Stop using -fcommon, xsrc should be fixed.

(joerg)

Avoid multiple definitions of the same variable.

(joerg)

PR toolchain/55837

Stop using enum for flags, as per request from joerg.

#define constants and #undef after use.

(rin)

Additional comment about the range (0-1025) that's needed for the PT page
reference count, and thus now many bits we need to scare up.

(thorpej)

After a comment by joerg@, go back to using a dedicated field for the
PT page reference count, but add an XXX comment stating the desire to
find a safely-unused field in the vm_page structure when pages are in-
use as PT pages, so that we can save the 8 bytes per page needed for
this.

(thorpej)

Include "opt_param.h" (ifdef _KERNEL_OPT) everywhere that MSGBUFSIZE is
referenced since some sources include <machine/param.h>.

(simonb)

Use <sys/param.h> in the synopsis rather than <machine/param.h>; an MI
interface should be pulled in via <sys/...>.

(simonb)

PR toolchain/55837

Bump LAST_REGISTER and LAST_RESTORE_REG to REGNO_ARM32_S31 for arm.

There are two numbering schemes for VFPv2 registers: s0-s31 and d0-d15.
The former is used by GCC, and the latter is by LLVM. Since libunwind was
derived from LLVM, it has never supported the former. This results in
crashes for GCC-compiled binaries in exception handler of C++, if it
encounters VFPv2 registers when unwinding frames.

This commit adds support for s0-s31 numbering to libunwind. I choose an
implementation in which VFPv2 registers are ``double-counted'' as s0-s31
AND d0-d15. This does not cause real problems, since the former is only
used by GCC, and the later is by LLVM. That is, different numbering
schemes cannot appear in a same frame. To make sure, assertions are added
in order to check this.

I've confirmed that no regression for ATF both for GCC- and LLVM-compiled
userlands.

(rin)

PR toolchain/55837

Fix logic error in copyFloatVectorRegister() for arm; copy s0-s31 or
d0-d31, not both.

(rin)

PR toolchain/55837

Fix pointer arithmetic when copying s0-s31 registers for arm.

(rin)

PR toolchain/55837

Fix DWARF/internal register numbers of s31 for arm.

(rin)

PR toolchain/55837

Fix for jumpto() armeb; use word-wise load for flags, instead of byte-wise one.

(rin)

PR toolchain/55837

copyFloatVectorRegister(): Assert register number is valid to make sure.

(rin)

PR toolchain/55837

Misc style fixes for clarity:

- Rename lazyVFP1() and lazyVFP3() to lazyVFPv2() and lazyVFPv3(),
  respectively. Note that VFPv1 was obsoleted and replaced by VFPv2.

- Introduce enum for flags.

- Add few comments.

No functional changes.

(rin)

Generalize boot.cfg workaround for ISO9660; fallback to default_device
whenever efi_file_path() fails (due to broken firmware), in addition to
the case of ISO9660 (for which efi_file_path() succeeds but does not
work correctly).

(rin)

uvm: Make uvm_extern.h (more) self-contained, needs sys/types.h.

(riastradh)

drm: sys/param.h, not machine/param.h, for DELAY.

(riastradh)

bump slightly harder

(martin)

MIPS n64 dynamic binaries have worked for a while now, remove -static
for these.

(simonb)

Use -fcommon

(joerg)

Various X drivers still need -fcommon, so add it.

(joerg)

Minimal change to suppress unaligned access errors on NetBSD/alpha; add
the UNALIGNED decoration to icmp6 related structures.

PR bin/56199.

(thorpej)

make: inline str_concat4

This function is only ever used for forming strings of the form
"archive(member)".

No functional change.

(rillig)

make: fix reported code coverage for *.h

In 'make test-coverage', the number of uncovered lines for inline
functions in headers was reported too high.  The cause for this is that
gcov reports the coverage for these functions multiple times, once per
translation unit.  If some of the translation units don't use these
inline functions, summing the lines containing '#####' quickly leads to
numbers that are obviously too high.

(rillig)

Minimal change to suppress unaligned access errors on NetBSD/alpha; add
the UNALIGNED decoration to struct icmp.

