CVS log for src/sys/arch/mac68k/mac68k/locore.s

Up to [cvs.NetBSD.org] / src / sys / arch / mac68k / mac68k

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

Pull up following revision(s) (requested by rin in ticket #724
to fix build fallout on the branch):

	sys/arch/m68k/m68k/compat_16_sigreturn14.s: revision 1.6
	sys/arch/mvme68k/mvme68k/locore.s: revision 1.129
	sys/arch/news68k/news68k/locore.s: revision 1.84
	sys/arch/hp300/hp300/locore.s: revision 1.183
	sys/arch/m68k/m68k/compat_13_sigreturn13.s: revision 1.8
	sys/arch/m68k/m68k/sigreturn.s: file removal
	sys/arch/m68k/include/pte_motorola.h: revision 1.10
	sys/arch/atari/atari/locore.s: revision 1.125
	sys/arch/amiga/amiga/locore.s: revision 1.169
	sys/arch/sun2/sun2/locore.s: revision 1.37
	sys/arch/next68k/next68k/locore.s: revision 1.84
	sys/arch/x68k/x68k/locore.s: revision 1.130
	sys/arch/sun3/sun3x/locore.s: revision 1.77
	sys/arch/cesfic/cesfic/locore.s: revision 1.45
	sys/arch/m68k/conf/files.m68k: revision 1.53
	sys/arch/sun3/sun3/locore.s: revision 1.109
	sys/arch/luna68k/luna68k/locore.s: revision 1.81
	sys/arch/mac68k/mac68k/locore.s: revision 1.182

Define PTE used in the pmap module int terms of the bit definitions
in mmu_{51,40}.h.

Make compat_13_sigreturn13.s and compat_16_sigreturn14.s build as their
own stand-alone files and G/C the now-empty sigreturn.s.

Fix up the mac68k interrupt handlers to work with the new style common
clockframe layout and remove the CLOCK_FORMAT0 work-around.  As a nice
side-effect, this also eliminates the super-sketchy stack unwinding used
by rtclock_intr to get at the interrupt stack frame.

Make sigcode.s and sunos_sigcode.s build as their own stand-alone files.

Make compat_13_sigreturn13.s and compat_16_sigreturn14.s build as their
own stand-alone files and G/C the now-empty sigreturn.s.

Build m68k/support.s on its own; don't include it from locore.s

Statically-initialize the '851 / '030 protorp with MMU51_SRP_BITS.  In
pmap_init(), re-initialize protorp with MMU51_CRP_BITS.  Remove the now-
redundant code that does this in each of the m68k platforms.

Remove duplicated / slightly-tweaked loadustp() (load user segment table)
routines from all of the m68k ports using the shared pmap.  Instead, in
pmap_init(), set up a function pointer to the appropriate mmu_load_urp*()
function in mmu_subr.s.

Define values for the 68030 TT registers in terms of the definitions
in <m68k/mmu_30.h> rather than using magic numbers.

Stop using magic numbers for the MMU root pointer attributes and the
Translation Control register, and also get rid of "#if PGSHIFT == ..."
where those magic numbers are used.

Instead, define new macros: MMU51_SRP_BITS, MMU51_CRP_BITS, MMU51_TCR_BITS,
and MMU40_TCR_BITS, in terms of the definitions in mmu_{40,51}.h.  These
automagically adapt to 8K and 4K pages based on the machine-specific value
of PGSHIFT.

loadustp(): Optimize for the overwhelmingly common case of "not the HP MMU".
Rather than converting the level 1 table address to a page number before
calling loadustp() only to have loadustp() convert it back to an address
for the '851, '030, '040, and '060, instead pass the address and convert
to a page number only in the case of the HP MMU.

This is a wash on HP MMU machines (9000/320 and 9000/350), and saves at
least 4 instructions (2x moveq + 2x lsXl) on every context switch on
everything else.

s/identifing/identifying/ and s/multipler/multiplier/ in comments.

Pull up following revision(s) (requested by rin in ticket #1255):

	sys/arch/mac68k/mac68k/locore.s: revision 1.174

Fix DJMEMCMAX option for Quadra/Centris 650/800.
- Use jeq instead of jra for conditional branch.
- Use cmpl instead of cmp (= cmpw) for int variables.

Now, my Quadra 800 recognizes full 520MB memory!

Sync with HEAD.

Fix DJMEMCMAX option for Quadra/Centris 650/800.

- Use jeq instead of jra for conditional branch.
- Use cmpl instead of cmp (= cmpw) for int variables.

Now, my Quadra 800 recognizes full 520MB memory!

For GCC8, do not omit frame pointer for intr_dispatch() and via1_intr()
(-fomit-frame-pointer is enabled for -O and higher for GCC8).

This is required by rtclock_intr() which unwinds stack frame of caller!

