remove prototypes with no matching function
ok mpi@

Remove unnecessary <sys/selinfo.h> includes.

ok jsg

Force comport initialization for some class of device

Some com@acpi devices will fail the comprobe1() check which can lead to
hang or reboot of the machine.

Input from deraadt@ kettenis@
OK miod@

Add another check to com probe

With this bentley@'s Steam Deck no longer attaches these:
com0 at isa0 port 0x3f8/8 irq 4: ns8250, no fifo
com1 at isa0 port 0x2f8/8 irq 3: ns8250, no fifo
com2 at isa0 port 0x3e8/8 irq 5: ns8250, no fifo

From Matthias Drochner in NetBSD rev 1.154
ok kettenis@

add Synopsys Degisnware UART (dw-apb-uart) support

To fix Allwinner H6's UART problem, need to add dw-apb-uart special code.
ok kettenis@

Fix bogus return statement.

From miod@

Re-commit Exar XR17V35x serial port support previously backed out:

The Exar XR17V354 has 4 ports that have a 256-byte FIFO, use a
frequency of 125Mhz, and have a unique sleep register. A custom
interrupt handler is setup in puc for these ports so it can check a
register which reports which ports triggered the interrupt, rather
than having to run comintr for every port every time.

To avoid probing for these ports in com_attach_subr which requires
access to registers that may be reserved on certain platforms, pass
the sc_uarttype from com_puc_attach since the port type is already
known based on the puc device vendor/id.

Input from kettenis, tested in snaps

Update the console device always when attaching the real device
driver. The information by the driver is supposed more reliable than
the information which was set up earlier.

ok patrick

Detach timeouts and the softint handler before freeing memory.

As more and more teardown functions include barriers, or any kind of
context change, it is unsafe to continue to assume that such code paths
are atomic. So a good practise is to only free descriptor when we're
sure that no other context can access them.

Found while looking at visa@'s ttkqflush() replacement diff. The same
pattern is present in many USB drivers as found with Peter Stuge.

ok visa@

ttysleep(): drop unused timeout parameter

All callers sleep indefinitely.

With help from visa@.

ok visa@, ratchov@, kn@

Add support for register shift/io-width to the console code as well.

ok visa@, patrick@

Revert previous commit; the XR17V35X probe that was added accesses registers
that aren't guaranteed to be there and may even belong to a different device.
This triggers a fault on hppa machines like the C3000 for example.

The Exar XR17V354 has 4 com ports that have a 256-byte FIFO, use a
frequency of 125Mhz, and have a unique sleep register. A custom
interrupt handler is setup in puc for these ports so it can check a
register which reports which ports triggered the interrupt, rather
than having to run comintr for every port every time.

ok mlarkin deraadt

Add support for register shift/io-width. This allows us to support UARTs
that are mostly NS16550 compatible but use 32-bit instead of 8-bit registers
such as the Synopsys Designware UARTs found on many armv7, arm64 and amd64
SoCs.

tested by florian@, blum@
ok deraadt@

Remove almost unused `flags' argument of suser().

The account flag `ASU' will no longer be set but that makes suser()
mpsafe since it no longer mess with a per-process field.

No objection from millert@, ok tedu@, bluhm@

Don't pull in <sys/file.h> just to get fcntl.h

ok deraadt@ krw@

Rename Debugger() into db_enter().

Using a name with the 'db_' prefix makes it invisible from the dynamic
profiler.

ok deraadt@, kettenis@, visa@

Unifdef KGDB.

It doesn't compile und hasn't been working during the last decade.

ok kettenis@, deraadt@

remove some more sparc remnants. ok deraadt

de-zaurus com.c
"looks correct" deraadt@ guenther@ on an earlier rev

Only probe the UART type if sc_uarttype is set to COM_UART_UNKNOWN.
This allows glue drivers to set the type based on other information (such as
Oopen Firmware or device tree properties) without going through to risky
code paths.

ok deraadt@, visa@

remove uneeded proc.h includes
ok mpi@ kspillner@

wrap a long line

At resume, do not spin flushing characters in from the chip. There shouldn't
be any characters. If the chip is not actually there for some reason, we'd
be spinning so early in the resume sequence, we'd probably go mad trying to
find the reason..
ok kettenis
request: people with serial ports on their laptops, try to run a "cu" over
a suspend/resume cycle, and see if you see "input noise"

At suspend/resume time cope with timeouts, DTR on resume in a bus
bus-independent fashion.
ok kettenis

No need to treat amd64 specially; always return CN_HIGHPRI.

ok miod@

Narrow the spltty/splx pair in com_activate to only enclose the minimum
operation which requires it.

Better detection of the st16650 v1 (with the broken fifo). Gets rid of
false positives like the DUART on the MPC8347 as found on socppc.

ok deraadt@

Turns out the v445 isn't happy at all if we touch the registers that deal
with sleep mode. So skip the wakeup code as well if we are a serial console.
The port on the v445 almost certainly isn't a real st16650 and probably
doesn't implement sleep mode (perhaps someone cut the soft core down a bit
too much?). But it is indistinguishable from it, so we have to deal with it.

ok deraadt@

Skip putting a port in sleep mode when it is used as serial console. In that
case the kernel will still be using the port even after the last process closes
it. And on machines like the v445 the firmware won't be happy if the console
port is in sleep mode when the kernel reboots or halts.

ok miod@

Remove unused variable.

ok mpi@, miod@

Remove config_activate() and DVACT_ACTIVATE. PCMCIA's the only thing
that's ever used it, and it's long since been changed to use
DVACT_{QUIESCE,SUSPEND,RESUME} instead.

ok deraadt@, dlg@; miod@ also agreed with this idea when I brought it
up a few weeks ago

Fix a problem found where one SOC has a uart implementation that causes bus
errors if the fifo is read when the fifo is empty. Dont read an empty fifo.
ok deraaadt@ fgsch@

only read from the UART if there is something to read. fix beagle, otherwise
it will crash with 'External Non-Linefetch Abort (S)'.
similar diff provided by drahn@. tested by jasper@, deraadt@ ok.

ansify function definitions, and constify a few arrays while there.
no functional change.

Add resume support for serial consoles.

