PR 51003 David Binderman: bzero struct before passing it around.

This is actually unnecessary as the call in question uses only fields
that have been set explicitly, but good practice regardless and it's
not like it's on a performance-critical path.

(dholland)

PR 51002 David Binderman: fix wrong printing code not enabled by default.

(dholland)

PR 51000 David Binderman: comment out unused assignment

(dholland)

Don't use the length return from snprintf to write out the result
buffer. If snprintf truncated the output, the length returned will be
greater and we'll write trash. Just call strlen instead. (And since
what we're doing is writing progress messages to the user, checking
carefully for truncation isn't really worthwhile either.)

Spotted when attending to PR 50998 from David Binderman; the issue
there (computation of an unused value) popped up because one of the
prints was already calling strlen.

(dholland)

PR 50792 David Binderman: make sure we don't divide by zero.

The loop that picks delay_divisor might conceivably reject all values,
particularly if the hardware is sulking for some reason; in that case
it'll be left zero. Use 1 instead of 0 so we don't then crash.

(dholland)

PR 50759 David Binderman: fix out of bounds array access.

If we don't find one of the expected device addresses, reject the
match, but do it by checking whether the loop matched something rather
than by testing an array entry that might be one past the end.

Note: I have also moved the test to be with the loop; since the call
to intio_map_allocate_region had gotten placed in between them, I've
moved it to go before rather than after them as (a) it doesn't
interact with the loop itself and (b) this seems like the best choice
given the history as it was deliberately added before the original
test.

I have not tried running this, not having an x68k, but as best I can
tell by reading the intio code it seems like it should be harmless
even if it's not really correct.

(dholland)

fix flagrantly wrong indent

(dholland)

Disable the code that tries to prepare a new partition table (but doesn't
do anything with it...) because it's zooming off the end of the array it's
trying to use.

It looks to me as if NEW_MAP_SIZE has been accidentally used as both
the number of blocks occupied by the new partition table and also the
number of entries in it. Or something. This needs platform knowledge
to sort out. XXX.

Workaround for PR 50757.

(dholland)

Add entries for fonts.conf.{0,5,html}

Finishes PR xsrc/49777

(pgoyette)

Install man-link from fonts-conf.5 --> fonts.conf.5 to deal with an
erroneous cross-ref in xclock's man-page.  suggested by wiz@.

Fixes PR xsrc/49777

(pgoyette)

PR 50756 David Binderman: avoid running off the end of an array in case
thing we're looking for isn't there. Which is probably impossible, but
that's not obvious.

(dholland)

PR 50755 David Binderman: assert about table entries only when they exist.

(dholland)

make sure we call sysmon_pswitch_event() from thread context

(macallan)

Add test for luhn library and utility, derived from the existing tests in the
Makefile using human2atf. Not very extensive, but gives a basic functionality
check. Results of running the tests:

[19:06:50] agc@netbsd-002 ...tests/usr.bin/luhn [3437] > sudo make test
*** WARNING: make test is experimental
***
*** Using this test does not preclude you from running the tests
*** installed in /usr/tests.  This test run may raise false
*** positives and/or false negatives.

Tests root: /usr/tests/usr.bin/luhn

t_luhn (1/1): 1 test cases
    luhn_testset_1_basic: [0.030007s] Passed.
[0.030780s]

Summary for 1 test programs:
    1 passed test cases.
    0 failed test cases.
    0 expected failed test cases.
    0 skipped test cases.

*** The verbatim output of atf-run has been saved to /usr/tests/usr.bin/luhn/atf-run.log
*** Once again, note that make test is unsupported.
[19:06:54] agc@netbsd-002 ...tests/usr.bin/luhn [3438]

(agc)

Convert the tests in the Makefile to atf, by means of human2atf, and add
the resulting Testspec file. Results of running atf checks:

[18:57:43] agc@netbsd-002 ...tests/usr.bin/delta [3390] > sudo make test
*** WARNING: make test is experimental
***
*** Using this test does not preclude you from running the tests
*** installed in /usr/tests.  This test run may raise false
*** positives and/or false negatives.