PR bin/56199.

(thorpej)

make: uncompress code in ApplyModifier_Unique

The algorithm is easier to understand when each line of code only
focuses on a single topic.

No change to the resulting binary, except for line numbers in assertion
messages.

(rillig)

make: inline VarUniq into ApplyModifier_Unique

No functional change.

(rillig)

tests/make: test whitespace normalization for modifier ':u'

(rillig)

Fix DEBUG build.

(thorpej)

Fix unused variable warning when DIAGNOSTIC is disabled.  Pointed out
by jklos.

(thorpej)

Implement pmap_remove_all().

(thorpej)

avoid dereferencing a constant string address as a UINT32 pointer,
KUBSAN complains about bad alignment.

Reported-by: syzbot+91187f4e33436ce83e56@syzkaller.appspotmail.com

(chs)

When removing mappings, hang PV entries to be freed off of the
pmap_tlb_context structure, and free them back in bulk after we
release all of our locks (as we do with PT pages that are freed).

(thorpej)

Keep track of a pmap's PV entries with a list hanging off the pmap.

(thorpej)

synaptics(4): New sysctl knob for debug output.

Set hw.synaptics.debug=1 to re-enable it; the compile-time DIAGNOSTIC
option was inappropriate for this.

(riastradh)

thinkpad(4): Fix evaluation of MHKA on version 2 devices.

Need to pass an argument.

(riastradh)

ld(4): Block requests while suspended until resumed.

Otherwise nothing stops us from continuing to feed I/O to the disk
controller when it expects that the queues are quiesced as it pokes
registers to change its power states.  Fixes resume during disk
activity on my T480 with nvme.

(riastradh)

PR port-amiga/56188

Include a prototype for __nbcompat_heapsort. The tools for amiga now
build on Darwin where one Amiga specific tool uses heapsort.

Despite being a one-liner, ./build.sh -m amiga tools was tested on
NetBSD, Darwin, FreeBSD & Linux with the change.

Discussed various options with jdc & thorpej. Perhaps in the future
we could look at using qsort instead of heapsort as qsort is
standardised and more likely to be available. We could then remove
heapsort from the compatibility library.

(cjep)

Bump image size slightly (for gcc 10)

(martin)

Include opt_param.h for MSGBUFSIZE ifdef _KERNEL_OPT.

(rin)

signal.7: minor clarification. Bump date (to when I wrote the change)

(dholland)

typo in comment

(dholland)

typo in comment

(dholland)

Store the modified / referenced attribute bits in the lower 2 bits of
the PV entry list pointer in struct vm_page_md.  This reduces the size
of that structure from 16 bytes to 8, and will go a fair way to making
up for increasing the size of struct pv_entry in a future commit.

(thorpej)

Add "root" command to pass a root specification.

(mlelstv)

Add "root" command to provide a BTINFO_ROOTDEVICE parameter.

(mlelstv)

Fix conversion between aarch64 and aarch32 fpreg's; in aarch32 mode,
d0-d31 are packed into v0-v15 (== q0-q15).

This fixes crashes in VFP-optimized codes running on COMPAT_NETBSD32.

OK ryo

(rin)

Define a macro, VM_MDPAGE_PVS(), for fetching the first pv entry
for a page.

(thorpej)

Pages that are in-use as page table pages should never be part of a
UVM loan transaction, so use the vm_page::loan_count field as the PT
page reference count.

(thorpej)

Remove old version of libc++, migrated to external/apache2/dist/libcxx
to match the rest of the mono-repo imports.

(joerg)

Remove files no longer used by 249b40b558955afe5ac2b549edcf2d7f859c8cc9

(joerg)

Note that compat_60 modules on i386 and amd64 once again include the
microcode-update functionality.

(pgoyette)

Drop a number of entries that no longer apply.

(joerg)

Don't use V8 atomic instruction for AA32 mode.

(joerg)

address-of-packed-member warning is no longer a problem here.

(joerg)

Drop explicit -msecure-plt, it's the default for both clang and gcc.

