kolibrios-fun/programs/develop/fasm/trunk/WHATSNEW.TXT

266 lines
10 KiB
Plaintext
Raw Normal View History

Visit http://flatassembler.net/ for more information.
version 1.70.01 (Apr 30, 2012)
[-] Corrected a recently introduced bug that caused some incorrect
address expressions to cause an error prematurely during the
parsing stage.
version 1.70 (Apr 17, 2012)
[+] Added support for AVX, AVX2, AES, CLMUL, FMA, RDRAND, FSGSBASE, F16C,
FMA4, XOP, MOVBE, BMI, TBM, INVPCID, HLE and RTM instruction sets.
[+] Added half-precision floating point values support.
[+] Extended the syntax of "rept" directive to allow numerical expressions
to be calculated by preprocessor in its arguments.
[+] Added "large" and "NX" settings from PE format.
[+] Allowed PE fixups to be resolved anywhere in the generated executable.
[+] Allowed to specify branding value (use 3 for Linux) after the
"format ELF executable" setting.
[+] Added "intepreter", "dynamic" and "note" keywords for creation of
special segments in ELF executables.
[-] Fixed long mode opcode generator to allow absolute addresses to be
generated with "qword" keyword inside square brackets.
[-] Disallowed negative immediates with "int", "enter", "ret" instructions.
[+] Allowed symbolic information dump file to be created even in case of error.
In such case it contains only the preprocessed source that can be extracted
with PREPSRC tool. If error occured during preprocessing, only the source up
to the point of error is provided.
[+] Added symbol references table to symbolic dump file.
[-] Corrected the "defined" and "used" flags in the symbols dump to reflect the
state from the final assembly pass.
[+] Added "assert" directive.
[-] Formatter symbols like "PE" or "readable" are now recognized only in the
context of formatter directives, and thus are no longer disallowed as
labels.
[+] Macroinstruction argument now can have default value, defined with "="
symbol followed by value after the argument name in definition.
[+] Added "relativeto" operator, which can be used in logical expressions
to test whether two values differ only by a constant and not relocatable
amount.
[-] Revised the expression calculator, it now is able to correctly perform
calculations in signed and unsigned ranges in full 64-bit. This fixes
a number of issues - the overflow will now be correctly detected for
64-bit values in cases, where previous versions could not distinguish
whether it was an overflow or not. The effect of these corrections is
that "dq" directive will now behave consistently with behavior of the
data directives for smaller sizes, and the same applies to all the
places where "qword" size for value is used.
version 1.68 (Jun 13, 2009)
[+] Added SSSE3 (Supplemental SSE3), SSE4.1, SSE4.2 and SSE4a instructions.
[+] Added the AMD SVM and Intel SMX instructions.
[+] Added "rdmsrq", "wrmsrq", "sysexitq" and "sysretq" mnemonics for the
64-bit variants of respective instructions.
[+] Added "fstenvw", "fstenvd", "fsavew", "fsaved", "frstorw" and "frstord"
mnemonics to allow choosing between 16-bit and 32-bit variants of
structures used by the "fstenv", "fsave" and "frstor" instructions.
[+] Added "plt" operator for the ELF output format.
[+] Allowed "rva" operator to be used in MS COFF object format, and also
added "static" keyword for the "public" directive.
[+] Added Intel-style aliases for the additional long mode 8-bit registers.
[-] The PE formatter now automatically detects whether relocatable labels
should be used, depending on whether the fixups directory is placed
somewhere into executable by programer, or not. This makes possible the
more flexible use of the addressing symbols in case of PE executable fixed
at some position.
[-] Added support for outputting the 32-bit address relocations in case of
64-bit object formats and PE executable. This makes some specific
instructions compilable, but it also forces linker to put such
generated code into the low 2 gigabytes of addressing space.
[+] Added "EFI", "EFIboot" and "EFIruntime" subsystem keywords for PE format.
[-] Corrected the precedence of operators of macroinstruction line maker.
The symbol escaping now has always the higher priority than symbol conversion,
and both have higher precedence than concatenation.
[+] Allowed to check "@b" and "@f" symbols with "defined" operator.
[+] Allowed "as" operator to specify the output file extension when
placed at the end of the "format" directive line.
[-] Definition of macro with the same name as one of the preprocessor's directives
is no longer allowed.
[+] Allowed single quote character to be put inside the number value,
to help improve long numbers readability.
[+] Added optional symbolic information output, and a set of tools that extract
various kinds of information from it.