XXXXXX
We need to get rid of this hackest hack for rtclock_intr(). This problem
was discussed back in 2014:

http://mail-index.netbsd.org/port-mac68k/2014/08/15/msg000595.html
http://mail-index.netbsd.org/port-mac68k/2014/08/17/msg000600.html
http://mail-index.netbsd.org/port-mac68k/2014/08/17/msg000601.html

However, unfortunately, the problem has been left untouched until today.

The patch attached in the third message works around the problem. But,
it adds hard-coded magic numbers to intr_dispatch() and via1_intr().
For real fix, we should probably reconsider whole interrupt handling.

Anyway, now kernel compiled by GCC8 works fine as far as I can see.

Align tmpstk to 4-byte boundary as required by System V (ELF) ABI.

Fix GCC8-compiled kernel crash during very early boot stage when
printf decimal numbers; __umoddi3 corrupts stack frame which is not
aligned properly.

Sync with HEAD

Sync with HEAD, resolve a few conflicts

Remove compat_svr4 and compat_svr4_32, as discussed on tech-kern@ recently,
but also as discussed several times in the past.

Rebase to HEAD as of a few days ago.

sync with head.

for a reference, the tree before this commit was tagged
as yamt-pagecache-tag8.

this commit was splitted into small chunks to avoid
a limitation of cvs.  ("Protocol error: too many arguments")

sync with head

Use common m68k/busaddrerr.s for bus error and address error handlers.

No binary changes on GENERIC.

Remove an instruction for nonexistent 060 branch prediction error accounting.

It's amiga and atari specific and copied from amiga/locore.s
to mac68k/locore.s in rev 1.80, then pasted into other ports.
Note this is harmless because all these ports don't have 68060 variants
(i.e. it's inside #if defined(M68060) block) and mvme68k (which has 68060)
has removed it since the initial import.

switch to ELF naming for local labels.

sync with head

Added DJMEMCMAX option for recognizing 128 meg SIMMs in Centris and Quadra
610, 650 and 800 systems. From Rob Braun,
http://synack.net/~bbraun/netbsdmac2.html

sync with head

merge to -current.

Merge m68881_save() and m68881_restore() (using most common mvme68k ones).
All functions are checked by my eyes, and all GENERIC kernels compile.

Make sure to call pflusha before (not after) enabling MMU on
68030/68851 initialization paths.

Pointed out by isaki@ per observation of ATC status on XM6i emulator.
Tested on TT030, HP362, NWS-1750, and X68030 by me.

XXX: probably we can remove existing pflusha or TBIA calls after MMU turned
XXX: on (and before enabling cache) but I leave them as is for now to avoid
XXX: unexcpected side effects.

Move spread getsfc() and getdfc() functions into common m68k/support.s.

Note:
- ENTRY_NOPROFILE() is okay since they are used only for debug printf
- they are declared to return int so no need to put a return value into %a0

Sync with HEAD.

sync with head

Sync with HEAD

Remove clause 3 (UCB advertising clause) from the University of Utah
copyright.  Confirmed by Mike Hibler, mike at cs.utah.edu - thanks!
Also, merge UCB and Utah copyright texts back into one, as they
originally were.

Extra verification by snj@.

Move counting of faults, traps, intrs, soft[intr]s, syscalls, and nswtch
from uvmexp to per-cpu cpu_data and move them to 64bits.  Remove unneeded
includes of <uvm/uvm_extern.h> and/or <uvm/uvm.h>.

Sync with HEAD.

sync with head.

sync with head

fix PR 6724 - convert m68k options to defflag's.  this means that
M680[12346] are now available from opt_m68k_arch.h.  FPSP meantioned
in the PR has already been fixed, and i could not find any more.

i built these kernels to ensure i did not break their builds:

amiga: GENERIC DRACO
atari: HADES FALCON MILAN-PCIIDE
mac68k: GENERIC
sun2: GENERIC
sun3: GENERIC GENERIC3X
cesfic: attempted GENERIC, does not build due to lack of machine/bus.h
hp300: GENERIC
luna68k: GENERIC
mvme68k: GENERIC
news68k: GENERIC
next68k: GENERIC
x68k: GENERIC

Sync with HEAD.

sync with head

s/cpu_fork/cpu_lwp_fork/ in comment.

Allow NetBSD/mac68k kernel work with PGSHIFT==13, i.e. 8KB/page.
Tested on LC630 with 68LC040.

Move uvm_setpagesize() calls from locore.s to pmap_bootstrap_finalize().
Also explicitly set uvmexp.pagesize to prepare 8KB/page settings.

XXX: There is #ifdef !defined(amiga) && !defined(atari) workaround
XXX: because they have different pmap initialization functions.

Use common pmap_bootstrap_finalize() to initialize lwp0 uarea etc.
Tested on LC630.

