I'm now maintaining aund separately from NetBSD, so it's no longer
appropriate for it to live in othersrc.

(bjh21)

Replace shutdownhooks with pmf in sec(4).

(bjh21)

Make arckbd(4) compile even if the kernel is lacking wsmouse(4), wskbd(4),
rnd(4), or any combination of thoses.  Thanks to Havard Eidnes for reporting
the build breakage.

(bjh21)

I've just discovered config_found_ia(), which means that wskbd(4) and
wsmouse(4) can attach directly at arckbd(4) and I can dispose of the
ugly arcwsmouse(4) and arcwskbd(4) interpositions.  Do that, and purge
them from the documentation as well.

(bjh21)

Split device_t from wd33c93_softc.  Reviewed by cube; tested on sgimips
by tsutsui.

(bjh21)

The introduction of config_stdsubmatch() lost an important feature of
upc_submatch() whereby it made sure that the correct driver attached.
Since this didn't really belong in the submatch function anyway,
reintroduce it in the match functions for upc's children.

This allows my A5000 to find at least one of its hard disks.

(bjh21)

Make sure that dp8390_intr() gets passed the softc pointer, not the device
pointer, when it's called.

(bjh21)

Convert more printfs to aprint_* and use the latter more consistently.

(bjh21)

Bump date for last change.

(bjh21)

sec(4): use device_t and cfdata_t rather than explicit structure pointers.

(bjh21)

Autoconfiguration cleanup for eca(4):
- Use device_t and cfdata_t.
- Split device_t out of eca_softc.
- Use aprint_*() rather than printf().
- Use device_xname().

(bjh21)

Split device_t from softc in arcvideo(4).  Also tidy a rogue printf().

(bjh21)

Split device_t out of softc, with one small fix where something wanted the
device_t but was getting the softc.

(bjh21)

Fix a bit of layout.

(bjh21)

Be consistent about what the void* cookies we pass to and from wscons contain,
in preparation for splitting softc and device_t.

(bjh21)

Use device_xname() rather than dv_xname.

(bjh21)

Autoconfiguration cleanup:
- Use device_t and cfdata_t rather than explicit structure pointers.
- Use aprint_*() rather than printf().
- Use device_*() rather than directly accessing structure members.
- Split device_t out of podulebus_softc.

(bjh21)

Split device_t from ioc_softc.

(bjh21)

Now that I've fixed the bootstrap IIC stuff, split device_t out of softc.

(bjh21)

Add ARM_LOCK_CAS_DEBUG.

(bjh21)

A commit I missed from the recent migration to arm/fusu.S.

(bjh21)

Adapt the arm32 fusu implementation for use on PROG26 configurations
and use it on acorn26.

(bjh21)

Move arch/arm/arm32/fusu.S to arch/arm/arm/fusu.S in preparation for making
it usable on acorn26 as well.

(bjh21)

Add what I hope is a useful feature whereby, in DIAGNOSTIC builds, trap
handlers leave a useful sentinel value in R14_svc instead of their own
return address.  This should mean that if something causes a trap in SVC
mode without saving R14 first, it'll cause a nice obvious panic rather than
a hang or worse.  Of course, if it's using R14 as a temporary, there may
still be some confusion.

(bjh21)

Need <arm/cpu.h> for GET_CURPCB.
Also, while it's true that at the moment struct user only contains
struct pcb, we probably shouldn't rely on that.

(bjh21)

Need <arm/cpu.h> for GET_CURPCB.

(bjh21)

Now that GET_CURPCB and GET_CURCPU are generally available, use them
unconditionally rather than rolling our own (often badly) all over the
place.

(bjh21)

Let the assembler generate my literal pools for me, since that makes things
more readable.

(bjh21)

Define a version of GET_CURCPU, GET_CURLWP, and GET_CURPCB for use on
old uniprocessor ARMs, and re-arrange the ifdefs so that it gets
used on 26-bit systems as well.  This allows us to uniformly use these
macros everywhere.

