Handle EINTR, from RVP.

(christos)

Remove jemalloc and gdb hacks for alpha; root cause has been addressed.

(thorpej)

Remote -O0 hack for alpha; root cause has been addressed.

(thorpej)

Remove -O0 hack for alpha; root cause has been addressed.

(thorpej)

Ensure that the stack is always 16-byte aligned by rounding sizes as needed.

All changes from rin@ except swapcontext.S, which is from me (added
symbolic constants to make the code clearer).

(thorpej)

new acpica

(christos)

merge changes for acpica-20210604

(christos)

- Define STACK_ALIGNBYTES to override the default and ensure that
  stacks are 16-byte aligned, an assumption made by the compiler
  and recommended by the Alpha Architecture Handbook.
- cpu_lwp_fork(): Ensure 16-byte stack alignment if the caller specified
  one.

Addresses root casue of PR port-alpha/54307 and PR toolchain/56153.

Many thanks to rin@ for performing the root cause analysis and testing
changes.

(thorpej)

Disable kernfs on live images -- it is not required.

(jmcneill)

build fix

(jmcneill)

Use raw write(2) instead of fwrite(3) to avoid stream corruption because
of the progress bar interrupts. From RVP.

(christos)

Assume all ARM11[37]6 r0 variants are missing ISAR and friends.  This is
certainly the case for r0 variants I could find TRMs for.

PR/50596 CPU_ID_ARM1136JS r0p4 does not support feature registers

(skrll)

One more s/pte/ptr/

(skrll)

tests/lint: document wrong handling of GCC __attribute__

(rillig)

lint: rearrange and rename grammar rules for parameter declarations

Place the notype/type variants close to each other to be able to compare
them visually.

No functional change.

(rillig)

lint: move the grammar rule array_size out of the type/notype section

An array size is used in several grammar rules for different types of
declarations, therefore it doesn't make sense to place that rule
somewhere in the middle, where it disrupted the flow of notype/type
rules.  The whole point of having the notype/type rules grouped is to be
able to quickly compare them, since they are almost equal.

No functional change.

(rillig)

lint: reduce grammar conflicts for GCC attributes

In all but one case, the use of type_attribute_list introduced an
unnecessary ambiguity in the grammar.  It appeared in a place where it
could be repeated either by the type_attribute_list or by the enclosing
rule.  Both variants have the same effect.

No functional change.

(rillig)

lint: clean up style in grammar

No functional change.

(rillig)

lint: rename type generic_association_types to generic_association

The word 'types' was misleading and unnecessary.

No functional change.

(rillig)

Tickets #1312 - #1317

(martin)

Pull up following revision(s) - all via patch -
(requested by riastradh in ticket #1317):

sys/uvm/uvm_page.c: revision 1.248
sys/uvm/uvm_anon.c: revision 1.80
sys/rump/librump/rumpvfs/vm_vfs.c: revision 1.40
sys/rump/librump/rumpvfs/vm_vfs.c: revision 1.41
sys/rump/librump/rumpkern/vm.c: revision 1.191
sys/uvm/uvm_pager.c: revision 1.130
external/cddl/osnet/dist/uts/common/fs/zfs/zfs_vnops.c: revision 1.71
tests/rump/rumpkern/t_vm.c: revision 1.5
tests/rump/rumpkern/t_vm.c: revision 1.6
sys/rump/librump/rumpvfs/vm_vfs.c: revision 1.39

Move the handling of PG_PAGEOUT from uvm_aio_aiodone_pages() to
uvm_page_unbusy() so that all callers of uvm_page_unbusy() don't need to
handle this flag separately.  Split out the pages part of uvm_aio_aiodone()
into uvm_aio_aiodone_pages() in rump just like in the real kernel.

In ZFS functions that can fail to copy data between the ARC and VM pages,
use uvm_aio_aiodone_pages() rather than uvm_page_unbusy() so that we can
handle these "I/O" errors.  Fixes PR 55702.

fix an incorrect assertion in the previous commit.

Handle PG_PAGEOUT in uvm_anon_release() too.

