Those instructions are not commutative since they only change the first
element in the vector and leave the rest unmodified.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Using rNm and x86inc's stack allocation with a negative value at the same
time isn't supported, and caused the original stack pointer to be clobbered
when using a compiler that doesn't support stack alignment.
The REP_RET workaround is only needed on old AMD cpus, and the labels clutter
up the symbol table and confuse debugging/profiling tools, so use EQU to
create SHN_ABS symbols instead of creating local labels. Furthermore, skip
the workaround completely in functions that definitely won't run on such cpus.
Note that EQU is just creating a local label when using nasm instead of yasm.
This is probably a bug, but at least it doesn't break anything.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
When allocating stack space with a larger alignment than the known stack
alignment a temporary register is used for storing the stack pointer.
Ensure that this isn't one of the registers used for passing arguments.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
* Correctly handle FMA instructions with memory operands.
* Print a warning if FMA instructions are used without the correct cpuflag.
* Simplify the instantiation code.
* Clarify documentation.
Only the last operand in FMA3 instructions can be a memory operand. When
converting FMA4 instructions to FMA3 instructions we can utilize the fact
that multiply is a commutative operation and reorder operands if necessary
to ensure that a memory operand is used only as the last operand.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
The REP_RET workaround is only needed on old AMD cpus, and the labels clutter
up the symbol table and confuse debugging/profiling tools, so use EQU to
create SHN_ABS symbols instead of creating local labels. Furthermore, skip
the workaround completely in functions that definitely won't run on such cpus.
Note that EQU is just creating a local label when using nasm instead of yasm.
This is probably a bug, but at least it doesn't break anything.
When allocating stack space with a larger alignment than the known stack
alignment a temporary register is used for storing the stack pointer.
Ensure that this isn't one of the registers used for passing arguments.
* Correctly handle FMA instructions with memory operands.
* Print a warning if FMA instructions are used without the correct cpuflag.
* Simplify the instantiation code.
* Clarify documentation.
Only the last operand in FMA3 instructions can be a memory operand. When
converting FMA4 instructions to FMA3 instructions we can utilize the fact
that multiply is a commutative operation and reorder operands if necessary
to ensure that a memory operand is used only as the last operand.
They're short enough that inlining them actually reduces code size due to
all the overhead associated with making a function call.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
The System V ABI on x86-64 specifies that the al register contains an upper
bound of the number of arguments passed in vector registers when calling
variadic functions, so we aren't allowed to clobber it.
checkasm_fail_func() is a variadic function so also zero al before calling it.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Tested functions are internally kept in a binary search tree for efficient
lookups. The downside of the current implementation is that the tree quickly
becomes unbalanced which causes an unneccessary amount of comparisons between
nodes. Improve this by changing the tree into a self-balancing left-leaning
red-black tree with a worst case lookup/insertion time complexity of O(log n).
Significantly reduces the recursion depth and makes the tests run around 10%
faster overall. The relative performance improvement compared to the existing
non-balanced tree will also most likely increase as more tests are added.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
The System V ABI on x86-64 specifies that the al register contains an upper
bound of the number of arguments passed in vector registers when calling
variadic functions, so we aren't allowed to clobber it.
checkasm_fail_func() is a variadic function so also zero al before calling it.
Tested functions are internally kept in a binary search tree for efficient
lookups. The downside of the current implementation is that the tree quickly
becomes unbalanced which causes an unneccessary amount of comparisons between
nodes. Improve this by changing the tree into a self-balancing left-leaning
red-black tree with a worst case lookup/insertion time complexity of O(log n).
Significantly reduces the recursion depth and makes the tests run around 10%
faster overall. The relative performance improvement compared to the existing
non-balanced tree will also most likely increase as more tests are added.
Now we no longer have to rely on function pointers intentionally
declared without specified argument types.