(joerg)

Don't use the return address hack with clang.

(joerg)

Don't warn about frame pointer use for clang either.

(joerg)

Always use C++17 string_view, the emulation is just broken.

(joerg)

Ignore pointer sign cast issues with clang.

(joerg)

Ignore pointer cast warnings for clang, too. Use -Wno-format more
selectively.

(joerg)

Update LLVM build system for 249b40b558955afe5ac2b549edcf2d7f859c8cc9

This enables the use of modules for a significant build performance gain
when building with clang as host compiler or when using HAVE_LLVM=yes.
Switch libc++ to using the copy from the mono-repo.

(joerg)

Support LLVM 13, mostly from upstream.

(joerg)

Track the PT pages allocated to a pmap in the pmap itself.

(thorpej)

Change pmap_physpage_alloc() to return the vm_page * so that the
callers have access to it.

(thorpej)

Define a pmap_pagelist LIST_HEAD and use it where we used ad hoc LIST_HEADs
of vm_page structures.  Define and use a generic routine to free such a list
back to UVM.

In pmap_remove_internal(), KASSERT that no PT pages are queued up to be
freed when removing mappings from the kernel pmap.

(thorpej)

Cast to enums via uintptr_t to avoid compiler warnings.

(joerg)

Unindent code to match the logical structure, ignoring the commented out if.

This code should be rewritten to use BUS_DMA_WAITOK, but that's a
project of its own.

(joerg)

Merge 47661d00cd4d6cd728ae31b0bb29a49a6c06272a

(joerg)

Handle recent ifconfig(8) outputs in the miniroot installation script.

- Remove netmask slash notation for IP addresses,
  which has been changed between NetBSD 7.x and 8.0:
  http://cvsweb.netbsd.org/bsdweb.cgi/src/sbin/ifconfig/af_inet.c#rev1.24
- Ignore inet6 entries, which miniroot scripts don't support

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

(tsutsui)

Move the pointer to the pmap's lev1map out of the shared pmap structure
and into each CPU's pmap_percpu area.  This pointer is constant for the
life of the pmap, and moving it gives us an additional 8 bytes in the
shared pmap structure.

Because the kernel pmap does not have per-CPU data, all places where we
need the kernel pmap's lev1map now just reference kernel_lev1map directly.

(thorpej)

Reduce diffs from GENERIC for maintainability.

No functional change intended.

(tsutsui)

Reorder entries and reorganize comments for maintainability.

No functional change intended.

(tsutsui)

Rather than tracking "needs I-sync on return to userspace" in a bitmap,
track it with a separate field in the pmap_percpu.  Not only does this
reduce cache line contention ever so slightly, it also frees up a field
in the shared portion of the pmap structure.

(thorpej)

Rename pmap_asn_info to pmap_percpu, and pmap::pm_asni to pmap::pm_percpu.
No functional change.

(thorpej)

Reduce maxusers to 8 as GENERIC.

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

(tsutsui)

Add missed 'nhpib at intio' for internal HP-IB.  Found on testing HPDisk.

Also fix comments for HP-IB devices.
Should be pulled up to netbsd-8 and netbsd-9.

(tsutsui)

Document netbsd-GENERIC.QEMU.gz

(thorpej)

Provide a binary/kernel/netbsd-GENERIC.QEMU.gz in the Alpha release
hierarchy.  This is identical to netbsd-GENERIC, except it has been
dbsym'd so that the kernel symbols are available when running under
Qemu (which loads the kernel directly, but does not load the symbol
table).

(thorpej)

first check, then copy

(christos)

Add birthtime support and make time 64 bit.

(christos)

Add some more Thinkpad hotkeys.

(riastradh)

thinkpad(4): Add dtrace probe for hotkey events.

(riastradh)

thinkpad(4): Attach at LEN0268 with v2 Thinkpad hotkeys too.

(riastradh)

xhci(4): Wait USB_RESUME_WAIT ms, not 20 ms.

Better to use the named constant, and although the spec says 20 ms is
enough, apparently for some devices it's not.

