Avoid the repeated inclusions of config-target.mak, which have
risks of namespace pollution, and instead build minimal configuration
files in a configuration script. The same configuration files can
also be included in Makefile and Makefile.qemu
[AJB 10/09/19]
In the original PR this had inadvertently enabled tests
for ppc64abi32. However as the rest of the multiarch tests work rather
than disabling the otherwise correctly functioning build I've just
skipped the failing linux-test test. For some reason I can't debug it
with TCG so I'm leaving that to the PPC maintainers to look at.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20190807143523.15917-4-pbonzini@redhat.com>
[AJB: s/docker/container/, rm last bits from configure, ppc6432abi hack]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Cc: Richard Henderson <rth@twiddle.net>
For i386 specifically, this allows using the host GCC
to compile the i386 tests. But, it should really be
done for all targets, unless we want to pass $(EXTRA_CFLAGS)
directly as part of $(CC).
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20190807143523.15917-2-pbonzini@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
The Exclusive Instructions provide a general-purpose mechanism for
atomic updates of memory-based synchronization variables that can be
used for exclusion algorithms.
Use cmpxchg-based implementation that is sufficient for the typical use
of exclusive access in atomic operations.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Drop test_fail: we know that exit simcall works. Now that it's not run
automatically there's no point in keeping it.
Drop test_pipeline: we're not modeling pipeline, we don't control ccount
and there's no plan to do so.
Enable test_boolean: it won't break on cores without boolean option, it
will do testing on cores with boolean option.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
break_dependency incorrectly handles the case of dependency on an opcode
that references the same register multiple times. E.g. the following
instruction is translated incorrectly:
{ or a2, a3, a3 ; or a3, a2, a2 }
This happens because resource indices of both dependency graph nodes are
incremented, and a copy for the second instance of the same register in
the ending node is not done.
Only increment resource index of the ending node of the dependency.
Add test.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Make windowed register tests conditional on the presence of this option.
Fix tests to work correctly for both 32 and 64 physical registers.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Make s32c1i tests conditional on the presence of this option. Initialize
ATOMCTL SR when it's present to allow RCW transactions on uncached
memory.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
SR tests generate instructions that the assembler does not recognize and
thus must take care about configuration endianness.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Make tests for specific special registers conditional on the presence of
the options that add these registers and test that the registers are not
accessible otherwise.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Make timer/CCOUNT tests conditional on the presence of timer option and
number of configured timers. Don't use hard coded interrupt levels for
timers, use configured values.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Make interrupt tests conditional on the presence of interrupt option and
on the presence of level-1 and high level software interrupts. Don't use
hard-coded interrupt level for the high level interrupt tests, choose
high level software IRQ and use its configured level.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Make tests for optional instruction groups conditional on the presence
of corresponding options in the config.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Make data/instruction tests conditional on the presence of
data/instruction cache, whether they're lockable and whether data cache
is writeback.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Make debug tests conditional on the presence of the debug option in the
config and tests that depend on the presence/number of instruction or
data breakpoint registers on the corresponding definitions. Use
configured debug interrupt level instead of the hardcoded value to set
up IRQ handler and access debug EPC register.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Uncomment test_boolean in the test makefile. Make actual tests code
conditional on the presence of boolean option in the config.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Use bbci.l/bbsi.l instead of bbci/bbsi, as they are assembly macros that
accept little-endian bit number and produce correct immediate for both
little and big endian configurations. Choose value loaded into register
for bbc/bbs opcodes based on configuration endianness.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Don't use 'loop' opcode in generic testsuite completion code, only use
core opcodes to make it work with any configuration.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Configurations with LITBASE register may use absolute literals by
default. Pass --no-absolute-literals option to assembler to use
PC-relative literals instead.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Configurations w/o vecbase may have vectors not grouped together and not
in fixed order. They may not always be grouped into single output
sections by assigning next offset to dot, as it may sometimes move dot
backwards and sometimes they may even belong to different memory region.
Don't group vectors into single output section. Instead put each vector
into its own section ant put it at its default virtual address.
Reserve 4KBytes from the default vectors base and put rest of the code
and data starting from there. Mark vectors sections as executable,
otherwise their contents is discarded. There may be as little as 16
bytes reserved for some vectors, load handler address into a0 and use
ret.n to jump there to make vector code fit into this 16 byte space.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
When test suite with multiple tests fails it's not obvious which test
failed. Pring "failed" in every invocation of test_fail. Do printing
when DEBUG preprocessor macro is defined.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Don't generate handlers for IRQ levels that are not defined for the CPU
or for window overflow/underflow exceptions for configs w/o windowed
registers.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Not all CPU configurations may have enough space for handler code
between exception/interrupt vectors. Leave jumps to the handlers at the
vectors, but move all handlers past the vectors area.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Failed memory transactions should raise exceptions 14 (for fetch) or 15
(for load/store) with XEA2.
