if we're building with GCC and MACHINE_CPU is aarch64, error out since
that isn't going to work.

doing this better in config(1) is difficult, but at least we can
detect and give a useful error message for this case instead of just
erroring out with some compiler options aren't supported. if we have
a good solution for config(1) we can remove this hack here..

might want to copy this hack for riscv too.

Simplify fix for PR toolchain/57146

Introduce ARCH_STRIP_SYMBOLS variable to centralize logic for debug
symbols from MD Makefile's to Makefile.kern.inc.

Fix kernel size inflation for arm and aarch64 (PR toolchain/57146)

For some conditions, SYSTEM_LD_TAIL is set for arm and aarch64.
Then, ctfmerge(1) in default SYSTEM_LD_TAIL is unintentionally
skipped, which results in the catastrophic kernel size inflation,
as reported in the PR.

Also, introduce and use OBJCOPY_STRIPFLAGS variable instead of
STRIPFLAGS, as strip(1) is replaced by objcopy(1) during MI
kernel build procedure.

XXX
For Makefile.{arm,aarch64}, weird logic is used to determine how
to handle debug symbols; MKDEBUG{,KERNEL} are taken into account
later in sys/conf/Makefile.kern.inc.

branches: 1.55.20;
Add support for KASAN on ARMv[67]

Thanks to maxv for many pointers and reviews.

Fix the armv[67] memory attributes for uncached memory. Previously it was
mapped as strongly-ordered which meant that unaligned accesses would fault.

armv7_generic_bs_map now maps pages with PMAP_DEV which is treated as SO

bus_dma continues to use PMAP_NOCACHE as appropriate, but this now get
mapped to the correct memory attribute bits for armv[67]

DEVMAP_ENTRY usees a new flag PTE_DEV.

The workaround for the unaligned access faults is now removed.

XXX Other armv[67] boards bus_space implementations should be checked.
XXX There is scope to reduce the difference to aarch64

branches: 1.53.2;
Use CFLAGS instead of COPTS for -mapcs-frame when using DDB.

Discussed with mrg@

gcc 8 needs -mapcs-frame when DDB is specified for backtrace to work

Update LLVM to 10.0.0git (01f3a59fb3e2542fce74c768718f594d0debd0da)

PR kern/54486

Workaround for alignment faults on ARMv6+, at least occur with
axe(4) and athn(4) drivers.

For ARMv6+, unaligned access is enabled by default. However, it
cannot be used for non-cacheable memory, which is used as DMA
buffers. This results in alignment faults above. A real fix is
to use cacheable memory as DMA buffers. However, it breaks some
drivers, awge(4) and vchiq(4) at least.

Until we figure out problems and fix them, we choose a fail-safe
workaround here; forbid unaligned memory access for whole kernel.
Affects on performance is negligibly small as far as we can see.

XXX
pullup netbsd-9

branches: 1.49.4;
- Determine KERN_AS automatically depending on whether OPT_MODULAR is
set or not, in the same way as libcompat.

- Specify OPT_MODULAR in the port Makefile instead of KERN_AS.

Now, KERN_AS=library is used for kernels without module(7) for all ports.

OK christos

branches: 1.48.16; 1.48.18;
Include splash with genfb.
splash.o is only included with options SPLASHSCREEN.
splash_image.o target moved to Makefile.kern.inc.

This commit was approved by matt@

Remove unintended whitespace.

Added a new option for embedding a splash screen into kernel.
Add: options SPLASHSCREEN
makeoptions SPLASHSCREEN_IMAGE="path/to/image"
to your config file. So far it will work on amd64 and RPI/RPI2.

This commit was with ideas, help, and OK from jmcneill@.

branches: 1.45.2;
Reorganize symbol table embedding. The existing option SYMTAB_SPACE is
replaced by the make option COPY_SYMTAB set to any value. The copy of
the symbol table is no longer put into a buffer in kern_ksyms.o, but a
small helper object. This object is build first with a dummy size, then
the kernel is linked to compute the real dimension of the symbol table
buffer. After that, the helper object is rebuild and the kernel linked
again.

Do the wildcard symbol dance for the debug case as well.

Use wildcards for stripping/preserving the mapping symbols on ARM and
AArch64. LLVM creates unique symbols in each file of the form $a.n etc.

branches: 1.42.4;
Explicitly enable VFP support for vfp_init.c. Fixes build of
INTEGRATOR_CP with clang.

Compile the armv4 cpufunc file with -mcpu=arm8 instead of arm9, to avoid
an armv4t marked object file which causes failure in the shark objcopy
step when creating the a.out version of the kernel.

Explicitly force a CPU for the assembler for the compatibility
cpu_func_* files. LLVM complains about the use of deprecated
instructions when -march=armv7 or equivalent is in use.

Pass -mfpu=vfp so GAS won't complain about vmsrne

Make sure certain files are always assembled with -marm

branches: 1.37.2;
Fix typo in option name

Always compile the kernel with -mfloat=soft

Make arm32_kvminit.o is complied with -fno-stack-protector

branches: 1.34.2;
Remove an '@' to hopefully see what's going on with the daily builds.

adjust the size of db_symtab space automatically.
proposed on tech-toolchain@, and got no objections.

branches: 1.32.20; 1.32.36; 1.32.40; 1.32.44;
Remove local -fno-strict-aliasing hack.
sys/conf/Makefile.kern.inc has it for gcc4.

branches: 1.31.6; 1.31.8;
Add arm version of cpu_in_cksum. Based on previous in_cksum4 code.

