When truncating a file purge the fat cache after setting the new size
and after all io but before actually updating the cluster chain.

Both uvm_vnp_zerorange() and vtruncbuf() call get/putpages -> bmap -> pcbmap
and here the fat cache gets updated with information no longer valid after
truncation.

(hannken)

check that nfsd works with kernel module autoloading

(pooka)

Remove unnecessary (and broken) cleanup routines.

The cleanup routines were being used to kill the rump process and to delete
the temporary image file.  These are things automatically done by atf-run,
but it looks like this code was added here to workaround a previous bug in
the atf-run code.

Note that, in the existing form, the cleanup routines segfault (haven't
spent the time to track down why).  atf-run does not care about this
(although it should), buy Kyua does.

As a side effect, this has a teeny-tiny performance speedup in the execution
of the tests including this file.

OKed by njoly@.

(jmmv)

Force cleanup parts to exit with a success status.  Failures in cleanup
should not be allowed by atf-run (although they currently are ignored).

(jmmv)

pnbuf_cache is used all over the place outside of vfs, so put it
in one place to avoid many definitions.

(pooka)

D-Link DSM-G600 support.

(phx)

Remove the sparc-v7 compatibility wrapper functions (.div, .mul, ... .urem),
the compiler never generates (or did generate in the past) 64bit code to call
them, and they are still present in our 32bit compat library.

(martin)

Sort sections.

(wiz)

Use 'offset indent' for the list of commands. Emphasize valid commands.
Remove PERFORMANCE (too small paragraph to warrant a section).

(jruoho)

Use .Ss for non-standard subtitles in the DESCRIPTION.

(jruoho)

* As is customary, append the parameter names from
  SYNOPSIS to the description of the functions.

* Use .Ss for non-standard subtitles.

* Fix two identical subtitles.

* Add missing .It's, fix few .Ft's.

* Use a little more markup.

* Improve offsets.

(jruoho)

* Remove saslc_strmech(), which does not appear in the sources.
  Instead, document saslc_sess_getmech().

* Add FUNCTIONS and describe the functions in a list for readability.

* Sort SYNOPSIS in the order of appearance in FUNCTIONS.

* Split couple of long paragraphs for readability.

* Split the code example into EXAMPLES.

* Add missing prototypes to SYNOPSIS.

* Small markup improvements.

No contextual change.

(jruoho)

Add slhci(4) and some usb devices.

(kiyohara)

Add dummy bus_dma(9) stuff.

(kiyohara)

Postfix 2.8.2 has released.

(taca)

Don't use #endif where #else is expected.

(enami)

pacify lint.

(christos)

Add Emin for the vax. Need to check

(christos)

cpu_need_resched: make AST if no __HAVE_PREEMPTION.  Change has no effect
since MP option is mandatory on x86, but makes code more logical.

(rmind)

Clean up setting ECN bit in TOS.  Fixes PR 44742

(matt)

Fix a bug in onchip_intr_name_lookup

(matt)

fix vax typo.

(christos)

Correct "832x624x74".
Update comments.
Reverse iteratiion, so that modes are in established timing order.

(jdc)

Clarify vrefresh comment (no functional change)

(jdc)

Regenerate for:

Add EDID established timing modelines for 720x400x70 and 720x400x88

(jdc)

Add EDID established timing modelines for 720x400x70 and 720x400x88

(jdc)

Add missing quotes.

(njoly)

Update LLVM/clang snapshot. This brings much improved support for
--sysroot and a cross-compiling toolchain.

(joerg)

update copyright

(pooka)

Update copyright statements.

no functional change.

(pooka)

remove historic test

(pooka)

this was moved to usr.bin ages ago

(pooka)

Use a better width for the list.

(jruoho)

Add missing quotes.