Tests root: /usr/tests/usr.bin/delta

t_delta (1/1): 3 test cases
    delta_testset_1_basics: [0.027398s] Passed.
    delta_testset_2_backwards: [0.029421s] Passed.
    delta_testset_3_more_extensive_changes: [0.027521s] Passed.
[0.089274s]

Summary for 1 test programs:
    3 passed test cases.
    0 failed test cases.
    0 expected failed test cases.
    0 skipped test cases.

*** The verbatim output of atf-run has been saved to /usr/tests/usr.bin/delta/atf-run.log
*** Once again, note that make test is unsupported.
[18:57:46] agc@netbsd-002 ...tests/usr.bin/delta [3391] >

(agc)

Handle PIC linking for tests

(christos)

Add printf-like annotation.

(joerg)

mark noreturn function exit_daemon() as __dead

(mlelstv)

Add 'a' to the stopwords list. Ok from christos

(abhinav)

Make sure that all messages end in '\n' in make_message_rcsvalid() and
compensate for it in add_rcs_file().

(christos)

PR 51190 David Binderman: simplify redundant conditionals.
Also add paranoia when looping on isdigit().

(dholland)

PR 51187 David Binderman: simplify redundant conditional

(dholland)

If MEMSIZE isn't set, #error instead of proceeding using uninitialized
stack garbage as the memory size. Please improve as needed; I don't know
anything about this hardware.

PR 51150 from David Binderman.

(dholland)

Remove undefined behavior in buf(); use buf() as intended in intarg().
While here also add includes to fix the build. Retires PR 50999 from
David Binderman.

(dholland)

PR 50997 David Binderman: fix format strings

(dholland)

PR 50709 David Binderman: memory leak

(dholland)

PR 50669 David Binderman: remove dead code

(dholland)

PR 50643 David Binderman: fix conditional order

(dholland)

Explicitly ignore errors in the hack for mucking with wedges; mark it XXX
for future attention. Other part of PR 50886 from David Binderman.

(dholland)

#if out some unreachable code that's apparently not yet supported.
PR 50886 from David Binderman.

(dholland)

bus_width is not used, so don't fetch it; but do leave the logic in place
commented out so it's there if anyone wants it in the future.
PR 50594 from David Binderman.

(dholland)

Use this program's Strdup() instead of strdup() so it exits on failure
instead of dumping core.

(dholland)

Use rewind() instead of freopen()'ing the same file for no reason. Also,
close the file when done instead of dropping it on the floor. Obsoletes
PR 50579.

(dholland)

Disable PR kern/51135 hack now that the problem is supposedly
fixed (to see if tests pass).

(pooka)

allocate cpuset structures needed by MP DDB.

(chs)

David Binderman in PR kern/51189: simplify loop conditions

(martin)

David Binderman in PR port-sparc/51188: simplify while condition

(martin)

Reduce diff with mainstream.

In the new sljt version, ppc_cache_flush() is guarded by
SLJIT_CACHE_FLUSH_OWN_IMPL. We can keep is as long as we
don't define SLJIT_CACHE_FLUSH_OWN_IMPL.

(alnsn)

Adapt to new sljit.

(nakayama)

WIP ucom support for devices with interrupt endpoints like umct

(skrll)

Unwrap some lines

(skrll)

Serialise commands

(skrll)

Restructure the abort code for TD based transfers (ctrl, bulk, intr).

In PR/22646 some TDs can be on the done queue when the abort start and,
if this is the case, they need to processed after the WDH interrupt.
Instead of waiting for WDH we release TDs that have been touched by the
HC and replace them with new ones.  Once WDH happens the floating TDs
will be returned to the free list.

(skrll)

regen

(nonaka)

Added some Intel wireless devices from OpenBSD.

(nonaka)

fix r1.409 mistake

(knakahara)

Note that i386 needs a PIE-related cleandir. Should avoid having PR 51174
happen to anyone else. :-/

(dholland)

further whitespace

(dholland)

KNF

(dholland)

Reindent.

