1
0
Fork 0
alistair23-linux/arch/m68k/q40
Linus Torvalds 7a932e5702 asm-generic: cross-architecture timer cleanup
This cleans up two ancient timer features that were never completed in
 the past, CONFIG_GENERIC_CLOCKEVENTS and CONFIG_ARCH_USES_GETTIMEOFFSET.
 
 There was only one user left for the ARCH_USES_GETTIMEOFFSET variant
 of clocksource implementations, the ARM EBSA110 platform. Rather than
 changing to use modern timekeeping, we remove the platform entirely as
 Russell no longer uses his machine and nobody else seems to have one
 any more.
 
 The conditional code for using arch_gettimeoffset() is removed as
 a result.
 
 For CONFIG_GENERIC_CLOCKEVENTS, there are still a couple of platforms
 not using clockevent drivers: parisc, ia64, most of m68k, and one
 Arm platform. These all do timer ticks slighly differently, and this
 gets cleaned up to the point they at least all call the same helper
 function. Instead of most platforms using 'select GENERIC_CLOCKEVENTS'
 in Kconfig, the polarity is now reversed, with the few remaining ones
 selecting LEGACY_TIMER_TICK instead.
 
 Signed-off-by: Arnd Bergmann <arnd@arndb.de>
 -----BEGIN PGP SIGNATURE-----
 
 iQIzBAABCgAdFiEEo6/YBQwIrVS28WGKmmx57+YAGNkFAl/Y1v8ACgkQmmx57+YA
 GNmCvQ/9EDlgCt92r8SB+LGafDtgB8TUQZeIrs9S2mByzdxwnw0lxObIXFCnhQgh
 RpG3dR+ONRDnC5eI149B377JOEFMZWe2+BtYHUHkFARtUEWatslQcz7yAGvVRK/l
 TS/qReb6piKltlzuanF1bMZbjy2OhlaDRcm+OlC3y5mALR33M4emb+rJ6cSdfk3K
 v1iZhrxtfQT77ztesh/oPkPiyQ6kNcz7SfpyYOb6f5VLlml2BZ7YwBSVyGY7urHk
 RL3XqOUP4KKlMEAI8w0E2nvft6Fk+luziBhrMYWK0GvbmI1OESENuX/c6tgT2OQ1
 DRaVHvcPG/EAY8adOKxxVyHhEJDSoz5GJV/EtjlOegsJk6RomczR1uuiT3Kvm7Ah
 PktMKv4xQht1E15KPSKbOvNIEP18w2s5z6gw+jVDv8pw42pVEQManm1D+BICqrhl
 fcpw6T1drf9UxAjwX4+zXtmNs+a+mqiFG8puU4VVgT4GpQ8umHvunXz2WUjZO0jc
 3m8ErJHBvtJwW5TOHGyXnjl9SkwPzHOfF6IcXTYWEDU4/gQIK9TwUvCjLc0lE27t
 FMCV2ds7/K1CXwRgpa5IrefSkb8yOXSbRZ56NqqF7Ekxw4J5bYRSaY7jb+qD/e+3
 5O1y+iPxFrpH+16hSahvzrtcdFNbLQvBBuRtEQOYuHLt2UJrNoU=
 =QpNs
 -----END PGP SIGNATURE-----

Merge tag 'asm-generic-timers-5.11' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic

Pull asm-generic cross-architecture timer cleanup from Arnd Bergmann:
 "This cleans up two ancient timer features that were never completed in
  the past, CONFIG_GENERIC_CLOCKEVENTS and CONFIG_ARCH_USES_GETTIMEOFFSET.

  There was only one user left for the ARCH_USES_GETTIMEOFFSET variant
  of clocksource implementations, the ARM EBSA110 platform. Rather than
  changing to use modern timekeeping, we remove the platform entirely as
  Russell no longer uses his machine and nobody else seems to have one
  any more.

  The conditional code for using arch_gettimeoffset() is removed as a
  result.

  For CONFIG_GENERIC_CLOCKEVENTS, there are still a couple of platforms
  not using clockevent drivers: parisc, ia64, most of m68k, and one Arm
  platform. These all do timer ticks slighly differently, and this gets
  cleaned up to the point they at least all call the same helper
  function.

  Instead of most platforms using 'select GENERIC_CLOCKEVENTS' in
  Kconfig, the polarity is now reversed, with the few remaining ones
  selecting LEGACY_TIMER_TICK instead"

* tag 'asm-generic-timers-5.11' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic:
  timekeeping: default GENERIC_CLOCKEVENTS to enabled
  timekeeping: remove xtime_update
  m68k: remove timer_interrupt() function
  m68k: change remaining timers to legacy_timer_tick
  m68k: m68328: use legacy_timer_tick()
  m68k: sun3/sun3c: use legacy_timer_tick
  m68k: split heartbeat out of timer function
  m68k: coldfire: use legacy_timer_tick()
  parisc: use legacy_timer_tick
  ARM: rpc: use legacy_timer_tick
  ia64: convert to legacy_timer_tick
  timekeeping: add CONFIG_LEGACY_TIMER_TICK
  timekeeping: remove arch_gettimeoffset
  net: remove am79c961a driver
  ARM: remove ebsa110 platform
2020-12-16 00:07:17 -08:00
..
Makefile treewide: Add SPDX license identifier - Makefile/Kconfig 2019-05-21 10:50:46 +02:00
README docs: ide: convert docs to ReST and rename to *.rst 2019-06-14 14:21:18 -06:00
config.c asm-generic: cross-architecture timer cleanup 2020-12-16 00:07:17 -08:00
q40ints.c m68k: remove timer_interrupt() function 2020-10-30 21:57:06 +01:00

README

Linux for the Q40
=================

You may try http://www.geocities.com/SiliconValley/Bay/2602/ for
some up to date information. Booter and other tools will be also
available from this place or http://ftp.uni-erlangen.de/pub/unix/Linux/680x0/q40/
and mirrors.

