tests/indent: migrate test token_case_label to lsym_case_label

(rillig)

When we build against DESTDIR, add the lint library path for DESTDIR during
the link phase for lint.

(christos)

For private and module libraries, build lint but do not install the lint
libraries. We want to lint the private library code and we want to be able
to link against the lint libraries, for example llib-largon2 from cgdconfig
which is accessed via PROGDPLIBS.

(christos)

tests/indent: migrate token_comma to lsym_comma

The section on initializer values is new.

(rillig)

tests/indent: move test for ':' to lsym_token.c

(rillig)

indent: treat L"string" as a single token

There is never whitespace between the 'L' and the string literal or the
character constant. There might be a backslash-newline between them, but
that case was not handled before either.

No functional change.

(rillig)

indent: clean up and document input handling

The transformation of moving comments from after an 'if (expr)' after
the following brace has a large implementation cost (about 300 lines of
code) and makes input handling quite complicated. Document the overall
idea to save future readers some time.

No functional change.

(rillig)

tests/lint: update history of exporting unnamed objects

(rillig)

lint: do not export unnamed objects to the .ln file

Since these objects are unnamed, they cannot be referenced by other
files or even create name clashes.

Seen in usr.sbin/cpuctl/aarch64.c, reported by Christos.

(rillig)

lint: allow Bison debug functions to be unused

The grammar is only compiled using Bison during development time. In the
official builds, it is built with byacc.

No functional change.

(rillig)

tests/lint1: reproduce parse error for accidentally exported name

Seen in usr.sbin/cpuctl/aarch64.c line 118.

error: aarch64.ln:857:
    not alnum or _:  (for '118d-1.0d1200000000_tmpA16PcC')

The name 00000000_tmp was not supposed to be exported.

(rillig)

lint2: in case of parse errors, output the offending line

This provides more of a clue than a simple '(not alnum or _: )',
especially in the output of build.sh.

While here, change the format of the error message to the standard
'%s:%d'. Since these are internal errors, they are not supposed to occur
often, so no need to change error(1).

(rillig)

sqlite3-3.37.0 out

(wiz)

tests/lint2: fix regular expressions containing backslash

In regular expressions, a backslash must be doubled. In this case, ATF
didn't complain because the single backslashes were used in the pattern
'\($2\)', where they produced '\(...\\)'. Omitting the backslash for the
closing parenthesis was apparently OK; other regex implementations
complain about this.

(rillig)

lint: move fgetln peculiarities out of read_ln_line

This allows the function parameter to be a pointer to const.

No functional change.

(rillig)

remove extra line

(christos)

Explain the usage errors.

(christos)

-lpthread to LDADD (fixes lint build)

(christos)

sys.mk: fix duplicate character in character class of LINTFLAGS

The character class contained two times U+0020, which makes one of them
redundant, so remove the character class. PR toolchain/18248 already
contained these two spaces.

Apparently there is no need to support tabs in this place, otherwise
someone would have noticed it in the 17 years since this pattern had
been added in sys.mk 1.83 from 2004-01-27.

No functional change.

(rillig)

make: with the option -de in jobs mode, output expanded commands

This helps tracking down the actual cause of build failures in build
systems like NetBSD's build.sh that use highly abstracted commands that
are mainly defined in terms of variables.

(rillig)

tests/make: demonstrate unhelpful debug log in case of errors

The releng build job that runs lint fails.  It outputs the usage message
of lint, which doesn't include the failing option.  After that, make
outputs:

*** Failed target: lint-cgdconfig
*** Failed commands:
${LINT} ${LINTFLAGS} \
    ${_LDFLAGS.${:Ucgdconfig}:C/-L[  ]*/-L/Wg:M-L*} \
    ${LOBJS.${:Ucgdconfig}} ${_LDADD.${:Ucgdconfig}}

Make doesn't output the expanded command, therefore the log doesn't show
the offending option '-pthread' that leads to the usage message.

(rillig)

Amend the comment about UsePAM; the ChallengeResponseAuthentication
setting is deprecated, replaced by KbdInteractiveAuthentication,
confirmed both by man page and code.

