doc: changes -- Add fixes for VMOVNTDQA, MOVNTDQA, MOVLPD

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

test: Add br978756

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

test: Add br3392259

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns: Fix MOVLPD

http://bugzilla.nasm.us/show_bug.cgi?id=978756

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns: Fix MOVNTDQA instruction

http://bugzilla.nasm.us/show_bug.cgi?id=3392259

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns: Fix VMOVNTDQA instruction

http://bugzilla.nasm.us/show_bug.cgi?id=3392259

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

changes.src: Prepare for 2.10.08

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

BR 3392260: Handle instructions only separated by vector SIB size

There are two instructions (VGATHERQPS, VPGATHERQD) where the only
separation between two forms is the vector length given to the vector
SIB.  This means the *matcher* has to be able to distinguish
instructions by vector SIB length and the matcher only operates on the
operands and the instruction flags, not on the bytecode.

Export the vector index-ness into the operand flags and add to the
matcher.

This resolves BR 3392260.

Reported-by: Agner <agner@anger.org>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>

parse_line: Shrink code with memset helper

We know that P_none = 0 thus instead of using
for() statement assign them all in one memset
call.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

ea_type: Fix typo in comment

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

autoconf: Warn a user if no asciidoc/xmlto found

It's been noticed on nasm forum that we don't complain
if no asciidoc/xmlto package installed on a system while
they are required for man pages generation.

So warn a user then, but still allow to build nasm.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

make: Fix nasm build with MSVC

http://bugzilla.nasm.us/show_bug.cgi?id=3392258

Signed-off-by: Marat Dukhan <maratek@gmail.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns: Fix vspllw instruction

http://bugzilla.nasm.us/show_bug.cgi?id=3392257

Signed-off-by: MITSUNARI Shigeo <herumi@nifty.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns: Fix vgatherqpd instruction

Signed-off-by: MITSUNARI Shigeo <herumi@nifty.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns: Fix VPMOVSXBQ instruction

http://bugzilla.nasm.us/show_bug.cgi?id=3392254

Signed-off-by: MITSUNARI Shigeo <herumi@nifty.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Make the errors for missing man page tools comprehensible

A lot of confusion was had due to missing asciidoc...

Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>

test: Add br3392252 test

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Add note about AMD TBM instructions

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Add BLCMSK

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Add BLCS

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Add BLSFILL

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Add BLCFILL

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Add BLCIC

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Add BLCI

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Add BLSIC

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Add immediate form of BEXTR

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Add T1MSKC intstruction

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Add TZMSK instruction

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Move TZCNT for alphabetical order

Other instructions in this section already sorted.
No func changes.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Make .git-ignore directory to ignore by scm

To stash various code which a developer not
need to be tracked.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Add IF_TBM flag

For TMB instructions

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

BR3392253: Fix potential buffer overflow in number conversion

Reported-by: franck.uberto@esrf.fr
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.dat: Udate year

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

br3392250: insns -- Allow byte size in PREFETCHTx instructions

The PREFETCHTx instructions do allow mem8 (byte).

Reported-by: Agner <agner@agner.org>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

BR3392244: docs -- Fix "respectively" typo

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

BR3392248: Update nasm -h output

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

BR3392248: Update nasm manual

Add -W description

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

ndisasm: man -- Add missing -p option

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

BR3392248: Update ndisasm man

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

BR3392248: Update ndisasm man

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

BR3392248: Update nasm manual

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Fixed two more resource leaks in rdoff/rdlib.c

Signed-off-by: Philipp Kloke <philipp.kloke@web.de>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Fixed three possible buffer overflows

Bug found by: CppCheck 1.59 (static source analysis tool)

Signed-off-by: Philipp Kloke <philipp.kloke@web.de>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Fixed several resource and memory leaks

Bug found by: CppCheck 1.59 (static source analysis tool)

Signed-off-by: Philipp Kloke <philipp.kloke@web.de>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Fixed wrong/redundant comparison according to documentation in comment above.

Bug found by: CppCheck 1.59 (static source analysis tool)

Signed-off-by: Philipp Kloke <philipp.kloke@web.de>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Fixed wrong format specifier in format string

Signed-off-by: Philipp Kloke <philipp.kloke@web.de>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Remove unnecessary calls to memset

The C standard guarantees that strncpy pads
the string with zeros if source string is smaller
than destination buffer.