Kill proc0paddr.  Use lwp0.l_addr instead.

sync with head.

Sync with HEAD.

Sync with HEAD.

Fix inverted logic on FPU type check in fpfault().  From OpenBSD.

Sync with HEAD.

Sync with head.

Sync w/ -current. 34 merge conflicts to follow.

Unify splraiseipl(9) implementation among m68k ports as per
suggestion from isaki@ on port-m68k.

For mac68k:
- rename mac68k_ipls[] to common ipl2psl_table[]
- make makeiplcookie() return PSL bits via ipl2psl_table[]
  rather than IPL_FOO indexes

sync with HEAD

Merge the ppcoea-renovation branch to HEAD.

This branch was a major cleanup and rototill of many of the various OEA
cpu based PPC ports that focused on sharing as much code as possible
between the various ports to eliminate near-identical copies of files in
every tree.  Additionally there is a new PIC system that unifies the
interface to interrupt code for all different OEA ppc arches.  The work
for this branch was done by a variety of people, too long to list here.

TODO:
bebox still needs work to complete the transition to -renovation.
ofppc still needs a bunch of work, which I will be looking at.
ev64260 still needs to be renovated
amigappc was not attempted.

NOTES:
pmppc was removed as an arch, and moved to a evbppc target.

sync with head.

Sync w/ NetBSD-4-RC_1

Sync with head.

Sync with head.

Sync with HEAD.

Pull up following revision(s) (requested by mhitch in ticket #728):
	sys/arch/mvme68k/mvme68k/locore.s: revision 1.98
	sys/arch/mac68k/mac68k/trap.c: revision 1.128
	sys/arch/next68k/next68k/locore.s: revision 1.48
	sys/arch/mac68k/mac68k/locore.s: revision 1.152
	sys/arch/sun3/sun3/locore.s: revision 1.88
	sys/arch/sun2/sun2/locore.s: revision 1.19
	sys/arch/sun2/sun2/trap.c: revision 1.32
	sys/arch/m68k/m68k/db_trace.c: revision 1.51
	sys/arch/mvme68k/mvme68k/trap.c: revision 1.90
	sys/arch/news68k/news68k/trap.c: revision 1.53
	sys/arch/luna68k/luna68k/locore.s: revision 1.27
	sys/arch/atari/atari/locore.s: revision 1.99
	sys/arch/sun3/sun3/trap.c: revision 1.130
	sys/arch/x68k/x68k/trap.c: revision 1.89
	sys/arch/next68k/next68k/trap.c: revision 1.67
	sys/arch/x68k/x68k/locore.s: revision 1.79
	sys/arch/news68k/news68k/locore.s: revision 1.43
	sys/arch/luna68k/luna68k/trap.c: revision 1.46
	sys/arch/hp300/hp300/locore.s: revision 1.140
	sys/arch/cesfic/cesfic/locore.s: revision 1.15
	sys/arch/cesfic/cesfic/trap.c: revision 1.35
	sys/arch/m68k/m68k/trap_subr.s: revision 1.12
	sys/arch/amiga/amiga/locore.s: revision 1.143
Pass a frame pointer to trap() rather than the 'entire frame' trick.  Gcc4
was optimizing away modifications to the frame contents (it's not nice to
trick gcc).  Pass the pointer as the first argument to reduce the number
of places that would be changed otherwise.  Fixes the getcwd regression
test on most m68k ports.

Pass a frame pointer to trap() rather than the 'entire frame' trick.  Gcc4
was optimizing away modifications to the frame contents (it's not nice to
trick gcc).  Pass the pointer as the first argument to reduce the number
of places that would be changed otherwise.  Fixes the getcwd regression
test on most m68k ports.

Sync with head.

Update to HEAD.

Remove two unused global variables.

Initialize curlwp as well as lwp0.l_addr right after MMU is enabled.

Follow the pattern suggested by mhitch for the atari port and
remove inclusion of m68k/m68k/proc_subr.s, restoring the kernel
to a buildable state.

Sync with HEAD.

MI softintr(9)'fy.
Probably problematic, but LC630 is running at single user.

Kill caddr_t; there will be some MI fallout, but it will be fixed shortly.

sync with head.

Sync with head.

Pull up following revision(s) (requested by chs in ticket #345):
	sys/arch/mac68k/mac68k/locore.s: revision 1.146
in doboot, use "longscratch" as a scratch memory location instead of
something in the (read-only) text section.  fixes PR 35068.

Pull up following revision(s) (requested by chs in ticket #1626):
	sys/arch/mac68k/mac68k/locore.s: revision 1.146
in doboot, use "longscratch" as a scratch memory location instead of
something in the (read-only) text section.  fixes PR 35068.