(riastradh)

use libera.chat-hosted kiwiirc

(maya)

Update the FPU register names and bit definitions to something somewhat
modern (MIPS32/MIPS64) and convert to __BIT/__BITS.

(simonb)

Switch mac68k to Xorg server 1.20. Only available driver is wsfb(4) both
for 1.10 and 1.20, and there is no reason to stay with 1.10.

XXX
1.20 works fine on WSFB* kernels with genfb(4), whereas both 1.10 and
1.20 fail for GENERIC* kernels with macfb(4) as some mandatory ioctl's
are missing. It would be better to make genfb(4) default before netbsd-10
is branched.

(rin)

Make wskbd(4) default for mac68k; no other protocol is available.

(rin)

Put the MSGBUFSIZE option in opt_param.h so that changes of its setting
in kernel config files will be picked up automatically without needing
a "make clean".  Added to opt_param.h since ports define their default
MSGBUFSIZE settings in their <arch/include/param.h> (or equivalent).

(simonb)

nvme(4): Move disestablishment of admin q interrupt to nvme_detach.

Nothing re-established this after suspend/resume, so attempting
suspend/resume/suspend would crash, and presumably we would miss
interrupts after resume.  This keeps the establish/disestablish more
symmetric in attach/detach.

(riastradh)

Appease compiler warnings derived from ancient HOST files.  PR/56187

Also fix misc styles and formats per KNF.

(tsutsui)

dwiic(4): Attribute output correctly and relegate to debug-level.

Tidy up a little while here.

(riastradh)

aiomixer: remove workaround for libcurses bug

(nia)

wresize: don't bound pads to the size of the screen

allows avoiding a workaround in aiomixer,

ok blymn uwe

(nia)

nvme(4): Add suspend/resume, derived from OpenBSD.

(riastradh)

i915drmkms: Fix LOCKDEBUG panic and potential deadlock.

This path is taken with a spin lock held, and possibly even in
interrupt context, where taking vmobjlock is not kosher, but we are
guaranteed to have the queue populated and unchanging.

XXX pullup-9

(riastradh)

tpm(4): Handle TPM 2.0 random source too, and loop on short reads.

Tested on ThinkPad T480.

(riastradh)

ualea(4): Null suspend/resume handler.

(riastradh)

Deal with the pmap limitation of maxproc in a more complete way and
recognise CPUs with only 8bit ASIDs.

(skrll)

Sort includes. NFCI.

(skrll)

report to the scheduler if we have different speed cpus.

(mrg)

PR 55413: 64-bit and OK police; grammar improvements.
From Michael Siegel.

(martin)

Add ixg* (commented out). Tested with rin@'s AlphaServer DS10.

(msaitoh)

G/C

(skrll)

I don't see any code I wrote left in here.

(dholland)

Restore return value of main, arbitrarily thrown away in -r1.4.

(dholland)

Update R6000 description.

(simonb)

The Tsunami / Typhoon chipsets have a static "monster window" for DMA
that allows for up to 32GB of RAM to be direct-mapped if the PCI device
can issue a 64-bit address (the monster window lives at 0x10000000000).
Enable this window and provide this to the PCI bus as a "dmat64".

(thorpej)

Define the arithmetic exception summary bits passed in a0 to entArith.

(thorpej)

kre@ does not like the subshell :-)

(christos)

mips has switched already.

(christos)

mips64 only works with gcc-10

(christos)

Don't nest comments.

(joerg)

Don't define MaxClip and junk twice, fixing build with -fno-common.

(joerg)

switch to MACHINE_ARCH i386 to include xen (thanks mlelstv)

(christos)

Switch "xen" to gcc 10 too, otherwise xen kernel builds fail because they
"think" we have gcc-9 where we have gcc-10 in reality (since MACHINE=xen),
and we don't add ${GCC_NO_RETURN_LOCAL_ADDR} COPTS.in_pcb.c

(christos)

Consistency and typo/grammar fixes from Michael Siegel in PR 55413,
With minor adjustments.

