Don't bother with gdbserver for archs that don't build it.

(christos)

fix and rege aarch64

(christos)

add aarch64 bits

(christos)

merge xorg-server 1.20.10.

(mrg)

make(1): indent parse.c with tabs instead of spaces

(rillig)

update the rump copy of uvm_page_unbusy() to match the real version,
in particular handle PG_PAGEOUT.  fixes a few atf tests.

(chs)

make(1): reduce indentation in ParseDependencySourceKeyword

(rillig)

make(1): extract ParseDependencySourceWait

(rillig)

Inform the default installation directory in the official ftp server.

(tsutsui)

Use proper release version strings ("9.1" rather than "91") in banners.

Also define and use "MACHINE" variable to describe port names
(no uname(1) or sysctl(8) in miniroot binary list by default).

I guess the short format like "91" by ${DISTRIBREV} was used only
for split sets for floppies in 1990's releases.

Worth to pullup to netbsd-9.

(tsutsui)

make(1): define constants for enum zero-values

(rillig)

Explicitly sort set names fetched via ftp nlist.

Several binary sets are stored as symbolic links on releases and
it seems some ftpd doesn't sort nlist outputs by name in such case.

Worth to pullup to netbsd-9 and netbsd-8.

(tsutsui)

Appease awk warnings on suppressing dmesg timestamps.

> awk: warning: escape sequence `\[' treated as plain `['
>  source line number 1
>  context is
>        { h=$0; >>>  gsub("^[ <<<
> awk: warning: escape sequence `\]' treated as plain `]'
>  source line number 1

Should be pulled up to netbsd-9.

(tsutsui)

Add missing md_get_partition_range() that causes installation failure.

Also reported by Martin Trusler on testing hp300 miniroot.
Should be pulled up to netbsd-9 and netbsd-8.

(tsutsui)

Add modules set for default sets on miniroot installation.

Should be pulled up to netbsd-9 and netbsd-8.

(tsutsui)

Refactor interval timers to make it possible to support types other than
the BSD/POSIX per-process timers:

- "struct ptimer" is split into "struct itimer" (common interval timer
  data) and "struct ptimer" (per-process timer data, which contains a
  "struct itimer").

- Introduce a new "struct itimer_ops" that supplies information about
  the specific kind of interval timer, including it's processing
  queue, the softint handle used to schedule processing, the function
  to call when the timer fires (which adds it to the queue), and an
  optional function to call when the CLOCK_REALTIME clock is changed by
  a call to clock_settime() or settimeofday().

- Rename some fuctions to clearly identify what they're operating on
  (ptimer vs itimer).

- Use kmem(9) to allocate ptimer-related structures, rather than having
  dedicated pools for them.

Welcome to NetBSD 9.99.77.

(thorpej)

The kernel and X sets should be specified in MD miniroot scripts.

Otherwise they are not listed at least on ftp installation.
The problem was reported by Martin Trusler on port-hp300@:
https://mail-index.netbsd.org/port-hp300/2020/11/21/msg000174.html

Should be pulled up to netbsd-9 and netbsd-8.

(tsutsui)

make(1): remove redundant assignments

(rillig)

make(1): add test for unknown modifier in :C modifier replacement

(rillig)

make(1): remove unused parameter from meta_needed

(rillig)

make(1): remove unused parameter from JobOutput

(rillig)

Remove unnecessary inclusion of <sys/timevar.h>

(thorpej)

Remove unnecessary inclusion of <sys/timevar.h>.

(thorpej)

Remove unnecessary inclusion of <sys/timevar.h>.

(thorpej)

make(1): refuse to use vsnprintf that is prone to buffer overflows

Several parts of make intentionally depend on the guarantee that
snprintf and vsnprintf do not overflow their buffer.  If an
implementation cannot provide this guarantee, refuse to use it.

(rillig)

make(1): remove unused variable from ArchFindMember

(rillig)

make(1): extract ExpandChildrenRegular from ExpandChildren

(rillig)

make(1): indent suff.c with tabs instead of spaces

ExpandChildren is way too deeply nested.

