H. Peter Anvin [Wed, 4 Jun 2008 22:13:02 +0000 (15:13 -0700)]
NASM 2.03rc5
H. Peter Anvin [Wed, 4 Jun 2008 22:11:23 +0000 (15:11 -0700)]
Fix double free in %depend
H. Peter Anvin [Wed, 4 Jun 2008 18:26:59 +0000 (11:26 -0700)]
Fix memory management issues with expanded %include
Ownership of the filename string was a bit fuzzy, with the result that
we were freeing it even though it was retained for use by __FILE__.
Clean up a number of other memory management issues with the new
quoting code, and change the stdscan implementation to one pass over
the string.
H. Peter Anvin [Wed, 4 Jun 2008 15:58:53 +0000 (08:58 -0700)]
Update .gitignore, remove bogus dependency
H. Peter Anvin [Wed, 4 Jun 2008 15:49:10 +0000 (08:49 -0700)]
Makefile/dependency updates
Add quote.c to all the auxiliary Makefiles, and run "make alldeps".
H. Peter Anvin [Tue, 3 Jun 2008 01:32:01 +0000 (18:32 -0700)]
doc: Remove "what's new" since it is out of date
H. Peter Anvin [Mon, 2 Jun 2008 20:59:09 +0000 (13:59 -0700)]
quote: Change the definition of escp
Semi-arbitrary change of the definition of escp to the beginning of
the argument sequence instead of the initiator character. This may
avoid an add in some code paths, and looks slightly cleaner to me.
H. Peter Anvin [Mon, 2 Jun 2008 17:38:54 +0000 (10:38 -0700)]
quote: be consistent in not using C escapes for bytes
We used numbers in nasm_unquote and C escapes in nasm_quote - use
numbers in both places, just in case some C compiler does something
weird with '\r' and (especially) '\n'.
H. Peter Anvin [Mon, 2 Jun 2008 17:02:36 +0000 (10:02 -0700)]
nasm_unquote: make code a little more uniform
Make the code a bit more consistent:
- ndig is now always a countdown, and we always to the (p > escp+1)
test to see if we got anything at all (this is to deal with stuff
like \x without a digit.)
- Add missing break; after 'v' (bug!).
- Preinitialize nval to zero.
H. Peter Anvin [Mon, 2 Jun 2008 16:57:46 +0000 (09:57 -0700)]
quote: massively simplify nasm_skip_string()
Greatly simplify nasm_skip_string() by observing that for the purpose
of string skipping, all states other than st_backslash are equivalent
to st_start.
H. Peter Anvin [Mon, 2 Jun 2008 06:00:23 +0000 (23:00 -0700)]
Document UTF-8 conversion better, use "byte" instead of "ASCII"
ASCII specifically refers to characters <= 127, so to use "ASCII" for
literal bytes is really confusing in a multibyte environment. Give an
example of using Unicode escapes.
H. Peter Anvin [Mon, 2 Jun 2008 05:53:15 +0000 (22:53 -0700)]
Document that underscores in numbers is permitted.
H. Peter Anvin [Mon, 2 Jun 2008 05:37:32 +0000 (22:37 -0700)]
NASM 2.03rc4
H. Peter Anvin [Mon, 2 Jun 2008 05:36:56 +0000 (22:36 -0700)]
Update CHANGES
H. Peter Anvin [Mon, 2 Jun 2008 05:35:47 +0000 (22:35 -0700)]
nasmdoc: document %substr, `...`
H. Peter Anvin [Mon, 2 Jun 2008 05:34:55 +0000 (22:34 -0700)]
Fix bug where the WinHelp backend corrupts the internal data
WinHelp/RTF needs to convert \ to \\, but did so on the global data,
so the DIP output was corrupted.
H. Peter Anvin [Mon, 2 Jun 2008 04:43:03 +0000 (21:43 -0700)]
qstring: fix unquoting in %pathsearch directive
%pathsearch unquoting should be done on the "t" token, not on the
"tline" token...
H. Peter Anvin [Mon, 2 Jun 2008 04:34:49 +0000 (21:34 -0700)]
qstring: backquoted strings seem to work now...
Hopefully backquoted strings should work correctly now.