In fact it's mostly the same code, with a different stub on it.

On a cats the regress/sys/net/in_cksum tests show that it takes around
50-60% of the time the C version takes. In some cases it takes as low as
20%.

Refactor in_cksum/in4_cksum/in6_cksum implementations:
- All three functions are included in the kernel by default.
They call a backend function cpu_in_cksum after possibly
computing the checksum of the pseudo header.
- cpu_in_cksum is the core to implement the one-complement sum.
The default implementation is moderate fast on most platforms
and provides a 32bit accumulator with 16bit addends for L32 platforms
and a 64bit accumulator with 32bit addends for L64 platforms.
It handles edge cases like very large mbuf chains (could happen with
native IPv6 in the future) and provides a good base for new native
implementations.
- Modify i386 and amd64 assembly to use the new interface.

This disables the MD implementations on !x86 until the conversion is
done. For Alpha, the portable version is faster.

branches: 1.29.16; 1.29.20; 1.29.22; 1.29.28;
src/sys/arch/arm32 is long gone, so stop referring to it.

branches: 1.28.2;
Add a port of Sam Leffler's IXP425 micro-engine support (from FreeBSD).

This is very much a work in progress. At the present time, only Ethernet
is supported.

branches: 1.27.22; 1.27.24;
partially revert previous commit stripping arm mapping symbols
it broke ddb when DEBUG wasn't set to -g.

strip arm mapping symbols for ddb using STRIPFLAGS instead of SYSTEM_LD_TAIL
this works even for kernels compiled with -g and doesn't interfere with gdb

branches: 1.25.2; 1.25.4;
Use objcopy (as suggested by Richard Earnshaw) to strip ARM mapping symbols
from the kernel image, as they interfere with ddb. But only do this if
'makeoptions DEBUG="-g"' was not specified, in case they're needed by gdb.

s/GENASSYM/GENASSYM_CONF/ so we can use "GENASSYM" as the program name.

Add isa_irq.o and spl.o to the list of files that depend on assym.h so
that "make -j<bignum>" works without a "make depend" beforehand.

Add dependancies on assym.h for some assembly files so that
"make -j <bignum>" works without having to do a "make depend" first.

Add the %MAKEOPTIONSAPPEND token at the end of the file, after the
common Makefile.kern.inc has been included.

branches: 1.20.4;
Fix a typo so that hack actually takes effect.

Compile sys/kern/uipc_socket.c -fno-strict-aliasing for now. PR 23044.

branches: 1.18.2;
Don't define -D${MACHINE} in Makefile.arm. Instead, let platforms
that care define it themselves. Note that evbarm NO LONGER defines
-D${MACHINE}.

add missing protos, clean up includes.

Set LOOSE_PROTOTYPES to no for cats. This means cats now uses stricter prototype checking.

Roll the rest of the ports over to the new MI kernel build machinery.
Any problems reported by testers have been fixed, and massive
cross-compiling of kernels has shown that any problems that remain
with actually building kernels are not related to this.

Allow port-specific Makefile fragments to specify a list of additional
genassym.cf fragments, if desired.

End the "new ARM includes strategy" experiment. The new strategy
broke too many assumptions makde by other parts of the source tree,
and the strategy and how it was supposed to work was never discussed
on tech-userlevel, nor was it applied consistently (to all ARM ports
and to other ports which have common MACHINE_ARCH code, such as MIPS,
m68k, powerpc).

Verified to complete a full "make build" on cats, dnard, evbarm,
and netwinder.

cleanup:
options SPACE TAB
makeoptions TAB
psuedo-device TAB
remove trailing whitespace
replace multiple spaces -> tabs
options "FOO" -> options FOO
options "FOO=bar" -> options FOO=bar
options "FOO=\"bar\"" -> options FOO="\"bar\""

branches: 1.11.2;
Make it possible for Makefile.${MACHINE}.inc to provide extra
stuff at the end of SYSTEM_LD_TAIL.

Change defaults for kernel compiles. Default all to USETOOLS?=no and have
the etc Makefile override that by putting USETOOLS into $.MAKEOVERRIDES
This way the default for kernel compiles is still to use the installed
toolchain instead of depending on $TOOLDIR. $TOOLDIR can be used by
simply adding USETOOLS=yes to the command line as usual.

Adjust each ports template to set the default no setting and also pull in
bsd.own.mk if they weren't already to ensure they'll build correctly
with the new toolchain setup.

branches: 1.9.2;
Use MACHINE, not TARGET_MACHINE.

Pull in <bsd.own.mk> to get the correct toolchain definitions for
the USE_NEW_TOOLCHAIN case.

Workaround "../../../.."

finish moving common arm32 bits out into arm/arm32.

Change -x into -X for linking, it makes traces more meaningful.

Gah!. Use the variable name (LINKENTRY).

Fix a few things for NetWinder. NetWinder needs to override the
std entry point so add a hook for that. Also pick up the port
specific .inc file from the right directory.

branches: 1.2.2; 1.2.4;
Only include dependency on include-dirs if the include directory doesn't
exist.

Add common Makefile for arm ports. Only include (for now) machine/io.h
if arm32 is defined. io.h is RISCPC specific and isn't needed for other
ports.