(rillig)

pkg_install updated to 20201205

(wiz)

Merge pkg_install-20201205

(wiz)

make(1): don't concatenate identifiers in Targ_PrintType

Concatenating identifiers makes it difficult to spot them when searching
the code.  This gets rid of the special case for OP_MEMBER and MEMBER.

The same pattern is applied in the DEBUG macro, by the way.

(rillig)

make(1): indent targ.c with tabs instead of spaces

Explain the tricky details of GNode_Free.

Invert a condition in Targ_PrintNode to reduce the overall indentation.

(rillig)

make(1): extract ShuffleStrings from ApplyModifier_Order

(rillig)

When matching, try a read in case this is an optional device and isn't
actually present.

(jdc)

Return "match_result", not "1".  Pointed out by jmcneill.

(jdc)

make(1): indent remaining code of var.c with tabs instead of spaces

(rillig)

Make sure that we set sc->sc_sme to NULL in failure cases in order to avoid
calling sysmon_envsys_unregister() with an invalid struct when we detach.

(jdc)

When matching, try a read in case this is an optional device and isn't
actually present.
Make sure that we set sc->sc_sme to NULL in failure cases and when we
detach in order to avoid calling sysmon_envsys_unregister() with an
invalid struct.

(jdc)

make(1): indent large parts of var.c with tabs instead of spaces

The few remaining functions need to be cleaned up before being indented
further, to reduce the overall indentation.

(rillig)

make(1): improve explanation of tests for the SysV modifier

(rillig)

spaces to tab

(skrll)

for boot -1, don't attach more than the boot CPU most others

(mrg)

Use .MAKE.OS if we can

(sjg)

Correct typo in previous.

(kre)

PR kern/55839:
handle multiple nvme_rescan()s correctly by doing the
name-space identify only once per nsid.

fixes issue where modloading triggers multiple
rescans.

(kardel)

make(1): extract UnexportEnv from Var_UnExport

(rillig)

make(1): extract UnexportVar from Var_UnExport

(rillig)

gicv3: Only use 1 of N SPI distribution when the feature is available.

A GICv3+ implementation is not guaranteed to support 1 of N SPI
distribution. Support for this feature is indicated in GICD_TYPER.No1N.

When No1N=1, route all interrupts to the primary PE by default and only
allow a single CPU target when updating affinity.

(jmcneill)

Add __null_sentinel

(christos)

add __null_sentinel (from FreeBSD)

(christos)

make(1): rename parse functions

The word "Do" was not necessary.

(rillig)

make(1): inline Lst_Enqueue

(rillig)

make(1): inline Vector_Done

(rillig)

new OpenSSH

(christos)

bump libssh

(christos)

Merge conflicts

(christos)

new libfido2new libfido2new libfido2new libfido2new libfido2new libfido2new libfido2new libfido2new libfido2

(christos)

merge changes

(christos)

new libfido2

(christos)

__sentinel__ = __sentinel__(0)

(christos)

make(1): improve variable names in Arch_ParseArchive

The variable buf used to be a Buffer, now it is a simple string pointer.

(rillig)

make(1): use consistent variable names for list nodes

(rillig)

make(1): use fixed format for debug output of the directory cache

The previous output format had a %-20s conversion specifier.  This
produced different output depending on the length of the pathname, which
was too difficult to normalize.  By moving the directory name to the
end, it is no longer necessary to fill up any space, and the numbers are
always aligned properly.

As a result, 3 of the unit tests no longer need any special
postprocessing of their output.

(rillig)

Remove unnecessary casts

(skrll)

Ensure translation table updates are visible to the hardware walker(s)
in pmapboot_enter.

(skrll)

Trailing whitespace

(skrll)

Use __func__ in panic calls (and misc style changes)

(skrll)

Update stats in the coherent case.  Reported by jmcneill.

(skrll)

Build ip_sync.c with -Wno-error to avoid failing due to excessive stack
usage.

(thorpej)

Build x86emu.c with -fno-inline, otherwise all of the auto-inlining
the compiler does causes stack usage to exceed the limits we've set.