Tested by sthen@, ok deraadt@

Initial stab at making com@puc suspend and resume properly. Works fine
for using cu(1) between two OpenBSD machines. Probably doesn't work for
serial consoles but we don't support those on puc(4) anyway.

ok deraadt@

Move common code for waking up writers on a tty into a function.

ok deraadt matthew millert

Allow tty drivers to request larger buffers at attach time using a
max-baud-rate hint. Adjust TTYHOG (the nearly full logic) to this new
situation. The larger buffers are required by the very high speed
KDDI devices in Japan (CF com, or USB ucom) so those are the only two
drivers which currently ask for a larger buffer size.
ok yasuoka miod

Use a more moderate fifo trigger level (4) for moderately quick
(sub-38400) port speeds. Inspired by netbsd. ok fgsch krw deraadt miod

Don't #include <sys/user.h> into files that don't need the stuff
it defines. In some cases, this means pulling in uvm.h or pcb.h
instead, but most of the inclusions were just noise. Tested on
alpha, amd64, armish, hppa, i386, macpcc, sgi, sparc64, and vax,
mostly by krw and naddy.
ok krw@

Some of the line disciplines want to check for suser. Better to pass them
a process instead of using curproc. ok deraadt

Every selwakeup() should have a matching KNOTE() (even if kqueue isn't
supported it doesn't do any harm), so put the KNOTE() in selwakeup() itself and
remove it from any occurences where both are used, except one for kqueue itself
and one in sys_pipe.c (where the selwakeup is under a PIPE_SEL flag).

Based on a diff from tedu.

ok deraadt

Get rid of __HAVE_GENERIC_SOFT_INTERRUPTS now that all our platforms support it.

ok jsing@, miod@

Call selwakeup()/KNOTE() even if the queue has emptied completely.

ok miod

Use suser when possible. Suggested by miod@.
miod@ deraadt@ ok.

Add missing KNOTE() calls after selwakeup(), until we decide if the KNOTE()
calls can go directly into selwakeup() safely
long discussion with nicm, murmers of consent from tedu and miod, noone
else seems to care of kqueue is busted as long as it makes their sockets
move data fast... pretty sad.

Get rid of devact enum, substitute it with an int and coresponding defines.

This is needed for the addition of further suspend/resume actions.

Okay deraadt@, marco@.

Actual final round of timeout_add(to, n * hz) -> timeout_add_sec(to, n)
conversions.

ok kettenis@
ok krw@ (possibly for the second time :)

pccom can finally die; ok kettenis dlg drahn, tested by okan

Second pass of simple timeout_add -> timeout_add_sec conversions
This should take care of the simpler ones (i.e., timeout values of
integer multiples of hz).

ok krw@, art@

Make serial console on non-primary ports work to on i386.

tested by jbg@, "it is right" deraadt@

the code that parses arguments from the boot loader on amd64 wasnt checking
what type of console was specified, so when a glass console was used it was
still configuring serial bits. this makes it so serial config only occurs
for serial console devices.

found by jolan@, thib@, and maybe sthen@
sorry guys

MD_ISA_IOT is not defined anywhere so get rid of it.

Spotted by drahn@, ok dlg@

Keep order of "#if defined..." consistent.

ok dlg@

Clean up comments.

Cleanup serial console handling and remove some of the MD code from com.c.
Specify the serial configuration from within the MD code, rather than
passing things via macros. This will allow other platforms to more readily
change the serial console configuration.

Committing this so it actually gets tested.

ok dlg@

Increase buffer size sent to the fifo, and clamp the size correctly.
Scrub the buffer afterwards, too

Cleanup cn_pri. Change constants to more meaningful names, rather than
the hp300 related ones currently in use. CN_NORMAL becomes CN_LOWPRI,
CN_INTERNAL becomes CN_MIDPRI and CN_REMOTE becomes CN_HIGHPRI.

ok miod@

Make comconsrate initialized to TTYDEF_SPEED at compile-time, for ports
which need to setup the chip before comcnattach() is invoked.

the rsc-console on the v880 operates at 115200, not the 9600 we think we
have to force all console devices to. this change passes the right speed
to the com code, and continues to use it when the device is opened as the
console device.

figured out by kettenis@
ok kettenis@ miod@

fix compilation with KDGB
from Nicholas Marriott

ok deraadt@

Delay switching to the real serial console until we attach the serial device.
The variety of different serial devices is just too big to make a decision up
front. This should make the second serial port on four-digit B/C/J-class
workstation work as a serial console too.

ok miod@

ansify/de-register
no binary change

backout while the amd64 puc situation is rethought

move com_activate() to where it belongs

put more foo_detach() routines where they are used

In a drivers activate() entry point, if on DVACT_DEACTIVATE it does
something, then it MUST return zero on DVACT_ACTIVATE, not EOPNOTSUPP;
this very popular bug has been cut and pasted a lot of times...
ok deraadt@ mickey@

2 lines of code in most drivers, to do the timestamping; ok miod kettenis

in detach, mark device with COM_SW_DEAD, so that close() will not go
tapping additional registes. allows it to work on macppc, where doing
such a tap after unplug is very bad. drahn helped fkr@bytemine.de
ok miod and drahn

Let cons_init() and cons_init_bell() initialize the whole consdev structure,
thus removing the need for drivers to initialize cn_pri to CN_DEAD when
hardware probe fails.

Move contents of sys/select.h to sys/selinfo.h in preparation for a
userland-visible sys/select.h. Consistent with what Net and Free do.
OK deraadt@, tested with full ports build by naddy@.

fix kgdb compilation; input and ok miod@.

Share com probe code between com and pccom; allows us to get rid of awkward
code duplication in cardbus, pcmcia and puc attachments.

Joint effort with fgs@; blessed deraadt@

fix sparc64 console; same diff as miod

COM_CONSOLE & DDB fix

Allow the console support code not to be compiled in; this will be
necessary for and only used on sparc, shortly.

Reluctantely ok deraadt@

Enable IR receive mode only for infrared capable PXA2x0 UARTs. Normal
reception is blocked while the IR receiver or transmitter is enabled.

Also enable the PXA2x0 UART before clearing the FIFO (suggested by
drahn@ some time ago), and disable the UART in compwroff().

ok deraadt@

xscale UART FIFOs are 32 bytes deep; found by deraadt@

- Merge changes from pxacom back into the common com driver.

Changes to the common com driver (partly from NetBSD):
- Multiply before divide in comspeed() to avoid overflow (from NetBSD).
- Don't call getc() and send -1 if the output queue is empty.
- Take UART clock frequency as a parameter instead of using COM_FREQ.

ok deraadt@

Use bus_addr_t instead of int for port address/offset, ok mickey

remove old arc stuff and add sgi support.
ok mickey@

serial console hack from tom. who, after having been told a few times
to commit it, has not. what is going on guys?

Remove OpenBSD/pegaos, this has been announced before but delayed until
the tree was stable. No point in code supporting a company which screws
its employees being in the OpenBSD tree.

Use isa iot for pegasos com console attachment, perhaps this could
be adopted MI?

Merge tty_attach() in ttymalloc() and tty_detach() in ttyfree(). The need for
separate tty registering is gone now that sparc has switched to wscons, and
this makes the code less error-prone.

Also, remove tests for ttymalloc() failure, since it uses M_WAITOK.

ok millert@ deraadt@, tested by various people as well besides me...

