Commit b920de1b authored by David Howells's avatar David Howells Committed by Linus Torvalds

mn10300: add the MN10300/AM33 architecture to the kernel

Add architecture support for the MN10300/AM33 CPUs produced by MEI to the
kernel.

This patch also adds board support for the ASB2303 with the ASB2308 daughter
board, and the ASB2305.  The only processor supported is the MN103E010, which
is an AM33v2 core plus on-chip devices.

[akpm@linux-foundation.org: nuke cvs control strings]
Signed-off-by: 's avatarMasakazu Urade <urade.masakazu@jp.panasonic.com>
Signed-off-by: 's avatarKoichi Yasutake <yasutake.koichi@jp.panasonic.com>
Signed-off-by: 's avatarDavid Howells <dhowells@redhat.com>
Signed-off-by: 's avatarAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: 's avatarLinus Torvalds <torvalds@linux-foundation.org>
parent ef3d5347
=========================
MN10300 FUNCTION CALL ABI
=========================
=======
GENERAL
=======
The MN10300/AM33 kernel runs in little-endian mode; big-endian mode is not
supported.
The stack grows downwards, and should always be 32-bit aligned. There are
separate stack pointer registers for userspace and the kernel.
================
ARGUMENT PASSING
================
The first two arguments (assuming up to 32-bits per argument) to a function are
passed in the D0 and D1 registers respectively; all other arguments are passed
on the stack.
If 64-bit arguments are being passed, then they are never split between
registers and the stack. If the first argument is a 64-bit value, it will be
passed in D0:D1. If the first argument is not a 64-bit value, but the second
is, the second will be passed entirely on the stack and D1 will be unused.
Arguments smaller than 32-bits are not coelesced within a register or a stack
word. For example, two byte-sized arguments will always be passed in separate
registers or word-sized stack slots.
=================
CALLING FUNCTIONS
=================
The caller must allocate twelve bytes on the stack for the callee's use before
it inserts a CALL instruction. The CALL instruction will write into the TOS
word, but won't actually modify the stack pointer; similarly, the RET
instruction reads from the TOS word of the stack, but doesn't move the stack
pointer beyond it.
Stack:
| |
| |
|---------------| SP+20
| 4th Arg |
|---------------| SP+16
| 3rd Arg |
|---------------| SP+12
| D1 Save Slot |
|---------------| SP+8
| D0 Save Slot |
|---------------| SP+4
| Return Addr |
|---------------| SP
| |
| |
The caller must leave space on the stack (hence an allocation of twelve bytes)
in which the callee may store the first two arguments.
============
RETURN VALUE
============
The return value is passed in D0 for an integer (or D0:D1 for a 64-bit value),
or A0 for a pointer.
If the return value is a value larger than 64-bits, or is a structure or an
array, then a hidden first argument will be passed to the callee by the caller:
this will point to a piece of memory large enough to hold the result of the
function. In this case, the callee will return the value in that piece of
memory, and no value will be returned in D0 or A0.
===================
REGISTER CLOBBERING
===================
The values in certain registers may be clobbered by the callee, and other
values must be saved:
Clobber: D0-D1, A0-A1, E0-E3
Save: D2-D3, A2-A3, E4-E7, SP
All other non-supervisor-only registers are clobberable (such as MDR, MCRL,
MCRH).
=================
SPECIAL REGISTERS
=================
Certain ordinary registers may carry special usage for the compiler:
A3: Frame pointer
E2: TLS pointer
==========
KERNEL ABI
==========
The kernel may use a slightly different ABI internally.
(*) E2
If CONFIG_MN10300_CURRENT_IN_E2 is defined, then the current task pointer
will be kept in the E2 register, and that register will be marked
unavailable for the compiler to use as a scratch register.
Normally the kernel uses something like:
MOV SP,An
AND 0xFFFFE000,An
MOV (An),Rm // Rm holds current
MOV (yyy,Rm) // Access current->yyy
To find the address of current; but since this option permits current to
be carried globally in an register, it can use:
MOV (yyy,E2) // Access current->yyy
instead.
===============
SYSTEM CALL ABI
===============
System calls are called with the following convention:
REGISTER ENTRY EXIT
=============== ======================= =======================
D0 Syscall number Return value
A0 1st syscall argument Saved
D1 2nd syscall argument Saved
A3 3rd syscall argument Saved
A2 4th syscall argument Saved
D3 5th syscall argument Saved
D2 6th syscall argument Saved
All other registers are saved. The layout is a consequence of the way the MOVM
instruction stores registers onto the stack.