(he)

usr.sbin: remove unnecessary CONSTCOND, lint no longer needs it

Since 2021-01-31, lint no longer requires a CONSTCOND comment in a
do-while-0 statement since this is a common code pattern, especially in
statement-like macros.

sed -i -E 's,} while \(/\* ?CONSTCOND ?\*/ ?0\),} while (0),' */*.[ch]

(rillig)

usr.bin: remove unnecessary CONSTCOND, lint no longer needs it

Since 2021-01-31, lint no longer requires a CONSTCOND comment in a
do-while-0 statement since this is a common code pattern, especially in
statement-like macros.

(rillig)

make: remove CONSTCOND comments, lint no longer needs them

(rillig)

indent: reword README

(rillig)

indent: accept a few formatting suggestions from indent

The remaining issues are still that the conditions look ambiguous even
with -eei, and that __attribute__ is broken into a separate line.

No functional change.

(rillig)

regen

(ryo)

Add pselect6(2) system call to COMPAT_LINUX32

(ryo)

indent: illustrate probably_looking_at_definition with examples

No functional change.

(rillig)

indent: fix out of bounds memory access (since 2021-11-25)

(rillig)

Add target attribute

(christos)

add options COMPAT_LINUX32 (commented out)

(ryo)

tests/indent: demonstrate strange formatting (since 2021-11-25)

(rillig)

/etc/shells - Add tmux.

We ship tmux in base and tmux can be used as user's default shell.

(uwe)

indent: rename dump functions to output

No functional change.

(rillig)

indent: inline switch_buffer

The function name was not accurate all the time. Now that
inp_from_comment is a separate function, it doesn't make sense anymore
to offload the 3 simple statements to a separate function.

No functional change.

(rillig)

indent: add assertions for input handling

Just to document the invariants; the code is already OK.

(rillig)

Force failure of the nfs_renamerace_cycle, p2k_ffs_renamerace_cycle,
and puffs_renamerace_cycle test cases as they fail only randomly or
only on some systems.

(gson)

kern: Add debug.crashme.null_jump.

(riastradh)

aarch64: Fix stack traces from jump-to-null.

(riastradh)

Trailing whitespace

(skrll)

Apply some errata workarounds for Cortex A17

(skrll)

Add the Cortex A17 Diagnostic control registers

(skrll)

Drive-by trailing whitespace removal.

(skrll)

Remove unnecessary #include

(skrll)

fix build for aarch64eb

(ryo)

Add "Gearmo USA-FTDI4X" (a 4-port USB serial adapter) to the list of
devices supported by uftdi(4).

(thorpej)

The wg_handshake_timeout test case was failing because it contained
atf_fail "failed to trigger PR kern/56252" without a corresponding
atf_expect_fail "PR kern/56252", which makes no sense.  Since the
test case does occasionally fail on real hardware, fix this by adding
the atf_expect_fail rather than by removing the atf_fail.

(gson)

Mask the bottom 5 bits in the standard pic way. NFC.

(skrll)

no need to shout

(skrll)

The changes on compat/linux/common/linux_termios.c should also
be applied to compat/linux32/common/linux_termios.c, and reduce diff.

- remove duplicate fd_putfile(). This will be done after "goto out;".
- catch up from compat/linux/common/linux_termios.c r1.26  TIOCSPTLCK
- catch up from compat/linux/common/linux_termios.c r1.37  (TCOOFF/TCOON) of the TCXONC
- catch up from compat/linux/common/linux_termios.c r1.39  FIOCLEX and FIONCLEX

(ryo)

ticket #1373 has not been removed, it was ticket #1377 (oops)

(martin)

Remove ticket #1373 (which has been undone)

(martin)

Backout previous pullup from ticket #1377:

etc/rc.d/npf_boot: revision 1.5
etc/rc.d/pf_boot: revision 1.8

(martin)

describe screenblank(1)

(nia)

use MNT_NFS4ACLS instead of MNT_ACLS (which was changed before to mean
MNT_POSIX1EACLS)

(christos)

Ticket #1377

(martin)

Pull up following revision(s) (requested by sborrill in ticket #1377):

etc/rc.d/npf_boot: revision 1.5
etc/rc.d/pf_boot: revision 1.8

Load rc configuration based on rcvar, not name, so that correct settings
in /etc/rc.conf.d are loaded.

