make(1): remove unnecessary cp-- from Arch_ParseArchive

(rillig)

Regenerate

(ryoon)

Add Buffalo BWC-35H01 or BWC-30L01

(ryoon)

make(1): save a strlen call in ForIterate

(rillig)

make(1): constify local variables in ForIterate

(rillig)

make(1): split local variable in ForIterate

(rillig)

make(1): remove redundant includes

(rillig)

make(1): convert ForEscapes from #define to enum

(rillig)

make(1): add local variable for escapes to for_substitute

Just for debugging.  No change in code size.

(rillig)

- Remove extra callout_stop() in ixgbe_detach(). Found by knakahara@.
- Rename ix{gbe,v}_free_workqueue() to ix{gbe,v}_free_deferred_handlers().
- Add KASSERT() to functions who are called from ixgbe_handle_admin().

(msaitoh)

make(1): clean up comments in job.c

(rillig)

make(1): fix archive test when .CURDIR != .PARSEDIR

(rillig)

__predict_false() should be __predict_true() for adapter->osdep.detaching
== false check.

(msaitoh)

Don't schedule admin work while shutdown like the recovery mode timer.

The admin workqueue also runs while !IFF_UP like the recovery mode timer.
Apply the same change of ixgbe.c 1.254 to ixgbe_schedule_admin_tasklet()
to prevent panic. Found by ozaki-r@.

(msaitoh)

Add the ability to set an alternate cnmagic in the kernel config
file, e.g.:

options    CNMAGIC="\"+++++\""

(thorpej)

Fix misspellings in comment.

(isaki)

remove GCC_NO_ADDR_OF_PACKED_MEMBER now lfs_accessors.h works.

(mrg)

remove __packed from struct netbsd32_lsdisable.  the structure
is already properly aligned (the embedded int64 already has
align(4) marker applied where needed.)

same size structure generated.

(mrg)

remove GCC_NO_ADDR_OF_PACKED_MEMBER for several subdir builds
that are now handled by lfs_accessors.h internally.

(mrg)

Add CPUID(EAX=07H, ECX=0) ECX bit 16 LA57 from maxv.

(msaitoh)

Suppress -Waddress-of-packed-member just for lfs_accessors.h.

We can remove -Wno-error=address-of-packed-member from various
makefiles now.

(riastradh)

enable -Wno-error=array-bounds for GCC 9 and hppa.  machdep.c has a
bunch of warnings that i can't workaround yet.

(mrg)

oops, move GCC_NO_ADDR_OF_PACKED_MEMBER into the right header.

(mrg)

avoid an uninit warning with GCC 9.

(mrg)

wg: Use threadpool(9) and workqueue(9) for asynchronous tasks.

- Using threadpool(9) job per interface to receive incoming handshake
  messages gives the same concurrency for active interfaces but
  doesn't waste kthreads for inactive ones.

  => Can't really do this with a global workqueue(9) because there's
    no bound on the amount of time wg_receive_packets() might run
    for; we really need separate threads or threadpool jobs in order
    to avoid having one interface starve all the others.

- Using a global workqueue(9) for asynchronous peer tasks avoids
  creating unnecessary kthreads.

  => Each task does a more or less bounded amount of work, so it's OK
    to share a global workqueue -- there's no advantage to adding
    concurrency for what is almost certainly going to be CPU-bound
    asymmetric crypto.

  => This way we don't need a thread per peer or iteration over a
    list of all peers, so the task mechanism should no longer be a
    bottleneck to scaling to thousands of peers.

XXX This doesn't distribute the load across CPUs -- it keeps it on
the same CPU where the packet came in.  Should consider doing
something to balance the load -- maybe note if the current CPU is
loaded, and if so, sort CPUs by queue length or some other measure of
load and pick the least loaded one or something.

(riastradh)

wg: Use a global pktqueue rather than a per-peer pcq.

- Improves scalability -- won't hit limit on softints no matter how
  many peers there are.
- Improves parallelism -- softint was kernel-locked to serialize
  access to the pcq.
- Requires per-peer queue on handshake init to avoid dropping first
  packet.
  . Per-peer queue is currently a single packet -- should serve well
    enough for pings, dns queries, tcp connections, &c.