=========================================
PART-SPECIFIC SOURCE COMPARTMENTALISATION
=========================================
The sources for various parts are compartmentalised at two different levels:
(1) Processor level
The "processor level" is a CPU core plus the other on-silicon
peripherals.
Processor-specific header files are divided among directories in a similar
way to the CPU level:
(*) include/asm-mn10300/proc-mn103e010/
Support for the AM33v2 CPU core.
The appropriate processor is selected by a CONFIG_MN10300_PROC_YYYY option
from the "Processor support" choice menu in the arch/mn10300/Kconfig file.
(2) Unit level
The "unit level" is a processor plus all the external peripherals
controlled by that processor.
Unit-specific header files are divided among directories in a similar way
to the CPU level; not only that, but specific sources may also be
segregated into separate directories under the arch directory:
(*) include/asm-mn10300/unit-asb2303/
(*) arch/mn10300/unit-asb2303/
Support for the ASB2303 board with an ASB2308 daughter board.
(*) include/asm-mn10300/unit-asb2305/
(*) arch/mn10300/unit-asb2305/
Support for the ASB2305 board.
The appropriate processor is selected by a CONFIG_MN10300_UNIT_ZZZZ option
from the "Unit type" choice menu in the arch/mn10300/Kconfig file.
============
COMPILE TIME
============
When the kernel is compiled, symbolic links will be made in the asm header file
directory for this arch:
include/asm-mn10300/proc => include/asm-mn10300/proc-YYYY/
include/asm-mn10300/unit => include/asm-mn10300/unit-ZZZZ/
So that the header files contained in those directories can be accessed without
lots of #ifdef-age.
The appropriate arch/mn10300/unit-ZZZZ directory will also be entered by the
compilation process; all other unit-specific directories will be ignored.
......@@ -2614,6 +2614,15 @@ L: linux-kernel@vger.kernel.org
W: http://www.linux-mm.org
S: Maintained
MEI MN10300/AM33 PORT
P: David Howells
M: dhowells@redhat.com
P: Koichi Yasutake
M: yasutake.koichi@jp.panasonic.com
L: linux-am33-list@redhat.com
W: ftp://ftp.redhat.com/pub/redhat/gnupro/AM33/
S: Maintained
MEMORY TECHNOLOGY DEVICES (MTD)
P: David Woodhouse
M: dwmw2@infradead.org
......
#
# For a description of the syntax of this configuration file,
# see Documentation/kbuild/kconfig-language.txt.
#
mainmenu "Linux Kernel Configuration"
config MN10300
def_bool y
config AM33
def_bool y
config MMU
def_bool y
config HIGHMEM
def_bool n
config NUMA
def_bool n
config UID16
def_bool y
config RWSEM_GENERIC_SPINLOCK
def_bool y
config RWSEM_XCHGADD_ALGORITHM
bool
config GENERIC_HARDIRQS_NO__DO_IRQ
def_bool y
config GENERIC_CALIBRATE_DELAY
def_bool y
config GENERIC_FIND_NEXT_BIT
def_bool y
config GENERIC_HWEIGHT
def_bool y
config GENERIC_TIME
def_bool y
config GENERIC_BUG
def_bool y
config QUICKLIST
def_bool y
config ARCH_HAS_ILOG2_U32
def_bool y
config ARCH_SUPPORTS_AOUT
def_bool n
# Use the generic interrupt handling code in kernel/irq/
config GENERIC_HARDIRQS
def_bool y
config HOTPLUG_CPU
def_bool n
mainmenu "Matsushita MN10300/AM33 Kernel Configuration"
source "init/Kconfig"
menu "Matsushita MN10300 system setup"
choice
prompt "Unit type"
default MN10300_UNIT_ASB2303
help
This option specifies board for which the kernel will be
compiled. It affects the external peripherals catered for.
config MN10300_UNIT_ASB2303
bool "ASB2303"
config MN10300_UNIT_ASB2305
bool "ASB2305"
endchoice
choice
prompt "Processor support"
default MN10300_PROC_MN103E010
help
This option specifies the processor for which the kernel will be
compiled. It affects the on-chip peripherals catered for.
config MN10300_PROC_MN103E010
bool "MN103E010"
depends on MN10300_UNIT_ASB2303 || MN10300_UNIT_ASB2305
select MN10300_PROC_HAS_TTYSM0
select MN10300_PROC_HAS_TTYSM1
select MN10300_PROC_HAS_TTYSM2
endchoice
choice
prompt "Processor core support"
default MN10300_CPU_AM33V2
help
This option specifies the processor core for which the kernel will be
compiled. It affects the instruction set used.
config MN10300_CPU_AM33V2
bool "AM33v2"
endchoice
config FPU
bool "FPU present"
default y
depends on MN10300_PROC_MN103E010
choice
prompt "CPU Caching mode"
default MN10300_CACHE_WBACK
help
This option determines the caching mode for the kernel.
