Apparently new source files need to be added to each Makefile.inc
individually. Who comes up with this kind of crappy build infra?

branches: 1.3.6;
Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.2.12; 1.2.22; 1.2.26;
include rumpcpu_generic

branches: 1.1.2; 1.1.4; 1.1.8;
Add md crud to allow libkern softfloat code to build.
And while here, make alpha use the real elf relocation code.

branches: 1.2.2;
Ansify function definitions w/o arguments. Generated with sed.

branches: 1.1.2; 1.1.4; 1.1.6;
Add md crud to allow libkern softfloat code to build.
And while here, make alpha use the real elf relocation code.

Apparently new source files need to be added to each Makefile.inc
individually. Who comes up with this kind of crappy build infra?

branches: 1.4.4; 1.4.6;
Antti objected to including rumpuser_sync_icache. Exclude it from the build.

Implement rumpuser_sync_icache hypercall.

branches: 1.2.2; 1.2.4;
Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.1.2;
Build kobj_machdep.c

branches: 1.1.2; 1.1.6;
Implement rumpuser_sync_icache hypercall.

Make the direct-map API always available, but fail if KASAN or rump.

(Only for architectures that support it at all; on others,
__HAVE_MM_MD_DIRECT_MAPPED_PHYS/IO are still undefined and the
functions unimplemented.)

This gives modules like zfs an opportunity to use it.

While here, fix the one caller of mm_md_direct_mapped_phys that
ignored the return value (and make sure to call pmap_kremove/update
before uvm_km_free).

branches: 1.2.16; 1.2.18;
Define rump_nativeabi_p() under arch, where one would logically expect
it to be defined.

branches: 1.1.4; 1.1.6; 1.1.10; 1.1.12;
Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.2.16;
Put the kernelside rump kernel headers into <rump-sys> instead of
sprinkling them around the faction directories. Avoids having
to add a CPPFLAGS (or several) to pretty much every component
Makefile.

Leave compat headers around in the old locations.

The commit changes some autogenerated files, but I'll fix the
generators shortly and regen.

branches: 1.1.2;
Define rump_nativeabi_p() under arch, where one would logically expect
it to be defined.

Put the kernelside rump kernel headers into <rump-sys> instead of
sprinkling them around the faction directories. Avoids having
to add a CPPFLAGS (or several) to pretty much every component
Makefile.

Leave compat headers around in the old locations.

The commit changes some autogenerated files, but I'll fix the
generators shortly and regen.

Apparently not all ports define struct clockframe in cpu.h, so we cannot
provide our definition that way. Instead, generate the struct clockframe
passed to hardclock() in MD code.

branches: 1.1.4; 1.1.6; 1.1.10; 1.1.12;
Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.3.4;
Merge the [pgoyette-compat] branch

Add missing RCSIDs.

branches: 1.1.2;
Make the direct-map API always available, but fail if KASAN or rump.

(Only for architectures that support it at all; on others,
__HAVE_MM_MD_DIRECT_MAPPED_PHYS/IO are still undefined and the
functions unimplemented.)

This gives modules like zfs an opportunity to use it.

While here, fix the one caller of mm_md_direct_mapped_phys that
ignored the return value (and make sure to call pmap_kremove/update
before uvm_km_free).

branches: 1.1.4; 1.1.6; 1.1.10;
Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

rump: Define pmap_resident_count, pmap_wired_count conditionally.

These definitions will not be used by anything yet because on every
architecture except x86 as of yeterday, pmap_resident_count and
pmap_wired_count are defined as macros anyway. But if more struct
pmaps are made private these definitions will get used.

Add stubs for pmap_resident_count, pmap_wired_count.

Put the kernelside rump kernel headers into <rump-sys> instead of
sprinkling them around the faction directories. Avoids having
to add a CPPFLAGS (or several) to pretty much every component
Makefile.

Leave compat headers around in the old locations.

The commit changes some autogenerated files, but I'll fix the
generators shortly and regen.

Give the kernel/local pmaps actual storage.

That way friendly neighborhood macros won't go scribbling
in the wrong places.

Use RUMP_PMAP_KERNEL and RUMP_PMAP_LOCAL to denote
kernel and local client pmaps, respectively.

branches: 1.2.4; 1.2.6; 1.2.10; 1.2.12;
Minor comment massage, zero functionality massage.

Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.6.8; 1.6.18; 1.6.22;
Reinstate the blanket pmap.h for archs which do not conform to the
kernel ABI (i.e. not i386 or amd64). Due to the "half function,
half macro, all noodles" nature of pmap.h, it's too entangling and
too brittle to keep up with an ifdeffy MI implementation.

The x86 kernel ABI depends on __cpu_simple_lock stuff being present.
Since they are practically never used (only when prehistoric code
uses simple_lock()), their efficiency doesn't matter that much and
we can simply adapt the versions from x86 lock.h.

add __HAVE_CPU_COUNTER stubs where possible (i.e. where the arch
doesn't think inlines are the second compiling)

branches: 1.3.2; 1.3.4; 1.3.8; 1.3.10; 1.3.12;
Define MODULAR for rump core components. This enables module
loading via the kernel module framework (instead of dlopen()).
For now it only works on amd64 and i386, but for the rest it should
just be a matter of including the relevant kobj_machdep.c modules
from the kernel sources.

Rename i386 and amd64 {cpu,spl}.c to rump{cpu,spl}.c to avoid
name collisions with other source files in the kernel tree.

Switch i386 away from rump/include/machine. This means that rump
on i386 now uses the native kernel ABI. This in turn means that
rump modules and kernel modules are binary equivalent and can be
used interchangeably.

Rename i386 and amd64 {cpu,spl}.c to rump{cpu,spl}.c to avoid
name collisions with other source files in the kernel tree.

Switch i386 away from rump/include/machine. This means that rump
on i386 now uses the native kernel ABI. This in turn means that
rump modules and kernel modules are binary equivalent and can be
used interchangeably.

Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.1.2; 1.1.4; 1.1.6; 1.1.14; 1.1.24; 1.1.28;
add __HAVE_CPU_COUNTER stubs where possible (i.e. where the arch
doesn't think inlines are the second compiling)

Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.3.2; 1.3.4;
Needs pmap_write_protect to finish DBG=-g build with clang.

branches: 1.2.2; 1.2.4; 1.2.6; 1.2.14;
Give the kernel pmap a container in case some tries to access it
via some pmap.h macro.

Reinstate the blanket pmap.h for archs which do not conform to the
kernel ABI (i.e. not i386 or amd64). Due to the "half function,
half macro, all noodles" nature of pmap.h, it's too entangling and
too brittle to keep up with an ifdeffy MI implementation.

Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.11.4;
Inform the rump kernel hypervisor of valid thread contexts so
that the implementation can allocate and release storage for them
in an optimal fashion, if necessary.

Stopgap fix to make rump cooperate with pserialize, may be revisited later.
Patch from pooka, ok: rmind. No related regressions in a complete atf test
run (which works again with this, even on non x86 SMP machines).

branches: 1.9.8; 1.9.18;
set first attached cpu as primary

Fix multiple virtual cpu support.

... or at least on x86. CPU_INFO_FOREACH() still isn't MI, and I
don't want to support 2^n different versions.

branches: 1.7.2; 1.7.4;
Give lwp usage some much-needed love: stop treating lwp0 as the
all-sink and make sure each separate thread in rump has its own
lwp. Happy-go-lucky callers will get scheduled a temporary lwp
on entry, while true lwp connoisseurs may request a stable lwp
for their purposes. Some more love may be required later down the
road, but for now different threads will stepping on each others
toes.

Add initial work on a rump virtual cpu scheduler. This is necessary
for kernel code which has been written to avoid MP contention by
using cpu-local storage (most prominently, select and pool_cache).

Instead of always assuming rump_cpu, the scheduler must now be run
(and unrun) on all entry points into rump. Likewise, rumpuser
unruns and re-runs the scheduler around each potentially blocking
operation. As an optimization, I modified some locking primitives
to try to get the lock without blocking before releasing the cpu.

Also, ltsleep was modified to assume that it is never called without
the biglock held and made to use the biglock as the sleep interlock.
Otherwise there is just too much drama with deadlocks. If some
kernel code wants to call ltsleep without the biglock, then, *snif*,
it's no longer supported and rump and should be modified to support
newstyle locks anyway.

"rumppriv" goes back to "rump" per internal interface naming change.