(njoly)

1. fix PR/44455, nonprintable character doesn't show hex-visual
  under big endian.
2. ":set octal" with nonprintable character causes nbwcurses move error.
3. moving (big)words by wW/eE/bB can't handle non-ascii characters.
4. toggle uppercase/lowercase by ~ can't handle non-ascii characters.
5. don't feed CHAR_T(=wchar_t) to is* function directly.
  is* funcs with over UCHAR_MAX value may cause undefined behavior.
  some ctype implementation(such as FreeBSD) return unexpected
  result(same result as isw* func).
6. using non-ascii digit character with count/line number of ex/vi
  command may causes unexpeced result.

(tnozaki)

Whitespace nits, use \- for a minus sign.

(wiz)

more de-linting.

(christos)

update names and locations to reflect current use.

(agc)

Initial import, into othersrc/crypto, of sharedkey, a library and
utility which generates a one-time key (by default RSA 2048bit), which
is used to encrypt a secret to be shared between a number of people.
The secret is encrypted to the generated, one-time key. That key is
then distributed using libssss(3). All encryption and decryption are
accomplished via libnetpgp(3).

The secret can be recovered by presenting a threshold of shares, and
using that to recover the secret key, which can then be used to
decrypt the original secret.

The key can optionally be further protected by a passphrase, which is
set at one-time-key generation time.  The net effect is to introduce a
gate to recovery - with a passphrase, secret recovery can be gated
through an individual who knows the passphrase - this protects against
collusion to recover the secret "behind everyone's backs". At the same
time, normal passphrase propagation rules apply, and there is a potential
for a Single Point of Failure.

There are a number of use cases for this functionality, which is a
generalised and much more flexible form of encryption to a number of
users.  Shares can be distributed in non-uniform amounts, for example,
so that more important players in the group can receive more than one
share.  Ad-hoc user groups can be established.  Shares can be
encrypted to other user's keys, and protected in transit that way.
Shares can be encrypted to the distributor's key before encrypting to
the recipient's key - this will protect against collusion and
accidental exposure of the secret, but again introduces a SPoF.
Backups can be generated which can be recovered only when a quorum of
known and authorised keys have authorised it. There are many other
use cases for this - they are left as an exercise for the reader.

The original public key is kept around to provide information on
the encrypted secret - who, when, how many shares, quorum, etc.
If this is deemed to be sensitive, this information can be redacted by
using the -a switch (to anonymize the public key).

There are still some loose ends to this code, but it is functional
at the present time. Further review is welcomed and encouraged.

A worked example, taken from the regression test for sharedkey(1)
distributes the /etc/group file on a machine to 3 shares, of which 2
make a quorum. The group file is then recovered by using two of the
generated shares:

% make t
cd /usr/othersrc/crypto/external/bsd/sharedkey/sharedkey && make t
make split
rm -rf /tmp/share.* /tmp/recover.*
./sharedkey -t 2/3 /etc/group
Generating a one-time key
/tmp/share.14828a/pubring.gpg: No such file or directory
Can't read pubring /tmp/share.14828a/pubring.gpg
Can't read pub keyring
signature  2048/RSA (Encrypt or Sign) 5eb658fbc4e0c0fc 2011-03-20
Key fingerprint: 1878 2f43 ff3b 0792 2f1c bacc 5eb6 58fb c4e0 c0fc
uid              /etc/group shared 2/3 by agc Sun Mar 20 03:48:28 2011
netpgp: generated keys in directory /tmp/share.14828a/5eb658fbc4e0c0fc
Enter passphrase for 5eb658fbc4e0c0fc:
Repeat passphrase for 5eb658fbc4e0c0fc:
Shared secrets are in: /tmp/share.14828a
tar tvzf /tmp/share.*/share000.tar.gz
drwx------  2 agc      wheel          0 Mar 20 03:48 share000
lrwxr-xr-x  1 agc      wheel          0 Mar 20 03:48 share000/keyid -> 5eb658fbc4e0c0fc
-rw-r--r--  1 agc      wheel        609 Mar 20 03:48 share000/secret.gpg
-rw-r--r--  1 agc      wheel        617 Mar 20 03:48 share000/pubring.gpg
-rw-r--r--  1 agc      wheel      1323 Mar 20 03:48 share000/secring.gpg.share
tar: ustar vol 1, 5 files, 10240 bytes read, 0 bytes written in 1 secs (10240 bytes/sec)
ls -laR /tmp/share.*
total 18
drwx------  2 agc  wheel  512 Mar 20 03:48 .
drwxrwxrwt  5 root  wheel  512 Mar 20 03:48 ..
-rw-r--r--  1 agc  wheel  617 Mar 20 03:48 pubring.gpg
-rw-------  1 agc  wheel  2908 Mar 20 03:48 share000.tar.gz
-rw-------  1 agc  wheel  2908 Mar 20 03:48 share001.tar.gz
-rw-------  1 agc  wheel  2910 Mar 20 03:48 share002.tar.gz
make join
./sharedkey -o group.recover -r /tmp/share.*/share002.tar.gz /tmp/share.*/share000.tar.gz
Error: Format error (ptag bit not set)
signature  2048/RSA (Encrypt or Sign) 5eb658fbc4e0c0fc 2011-03-20
Key fingerprint: 1878 2f43 ff3b 0792 2f1c bacc 5eb6 58fb c4e0 c0fc
uid              /etc/group shared 2/3 by agc Sun Mar 20 03:48:28 2011
netpgp passphrase:
%


Status:

Vendor Tag: CROOKS
Release Tags: sharedkey-base

(agc)

Initial import of alternative threshold code to othersrc/crypto to
perform Shamir's Secret Sharing Scheme.

For more information on Sharmir's Secret Sharing Scheme (SSSS or S4),
please see:

http://en.wikipedia.org/wiki/Shamir's_Secret_Sharing

To re-iterate: this is a threshold scheme where an original secret is
split into shares. When a threshold of shares is presented, the original
secret can be reconstructed.

This is closely related to the threshold routines previously imported
into othersrc; the difference is that this code does not reveal any of
the information in the original secret until a threshold of shares is
presented. Compared to IDA, for example, an SSSS share is the same size
as the original data (compared to originalsize / threshold for IDA), but
provides privacy which IDA does not do.

This implementation of SSSS keeps the same interface as the threshold(1)
and libthreshold(3), for convenience sake.


Status:

Vendor Tag: CROOKS
Release Tags: ssss-base

(agc)

Include bsd.own.mk before making decisions based on mk.conf.

(joerg)

Create vers.c with -r, there is no point in embedding an ever changing
token that isn't even used.

(joerg)

fix some lint on i386

(christos)

Fix xref; security(8) -> security(7).

(jruoho)

--with-clang-resource-dir is no longer needed

(joerg)

enable pci@pryo@mainbus and ebus@mainbus attachments.

pryo still isn't fully working, but i'm able to talk to disk and com
properly.  bge, ohci/ehci, aceride and wm (pcie card) interrupts
are not working yet.

(mrg)

import latest

(christos)

Optimise mutex_onproc() and rw_onproc() by making them O(1), instead
of O(ncpu) for adaptive paths.  Add an LWP destructor, lwp_dtor() with
a comment describing the principle of this barrier.

Reviewed by yamt@ and ad@.

(rmind)

Add ld formats for the unit-tests.

(christos)

Merge the new gdtoa: Note that the built-in tests fail the same way as with
the old gdtoa.

(christos)

re-add most of the pryo-supporting code, and some general clean up:
- handle setting tsb size in the ptsb via IOMMU_TSBSIZE_IN_PTSB (partly from
  openbsd)
- fix IDB_INFO for systems without a STC that fault on accesses to the
  flush register (like pyro)
- move iommu_reset() to the end of iommu_init()
- use IOMMUREG_WRITE() in a few places
- add a missing membar_lookaside() (from openbsd)
- if pmap_extract() on the flush buffer fails, disable flushing (from openbsd)
- flush the pyro-style iommu when IOMMU_FLUSH_CACHE is set (partly from openbsd)
- clear up a bit of debugging code so it spew a little less (sometimes you will
  get kernel lock spinouts due to long scrolling printfs)
- use __func__ in several places

(mrg)

add pci@pryo@mainbus and ebus@mainbus support.