Pull up following revision(s) (requested by chs in ticket #1626):
	sys/arch/mac68k/mac68k/locore.s: revision 1.146
in doboot, use "longscratch" as a scratch memory location instead of
something in the (read-only) text section.  fixes PR 35068.

Pull up following revision(s) (requested by chs in ticket #1626):
	sys/arch/mac68k/mac68k/locore.s: revision 1.146
in doboot, use "longscratch" as a scratch memory location instead of
something in the (read-only) text section.  fixes PR 35068.

in doboot, use "longscratch" as a scratch memory location instead of
something in the (read-only) text section.  fixes PR 35068.

merge ktrace-lwp.

Fix the sync with head I botched.

Sync with HEAD.

Sync with HEAD

Centralize identical copies of proc_trampoline code.

(lots more opportunites here for the bored hacker, such as rei and trap0...)

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22364, verified by myself.

Make sure to include opt_foo.h if a defflag option FOO is used.

Use PAGE_SIZE rather than NBPG.

Merge the nathanw_sa branch.

LWPify.  From hp300.

catch up with -current on kqueue branch

Catch up to -current.

Use m68k/m68k/cacheops.[ch]
Use m68k/m68k/bus_dma.c
	- Add mainbus_attach_args with bus_space_tag_t and bus_dma_tag_t.
	- Use passed-in tags for nubus scan.

Catch up with -current.

Catch up to -current.

Sync w/ -current.

s/seperate/separate/

Frederic Bruckman spotted a missing _C_LABEL.

Sync kqueue branch with -current.

Catch up to -current.

prepare for ELF:
 - add "%" prefix to register names in assembly code.
 - use _C_LABEL() where necessary.
 - use packed structures where necessary to match the old ABI.

MD Scheduler Activation bits for Mac68k.

XXX: I'm unable to even compile-test these changes since Mac68k hasn't
XXX: transitioned to ELF, and I don't have a -current m68k a.out toolchain
XXX: handy.

update to -current

file locore.s was added on branch nathanw_sa on 2001-11-18 18:12:02 +0000

seperate -> separate

Catch up to -current.

add missing   #include "opt_kgdb.h"

Pull up revision 1.133 (requested by briggs):
 Improve performance of the pseudo-DMA mode of the esp driver, and handle
 residual FIFO data on write.

Sync with HEAD

Rework the esp driver a little bit to improve performance.  This requires
an ugly bit where we record the value of a2 while handling a "nofault" bus
error in locore.s in order to know how far the transaction got before a
disconnect (see esp.c comments for more details on this).
Also get rid of a loop (and delay()!) in the interrupt routine--hinted at
and researched by Chas Williams <chas@cmf.nrl.navy.mil>.

Sync with HEAD.

split m68k SunOS and SVR4-specific sigcode.s code to sunos_sigcode.s
and svr4_sigcode.s respectively
adjust individual m68k port's locore.s to include new sigcode files if
appropriate

Define proper macros for FP frame offsets rather than magic numbers.

Update thorpej_scsipi to -current as of a month ago
A i386 GENERIC kernel compiles without the siop, ahc and bha drivers
(will be updated later). i386 IDE/ATAPI and ncr work, as well as
sparc/esp_sbus. alpha should work as well (untested yet).
siop, ahc and bha will be updated once I've updated the branch to current
-current, as well as machine-dependant code.

Add a lock around the scheduler, and use it as necessary, including
in the non-MULTIPROCESSOR case (LOCKDEBUG requires it).  Scheduler
lock is held upon entry to mi_switch() and cpu_switch(), and
cpu_switch() releases the lock before returning.

Largely from Bill Sommerfeld, with some minor bug fixes and
machine-dependent code hacking from me.

Sync w/ netbsd-1-5-base.

Add a comment on single-processor systems about p_cpu being initialized
by fork1().

First sweep at scheduler state cleanup.  Collect MI scheduler
state into global and per-CPU scheduler state:

	- Global state: sched_qs (run queues), sched_whichqs (bitmap
	  of non-empty run queues), sched_slpque (sleep queues).
	  NOTE: These may collectively move into a struct schedstate
	  at some point in the future.

	- Per-CPU state, struct schedstate_percpu: spc_runtime
	  (time process on this CPU started running), spc_flags
	  (replaces struct proc's p_schedflags), and
	  spc_curpriority (usrpri of processes on this CPU).

	- Every platform must now supply a struct cpu_info and
	  a curcpu() macro.  Simplify existing cpu_info declarations
	  where appropriate.

	- All references to per-CPU scheduler state now made through
	  curcpu().  NOTE: this will likely be adjusted in the future
	  after further changes to struct proc are made.

Tested on i386 and Alpha.  Changes are mostly mechanical, but apologies
in advance if it doesn't compile on a particular platform.