Adjust rump sources for external/internal interfaces.
No functional change.

branches: 1.3.2;
Ansify function definitions w/o arguments. Generated with sed.

branches: 1.2.2; 1.2.4; 1.2.6;
Define MODULAR for rump core components. This enables module
loading via the kernel module framework (instead of dlopen()).
For now it only works on amd64 and i386, but for the rest it should
just be a matter of including the relevant kobj_machdep.c modules
from the kernel sources.

Rename i386 and amd64 {cpu,spl}.c to rump{cpu,spl}.c to avoid
name collisions with other source files in the kernel tree.

Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.2.4; 1.2.16; 1.2.26; 1.2.30;
Rip out the rwlock spl emulation code. It never did anything useful
except keep my feet warm by consuming an insane amount of cpu
cycles -- in rump our current "cpu" context is never interrupted
& we have MULTIPROCESSOR.

(itch i'm scratching: it made the networking stack 10-20% slower.
this is one of the places where fast code actually matters)

branches: 1.1.2; 1.1.4;
Rename i386 and amd64 {cpu,spl}.c to rump{cpu,spl}.c to avoid
name collisions with other source files in the kernel tree.

Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.2.2;
use the definitions from lock.h

branches: 1.1.2; 1.1.4; 1.1.6; 1.1.14; 1.1.24;
The x86 kernel ABI depends on __cpu_simple_lock stuff being present.
Since they are practically never used (only when prehistoric code
uses simple_lock()), their efficiency doesn't matter that much and
we can simply adapt the versions from x86 lock.h.

Rename i386 and amd64 {cpu,spl}.c to rump{cpu,spl}.c to avoid
name collisions with other source files in the kernel tree.

Switch i386 away from rump/include/machine. This means that rump
on i386 now uses the native kernel ABI. This in turn means that
rump modules and kernel modules are binary equivalent and can be
used interchangeably.

mipsn64* is native 64

Apparently new source files need to be added to each Makefile.inc
individually. Who comes up with this kind of crappy build infra?

branches: 1.2.6;
Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.1.4; 1.1.14; 1.1.24; 1.1.28;
Make librump play with mips nicely. Define ARCH_ELFSIZE for mips to be 32.
This works for N64 kernels because objcopy them to be 32bit to the bootloaders
can handle them.

Apparently new source files need to be added to each Makefile.inc
individually. Who comes up with this kind of crappy build infra?

branches: 1.5.6;
Be a little more careful checking MLIBDIR

Use ARCH_ELFSIZE=32 is MLIBDIR == "powerpc"

Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.2.2;
File is common between pp6 and ppc64 ports, so define ARCH_ELFSIZE
accordingly.

From christos@, resolves the latest build break

branches: 1.1.2; 1.1.12;
define ARCH_ELFSIZE=32 and add kobj_stubs.c rumpcpu_generic.c.

Define rump_nativeabi_p() under arch, where one would logically expect
it to be defined.

branches: 1.1.4; 1.1.6; 1.1.10; 1.1.12;
Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.2.16;
Put the kernelside rump kernel headers into <rump-sys> instead of
sprinkling them around the faction directories. Avoids having
to add a CPPFLAGS (or several) to pretty much every component
Makefile.

Leave compat headers around in the old locations.

The commit changes some autogenerated files, but I'll fix the
generators shortly and regen.

branches: 1.1.2;
Define rump_nativeabi_p() under arch, where one would logically expect
it to be defined.

rump/x86: Need x86/cpufunc.h for declaration of wbinvd.

Put the kernelside rump kernel headers into <rump-sys> instead of
sprinkling them around the faction directories. Avoids having
to add a CPPFLAGS (or several) to pretty much every component
Makefile.

Leave compat headers around in the old locations.

The commit changes some autogenerated files, but I'll fix the
generators shortly and regen.

Apparently not all ports define struct clockframe in cpu.h, so we cannot
provide our definition that way. Instead, generate the struct clockframe
passed to hardclock() in MD code.

branches: 1.2.4; 1.2.6; 1.2.10; 1.2.12;
Allow compile-time optimizations to curlwp. This can have a pretty
staggering impact on performance. When running sendto() in a loop,
the improvement is 200k more calls per second with an inlined __thread
curlwp as opposed to the default. In other words, it shaves off hundreds
of CPU cycles per call (~20%). Even just eliminating the x86_curlwp()
call in favor of an inline gives an improvement of 60k calls per second.

Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

Serialize rdtsc using with lfence, mfence or cpuid to read TSC more precisely.

x86/x86/tsc.c rev. 1.67 reduced cache problem and got big improvement, but it
still has room. I measured the effect of lfence, mfence, cpuid and rdtscp.
The impact to TSC skew and/or drift is:

AMD: mfence > rdtscp > cpuid > lfence-serialize > lfence = nomodify
Intel: lfence > rdtscp > cpuid > nomodify

So, mfence is the best on AMD and lfence is the best on Intel. If it has no
SSE2, we can use cpuid.

NOTE:
- An AMD's document says DE_CFG_LFENCE_SERIALIZE bit can be used for
serializing, but it's not so good.
- On Intel i386(not amd64), it seems the improvement is very little.
- rdtscp instruct can be used as serializing instruction + rdtsc, but
it's not good as [lm]fence. Both Intel and AMD's document say that
the latency of rdtscp is bigger than rdtsc, so I suspect the difference
of the result comes from it.

branches: 1.1.4; 1.1.6; 1.1.10;
Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

rump/x86: Add stubs for pmap_resident_count, pmap_wired_count.

These are no longer static inlines, now that struct pmap is private.

Put the kernelside rump kernel headers into <rump-sys> instead of
sprinkling them around the faction directories. Avoids having
to add a CPPFLAGS (or several) to pretty much every component
Makefile.

Leave compat headers around in the old locations.

The commit changes some autogenerated files, but I'll fix the
generators shortly and regen.

Give the kernel/local pmaps actual storage.

That way friendly neighborhood macros won't go scribbling
in the wrong places.

Use RUMP_PMAP_KERNEL and RUMP_PMAP_LOCAL to denote
kernel and local client pmaps, respectively.

branches: 1.1.4; 1.1.6; 1.1.10; 1.1.12;
Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.1.4; 1.1.6; 1.1.10;
Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

Put the kernelside rump kernel headers into <rump-sys> instead of
sprinkling them around the faction directories. Avoids having
to add a CPPFLAGS (or several) to pretty much every component
Makefile.

Leave compat headers around in the old locations.

The commit changes some autogenerated files, but I'll fix the
generators shortly and regen.

branches: 1.1.4; 1.1.6; 1.1.10; 1.1.12;
Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

Make the direct-map API always available, but fail if KASAN or rump.

(Only for architectures that support it at all; on others,
__HAVE_MM_MD_DIRECT_MAPPED_PHYS/IO are still undefined and the
functions unimplemented.)

This gives modules like zfs an opportunity to use it.

While here, fix the one caller of mm_md_direct_mapped_phys that
ignored the return value (and make sure to call pmap_kremove/update
before uvm_km_free).

branches: 1.8.28; 1.8.30;
Rototill a bit, and attempt to disguise it as non-gratuitous.

Add arch/generic and move non-x86 files from rumpkern/ there. Also,
move files from arch/i386 to arch/x86, and make both i386 and x86_64
use those.

This clarifies the situation with what is MD vs. MI code.

renames:
rumpcpu_generic,kobj_stubs,pmap_stubs => arch/generic/rump_generic_$x
arch/i386/* => arch/x86/rump_x86_$x

(for those who forget, x86 requires MD code because rump kernels
use the same ABI as kernel modules)

branches: 1.7.8; 1.7.18; 1.7.22;
Reinstate the blanket pmap.h for archs which do not conform to the
kernel ABI (i.e. not i386 or amd64). Due to the "half function,
half macro, all noodles" nature of pmap.h, it's too entangling and
too brittle to keep up with an ifdeffy MI implementation.

The x86 kernel ABI depends on __cpu_simple_lock stuff being present.
Since they are practically never used (only when prehistoric code
uses simple_lock()), their efficiency doesn't matter that much and
we can simply adapt the versions from x86 lock.h.

add __HAVE_CPU_COUNTER stubs where possible (i.e. where the arch
doesn't think inlines are the second compiling)

branches: 1.4.2; 1.4.4; 1.4.8; 1.4.10; 1.4.12;
add missing "arch/" as noted by a few people

Define MODULAR for rump core components. This enables module
loading via the kernel module framework (instead of dlopen()).
For now it only works on amd64 and i386, but for the rest it should
just be a matter of including the relevant kobj_machdep.c modules
from the kernel sources.

Rename i386 and amd64 {cpu,spl}.c to rump{cpu,spl}.c to avoid
name collisions with other source files in the kernel tree.

Switch amd64 rumps to the kernel ABI.