Memory accesses that result in TLB miss followed by an attempt to load
PTE from physical memory which fails should raise InstTLBMiss or
LoadStoreTLBMiss with XEA2.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
The value of CCOUNT special register is calculated as time elapsed
since CCOUNT == 0 multiplied by the core frequency. In icount mode time
increment between consecutive instructions that don't involve time
warps is constant, but unless the result of multiplication of this
constant by the core frequency is a whole number the CCOUNT increment
between these instructions may not be constant. E.g. with icount=7 each
instruction takes 128ns, with core clock of 10MHz CCOUNT values for
consecutive instructions are:
502: (128 * 502 * 10000000) / 1000000000 = 642.56
503: (128 * 503 * 10000000) / 1000000000 = 643.84
504: (128 * 504 * 10000000) / 1000000000 = 645.12
I.e.the CCOUNT increments depend on the absolute time. This results in
varying CCOUNT differences for consecutive instructions in tests that
involve time warps and don't set CCOUNT explicitly.
Change frequency of the core used in tests so that clock cycle takes
exactly 64ns. Change icount power used in tests to 6, so that each
instruction takes exactly 1 clock cycle. With these changes CCOUNT
increments only depend on the number of executed instructions and that's
what timer tests expect, so they work correctly.
Longer story:
http://lists.nongnu.org/archive/html/qemu-devel/2018-03/msg04326.html
Cc: Pavel Dovgaluk <Pavel.Dovgaluk@ispras.ru>
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
memctl SR is not available on dc232b, as it was introduced in more
recent hardware release. Now that this information is available through
the libisa the test fails. Fix the test.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Don't use hardcoded software interrupt masks, use XCHAL macros.
Mask off timer interrupt bits that are not checked for.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Don't expect that CCOUNT increments are equal to the number of executed
instructions. Verify that timer interrupt does not fire before the
programmed CCOMPARE value and does fire after.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Timer tests expect certain determinism in CCOUNT updates and timer
interrupts firing. Run QEMU with -icount to get deterministic results.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
We've currently got 18 architectures in QEMU, and thus 18 target-xxx
folders in the root folder of the QEMU source tree. More architectures
(e.g. RISC-V, AVR) are likely to be included soon, too, so the main
folder of the QEMU sources slowly gets quite overcrowded with the
target-xxx folders.
To disburden the main folder a little bit, let's move the target-xxx
folders into a dedicated target/ folder, so that target-xxx/ simply
becomes target/xxx/ instead.
Acked-by: Laurent Vivier <laurent@vivier.eu> [m68k part]
Acked-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de> [tricore part]
Acked-by: Michael Walle <michael@walle.cc> [lm32 part]
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com> [s390x part]
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com> [s390x part]
Acked-by: Eduardo Habkost <ehabkost@redhat.com> [i386 part]
Acked-by: Artyom Tarasenko <atar4qemu@gmail.com> [sparc part]
Acked-by: Richard Henderson <rth@twiddle.net> [alpha part]
Acked-by: Max Filippov <jcmvbkbc@gmail.com> [xtensa part]
Reviewed-by: David Gibson <david@gibson.dropbear.id.au> [ppc part]
Acked-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> [crisµblaze part]
Acked-by: Guan Xuetao <gxt@mprc.pku.edu.cn> [unicore32 part]
Signed-off-by: Thomas Huth <thuth@redhat.com>
Tracked down with an ugly, brittle and probably buggy Perl script.
Also move includes converted to <...> up so they get included before
ours where that's obviously okay.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Tested-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Check that entry instruction raises window overflow exception when
PS.CALLINC points to live registers.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Xtensa cores have configurable interrupt vectors and endiannes. This
information is needed to link executable images correctly for a specific
core configuration. Instead of hard-coding dc232 defaults pull endianness,
number of high-priority interrupts and location of vectors from the core
configuration and pass it through the C preprocessor.
While at it clean up tabs and align the initial stack on 16 bytes.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Test that non-locking prefetch operations don't cause exceptions on
missing TLB and that other 'hit' cache operations do.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Add path to the core configuration directory to test build command and
replace .include asm directive with #include to enable preprocessing.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Some test suites, like MMU, need per-test initialization. Don't make them
redefine test macro, add test_init for that purpose.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