Introduce a new process state distinct from SRUN called SONPROC
which indicates that the process is actually running on a
processor.  Test against SONPROC as appropriate rather than
combinations of SRUN and curproc.  Update all context switch code
to properly set SONPROC when the process becomes the current
process on the CPU.

defopt FPSP.

Back out previous; it's now dead code.

Add get_pte_s() to search the supervisor space.

Pull up to last week's -current.

bring in latest KAME (as of 19991130, KAME/NetBSD141) into kame branch
just for reference purposes.
This commit includes 1.4 -> 1.4.1 sync for kame branch.

The branch does not compile at all (due to the lack of ALTQ and some other
source code).  Please do not try to modify the branch, this is just for
referenre purposes.

synchronization to latest KAME will take place on HEAD branch soon.

Sync with -current

Sync with main branch.

Update an outdated comment.

Since nothing depends on interrupts during autoconfig proper
any longer, we can finally avoid enabling interrupts until the
end of cpu_configure().  This resolves a long-standing issue that
has caused us trouble several times in the past.

Sync with main branch.

Sync trace trap handling with other m68k ports, originated by
ITOH Yasufumi.  Coincidentally closes PR 7738.

Centralize the declaration and clearing of `cold'.

fix botch in pullup:
>pullup 1.114->1.117 (scottr): re-implement delay()
Requested by briggs.

Deal with the fact that ipls (specifically mac68k_clockipl) are now in an array.

Sync w/ -current.

pullup 1.114->1.117 (scottr): re-implement delay()

Unfortunately, several changes that are intermingled:
- Add initial IOP support.  ADB doesn't work yet for me, but it's here so
  that others will be encouraged to work on it.  ADB_HW_IOP basically
  is configured as a NOP so that serial consoles will continue to work.
- Roll via1_intr and via2_intr into the intr.c scheme--this also required
  changing rtclock_intr to grovel the stack differently so that hardclock
  gets the right arguments and softclock() doesn't get all reentrant.
- Make via1 interrupts parallel to via2 interrupts--handlers get a pass-
  through pointer and we can register handlers.  Register via1 interrupt
  with intr_establish()--normally level 1, level 6 for A/UX scheme.
- Use intr_establish() to set real via2 interrupt handler instead of the
  hacked function pointer.
- Reorganize adb-direct interrupts so that a function call is removed.
- Implement A/UX interrupts for all Quadras right now.  We may need to
  special case some Quadras, but Linux folks are reporting success on
  several models.
- Fix intrnames to be accurate for the normal, PSC, and A/UX interrupt
  configurations.

Sync w/ -current.

Sync with main branch.

Eliminate more overhead in delay(); there's no point in doing the same
thing in both the calibrator function and the actual delay function.  At
this point, _delay() does all of the work, so we can garbage collect
dummy_delay() and make _delay() an alternate entry point for delay().

This results in a small (but measurable) improvement on the IIci.

Sync with main branch.

Tweak previous:  use shorter instructions to accomplish the same
#task (minimizes overhead).

Sync with main branch.

Write a new low-level assembly help for delay() and the calibrator.  This
eliminates stalls during instruction prefetch and makes the delay
consistent regardless of kernel configuration.

Sync with main branch.

pullup 1.113->1.114 (christos): conflicts manually edited

Include opt_compat_sunos.h

Add COMPAT_SVR4 for m68k.

completely remove Mach VM support.  all that is left is the all the
header files as UVM still uses (most of) these.

mac68k part of pr 6152 fix, and missing bits for the others

Pull up a patch consisting of revs 1.99 and 1.106, backported to 1.3-branch.
Rewrite doboot() so that it turns off the MMU, rather than relying
on the ROM space being mapped where it would like to be.  (scottr)

Changes to support fork_kthread():
- cpu_set_kpc() now takes void *arg third argument, passed to the
  entry point.
- cpu_fork() allows parent to be non-curproc iff parent is proc0.
  When forking non-curproc, assume its state has already been saved.
- Adjust various pieces of machine-dependent code to account of all of this.

Garbage collect trap #1 and trap #2 handlers; they're moved to
trap_subr.s

Pull in opt_compat_netbsd.h

Adjust for the new "reaper" kernel thread: do not free the vmspace and
u-area in machine-dependent code.  Instead, call exit2() to schedule
the reaper to free them for us, once it is safe to do so (i.e. we are
no longer running on the dead proc's vmspace and stack).

Remove the 'fast' return from intrhand().

New framework for handling processor interrupts, derived in part from
the hp300 port.

 - Interrupts 3-6 use this immediately.  Interrupt 7 is a special case,
   and the VIA interrupts (1 and 2) will be addressed when that code is
   rototilled.

 - Modify the zs front end to register with the appropriate interrupt
   controller:  through the PSC on the AV Quadras, and direct to
   interrupt 4 on the rest.  Arrange to have the appropriate zsc_softc
   supplied to us at interrupt time.