H. Peter Anvin [Mon, 2 Jun 2008 00:23:51 +0000 (17:23 -0700)]
qstring: first cut at full quoted string support in the preprocessor
First attempt at properly handle quoted strings in the preprocessor.
This also adds range support in %substr.
No support in the assembler yet.
H. Peter Anvin [Sun, 1 Jun 2008 23:07:48 +0000 (16:07 -0700)]
qstring: add nasm_unquote() supporting `...`
Add a nasm_unquote() function supporting the intended `...` syntax.
H. Peter Anvin [Fri, 30 May 2008 21:32:16 +0000 (14:32 -0700)]
doc: Single section for all the date/time macros
Move all the date/time macros to a common section so we can give a
common example. We don't have support for a fourth level of headers,
so just use a bulleted list.
H. Peter Anvin [Fri, 30 May 2008 21:11:40 +0000 (14:11 -0700)]
doc: fix __UTC_DATE__ and __UTC_TIME__ description
H. Peter Anvin [Fri, 30 May 2008 18:31:35 +0000 (11:31 -0700)]
incbin: use the macro name itself as context name
Use the macro name itself as the context name, this is consistent with
other standard macros.
H. Peter Anvin [Fri, 30 May 2008 18:08:14 +0000 (11:08 -0700)]
Update CHANGES
H. Peter Anvin [Fri, 30 May 2008 18:06:28 +0000 (11:06 -0700)]
Add missing %rotate directive to the FLOAT macro.
H. Peter Anvin [Fri, 30 May 2008 18:04:07 +0000 (11:04 -0700)]
incbin definition: use %?
H. Peter Anvin [Fri, 30 May 2008 18:03:56 +0000 (11:03 -0700)]
Document %pathsearch and %depend
H. Peter Anvin [Fri, 30 May 2008 17:42:30 +0000 (10:42 -0700)]
Introduce %depend and %pathsearch, and make incbin a macro
Introduce new preprocessor directives %depend and %pathsearch, and
make incbin a standard macro using these filenames. This lets us
remove the code that makes incbin search the path.
H. Peter Anvin [Fri, 30 May 2008 17:26:28 +0000 (10:26 -0700)]
stdscan.c: clarify comment
There is a reason rn_warn isn't checked in this particular case...
H. Peter Anvin [Fri, 30 May 2008 17:09:45 +0000 (10:09 -0700)]
preproc.c: %include: use expand_smacros() not expand_smacros_in_string()
Call expand_smacros() early instead of expand_smacros_in_string()
late. expand_smacros_in_string() seems like a prodigiously bad idea
and a sheer brainfart in my opinion.
H. Peter Anvin [Fri, 30 May 2008 17:09:24 +0000 (10:09 -0700)]
doc: consistent capitalization (and indexing!) of INCBIN
H. Peter Anvin [Fri, 30 May 2008 04:38:00 +0000 (21:38 -0700)]
Add the -MP option to emit phony targets
Add the -MP option to emit phony targets. Since this means each
header file has to be visited more than once, change the
implementation to use an internal list of all the dependencies, and
centralize the emission of the dependency files.
H. Peter Anvin [Fri, 30 May 2008 02:29:22 +0000 (19:29 -0700)]
Add new options to CHANGES
H. Peter Anvin [Fri, 30 May 2008 02:27:35 +0000 (19:27 -0700)]
NASM 2.03rc3
H. Peter Anvin [Fri, 30 May 2008 02:27:05 +0000 (19:27 -0700)]
Document the new dependency options.
H. Peter Anvin [Fri, 30 May 2008 02:09:11 +0000 (19:09 -0700)]
Implement -MD, -MF, -MT, -MQ
Implement the dependency options:
-MF: set the file to which dependencies are written.
-MD: generate dependencies in parallel with compilation.
-MT: set the name of the dependency target.
-MQ: same as -MT, but *attempt* to quote it for Makefile safety.
H. Peter Anvin [Fri, 30 May 2008 02:08:08 +0000 (19:08 -0700)]
preproc.c: get_ctx() can return NULL...
Handle the case where we would attempt to look up a possible local
context just to find that one doesn't exist.