(riastradh)

threadpool: Simplify job reference-counting logic.

Use atomic_load_relaxed while here.

(riastradh)

threadpool: Fix man page for threadpool_job_init.

This is threadpool_job_init(job, fn, interlock, fmt, args...), not
just threadpool_job_init(job, fn, interlock).

(riastradh)

make some prototypes match the builtin properly.  GCC 9 complains
with the old version, GCC 8 is happy with this version.

tested on sparc.

(mrg)

avoid calling printf %s with NULL.  fix idea from christos.

(mrg)

adjust some fallback code to be slightly less ugly and avoid
upsetting GCC 9.

(mrg)

avoid a GCC stringop truncate warning.

(mrg)

update status for several ports that now mostly build (perhaps with
some uncommited changes.)

(mrg)

re-gen.

(mrg)

Avoid strict aliasing issue by using a separate buffer..

(christos)

wg: Fix debug output now that the priority is mixed into it.

(riastradh)

apply GCC_NO_ADDR_OF_PACKED_MEMBER

(mrg)

don't call printf %s with a NULL.

(mrg)

wg: Fix non-DIAGNOSTIC build.

(riastradh)

pass a pointer conversion via uintptr_t *and* the desired type casts.
avoids new GCC 9 warnings with C++ code.

(mrg)

avoid uninit memory and use memmove vs strncpy for a will be
truncated on purpose string.

(mrg)

avoid new GCC 9 warnings.

(mrg)

don't pass NULL to printf(), but use "<memory>" to signify this
test is operating on an in-memory only database.

(mrg)

apply GCC_NO_ADDR_OF_PACKED_MEMBER

(mrg)

use GCC_NO_ADDR_OF_PACKED_MEMBER.

(mrg)

apply -Wno-error=alloca-larger-than= for GCC 9.

(mrg)

fix for netbsd/mips.

(mrg)

fix mips and powerpc GCC 9 builds.

(mrg)

apply some GCC 9 -Wno-error*

(mrg)

fix unaligned access

(christos)

mknative-gcc for gcc 9.3.0 and aarch64eb

(jakllsch)

mark GCC 9 as (so far) netbsd 20200907 version.  this should fix the
compiler saying it is "gcc version 9.3.0 ()".

(needs mknative redo.)

(mrg)

Stop using alloca

(christos)

Adjust string sizes

(christos)

make(1): add const for Parse_IsVar

(rillig)

make(1): add const to For_Eval and For_Accum

(rillig)

make(1): clean up For_Eval

* Reduce scope of local variables.
* Remove unnecessary null character test before positive isspace call.

(rillig)

make(1): properly initialize For structure in For_Eval

Initializing a Buffer or a strlist_t with zero-valued bytes only works
by conincidence, but because it would be the correct way.  In the code
path "missing `in' in for", that zero-filled Buffer is freed using
Buf_Destroy, which could have invoked undefined behavior.

(rillig)

make(1): improve documentation in For_Eval

(rillig)

make(1): fix type of For.short_var

(rillig)

need swap for aarch64be

(ryo)

Fix panic caused by modload. http://mail-index.netbsd.org/port-arm/2020/08/30/msg006960.html

The address space reserved for modules may not be mapped in L1-L3.

(ryo)

uvm/uvm_extern.h, not machine/pmap.h, for pmap(9).  Sort includes.

(riastradh)

Fix some more uvm header fallout.  Add include guards while here.

(riastradh)

comment out ioctl that is only defined for the kernel

(christos)

fix null argument

(christos)

fix zeroing

(christos)

Appease gcc-9

(christos)

Fix some more uvm.h cleanup fallout.

(riastradh)

Note dhcpcd update

(roy)

dhcpcd: Use ppoll(2)

(roy)

Sync

(roy)

note that aarch64, earm*, sh3*, sparc*, vax, and x86-64 are
slightly updated in status.  x86-64 and sparc and mips64-eb
all appear to work well enough to complete or get well into
an atf run (still going.)

note first ICE segv on sh3

(mrg)

obsolete gcc-7 dirs.  add gcc-9 dirs.

(mrg)

re-port to netbsd/sparc*
fix build on non-constant pagesize platforms and for dkbad.