(dholland)

PR 51185 David Binderman: simplify redundant conditional

(dholland)

PR 51184 David Binderman: simplify redundant conditional

(dholland)

PR 51183 David Binderman: simplify redundant conditional

(dholland)

PR 51182 David Binderman: fix redundant conditional

(dholland)

Update sljit entry.

(alnsn)

Call the path for makewhatis _PATH_MAKEWHATIS instead of _PATH_WHATIS,
for clarity.

(dholland)

Add notes on how to lift the BUGS entry in the man page (about not
supporting hardlinks) in case anyone thinks it's worth doing sometime.

(dholland)

Add missing rcsid.

(dholland)

Add human2atf, a means of generating ATF test scripts through a short test
specification file.

human2atf is a frontend for the ATF test harness. The idea is to provide
an easy/simple way to sepcify tests, and to provide the correct
environment for running these test cases.

In order to accomplish this, a specfile, (by default Testspec) is
given to human2atf, which is a shell script which produces the test
Makefile, and t_*.sh shell script used by atf.  The Testspec file is
itself a shell file fragment.  It specifies the names of the files
which are to be used in testing, which are typically input files to
the command being tested, and expected output files.  Care should be
taken in naming test sets so that a name that is a shell variable is
used.

It is advised that individual test cases are grouped into
functionally-related test sets.  This is not enforced, but is good
practice since the tests are used together as a regression test suite,
and so changes to utilities may show up in certain parts of the
utility being tested, but not in others.  Grouping indivual
functionally-related test cases into a test set for that functionality
is thus a good idea.

Some hand-rolled tests in the tree use shell "here" documents, which
can interfere with control flow unless used carefully.  To avoid this,
human2atf encodes its input and expected output files using base64
uuencoding, so that the contents of such files do not affect the way
that the human2atf shell script itself works.

For a test set, all that need to be specified are:
+ the testset name
+ any input and expected output files
+ the test cases themselves

To use a worked example, to enable atf tests for a (possibly fictional)
utility called b2e, a Testspec file was drawn up:

TESTNAME=b2e

TESTSET_1_NAME=basics
TESTSET_1_FILES='
1.in
1.expected
2.in
2.expected
'
TESTSET_1_CASE_1="-s eq:0 -o file:1.expected -e empty b2e < 1.in"
TESTSET_1_CASE_2="-s eq:0 -o file:2.expected -e empty b2e < 2.in"

TESTSET_2_NAME=extended
TESTSET_2_FILES='
3.in
3.expected
4.in
4.expected
5.in
5.expected
'
TESTSET_2_CASE_1="-s eq:0 -o file:3.expected -e empty b2e < 3.in"
TESTSET_2_CASE_2="-s eq:0 -o file:4.expected -e empty b2e < 4.in"
DISABLE_TESTSET_2_CASE_3="-s eq:0 -o file:5.expected -e empty b2e < 5.in"

This specifies 2 test sets, the first set containing 2 test cases, and
the second set containing 3 test cases.  The first test set is for
basic test cases, the second set is for extended tests (see the
TESTSET_n_NAME definition).  Test cases must be numbered in increasing
order from 1, and may not miss numbers.  (When the human2atf test
script finds that a test case is not defined, it stops processing).

The same numbering restrictions are in place for test sets.

To provide the tests a useful environment in which to function, input
and expected output files are specified.  These are given in the
TESTSET_n_FILES definition.

Individual test cases are specified using the TESTSET_n_CASE_m style of
definition.

Occasionally, we will want to add a test and its supporting files, but
leave it disabled (until appropriate functionality is written or
debugged fully, for example) - for that case, the
DISABLE_TESTSET_n_CASE_m definition is used.

I opted to leave the atf-style of command invocation in there as the
expected output file is specified in the test case itself but not as
part of the command, so it seemed simpler to do it that way.