Hints to documentation usually refer to the linux source tree in
/usr/src/linux/Documentation unless URL given.

It seems IRQ unmasking can't be safely done on a Q40. IRQ probing
is not implemented - do not try it! (See below)

For a list of kernel command-line options read the documentation for the
particular device drivers.

The floppy imposes a very high interrupt load on the CPU, approx 30K/s.
When something blocks interrupts (HD) it will lose some of them, so far
this is not known to have caused any data loss. On highly loaded systems
it can make the floppy very slow or practically stop. Other Q40 OS' simply
poll the floppy for this reason - something that can't be done in Linux.
Only possible cure is getting a 82072 controller with fifo instead of
the 8272A.

drivers used by the Q40, apart from the very obvious (console etc.):
	drivers/char/q40_keyb.c		# use PC keymaps for national keyboards
		     serial.c		# normal PC driver - any speed
	             lp.c		# printer driver
		     genrtc.c		# RTC
		char/joystick/*		# most of this should work, not
				        # in default config.in
	        block/q40ide.c		# startup for ide
		      ide*		# see Documentation/ide/ide.rst
		      floppy.c		# normal PC driver, DMA emu in asm/floppy.h
					# and arch/m68k/kernel/entry.S
					# see drivers/block/README.fd
		net/ne.c
		video/q40fb.c
		parport/*
		sound/dmasound_core.c
		      dmasound_q40.c

Various other PC drivers can be enabled simply by adding them to
arch/m68k/config.in, especially 8 bit devices should be without any
problems. For cards using 16bit io/mem more care is required, like
checking byte order issues, hacking memcpy_*_io etc.


Debugging
=========

Upon startup the kernel will usually output "ABCQGHIJ" into the SRAM,
preceded by the booter signature. This is a trace just in case something
went wrong during earliest setup stages of head.S.
**Changed** to preserve SRAM contents by default, this is only done when
requested - SRAM must start with '%LX$' signature to do this. '-d' option
to 'lxx' loader enables this.

SRAM can also be used as additional console device, use debug=mem.
This will save kernel startup msgs into SRAM, the screen will display
only the penguin - and shell prompt if it gets that far..
Unfortunately only 2000 bytes are available.

Serial console works and can also be used for debugging, see loader_txt

Most problems seem to be caused by fawlty or badly configured io-cards or
hard drives anyway.
Make sure to configure the parallel port as SPP and remove IRQ/DMA jumpers
for first testing. The Q40 does not support DMA and may have trouble with
parallel ports version of interrupts.


Q40 Hardware Description
========================

This is just an overview, see asm-m68k/* for details ask if you have any
questions.

The Q40 consists of a 68040@40 MHz, 1MB video RAM, up to 32MB RAM, AT-style
keyboard interface, 1 Programmable LED, 2x8bit DACs and up to 1MB ROM, 1MB
shadow ROM.
The Q60 has any of 68060 or 68LC060 and up to 128 MB RAM.

Most interfacing like floppy, IDE, serial and parallel ports is done via ISA
slots. The ISA io and mem range is mapped (sparse&byteswapped!) into separate
regions of the memory.
The main interrupt register IIRQ_REG will indicate whether an IRQ was internal
or from some ISA devices, EIRQ_REG can distinguish up to 8 ISA IRQs.

The Q40 custom chip is programmable to provide 2 periodic timers:
	- 50 or 200 Hz - level 2, !!THIS CAN'T BE DISABLED!!
	- 10 or 20 KHz - level 4, used for dma-sound

Linux uses the 200 Hz interrupt for timer and beep by default.


Interrupts
==========

q40 master chip handles only a subset of level triggered interrupts.

Linux has some requirements wrt interrupt architecture, these are
to my knowledge:
	(a) interrupt handler must not be reentered even when sti() is called
	    from within handler
	(b) working enable/disable_irq

Luckily these requirements are only important for drivers shared
with other architectures - ide,serial,parallel, ethernet.
q40ints.c now contains a trivial hack for (a), (b) is more difficult
because only irq's 4-15 can be disabled - and only all of them at once.
Thus disable_irq() can effectively block the machine if the driver goes
asleep.
One thing to keep in mind when hacking around the interrupt code is
that there is no way to find out which IRQ caused a request, [EI]IRQ_REG
displays current state of the various IRQ lines.

Keyboard
========

q40 receives AT make/break codes from the keyboard, these are translated to
the PC scancodes x86 Linux uses. So by theory every national keyboard should
work just by loading the appropriate x86 keytable - see any national-HOWTO.

Unfortunately the AT->PC translation isn't quite trivial and even worse, my
documentation of it is absolutely minimal - thus some exotic keys may not
behave exactly as expected.

There is still hope that it can be fixed completely though. If you encounter
problems, email me ideally this:
	- exact keypress/release sequence
	- 'showkey -s' run on q40, non-X session
	- 'showkey -s' run on a PC, non-X session
	- AT codes as displayed by the q40 debugging ROM
btw if the showkey output from PC and Q40 doesn't differ then you have some
classic configuration problem - don't send me anything in this case