Usually this does not matter as rcvar and name are set to the same value.

For pf_boot and npf_boot, rcvar is set to pf and npf respectively.

Prior to the change, if:
rc.conf contains nfp=YES
rc.conf.d/npf does not exist
Then:
/etc/rc.d/npf_boot rcvar
outputs:
$npf=YES

If:
rc.conf contains npf=NO (or is not set)
rc.conf.d/npf contains npf=YES
Then:
/etc/rc.d/npf_boot rcvar
outputs:
$npf=NO

This means that in the latter case, at boot time the npfctl start command
is never run and the firewall is not operational.

(martin)

Delete trailing whitespace

(gson)

indent: enhance debug logging for input handling

(rillig)

indent: add buf_add_range for adding characters to a buffer

No functional change.

(rillig)

indent: replace inp_enlarge with inp_add

Previously, inbuf.inp.s was only updated at the very end of reading a
line from the input file, which meant that during debugging, it pointed
to invalid memory. Updating all fields in inbuf.inp after every
reallocation makes the code less tricky to understand.

No functional change.

(rillig)

indent: split inp_read_line into smaller functions

No functional change.

(rillig)

indent: extract inp_from_file from inp_read_line

No functional change.

(rillig)

indent: remove code that fixes malformed preprocessor directives

(rillig)

indent: move ind_add from io.c to indent.c

It's a general-purpose function that is not directly related to input or
output.

(rillig)

Remove #error, fixes the build. I don't understand why this would not work
on big endian. We use this construct to convert pointers everywhere else.

(christos)

declare xc

(christos)

Load rc configuration based on rcvar, not name, so that correct settings
in /etc/rc.conf.d are loaded.

Usually this does not matter as rcvar and name are set to the same value.
For pf_boot and npf_boot, rcvar is set to pf and npf respectively.

Prior to the change, if:
rc.conf contains nfp=YES
rc.conf.d/npf does not exist

Then:
/etc/rc.d/npf_boot rcvar
outputs:
# npf_boot
$npf=YES

If:
rc.conf contains npf=NO (or is not set)
rc.conf.d/npf contains npf=YES

Then:
/etc/rc.d/npf_boot rcvar
outputs:
# npf_boot
$npf=NO

This means that in the latter case, at boot time the npfctl start command
is never run and the firewall is not operational.

(sborrill)

Use rcvar=$name syntax consistently in rc.d scripts which makes it easier to
search for rc.d scripts where rcvar differs from name. No functional change.

(sborrill)

The internal architecture name of golang is "arm" or "arm64", which has
different naming conventions from netbsd's machine[] or machine_arch[]
("earmv7", "aarch64", etc.) and cannot be used as is.
Instead, use define LINUX_GO_RT0_SIGNATURE_ARCH{32,64}.

(ryo)

- LINUX_GO_RT0_SIGNATURE for compat_linux is no longer enabled by default.
  it is now defined in each arch same as the other *_SIGNATURE definitions.
- add new LINUX32_GO_RT0_SIGNATURE for compat_linux32, and is defined in amd64 and aarch64.

(ryo)

Fix anonymous memory object leak for sigcode.

- Repeating "modload compat_linux && /emul/linux/bin/ls && modunload compat_linux"
  will reproduce this problem.
- It cause in exec_sigcode_map(), anon-object for sigcode was created at
  first exec, but it remained even after exec_remove.
- Fixed that the anon-object for sigcode is created at exec_add(), and the
  anon-object reference is removed at exec_remove().
- sigobject_lock is no longer needed since it is locked by exec_lock.
- The compat_16 module rewrites the e_sigcode entry in emul_netbsd directly and
  does not use exec_add()/exec_remove(), so it needs to call
  sigcode_alloc()/sigcode_free() on its own.