Test execution (the directory and Makefile in tests/usr.bin had
already been set up):

[12:34:14] agc@netbsd-002 ~/local/human2atf-20160528 [3068] > ./human2atf -f Testspec2
Generating Makefile for b2e tests as tests-Makefile
Generating test harness t_b2e.sh for b2e
[12:34:19] agc@netbsd-002 ~/local/human2atf-20160528 [3069] > sudo cp t_b2e.sh =1
[12:34:30] agc@netbsd-002 ~/local/human2atf-20160528 [3070] > pushd +1
/usr/tests/usr.bin/b2e ~/local/human2atf-20160528
[12:34:33] agc@netbsd-002 ...tests/usr.bin/b2e [3071] > l
total 20
drwxr-xr-x  2 root  wheel  512 May 28 23:23 .
drwxr-xr-x  35 root  wheel  1024 May 28 19:17 ..
-rw-r--r--  1 root  wheel  136 May 28 19:18 Atffile
-rw-r--r--  1 root  wheel  112 May 28 19:18 Makefile
-rw-r--r--  1 root  wheel  2184 May 28 23:23 atf-run.log
-rwxr-xr-x  1 root  wheel  3368 May 28 23:23 t_b2e
-rw-r--r--  1 root  wheel  3751 May 29 12:34 t_b2e.sh
[12:34:34] agc@netbsd-002 ...tests/usr.bin/b2e [3072] > sudo make
#    build  b2e/t_b2e
echo '#! /usr/bin/atf-sh' >t_b2e.tmp
cat t_b2e.sh >>t_b2e.tmp
chmod +x t_b2e.tmp
mv t_b2e.tmp t_b2e
[12:34:39] agc@netbsd-002 ...tests/usr.bin/b2e [3073] > sudo make test
*** WARNING: make test is experimental
***
*** Using this test does not preclude you from running the tests
*** installed in /usr/tests.  This test run may raise false
*** positives and/or false negatives.

Tests root: /usr/tests/usr.bin/b2e

t_b2e (1/1): 2 test cases
    b2e_testset_1_basics: [0.031023s] Passed.
    b2e_testset_2_extended: [0.026240s] Passed.
[0.060611s]

Summary for 1 test programs:
    2 passed test cases.
    0 failed test cases.
    0 expected failed test cases.
    0 skipped test cases.

*** The verbatim output of atf-run has been saved to /usr/tests/usr.bin/b2e/atf-run.log
*** Once again, note that make test is unsupported.
[12:34:43] agc@netbsd-002 ...tests/usr.bin/b2e [3074] >

Alistair Crooks
Sun May 29 12:41:19 PDT 2016

(agc)

Update version in the comment too.

(alnsn)

Version bump after sljit import.

New version of sljit is not binary compatible with the previous
version.

(alnsn)

Update sljit revision to r313 after the import.

(alnsn)

Adapt to the new version of sljit@r313.

(alnsn)

Alloc/free macros in the new sljit version take two arguments.

(alnsn)

Fix a pilot error in the manual conflict handling..

(alnsn)

Resolve conflicts.

(alnsn)

Switch to elf notes for amd64 instead of the old key=value list to describe the
  guest requirements and support.
Add infrastructure to query the hypervisor about features support.
For verbose boot, print the features suppoted by the hypervisor for this
  guest.

(bouyer)

Don't make crunch binaries PIE

(christos)

Don't build PIE binaries for install media

(christos)

Several improvements to the ISCSI driver.

- Enable debug messages but set log level to be quiet. Provide a
  system (hw.iscsi.debug) to set the log level at run time.
- Replace old tsleep/wakeup synchronization with mutexes and condvars.
- Defer actions from callouts (basically timeouts) to the cleanup thread.
- Protect lists and unique ids with mutexes. protect connection usecount
  by using atomic operations.
- Assert kernel lock when calling into scsipi and network code.
- Use this to make send/receive/cleanup threads MPSAFE.

- Fix handling of out-of-CCB/out-of-PDU conditions against the scsipi layer.
- Bump number of PDUs to 128 to avoid virtually all out-of-PDU conditions

- Make use of softc structure for attach/detach operations.
- Track open file handles to prevent detach when busy.