Replace select backends with poll backends. selscan() and pollscan()
now call the poll backend. With this change we implement greater
poll(2) functionality instead of emulating it via the select backend.
Adapted from NetBSD and including some changes from FreeBSD.
Tested by many, deraadt@ OK

change arguments to suser. suser now takes the process, and a flags
argument. old cred only calls user suser_ucred. this will allow future
work to more flexibly implement the idea of a root process. looks like
something i saw in freebsd, but a little different.
use of suser_ucred vs suser in file system code should be looked at again,
for the moment semantics remain unchanged.
review and input from art@ testing and further review miod@

move prototype for com_raisedtr() to comvar.h (just like pccom).

terms 3 & 4 cleanup based on "terms" file

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

comparam() does not need an spltty() for it, already called there.
time-bound loops in com_common_cnputc() and lower to spltty(),
which i guess is left from times when timeouts were processed at splhigh().
jason@ tested and ok, art@ ok

ifndef __HAVE_GENERIC_SOFT_INTERRUPTS leaved a label without a statement after it. Fix.

Remove kernel support for NTP. ok deraadt@ and tholo@

com.c is not a block device, no bdev_decl() for it.

Don't enable interrupts until an open() is done. Until then the driver
is not handling them and may throw the kernel into a spurious interrupt
loop. ok art@

First round of __P removal in sys

compare pointers with NULL not 0

restore CN_REMOTE priority, as it always was there.
anyway, only i386 uses that path and only if forced anyway.
also do not do comprobe1() on hppa since that hoses the port.

Move SET/CLR/ISSET macros to param.h. fgsch@ and millert@ ok

CONADDR is defined elsewhere for hppa

oops, make the kgdb var's non-static

allow this to compile with KGDB

branches: 1.76.4;
restore some of that dead code needed for dev/cons functionality

#ifdef FOO works much better when FOO isn't misspelled.

Enable break interrupt earlier.

Next round of cleanups.
- some KNF.
- disable interrupts earlier in attach.
- give it some more time in attach to finish earlier output.
- explicit bus_space_barrier in comcnputc.
- dead code.

Remove dead code.

Remove the COM_HAYESP code. It doesn't appear to have been ever used and
it's very intrusive and makes the code less readable.

Move out the com at commulti attachment into an ownfile.
Same story as isapnp - not tested because i386 uses pccom.

Move out com at isapnp attachment into an own file.
Not tested because the only system with isapnp is i386 and i386 uses pccom.

Next step in making the com driver improvable.
Split out the com at isa attachment into its own file.

- compoll is now scheduled once for every sc instead of just one timeout for
all ports and looping over them.
- support for generic soft interrupts.

Simplify comstart a bit.

Enable #if 0'd code in comcnputc() so serial console works correctly;
art@openbsd.org

Split out comattach that does isa and isapnp attachment and initialization,
into comattach that does the isa stuff and com_attach_subr that does the
init.

Implement cnbell(), an optional entrypoint that rings the console bell; from
NetBSD. deraadt@ ok

Put back new timeouts. This time with correct timeout for compoll.

typo

back out new timeouts until I find out why they break alpha.

support puc devices with higher speeds (not tested yet)

new timeouts

KGDB for i386 from NetBSD via niklas, with tweaks and accommodations.
Tested by various, vetted by mickey & theo.

Add comcnattach that does the basic steps needed to attach
a com as a serial console. (idea from NetBSD).

If cn_tab->cn_dev == NODEV, try to find comopen in cdevsw and set cn_dev
to the right device.

branches: 1.55.2;
isa-avoidance

a few ifdef accomodations for hppa, also a few knf spaces

Enable TI16750 UART support.

in the ifdef POWERPC section, do use ppc, not p4e, for the iohandle
for the console configuration.

This file has my code in it, attach my copyright.

pasto from pccom.c

revert function accidentally committed

Detach support for com; NetBSD and me

Move over the ISA PnP support from pccom.c to com.c

fix my last commit where I committed more than intended

Remove obsolete detach code that has been dead for very long

Don't turn off uart when closing cua device if tty is waiting.
Instead just drop dtr briefly.

Fix cua device to not block on open.
Turn off chip (and dtr) on interrupted open.

panic prints a newline for you, don't do it in the panic string

Powerpc port uses com.c not just power4e.

Remove clearing of interrupt after character output in cnputci(). Clearing
the int will make the tty stop if output is currently active.

com.c needs to know about mem mapped ports in ppc systems.

ncr.c and if_de.c needs some hooks for systems with ncr and dec
ether chips on the motherboard that runs on odd clocks and doesn't
support a srom. srom is emulated in machine dependent code.

big endian support for ncr and dec ether finished. the 21040
have some more coding requirements if it is to run BE since
it does not have the descriptor endian swap configuration bit.

New probe routine and support for ST16650.

Compiles, but completely untested.

rename ddb.panic_ddb, and add ddb.console. Now you can stop console ddb entry
with a sysctl. There will be architectures and drivers that lack function,
and I trust the maintainers of those will forget to add the code..

fix cua/tty locking

Removed an unused arg

Need to separate i/o and mem on mips

Serial console trouble again...

The stupid "use console before configured" scheeme strikes again :-)

Initial implementation of support for calling hardpps() on carrier enable

the real changes by niklas... I did not merge correctly

clr TS_WOPEN if bailing on cua opens
As well, at this point I've also done fancy cvs repository mangling to
ensure that the com* files in sys/dev/ic maintain the history we had
before. (Yes, I cleaned up the 2.0 tags too..)

Missing pieces of new bus.h that CVS lost behind my back

s/bus.h/bus.old.h/ to make the transit to the bus_space bus.h an easy road
possibly taken in small steps

Break shall be handled as data == 0 and FE bit set; from Theo

cominit is used from the outside (the alpha port)

patch com speed initialization

contextual usage

unused variables

sunos-like cua behaviour; inspired by budd@cs.bu.edu in his scn.c

better FIFO enabling code; netbsd pr#2574, cliff@wport.com

pica changes backed out. does not belong here

Pica bus support added

Some cleanup for -Wall -Wstrict-prototypes

tty_attach()

sync 0521

proto change

sync with 0504 -- prototypes and bus.h

Pull in John Kohl's [jtk@netbsd.org] most recent (15Apr96) APM and PCMCIA work
(original PCMCIA framework by Stefan Grefen [grefen@convex.com]).

partial sync with netbsd 960418, more to come

NetBSD 960317 merge

Fix back wrong patches.

Merging w/ NetBSD 021796.
speaker upgraded to the current.
some changes to the VM stuff (ie kern_thread.c added and so).