(bjh21)

Clear out the vestigial support for the Simtec Hydra board.  Now that we're
getting support for proper SMP ARMs, keeping non-functional support for
a seriously crippled not-really-SMP system is likely to cause more problems
than it's worth, even if it was fun for a while.

(bjh21)

Use more fields from curcup(), specifically ci_dev (replacing the local
the_cpu) and the various fields set from the CPU ID register.

(bjh21)

Small tweak to kernel RAS restart code: there's no need to restart if PC
is already at the first instruction of the RAS.

(bjh21)

Remove the implementation of cpu_intr_p() from here now that it's in
arm/arm_machdep.h.  Should fix builds on arm32 platforms.  Thanks to
Matt Fleming for reporting the problem.

(bjh21)

Simplify the pre-ARMv6 code.  Specifically:
* Take advantage of conditional instructions to avoid branching.
* Tweak register allocation to remove some MOVs from the debug code.
* In the debug code, only the address of the counter varies between
  the success and failure cases, so make the rest of the code common.

Also correct the initial comment: _lock_cas returns the old value, not a
boolean.

(bjh21)

cpu_info_store, machine, and machine_arch can all be shared by acorn26 and
arm32, so share them.

(bjh21)

acorn26 and arm32 have identical versions of cpu_need_resched().  Unify them.
Similarly, with a small tweak acorn26 can use the same cpu_intr_p() as arm32,
so those can be unified too.

(bjh21)

Remove a couple of spurious instructions from irq_entry -- there's
no need to copy SP into R0 to pass to irq_handler() when R0 already
contains SP from a few instructions earlier.

(bjh21)

ILL_ILLTRP seems like the right code for a SWI that's not in the range
assigned for syscalls, so use it.

(bjh21)

Finally purge __P() from sys/arch/acorn26.

(bjh21)

Nothing uses set_r13_irq().  Indeed, NetBSD/acorn26 doesn't use R13_irq
at all.  G/c the function.

(bjh21)

atomic_set_bit() and atomic_clear_bit() are unused.  Garbage-collect them.
If anyone wants such functionality, it's trivially synthesised with
atomic_cas_uint() anyway.

(bjh21)

Use a compatible format for printing suseconds_t.

(bjh21)

Make the restartable atomic sequence for _lock_cas() actually atomic and
restartable on acorn26.

(bjh21)

Add support for encoding RISC OS metadata into ISO 9660 filesystems.
This is done by adding the general ability to stuff non-SUSP data into
the end of the System Use field of a Directory Record, which required
some amount of rewriting of the SUSP support.  I think the result is
at least as good as what came before, and I've fixed at least one bug
along the way.  Tested against RISC OS 3.70 on my Risc PC.

Oh, why do we want it?  It should allow us to make acorn{26,32} CDs that
can be booted directly from RISC OS without mucking around copying the
bootloader to a native filing system.

(bjh21)

Fix another bug in SUSP "CE" handling: when the SUSP records fit into
the System Use field with fewer then 28 bytes to spare, we were
remembering the wrong length for the System Use field and hence
emitting a corrupt directory entry.  This could be triggered by trying
to build a filesystem containing a regular file with a 120-byte name.
Now we're a little more careful.

(bjh21)

When splitting a System Use Area between a System Use field and a
Continuation Area, be sure to insert the "CE" record at the correct
point in the chain of SUSP records.  Without this, makefs emits
an invalid image if any file needs a Continuation Area.  The bug
seems to have been introduced with the conversion to TAILQs.

(bjh21)

Typo: debuging -> debugging.

(bjh21)

If the user specifies a cd9660-specific option that needs an integer value
without supplying a value, complain at them instead of segfaulting.
Or, to put it another way, check that the value is non-NULL before calling
set_option().

(bjh21)

Autoconfiguration cleanup for arcvideo(4):
- Use device_t and cfdata_t rather than explicit structure pointers.
- Use device_private() rather than a cast to convert device_t to a softc.
- Use device_*() to get at the innards of device_t.
device/softc split not done yet because of worrying about bootstrap console
stuff.

