asserts on seL4_Errors now use test_error_eq

changed test_asserts that check seL4_Errors
to use test_error_eq to output errors in failed cases
that would otherwise be suppressed in a
KernelInvocationReportErrorIPC build config

SELFOUR-161: Remove Remap and use Map instead

Update usages of Page_Remap to instead use Page_Map.

sel4test-tests: Disable device mem tests for spike

Spike does not have any device untypes exported to user level, so the
tests that require a device untyped are disabled.

sel4test-tests: Fix frame tests

Previously, these tests relied on using initialised timer device frames
that were supposedly copied across from the driver. This wasn't the
case, so this commit fixes this behaviour by using the device frame that
is copied across.

sel4test: explicitly link sel4_autoconf

since the global conf lib is going be removed, we need to explicitly
linke each project with sel4_autoconf and other config libs
this commit also removed a bunch of unused includes as well as
HAVE_AUTOCONF ifdef checking.

Trivial: style fix

make the style checker happy

MCS: use unique reply objects for each thread

Using the same reply object for recv does not work: multiple threads
cannot use the same reply object to block at the same time. This was
working by luck / concurrency before, but now the kernel actually
checks this condition.

SELFOUR-1277: undo s/Remap/Map in FRAMEEXPORT test.

This is no longer needed for riscv now that Remap is implemented.

RISCV port

Use seL4_TCB_SetTimeoutEndpoint

- remote timeout endpoint from api_tcb_set_space calls
- update helper for setting timeout endpoint

Remove deleted seL4_PrioProps from frames.c

SELFOUR-1016: Update for TCB api change

Changes for removal of seL4_CapData_t type in libsel4

This updates sel4test to not use what are now deprecated functions and types by following
the upgrade notes in the kernel CHANGES file.

Improve performance of frameexport test

Testing every single byte of every single mapping is extremely time consuming on systems
with a large amount of memory, and is also completely unnecessary

SELFOUR-1130 Refactor tests to use new DEFINE_TEST

These changes turn tests on and off with the new `enabled` argument in
DEFINE_TEST instead of using #ifdefs.

SELFOUR-1083 refactor tests because DEFINE_TEST_MAYBE was renamed to DEFINE_TEST

SELFOUR-894 run config-controlled tests with DEFINE_TEST_MAYBE

tests: correct usage of macros

s/test_assert_fatal/test_assert: none of these tests should require the
test suite to abort in their current state as they are running in a
different process.

Fix cases where test_assert would return a value from a void function as
appropriate.

Add test for ensuring deleting frame cap performs an unmap

As part of this change a fault helper function from the XN tests are moved into the
common source, instead of being behind an ifdef, so this new test can use it.

MCS: use api_* wrappers for recv operations

- on the MCS kernel, recv operations that will use call / replyrecv
semantics require a reply object, provide the reply object when needed
if CONFIG_KERNEL_RT is enabled. If not, it will always be seL4_CapNull
and is not used.
- use the api_recv etc wrappers from sel4utils/api.h

MCS: use tcb wrappers from sel4tutils/api.h

- replace calls of seL4_TCB_SetSpace with api_tcb_set_space
- replace calls of seL4_TCB_Configure with api_tcb_configure
- update some calls from seL4_TCB_Configure to api_tcb_set_space as a
full TCB configure is not required in that specific case

Use seL4_Wait instead of seL4_Recv where possible

On the master kernel seL4_Wait just calls seL4_Recv and does not return
a message tag. On the mcs kernel, seL4_Wait is a first class system call which
waits on an object without requiring a reply object, but does return a
message tag. Therefore seL4_Wait can be
used equivalently on both branches as long as the tag is not required.

This change replaces calls to seL4_Recv where the return value is
ignored with seL4_Wait.

Provide and use wrappers for getting helper thread information

Reaching directly into the helper thread struct makes it difficult to change how helper threads
work without breaking lots of code.

Use ltimer interface in sel4test

- remove all plat specific timer code
- new timers don't need any code added to sel4test!
- also remove plat_copy_serial caps as it is empty for all cases

Add env argument to helper_set_priority, mcp

- it will be needed in future when SELFOUR-1016 is merged
- merge it now to avoid the conflicts inherent on a long running branch

trivial: fix warning

Fix licenses

More robust test that frames are mapped and remapped correctly

Ensure we test each differently sized frame at least once

aarch64: update related tests

SELFOUR-114: use new fault API

SELFOUR-192: remove recycle, add cancelSends

SELFOUR-527: Unify instruction cache for self modifying code

sel4test-tests: init timer if CONFIG_HAVE_TIMER=y

This avoids timer initialisation failures on simulators.
We don't run the timer tests on simulators anyway due to
inaccuarcy of interrupt delivery.

A better solution would be to fall back on different timer
implementations (the current failure is that on x86 our hpet
driver relies on fsb delivery which qemu does not support).

User device untyped and s/PIT/HPET for x86

This commit changes the way sel4test-driver passes
device memory to sel4test-tests: instead of copying the frame
caps it copies untypeds, to be compataible with how seL4_libs
has changed after device untyped. sel4test-tests now gives
the relevant untyped to its allocator so seL4_libs can
initialise the relevant devices.

Additionally, this commit changes to use the HPET and IOAPIC
instead of the legacy PIT/PIC on x86, allowing us much greater
timer accuracy in tests and faster tests in general.

SELFOUR-276: add maximum controlled priority (MCP)

SELFOUR-276: add support for MCPs

Where MCP = Maximum controlled priority.

This commit:

* updates the arguments to seL4_TCB_Configure
* updates the set_priority tests to test MCPs
* adds set_helper_mcp to helpers.h

SELFOUR-421 Tests for using a device frame as an IPC buffer

The kernel should not permit using a device frame as an IPC buffer.
This change introduces two tests to check for that

Remove references to experimental branch (stable)

Test XN kernel feature

The XN tests were #ifdef'ed out in such a way that it was never being
run. This commit enables them and unbitrots them.

Support x64 tests

This fixes various tests that were not compatible with x64. The real
changes are to tests that create/map/recycle frames and these have
been rewritten to be clearer and support both 32 and 64bit.

Other tests have minor changes for type compatibility and reserving
additional amounts of memory needed for the changes.

update for s/Wait/Recv

Avoid shadowing variables.

This change has no effect on functionality, but helps poor users such as myself
who switch on -Wshadow, only to be buffeted with spurious warnings.

Remove 'args' parameter to test cases.

This commit removes the state parameter to test cases, which was unused in all
tests. In three years, no test has required this parameter. I think it's time
to ditch this dead weight. It can be reintroduced if necessary later.

Note that this change is interdependent with changes to libsel4test.

bandaid to get sel4test to output the correct number of tests passing

Test that cache operations can be performed on all page sizes

Do not assume large frames on x86 are 4M

Test if mapped frames can be remapped

Fixes for vspace API change

Release snapshot