 - Modify the direct ADB driver (and its PowerManager cousin) to call
   intr_dispatch(), rather than zshard().  XXX This is a kludge, but at
   least limits the brokenness to the ADB drivers, now.

As a side effect, this should fix PR 5590.  Thanks to Bill Studenmund for
correctly determining the cause of the problem reported there.

New and improved console framebuffer initialization and autoconfig.  This
resolves a great many issues, including Performa 58x interrupt handling
and offset displays on some models.

Programs that depend on the old (pre-NetBSD 1.1) grf interface may break.
That's actually a separate issue uncovered by this code, not caused by it.

Fix a problem with the reworked reboot code on the IIci:  don't try to
reference variables after the MMU is disabled.

defopt DDB.

Fix warnings from recent ddb_init() change: handle end and esym
consistently.

Back out the defopt for M680?0. As pointed out to me, this breaks ports which
dont use the M680?0 the way Amiga, Atari, Mac68k, and X68k do. I must have
missed this in the past discussion about this project.

Rewrite doboot() so that it turns off the MMU, rather than relying
on the ROM space being mapped where it would like to be.

Move M680[2346]0 to opt_m68kcpu.h.
XXX Some explicit dependencies could be removed now from the individual
arch/$ARCH/conf/Makefile.$ARCH files, but this is still to be done.

Checkpoint of UVM work to date.  This includes a fully-functioning
MACHINE_NEW_NONCONTIG interface implementation, which is now enabled
on all systems.  Thanks to Jason Thorpe for his work on the hp300 port,
from which most of the code used here was derived.

XXX In spite of the fact that it works on the hp300, UVM does not (yet)
work well under load on mac68k.

Lay groundwork for KGDB by pulling in the trap 15 handler from hp300.

Minor adjustment for genassym.cf change.

Make space for want_resched here, like most of the other m68k-based ports.

Fix a bogosity apparently inherited from when the Utah 4.3BSD code base
was converted to use Mach VM for Net2/4.4BSD.  The user segment table
pointer was originally stored in the PCB.  When Mach VM came along,
however, it was also stored in the pmap, and loaded into the PCB in
pmap_activate().  pmap_activate() would then note that the PCB's USTP
was now in sync with the pmap's USTP, and the low-level context switch
code would use the value from the PCB.

However, pmap_activate() would also load the hardware MMU context if
the pmap was the current pmap (or, in the case where pmaps can be shared,
such as in NetBSD, if the proc was the current proc).  The low-level
context switch code would then reload the hardware _again_ using the
USTP from the PCB.

However, the optimization of not calling pmap_activate() if "stchanged"
was false ended up causing some processes to use stale USTP values from
the PCB when the low-level context switch code reloaded the hardware!
This was noticed by using a real vfork(2) (which worked for some time
before failing, surprisingly!)

Since I'm hard pressed to find any real optimization here (since the
hardware was always reloaded once, sometimes twice!), the code now always
calls pmap_activate(), which uses the correct USTP value (the one in the
pmap).  The PCB's USTP is now ignored, and should eventually be g/c'd.

Another optimization can actually be performed, and I have added a comment
describing what it is, but have not yet implemented it.

Also note that most of the loadustp() functions where actually incomplete.
This has been corrected.  These functions should probably be split up into
MMU-specific operations, and called indirectly, rather than doing constant
run-time decision making based on values that will never change during the
course of a boot's lifetime.

- Make pmap_activate() and pmap_deactivate() take a struct proc *.
- Define active_user_pmap() and use it in the appropriate places (from hp300
  port).

Pull rev 1.90 and 1.91 up from trunk (scottr)

For level 4 (serial) interrupts: If the handler returns 0, take the
normal rei course.  If the handler returns non-zero, just rte.
This should allow better MACE response-time and still keep serial
interrupt overhead to a minimum on older, slower machines.

Add in the if_mc ethernet driver for the AMD "MACE" part.  This is the
internal ethernet on the Quadra/Centris 660av/840av.
Add initial support for the PSC (DMA controller) to support the above
(DMA SCSI remains unsupported).  This involved also changing the way
that several interrupts are handled.
Above from David Huang <khym@bga.com>
Since the interrupts changed somewhat, we must also make the ipls
dynamic, defaulting to their prior levels and adjusted for the AVs.
I modelled this on the hp300.

Update marc-pcmcia branch from trunk.

For consistency:  VIA_initialize() -> via_init()

Move VIA_initialize() from configure() to locore.s, shortly after
setting up and enabling the MMU.  This has the effect of ensuring
that any stray interrupts that get posted before the IPL is lowered
will be handled gracefully, i.e. without an access fault or illegal
instruction exception.

Update marc-pcmcia branch from trunk.

Initialize the bus error and address error vectors explicitly for all
CPUs, and set up the VBR just to be certain.