(thorpej)

In pfr_fix_anchor(), change an overlapping bcopy() call to a memmove()
call.

(thorpej)

- malloc(9) -> kmem(9)
- In ibm561_set_cmap(), allocating 3K on the stack is not polite; allocate
  a temporary buffer for the cmap data using kmem_alloc().

(thorpej)

Rewrite linux_sys_alarm() to use dogetitimer() / dosetitimer(), rather
than fiddling with process timers directly.

(thorpej)

Add some missing channel order related ioctl defines.

These are no-ops (unimplemented) on both FreeBSD and Solaris
and the one piece of code I've found that uses it seems to assume the
call will fail so it should be safe to leave this returning EINVAL.

However, it does need the definitions to compile...

(nia)

make(1): format coverage summary as a table

This is only used interactively, not in the official builds, therefore
the additional dependency on Perl doesn't matter.  The same result could
have been achieved in any other programming language, but probably not
as concisely.

(rillig)

tlb_print_*() are used only if _ARM_ARCH_7; caught by clang -Wunused-function.

(rin)

Fix build

(skrll)

PR bin/55834

count blocks written in unsigned 64 bit counter
rather than signed int which overflows after 2^31-1
blocks (2TiB) after which neither the 5 minute
status updates or SIGINFO (^T) reports are issued
until the negative numbers increase past 0 and
wildly inaccurate reports would be written.

(kre)

Provide and use a sev() macro for the sev instruction.

While here use the correct barrier to ensure completion of memory accesses
before a couple of the sev() calls.

(skrll)

comparision -> comparison

(wiz)

Update default pkgsrc database location from /var/db/pkg to /usr/pkg/pkgdb.

(wiz)

Mention pkg_install update.

(wiz)

Define PREFIX for pkg_install build.

(wiz)

merge pkg_install after import of 20201202 version

(wiz)

Explicitly link liblber and libssl when ldap support is enabled.
Fix sun2, i.e., MKPIC=no build.

(rin)

make(1): add test and tutorial for indirect modifiers

(rillig)

Add missig date for previous.

(rin)

make(1): rename local variable in Dir_SetPATH

The variable name should reflect the close relationship to the .DOTLAST
keyword that can be used in search paths.

(rillig)

make(1): add test for the special .MAKEFLAGS variable

(rillig)

make(1): add test for parse errors in variable name in Var_SetWithFlags

(rillig)

make(1): document handling of irrelevant conditions before 2020-07-09

(rillig)

make(1): adjust condition descriptions in test cond-short

Before, the conditions in the output had been expanded, which made them
illegible.  The expanded conditions were unrealistic as well since the
evaluation flags differ between a condition and normal evaluation
(VARE_WANTRES, VARE_UNDEFERR).

(rillig)

make(1): remove Dir_InitDir

The function name had been too ambiguous since it didn't mention the
particular directory that was initialized.  Instead of that function,
Dir_InitCur is called directly from main_Init.

The pseudo CachedDir entry ".DOTLAST" is initialized at the very
beginning.  The observable behavior is unchanged since this a
memory-only object with no connection to the file system.

(rillig)

make(1): add tests for suppressing "stopped in"

These tests demonstrate the unwanted behavior described in PR bin/55578
and PR bin/55832.

(rillig)

G/C MUTEX_PAUSE_{PAUSE,WAKE}

(skrll)

Trailing whitespace

(skrll)

Simplify the simplification... basically always set sctlr.u before setting
cpsr.e

(skrll)

Enable ldap support in racoon(8). Proposed on tech-userlevel,
"Go for it" from christos@

(bouyer)

Trailing whitespace

(skrll)

Simplify previous

(skrll)

Dequeue aqb from sc_atq_live even when the last command is failed

iavf(4) didn't dequeue aqb from sc_atq_live that is a list for
buffer in use when a command is failed by ETIMEDOUT.