(mrg)

add a stdio neutral getline/getdelim implementation.

(christos)

remove pyro_xir().  we don't have the MD support for it (yet?).
clean up some debug code.  clean up the DPRINTF() code, and expand
the interrupt mapping debug code to match schizo and psycho.
set IOMMU_TSBSIZE_IN_PTSB, and if oberon, IOMMU_FLUSH_CACHE.
disable the STC explicitly.

(mrg)

add a new is_flags member to strut iommu_state.  add two flags:
IOMMU_FLUSH_CACHE    - extra flushes needed for some pyro
IOMMU_TSBSIZE_IN_PTSB - different ptsb assignment for pyro

partly from openbsd.

(mrg)

enable the tomatillo iocache.
garbage collect struct iommureg_schizo and use struct iommureg2.

(mrg)

define 'struct iommureg2', that includes the full set of regs for the
schizo and pyro iommu's.  use these in IOMMUREG().

add back IOMMUREG_READ() and IOMMUREG_WRITE() macros.

(mrg)

remove unnecessary iommuvar.h and psycho{reg,var}.h includes.

(mrg)

provide more informative error messages

(plunky)

Add usb_dma attribute for DMA capable USB host controllers
and specify that usb_dma attribute to dev/usb/usb_mem.c.

usb_mem.c uses bus_dma(9) for DMA memory allocation,
but non-DMA capable USB host controllers like slhci(4)
doesn't need them at all, and some ports don't bother to
prepare MD bus_dma(9) implementation (yet).

Discussed on current-users
http://mail-index.NetBSD.org/current-users/2011/03/13/msg015995.html
and usb_dma attribute is suggested by bouyer@.
Tested by kiyohara@ on mmeye with slhci at pcmcia.

(tsutsui)

Move MALLOC_DEFINE()s from usb_mem.c to usb_subr.c.
usb_mem.c provides functions for DMA memory allocation
that is required by DMA capable host controllers only,
but MALLOC_DEFINE() isn't DMA specific and could be required
by any USB drivers.

Discussed on current-users.

(tsutsui)

Shrink more options, file-systems and device.

(tsutsui)

More networking options.

(phx)

Now that this is a generic page, clarify the AUTHORS section a little.

(jruoho)

Use sysctl-node machdep.prodfamily to decide about the kernel to install and
to fix the console speed in /etc/ttys.

(phx)

Added missing devices: usb, sysmon, altq

(phx)

Fix use-after-free. Discovered by running tests with MALLOC_OPTIONS=J
(should this be the default for tests ?), pointed out by njoly@

(bouyer)

When extending a file, either by truncating or by writing past EOF make
sure the unallocated remainder of the last page gets zeroed.

Detected by fsx.

(hannken)

initialise memory allocated for uquot_user_hash & uquot_group_hash.
Pointed out by Nicolas Joly.

(bouyer)

avoid drmmode_get_height_align() in non XF86MODE (KMS) code.
fixes this on r6xx/r7xx.  (other cards wouldn't have called
into the path that failed.)

(mrg)

Add /usr/bin/getopt, it is needed by /usr/mdec/binstall.
Noticed by Jaime Fournier.

(martin)

update for xf86-video-ati 6.14.1

(mrg)

merge xf86-video-ati 6.14.1

(mrg)

initial import of xf86-video-ati-6.14.1.  changes include:

- several Xv fixes
- lots of kms/other stuff we can't use yet
- fixes for some crashes

Status:

Vendor Tag: xorg
Release Tags: xf86-video-ati-6-14-1

(mrg)

Fix markbad command

(ahoka)

if poll card detect is enabled, don't show "couldn't enable card".

