocked by
the caller of ffs_snapshot().

Resuming the file system before expunging all snapshots and syncing the
snapshot creates races and deadlocks with journaling file systems at least.

(hannken)

Tickets 1422, 1423, 1428

(bouyer)

Pull up following revision(s) (requested by christos in ticket #1428):
lib/libc/db/hash/hash_page.c: revision 1.24
lib/libc/db/btree/bt_open.c: revision 1.25
Correct check for snprintf() overflow via Maksymilian Arciemowicz from FreeBSD.
(the bt one was ok, but set errno and make it the same for consistency).
[to be pulled up]

(bouyer)

Pull up following revision(s) (requested by fair in ticket #1423):
distrib/sets/lists/man/mi: revision 1.1305 via patch
I missed the .0 and .html changes for mr.4 replacement by gtp.4 in PR =
kern/40070
Now corrected.

(bouyer)

Pull up following revision(s) (requested by fair in ticket #1422):
share/man/man4/gtp.4: revision 1.1
share/man/man4/mr.4: delete
distrib/sets/lists/man/mi: revision 1.1304
share/man/man4/Makefile: revision 1.556
The mr(4) driver was replaced by gtp(4) in 2002, but the documentation
did not follow; fix PR kern/40070

(bouyer)

Pull up following revision(s) (requested by fair in ticket #1422):
share/man/man4/gtp.4: revision 1.1
distrib/sets/lists/man/mi: revision 1.1304
share/man/man4/Makefile: revision 1.556
The mr(4) driver was replaced by gtp(4) in 2002, but the documentation
did not follow; fix PR kern/40070

(bouyer)

Fix parsing of -A option so that the [/ptn_0_offset] part is optional, as
described in the man page.

(pgoyette)

aarpprobe() is called before MOWNER_ATTACH(&aarp_mowner), leading
to a DIAGNOSTIC panic when MBUFTRACE is defined and atalkd is started.
Fix by moving MOWNER_ATTACH(&aarp_mowner) to ddp_init().
Should fix PR kern/44734

(bouyer)

babylon5 ssh key changed (due to reinstall)

(spz)

Add a no_stack_protector function attribute to localize the effect
of disabling stack protection on a function-by-function level, as
opposed to per source file.

(christos)

Print the PGD address in the debug message.

(jym)

no c99 please.

(christos)

Do as the comment says, use ilog2(). This gets optimized directly at
compile time, no call to fls() is needed.

(jym)

Use a correct register to save/restore MIPS_COP_0_TLB_HI in
mips1_tlb_record_asids().

(tsutsui)

minor formatting and explanation changes

(agc)

Tweak some comments in mipsN_tlb_record_asids() to reduce diffs.

(tsutsui)

Remove trailing spaces and tabs.

(tsutsui)

Fix the (recurring) problem with APDPs and Xen: unmap all of them at
suspend, and let MD parts remap them when they are needed.

The issue is not PAE specific, this can be triggered with i386 and amd64.
Xen evaluates mappings in a lazy fashion, and it can incorrectly detects
recursive ones when they are pointing to inactive pmaps.

Move a comment that explains the L2 shadow page unmapping code closer to
the associated function, it makes more sense.

Now, you can save/suspend all kind of NetBSD domUs, with xbd(4) and
xennet(4) devices. Remaining bugs are in xbd(4) and xennet(4) resuming,
where the mappings have to be updated before issuing more I/Os. More
Linux code reading I guess... Stay tuned.

XXX (note to myself): move away from the machdep.sleep_state sysctl.

(jym)

Fix misc comments.

(tsutsui)

When it successfully resume a device, return true instead of 0 aka false.

(jym)

KNF.

(jym)

Implement pmf(9) suspend/resume routines for balloon(4). Trivial, heh.

(jym)

Fix a MIPS3 dependent KASSERT.

(tsutsui)

Add myself

(cheusov)

Grammar patrol

(dholland)

Fixup pre-req lib path now that this is in-tree.

(agc)

Fixup pre-req lib path now that this is in-tree.

(agc)

Initial import of libcirca(3) and circa(1) into othersrc.