Update marc-pcmcia branch from trunk.

Do run-time determination whether to save the TC value for AV
Macs; this way at least those models can configure intvid0.
From Dave Huang.

Update marc-pcmcia branch from trunk.

Get the correct fault address on a bus error on a 68040.

Remove a comment that has not been applicable for some time.

addql does set condition codes, so we are not allowed to put it in between
the andl and the jeq.
Mea maxima culpa.

Optimize the bus error handlers a bit; save a couple cycles by dropping
through a branch rather than taking it in the case of an MMU fault,
which also has the desirable side effect of possibly avoiding an
I-cache miss.

Split up CPU-dependent bus error/address error handlers.  Mostly from
the Amiga port, but with an extension to save the actual fault
address (in m68k_fault_addr) when we catch a bus error.

Get several cleanup chores out of the way.  The code is functionally
identical to the previous incarnation.

 - Update using m68k asm.h macros
 - Move initialization towards the front of the file
 - Rename mac68k_buserr_addr to m68k_fault_addr
 - Reorganize trap 15 handler, similar in structure to -- though not as
   complete as -- the hp300 version
 - Reorganize doboot() for easier integration of external cache, and
   make room for the latter (#ifdef __notyet__)
 - General garbage collection of unused code/data

Eliminate references to vmspace.vm_pmap

Adjust for <m68k/m68k/trap_subr.s>

- Use common mac68k sigcode.
- Garbage-collect icode; hasn't been used in eons.

Use <m68k/m68k/proc_subr.s>

Use <m68k/m68k/support.s>

Garbage-collect extra copypage()

Use only FC $1 in TT0 in get_pte(), and map the entire address space.  We
can then access the page table entries directly using user data space with
the MOVES instruction.  This should resolve the problem on the Duo 2x0
series with hard hangs in this function.

Warning:  do not call any function that accesses user data space (FC $1)
from get_pte()!

Sync with current 040-related development:

 - Figure out what CPU we have, rather than assuming the Booter knows,
   and set cputype variable accordingly.
 - Carefully rearrange the order of various MMU/ATC operations so as to
   eliminate possible side effects.  Also, re-order the initial bootstrap
   code to make clear what various sections of this code do.
 - Adjust to combined initenv()/genenvvars() (the latter has assimilated
   the former).
 - Add a function code argument to ptest040().
 - Use symbolic constants from assym.h where appropriate (and defined).
 - Some style cleanup.

Fix an inconsistency that came in with Lite: setrq() was renamed to
setrunqueue(), but remrq() was never renamed.  Rename remrq() to
remrunqueue().

Some changes from hp300 (thanks Jason!):
- Initialize proc0.p_addr just after setting up the kernel stack, to avoid
  getting NULL pointers in trap().  Change suggested by Gordon Ross.
- Panic if main() returns.

Implement ptest040(), a helper for get_physical().

Disable 040 caches in doboot(), and some minor stylistic changes to make
the hand-coded assembly consistent throughout.

Convert `fpu_type' to the more consistent `fputype', and use the new
#defines from m68k/include/cpu.h.

Sync with m68k/cpu.h changes.  Also, test fpu_type instead of mmutype
when handling fp exceptions.

port-m68k/2547: wrong bus error detection from is@beverly.rhein.de.

From HEAD:
Implement suggestion from is -- handle _fpfault differently for 040 and better.

Implement suggestion from is -- handle _fpfault differently for 040 and better.

Set _kernel_text as recommended by Gordon--set before vector table.

Note that rtclock_intr() could be tweaked, but why it isn't.

Wrap rtclock_intr in splclock()/splx() to make sure that hardclock is running
at splclock.  Passed along from Gordon via Scott.

Optimize user-side TLB invalidation when !defined(M68020).

Use 851-specific instructions only #if defined(M68020).

For a bus error in nofault, if it's an 040, load the address from the
correct frame offset.  This is required for the 5380 driver to work
properly.

Back out last change.  Bad karma to initialize the VIA that early.

New delay() implementation from Scott Reynolds (scottr@edsi.org).
Also some minor rearranging of the VIA initialization (init earlier).

Fix #includes.

assym.s -> assym.h
(Some ports did this already.)

Fixes from PR #1931 w/ minor mod in vm_machdep.c

Move bcopy/ovbcopy/memcpy into m68/m68k/copy.s, which is used by all
m68k ports already anyhow.  No use in keeping 6 identical copies of this
function around.

Nuke blkclr and use thereof.

Get rid of libkern duplicates.

Fix a spelling slight.  Grab the faulting address if nofault in bus error.

Simplify doboot().

Very minor cleanup.

Load the old MacOS ROMBase value back in before we jump to reset.

Fix oversight in last.

Get rid of cpu040 for good.