(nonaka)

added ld0 for SD/MMC.

(nonaka)

The DSM-G600 U-Boot is so restricted that there is no possibility to pass
any bootargs. So we will just do the default multiuser boot from wd0: when
altboot was started together with a Linux initrd image.

(phx)

don't need these and they break the build

(christos)

Remove a vnode reference leak from msdosfs_rename. Release tdvp if either
doscheckpath() or relookup() fails.

Adjust test fs/vfs/t_vnops.c and remove the link count test for msdos.

Fixes PR #44661

(hannken)

DLink DSM-G600 IDSEL-14 (USB) needs special handling: Pin A is mapped to
EPIC-IRQ 1, but Pin C must be mapped to IRQ 2.

(phx)

Fix typo, adjust less list address.

(jakllsch)

from www.netlib.org

Status:

Vendor Tag: DMG
Release Tags: gdtoa-2011-03-19

(christos)

Add announce mailing list address for less.  This and previous commits changes
prompted by Snader_LB.

(jakllsch)

Add ral(4) and stge(4) for DLink DSM-G600.
Disabled DIAGNOSTIC.

(phx)

Less is now at version 436.

(jakllsch)

include flashio.h

(cliff)

Like in rest of the acpi(4) stack, queue all resume hooks.

(jruoho)

Sort options; use complete sentences in all option descriptions; add
EXIT STATUS section; use Ss for non-standard sections; upgrade history;
use standard section headers; some whitespace and punctuation nits.

(wiz)

Remove duplicate word.

(wiz)

Add missing comma.

(wiz)

Whitespace.

(skrll)

Add a simple test file for <sys/bitops.h>. For now, only ilog2(3) is tested.

(jruoho)

Dont try to read MSR_TEMPERATURE_TARGET on Core Duo Yonah

(ahoka)

Add missing period.

(jruoho)

Briefly document __predict_true() and __predict_false().

(jruoho)

Initial import of a find2 utility into othersrc.

This library and utility provide all the functionality of find(1) and more,
but with a more Unix/C syntax. e.g.

% find2 'fnmatch == *.[ch]' .
./dist/defs.h
./dist/find2.c
./dist/find2.h
./dist/main.c
% find2 'fnmatch == *.[ch] && size > 10k' .
./dist/find2.c
% find2 'mode & 04000' /sbin /bin
/sbin/shutdown
/sbin/ping
/sbin/ping6
/bin/rcmd
% find2 -l 'uid == ./Makefile' find2
812131      8 -rw-r--r--    1 agc      agc          2481 Mar 18 01:07 find2/Makefile
% find2 -l 'mtime < ./Makefile' find2
812131      8 -rw-r--r--    1 agc      agc          2481 Mar 18 01:07 find2/Makefile
% find2 'mage < 1days' find2
find2/Makefile
% find2 'mage < 2days && fnmatch == *.[ch]' dist
dist/defs.h
dist/find2.c
dist/find2.h
dist/main.c
% find2 -l 'fnmatch == *.h && mage < 1days && size > 2800' dist
812120      8 -rw-r--r--    1 agc      agc          2981 Mar 17 17:23 dist/defs.h
%


Status:

Vendor Tag: CROOKS
Release Tags: find2-base

(agc)

compat/netbsd32 ioctl support for FLASH_GET_INFO

(cliff)

Fix memory leak introduced with the struct pathbuf changes. Hi, me.
Closes PR 44625.

(dholland)

check for the appropriate terminating character

(agc)

apply the fix_label hack to partitionSizeHi as well.  it's needed there.
to do so, move the call to fix the label inside of rf_reasonable_label()
itself, so we can fix the partition sizes before calling
rf_component_label_partitionsize() itself.

fixes the failure mode where i had garbage not in numBlocksHi but in
partitionSizeHi, and the check against rf_component_label_partitionsize()
would fail and my raid would not auto-configure.