This causes a panic in the following sequence:

1. enqueue an aqb to sc_atq_live at iavf_aqb_post()
2. the last command is failed by ETIMEDOUT
3. enqueue the aqb used in the failed command to sc_atq_idle
    at an error handling in iavf_attach()
4. dequeue the same aqb from sc_atq_live and enqueue sc_atq_idle
    again at iavf_cleanup_admin_queue()
  - sc_atq_idle is broken at that time
5. free the aqb in sc_atq_idle more than once

Fix PR/55822

reviewed by knakahara@n.o.

(yamaguchi)

Raspberry Pi [0-3] have been supported in big-endian mode.

Thanks ryo and skrll for their kind review!

(rin)

G/C unused.

(rin)

rpi.img: Bump boot partition in a similar manner to other images.

(rin)

Build RPI and RPI2 kernels also for earmv6{,hf}eb.

(rin)

Do not attach vchiq(4) in big-endian mode.

For big-endian mode, all data written to and read from DMA buffer must be
byte-swapped. This requires heavy modifications to third party codes.

Also, we will switch to vc4 drm driver, hopefully soon.

(rin)

Link RPI and RPI2 kernels in BE8 format for big-endian variants.

(rin)

bcm2835-fb seems byte-swapped to CPU when running in big-endian mode.

(rin)

Data written to and read from bcm283x UART registers should be in little-endian.

(rin)

Data written to bcmdmac(4) should be in little-endian.

(rin)

Data written to and read from bcmmbox(4) should be in little-endian.

(rin)

Fix style of error message.

(rin)

earmv6{,hf}eb has been switched to BE8.

Thanks for discussion in port-arm and tech-toolchain, especially
thorpej and mrg for useful comments.

(rin)

Fix confusions related to encoding of instructions.

For __ARMEB__ in BE8 mode, instructions are encoded in little-endian.
Therefore, we need to swap bytes against these in native byte order.

In other cases, i.e., __ARMEL__ and __ARMEB__ in BE32 mode, instructions
are in native byte order, and we don't need to swap them.

(rin)

Fix earmv6{,hf}eb start-up routines:

- Turn on U-bit in SCTLR before E-bit is turned on by ``setend be'',
  in order to avoid undefined condition. ARM1176JZF-S, at least, halts
  if only E-bit is turned on.

- Turn on EE-bit in SCTLR instead of B-bit as we've switched to BE8.

(rin)

Be consistent with mark-up of vop_reclaim

(pgoyette)

Switch earmv6{,hf}eb to BE8 in kernel side.

My strategy here is to define _ARM_ARCH_BE8 macro in arm/cdefs.h, if
__ARMEB__ && _ARM_ARCH_6.

Use this macro to determine whether system is compiled for and running on
BE8 mode or not.

Note that, for __ARMEB__, some conditions become compile-time constants
determined by _ARM_ARCH_BE8 macro, e.g., whether BE8 or BE32 are accepted
as a userland binary, or unaligned memory access is possible or not.

(rin)

gcc.old: Switch earmv6{,hf}eb to BE8, ported from gcc.

(rin)

Switch earmv6{,hf}eb from BE32 (deprecated) to BE8 (new), which enables

- binary compatibility with earmv7{,hf}eb and later, as well as
  COMPAT_NETBSD32 on aarch64eb
- unaligned memory access

whereas compatibility with earmv5{,hf}eb and prior is lost.

As we have never released kernel and userland for earmv6{,hf}eb yet,
this does not causes any compatibility problems.

Discussed on port-arm and tech-toolchain.

(rin)

Unbreak the lint build (sync to elftoolchain [r3912]).

(jkoshy)

Added meta-cmd-cmp

(sjg)

make(1): fix tests for :gmtime and :localtime for 32-bit time_t

https://bugs.freebsd.org/251465

(rillig)

Ensure stores are observed before the 'sev' instructions.

(skrll)

make(1): clean up variable names in Dir_FindFile

The special path entry is called .DOTLAST, therefore the local variable
should have the same name.

A variable named 'base' must not point to the slash of a pathname.  It
may only point to the character after the slash, everything else is
confusing, even if it's only for a brief moment.

