replacing umlaute and fixing address in headers Haldeneggsteig -> Universitaetsstrasse

Signed-off-by: Reto Achermann <reto.achermann@inf.ethz.ch>

armv7: remove gic dist debug

Signed-off-by: Lukas Humbel <lukas.humbel@inf.ethz.ch>

Turn serial driver into module and fix userspace GIC

Signed-off-by: Lukas Humbel <lukas.humbel@inf.ethz.ch>

armv7: Move distributor driver to userspace

Signed-off-by: Lukas Humbel <lukas.humbel@inf.ethz.ch>

systime: adding a relative timeout function

Signed-off-by: Adam Turowski <adam.turowski@inf.ethz.ch>

kernel,libbarrelfish: Remove Frame identify and Vnode identify invocation handlers and commands

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

Remove invoke_frame_identify, replaced with frame_identify().

This commit fully removes the Frame_Identify invocation and replaces the
calls with frame_identify().
Additionally, because the KCB is special, this commit introduces a new
invocation kcb_identify() which translates the lvaddr in the KCB
capability into the pysical address of the kernel control block.

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

libbarrelfish: endpoints: refactor to make EP management invocations unnecessary.

This commit implements endpoint_identify using cap_direct_identify and
endpoint_set_iftype using cap_mint.

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

kernel,monitor: implement reclaiming of previously dropped ram caps.

Until now, if the kernel couldn't return a freed region of physical
memory to the monitor, e.g. because the monitor was not ready yet, or
the monitor's receive slot is already occupied, it would drop the range
and print a diagnostic message.

With this change, the cpu driver has 4 slots per KCB in which it can
store a range that would be dropped. The monitor gets a new kernel cap
invocation which allows it to retrieve one of the stored slots, if any.
The monitor runs a deferred event loop which checks if there's a range
to retrieve every second, and forwards any retrieved ranges to the
memory server.

The retrieval is very similar to a capops delete step, but does not
happen inside the capops machinery because the need for an action are
not coupled to a capability operation.

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

libbf: use cap_direct_identify for frame_identify and co and replace calls to debug_cap_identify where appropriate

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

ARMv7: syscall endpoint invoke fixed

Signed-off-by: Roni Häcki <roni.haecki@inf.ethz.ch>

kernel: adding invoke_endpoint syscalls to armv7 + armv8

Signed-off-by: Roni Häcki <roni.haecki@inf.ethz.ch>

armv7: make armv7 compile again on VF branch

Signed-off-by: Roni Häcki <roni.haecki@inf.ethz.ch>

caps: adding endpoint cap type for UMP endpoints

Signed-off-by: Reto Achermann <reto.achermann@inf.ethz.ch>

devidcaps: pass all arguemnts to the kernel to fill the cap

Signed-off-by: Reto Achermann <reto.achermann@inf.ethz.ch>

caps: adding invocation handlers for the new captypes to the cpudriver

Signed-off-by: Reto Achermann <reto.achermann@inf.ethz.ch>

Kaluga: Make unified irq src caps work on armv7

Signed-off-by: Lukas Humbel <lukas.humbel@inf.ethz.ch>

ARM: unifying coreboot on ARMv7 and ARMv8

Coreboot:
* use of hwid_t instead of coreid_t to ensure enough size
* adding context id parameter to the start function

ARMv8:
* removing PSCI debug syscall and use ipi_cap instead
* implementaton of platform specific call functions

ARMv7:
* adapting call sites to deal with new argument
* adaption of platform boot function

Signed-off-by: Reto Achermann <reto.achermann@inf.ethz.ch>

systime: introducing system time based on time stamp counter/global timer, removing kernel_now

Signed-off-by: Adam Turowski <adam.turowski@inf.ethz.ch>

armv7: handle VNodeCmd_Identify

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

Make debug_my_cspace work with new CSpace layout

Signed-off-by: Lukas Humbel <lukas.humbel@inf.ethz.ch>

kernel: arm: fix modify_flags invocation

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

ARMv7: Manual coreboot works on Pandaboard

Module that fact the UMP isn't working properly with ARM caches.

Signed-off-by: David Cock <david.cock@inf.ethz.ch>

ARMv7: Clone CPU driver cmdline correctly.

Signed-off-by: David Cock <david.cock@inf.ethz.ch>

armv7: kernel: monitor_delete_last: fix argument unmarshalling

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

ARMv7: Find and pass the GOT base to the APP core.

Locating it in the source ELF still doesn't quite work.

Signed-off-by: David Cock <david.cock@inf.ethz.ch>

ARMv7: Coreboot makes it into the APP CPU driver.

Signed-off-by: David Cock <david.cock@inf.ethz.ch>

T266: armv7: add kernel-side handlers for cnode resizing

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

ARMv7: CPU driver relocations done.

Signed-off-by: David Cock <david.cock@inf.ethz.ch>

T258: Delete GPT cspace functionality and rename new functions

This commit deletes all the old GPT cnode code, and renames the new two-level
functions to the "old" names (e.g. caps_lookup_slot() is now a two-level
cspace function).

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

T276: armv7: Add new monitor invocation for retypeability check

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

T273: armv7: Fix monitor invocations for capability retype and creation

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

ARMv7: Rewrote serial drivers on ARM.

They're now fairly well compartmentalised on both ARM and x86, and can start
automatically from Kaluga. There's also a 'kernel' serial driver, which uses
the syscall interface to read and write on the kernel console. This avoids
having user level and kernel code racing on access to the UART. Fish works
(minus line editing) on all Versatile Express platforms.