(ryo)

indent: prevent undefined behavior in inp_line_start

No functional change.

(rillig)

indent: make error message for missing command line arguments clearer

(rillig)

tests/indent: use escape sequence for newline characters

No functional change.

(rillig)

tests/indent: test in-place formatting with parse errors

The file gets formatted until the end, despite the parse error. This may
destroy layout details of the code but usually preserves the overall
structure. Since all source code is supposed to be under version
control, this is not a problem.

(rillig)

indent: make fits_in_one_line independent of ps

This makes it easier to understand the function. Having the dependency
to the input line handling is already complicated enough.

No functional change.

(rillig)

indent: fix accidentally joined and broken comments (since 2019-04-04)

The fixed version is not perfect as it gets the indentation of the last
line of the first comment wrong, but at least indent doesn't generate
malformed output anymore.

(rillig)

tests/indent: demonstrate bugs in comment processing

(rillig)

tests/indent: fix misleading typedef name

(rillig)

indent: update cross-reference comments for bug in comment handling

The function was renamed in io.c 1.122 from 2021-11-19.

(rillig)

tests/indent: rename ps.in_parameter_declaration in test as well

(rillig)

indent: rename ps.in_function_parameters to match reality

This flag is only set while parsing the parameters of a function
definition, but not for a function declaration. See buffer_add in the
test fmt_decl.

No functional change.

(rillig)

indent: improve heuristic for spaces around '*' in declarations

(rillig)

tests/indent: demonstrate more example for broken '*' in declarations

Since indent-2021.11.25.16.41.33.

(rillig)

tests/indent: demonstrate 'sizeof(int) *expr' (since 2021-11-25)

Broken in indent-2021.11.25.16.41.33, while fixing the spacing around
'*' in declarations.

(rillig)

Ticket #1376

(martin)

indent: eliminate 3 negations in tokenizer

No functional change.

(rillig)

Pull up following revision(s) (requested by gutteridge in ticket #1376):

distrib/notes/common/main: revision 1.566
distrib/notes/common/main: revision 1.567

main: add note about amd64 now having a "bios" image, too

main: add .gz extension to .img references for consistency

(martin)

indent: extract lex_asterisk_unary into separate function

No functional change.

(rillig)

indent: condense code for building tokens from characters

No functional change.

(rillig)

indent: in lexi, assign lsym and next_unary in consistent order

No functional change.

(rillig)

indent: fix heuristic for declaration/definition to post-1990 reality

(rillig)

tests/indent: remove unnecessary option for testing '-pcs'

Before lexi.c 1.157 from 2021-11-25, the option '-npsl' had been
necessary to prevent a wrong line break, as indent wrongly interpreted
the function declaration as a function definition, just because the
first ')' in the declaration line was not immediately followed by a ','
or ';'.

(rillig)

indent: fix space after function name for option '-pcs'

(rillig)

indent: fix spacing for unknown type names in declarations

(rillig)

indent: extract probably_looking_at_definition to separate function

This heuristic guesses wrong in many cases and will some cleanups.

No functional change.

(rillig)

tests/indent: test '*' in function declarations

(rillig)

Consistently use "drvctl -l qemufwcfg0" to check if
running under qemu in general.

(hannken)

Reverte my previous changes kern_exec.c r1.512. It panics.

This changes was insufficient because es_emul is referenced by multiple execsw.

(ryo)

More / improved debug

(skrll)

Improve error handling.

Hypervisors can return a PMCR.N of 0.

(skrll)

indent: merge duplicate code for parsing 'struct s *'

No functional change.

(rillig)

indent: fix formatting of a few declarations involving unknown types

(rillig)

indent: fix comment for ps.in_decl

In C, there are no declaration statements. There are declarations and
statements, but no combination thereof.

(rillig)

indent: rename ps.in_stmt to in_stmt_or_decl

The previous name didn't match reality.

No functional change.