H. Peter Anvin [Thu, 29 May 2008 05:11:13 +0000 (22:11 -0700)]
Missing % in %ifmacro
H. Peter Anvin [Wed, 28 May 2008 19:28:58 +0000 (12:28 -0700)]
hash user allocates struct hash_table
struct hash_table, a fixed-sized structure, is now allocated by the
caller. This lets us integrate it into the Context structure, thus
avoiding an additional dynamically allocated object for no good
reason.
Add some minor code collapsing: make it more obvious that all that
differs is a pointer value, rather than relying on the compiler to do
tail merging.
H. Peter Anvin [Wed, 28 May 2008 01:15:39 +0000 (18:15 -0700)]
Merge commit 'autotest/master'
H. Peter Anvin [Tue, 27 May 2008 21:43:48 +0000 (14:43 -0700)]
tokhash.pl: fix comment
Correct filename in comment
H. Peter Anvin [Tue, 27 May 2008 21:43:14 +0000 (14:43 -0700)]
regs.pl: add include guards to generated header files
Add include guards to generated header files, and make sure the
appropriate prerequisite headers are included if necessary.
H. Peter Anvin [Tue, 27 May 2008 21:34:53 +0000 (14:34 -0700)]
make alldeps
H. Peter Anvin [Tue, 27 May 2008 21:32:55 +0000 (14:32 -0700)]
regs.pl: regdis.h needs "extern"; minor cleanups
The declarations in regdis.h need to be "extern", since they are
external data.
regdis.c should include regdis.h.
Minor stylistic cleanups in regs.c.
H. Peter Anvin [Tue, 27 May 2008 21:23:20 +0000 (14:23 -0700)]
NASM 2.03rc2
H. Peter Anvin [Tue, 27 May 2008 21:22:19 +0000 (14:22 -0700)]
insns.dat: whitespace cleanup
H. Peter Anvin [Tue, 27 May 2008 21:20:21 +0000 (14:20 -0700)]
Fix double 66 prefixes on INVEPT/INVVPID (BR 1956955)
Fix double 66 prefixes on INVEPT/INVVPID in 16-bit mode, per BR
1956955.
Andy Polyakov [Tue, 27 May 2008 21:03:09 +0000 (14:03 -0700)]
doc: document Win32/64 SEH extensions
Document COFF extensions for Windows SEH
Andy Polyakov [Tue, 27 May 2008 21:01:25 +0000 (14:01 -0700)]
outcoff.c: remove dead code, add check for imagerel
Remove commented-out (dead) code, add check that the imagerel
references are valid.
H. Peter Anvin [Tue, 27 May 2008 05:52:57 +0000 (22:52 -0700)]
avx.bin: clean up screwy whitespace
H. Peter Anvin [Tue, 27 May 2008 05:48:51 +0000 (22:48 -0700)]
VCVTPD2PS, VCVTPD2DQ, VCVTTPD2DQ mem need explicit op size (BR 1974170)
BR 1974170: VCVTPD2PS, VCVTPD2DQ, VCVTTPD2DQ with a memory operand are
ambiguous without a specific operand size, so force one to be added.
Split the instruction pattern due to our current clunky handling of
MMX/XMM/YMM registers together with sizes. Fix in the future, please!
H. Peter Anvin [Tue, 27 May 2008 05:42:02 +0000 (22:42 -0700)]
Fix parameters to VCVTPD2DQ (BR 1974159)
H. Peter Anvin [Tue, 27 May 2008 02:32:52 +0000 (19:32 -0700)]
regs.pl: add comments to regflags.c and regvals.c
Add comments to regflags.c and regvals.c, to make it easier to spot
errors.
H. Peter Anvin [Tue, 27 May 2008 02:23:01 +0000 (19:23 -0700)]
Fix register numbers for ymm1-15!
ymm1-15 were incorrectly listed as starting at register number 0, with
obviously disastrous consequences...
H. Peter Anvin [Mon, 26 May 2008 19:03:55 +0000 (12:03 -0700)]
ndisasm: the high bit of is4 bytes is ignored in 32-bit mode
Mask the high bit of is4 bytes in 32-bit mode. Provide a generic
"regmask" variable that we can use for equivalent purposes as needed.