Signed-off-by: Philipp Kloke <philipp.kloke@web.de>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Make F2 and F3 SSE prefixes override 66

According to XED and experimentation, the 66 is ignored.

Signed-off-by: Ben Rudiak-Gould <benrudiak@gmail.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Drop SAME_AS flag from instruction matcher

It was there to support the SSE5 DREX encoding,
which as far as I know is dead forever.

Signed-off-by: Ben Rudiak-Gould <benrudiak@gmail.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns: Remove pushseg/popseg internal bytecodes

This patch is getting rid of the following bytecodes
'pushseg','popseg','pushseg2','popseg2' and simplifies
overall code.

[gorcunov@: a few style fixes]
Signed-off-by: Ben Rudiak-Gould <benrudiak@gmail.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Move conditional opcodes close to enum ccode definition

Thus if someone need to rework this code he won't need
to jump between files trying to figure out where enum
and opcodes lay.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

insns.pl: Various style polyshing

 - convert tabs to spaces
 - align octal opcodes

No func changes.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

assemble: Make emit_rex being a function

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

assemble: Use case3/4 where appropriate

This allows to shrink code a bit.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

assemble: Add case3 helper

Signed-off-by: cyrill <cyrill@cyrills-MacBook-Pro.local>

process_ea: Drop redundant variable

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Remove +s

It doesn't seem worth >200 lines of C and Perl to save ~50 lines in insns.dat.

In order to make this work I had to rename sbyte16/sbyte32 so that
they can take an ordinary size suffix (their size suffix was formerly
treated specially).

This fixes one disassembly bug: 48C7C000000080 disassembles to mov
rax,0x80000000, which reassembles to B800000080, which loads a
different value.

Signed-off-by: Ben Rudiak-Gould <benrudiak@gmail.com>
Acked-by: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Fix warnings generated by clang 3.0

Fix warnings like this:

output/outelf32.c:2120:33: warning: equality comparison with extraneous
      parentheses [-Wparentheses-equality]
            if ((match->section == index)) {
                 ~~~~~~~~~~~~~~~^~~~~~~~
output/outelf32.c:2120:33: note: remove extraneous parentheses around the
      comparison to silence this warning
            if ((match->section == index)) {
                ~               ^       ~
output/outelf32.c:2120:33: note: use '=' to turn this equality comparison into
      an assignment
            if ((match->section == index)) {
                                ^~
                                =
1 warning generated.

Signed-off-by: Andrew Nayenko <resver@gmail.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

BR3392242: insns.dat -- Support AMD SVM instructions in 32bit mode

AMD CPUs do support SVM instructions in 32-bit mode thus drop X64
restriction from instructions template where appropriate.

Signed-off-by: Andrew Nayenko <resver@gmail.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Fix jmp/call near offsets in long mode

In long mode relative offsets are always 32 bits sign-extended to 64
bits and absolute near addresses are always 64 bits, regardless of the
operand size.

Signed-off-by: Ben Rudiak-Gould <benrudiak@gmail.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Add np and similar prefixes to instructions that should have them

This adds "np" to a bunch of SSE-style instructions that should have
it, "norep" (which was implemented but unused) on quasi-SSE instructions
that use F2 and F3 as instruction extensions but 66 for operand size,
"nof3" (newly implemented) on a few instructions, "norexw" on some
instructions that have only 32-bit and 64-bit versions, and one NOLONG.

It also removes some incorrect "np"s, changes some "f3"s to "f3i"s,
and fixes the decoding of the XCHG/NOP/PAUSE mess: F390 is always
PAUSE even when rex.b=1 (at least according to XED).