(martin)

Rename the unhelpfully named mips_emul_lwc0() and mips_emul_swc0() to
mips_emul_ll() and mips_emul_sc(); make these static to mips_emul.c.

(simonb)

Simplify; no need to special case the small buffer zero src_len.
lla_snprintf1 never returns -1.

(christos)

Don't use the stack, print to the buffer directly (this was one of the
biggest stack users).

(christos)

Print the CP0 status register too in the debug trap code.

(simonb)

remove dup line (thanks rillig)

(christos)

Fix error handling botch in 1.138. Thanks to jmcneill@ for spotting it.

(skrll)

Honour USBMALLOC_ZERO when reusing blocks from the freelist.

From sc.dying on tech-kern

(jmcneill)

Remove nfs_putpages() prototype; it's not defined anywhere.

(simonb)

Switch arm64.img to big-endian for aarch64eb, as libsa/efiboot support
bi-endian FFS/disklabel now.

Many thanks to mrg@ for working on this!!

(rin)

If we are building date earlier, we need to use an uninstalled copy
of host-mkdep and libnbcompat otherwise the build process looks for
tools that might not be there yet (e.g. fresh build).
Reviewed by martin.

(cjep)

sparc raid boot was 2006 (i had it right originally.)  noted by jdc.

(mrg)

Use \(em. Remove superfluous Pp.

(wiz)

Avoid using Pq.

In this case, the parentheses were wrong when a line was extended.

(wiz)

updated latest date for GCC.

(mrg)

note recent raid and GCC updates.

(mrg)

switch mips* and i386 to GCC 10.

arm32 (significant issues remaining), m68k (seems fine, but x68k
needs to reduce support in loadbsd), and sh3 remaining.

(mrg)

add bi-endian support to the libsa ufs reader and enable it in efiboot.

ffs frontends to "ufs.c" now also define ufs_dinode_swap, ufs_indp_swap,
and FS_MAGIC (moved from ufs.c #if segments.)  these are used to call
the right (32/64 bit) ffsv1/v2 version.

ufs.c 'struct file' gains f_swapped member.  accessors for d_magic,
d_reclen, and d_ino are introduced (they need to be swapped.)  sfter
reading an inode from disk, read_inode() may call ufs_dinode_swap().
indirect block number and caches may be swapped.

error handling in ffs_find_superblock() is cleaned up.  (size is slightly
reduced on some ports with this part.)

defaults for new defines added to ufs.c.  (XXX: we build ufs.c but i think
all the consumers don't use it, and we can stop building it.)

LFS support is not included.

add a cut-down copy of ffs_bswap.c from the kernel.

also enable bi-endian disklabel support in efiboot.

most ports build and sizes compared for platforms that don't enable this
code and all but one saw reduced code size.  booted several platforms with
new boot code.

(mrg)

move the basic history i wrote from raidctl(8) to raid(4), and
fix the x86 (thanks simonb) and sparc bootable dates, a couple
of spelling errors.

(mrg)

In gcc10, -msign-return-address is no longer supported.
Instead, (LLVM-compatible) -mbranch-protection option is supported.

(ryo)

Do no_jmbuf++ when ixgbe_getjcl() failed in ixgbe_setup_receive_ring(), too.

(msaitoh)

fix build error with options ARMV85_BTI

(ryo)

Restore the pnozz Xorg driver, now that it has EXA support.

(jdc)

Restore the build of the pnozz driver on sparc, now that it has EXA support.

(jdc)

Update the pnozz files list for EXA support.

(jdc)

Add support for EXA, so that the pnozz driver works again with Xorg.
Note, that only minimal acceleration is currently supported.

Thanks to macallan for the pointers to get this working.

(jdc)

Move the send queue checking to a new function, and also call this
in the rx interrupt path.  Measureable improvement on a NFS "create
small files" test.

(simonb)

Schedule the send cleanup function for next tick in cnmac_start().  In
the send cleanup function, schedule for the next tick instead of waiting
for HZ ticks if there are still send requests outstanding.  Greatly
increases NFS throughput, perhaps other types of network traffic.

Use callout_setfunc() after callout_init() instead of callout_reset().