(rillig)

make(1): document difference between 'cur' and 'dot'

(rillig)

make(1): add test for directory cache and exists()

(rillig)

make(1): fix memory leak for lstat cache in -DCLEANUP mode

(rillig)

Add some tests for meta mode

(sjg)

make(1): document how short-circuit evaluation behaved on 2020-07-03

(rillig)

Work in progress on dom0 PVH support. kernel boots and xl info works,
but we can't start a domU yet.

(bouyer)

Switch some debug/informational printf to device_printf(). Usefull
for multiprocessor environnement.

(bouyer)

Introduce smap_enable()/smap_disable() functions, to be used from
C code.

(bouyer)

man.conf(5): Drop support for .me files

The .me troff macros are unused for any known man-pages.

(kamil)

Apply ixgbe.c rev. 1.204 which was for ixgbe_msix_admin() to
ixgbe_legacy_irq(), too.

>  An interrupt might not arrive when a module is inserted. When an link
> status change interrupt occurred and the driver still regard SFP as
> unplugged, issue the module softint before issuing LSC interrupt.

TODO: Reduce duplicated code.

(msaitoh)

s/ we we / we /

(msaitoh)

s/ we we / we /

(msaitoh)

uhid(4): Omit needless softint indirection.

USB xfer callbacks already run in softint context at IPL_SOFTSERIAL,
and I see no reason why the call to psignal must happen instead at
the lower priority of IPL_SOFTCLOCK, so let's avoid using up the
scarce resource of softints for something that doesn't need 'em.

While here, use atomic_store_relaxed to update sc->sc_async and
atomic_load_relaxed to optimisitcally test it without acquiring
proc_lock.

(riastradh)

Note that open(O_ASYNC) does not work as advertised.

This matches Linux, which also advertises the same semantics and
fails to implement it, according to:

https://man7.org/linux/man-pages/man2/fcntl.2.html

(At what point should we collectively stop advertising the semantics
that doesn't work and call the bug a feature?)

(riastradh)

Add sgx, sgx_lc, serialize and tsxldtrk.

(msaitoh)

usb: Overhaul uhid(4) and uhidev(4) locking.

- uhidev API rules:

  1. Call uhidev_open when you want exclusive use of a report id.
    After it succeeds, you will get interrupts.

  2. Call uhidev_close when done with exclusive use of a report id.
    After it returns, you will no longer get interrupts.

    => uhidev_open/close do not nest.

  3. uhidev_write no longer requires the caller to have exclusive
    access -- if there is a write in progress, it will block
    interruptibly until done.  This way drivers for individual
    report ids need not work separately to coordinate their writes.

  4. You must uhidev_stop to abort any pending writes on the same
    report id.  (uhidev_stop no longer does anything else -- to
    ensure no more interrupts, just use uhidev_close.)

- Fix uhidev_open/close locking -- uhidev now has an interruptible
  config lock held only on first open and last close by any report id
  in the device, to serialize the transition between zero and nonzero
  numbers of references which requires opening/closing pipes and
  allocating/freeing buffers.

- Make /dev/uhidN selnotify(POLLHUP) when the device is yanked.

- Factor uhid device lookup and reference counting and dying
  detection and so on into uhid_enter/exit.

- Nix struct uhid_softc::sc_access_lock.  This served no purpose but
  to confuse me when trying to understand the logic of this beast
  (and to ensure uhidev_write exclusion, but it was uninterruptible,
  which is wrong for something that implements userland operations,
  and didn't actually work because uhidev_write did nothing to
  coordinate between different report ids).

- Fix locking in select/poll.

- Use atomics to manage UHID_IMMED to keep it simple.  (sc_lock would
  be fine too but it makes the code more verbose.)

- Omit needless UHID_ASLP -- cv_broadcast already has this
  micro-optimization.

With these changes, my Pinebook survives

for i in `jot 100`; do
        echo '###' $i
        for j in `jot 16`; do
                usbhidctl -rf /dev/uhid$j >/dev/null &
        done
        wait
done

while plugging and unplugging uhid(4) devices (U2F keys), and the U2F
keys still work as U2F keys.

ok nick, mrg

XXX pullup-9
XXX pullup-8?