(bjh21)

Autoconfiguration cleanup for arcpp(4):
- Split device_t out of softc.
- Use device_t and cfdata_t rather than explicit structure pointers.
- Use device_*() rather than accessing parts of device_t directly.
- Use aprint_*() rather than printf().
Also ANSIfy arcpppushbytes.

(bjh21)

Add arcpp(4) and eca(4) since they are aren't horribly broken.

(bjh21)

Rework the bootstrap IIC stuff used by cmos_read() so that it works
properly, albeit only after iociic0 has been attached.  Happily,
cmos_read() isn't needed before that anyway.

(bjh21)

Make Econet code compile again.

(bjh21)

Autoconfiguration cleanup for latches(4):
- Split struct device out of the softc.
- Use device_t and cfdata_t rather than explicit structure pointers.
- Use aprint_*() rather than printf().
- Use device_private() to convert device_t into softc.

(bjh21)

Autoconfiguration cleanup for iociic(4):
- Use device_t and cfdata_t rather than explicit structure pointers.
- Use aprint_*() rather than printf().
- Pass "self" to config_found_ia rather than &sc->sc_dev.
Also an amount of KNF.
I haven't actually split the device structure off from the softc because
iociic_bootstrap_cookie() expects to allocate fake versions of both at once,
and I've not yet decided what it's best to do about that.

(bjh21)

Autoconfiguration cleanup for arckbd(4):
- Use device_t and cfdata_t rather than explicit structure pointers.
- Split device and (non-existent) softc for arcwskbd and arcwsmouse.
- Use aprint_*() rather than printf().
- Use device_*() to access fields of device_t.
- Use device_private() rather than a cast to convert from device_t to softc.
device/softc split not done for arckbd(4) itself because there are lots
of void* cookies that get treated as arckbd_softc at one end and device_t
at the other, and they'll need a little thought to clean up.

(bjh21)

Autoconfiguration cleanup for upc(4) at iobus(4):
- Use device_t and cfdata_t rather than explicit structure pointers.
- Use device_*() to access parts of device_t.
- Use device_private() to convert device_t into a softc.
device and softc not split because that will involve venturing outside
sys/arch/acorn26 and I want to get that cleaned up first.

(bjh21)

Autoconfiguration cleanup for ioc(4):
- Use aprint_*() rather than printf().
- Use device_t and cfdata_t rather than explicit pointer types.
- Use device_* to get at parts of device_t.
- Use device_private to convert device_t into ioc_softc *.
Note that I haven't yet separated the softc from the device structure because
this driver will need a little care in doing that.

(bjh21)

Autoconfiguration cleanup:
- Use aprint_*() rather than printf().
- Split device_t from iobus_softc and remove the latter since it's empty.
- Use device_t and cfdata_t rather than explicit structure pointers.

(bjh21)

Autoconfiguration cleanup:
- Use aprint_*() rather than printf().
- Split device_t out of cpu_softc, then remove the latter entirely since that
  left it empty.
- Use device_t and cfdata_t rather than explicit structure types.

(bjh21)

Replace the local current_spl with the ARM-wide curcpu()->ci_cpl, since
otherwise a DEBUG kernel panics at the first AST, though I don't think
any non-DEBUG code cares.

(bjh21)

Add a TESTING kernel configuration that includes GENERIC with useful
diagnostic options turned on and any local drivers that aren't ready
for GENERIC yet but still ought to be tested.

(bjh21)

device_t/softc split for experimental powerrom driver.

(bjh21)

device_t/softc split for unixbp(4)

(bjh21)

Initialise cpu_info_store so that we start up with a valid curlwp.

(bjh21)

Fix a few bugs pointed out by GCC's warnings.

(bjh21)

Add a comment explaining what the RISC OS 3 PRM says about *Info.

(bjh21)

Cope with longer hostnames, including FQDNs as recommended by sysinst.

(bjh21)

Include <string.h> for str*().

(bjh21)