- Move some global variables to make them static.

- Fix 'Overlapping Commands Attempted' error by marking commands as
  simply ordered (ATTR_SIMPLE) like FreeBSD.

(mlelstv)

Several improvents to iscsid

- debug log is now using syslog
- seperate options for log level and foreground mode
- writes a pidfile so that /etc/rc.d/iscsid works
  Now links with libutil for pidfile(), the functions login() and logout()
  needed to be renamed to avoid a conflict.
- drops the nothreads option
- handles signals to shut down gracefully
- the driver may also shut down the daemon when it terminates
  Currently this cannot work as the driver can only terminate when
  the daemon has closed the driver file handle.

(mlelstv)

missed one exit path with the previous change.

(mlelstv)

release openlock mutex before closing parent device.

(mlelstv)

Define tablesize. Useful when debugging.

(maxv)

Revert rev1.94. It apparently raises a page fault from SMEP. I need to
investigate the whole kernel mappings anyway, so I'll recommit this
patch later.

(maxv)

Sync with HEAD

(skrll)

Fix compile - missed commit.

(skrll)

Pull across fix from head for BE short reads

(skrll)

Use sparc64 code always on 32-bit sparc64 kernels since %psr read/write
instructions used in sparc's atomic_cas.S don't exist in SPARC-V9.

(nakayama)

Add missing transform for xpresent.pc.

(nakayama)

npftest needs to disable mprotect because it uses bpfjit

(christos)

As instructed by Martin@, disable DKWEDGE_METHOD_APPLE for now as it causes a
fresh install in a dual boot scenario to fail. This is because fstab will be
referencing wd(4) and the kernel expecting dk(4).

Closes PR port-macppc/51160

(sevan)

Fix mprotect for the helper.

(christos)

Skip the mprotect_exec test if PaX MPROTECT restrictions are enabled.

(christos)

Define fillkpt_blank, which creates blank entries in a page table. Use
it to map the first MB. No functional change.

(maxv)

Move proc0's stack out of the BOOTSTRAP TABLES, and map it independently
with RW permissions. Reduces the impact of a stack overflow.

(maxv)

The UART in the allwiner SoCs is not full-compatible with the 16550, and
it's not a 16750 either. Like the 16750 it has the IIR_BUSY interrupt,
which is triggered when writing to LCR while the chip
can't accept it. But unlike the 16750, it has a specific register,
HALT, to allow writing to the LCR and divisor registers, and then
commit the changes.
Tested on an A20 SoC, changing the baud rate while keeping the
tty device open and incoming data.

(bouyer)

make hostzerobroadcast default to "no".

(christos)

fix compilation without PAX_MPROTECT

(christos)

usage nit

(dholland)

treat IPIs like regular interrupts at IPL_HIGH
should fix port-powerpc/44387
tested by chuq

(macallan)

new bind

(christos)

merge conflicts

(christos)

Use vnode state to replace VI_MARKER, VI_CHANGING, VI_XLOCK and VI_CLEAN.

Presented on tech-kern@

(hannken)

Add vnode state and supporting operations and diagnostics.

Presented on tech-kern@

(hannken)

Merge the vnode and its corresponding vcache_node into one
vcache_node structure.

Print the vcache_node part in vprint() and vfs_vnode_print().

Presented on tech-kern@

(hannken)

Add more CardBus devices.

(nakayama)

Handle truncated DHCP messages, provided only the BOOTP vendor area
is truncated.  [3fd740f3ed]
OK from roy@

(prlw1)

Changing GLU_TESS_MAX_COORD to DBL_MAX causes internal constants derived
from GLU_TESS_MAX_COORD used in libtess to be +/-inf, leading to
logic errors. See
http://mail-index.netbsd.org/tech-x11/2016/05/25/msg001733.html
and
http://willkamp.com/opencpn/flyspray/index.php?do=details&task_id=2076
for details.
Change back GLU_TESS_MAX_COORD to its upstream value if it fits
in a double, or 1e15 if DBL_MAX_10_EXP is not large enough
(fix proposed by christos@)

(bouyer)

Remove a KASSERT() which is A20-specific, as well as the local variable
used here.
Fix "error: unused variable 'grp'" for non-DIAGNOSTIC kernels,
reported by Rin Okuyama.