Write-Back caching mode involves the all reads and writes causing
the affected cacheline to be read into the cache first before being
operated upon. Memory is not then updated by a write until the cache
is filled and a cacheline needs to be displaced from the cache to
make room. Only at that point is it written back.
Write-Through caching only fetches cachelines from memory on a
read. Writes always get written directly to memory. If the affected
cacheline is also in cache, it will be updated too.
The final option is to turn of caching entirely.
config MN10300_CACHE_WBACK
bool "Write-Back"
config MN10300_CACHE_WTHRU
bool "Write-Through"
config MN10300_CACHE_DISABLED
bool "Disabled"
endchoice
menu "Memory layout options"
config KERNEL_RAM_BASE_ADDRESS
hex "Base address of kernel RAM"
default "0x90000000"
config INTERRUPT_VECTOR_BASE
hex "Base address of vector table"
default "0x90000000"
help
The base address of the vector table will be programmed into
the TBR register. It must be on 16MiB address boundary.
config KERNEL_TEXT_ADDRESS
hex "Base address of kernel"
default "0x90001000"
config KERNEL_ZIMAGE_BASE_ADDRESS
hex "Base address of compressed vmlinux image"
default "0x90700000"
endmenu
config PREEMPT
bool "Preemptible Kernel"
help
This option reduces the latency of the kernel when reacting to
real-time or interactive events by allowing a low priority process to
be preempted even if it is in kernel mode executing a system call.
This allows applications to run more reliably even when the system is
under load.
Say Y here if you are building a kernel for a desktop, embedded
or real-time system. Say N if you are unsure.
config PREEMPT_BKL
bool "Preempt The Big Kernel Lock"
depends on PREEMPT
default y
help
This option reduces the latency of the kernel by making the
big kernel lock preemptible.
Say Y here if you are building a kernel for a desktop system.
Say N if you are unsure.
config MN10300_CURRENT_IN_E2
bool "Hold current task address in E2 register"
default y
help
This option removes the E2/R2 register from the set available to gcc
for normal use and instead uses it to store the address of the
current process's task_struct whilst in the kernel.
This means the kernel doesn't need to calculate the address each time
"current" is used (take SP, AND with mask and dereference pointer
just to get the address), and instead can just use E2+offset
addressing each time.
This has no effect on userspace.
config MN10300_USING_JTAG
bool "Using JTAG to debug kernel"
default y
help
This options indicates that JTAG will be used to debug the kernel. It
suppresses the use of certain hardware debugging features, such as
single-stepping, which are taken over completely by the JTAG unit.
config MN10300_RTC
bool "Using MN10300 RTC"
depends on MN10300_PROC_MN103E010
default n
help
This option enables support for the RTC, thus enabling time to be
tracked, even when system is powered down. This is available on-chip
on the MN103E010.
config MN10300_WD_TIMER
bool "Using MN10300 watchdog timer"
default y
help
This options indicates that the watchdog timer will be used.
config PCI
bool "Use PCI"
depends on MN10300_UNIT_ASB2305
default y
help
Some systems (such as the ASB2305) have PCI onboard. If you have one
of these boards and you wish to use the PCI facilities, say Y here.
The PCI-HOWTO, available from
<http://www.tldp.org/docs.html#howto>, contains valuable
information about which PCI hardware does work under Linux and which
doesn't.
source "drivers/pci/Kconfig"
source "drivers/pcmcia/Kconfig"
menu "MN10300 internal serial options"
config MN10300_PROC_HAS_TTYSM0
bool
default n
config MN10300_PROC_HAS_TTYSM1
bool
default n
config MN10300_PROC_HAS_TTYSM2
bool
default n
config MN10300_TTYSM
bool "Support for ttySM serial ports"
depends on MN10300
default y
select SERIAL_CORE
help
This option enables support for the on-chip serial ports that the
MN10300 has available.
config MN10300_TTYSM_CONSOLE
bool "Support for console on ttySM serial ports"
depends on MN10300_TTYSM
select SERIAL_CORE_CONSOLE
help
This option enables support for a console on the on-chip serial ports
that the MN10300 has available.