The circa(3) library encodes data, protecting against damage in
transit or at rest.  The data size expands by 4/3 to provide
protection.  The library is based loosely on the protection scheme
used by CD-ROMs, and employs two Reed-Solomon erasure codings (one
inner and one outer), and a number of dispersal, delay lines and
scattering of bytes.

There are numerous areas in which single bit errors can cause
re-transmission of large amounts of data, or cause file system
data errors - encoding with libcirca(3) can protect against this.

To illustrate its use:

% wc dist/circa.h
      81    425    2898 dist/circa.h
% circa -o h2 dist/circa.h
% hd h2 | grep n | wc -l
    101
% sed -e 's|n|o|g' h2 > h2e
% hd h2e | grep n | wc -l
      0
% hd h2e | grep o | wc -l
    157
% circa -d -o h3 h2e
% diff dist/circa.h h3
%

i.e. circa(1) can recover completely from all 101 instances of the 'n'
character being transformed into 'o', in a file of 2898 characters.

(hd is a shell alias for the "codecs hexdump" transformation described
elsewhere).

Taken from the libcirca(3) man page:


DESCRIPTION
    The libcirca library implements a number of functions designed to protect
    data against errors and erasures.  It uses a number of methods of doing
    this, and is loosely based on Cross-Interleaved Reed-Solomon Coding, as
    found in CDs.  The libcirca draws heavily on the librs(3) library inter-
    nally.

    The aim of this library is to encode data in preparation for transmission
    or storage.  Various erasure codes and checksums are added to the origi-
    nal data, in an effort to recover the original data in the face of errors
    in individual bits and bytes in the encoded data.  This is the same idea
    as implemented in CD CIRC encoding.

    Internally, two Reed-Solomon erasure codes are employed, a 28,24 erasure
    code acting on the original data, and a 32,28 erasure code acting on the
    output from that.  The net effect of these two erasure codes is that the
    size of the encoded data increases by (32 / 24) or (4 / 3)

    The circa_init() function should be called prior to using the encoding
    and decoding functions.  circa_encode() is used to encode data so that it
    is protected against data corruption.  It returns the number of bytes
    produced in the out array.  To decode the original data, circa_decode()
    is called.

    When creating output files, the two functions circa_get_header() and
    circa_put_header() are used.

INTERNALS
    The libcirca library performs a number of transformations, when encoding
    data.  These transformations are modelled on the Cross Interleaved
    Reed-Solomon Coding as used in CDs.  Delay lines can straddle sectors in
    CIRC encoding for CDs, whilst libcirca strictly keeps all sector data
    within the same sector.  This has implications about integrity of data
    cross-sector, although other means, such as libthreshold(3) and
    libssss(3) can be used to replicate data at a sector level, to say noth-
    ing of protection by other means such as RAID raid(4).

    A default sector is 3136 bytes in length, although this can be changed at
    circa_init() time with the sectorsize argument.

    Each 24 bytes of input data is transformed into 32 bytes of encoded data.
    The 24 bytes of input data is known as a C3 frame, and the 32 bytes of
    encoded data is known as a C1 frame.

    The transformations are, in encoding order:

    delay1    Every other 4 bytes in the input stream are put into a delay
              slot of 1 frame.

    scramble  Within a C3 frame, the bytes are changed to occupy different
              positions.  This is done to mitigate the problem where a number
              of bytes next to each other are the subject of errors.

    Q-Parity  A Reed Solomon 28,24 erasure code is calculated over the 24
              bytes of the C3 frame.  This erasure code is inserted at the
              half-way point of the C3 frame.  This creates a C2 frame of 28
              bytes.

    Dispersal
              Each bytes is striped across the sector, according to the byte
              offset from the start of the C2 frame.  Byte b in C2 frame f
              will end up in frame (f + b) and byte b in the output sector.

    P-Parity  A Reed Solomon 32,28 erasure code is calculated on the C2
              frame, to create a 32-byte C1 frame.

    delay2    Even bytes are delayed 1 frame in the output.

    At decoding time, the transformations to the bytes occur in the opposite
    order.