From NetBSD: merge of 960217

Pass device name to interrupt establish routines so it can be recorded in
the device interrupt chain structures (isa, pci)

Move interrupt chain structure definition to <machine/psl.h> so vmstat can
get at it (i386)

Remove hack to count interrupts the old way (i386)

Last of the glue and devices for for PCMCIA support from Stefan Grefen
<grefen@convex.com> with modifications by John Kohl <jtk@kolvir.blrc.ma.us>

Patches to use Hayes ESP with com.c; from banshee@gabriella.resort.com;
netbsd pr#1865. the ESP card code is always enabled -- we will have to
see if that fails on any chipsets.

from netbsd:
The IST_* and IPL_* constants are not bus-specific; don't treat them as such.
Change splimp -> splnet in Ethernet, ARCnet, and FDDI drivers.

update from netbsd

branches: 1.1.1;
Initial revision

Remove unnecessary <sys/selinfo.h> includes.

ok jsg

Force comport initialization for some class of device

Some com@acpi devices will fail the comprobe1() check which can lead to
hang or reboot of the machine.

Input from deraadt@ kettenis@
OK miod@

Add another check to com probe

With this bentley@'s Steam Deck no longer attaches these:
com0 at isa0 port 0x3f8/8 irq 4: ns8250, no fifo
com1 at isa0 port 0x2f8/8 irq 3: ns8250, no fifo
com2 at isa0 port 0x3e8/8 irq 5: ns8250, no fifo

From Matthias Drochner in NetBSD rev 1.154
ok kettenis@

add Synopsys Degisnware UART (dw-apb-uart) support

To fix Allwinner H6's UART problem, need to add dw-apb-uart special code.
ok kettenis@

Fix bogus return statement.

From miod@

Re-commit Exar XR17V35x serial port support previously backed out:

The Exar XR17V354 has 4 ports that have a 256-byte FIFO, use a
frequency of 125Mhz, and have a unique sleep register. A custom
interrupt handler is setup in puc for these ports so it can check a
register which reports which ports triggered the interrupt, rather
than having to run comintr for every port every time.

To avoid probing for these ports in com_attach_subr which requires
access to registers that may be reserved on certain platforms, pass
the sc_uarttype from com_puc_attach since the port type is already
known based on the puc device vendor/id.

Input from kettenis, tested in snaps

Update the console device always when attaching the real device
driver. The information by the driver is supposed more reliable than
the information which was set up earlier.

ok patrick

Detach timeouts and the softint handler before freeing memory.

As more and more teardown functions include barriers, or any kind of
context change, it is unsafe to continue to assume that such code paths
are atomic. So a good practise is to only free descriptor when we're
sure that no other context can access them.

Found while looking at visa@'s ttkqflush() replacement diff. The same
pattern is present in many USB drivers as found with Peter Stuge.

ok visa@

ttysleep(): drop unused timeout parameter

All callers sleep indefinitely.

With help from visa@.

ok visa@, ratchov@, kn@

Add support for register shift/io-width to the console code as well.

ok visa@, patrick@

Revert previous commit; the XR17V35X probe that was added accesses registers
that aren't guaranteed to be there and may even belong to a different device.
This triggers a fault on hppa machines like the C3000 for example.

The Exar XR17V354 has 4 com ports that have a 256-byte FIFO, use a
frequency of 125Mhz, and have a unique sleep register. A custom
interrupt handler is setup in puc for these ports so it can check a
register which reports which ports triggered the interrupt, rather
than having to run comintr for every port every time.

ok mlarkin deraadt

Add support for register shift/io-width. This allows us to support UARTs
that are mostly NS16550 compatible but use 32-bit instead of 8-bit registers
such as the Synopsys Designware UARTs found on many armv7, arm64 and amd64
SoCs.

tested by florian@, blum@
ok deraadt@

Remove almost unused `flags' argument of suser().

The account flag `ASU' will no longer be set but that makes suser()
mpsafe since it no longer mess with a per-process field.

No objection from millert@, ok tedu@, bluhm@

Don't pull in <sys/file.h> just to get fcntl.h

ok deraadt@ krw@

Rename Debugger() into db_enter().

Using a name with the 'db_' prefix makes it invisible from the dynamic
profiler.

ok deraadt@, kettenis@, visa@

Unifdef KGDB.

It doesn't compile und hasn't been working during the last decade.

ok kettenis@, deraadt@

remove some more sparc remnants. ok deraadt

de-zaurus com.c
"looks correct" deraadt@ guenther@ on an earlier rev

Only probe the UART type if sc_uarttype is set to COM_UART_UNKNOWN.
This allows glue drivers to set the type based on other information (such as
Oopen Firmware or device tree properties) without going through to risky
code paths.

ok deraadt@, visa@

remove uneeded proc.h includes
ok mpi@ kspillner@

wrap a long line

At resume, do not spin flushing characters in from the chip. There shouldn't
be any characters. If the chip is not actually there for some reason, we'd
be spinning so early in the resume sequence, we'd probably go mad trying to
find the reason..
ok kettenis
request: people with serial ports on their laptops, try to run a "cu" over
a suspend/resume cycle, and see if you see "input noise"

At suspend/resume time cope with timeouts, DTR on resume in a bus
bus-independent fashion.
ok kettenis

No need to treat amd64 specially; always return CN_HIGHPRI.

ok miod@

Narrow the spltty/splx pair in com_activate to only enclose the minimum
operation which requires it.

Better detection of the st16650 v1 (with the broken fifo). Gets rid of
false positives like the DUART on the MPC8347 as found on socppc.

ok deraadt@

Turns out the v445 isn't happy at all if we touch the registers that deal
with sleep mode. So skip the wakeup code as well if we are a serial console.
The port on the v445 almost certainly isn't a real st16650 and probably
doesn't implement sleep mode (perhaps someone cut the soft core down a bit
too much?). But it is indistinguishable from it, so we have to deal with it.

ok deraadt@

Skip putting a port in sleep mode when it is used as serial console. In that
case the kernel will still be using the port even after the last process closes
it. And on machines like the v445 the firmware won't be happy if the console
port is in sleep mode when the kernel reboots or halts.

ok miod@

Remove unused variable.

ok mpi@, miod@

Remove config_activate() and DVACT_ACTIVATE. PCMCIA's the only thing
that's ever used it, and it's long since been changed to use
DVACT_{QUIESCE,SUSPEND,RESUME} instead.

ok deraadt@, dlg@; miod@ also agreed with this idea when I brought it
up a few weeks ago

Fix a problem found where one SOC has a uart implementation that causes bus
errors if the fifo is read when the fifo is empty. Dont read an empty fifo.
ok deraaadt@ fgsch@

only read from the UART if there is something to read. fix beagle, otherwise
it will crash with 'External Non-Linefetch Abort (S)'.
similar diff provided by drahn@. tested by jasper@, deraadt@ ok.

ansify function definitions, and constify a few arrays while there.
no functional change.

Add resume support for serial consoles.