Commit the ZFS file that I forgot in this previous commit:

Move the handling of PG_PAGEOUT from uvm_aio_aiodone_pages() to
uvm_page_unbusy() so that all callers of uvm_page_unbusy() don't need to
handle this flag separately.  Split out the pages part of uvm_aio_aiodone()
into uvm_aio_aiodone_pages() in rump just like in the real kernel.

In ZFS functions that can fail to copy data between the ARC and VM pages,
use uvm_aio_aiodone_pages() rather than uvm_page_unbusy() so that we can
handle these "I/O" errors.  Fixes PR 55702.
update the rump copy of uvm_page_unbusy() to match the real version,
in particular handle PG_PAGEOUT.  fixes a few atf tests.
the busypage test is buggy, expect it to fail.

make rump's uvm_aio_aiodone_pages() look more like the kernel version.
fixes some more rumpy assertions.

for the busypage test, replace atf_tc_expect_fail() with atf_tc_skip()
because atf apparently has no way to expect a test program to crash.
fixes PR 55945.

(martin)

Pull up following revision(s) (requested by riastradh in ticket #1316):

external/cddl/osnet/dist/uts/common/dtrace/dtrace.c: revision 1.41

Remove a pointless printf.

(martin)

Pull up following revision(s) (requested by riastradh in ticket #1315):

external/cddl/osnet/sys/kern/printf.c: revision 1.3

Use vpanic, not vprintf and then panic.

(martin)

Pull up following revision(s) (requested by skrll in ticket #1314):

common/lib/libc/arch/aarch64/atomic/atomic_nand_8.S: revision 1.3
common/lib/libc/arch/aarch64/atomic/atomic_nand_8.S: revision 1.4
common/lib/libc/arch/aarch64/atomic/atomic_nand_64.S: revision 1.3
common/lib/libc/arch/aarch64/atomic/atomic_nand_64.S: revision 1.4
common/lib/libc/arch/aarch64/atomic/atomic_nand_16.S: revision 1.3
common/lib/libc/arch/aarch64/atomic/atomic_nand_32.S: revision 1.3
common/lib/libc/arch/aarch64/atomic/atomic_nand_32.S: revision 1.4

Fix the logic operation for atomic_nand_{8,16,32,64}

From the gcc docs the operations are as follows
  { tmp = *ptr; *ptr = ~(tmp & value); return tmp; }  // nand
  { tmp = ~(*ptr & value); *ptr = tmp; return *ptr; }  // nand

yes, this is really rather strange.

typo in comment s/pte/ptr/

(martin)

Pull up following revision(s) (requested by skrll in ticket #1313):

common/lib/libc/arch/arm/atomic/atomic_xor_64.S: revision 1.5
common/lib/libc/arch/arm/atomic/atomic_nand_64.S: revision 1.5
common/lib/libc/arch/arm/atomic/atomic_or_64.S: revision 1.12
common/lib/libc/arch/arm/atomic/atomic_cas_8.S: revision 1.8
common/lib/libc/arch/arm/atomic/atomic_sub_64.S: revision 1.3
common/lib/libc/arch/arm/atomic/atomic_and_64.S: revision 1.11
common/lib/libc/arch/arm/atomic/atomic_swap_64.S: revision 1.13
common/lib/libc/arch/arm/atomic/atomic_add_64.S: revision 1.12

Whitespace

(martin)

Pull up following revision(s) (requested by macallan in ticket #1312):

sys/arch/hppa/hppa/intr.c: revision 1.4

fix off by one which resulted in all idle time reported as interrupt time
final fix from nick@

(martin)

Ticket #1318

(martin)

Pull up following revision(s) (requested by dholland in ticket #1318):

sys/miscfs/kernfs/kernfs_vnops.c: revision 1.169
sys/miscfs/kernfs/kernfs_vnops.c: revision 1.170

Add missing VOP_KQFILTER to kernfs.

Not sure if lack of it can be used for local DoS or not, but best to
fix.

-

Fix perms on /kern/{r,}rootdev.

(martin)

Fix perms on /kern/{r,}rootdev.