(bouyer)

There is an issue in the way the fillkpt macro sets up pages on both
amd64 and i386.

The fillkpt loop is equivalent to the following:

do {
/* fill in the slot */
/* increment %ebx to the next slot */
/* increment %eax to the next pa */
} while (%ecx > 0)

The issue here is that if %ecx = 0 (i.e., the chunk we are trying to
map is zero-sized), there is still one entry created in the page table.
The kernel expects the va<->pa translation to be linear in low memory.
If there is a zero-sized chunk, the dead entry creates a +4096 offset in
the virtual space, with two consecutive entries that point to the same
physical address. In other words, the mappings are not linear anymore,
which causes the kernel to die.

Before my recent changes, there were only two big chunks that were
mapped, and neither of these could be zero-sized. Now, with multiple,
fine-grained chunks, it is possible that the [SYMS]+[PRELOADED_MODULES]
chunk could be zero-sized.

[PRELOADED_MODULES] is almost never here, and [SYMS] is always here on
default kernels. Except for floppies, where the bootloader does not load
[SYMS].

Should fix PR 51148.

(maxv)

Introduce M_CLEARCTX and use it instead of open-coding rcvif

No functional change.

(ozaki-r)

Use M_GETCTX

No functional change.

(ozaki-r)

sort

(nakayama)

Fix building GCC-5.3 with GCC-6.1

Cherry-pick upstream patch:

From 1e5f1089dec3af328fd03125d6778f666d0bd4e4 Mon Sep 17 00:00:00 2001
From: edlinger <edlinger@138bc75d-0d04-0410-961f-82ee72b054a4>
Date: Thu, 25 Feb 2016 15:33:50 +0000
Subject: [PATCH] 2016-02-25  Bernd Edlinger  <bernd.edlinger@hotmail.de>

        Backported from mainline
        2016-02-19  Jakub Jelinek  <jakub@redhat.com>
                    Bernd Edlinger  <bernd.edlinger@hotmail.de>

        * Make-lang.in: Invoke gperf with -L C++.
        * cfns.gperf: Remove prototypes for hash and libc_name_p
        inlines.
        * cfns.h: Regenerated.
        * except.c (nothrow_libfn_p): Adjust.

git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/branches/gcc-5-branch@233720 138bc75d-0d04-0410-961f-82ee72b054a4

Verified to work with GCC 5.3 and 6.1.

(kamil)

Consistent indent.

(wiz)

Punctuation nit.

(wiz)

Give 0,1,2 for security.pax.mprotect.ptrace and make it default to 1
as documented in sysctl(7):
0 - ptrace does not affect mprotect
1 - (default) mprotect is disabled for processes that start executing from
    the debugger (being traced)
2 - mprotect restrictions are relaxed for traced processes

(christos)

tabify, fix indent and remove trailing whitespace.

(nakayama)

Document security.pax.mprotect.ptrace

(christos)

Apply patch https://github.com/acpica/acpica/commit/\
    c450c13615f7af0673230041da4216b3de5bc4d3.patch

This patch fixes 2 issues in AccessWidth/BitOffset support and adds
comments to justify the BitOffset/BitWidth style macro usages. Lv Zheng.

This patch introduces ACPI_IS_ALIGNED() macro. Lv Zheng.

The old register descriptors are translated in AcpiTbInitGenericAddress()
with AccessWidth being filled with 0. This breaks code in
AcpiHwGetAccessBitWidth() when the registers are 16-bit IO ports and their
BitWidth fields are filled with 16. The rapid fix is meant to make code
written for AcpiHwGetAccessBitWidth() regression safer before the issue is
correctly fixed from AcpiTbInitGenericAddress(). Reported by
John Baldwin <jhb@freebsd.org>, Fixed by Lv Zheng <lv.zheng@intel.com>,
Tested by Jung-uk Kim <jkim@freebsd.org>.