Tested by sthen@, ok deraadt@

Initial stab at making com@puc suspend and resume properly. Works fine
for using cu(1) between two OpenBSD machines. Probably doesn't work for
serial consoles but we don't support those on puc(4) anyway.

ok deraadt@

Move common code for waking up writers on a tty into a function.

ok deraadt matthew millert

Allow tty drivers to request larger buffers at attach time using a
max-baud-rate hint. Adjust TTYHOG (the nearly full logic) to this new
situation. The larger buffers are required by the very high speed
KDDI devices in Japan (CF com, or USB ucom) so those are the only two
drivers which currently ask for a larger buffer size.
ok yasuoka miod

Use a more moderate fifo trigger level (4) for moderately quick
(sub-38400) port speeds. Inspired by netbsd. ok fgsch krw deraadt miod

Don't #include <sys/user.h> into files that don't need the stuff
it defines. In some cases, this means pulling in uvm.h or pcb.h
instead, but most of the inclusions were just noise. Tested on
alpha, amd64, armish, hppa, i386, macpcc, sgi, sparc64, and vax,
mostly by krw and naddy.
ok krw@

Some of the line disciplines want to check for suser. Better to pass them
a process instead of using curproc. ok deraadt

Every selwakeup() should have a matching KNOTE() (even if kqueue isn't
supported it doesn't do any harm), so put the KNOTE() in selwakeup() itself and
remove it from any occurences where both are used, except one for kqueue itself
and one in sys_pipe.c (where the selwakeup is under a PIPE_SEL flag).

Based on a diff from tedu.

ok deraadt

Get rid of __HAVE_GENERIC_SOFT_INTERRUPTS now that all our platforms support it.

ok jsing@, miod@

Call selwakeup()/KNOTE() even if the queue has emptied completely.

ok miod

Use suser when possible. Suggested by miod@.
miod@ deraadt@ ok.

Add missing KNOTE() calls after selwakeup(), until we decide if the KNOTE()
calls can go directly into selwakeup() safely
long discussion with nicm, murmers of consent from tedu and miod, noone
else seems to care of kqueue is busted as long as it makes their sockets
move data fast... pretty sad.

Get rid of devact enum, substitute it with an int and coresponding defines.

This is needed for the addition of further suspend/resume actions.

Okay deraadt@, marco@.

Actual final round of timeout_add(to, n * hz) -> timeout_add_sec(to, n)
conversions.

ok kettenis@
ok krw@ (possibly for the second time :)

pccom can finally die; ok kettenis dlg drahn, tested by okan

Second pass of simple timeout_add -> timeout_add_sec conversions
This should take care of the simpler ones (i.e., timeout values of
integer multiples of hz).

ok krw@, art@

Make serial console on non-primary ports work to on i386.

tested by jbg@, "it is right" deraadt@

the code that parses arguments from the boot loader on amd64 wasnt checking
what type of console was specified, so when a glass console was used it was
still configuring serial bits. this makes it so serial config only occurs
for serial console devices.

found by jolan@, thib@, and maybe sthen@
sorry guys

MD_ISA_IOT is not defined anywhere so get rid of it.

Spotted by drahn@, ok dlg@

Keep order of "#if defined..." consistent.

ok dlg@

Clean up comments.

Cleanup serial console handling and remove some of the MD code from com.c.
Specify the serial configuration from within the MD code, rather than
passing things via macros. This will allow other platforms to more readily
change the serial console configuration.

Committing this so it actually gets tested.

ok dlg@

Increase buffer size sent to the fifo, and clamp the size correctly.
Scrub the buffer afterwards, too

Cleanup cn_pri. Change constants to more meaningful names, rather than
the hp300 related ones currently in use. CN_NORMAL becomes CN_LOWPRI,
CN_INTERNAL becomes CN_MIDPRI and CN_REMOTE becomes CN_HIGHPRI.

ok miod@

Make comconsrate initialized to TTYDEF_SPEED at compile-time, for ports
which need to setup the chip before comcnattach() is invoked.

the rsc-console on the v880 operates at 115200, not the 9600 we think we
have to force all console devices to. this change passes the right speed
to the com code, and continues to use it when the device is opened as the
console device.

figured out by kettenis@
ok kettenis@ miod@

fix compilation with KDGB
from Nicholas Marriott

ok deraadt@

Delay switching to the real serial console until we attach the serial device.
The variety of different serial devices is just too big to make a decision up
front. This should make the second serial port on four-digit B/C/J-class
workstation work as a serial console too.

ok miod@

ansify/de-register
no binary change

backout while the amd64 puc situation is rethought

move com_activate() to where it belongs

put more foo_detach() routines where they are used

In a drivers activate() entry point, if on DVACT_DEACTIVATE it does
something, then it MUST return zero on DVACT_ACTIVATE, not EOPNOTSUPP;
this very popular bug has been cut and pasted a lot of times...
ok deraadt@ mickey@

2 lines of code in most drivers, to do the timestamping; ok miod kettenis

in detach, mark device with COM_SW_DEAD, so that close() will not go
tapping additional registes. allows it to work on macppc, where doing
such a tap after unplug is very bad. drahn helped fkr@bytemine.de
ok miod and drahn

Let cons_init() and cons_init_bell() initialize the whole consdev structure,
thus removing the need for drivers to initialize cn_pri to CN_DEAD when
hardware probe fails.

Move contents of sys/select.h to sys/selinfo.h in preparation for a
userland-visible sys/select.h. Consistent with what Net and Free do.
OK deraadt@, tested with full ports build by naddy@.

fix kgdb compilation; input and ok miod@.

Share com probe code between com and pccom; allows us to get rid of awkward
code duplication in cardbus, pcmcia and puc attachments.

Joint effort with fgs@; blessed deraadt@

fix sparc64 console; same diff as miod

COM_CONSOLE & DDB fix

Allow the console support code not to be compiled in; this will be
necessary for and only used on sparc, shortly.

Reluctantely ok deraadt@

Enable IR receive mode only for infrared capable PXA2x0 UARTs. Normal
reception is blocked while the IR receiver or transmitter is enabled.

Also enable the PXA2x0 UART before clearing the FIFO (suggested by
drahn@ some time ago), and disable the UART in compwroff().

ok deraadt@

xscale UART FIFOs are 32 bytes deep; found by deraadt@

- Merge changes from pxacom back into the common com driver.

Changes to the common com driver (partly from NetBSD):
- Multiply before divide in comspeed() to avoid overflow (from NetBSD).
- Don't call getc() and send -1 if the output queue is empty.
- Take UART clock frequency as a parameter instead of using COM_FREQ.

ok deraadt@

Use bus_addr_t instead of int for port address/offset, ok mickey

remove old arc stuff and add sgi support.
ok mickey@

serial console hack from tom. who, after having been told a few times
to commit it, has not. what is going on guys?

Remove OpenBSD/pegaos, this has been announced before but delayed until
the tree was stable. No point in code supporting a company which screws
its employees being in the OpenBSD tree.

Use isa iot for pegasos com console attachment, perhaps this could
be adopted MI?

