The crc32 test now does produce the correct output one would expect
for running with optimized assembly.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
The CRC32 byte patterns are documented to be able to take a
(pointless) REX.W to "extend" the pattern to a 64-bit
destination... but that never changes the result.
However, be excrutiatingly correct and add those patterns when not
using optimization.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
The crc32 instructions are kind of odd; the operand size really only
applies to the right (source) operand.
The APX forms are also kind of messy.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
Update the apx.bin.t file, since it now produces the same output as it
does with -DSRC and -DBIN.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
Memory references should have RN_L16 set if and only if they are
compatible with instruction patterns requiring register numbers below
16.
Add a "vex+" encoding pattern for VEX-encoded instructions that should
be promoted to EVEX when AVX-encoded.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
If there are no labels in the default segment, then NASM would crash
if anything was emitted in it, as the label would be defined on the
final pass only.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
Along with C and other languages, the current trend is to be able to
probe for features rather than relying on version numbers. This is
motivated in part by the intent of bumping the major version number to
3.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
Add the %null(), %note(), %warning(), %error(), and %fatal()
functions. They behave like the corresponding directives, then expands
to nothing.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
Add INVPCI, INVEPT, INVVPID APX instruction variants to the database and
relevant tests.
Signed-off-by: Maciej Wieczor-Retman <maciej.wieczor-retman@intel.com>
Add the test cases for the IMUL instruction.
Fix the 0x6B database bug that used the same bit width of the immediate
operand as the other operands. Make it 8 bit wide, just like in the
legacy instruction.
Signed-off-by: Maciej Wieczor-Retman <maciej.wieczor-retman@intel.com>
Looks like the wwflag variable was set for both ww and w1 which caused
some VEX.W bits to not get set. Don't set the wwflag for w0 or w1 cases
in the script.
Signed-off-by: Maciej Wieczor-Retman <maciej.wieczor-retman@intel.com>
Clean up CRC32 with $dq macro and add the APX variants.
Also add tests for legacy and APX variants.
Signed-off-by: Maciej Wieczor-Retman <maciej.wieczor-retman@intel.com>
The ADD, ADC, ADCX, ADOX, AND instruction testcases.
Part of ADCX tests is commented out because addressing with R16-R31 is
currently bugged without some other mechanism indicating using the APX.
Signed-off-by: Maciej Wieczor-Retman <maciej.wieczor-retman@intel.com>
Add the AOR instruction and use $bwdq macros on the other new atomic
instructions: AADD, AAND and AXOR.
Add tests.
Signed-off-by: Maciej Wieczor-Retman <maciej.wieczor-retman@intel.com>
This json allows assembling the first column in the testcase macro when
updating the apx.bin.t file, and assembling the second column when
running the test.
Signed-off-by: Maciej Wieczor-Retman <maciej.wieczor-retman@intel.com>
Due to ib in the apx arithmetic instruction preprocessed implementation
only byte size immediate operands were possible. Changing ib to i# fixes
the issue.
Signed-off-by: Maciej Wieczor-Retman <maciej.wieczor-retman@intel.com>
Check a few more corner cases, including $ and $$, as well as parsing
in the assembler (dd) and the preprocessor (%assign).
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
$$ is TOKEN_BASE, not a symbol. If this is done incorrectly, ppscan
chokes on $$ as it ends up being an invalid symbol.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
Segment references can come from two places: either an explicit SEG
operator or from a far expression. In the former case, the segment can
have a programmer-provided offset, but in the latter case, it
cannot.
The fix for bug 3392949 fixed the latter case, but broke the
former. This patch hopefully makes both work.
Rename out_segment() to out_farseg() and add a comment to explain the
logic behind the difference.
Reported-by: E. C. Masloch <pushbx@ulukai.org>
Fixes: https://bugzilla.nasm.us/show_bug.cgi?id=3392950
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
It is possible for m->mstk.mmac to point back to itself. In that case,
don't terminate the macro debug invocation just yet.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
- Add improvements to omfdump from Bernd Böckmann.
- Add misc/Makefile with option to install omfdump and auxiliary NASM
data files.
- Use compiler.h and friends in the misc directory.
- Use -std=c23 if the C compiler supports it.
Cc: Bernd Boeckmann <bernd-github@boeckmann.io>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Micro-optimize the set_bit() function. Using size_t (in particular,
using an unsigned type) produces better code.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
The documentation actually states that $ as a hex prefix is only valid
when followed by a digit, which at least makes the syntax conflict
less complicated. Actually match the documentation.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
To avoid repeated warnings, there is no warning issued for when
tokenizing a number starting with $ in the preprocessor, but issue a
warning if such a number is *consumed* (used in arithmetic) in the
preprocessor.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
The use of $ prefixes for hexadecimal numbers conflicts with
the use of $ to escape symbols. Add a directive to disable
$ for hexadecimal numbers so that those escapes work OK.
As a result, allow escaped symbols to start with a digit.
Add a warning that this syntax is deprecated.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