(rillig)

indent: rename ps.ind_stmt to in_stmt_cont

This makes a comment redundant.

No functional change.

(rillig)

indent: clean up style

No functional change.

(rillig)

Add support COMPAT_LINUX32 for aarch64

(ryo)

add support COMPAT_LINUX32 for aarch64

(ryo)

Extra arguments for AT_HWCAP and AT_PLATFORM are now passed when executing linux32 binaries.

AT_SYSINFO (vsyscall) feature seems to be broken, so leave it commented out.

(ryo)

The default name of the `exec_setup_stack' function for linux32 should be `linux32_exec_setup_stack',
and it should be defined in each arch.
declaration of linux32_exec_setup_stack() in linux32/amd64 has been removed because it does not exist.

NFC.

(ryo)

Fix anonymous memory object leak for sigcode.

- Repeating "modload compat_linux && /emul/linux/bin/ls && modunload compat_linux"
  will reproduce this problem.
- It cause in exec_sigcode_map(), anon-object for sigcode was created at
  first exec, but it remained even after exec_remove.
- Fixed that the anon-object for sigcode is created at exec_add(), and the
  anon-object reference is removed at exec_remove().
- sigobject_lock is no longer needed since it is locked by exec_lock.

(ryo)

add missing options EXEC_ELF32

(ryo)

- Use LINUX32_F_xxx instead of LINUX_F_xxx for compat_linux32.
  aarch64 and arm have different values.
- Compile linux32_sys_fadvise64() only if the syscall is required.

(ryo)

regen

(ryo)

add statx(2) syscall entry

(ryo)

- Add compat_linux statx(2) syscall.
- The AT_EMPTY_PATH processing from the modification of
  sys/compat/linux/common/linux_file64.c r1.63 has been separated, and made
  common to linux_statat(), so that it can be used not only by
  linux32_sys_fstatat64() but also by other *statat() variants.

(ryo)

rename bsd_to_linux_stat() to bsd_to_linux_stat64() to avoid confusion.
There is a static function with the same name in linux/common/linux_file.c.

NFC.

(ryo)

rename bsd_to_linux_stat() to bsd_to_linux_stat64() to avoid confusion.
There is a static function with the same name in linux/common/linux_file.c.

NFC.

(ryo)

Better counting for ierrors, iqdrops and noproto in ether_input().

- Use if_noproto for unknown or unsupported protocols.
- Use if_ierrors for wrong mbuf or oversized frame.

(msaitoh)

tests/indent: demonstrate several real-life formatting bugs

Collected by indenting usr.bin/make with the default profile.

The heuristic for distinguishing between type names and other
identifiers is way too primitive, it seems to have stopped evolving
somewhere before function prototypes were standardized in C90, at least
it handles function prototypes poorly.

indent-2014.09.04.04.06.07 added seemingly random spaces to
declarations, which were fixed in indent-2019.04.04.15.27.35, which came
5 years later. The latter commit introduced many new bugs, but it also
fixed this one.

The wrongly indented struct member declaration for HashTable has been
there since at least indent-2000.10.11.14.46.04.

The wrongly indented initializer expression was added in
indent-2019.04.04.15.27.35 (see above).

(rillig)

regen

(ryo)

add getrandom(2) syscall entry

(ryo)

add missing SUBDIR aarch64

(ryo)

Fix "restart" semantics -- restart is terminal, so don't clear the
condition when previous waiters have drained.  ("restart" is a bad
name for that this function does, fwiw.)

This should address a kernel assertion failure reported by Chavdar Ivanov
on current-users.

(thorpej)

Decorate the VAX signal trampoline with the appropriate .cfi
directives to allow exception unwind / backtrace across a signal
handler.

N.B. This is currently disabled, as these .cfi directives cause
linker warnings about incompatible TEXTREL relocations in .eh_frame.

(thorpej)

wsmoused: support absolute mouse position events

Tested with VirtualBox Guest Addtions.