(dholland)

Add missing VOP_KQFILTER to kernfs.

Not sure if lack of it can be used for local DoS or not, but best to
fix.

(dholland)

Drop unicast packets that are not for us
when vlan(4) is not in promisc

(yamaguchi)

vlan: added NULL check for the parent interface

The pointer may set to NULL by vlan_unconfig
while packet processing

(yamaguchi)

vlan: added checks of linkstate

(yamaguchi)

vlan: set the link state to DOWN when its parent detaches

(yamaguchi)

whitespace

(dholland)

lint: refer to documentation about the 'notype' rules in the grammar

(rillig)

lint: reduce noise in storage_class_name

No functional change.

(rillig)

lint: rename message 340 to talk about "GCC extension", not "GNU"

(rillig)

lint: rename dinfo_t.d_stralign to d_sou_align_in_bits

When talking about alignment, offset and size of a type, the measurement
unit must be mentioned in the variable name, especially when it differs
from the standard unit of measurement, which is a byte, not a bit.

No functional change.

(rillig)

tests/lint: fix typos in comment in expr_range

(rillig)

lint: fix type warnings in emit1.c

The plain character strings in strg_t are saved as an array of unsigned
char.  When such a character is passed to ch_isdigit, it requires a
cast.  This is exactly the situation that ch_isdigit tried to avoid in
the first place.

No functional change.

(rillig)

tests/lint: fix check-expect.lua for empty .exp file

An absent .exp file is equivalent to an empty .exp file.  In neither of
these cases must the corresponding .c file declare any expected
diagnostics.

(rillig)

tests/lint: remove unused variables

(rillig)

sun4v: update status on S7 systems

(palle)

Instrument the number of calls to pmap_growkernel().

(thorpej)

Cleanup duplicated CS/80 indentify structures.  From OpenBSD.

https://marc.info/?l=openbsd-cvs&m=113227249626888&w=2
> Define the CS/80 identify structure only once and correctly, instead of
> duplicating it in every CS/80 driver and using an hardcoded number for
> its size.
> No functional change.

https://marc.info/?l=openbsd-cvs&m=113273001020159&w=2
> Pick HP-IB describe structures changes from main kernel code here as well.

(tsutsui)

Move attach messages from common rdident() to explicit rdattach().

(tsutsui)

Pull HP-IB probe fixes from OpenBSD/hp300.

https://marc.info/?l=openbsd-cvs&m=113217630426615&w=2
> Overhaul the way HP-IB devices are probed. We will now do an exhaustive
> probe of the (slave, punit) tuple space, since this is the only way we
> can get a dual disk or dual tape enclosure to attach two devices of the
> same kind.

This allows using multiple rd(4) disk images on the same slave emulated
by HPDisk (and probably the real 9122D with dual floppy disk drives).

Thanks to Miod Vallat for suggesting this fix.

(tsutsui)

Get rid of obsolete grfreg.h.  From OpenBSD.

(tsutsui)

Update gcc9-sh3-lint - instead of disabling optimization for lex.c on
sh3, rather revert the gcc change that (seemingly accidentally) broke
this.

(hgutch)

GCC git commit 91f66e78cc141da77ff9e0e3c8519e1af3f26c07[1] introduced
a regression in sh.  In addition to the intended change (based on the
commit message), an apparently unintended change was made, inverting a
comparison.  This broke sh builds and our workaround (so far) was to
compile xlint/lint1 with -O0.

Revert the comparison to what it was before and remove the -O0 hack
from xlint/lint1.

[1] https://gcc.gnu.org/git/?p=gcc.git;a=commitdiff;h=91f66e78cc141da77ff9e0e3c8519e1af3f26c07

(hgutch)

Fix a bug introduced in pmap.c,v 1.287 where, when creating the PTE
in pmap_enter(), we would erroneously disregard MOD/REF attributes
already present on the page, thus causing FOW/FOR to be set incorrectly.
Normally this is not a big problem (an extra page fault will be taken
to resolve it), but if you access the mapping in interrupt context
(such as during IDE PIO, for example), a KASSERT can fire due to
acquiring an rwlock in interrupt context while servicing that fault.