Signed-off-by: David Cock <david.cock@inf.ethz.ch>

T273: armv7: monitor invocation handlers: rename bits to level for cspace level arguments

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

T273: fix monitor_retype and adjust arguments in direct retype

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

T272: armv7: implement two-level cspace layout

Closes T272.

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

ARMv7: Moved MMU initialisation into C

Signed-off-by: David Cock <david.cock@inf.ethz.ch>

T255: kernel cap invocation and monitor RPC to discover architecture and platform.

Closes T255.

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

Rationalised the ARM Platform code.

We now have separate targets for the GEM5 VE and FVP VE, as the GEM5 version
actually simulates an A15, and has a few quirks. The FVP VE and Pandaboard
targets share most of their code. Thus we've got three ARMv7-A targets now,
and it should be easy to add more, as soon as we've got the simulators.

Signed-off-by: David Cock <david.cock@inf.ethz.ch>

Documentation, and cleaner dispatcher mode tracking code. Still bug with yield.

Signed-off-by: Timothy Roscoe <troscoe@inf.ethz.ch>

Refactored ARMv7-A code.
Cleaner startup process.
Cleaner cross-platform portability (platform.h).
Removed ARMv5.
Removed ARMv7-m
Removed XScale
Changed kernel device paging interface and fixed bugs.

Signed-off-by: Timothy Roscoe <troscoe@inf.ethz.ch>

Fixed usage of kernel cap in armboot. Using IPICmd_Send_Start to call Spawn_core for now.

Boot driver for ARM can now boot a pandaboard core.

Make sure ARM kernels are working agian.

arm: added missing bits for ID cap creation.

Split CPU_ARM into CPU_ARM5 and CPU_ARM7

Fixed monitor arm code.

armv7: use A9 timer functions only in pandaboard

These functions are only defined for pandaboard:
collect2: ld terminated with signal 11 [Segmentation fault]
./armv7/kernel/arch/armv7/syscall.o: In function `sys_syscall':
syscall.c:(.text+0x9ac): undefined reference to `gt_read_high'
syscall.c:(.text+0x9c0): undefined reference to `gt_read_low'

Guard them with an #if for now.

temporarily disable argc check in invocation macro.

armv7: fixed stuff and added missing invocation handlers.

armv7: compiles.

arm: added missing bits for ID cap creation.

Split CPU_ARM into CPU_ARM5 and CPU_ARM7

Added basic support for the Cortex A9 Global Timer

The Cortex A9 global timer is a 64-bit incrementing counter, which
is memory mapped in the private memory region. Thus, the code for
the timer is currently all in the CPU driver.

To read out the timer in userspace, two system calls are needed,
one for the lower and and one for the upper half of the counter.
The following helper function is provided to correctly read out
the timer value in two 32 bit reads as suggested in the ARM TRM:

errval_t sys_debug_hardware_global_timer_read(uint64_t *ret)

Refactoring core boot-up logic in kernel.

Move architecture independent logic into single arch-independent
syscall for all platforms. Handle arch specific boot-up with
specialized function (usually defined in start_aps.c).

Redesigned hw page table dumping to be an invocation on a dispatcher cap.

This has two benefits over the old design where dumping hw page tables was a
kernel cap invocation and took a dispatcher cap as an argument:
1) This is probably the most common use case, you don't have to go through
the monitor if you just want to dump your own hw page tables.
2) A domain which has the dispatcher cap of another domain (mostly because
it's going to spawn said domain) can dump the hw page tables of the
to-be-spawned domain without capability address contortions.

timer interrupts enabled

fixed interrupt handler writing to trap area for non-trap interrupts
enabled timer interrupts at reasonable intervals (reasonable for -O2 and no caches)
added system call for restoring a context (just for armv7-m, replacing a x86-specific interrupt),
because the ONLY way to restore an IT block is by exiting handler mode (there is literally no other mechanism)
(will only be called if userspace tries to restore a context that was saved by the kernel, and comes from interrupting an IT block or a ldm/stm instruction)

You may want to use -O2 from now on (because we use interrupts, and thrashing is possible)
The compiler we use will probably complain up to 3 times ("unable to find a register to spill in class 'GENERAL_REGS'")
just retry the faulting compile instruction with -O1, and then restart make

basic armv7-m support

added new hake architecture "armv7-m"
added armv7-m port of kernel and libbarrelfish
fixed small arm_molly issues with parsing paths containing '-'

the port does not actually handle interrupts yet, but scheduling, spawning, and starting processes works
this update does not yet contain the armv7-a code to start the m3 core

Enable user-level interrupt forwarding on ARM.

IRQ handling with hardcoded IRQ id (ehci interrupt) in monitor.

Interrupt handling on usb_manager level i.e. the host controller interrupt handlers
are called.

--HG--
rename : include/usb/class/usb_hub_request.h => include/usb/class/usb_hub.h
rename : usr/drivers/usb/usb_manager/usb_hub.c => usr/drivers/usb/usb_manager/hub/usb_hub.c
rename : usr/drivers/usb/usb_manager/usb_hub.h => usr/drivers/usb/usb_manager/hub/usb_hub.h

armv7: added new modify_flags invocation.

armv7: fixed merge errors and implemented new memory interface for omap44xx.

armv5,armv7: Fix unmarshalling invoke_vbits in handle_invoke().

--HG--
extra : rebase_source : ee2de7b9445cf9deb273d914fb9c6aa4abc4b772

Renamed ARM to ARMv5.

Renamed arm_gem5 to armv7.

TODO:
* Figure out if the __GEM5__ define actually refers to gem5 or just to ARMv7.