Merge tty_attach() in ttymalloc() and tty_detach() in ttyfree(). The need for
separate tty registering is gone now that sparc has switched to wscons, and
this makes the code less error-prone.

Also, remove tests for ttymalloc() failure, since it uses M_WAITOK.

ok millert@ deraadt@, tested by various people as well besides me...

Replace select backends with poll backends. selscan() and pollscan()
now call the poll backend. With this change we implement greater
poll(2) functionality instead of emulating it via the select backend.
Adapted from NetBSD and including some changes from FreeBSD.
Tested by many, deraadt@ OK

change arguments to suser. suser now takes the process, and a flags
argument. old cred only calls user suser_ucred. this will allow future
work to more flexibly implement the idea of a root process. looks like
something i saw in freebsd, but a little different.
use of suser_ucred vs suser in file system code should be looked at again,
for the moment semantics remain unchanged.
review and input from art@ testing and further review miod@

move prototype for com_raisedtr() to comvar.h (just like pccom).

terms 3 & 4 cleanup based on "terms" file

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

comparam() does not need an spltty() for it, already called there.
time-bound loops in com_common_cnputc() and lower to spltty(),
which i guess is left from times when timeouts were processed at splhigh().
jason@ tested and ok, art@ ok

ifndef __HAVE_GENERIC_SOFT_INTERRUPTS leaved a label without a statement after it. Fix.

Remove kernel support for NTP. ok deraadt@ and tholo@

com.c is not a block device, no bdev_decl() for it.

Don't enable interrupts until an open() is done. Until then the driver
is not handling them and may throw the kernel into a spurious interrupt
loop. ok art@

First round of __P removal in sys

compare pointers with NULL not 0

restore CN_REMOTE priority, as it always was there.
anyway, only i386 uses that path and only if forced anyway.
also do not do comprobe1() on hppa since that hoses the port.

Move SET/CLR/ISSET macros to param.h. fgsch@ and millert@ ok

CONADDR is defined elsewhere for hppa

oops, make the kgdb var's non-static

allow this to compile with KGDB

branches: 1.76.4;
restore some of that dead code needed for dev/cons functionality

#ifdef FOO works much better when FOO isn't misspelled.

Enable break interrupt earlier.

Next round of cleanups.
- some KNF.
- disable interrupts earlier in attach.
- give it some more time in attach to finish earlier output.
- explicit bus_space_barrier in comcnputc.
- dead code.

Remove dead code.

Remove the COM_HAYESP code. It doesn't appear to have been ever used and
it's very intrusive and makes the code less readable.

Move out the com at commulti attachment into an ownfile.
Same story as isapnp - not tested because i386 uses pccom.

Move out com at isapnp attachment into an own file.
Not tested because the only system with isapnp is i386 and i386 uses pccom.

Next step in making the com driver improvable.
Split out the com at isa attachment into its own file.

- compoll is now scheduled once for every sc instead of just one timeout for
all ports and looping over them.
- support for generic soft interrupts.

Simplify comstart a bit.

Enable #if 0'd code in comcnputc() so serial console works correctly;
art@openbsd.org

Split out comattach that does isa and isapnp attachment and initialization,
into comattach that does the isa stuff and com_attach_subr that does the
init.

Implement cnbell(), an optional entrypoint that rings the console bell; from
NetBSD. deraadt@ ok

Put back new timeouts. This time with correct timeout for compoll.

typo

back out new timeouts until I find out why they break alpha.

support puc devices with higher speeds (not tested yet)

new timeouts

KGDB for i386 from NetBSD via niklas, with tweaks and accommodations.
Tested by various, vetted by mickey & theo.

Add comcnattach that does the basic steps needed to attach
a com as a serial console. (idea from NetBSD).

If cn_tab->cn_dev == NODEV, try to find comopen in cdevsw and set cn_dev
to the right device.

branches: 1.55.2;
isa-avoidance

a few ifdef accomodations for hppa, also a few knf spaces

Enable TI16750 UART support.

in the ifdef POWERPC section, do use ppc, not p4e, for the iohandle
for the console configuration.

This file has my code in it, attach my copyright.

pasto from pccom.c

revert function accidentally committed

Detach support for com; NetBSD and me

Move over the ISA PnP support from pccom.c to com.c

fix my last commit where I committed more than intended

Remove obsolete detach code that has been dead for very long

Don't turn off uart when closing cua device if tty is waiting.
Instead just drop dtr briefly.

Fix cua device to not block on open.
Turn off chip (and dtr) on interrupted open.

panic prints a newline for you, don't do it in the panic string

Powerpc port uses com.c not just power4e.

Remove clearing of interrupt after character output in cnputci(). Clearing
the int will make the tty stop if output is currently active.

com.c needs to know about mem mapped ports in ppc systems.

ncr.c and if_de.c needs some hooks for systems with ncr and dec
ether chips on the motherboard that runs on odd clocks and doesn't
support a srom. srom is emulated in machine dependent code.

big endian support for ncr and dec ether finished. the 21040
have some more coding requirements if it is to run BE since
it does not have the descriptor endian swap configuration bit.

New probe routine and support for ST16650.

Compiles, but completely untested.

rename ddb.panic_ddb, and add ddb.console. Now you can stop console ddb entry
with a sysctl. There will be architectures and drivers that lack function,
and I trust the maintainers of those will forget to add the code..

fix cua/tty locking

Removed an unused arg

Need to separate i/o and mem on mips

Serial console trouble again...

The stupid "use console before configured" scheeme strikes again :-)

Initial implementation of support for calling hardpps() on carrier enable

the real changes by niklas... I did not merge correctly

clr TS_WOPEN if bailing on cua opens
As well, at this point I've also done fancy cvs repository mangling to
ensure that the com* files in sys/dev/ic maintain the history we had
before. (Yes, I cleaned up the 2.0 tags too..)

Missing pieces of new bus.h that CVS lost behind my back

s/bus.h/bus.old.h/ to make the transit to the bus_space bus.h an easy road
possibly taken in small steps

Break shall be handled as data == 0 and FE bit set; from Theo

cominit is used from the outside (the alpha port)

patch com speed initialization

contextual usage

unused variables

sunos-like cua behaviour; inspired by budd@cs.bu.edu in his scn.c

better FIFO enabling code; netbsd pr#2574, cliff@wport.com

pica changes backed out. does not belong here

Pica bus support added

Some cleanup for -Wall -Wstrict-prototypes

tty_attach()

sync 0521

proto change

sync with 0504 -- prototypes and bus.h

Pull in John Kohl's [jtk@netbsd.org] most recent (15Apr96) APM and PCMCIA work
(original PCMCIA framework by Stefan Grefen [grefen@convex.com]).

partial sync with netbsd 960418, more to come

NetBSD 960317 merge

Fix back wrong patches.

Merging w/ NetBSD 021796.
speaker upgraded to the current.
some changes to the VM stuff (ie kern_thread.c added and so).