Reported and bisected-to-commit by rin@.

(thorpej)

typo in comment s/pte/ptr/

(skrll)

Account for partial writes when interrupted (from FreeBSD).

(christos)

Remove unnecessary #include <sys/malloc.h>

(thorpej)

Reduce code duplication when setting up the interrupt handler data
structures:
- alpha_shared_intr_alloc() no longer takes a "string length" argument,
  and just uses kmem_asprintf() to create an "irq %u" string by default.
  This is suitable for nearly every caller.
- Add a alpha_shared_intr_set_string() that allows callers to override
  the default IRQ description string.
- Related: make alpha_shared_intr_string() return a const char *, since
  no callers should need to modify the string directly now.
- Re-factor PCI shared interrupt structure allocation / initialization
  into a new alpha_pci_intr_alloc(), which is suitable for nearly every
  Alpha PCI platform.  Callers are expected to first have initialized
  the interrupt hardware to the quiescent state.

Adjust various call sites of above functions to account for changes,
even if they are not able to use the newly re-factored code.

(thorpej)

test/lint: demonstrate wrong warnings for 128-bit shifts

(rillig)

tests/lint: align tests for argument conversion to each other

(rillig)

lint: replace wrong assumption with citation from C99

No functional change.

(rillig)

lint: document where to fix missing const in lvalue-to-rvalue

No functional change.

(rillig)

lint: align messages for invalid left operand of '.' or '->'

(rillig)

lint: clean up new_tnode

No functional change.

(rillig)

libc: remove special handling for lint

In namespace.h 1.102 from 2005-07-16, the #ifndef was added to fix the
broken lint2 pass.  Somewhere between 2005 and now, lint has been
changed so that it does not need this workaround anymore.

After usr.bin/xlint/lint1/tree.c 1.303 from 2021-06-30, the #ifndef in
namespace.h caused the only warning about an implicit function
declaration in the whole NetBSD tree, in lib/libc/gen/sysconf.c:149,
where _getpagesize was not defined, but only getpagesize.

(rillig)

tests/lint: fix unintended character constant in test

I originally intended to type ASCII DEL as an example of a nonprintable
character code.  The actual character that landed in the code was
instead U+2303 "house", which looks similar to the typical
representation of the DEL character.  The UTF-8 byte sequence for that
code point started with \xE2, or in octal \342.

(rillig)

tests/lint: document why lint does not need to detect wrong types

(rillig)

lint: improve wording for incomplete struct or union

(rillig)

lint: remove remaining support for lvalue casts

These had been GCC extensions until GCC 3.4, they were removed in GCC
4.0.

(rillig)

LLVM RT library will actually cross-build from OpenBSD provided sufficient
limits are in place for the build user.

(cjep)

lint: remove support for obsolete GCC-style lvalue casts

Since GCC 4.0, a cast results in an rvalue.

(rillig)

Add VOP_PARSEPATH to zfs control dir vnode op table.

(hannken)

Fix union_parsepath(), either the upper or the lower dvp may be NULL.

(hannken)

lint: in strict bool mode, allow mixed types in generated C code

This allows flex lexers to be run through lint in strict bool mode.

(rillig)

tests/lint: re-enable test c99_bool_strict_suppressed

It had been disabled accidentally in the previous commit.

(rillig)

lint: include scan.c in 'make lint'

This theoretically enables strict bool mode for the few remaining code
in scan.l.  Since scan.l is not yet detected as generated code, all
interesting errors have to be suppressed though.

(rillig)

lint: remove outdated assertion

Since err.c 1.12 from 2000-07-06, lint allows to suppress individual
error messages.  Suppressed error messages do not increment nerr.
Keeping nerr at 0 had triggered the assertion.

(rillig)

tests/lint: add test for suppressing errors in strict bool mode

(rillig)

lint: in strict bool mode, continue after error message

If a controlling expression is not of type bool but of any other scalar
type, keep the expression.  Its value is still useful for control flow
analysis.