(mrg)

get rid of debugging comment

(agc)

Update LLVM snapshot. This allows deriving the target from the binary
name, provides a proper CPP mode and fixes a number of compat issues
in the integrated assembler.

Build the toolchain compiler optimized and without assertions now.

(joerg)

Note also tests(7).

(jruoho)

Fix xref; security(8) -> security(7).

(jruoho)

Remove ENHANCED_SPEEDSTEP, EST_FREQ_USERWRITE, INTEL_ONDEMAND_CLOCKMOD,
POWERNOW_K7, POWERNOW_K8, and VIA_PADLOCK.

(jruoho)

Xref x86/ichlpcib(4) instead of ichlpcib(4).

(jruoho)

Add some datasheets to SEE ALSO.

(jruoho)

Comment out IA32_TEMPERATURE_TARGET temporarily.

(jruoho)

fix typo

(cliff)

style improvements, input from matt@

(cliff)

Simplify tags target to reduce diffs with amd64/Makefile tags target.

(dyoung)

Fix tags target.

(dyoung)

Obey MKNLS=no.

(dyoung)

Honor bp_timeout even if someone sets X86_BP_FLAGS_NOBOOTCONF.

(jakllsch)

Honor bp_timeout even when X86_BP_FLAGS_NOBOOTCONF is set.
Or, in other words, bring back the 5 second countdown in pxeboot(8).

(jakllsch)

Automatically adjust pxeboot(8) stack based on the end of .bss, like is already
done in biosboot.S for boot(8).
(The heap location will need to be adjusted if pxeboot expands much more.)

(jakllsch)

Bump symtab space.

(jakllsch)

use %zu for size_t, also add rcsid

(ahoka)

- include <mips/mips_param.h> after MACHINE is defined
- remove MACHINE_ARCH and MID_MACHINE that are defined in <mips/mips_param.h>
- remove redundant comment

(tsutsui)

- include <mips/mips_param.h> after MACHINE is defined
- remove redundant comment

(tsutsui)

- include <mips/mips_param.h> after MACHINE is defined
- remove redundant comment

(tsutsui)

- include <mips/mips_param.h> after MACHINE is defined
- remove redundant comment

(tsutsui)

Include <mips/mips_param.h> after MACHINE is defined.

(tsutsui)

- include <mips/mips_param.h> after MACHINE is defined
- remove redundant comment

(tsutsui)

- include <mips/mips_param.h> after MACHINE is defined
- remove redundant comment

(tsutsui)

- include <mips/mips_param.h> after MACHINE is defined
- remove redundant comment

(tsutsui)

Include <mips/mips_param.h> after MACHINE is defined.

(tsutsui)

Fix xref; security(8) -> security(7).

(jruoho)

- include <mips/mips_param.h> after MACHINE is defined
- remove redundant comment

(tsutsui)

Fix xref; security(8) -> security(7).

(jruoho)

Include <mips/mips_param.h> after MACHINE is defined.

(tsutsui)

Fix xref; security(8) -> security(7).

(jruoho)

- no need to define MACHINE_ARCH in MD header
- include <mips/mips_param.h> after MACHINE is defined

(tsutsui)

Remove xref to nonexistent option(4).

(jruoho)

Note the previous also in the lead paragraph.

(jruoho)

Note the 'fetch_pkg_vulnerabilities=YES' also here. In lack of a proper
name, put this under "administrative security".

(jruoho)

Use .Ss for non-standard subtitles.

(jruoho)

Use explicit size with fld.

(joerg)

Use vaddr_t for pc instead of uint32_t.

(tsutsui)

Move security(8) to the section 7. Discussed on source-changes a while back.
Should address PR # 35718 at least partially.

(jruoho)

Make this compile with options DEBUG and MIPS1 only config.