Note on ABI and pullups: This changes the layout of struct
uhidev_softc, but with the sole exception of ucycom(4) -- which at
the moment is completely broken and unusable -- the only members that
USB HID drivers use are sc_udev and sc_iface, which haven't changed.
The layout of struct uhidev, which is allocated by each USB HID
driver in its own softc structure, is unchanged.

(riastradh)

make(1): initialize global variables in dir.c

Calling CachedDir_Assign requires that the variable be initialized.  On
most systems, NULL is represented as all-zero bits already.  This change
is only for the few other systems.

Add some comments explaining the implementation of Dir_AddDir since that
is tricky to read from the code alone.

(rillig)

make(1): reduce memory allocation in meta_oodate

(rillig)

make(1): reduce memory allocation in meta mode

(rillig)

make(1): add test variant for reference counting logging

(rillig)

make(1): clean up memory management for CachedDirs

Previously, the reference count for a newly created CacheDir had been
set to 1 in CacheNewDir.  This was wrong because at that point, the
object had not been referenced by any nonlocal variable.  The reference
count is no longer incremented at this point.

All callers of CacheNewDir either append the newly created CachedDir to
a SearchPath via Lst_Append and CachedDir_Ref, or they assign it to a
global variable via CachedDir_Assign.

Since the reference count is no longer wrongly incremented, it does not
need to be decremented more than necessary in Dir_End.  To keep the code
simple and maintainable, all assignments to global variables are now
handled by CachedDir_Assign.  Adding a CachedDir to a list is still done
manually via Lst_Append, and the corresponding code for decrementing is
in SearchPath_Clean and SearchPath_Free.  These details may be cleaned
up in a follow-up commit.

As a result, when OpenDirs_Done is called in the unit tests, the list of
open directories is empty.  It had been non-empty in a single unit test
before (dep-wildcards.mk), as a result of calling Dir_Expand.

The additional debug logging for the reference counting is not enabled
by default since it contains memory addresses, which makes the output
dependent on the memory allocator.

The function CachedDir_Destroy has been merged into CachedDir_Undef,
which had only been used in Dir_End before.  The new name emphasizes
that it corresponds to CachedDir_Ref.

(rillig)

make(1): fix the reference count of dotLast going negative

The memory management for dotLast is quite simple.  It is initialized
exactly once main_Init > Init_Objdir > Dir_InitDir and freed exactly
once in main_CleanUp > Dir_End.  Previously, dotLast was not freed at all.

The first call to CachedDir_Unref decremented the refCount to 0 but
didn't free anything.  Next, CachedDir_Destroy was called, which
decremented the reference count to -1, therefore skipping the actual
freeing.  This was probably an implementation mistake.

Since Dir_End is called at the very end of main_CleanUp, no code
accesses dotLast after it has been freed.

(rillig)

make(1): move CachedDir_Destroy up to the related functions

(rillig)

make(1): extract CachedDir_Free0 from CachedDir_Destroy

(rillig)

make(1): remove wrong comment in Dir_InitCur

In a makefile with repeated ".CURDIR=." lines, Dir_AddDir is called with
a NULL path, once per line.  Since the path is NULL, the search for
OpenDirs_Find is skipped and the directory is always read from disk.
The freshly read directory has a refCount of 1, and the refCount never
raises above 2.

In Dir_InitCur, the directory of the previous .CURDIR has a refCount of
2, which is decremented twice and then freed.  After this, the new
directory is placed in the global 'cur', after incrementing its refCount
to 2.

It still seems wrong that the refCount of 'cur' is 2 instead of 1, but
it works well.

(rillig)

make(1): add debug logging for OpenDirs_Done

(rillig)

Message buffer is too small, since timestamps were enforced.

(phx)

make(1): extract CacheNewDir from Dir_AddDir

Change the debug output for directories that are not found.

(rillig)

Tickets #1135 - #1138

(martin)