(uwe)

arm64: acpi: Set capacity_dmips_mhz for CPUs

The GICC structure describes a relative power efficiency for each
processor. Use this value as-is for the capacity_dmips_mhz value of a
cpu. This makes the assumption that "more efficient" means "slower".

(jmcneill)

Decorate the SuperH signal trampoline with the appropriate .cfi
directives to allow exception unwind / backtrace across a signal
handler.

(thorpej)

Teach the LLVM-derived unwinder about the DWARF pseudo-registers defined
by GCC for SuperH for GBR, MACH, MACL, and SR.

(thorpej)

Temporary hack to work around incorrect/missing unwind info for
the t_sig_backtrace test case on ARM when compiled with optimization.

(thorpej)

Decorate the ARM signal trampoline with the appropriate .cfi
directives to allow exception unwind / backtrace across a signal
handler.

(thorpej)

Make LINUX_FIOCLEX and LINUX_FIONCLEX work

These ioctls were declared in linux_termios.h but were not actually
handled.

(pho)

Because the PC is not a general purpose register on aarch64, we need
to use DWARF pseudo-register for the signal trampoline return address.

Adjust the style to match other platforms.

(thorpej)

Don't waste time setting the dirty flag multiple times.

(blymn)

Ticket #1375

(martin)

Pull up following revision(s) (requested by mrg in ticket #1375):

usr.sbin/cpuctl/arch/i386.c: revision 1.123
sys/arch/x86/x86/cpu_topology.c: revision 1.20

decode SMT parts for AMD family >= 0x17, not just 0x17.

now zen3 systems are properly identified by cpu topology for the
scheduler and cpuctl identify.

(martin)

Revert unintended commit.

(thorpej)

Gah, fix two typos.

(thorpej)

Teach the LLVM-derived unwinder about the DWARF pseudo-register defined
by GCC for AArch64 to hold the return address where the signal trampoline
will resume.

(thorpej)

doc: new key generation method for cgdconfig(8)

(nia)

cgdconfig(8): add some console feedback when calculating the number
of pkcs5_pbkdf2 iterations

(nia)

undef a couple more.

(christos)

cgdconfig(8): Add an argon2id password-based key generation method

This provides an extra level of side-channel and cracking resistance
compared to the pre-existing pkcs5_pbkdf2/sha1 method used for
password-based disk encryption.

Several new keygen parameters are supported:

memory (integer, in kilobytes)
parallelism (integer, usually the number of CPU cores)
version (integer, usually 19...)

We do our best to calibrate these automatically when the paramsfile
is initially generated.

lgtm riastradh@

(nia)

crypt(3): document some of the memory units we're dealing with

(nia)

Fix typo. Pointed out by bch - thanks!

(martin)

Add LLDP and MACSec.

(msaitoh)

Modify comment:

s/Netbios/NetBIOS/
s/PPPOE/PPPoE/

(msaitoh)

s/repsonse/response/ in comment.

(msaitoh)

PR bin/53550

/bin/sh's processing of here doc expansions has changed.  Now it happens
in the context of the parent shell, so side effects are visible there,
just like all other redirection expansions.

We need to stop testing that that doesn't happen, and instead test
that it does.  This is that change.

Add another test case which is testing exactly the example from the PR
(well, with a different exit status, 1 is too generic and could happen
by accident) to make sure we don't reintroduce that bug sometime.

(kre)

PR bin/53550

Here we go again...  One more time to redo how here docs are
processed (it has been a few years since the last time!)

This is actually a relatively minor change, mostly to timimg
(to just when things happen).  Now here docs are expanded at the
same time the "filename" word in a redirect is expanded, rather than
later when the heredoc was being sent to its process.  This actually
makes things more consistent - but does break one of the ATF tests
which was testing that we were (effectively) internally inconsistent
in this area.

Not all shells agree on the context in which redirection expansions
should happen, some make any side effects visible to the parent shell
(the majority do) others do the redirection expansions in a subshell
so any side effcts are lost.  We used to have a foot in each camp,
with the majority for everything but here docs, and the minority for
here docs.  Now we're all the way with LBJ ... (or something like that).

(kre)