This prevents an assertion failure when running lint on the generated
scan.c, which contains a "while (1)" that does not stem from a system
header.  If it did, lint would accept it, see tn_from_system_header. But
"scan.c" is not considered a system header.  Maybe lint's definition of
a system header needs to be revisited.

After fixing this, there is another assertion failure though, so scan.c
is not yet ready to be inspected by lint.

(rillig)

elide cast warning: error: cast from pointer to integer of different size

(christos)

Fix the logic operation for atomic_nand_{8,16,32,64}

From the gcc docs the operations are as follows

{ tmp = *ptr; *ptr = ~(tmp & value); return tmp; }  // nand
{ tmp = ~(*ptr & value); *ptr = tmp; return *ptr; }  // nand

yes, this is really rather strange.

(skrll)

lint: sync stdout and stderr when compiled in debug mode

If lint is compiled with -DDEBUG, its debug output goes to stdout, no
matter whether the option -d is given or not.

(rillig)

Do not assume /tmp is safe to use if TMPDIR is set

(sjg)

tests/make: make test job-output-null more portable

Previously, the test dependend on implementation details of the system's
printf command.

Thank you sjg for the detailed analysis on macOS, FreeBSD and Linux.

(rillig)

lint: do not define 'long double' GCC builtins for traditional C

(rillig)

lint: guard against segmentation fault when parsing struct

Found by afl.  The code that triggers this case is:

struct{const}

(rillig)

lint: fix segmentation fault on malformed preprocessor output

# 0 ""
# 0 ""2

Found using afl, does not happen in practice though since every C
preprocessor generates balanced 'enter' (1) and 'leave' (2) events.

(rillig)

lint: make stdout unbuffered in debug mode

Some debug log messages are on stdout, others on stderr.  Make sure that
they occur in the correct order, even when the output is redirected.

(rillig)

lint: fix assertion failure in display_expression for bool constants

(rillig)

sun4v: request hypervisor api major version 3 for the interrupt group - should solve issues when setting up interrupts on S7 systems

(palle)

tests/lint: fix test d_cast_init on macppc

On macppc, char == unsigned, which generated the following unintended
warnings:

d_cast_init.c(18): warning: initialization of unsigned with negative
    constant [221]
d_cast_init.c(18): warning: initialization of unsigned with negative
    constant [221]
d_cast_init.c(18): warning: initialization of unsigned with negative
    constant [221]
d_cast_init.c(18): warning: initialization of unsigned with negative
    constant [221]

(rillig)

tests/lint: fix test for character comparison on macppc

On macppc, char == unsigned char, which generates one more warning than
on platforms where char == signed char.

(rillig)

sun4v: add support for hypervisor soft state - from OpenBSD

(palle)

Treat failed chdir/chroot for guest and chroot accounts as fatal.
Also treat failed set{e,}(u,g}id calls as fatal.
Addresses CVE-2020-7468, via FreeBSD.

(christos)

PR/56246: Yuichiro Naito: gprof(1) never show call graphs when profiling
multi-threaded application

(christos)

PR/56250: Michael Forney: mkstemp/mkdtemp not declared with
_POSIX_C_SOURCE=200809L

(christos)

Ticket #1311

(martin)

Pull up following revision(s) (requested by thorpej in ticket #1311):

sys/dev/ic/dp8390.c: revision 1.99
sys/dev/ic/dl10019.c: revision 1.17
sys/dev/ic/ax88190.c: revision 1.18

Make sure the media / mii members in struct ethercom are initialized
so that the media-related ioctls work.  Problem reported by Bj�rn Johannesson
on current-users@.

XXX pullup-9

(martin)

Add missing include of compiler.h and make all declarations of
IOPortBase consistent.

(martin)

Return error from fd_dupopen.

(mlelstv)

itesio(4): Added IT8625E support.

(nonaka)

tests/lint: encode the array length in the diagnostic

(rillig)

lint: fix initialization of array of unknown size

The size of the resulting array was computed wrong.

(rillig)

tests/lint: add test for array subscripts in C99 initialization

(rillig)

lint: in strict bool mode, make (flag & FLAG) simpler

This is a tiny change in an edge case that does not occur in practice,
which is that the left-hand side of the '&' is explicitly cast to an
enum type.  The apparent "loss of information" from the deleted comment
has already been explained in the previous commit.