H. Peter Anvin [Mon, 26 May 2008 18:39:00 +0000 (11:39 -0700)]
macros.pl: more readable comments
H. Peter Anvin [Mon, 26 May 2008 18:21:25 +0000 (11:21 -0700)]
Add some comments to the output of macros.pl
Add some comments to the output of macros.pl, which might make it a
little bit easier to debug issues if there should be any.
H. Peter Anvin [Mon, 26 May 2008 18:14:38 +0000 (11:14 -0700)]
preproc.c: prevent the compiler from reducing enum pp_token_type
The compiler is free to store enum pp_token_type into any size integer
small enough to contain all the values up to 2^n-1 for the smallest n
which contains all the values. Force it to size it to integer size,
since we use it to hold macro positional parameters.
H. Peter Anvin [Mon, 26 May 2008 05:29:00 +0000 (22:29 -0700)]
NASM 2.03rc1
H. Peter Anvin [Mon, 26 May 2008 05:11:36 +0000 (22:11 -0700)]
CHANGES entry for %? and %??
H. Peter Anvin [Mon, 26 May 2008 05:09:17 +0000 (22:09 -0700)]
Document %? and %??
H. Peter Anvin [Mon, 26 May 2008 04:57:00 +0000 (21:57 -0700)]
Implement %? and %?? for mmacros
Same as before, but for mmacros this time
H. Peter Anvin [Mon, 26 May 2008 04:46:17 +0000 (21:46 -0700)]
Implement %? and %?? for smacros
%? - substitute the macro name as invoked
%?? - substitute the macro name as defined
In particular:
%idefine keyword $%?
... can be used to make a new keyword "disappear".
H. Peter Anvin [Mon, 26 May 2008 01:44:44 +0000 (18:44 -0700)]
phash: canonicalize order, fix handling of ignored duplicates
Canonicalize the order of the prehash entries, so we don't have to
worry about looking up both pairs of edges.
When we find a collision that we decide to ignore, there is no point
in adding the same edge into the array again; instead, just skip the
current edge.
H. Peter Anvin [Mon, 26 May 2008 01:17:49 +0000 (18:17 -0700)]
phash: allow collisions if the hash target is the same
If the hash target is the same value, we can permit collisions. This
isn't relevant for the current applications of the hash generator, but
is useful for cases where one have a number of sources for the same
target. It's easy to check, either way.
H. Peter Anvin [Mon, 26 May 2008 01:10:57 +0000 (18:10 -0700)]
phash: massively speed up the perfect hash generator
Make the perfect hash generator about 200x faster by using a very
simple custom graph adjacency representation instead of using
Graph::Undirected.
H. Peter Anvin [Sun, 25 May 2008 16:25:47 +0000 (09:25 -0700)]
Fix mnemnonics for SSE5 PCOMU instructions
H. Peter Anvin [Sun, 25 May 2008 05:15:56 +0000 (22:15 -0700)]
Fix mnemonics for VTESTP[SD] (BR 1971570)
Incorrectly entered as VPTEST* due to illogical placement in the manual.
H. Peter Anvin [Sun, 25 May 2008 05:13:33 +0000 (22:13 -0700)]
Fix the VPSHUF*W instructions (BR 1971567)
The VPSHUF*W instructions had both wrong mnemonics and opcodes.
H. Peter Anvin [Sun, 25 May 2008 05:11:44 +0000 (22:11 -0700)]
Fix typo in VPCMPESTRM instruction (BR 1971565)
The VPCMPESTRM instruction was typoed.
H. Peter Anvin [Sun, 25 May 2008 05:09:51 +0000 (22:09 -0700)]
Add VCVTSI2SS (BR 1971564)
The VCVTSI2SS instruction was missing.
H. Peter Anvin [Sun, 25 May 2008 05:07:03 +0000 (22:07 -0700)]
Fix immediate for PCLMULHQ* instructions (BR 1971555)
The immediate for the PCLMULHQ* instructions was wrong.
H. Peter Anvin [Sun, 25 May 2008 05:04:23 +0000 (22:04 -0700)]
Remove imm from specific versions of VCMPxx
For the versions of VCMPxx which already embed their condition code,
we do not want an extra immediate argument.