(mrg)

mostly support GCC 9 builds.  some what tested with GCC 9 but
tested to not break GCC 8 builds for x86, arm*, vax, and sparc*.

(mrg)

apply ${GCC_NO_FORMAT_TRUNCATION} to scores.c

(mrg)

build-hash-table.lo and build-vec.lo depend upon $HASH_TABLE_H not $GGC_H

(mrg)

forward declare struct vm_page, and include uvm_prot.h and uvm_pmap.h.

fixes build, presumeably after uvm.h external rototill.

(mrg)

remove unused NULL pointer that was passed to printf %s.

(mrg)

use struct zero initialiser and avoid calling memset() on a pointer
of the wrong size.

(mrg)

add support for new GCC 9 warnings that may be too much to fix
right now.  new address-of-packed-member and format-overflow
warnings have new GCC_NO_ADDR_OF_PACKED_MEMBER amd
GCC_NO_FORMAT_OVERFLOW variables to remove these warnings.

apply to a bunch of the tree.  mostly, these are real bugs that
should be fixed, but in many cases, only by removing the 'packed'
attribute from some structure that doesn't really need it.  (i
looked at many different ones, and while perhaps 60-80% were
already properly aligned, it wasn't clear to me that the uses
were always coming from sane data vs network alignment, so it
doesn't seem safe to remove packed without careful research for
each affect struct.)  clang already warned (and was not erroring)
for many of these cases, but gcc picked up dozens more.

(mrg)

avoid trying to printf() a NULL as %s.  fixes likely bug.

(mrg)

fix merge botch: netbsd doesn't want any of this.

(mrg)

make(1): add test for the -de option

(rillig)

rework error message to never call printf() %s with NULL.

(mrg)

avoid calling printf() %s with NULL.

(mrg)

mknative-gcc for gcc 9.3.0 and earmv5*.

(mrg)

pass down earm* not earmv5* to mknative-gcc.

(mrg)

Fix fallout from previous uvm.h cleanup.

- pmap(9) needs uvm/uvm_extern.h.

- x86/pmap.h is not usable on its own; it is only usable if included
  via uvm/uvm_extern.h (-> uvm/uvm_pmap.h -> machine/pmap.h).

- Make nvmm.h and nvmm_internal.h standalone.

(riastradh)

update mknative-gcc status.  everything but earmv5 is done, though
few have been tested to actually build yet.

(mrg)

mknative-gcc for gcc 9.3.0 and arm, earmv[467]*, i386, mipseb, mips64*,
and powerpc64.

(mrg)

don't include vxworks-dummy.h twice

(mrg)

mknative-gcc for gcc 9.3.0 and arm64, alpha, hppa, ia64, m68*, mipseb,
powerpc, riscv*, sh3*, sparc*, vax.

move data from README.gcc8 to README.gcc9.

(mrg)

Fix a broken corner case of strlen()/strnlen() on aarch64eb

Previously a string such as "\x1\x1\x1\x1\x1\x1\x1" would count as
0 instead of 7 on BE.

(jakllsch)

lists/base/mi: fix lines with 4 fields

ok mrg

(rillig)

make(1): replay the changes from v1.283

I accidentally reverted them in v1.284.

(rillig)

make(1): clean up comments about parsing

(rillig)

make(1): fix comments about setting .PARSEDIR and .PARSEFILE

(rillig)

make(1): make GetActuallyIncludingFile faster

In deeply nested includes, starting the search from the inner end is
faster since it needs fewer comparisons.

(rillig)

make(1): fix .INCLUDEDFROMDIR/.INCLUDEDFROMFILE

(rillig)

make(1): add test for .INCLUDEDFILE combined with .for loops

The .for loops are implemented as a special kind of .include, therefore
they affect the .INCLUDEDFROM variable.

(rillig)

Track the SSIR per-cpu, rather than globally.

(thorpej)

aarch64: switch CPU to the kernel's byte order during boot

(jakllsch)

Suppress pool redzone message unless booted with debug.

(riastradh)

make(1): make test for .INCLUDEDFROMDIR simpler

The .info and .warning directives provide exactly the early expansion
that this test needs.  No more .for for getting a snapshot of a
variable.