(simonb)

new isc-dhcp

(christos)

merge local changes between dhcp-4.4.2 and dhcp-4.4.2-P1

(christos)

xhci: Fix logic in waiting for command queue access.

_Either_ an existing command in progress, _or_ an existing suspend in
progress that is not done by us, should block us; the logic I wrote
previously erroneously blocked only if both conditions happened at
the same time.

Should fix issue reported by Andrius V in the PR kern/56050 followup
discussion.

(riastradh)

suggest libera.chat instead

(maya)

handle different flavors of date(1)

(christos)

use the date tool

(christos)

Use the date tool

(christos)

Add date as a tool

(christos)

tool changes

(christos)

Add a date tool

(christos)

Fix ip_nat memory leak and use-after-free, wrong element freed (Cy Schubert)
https://cgit.freebsd.org/src/commit/?id=323a4e2c4e285e6f8eee8db3fe2cb74

(christos)

add basic raidframe support to efiboot.

if raid disklabel or gpt is found, add this partition with
the offset/size adjusted by RF_PROTECTED_SECTORS.  note

don't le32toh() the disklabel.  if it was wrong-endian, then
getdisklabel() will have swapped it.

ok jmcneill thorpej.

(mrg)

update the list of platforms supporting loading kernels from raid.
expand the HISTORY section to include notable RF moments in netbsd.
bump date.

(mrg)

correct indentation. spotted by Roland.

(cjep)

support different endian raidframe component label.

there are two on-disk formats in use in raidframe:
- the component label
- the parity map

the parity map is a bitmap implemented as bytes, so it has no
endian issue.  the component label is the problem, as most of
the fields are 32 bit.  this change only supports version 2 of
raidframe (active since the year 2000.)

as component labels are read and used before a raidPtr for the
raid set has been created, there is no obvious storage for the
swapped indicator, so the in-core version remains the on-disk
version, while the rest of in-core label is swapped.

in raidread_component_label() and raidwrite_component_label(),
check if the swapped version, and if so, call new rf_swap_label()
and ensure that the in-core label is native-byte order.  for the
write method, an on-stack copy is modified before writing, so
that the in-core version remains valid.  (this stack usage is
below other stack usage in similar functions here.)

adjust the label ioctls RAIDFRAME_GET_COMPONENT_LABEL and
RAIDFRAME_GET_COMPONENT_LABEL80 to return the byte-swapped
version so that eg, raidctl -s reports the right version.

when performing final configuration of a raidset, report if a
label swapped, and also complain if there are differently swapped
versions on the other components.

tested on arm64, sparc64 and amd64
ok @oster

(mrg)

in getdisklabel() swap the other-endian disklabel, #ifdef LIBSA_DISKLABEL_EI

(mrg)

tests/lint: make test d_gcc_extension platform-independent

That test was intended to test the keywords '__extension__' and
'__typeof'.  The GCC builtin functions were just a side-effect.

These built-in functions generated error messages on platforms such as
amd64 where sizeof(long double) != sizeof(double), but not on others
such as sparc.

The current infrastructure for the lint tests cannot handle tests with
platform-dependent outcome.

(rillig)

tests/lint: make test 130 platform-independent

On 32-bit platforms such as i386 and sparc, sizeof(int) == sizeof(long),
which produced an additional unintended lint error message:

msg_130.c(78): error: duplicate case in switch: 4 [199]

(rillig)

PR port-xen/53130 was fixed and closed some time ago, so now we can
belatedly re-enable inclusion of the compat_60 x86 microcode code in
the loadable module.  (It was already being included in the built-in
version of the module.)

(pgoyette)

The long lines test is now expected to pass

(cjep)

While we do not yet have any 64bit kernels, do not build
floppy or CD images for installation.

(martin)

As per OpenBSD, use malloc for the line buffer. Fixes the known issue
with long lines and makes our ATF test suite pass fully.
Closes PR bin/54620 from coypu who suggested the approach.
Reviewed by christos.

(cjep)

Provide support to build 32bit crunched ramdisks and use that for
macppc64. With help from rin.

(martin)