Todo: fix bytecode compiler to complain more about these.
H. Peter Anvin [Sun, 25 May 2008 04:58:59 +0000 (21:58 -0700)]
Add VLDQQU as an alias for 256-bit VLDDQU (BR 1971539)
Accept VLDQQU as an alias for VLDDQU when used with 256-bit values.
H. Peter Anvin [Sun, 25 May 2008 04:54:09 +0000 (21:54 -0700)]
VFMSUBADDP[SD], not VFMADDSUBS[SD] (BR 1971573)
There are VFMSUBADDP instructions, but there are no VFMADDSUBS
instructions.
H. Peter Anvin [Sat, 24 May 2008 04:51:21 +0000 (21:51 -0700)]
Add note about SEH support
Still need documentation from someone who actually knows anything
about how this works.
Andy Polyakov [Sat, 24 May 2008 02:45:39 +0000 (19:45 -0700)]
Win64: IMAGEREL references (SEH support for Win64)
Guess what, SEH again, but in Win64 context, which is completely
different matter from Win32. At lowest level this one boils down to
putting so called imagerel references, or in practical terms
relocations of type ADDR32NB, 0x0003, into .pdata and .xdata
segments. Two possibilities. 1. implement say 'wrt ..imagerel' or 'wrt
..imagebase'. 2. silently enforce ADDR32NB relocations in .pdata and
.xdata segments.
H. Peter Anvin [Sat, 24 May 2008 01:38:51 +0000 (18:38 -0700)]
Make SAFESEH syntax error ERR_NONFATAL
Incorrect use of SAFESEH is more suited for ERR_NONFATAL than ERR_FATAL
Andy Polyakov [Sat, 24 May 2008 01:37:52 +0000 (18:37 -0700)]
Support for Win32 safeseh (BR 195417)
This is basically not a bug report, but a feature request.
It's desired to be able to link .obj modules compiled with 'nasm -f
win32' with Microsoft 'link /safeseh'. As well as to register symbols
(commonly subroutine's entry points or even external symbols) as "safe
handlers." In order to achieve this, several points are required.
First of all, object module has to have absolute symbol named @feat.00
with value of 1. This can actually be achived by adding 'absolute 1'
and '@feat.00:' to source code, but it's desirable that it's
autogenerated for win32 modules.
Handler registration is essentially symbol's *index* in current
module's symbol table in .sxdata, segment with 0x200 segment flags, an
"info" segment. It's also essential that symbol has type 0x20 (see
below). All this is depicted in following framgents of 'objdump -xD'
output:
Sections:
Idx Name Size VMA LMA File off Algn
0 .sxdata
00000004 00000000 00000000 0000003c 2**2
CONTENTS, READONLY, DEBUGGING
SYMBOL TABLE:
...
[ 5](sec -1)(fl 0x00)(ty 0)(scl 3) (nx 0) 0x00000001 @feat.00
[ 6](sec 0)(fl 0x00)(ty 20)(scl 2) (nx 0) 0x00000000 _handler
...
<.sxdata>
0: 06 00 00 00
Note [6] and (ty 20) in _handle line in SYMBOL TABLE. "06 00 00 00" in
.sxdata is little-endian 6, _handler's index. This is what makes up
"registration." It's impossible to achieve this with current nasm
facilities and it's probably appropriate to introduce a directive for
it, 'safeseh _handler' is probably most natural choice.
H. Peter Anvin [Sat, 24 May 2008 01:24:03 +0000 (18:24 -0700)]
doc: another reference to DY
H. Peter Anvin [Sat, 24 May 2008 01:03:26 +0000 (18:03 -0700)]
CHANGES since 2.02, so far...
H. Peter Anvin [Sat, 24 May 2008 00:58:01 +0000 (17:58 -0700)]
Document DY, RESY and YWORD
H. Peter Anvin [Sat, 24 May 2008 00:46:08 +0000 (17:46 -0700)]
AVX FMA: Instruction table for the AVX FMA instructions
This adds the AVX FMA instructions to the instruction table, which
should complete the AVX work.