Signed-off-by: Ben Rudiak-Gould <benrudiak@gmail.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

eval: Use is_power2 helper instead of open code

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

make: Don't forget to generate manpage for all target

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

man: Generate manpages from asciidoc format

The asciidoc format is a way more easier to read
by a human.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Update .gitignore

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

nasm.c: Convert GET_CURR_OFFS/SET_CURR_OFFS to functions

In future better to not work with global variables but
rather pass arguments.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

nasm.c: Update year in header

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

nasm.c: Tabs to spaces in emit_dependencies

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

nasm.c: Tabs to spaces in main

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

nasm.c: Tabs to spaces in copy_filename

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

nasm.c: Tabs to spaces in copy_filename

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

nasm.c: Tabs to spaces in process_arg

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

nasm.c: Tabs to spaces in process_response_file

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

nasm.c: Tabs to spaces in assemble_file

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

nasm.c: Tabs to spaces in nasm_verror_gnu

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

nasm.c: Tabs to spaces in nasm_verror_common

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

nasm.c: Tabs to spaces in define_macros_early

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Drop trailing space from directive.pl

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

Align enum prefixes members

It's a way easier to read.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

BR3392240: preproc: Don't fail on pasting of space expanded rvalue tokens

Reported-by: KO Myung-Hun <komh@chollian.net>
Tested-by: KO Myung-Hun <komh@chollian.net>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

NASM 2.10.07

docs: Update changes

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

br3392236: Don't treat \Space after \BackSlash as a sign for line continuation

In commit f1fe4fdeabeaf2e5e4d02ef43beeb09a6fbfed1b I occasionally
made a \Space after \BackSlash being a sign of line continuation.

Fix it.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

NASM 2.10.06

changes: document fix of macho64

Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>

BR3392232: Fix relocations in MachO64

Signed-off-by: Keith Kanios <keith@kanios.net>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

output: Add more Elf unification

One day the elf output routines would be abstracted
enough to be merged in one file. This patch simply
removes some differences from elf32/64 code.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

BR3392231: Fix get_closest_section_symbol_by_offset

This patch changes get_closest_section_symbol_by_offset
logic to lookup only the closest symbols which are at
or before the supplied offset.

Signed-off-by: Keith Kanios <keith@kanios.net>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

BR3392226 preproc: Rework line readin procedure

It's been reported that we handle MacOS eol wrong.
This patch fixes the problem.

http://bugzilla.nasm.us/show_bug.cgi?id=3392226

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

doc/changes.src: Update to include recent changes.

Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>

BR 3327107: fix assembly of VPCMPGTQ

Fix incorrect instruction encoding for VPCMPGTQ.

Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>

insns.dat: Mark the immediate for shift instructions as imm8

Allow the form:

      mov <rm>,byte 1

... to generate the explicit byte form.  An unfortunate side effect is
that disassembly is ugly; this could be fixed by making a special byte
code that acts the same for the assembler but disassembles specially.

Signed-off-by: H. Peter Anvin <hpa@zytor.com>

BR 3392227: Remove SB flag from SHIFT rm,imm

Remove the SB flag from the rm,imm forms of the shift/rotate
instructions; presumably the intent was to mark the immediate as a
byte immediate, but it instead caused the memory operand to be
accepted without a size, but generate the incorrect code as a result.

Note: it would be good to modify the UNITY filter so that the explicit
form:

shl eax,byte 1

... forces the imm form rather than the ,1 form.

Signed-off-by: H. Peter Anvin <hpa@zytor.com>

opflags: Rework opflags bits with OP_ macros

In this path the opflags bits are completely reworked
in a sake of simplier extension. Inparticular for Knights
Corener instructions we will need new registers and new
sizes.

What's done

 - all bits are grouped in sequences, and start using OP_
   macros, thus if one need to extend some field -- just
   tune up @shift and @bits where needed

 - the #define we use in code are OR'ed in symbols, this
   should be a way more convenient to deal with instead of
   pure hex numbers.

The tests are passed but more eyes needed to review this
rather big and intrusive patch. The reason why it's done
in one single path -- for revertability in one command.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

opflags: Introduce opflags generating macros

This patch introduces two macros

 - OP_GENMASK, to generate masks in opflags
 - OP_GENBIT, to generate bit at specified position

we will use them with extended 64 bits opflags.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

opflags: Drop never used REG_RIP/REG_EIP

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

When generate dependency names internally, quote filenames

Quote filenames for Make when generated for filenames internally.
Only skip quoting when using the -MT option (rather than -MQ).

Signed-off-by: H. Peter Anvin <hpa@zytor.com>

changes.src: fix grammar error (support -> supports)

Signed-off-by: H. Peter Anvin <hpa@zytor.com>

opflags: Extend opflags_t to 64 bits

Soon we will need to encode 512 bits values
thus there is no space left in our opflags_t
which is 32 bitfield.

Extend it to 64 bits width.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

NASM 2.10.05