From NetBSD: merge of 960217

Pass device name to interrupt establish routines so it can be recorded in
the device interrupt chain structures (isa, pci)

Move interrupt chain structure definition to <machine/psl.h> so vmstat can
get at it (i386)

Remove hack to count interrupts the old way (i386)

Last of the glue and devices for for PCMCIA support from Stefan Grefen
<grefen@convex.com> with modifications by John Kohl <jtk@kolvir.blrc.ma.us>

Patches to use Hayes ESP with com.c; from banshee@gabriella.resort.com;
netbsd pr#1865. the ESP card code is always enabled -- we will have to
see if that fails on any chipsets.

from netbsd:
The IST_* and IPL_* constants are not bus-specific; don't treat them as such.
Change splimp -> splnet in Ethernet, ARCnet, and FDDI drivers.

update from netbsd

branches: 1.1.1;
Initial revision

Force comport initialization for some class of device

Some com@acpi devices will fail the comprobe1() check which can lead to
hang or reboot of the machine.

Input from deraadt@ kettenis@
OK miod@

Add another check to com probe

With this bentley@'s Steam Deck no longer attaches these:
com0 at isa0 port 0x3f8/8 irq 4: ns8250, no fifo
com1 at isa0 port 0x2f8/8 irq 3: ns8250, no fifo
com2 at isa0 port 0x3e8/8 irq 5: ns8250, no fifo

From Matthias Drochner in NetBSD rev 1.154
ok kettenis@

add Synopsys Degisnware UART (dw-apb-uart) support

To fix Allwinner H6's UART problem, need to add dw-apb-uart special code.
ok kettenis@

Fix bogus return statement.

From miod@

Re-commit Exar XR17V35x serial port support previously backed out:

The Exar XR17V354 has 4 ports that have a 256-byte FIFO, use a
frequency of 125Mhz, and have a unique sleep register. A custom
interrupt handler is setup in puc for these ports so it can check a
register which reports which ports triggered the interrupt, rather
than having to run comintr for every port every time.

To avoid probing for these ports in com_attach_subr which requires
access to registers that may be reserved on certain platforms, pass
the sc_uarttype from com_puc_attach since the port type is already
known based on the puc device vendor/id.

Input from kettenis, tested in snaps

Update the console device always when attaching the real device
driver. The information by the driver is supposed more reliable than
the information which was set up earlier.

ok patrick

Detach timeouts and the softint handler before freeing memory.

As more and more teardown functions include barriers, or any kind of
context change, it is unsafe to continue to assume that such code paths
are atomic. So a good practise is to only free descriptor when we're
sure that no other context can access them.

Found while looking at visa@'s ttkqflush() replacement diff. The same
pattern is present in many USB drivers as found with Peter Stuge.

ok visa@

ttysleep(): drop unused timeout parameter

All callers sleep indefinitely.

With help from visa@.

ok visa@, ratchov@, kn@

Add support for register shift/io-width to the console code as well.

ok visa@, patrick@

Revert previous commit; the XR17V35X probe that was added accesses registers
that aren't guaranteed to be there and may even belong to a different device.
This triggers a fault on hppa machines like the C3000 for example.

The Exar XR17V354 has 4 com ports that have a 256-byte FIFO, use a
frequency of 125Mhz, and have a unique sleep register. A custom
interrupt handler is setup in puc for these ports so it can check a
register which reports which ports triggered the interrupt, rather
than having to run comintr for every port every time.

ok mlarkin deraadt

Add support for register shift/io-width. This allows us to support UARTs
that are mostly NS16550 compatible but use 32-bit instead of 8-bit registers
such as the Synopsys Designware UARTs found on many armv7, arm64 and amd64
SoCs.

tested by florian@, blum@
ok deraadt@

Remove almost unused `flags' argument of suser().

The account flag `ASU' will no longer be set but that makes suser()
mpsafe since it no longer mess with a per-process field.

No objection from millert@, ok tedu@, bluhm@

Don't pull in <sys/file.h> just to get fcntl.h

ok deraadt@ krw@

Rename Debugger() into db_enter().

Using a name with the 'db_' prefix makes it invisible from the dynamic
profiler.

ok deraadt@, kettenis@, visa@

Unifdef KGDB.

It doesn't compile und hasn't been working during the last decade.

ok kettenis@, deraadt@

remove some more sparc remnants. ok deraadt

de-zaurus com.c
"looks correct" deraadt@ guenther@ on an earlier rev

Only probe the UART type if sc_uarttype is set to COM_UART_UNKNOWN.
This allows glue drivers to set the type based on other information (such as
Oopen Firmware or device tree properties) without going through to risky
code paths.

ok deraadt@, visa@

remove uneeded proc.h includes
ok mpi@ kspillner@

wrap a long line

At resume, do not spin flushing characters in from the chip. There shouldn't
be any characters. If the chip is not actually there for some reason, we'd
be spinning so early in the resume sequence, we'd probably go mad trying to
find the reason..
ok kettenis
request: people with serial ports on their laptops, try to run a "cu" over
a suspend/resume cycle, and see if you see "input noise"

At suspend/resume time cope with timeouts, DTR on resume in a bus
bus-independent fashion.
ok kettenis

No need to treat amd64 specially; always return CN_HIGHPRI.

ok miod@

Narrow the spltty/splx pair in com_activate to only enclose the minimum
operation which requires it.

Better detection of the st16650 v1 (with the broken fifo). Gets rid of
false positives like the DUART on the MPC8347 as found on socppc.

ok deraadt@

Turns out the v445 isn't happy at all if we touch the registers that deal
with sleep mode. So skip the wakeup code as well if we are a serial console.
The port on the v445 almost certainly isn't a real st16650 and probably
doesn't implement sleep mode (perhaps someone cut the soft core down a bit
too much?). But it is indistinguishable from it, so we have to deal with it.

ok deraadt@

Skip putting a port in sleep mode when it is used as serial console. In that
case the kernel will still be using the port even after the last process closes
it. And on machines like the v445 the firmware won't be happy if the console
port is in sleep mode when the kernel reboots or halts.

ok miod@

Remove unused variable.

ok mpi@, miod@

Remove config_activate() and DVACT_ACTIVATE. PCMCIA's the only thing
that's ever used it, and it's long since been changed to use
DVACT_{QUIESCE,SUSPEND,RESUME} instead.

ok deraadt@, dlg@; miod@ also agreed with this idea when I brought it
up a few weeks ago

Fix a problem found where one SOC has a uart implementation that causes bus
errors if the fifo is read when the fifo is empty. Dont read an empty fifo.
ok deraaadt@ fgsch@

only read from the UART if there is something to read. fix beagle, otherwise
it will crash with 'External Non-Linefetch Abort (S)'.
similar diff provided by drahn@. tested by jasper@, deraadt@ ok.

ansify function definitions, and constify a few arrays while there.
no functional change.