#
# /dev/ttySM0
#
config MN10300_TTYSM0
bool "Enable SIF0 (/dev/ttySM0)"
depends on MN10300_TTYSM && MN10300_PROC_HAS_TTYSM0
help
Enable access to SIF0 through /dev/ttySM0 or gdb-stub
choice
prompt "Select the timer to supply the clock for SIF0"
default MN10300_TTYSM0_TIMER8
depends on MN10300_TTYSM0
config MN10300_TTYSM0_TIMER8
bool "Use timer 8 (16-bit)"
config MN10300_TTYSM0_TIMER2
bool "Use timer 2 (8-bit)"
endchoice
#
# /dev/ttySM1
#
config MN10300_TTYSM1
bool "Enable SIF1 (/dev/ttySM1)"
depends on MN10300_TTYSM && MN10300_PROC_HAS_TTYSM1
help
Enable access to SIF1 through /dev/ttySM1 or gdb-stub
choice
prompt "Select the timer to supply the clock for SIF1"
default MN10300_TTYSM0_TIMER9
depends on MN10300_TTYSM1
config MN10300_TTYSM1_TIMER9
bool "Use timer 9 (16-bit)"
config MN10300_TTYSM1_TIMER3
bool "Use timer 3 (8-bit)"
endchoice
#
# /dev/ttySM2
#
config MN10300_TTYSM2
bool "Enable SIF2 (/dev/ttySM2)"
depends on MN10300_TTYSM && MN10300_PROC_HAS_TTYSM2
help
Enable access to SIF2 through /dev/ttySM2 or gdb-stub
choice
prompt "Select the timer to supply the clock for SIF2"
default MN10300_TTYSM0_TIMER10
depends on MN10300_TTYSM2
config MN10300_TTYSM2_TIMER10
bool "Use timer 10 (16-bit)"
endchoice
config MN10300_TTYSM2_CTS
bool "Enable the use of the CTS line /dev/ttySM2"
depends on MN10300_TTYSM2
endmenu
source "mm/Kconfig"
menu "Power management options"
source kernel/power/Kconfig
endmenu
endmenu
menu "Executable formats"
source "fs/Kconfig.binfmt"
endmenu
source "net/Kconfig"
source "drivers/Kconfig"
source "fs/Kconfig"
source "arch/mn10300/Kconfig.debug"
source "security/Kconfig"
source "crypto/Kconfig"
source "lib/Kconfig"
source "arch/mn10300/oprofile/Kconfig"
menu "Kernel hacking"
source "lib/Kconfig.debug"
config DEBUG_STACKOVERFLOW
bool "Check for stack overflows"
depends on DEBUG_KERNEL
config DEBUG_DECOMPRESS_KERNEL
bool "Using serial port during decompressing kernel"
depends on DEBUG_KERNEL
default n
help
If you say Y here you will confirm the start and the end of
decompressing Linux seeing "Uncompressing Linux... " and
"Ok, booting the kernel.\n" on console.
config KPROBES
bool "Kprobes"
depends on DEBUG_KERNEL
help
Kprobes allows you to trap at almost any kernel address and
execute a callback function. register_kprobe() establishes
a probepoint and specifies the callback. Kprobes is useful
for kernel debugging, non-intrusive instrumentation and testing.
If in doubt, say "N".
config GDBSTUB
bool "Remote GDB kernel debugging"
depends on DEBUG_KERNEL
select DEBUG_INFO
select FRAME_POINTER
help
If you say Y here, it will be possible to remotely debug the kernel
using gdb. This enlarges your kernel ELF image disk size by several
megabytes and requires a machine with more than 16 MB, better 32 MB
RAM to avoid excessive linking time. This is only useful for kernel
hackers. If unsure, say N.
config GDBSTUB_IMMEDIATE
bool "Break into GDB stub immediately"
depends on GDBSTUB
help
If you say Y here, GDB stub will break into the program as soon as
possible, leaving the program counter at the beginning of
start_kernel() in init/main.c.
config GDB_CONSOLE
bool "Console output to GDB"
depends on GDBSTUB
help
If you are using GDB for remote debugging over a serial port and
would like kernel messages to be formatted into GDB $O packets so
that GDB prints them as program output, say 'Y'.
config GDBSTUB_DEBUGGING
bool "Debug GDB stub by messages to serial port"
depends on GDBSTUB
help
This causes debugging messages to be displayed at various points
during execution of the GDB stub routines. Such messages will be
displayed on ttyS0 if that isn't the GDB stub's port, or ttySM0
otherwise.