(rillig)

use PORT_SIZE from compiler.h instead of int.

(christos)

lint: no special check for unary operators in strict bool mode

All interesting constellations regarding unary operators are already
covered by the simple model from ops.def.

No functional change.

(rillig)

lint: document lint1_type.t_is_enum

Enum types are subject to implicit conversions, as opposed to struct,
union, pointer and function types.

(rillig)

Add VOP_PARSEPATH to zfs's vnode table. "oops"

zfs not being under src/sys strikes again :-(

(dholland)

Regen.

(dholland)

Turn off fstrans for VOP_PARSEPATH. Fixes t_umountstress deadlock.

Diagnosis by hannken@.

(dholland)

port-macppc/56289: kernel spinout on macppc when runing ATF tests with LOCKDEBUG

Move call to snapper_set_rate from trigger_output to commit_settings,
since the intr lock is not held when calling the latter.

(jmcneill)

Be consistent about #ifndef ARM32_DISABLE_ALIGNMENT_FAULTS.  NFCI.

(skrll)

Added tests for changing a MTU when the vlan(4) is added to bridge(4)

The tests is for PR kern/56292

(yamaguchi)

Use if_ioctl() for changing MTU, not ether_ioctl to prevent panic

Fix PR kern/56292

(yamaguchi)

fix off by one which resulted in all idle time reported as interrupt time
final fix from nick@

(macallan)

Back out fix for kern_pmf.c calling a null if_stop and apply a fix
suggested by Jared McNeill which sets if_stop to a stub function
which means that more than just the pmf is protected from the NULL call.

(blymn)

Make sure the media / mii members in struct ethercom are initialized
so that the media-related ioctls work.  Problem reported by Bj旦rn Johannesson
on current-users@.

XXX pullup-9

(thorpej)

AWS marketplace does not allow root ssh logins. Create an ec2-user account
and install the ssh key in that user's home directory instead.

(jmcneill)

port-evbarm/56274: no network on ec2 arm64 9.99.85

Add -w to dhcpcd_flags when running on EC2, since we need to wait for the
network to come up before contacting the metadata service.

(jmcneill)

No need for ec2_init on arm64mbr

(jmcneill)

port-evbarm/56274: no network on ec2 arm64 9.99.85

Remove custom SIOCSIFFLAGS handling and rely on ether_ioctl to DTRT

(jmcneill)

gcc (with some options) eroneously claims we would use "vp" uninitialized,
so initialize it as NULL.

(martin)

in uvm_mapent_forkzero(), if the old entry was an object mapping,
appease a debug check by setting the new entry offset to zero along with
setting the new entry object pointer to NULL.

Reported-by: syzbot+de8e4b223a3838c7307b@syzkaller.appspotmail.com
Reported-by: syzbot+efaea991addfdcc5abd4@syzkaller.appspotmail.com
Reported-by: syzbot+15d1e19dff9209c2e40b@syzkaller.appspotmail.com

(chs)

don't clear the error before we use it to determine if we are moving or duping.

(christos)

Fix hardcoded size for total_patckets array in synaptics_softc, it
needs to be SYN_MAX_FINGERS because the driver expects it to be.

(blymn)

Prevent kernel panic on sleep if network interface driver does not
have if_stop defined.  This is a common problem with usb adaptors.
This is a temporary fix, the printf needs to be converted to an
assert once the drivers are cleaned up.

(blymn)

Do previous differtly by pushing two registers in the same way as the
_INVOKE_CERROR macro

(skrll)

Do previous differently by pushing even number of registers and remove
strange r7 usage.

(skrll)

Fix a couple of problems with MII-equipped NE2000 derivatives reported by
Bj旦rn Johannesson on current-users@:

- Re-factor dp8390_stop() into dp8390_halt() (that does the stuff to
  halt the hardware) and dp8390_stop() (which alls dp8390_halt() before
  calling mii_down() via sc->sc_stop_card()).  This prevents us from
  calling mii_down() before all of the interface data structures have
  been set up, which these days can trip a KASSERT().