Add resume support for serial consoles.

Tested by sthen@, ok deraadt@

Initial stab at making com@puc suspend and resume properly. Works fine
for using cu(1) between two OpenBSD machines. Probably doesn't work for
serial consoles but we don't support those on puc(4) anyway.

ok deraadt@

Move common code for waking up writers on a tty into a function.

ok deraadt matthew millert

Allow tty drivers to request larger buffers at attach time using a
max-baud-rate hint. Adjust TTYHOG (the nearly full logic) to this new
situation. The larger buffers are required by the very high speed
KDDI devices in Japan (CF com, or USB ucom) so those are the only two
drivers which currently ask for a larger buffer size.
ok yasuoka miod

Use a more moderate fifo trigger level (4) for moderately quick
(sub-38400) port speeds. Inspired by netbsd. ok fgsch krw deraadt miod

Don't #include <sys/user.h> into files that don't need the stuff
it defines. In some cases, this means pulling in uvm.h or pcb.h
instead, but most of the inclusions were just noise. Tested on
alpha, amd64, armish, hppa, i386, macpcc, sgi, sparc64, and vax,
mostly by krw and naddy.
ok krw@

Some of the line disciplines want to check for suser. Better to pass them
a process instead of using curproc. ok deraadt

Every selwakeup() should have a matching KNOTE() (even if kqueue isn't
supported it doesn't do any harm), so put the KNOTE() in selwakeup() itself and
remove it from any occurences where both are used, except one for kqueue itself
and one in sys_pipe.c (where the selwakeup is under a PIPE_SEL flag).

Based on a diff from tedu.

ok deraadt

Get rid of __HAVE_GENERIC_SOFT_INTERRUPTS now that all our platforms support it.

ok jsing@, miod@

Call selwakeup()/KNOTE() even if the queue has emptied completely.

ok miod

Use suser when possible. Suggested by miod@.
miod@ deraadt@ ok.

Add missing KNOTE() calls after selwakeup(), until we decide if the KNOTE()
calls can go directly into selwakeup() safely
long discussion with nicm, murmers of consent from tedu and miod, noone
else seems to care of kqueue is busted as long as it makes their sockets
move data fast... pretty sad.

Get rid of devact enum, substitute it with an int and coresponding defines.

This is needed for the addition of further suspend/resume actions.

Okay deraadt@, marco@.

Actual final round of timeout_add(to, n * hz) -> timeout_add_sec(to, n)
conversions.

ok kettenis@
ok krw@ (possibly for the second time :)

pccom can finally die; ok kettenis dlg drahn, tested by okan

Second pass of simple timeout_add -> timeout_add_sec conversions
This should take care of the simpler ones (i.e., timeout values of
integer multiples of hz).

ok krw@, art@

Make serial console on non-primary ports work to on i386.

tested by jbg@, "it is right" deraadt@

the code that parses arguments from the boot loader on amd64 wasnt checking
what type of console was specified, so when a glass console was used it was
still configuring serial bits. this makes it so serial config only occurs
for serial console devices.

found by jolan@, thib@, and maybe sthen@
sorry guys

MD_ISA_IOT is not defined anywhere so get rid of it.

Spotted by drahn@, ok dlg@

Keep order of "#if defined..." consistent.

ok dlg@

Clean up comments.

Cleanup serial console handling and remove some of the MD code from com.c.
Specify the serial configuration from within the MD code, rather than
passing things via macros. This will allow other platforms to more readily
change the serial console configuration.

Committing this so it actually gets tested.

ok dlg@

Increase buffer size sent to the fifo, and clamp the size correctly.
Scrub the buffer afterwards, too

Cleanup cn_pri. Change constants to more meaningful names, rather than
the hp300 related ones currently in use. CN_NORMAL becomes CN_LOWPRI,
CN_INTERNAL becomes CN_MIDPRI and CN_REMOTE becomes CN_HIGHPRI.

ok miod@

Make comconsrate initialized to TTYDEF_SPEED at compile-time, for ports
which need to setup the chip before comcnattach() is invoked.

the rsc-console on the v880 operates at 115200, not the 9600 we think we
have to force all console devices to. this change passes the right speed
to the com code, and continues to use it when the device is opened as the
console device.

figured out by kettenis@
ok kettenis@ miod@

fix compilation with KDGB
from Nicholas Marriott

ok deraadt@

Delay switching to the real serial console until we attach the serial device.
The variety of different serial devices is just too big to make a decision up
front. This should make the second serial port on four-digit B/C/J-class
workstation work as a serial console too.

ok miod@

ansify/de-register
no binary change

backout while the amd64 puc situation is rethought

move com_activate() to where it belongs

put more foo_detach() routines where they are used

In a drivers activate() entry point, if on DVACT_DEACTIVATE it does
something, then it MUST return zero on DVACT_ACTIVATE, not EOPNOTSUPP;
this very popular bug has been cut and pasted a lot of times...
ok deraadt@ mickey@

2 lines of code in most drivers, to do the timestamping; ok miod kettenis

in detach, mark device with COM_SW_DEAD, so that close() will not go
tapping additional registes. allows it to work on macppc, where doing
such a tap after unplug is very bad. drahn helped fkr@bytemine.de
ok miod and drahn

Let cons_init() and cons_init_bell() initialize the whole consdev structure,
thus removing the need for drivers to initialize cn_pri to CN_DEAD when
hardware probe fails.

Move contents of sys/select.h to sys/selinfo.h in preparation for a
userland-visible sys/select.h. Consistent with what Net and Free do.
OK deraadt@, tested with full ports build by naddy@.

fix kgdb compilation; input and ok miod@.

Share com probe code between com and pccom; allows us to get rid of awkward
code duplication in cardbus, pcmcia and puc attachments.

Joint effort with fgs@; blessed deraadt@

fix sparc64 console; same diff as miod

COM_CONSOLE & DDB fix

Allow the console support code not to be compiled in; this will be
necessary for and only used on sparc, shortly.

Reluctantely ok deraadt@

Enable IR receive mode only for infrared capable PXA2x0 UARTs. Normal
reception is blocked while the IR receiver or transmitter is enabled.

Also enable the PXA2x0 UART before clearing the FIFO (suggested by
drahn@ some time ago), and disable the UART in compwroff().

ok deraadt@

xscale UART FIFOs are 32 bytes deep; found by deraadt@

- Merge changes from pxacom back into the common com driver.

Changes to the common com driver (partly from NetBSD):
- Multiply before divide in comspeed() to avoid overflow (from NetBSD).
- Don't call getc() and send -1 if the output queue is empty.
- Take UART clock frequency as a parameter instead of using COM_FREQ.

ok deraadt@

Use bus_addr_t instead of int for port address/offset, ok mickey

remove old arc stuff and add sgi support.
ok mickey@