(tsutsui)

Remove static branch prediction hints. They pessimize code for all CPUs
but Pentium IV. Discussed with ad@ and dsl@

(joerg)

tools/mklocale has not required ctype_bits.h since the following fix:
http://mail-index.NetBSD.org/source-changes/2010/06/20/msg011122.html

Tested build.sh tools on CentOS 5.

(tsutsui)

It would be nice if lint knew that abort() will never return.
For now just insert a /* NOTREACHED */ comment to fix the build failure.

(he)

disable use of gcc-specific "label as values" feature for HAVE_PCC

(plunky)

Make xcb_icccm.h use the same type as icccm.c for win_gravity parameter,
fixes build with PCC which does not use "unsigned int" for enum type

from upstream:

    http://cgit.freedesktop.org/xcb/util-old/commit/?id=155c99470961b38ed090ef1b4767c87ec1042386

(plunky)

Simplify some conversion by usig pre formatted strings and a function
to print a parameter.

(roy)

re-add local ebusvar.h, and use it to share between ebus.c and ebus_mainbus.c

(mrg)

Pull up fix from https://bugs.freedesktop.org/show_bug.cgi?id=28181 to
prevent random segfaults of the X server when using composite with the
ati driver.

This makes my X session rock solid when running KDE 4.5 on a MacBookPro
2,2 with an ATI Radeon Mobility X1600.  The X server was crashing very
frequently before.

OKed by mrg@ and macallan@.  Addresses PR xsrc/44730.

(jmmv)

Test the proper array index. Discovered by sparc64 automatic test runs.

(martin)

Add the specification to the SEE ALSO.

(jruoho)

RMIXL_FLASH_BAR_MASK_MAX is __BITS(34,0)
so avoid using extent if that value overflows 'long'

(cliff)

Kill a redundant lvalue cast in assembler constrained.

(joerg)

- add config for gpio, iobus, nand, flash
- config for pcic, pcmcia stubbed for now (drivers TBD)
- ensure space-tab is used to maintain format when commented

(cliff)

- add config for gpio
- add config for iobus, nand, flash

(cliff)

cosmetics

(cliff)

- fix typo when clearing pre-existing intrs in attach

(cliff)

- add support for Peripherals IO Bus for RMI XL* chips

(cliff)

- add NAND suport for RMI XL* chips

(cliff)

- add support for Peripherals IO Bus for RMI XL* chips

(cliff)

add gpio support for RMI XL* chips

(cliff)

- remove obsolete "TBD" comment

(cliff)

- add to struct rmixl_config storage for IO Peripherals Bus
base address, address mask, and bus space.

(cliff)

- add register & bit defines for GPIO, Peripherals IO Bus, Flash, NAND

(cliff)

- add suport for MT29FxGx family devices

(cliff)

Add missing MLINKS. From Stathis Kamperis.

(joerg)

Note HPPA TLS support.

(skrll)

HPPA TLS support.

(skrll)

Appease the hppa linker.

(skrll)

Another TLS reloc. Thanks to joerg for a hint.

(skrll)

Remove hack introduced in revision 1.6 to workaround a bug in atf-run that
would cause it to lock up while waiting for the p2k_ffs_fillfs test case
to complete (in some cases only).

This has been fixed by the upstream revision
3dd2481ec97b2fde76521939b6451d03ce989745 which I have just pulled into
our copy of atf.

(jmmv)

Pull up revision 3dd2481ec97b2fde76521939b6451d03ce989745 from upstream:

Really kill subprocesses of a test case before waiting for its completion

Before waiting for any leftover output of the test case after it terminates,
we must ensure that all of its subprocess are really, really dead.  Otherwise,
these subprocesses may be sharing the stdout of the test case, in which case
our wait will block (potentially indefinitely).

This finally (well, hopefully) fixes some random lockups exposed by the
NetBSD test suite.  Reported by Antti Kantee after
tests/fs/vfs/t_full:p2k_ffs_fillfs was exposing this problem in a pretty
reproducible manner.

(jmmv)