(christos)

Introduce security.pax.mprotect.ptrace sysctl which can be used to bypass
mprotect settings so that debuggers can write to the text segment of traced
processes so that they can insert breakpoints. Turned off by default.
Ok: chuq (for now)

(christos)

randomize the location of the rtld.

(christos)

abstract read code to a single function (Ingo Schwarze)

(christos)

Fix get_specifier_by_index to work with indexes greater than 0.

(jmcneill)

Use M_GETCTX and M_SETCTX

No functional change.

(ozaki-r)

Give up and use TOOL_SED again.

(christos)

Fix SED issue differently.

(christos)

Export the variables so that they propagate via sets.subr -> make -> mkvars.mk

(christos)

Sync and allow arm to natively build

(skrll)

As proposed in:

http://mail-index.netbsd.org/tech-userlevel/2016/05/18/msg009999.html

and

https://www.netbsd.org/~agc/bozo-20160517.diff

add a patch to httpd to return the version string of httpd itself, and use the
-G option on the command line to enable this. This gives httpd the ability to
show, from the command line, what version is running.

% /usr/build/obj/x86_64/usr/src/libexec/httpd/bozohttpd -G
bozohttpd version bozohttpd/20160415
%

(agc)

Put the name of the dynamic linker in allocated memory, so that it becomes
part of the core file link-map, so that gdb can find it.

(christos)

PR kern/50985: use the runtime limits of the vmspace in range_test()
instead of the compile time defaults for it.

(martin)

This should be SED not TOOL_SED since this is what is set from the ${SETSENV}
command in the Makefile!

(christos)

el_map.alt can't be NULL here (Ingo Schwarze)

(christos)

Bring man pages in sync with reality.

Remove man pages run_query_html.3 and run_query_pager.3 as the corresponding
functions have been removed from apropos-utils.c
Ok by wiz@

(abhinav)

remove debug read (Ingo Schwarze)

(christos)

Effectively disable aslr for non-topdown-VA binaries (unless they are
compat32, which we deal with properly). It would be possible to get
those working too, but it is not worth the code complexity.

This makes binaries compiled with -mcmodel=medlow (and ancient binaries)
work again on sparc64, smoothing the upgrade path.

ok: christos

(martin)

Lua 5.3.2: update doc/3RDPARTY and add entry to doc/CHANGES

(salazar)

Fix armv3 segv compiling mDNS.c, from:

    https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62254
    https://gcc.gnu.org/ml/gcc-patches/2016-02/msg01095.html

(christos)

From Ingo Schwarze:

Reduce obfuscation of errno handling. There is only one purpose
non-local errno handling is needed for:  Inside el_wgets(), several
functions call down indirectly to el_wgetc(), many of them via the
dispatch table.  When el_wgetc() fails, it does properly report
failure, but then various cleanup is done which may clobber errno.
But when returning due to failure, el_wgets() wants to have errno
set to the reason of the original read failure, not to the reason
of some subsequent failure of some cleanup operation.  So el_wgetc()
needs to save errno, and if it's non-zero, el_wgets() needs to
restore it on failure.

This core logic is currently obscured by the fact that el_errno
is set and inspected at some additional places where it isn't needed.
Besides, since el_wgetc() and and el_wgets() are both in read.c,
el_errno does not need to be in struct editline, it can and should
be local to read.c in struct el_read_t.

Let's look at what can be simplified.

1. keymacro_get() abuses el_errno instead of having a proper
    error return code.  Adding that error return code is easy
    because node_trav() already detects the condition and an
    adequate code is already defined.  Returning it, testing
    for it in read_getcmd(), and returning with error from there
    removes the need to inspect el_errno from el_wgets() after
    calling read_getcmd().
    Note that resetting lastchar and cursor and clearing buffer[0]
    is irrelevant.  The code returns from el_wgets() right afterwards.
    Outside el_wgets(), these variables are no longer relevant.
    When el_wgets() is called the next time, it will call ch_reset()
    anyway, resetting the two pointers.  And as long as lastchar
    points to the beginning of the buffer, the contents of the
    buffer won't be used for anything.