- Add a 1-second timer to call mii_tick(), and enable it in the
  sc->sc_init_card() callback, and cancel it in the sc->sc_stop_card()
  and sc->sc_media_fini() callbacks.  This is actually a long-standing
  bug that previously didn't have much practical effect, but causes
  problems with dhcpcd's link live-ness detection logic.

(thorpej)

Improve Christos's vn_open fix.

- assert about api misuse up front (suggested by riastradh)
- restore the behavior of returning EOPNOTSUPP if ret_fd is NULL and we
  get a fd back (otherwise things like ktruss -o /dev/stderr panic)
- clear error to 0 for the EDUPFD and EMOVEFD cases so opening a
  cloner succeeds

(dholland)

devlist2h.awk: Nix trailing whitespace.

No functional change intended.

(riastradh)

No need to bother to use aprint_debug(9) inside #ifdef DEBUG block.

(tsutsui)

lint: do not warn about use of implicitly declared GCC builtins

(rillig)

lint: in C99 mode, only warn about implicit function declarations

Since tree.c 1.294 from 2021-06-28 (two days ago), lint errored out on
an implicit function declaration.  In principle it is correct to do so
since C99 requires it, but in practice there are a several functions
that are not declared in the translation unit itself since they are
provided by the compiler.  Typical examples for GCC and Clang are the
various functions named '__builtin_*' and '__atomic_*'.

For now, only warn about these but don't error out.

(rillig)

lint: mention the name of an implicitly declared function

In the regular NetBSD builds, this happened in swab.c:65.  That line
contains __predict_false, which may or may not be a macro.  In other
cases, there may be more than one function call in a single line.

(rillig)

tests/lint: extend test for message 215 about implicit function

(rillig)

lint: fix type name in message 101 for wrong member name

(rillig)

tests/lint: extend tests for message 101

This demonstrates the wrong type name in the message.

(rillig)

tests/lint: extend and explain test for implicit function declaration

(rillig)

lint: improve comment about INT keeping t_is_enum

(rillig)

tests/lint: add test for type name of enum converted to int

(rillig)

lint: invert condition in control flow analysis of switch statements

The condition previously contained many negations, which was hard to
understand.

No functional change.

(rillig)

PR/56286: Martin Husemann: Fix NULL deref on kmod load.
- No need to set ret_domove and ret_fd in the regular case, they are meaningless
- KASSERT instead of setting errno and then doing the NULL deref.

(christos)

lint: fix bit-size of long double _Complex in portable mode

C99 6.2.5p13 says that LCOMPLEX has the same representation and
alignment requirements as an array type containing exactly two LDOUBLE.

When support for _Complex was added to lint in inittyp.c 1.10 from
2008-09-27, there was no explanation for making the bit-size of LCOMPLEX
different from what C99 says, so it was probably a mistake that went
unnoticed for more than 12 years.  It's an edge case anyway.

(rillig)

lint: inline 'sign' into convert_integer

No functional change.

(rillig)

The _INVOKE_CERROR macro deals with thumb so simplify the code (at the
expense of a couple more instructions).

(skrll)

Trailing whitespace

(skrll)

lagg: fix an uninitialize variable

pointed out by tnn@n.o., thanks.

(yamaguchi)

Revert previous. Apparently it's wrong.

(skrll)

Fix previous. For Thumb-1:
- sp cannot be manipulated directly
- {add,sub}s should be used instead of {add,sub}

(rin)

Align sp to 8-byte boundary as required by EABI.

IIUC, this change only affects libc compiled for ``Thumb-mode userland'',
which we've not officially supported yet.

(rin)

Align sp to 8-byte boundary as required by EABI.

This is especially important for non-leaf functions; GCC optimizes codes
based on assumption that sp is aligned properly.

Mostly fix broken earmv5 userland compiled by GCC10 due to alignment
faults in ld.elf_so, where {ld,st}rd are used for [sp, #8x].

No regression for ATF is observed for earmv[67]{,hf}{,eb}.

(rin)