The circa(1) utility is also provided to encode and decode data. Another example
of its use is provided below:

    % circa Makefile > Makefile.circa
    % hd Makefile.circa | head -20
    00000 | 00 a0 4c c1 40 0c 00 00 00 5a 31 74 63 31 00 fd  | ..L.@....Z1tc1..
    00016 | 00 00 00 00 00 35 00 3f 00 00 00 63 7f 00 00 7d  | .....5.?...c...}
    00032 | 00 00 00 00 ac 5c e4 9c 37 00 00 0a 61 49 00 00  | .......7...aI..
    00048 | 42 00 00 00 7e 32 99 00 63 00 00 64 00 00 00 00  | B...~2..c..d....
    00064 | 55 00 00 00 fd 1d 05 71 00 20 9f 24 20 17 31 00  | U......q. .$ .1.
    00080 | 2f fd 00 ff 52 00 16 00 63 00 00 52 00 00 00 00  | /...R...c..R....
    00096 | 69 4d 00 00 c3 9a 12 cc 97 20 70 2e 28 c1 61 2e  | iM....... p.(.a.
    00112 | 24 53 60 66 af 00 7d 00 20 00 00 00 69 00 00 00  | $S`f..}. ...i...
    00128 | 69 36 00 00 af 71 b1 84 76 63 57 55 90 6b 69 67  | i6...q..vcWU.kig
    00144 | 49 30 ff 3a a3 00 ff 00 20 20 00 00 34 00 00 00  | I0.:....  ..4...
    00160 | 7d 42 00 00 22 18 74 74 36 75 64 74 00 00 c0 72  | }B..".tt6udt...r
    00176 | 73 0a 65 2b ea 00 c2 00 41 20 00 00 3d 00 00 00  | s.e+....A ..=...
    00192 | 00 72 00 00 4e 20 0f 72 62 09 43 00 00 00 00 65  | .r..N .rb.C....e
    00208 | 52 54 35 0a 64 c2 0c 00 00 20 00 00 2e 00 00 00  | RT5.d.... ......
    00224 | 00 72 00 00 2c 1e 13 5f 6c 24 20 00 00 00 00 00  | .r..,.._l$ .....
    00240 | 40 75 52 3e 22 e9 f1 3c 00 61 00 00 0a 6c 00 00  | @uR>"..<.a...l..
    00256 | 00 2f 65 00 38 b9 86 3b 0a 00 00 00 00 00 00 00  | ./e.8..;........
    00272 | c0 44 0a 63 00 54 63 e2 00 3c 00 00 61 31 00 00  | .D.c.Tc..<..a1..
    00288 | 00 00 3a 00 09 69 9d 52 20 00 00 00 00 00 00 00  | ..:..i.R .......
    00304 | 00 00 6b 00 00 41 ad 27 00 24 00 00 00 20 00 00  | ..k..A.'.$... ..
    % sed -e 's|M|N|g' Makefile.circa > Makefile.damaged
    % cmp Makefile.circa Makefile.damaged
    Makefile.circa Makefile.damaged differ: char 98, line 2
    % circa -d -o Makefile.recon Makefile.damaged
    % diff Makefile Makefile.recon
    %


Status:

Vendor Tag: CROOKS
Release Tags: circa-base

(agc)

Do not return for early failures; otherwise the testcase will be
reported as passed, even if not really run.

(njoly)

write-only mapping is not portable, change it to read/write. Unbreaks
mprotect_write testcase on alpha.

(njoly)

As a first step towards more fine-grained locking, don't require
crypto_{new.free}session() to be called with the "crypto_mtx"
spinlock held.
This doesn't change much for now because these functions acquire
the said mutex first on entry now, but at least it keeps the nasty
locks local to the opencrypto core.

(drochner)

Specify DGB=-Os to make libhack a bit smaller.

(tsutsui)

Use /sbin/dmesg directly rather than kernfs and /kern/msgbuf as other ports.

(tsutsui)

remove excess newlines in debug output

(drochner)

BIND 9.8.0-P1 has released.

(taca)