(rillig)

Adjust aarch64 bus_space tags to also work on aarch64eb

(jakllsch)

Round of uvm.h cleanup.

The poorly named uvm.h is generally supposed to be for uvm-internal
users only.

- Narrow it to files that actually need it -- mostly files that need
  to query whether curlwp is the pagedaemon, which should maybe be
  exposed by an external header.

- Use uvm_extern.h where feasible and uvm_*.h for things not exposed
  by it.  We should split up uvm_extern.h but this will serve for now
  to reduce the uvm.h dependencies.

- Use uvm_stat.h and #ifdef UVMHIST uvm.h for files that use
  UVMHIST(ubchist), since ubchist is declared in uvm.h but the
  reference evaporates if UVMHIST is not defined, so we reduce header
  file dependencies.

- Make uvm_device.h and uvm_swap.h independently includable while
  here.

ok chs@

(riastradh)

- Document all of the various interrupt levels in the Processor Stataus
  register, and provide symbolic names for them as well.
- Use ALPHA_PSL_IPL_* values directly for IPL_*.

(thorpej)

Add missing prototypes for __bs_c(f,_bs_sm_?_swap)

(implementations of these already exist for aarch64)

(jakllsch)

AArch64 instructions are always LE: swap if we're BE

(jakllsch)

make(1): add tests for some of the special sources

(rillig)

make(1): fix local variable type in ParseIsEscaped

(rillig)

make(1): fix return type of ParseIsEscaped

(rillig)

make(1): remove redundant prototype for ParseMark

(rillig)

make(1): remove redundant prototypes for local functions from parse.c

(rillig)

make GCC 9.3.0 build itself:
- common-target is needed before common now
- complete build-sort.o framework
- fix cyclic dep
- fix .c vs .cc issues

(mrg)

make(1): remove initial size argument from Hash_InitTable

In all but one case this argument was set to auto-detect anyway.  The
one case where it was set was not worth keeping this complicated API.

(rillig)

mknative-gcc for gcc 9.3.0 and amd64.

the tree has quite a number of issues to resolve before it builds.

(mrg)

disable -Wabi for GCC 9.

(mrg)

avoid passing NULL pointers to printf() like functions and %s.

(mrg)

make the 2nd argument to __memmove_chk() const

(mrg)

work around various headers not being fully open without some
special #define being present (ie, define them.)

(mrg)

make(1): make Hash_Table independent from -funsigned-char

This only makes a difference for Hash_Table keys outside the ASCII
character set, and these are barely used in practice, if at all.

The effects of this change can only be seen in debug mode, when printing
the full contents of the variable namespaces.  In this output, the order
of the entries might change.

All other use cases stay the same as before.

(rillig)

fix various merge botches; we may need to re-port the ThreadLister code.

(mrg)

avoid always passing NULL to printf()-like functions

(mrg)

make(1): add expected test result for hashing variable names

(rillig)

make(1): add test for intentional hash collision for variable names

Hash collisions may slow down make in certain special situations.  There
is no point though in maliciously triggering such a situation since
anyone who can inject values into makefiles can easily run shell
commands using the :!cmd! modifier or similar mechanisms.  Crafting
variable names just to slow down make is thus not an attack vector.

(rillig)

find header files in /usr/include/gcc-9 and the gcc cpu dir.

(mrg)

endfor not endif

(mrg)

add distrib/sets/fmt-list to format the file lists consistently

This program is much more complicated than sort-list in the same
directory.  It takes care of aligning the fields of the lines so that
lines from the same directory are aligned to each other.  This reduces
horizontal jumps for the category and flags fields.

(rillig)

fix alignment in ad.sh3

(rillig)

disable fetching stuff from libsanitizer for now.  it needs libstdc++
to build in mknative using gcc makefiles, and our locale changes break
the "normal" build of this.

make sure MKNATIVE_TARGET is always set.  since mknative-gcc doesn't
care about the version any more (we keep two versions for gcc/gcc.old)
simply set it to "gcc".

(mrg)

port reachover framework to GCC 9.

(mrg)

pull out a few more things for GCC 9.

(mrg)

port to GCC 9.3.

(mrg)

merge GCC 9.3.0.

(mrg)