H. Peter Anvin [Sat, 24 May 2008 00:27:15 +0000 (17:27 -0700)]
AVX: Remaining AVX instructions (still need FMA)
Implement the remaining set of AVX instructions
H. Peter Anvin [Sat, 24 May 2008 00:25:54 +0000 (17:25 -0700)]
insns.pl: better error messages, handle no-operand instructions better
Better error messages, and allow "void" instructions to omit the
operand colon.
H. Peter Anvin [Sat, 24 May 2008 00:03:30 +0000 (17:03 -0700)]
AVX instruction table through "P"
AVX instruction table through the letter P in the manual
H. Peter Anvin [Fri, 23 May 2008 23:59:59 +0000 (16:59 -0700)]
fmtinsns.pl: handle compiler fields; do some case mangling
Handle compiler-generated fields, and at least try to do some case
mangling.
H. Peter Anvin [Thu, 22 May 2008 20:17:51 +0000 (13:17 -0700)]
Use hash tables even for context-sensitive macros
Normally, contexts aren't used with a large number of macros, but in
case someone does, do use hash tables for those as well. This
simplifies the code somewhat, since *all* handling of macros is now
done via hash tables.
Future note: consider if it wouldn't be better to allow struct
hash_table to be allocated by the caller, instead of being allocated
by the hash table routine.
H. Peter Anvin [Thu, 22 May 2008 18:24:35 +0000 (11:24 -0700)]
Add tokens vex.ww and vex.wx; vex.wx is the default
Add vex.ww (for VEX.W follows REX.W) and vex.wx (for VEX.W is a don't
care); vex.wx is the default since that seems to match existing usage
better.
H. Peter Anvin [Thu, 22 May 2008 06:21:57 +0000 (23:21 -0700)]
AVX: instruction table up to PE
Complete the instruction table up to and including PE (document
319433-002, start next on page 5-330).
H. Peter Anvin [Thu, 22 May 2008 01:10:09 +0000 (18:10 -0700)]
AVX: instruction table through M
Implement the AVX instruction table through the letter M.
H. Peter Anvin [Thu, 22 May 2008 01:09:17 +0000 (18:09 -0700)]
insns.pl: error if we have a 'v' operand and no vex.nds/ndd
Although redundant, the presence of a 'v' operand should match the
existence of a nds or ndd flag on vex; this should help catch typos.
H. Peter Anvin [Wed, 21 May 2008 22:09:31 +0000 (15:09 -0700)]
insnsn.c: cleaner to *not* separate out conditional instructions
The disassembler code gets cleaner if we do *not* separate out the
conditional instructions; instead, rely on the fact that the
conditionals are always at the end and use FIRST_COND_OPCODE as a
barrier.
H. Peter Anvin [Wed, 21 May 2008 22:03:26 +0000 (15:03 -0700)]
Add legacy blendvpd to avx.asm, as a disassembler test.
H. Peter Anvin [Wed, 21 May 2008 22:02:30 +0000 (15:02 -0700)]
Implement aliases for specific SSE5 compare operations
Implement aliases for specific SSE5 compare operations, per BR 1930630.
H. Peter Anvin [Wed, 21 May 2008 21:45:46 +0000 (14:45 -0700)]
insns.dat: reimplement SSE5 compares using the bytecode compiler
Use the bytecode compiler for the SSE5 compare instructions. While we
are at it, give it correct flags.
H. Peter Anvin [Wed, 21 May 2008 21:44:42 +0000 (14:44 -0700)]
insns.pl: match AMD documentation for DREX syntax
Adjust the compiler to give a syntax for DREX instructions that
matches the AMD documentation.
H. Peter Anvin [Wed, 21 May 2008 21:28:42 +0000 (14:28 -0700)]
Add the PCLMUL instructions (BR 1933742)
Add the PCLMUL group instructions, from the AVX spec but not actually
AVX instructions.
H. Peter Anvin [Wed, 21 May 2008 21:17:33 +0000 (14:17 -0700)]
Add INVEPT and INVVPID (BR 1956955)
H. Peter Anvin [Wed, 21 May 2008 21:16:49 +0000 (14:16 -0700)]
Use "o64nw" instead of "o64i" for o64 without REX.W
Use the somewhat more mnemonic "o64nw" for 64-bit operand size sans
REX.W.