This makes it easier to support functions with floating point parameters
or return values as well as functions returning 64-bit values on 32-bit
architectures. It also avoids having to explicitly cast strides to
ptrdiff_t for example.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
If the return value doesn't fit in a single register rdx/edx can in some
cases be used in addition to rax/eax.
Doesn't affect any of the existing checkasm tests but might be useful later.
Also comment the relevant code a bit better.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
If the return value doesn't fit in a single register rdx/edx can in some
cases be used in addition to rax/eax.
Doesn't affect any of the existing checkasm tests but might be useful later.
Also comment the relevant code a bit better.
Now we no longer have to rely on function pointers intentionally
declared without specified argument types.
This makes it easier to support functions with floating point parameters
or return values as well as functions returning 64-bit values on 32-bit
architectures. It also avoids having to explicitly cast strides to
ptrdiff_t for example.
The .text section is already 16-byte aligned by default on all supported
platforms so `SECTION_TEXT` isn't any different from `SECTION .text`.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Change ALLOC_STACK to always align the stack before allocating stack space for
consistency. Previously alignment would occur either before or after allocating
stack space depending on whether manual alignment was required or not.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Change ALLOC_STACK to always align the stack before allocating stack space for
consistency. Previously alignment would occur either before or after allocating
stack space depending on whether manual alignment was required or not.
The bug was fixed in 1.3.0, so only perform the workaround in earlier versions.
Reviewed-by: "Ronald S. Bultje" <rsbultje@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
There is an SSE2 implementation so the SSE version is never used. The "SSE"
version also happens to contain SSE2 instructions on x86-64.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
The upper halves are not guaranteed to be zero in x86-64.
Also use `test` instead of `and` when the result isn't used for anything other
than as a branch condition, this allows some register moves to be eliminated.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
It provides the following features:
* verify correctness by comparing output to the C version.
* detect failure to save and restore clobbered callee-saved registers.
* detect 32-bit parameters being used as if they were 64-bit in x86-64
(the upper halves are not guaranteed to be zero - but in practice
they very often are, which makes those bugs hard to spot otherwise).
* easy benchmarking.
Compile by running 'make checkasm'.
Execute by running 'tests/checkasm/checkasm'.
Optional arguments are '--bench' to run benchmarks for all functions,
'--bench=<pattern>' to run benchmarks for all functions that starts with
<pattern>, and '<integer>' to seed the PRNG for reproducible results.
Contains unit tests for most h264pred functions to get started, more tests
can be added afterwards using those as a reference.
Loosely based on code from x264. Currently only supports x86 and x86-64,
but additional architectures shouldn't be too much of an obstacle to add.
Note that functions with floating point parameters or floating point
return values are not supported. Some compiler-specific features or
preprocessor hacks would likely be required to add support for that.
Signed-off-by: Janne Grunau <janne-libav@jannau.net>
Improves the accuracy of measurements, especially in short sections.
To quote the Intel 64 and IA-32 Architectures Software Developer's Manual:
"The RDTSC instruction is not a serializing instruction. It does not necessarily
wait until all previous instructions have been executed before reading the counter.
Similarly, subsequent instructions may begin execution before the read operation
is performed. If software requires RDTSC to be executed only after all previous
instructions have completed locally, it can either use RDTSCP (if the processor
supports that instruction) or execute the sequence LFENCE;RDTSC."
SSE2 is a requirement for lfence so only use it on SSE2-capable systems.
Prefer lfence;rdtsc over rdtscp since rdtscp is supported on fewer systems.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Prevents a crash if the misaligned exception mask bit is
cleared for some reason.
Misaligned SSE functions are only used on AMD Phenom CPUs
and the benefit is miniscule. They also require modifying
the MXCSR control register and by removing those functions
we can get rid of that complexity altogether.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Store XMM6 and XMM7 in the shadow space in functions that
clobbers them. This way we don't have to adjust the stack
pointer as often, reducing the number of instructions as
well as code size.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