[+] Added "err" directive that allows to signalize error from the source.
version 1.66 (May 7, 2006)
[+] Added "define" directive to preprocessor, which defines symbolic constants,
the same kind as "equ" directive, however there's an important difference
that "define" doesn't process symbolic constants in the value before
assigning it. For example:
a equ 1
a equ a+a
define b 1
define b b+b
defines the "a" constant with value "1+1", but the "b" is defined with
value "b+b". This directive may be useful in some advanced
macroinstructions.
[-] Moved part of the conditional expression processing into parser,
for slightly better performance and lesser memory usage by assembler.
The logical values defined with "eq", "eqtype" and "in" operators are now
evaluated by the parser and if they are enough to determine the condition,
the whole block is processed accordingly. Thus this block:
if eax eq EAX | 0/0
nop
end if
is parsed into just "nop" instruction, since parser is able to determine
that the condition is true, even though one of the logical values makes no
sense - but since this is none of the "eq", "eqtype" and "in" expressions,
the parser doesn't investigate.
[-] Also the assembler is now calculating only as many logical values as it
needs to determine the condition. So this block:
if defined alpha & alpha
end if
will not cause error when "alpha" is not defined, as it would with previous
versions. This is because after checking that "defined alpha" is false
condition it doesn't need to know the second logical value to determine the
value of conjunction.
[+] Added "short" keyword for specifying jump type, the "jmp byte" form is now
obsolete and no longer correct - use "jmp short" instead.
[-] The size operator applied to jump no longer applies to the size of relative
displacement - now it applies to the size of target address.
[-] The "ret" instruction with 0 parameter is now assembled into short form,
unless you force using the 16-bit immediate with "word" operator.
[+] Added missing extended registers for the 32-bit addressing in long mode.
[+] Added "linkremove" and "linkinfo" section flags for MS COFF output.
[+] Added support for GOT offsets in ELF object formatter, which can be useful
when making position-independent code for shared libraries. For any label
you can get its offset relative to GOT by preceding it with "rva" operator
(the same keyword as for PE format is used, to avoid adding a new one,
while this one has very similar meaning).
[-] Changed ELF executable to use "segment" directive in place of "section",
to make the distinction between the run-time segments and linkable
sections. If you had a "section" directive in your ELF executables and they
no longer assemble, replace it with "segment".
[-] The PE formatter now always creates the fixups directory when told to -
even when there are no fixups to be put there (in such case it creates the
directory with one empty block).
[-] Some of the internal structures have been extended to provide the
possibility of making extensive symbol dumps.
[-] Corrected "fix" directive to keep the value intact before assigning it to the
prioritized constant.
[+] The ` operator now works with any kind of symbol; when used with quoted
string it simply does nothing. Thus the sequence of ` operators applied to
one symbol work the same as if there was just one. In similar manner, the
sequence of # operators now works as if it was a single one - using such a
sequence instead of escaping, which was kept for some backward
compatibility, is now deprecated.
[-] Corrected order of identifying assembler directives ("if db eq db" was
incorrectly interpreted as data definition).
[-] Many other small bugs fixed.
version 1.64 (Aug 8, 2005)
[+] Output of PE executables for Win64 architecture (with "format PE64"
setting).
[+] Added "while" and "break" directives.
[+] Added "irp" and "irps" directives.
[+] The macro arguments can be marked as required with the "*" character.
[-] Fixed checking for overflow when multiplying 64-bit values - the result
must always fit in the range of signed 64 integer now.
[-] Segment prefixes were generated incorrectly in 16-bit mode when BP was used
as a second addressing register - fixed.
[-] The "local" directive was not creating unique labels in some cases - fixed.
[-] The "not encodable with long immediate" error in 64-bit mode was sometimes
wrongly signaled - fixed.
[-] Other minor fixes and corrections.
version 1.62 (Jun 14, 2005)
[+] Escaping of symbols inside macroinstructions with backslash.
[+] Ability of outputting the COFF object files for Win64 architecture
(with "format MS64 COFF" setting).
[+] New preprocessor directives: "restruc", "rept" and "match"
[+] VMX instructions support (not documented).
[+] Extended data directives to allow use of the "dup" operator.
[+] Extended "struc" features to allow custom definitions of main structure's
label.
[-] When building resources from the the .RES file that contained more
than one resource of the same string name, the separate resource
directories were created with the same names - fixed.
[-] Several bugs in the ELF64 object output has been fixed.
[-] Corrected behavior of "fix" directive to more straightforward.
[-] Fixed bug in "include" directive, which caused files included from within
macros to be processed the wrong way.