Fix a few 040-isms, remove an old comment, and add a few things for the future.

Stop recursive jmp0panics.

Clean up a lot.  Run machdep.c through indent.
Remove a bunch of old debugging code and some dead code.
Put in some new comments.  Need more.

Ooops.

Some random cleanup and some cleanup of the dump routines, modelled on
i386.  Untested.  Remove old mac68k-specific stack trace code, etc. as
the cleanup.

Do something a lot closer to the right thing for FPU_EMULATE/FPSP/040
and subsets thereof.

Bring in 040 code from hp300 port.
Remove memory mapping code.
Update cpu_switch and a few other functions from hp300 port.

Add a memcpy entry point as at least a temp fix for compiling with a
recent gcc2 snapshot.

Nuke FPCOPROC and add fpu device driver and FPU_EMULATE from sun3.
Still need to figure out how to handle this w/ FPSP on 68040's but
since we're not really using the 68040 yet...

Clean up via.c a bit.  Still needs lots of work.

Trim the interrupt counter space some and add _eintrnames and _eintrcnt
for vmstat.

Align sigcode.

Fix major brain damage from a while back.  Old damage broke jump to zero
panic, leading to mass confusion.

Mac ROM Glue changes from grantham@tenon.com.

Fixup a couple of missed fixes when updating to hp300/amiga
interpretation of mmutype.  Thanks to Lawrence Kesteloot for
catching the first of these.

Make this work.  Kludge from grantham.

new RCS ID format.

from trunk, per briggs

Snarf chopps changes for bus/addrerr handling.

update from trunk

Added Sysseg as argument to remap_MMU() so it can map the ROMs.

We used to disable the MMU before jumping to the ROM routine to
reset the machine.  This doesn't work on some '030s because maybe
not PA == VA where PC is.  Now if '030, don't disable MMU, just
use tt0 register to map ROM segment transparently.  Also removed
superfluous @ from new get_pte code.

Make this assemble...  pte_tmp1 -> pte_tmp and remove ref to
serial_boot_echo.

Added get_pte() routine to get a pte from a logical address using ptest.
Cleaned up startup code in a big way.  Now calling get_mapping() if
'030 and MMU is on, and post-processing our page tables with remap_MMU().

update, as requested by briggs

Be somewhat better at machine dependence, clean up a hair, and get rid
of addupc.

Comment out macserinit and macinit until we can replace them.  They're
for debugging before we have an initialized console, primarily.

file locore.s was added on branch netbsd-1-0 on 1994-07-24 01:23:32 +0000

Whoops.  Forgot to make rei globl in this copy.

Use newer fpsp code.  Nuke lots of glue in locore.s, change vectors
directly if FPSP and move back to HP numbering for mmutype which is
used as cputype for FPSP.

Make sure TT's are disabled along with the MMU.
TBIA for paranoia's sake.

ifdef out FPSP support to avoid having to use fpspnull.
Fix a bogon.

Turn off MMU early on '040 machines.  Put gray_bars around this in case
it hangs on some machine.  (Q700 is mapped PA==LA in MacOS--at least in
this version of the MacOS).

Huge, gigantic changes...  Including:
68040 support--some by me, much from Amiga...
some cleanup
The hp300 rei
A number of frame-related fixes.  Finally all found, I think, thanks to
the sharp eyes/mind of Brad Grantham, Lawrence Kesteloot, and Steve Allen.

The upshot is that we're finally booting again on the machines that
used to work.

Get things to compile with latest changes.

Add '040 stuff from Amiga.
Modified TBIA and friends which were all basically flushing everything
to satisfy year-old paranoia.

Small mods.  Needs lots of cleanup.
Added one change to make sure we get symbols in DDB.

Numerous changes by lkestel and grantham to match those in machdep.c.

Fixup calls to profclock and softclock.

Call hardclock w/ clockframe * instead of clockframe.

DTRT w.r.t. root and boot disk devices.  This could still use some work,
but it's a lot better than depending on root being on sd0 (and swapping
if it isn't).

Ditch ENTRY/ALTENTRY defines and include "m68k/asm.h" instead.

ifdef out copyinstr/copyoutstr/... functions that are now defined in copy.s.

Cleanup.  Use mac68k_xxx macros instead of macII, etc.  Cleanup via.h.
Add minimal support for ddb.
BIG CHANGE: Use proper offsets for new gas in locore.s!!!

Use /**/ instead of ## for concatenation.  Other way doesn't work right.

Update from 0.9-ish to -current.  Including a switch to use config.new.
The new functions (ptrace, etc) are stubbed.
Some cleanup, too.

The current MacBSD architecture sub-directory for NetBSD-0.9--more or
less.  It needs some help to get to be -current.  Support for minimal
hardware on the SE/30, II, IIx, and IIcx exists.

Initial revision