2. read_getcmd() doesn't need to set el_errno again after el_wgetc()
    failure since el_wgetc() already did so.  While here, remove
    the silly "if EOF or error" comments from the el_wgetc()
    return value tests.  It's a public interface documented in a
    manual, so people working on the implementation can obviously
    be expected to know how it works.  It's a case of

      count++;  /* Increment count. */

3. In the two code paths of el_wgets() that lead up to "goto noedit",
    there is no need to save the errno because nothing that might
    change it happens before returning.

For clarity, since el_wgets() is the function restoring the errno,
also move initializing it to the same function.

Finally, note that restoring errno when the saved value is zero is
wrong.  No library code is ever allowed to clear a previously set
value of errno.  Only application programs are allowed to do that,
and even they usually don't need to do so, except when using certain
ill-designed interfaces like strtol(3).

I tested that the behaviour remains sane in the following cases,
all during execution of el_wgets(3) and with a signal handler
for USR1 installed without SA_RESTART.

* Enter some text and maybe move around a bit.
  Then send a USR1 signal.
  The signal gets processed, then read_char() resumes reading.
  Send another USR1 signal.
  Now el_wgets() sets errno=EINTR and returns -1.

* Press Ctrl-V to activate ed-quoted-insert.
  Then send a USR1 signal.
  The signal gets processed, then read_char() resumes reading.
  Send another USR1 signal.
  ed_quoted_insert() returns ed_end_of_file(), i.e. CC_EOF,
  and el_wgets() returns 0.

* Press a key starting a keyboard macro.
  Then send a USR1 signal.
  The signal gets processed, then read_char() resumes reading.
  Send another USR1 signal.
  Now el_wgets() sets errno=EINTR and returns -1.

* Press : to enter builtin command mode.
  Start typing a command.
  Then send a USR1 signal.
  The signal gets processed, then read_char() resumes reading.
  Send another USR1 signal.
  Now c_gets() returns -1, ed_command() beeps and returns CC_REFRESH,
  and el_wgets() resumes operation as it should.

I also tested with "el_set(el, EL_EDITMODE, 0)", and it returns
the right value and sets errno correctly.

(christos)

Update comment explaining why PAX_MPROTECT is disabled here

(martin)

More mknative re-runs

(skrll)

More mknative re-run.

(skrll)

Disable PAX mprotect to make just-in-time-compile tests work again.

Ok: Christos Zoulas

(hannken)

Cleanup VAT writout. To prevent issues with the sequential writing strategy
trying to write on blocks that are lost due to the synchronisation, don't just
bluntly do synchronize device caches, but split out on strategies.

(reinoud)

Re-tun mknative for earm gdb

(skrll)

md5 is a hashing algorithm, don't say it is used for encryption.
Ok by wiz@

(abhinav)

Fix typo. From Michael Scherer in PR 51162.

(wiz)

Read and load the aux vector from a core file

(christos)

Add a note for the auxv array so we can find our load location from a
core file of a PIE binary.

(christos)

Update TODO: evcnt

(knakahara)

New Rawrite32 release

(martin)

IST_MPSAFE is not a valid flag for fdtbus_intr_establish; use
FDT_INTR_MPSAFE instead.

(jmcneill)

Fix a longstanding problem with accept filters noticed by Timo Buhrmester:
sockets sitting in the accept filter can consume the entire listen queue,
such that the application is never able to handle any connections.  Handle
this by simply passing through the oldest queued cxn when the queue is full.

This is fair because the longer a cxn lingers in the queue (stays connected
but does not meet the requirements of the filter for passage) the more likely
it is to be passed through, at which point the application can dispose of it.

Works because none of our accept filters actually allocate private state
per-cxn.  If they did, we'd have to fix the API bug that there is presently
no way to tell an accf to finish/deallocate for a single cxn (accf_destroy
kills off the entire filter instance for a given listen socket).

(tls)