Some important fixes:
- Fix incorrect labels offset for VEX intructions
- Eliminate bogus warning on implicit operand size override.
- %if term could not handle 64 bit numbers.
- The COFF backend was limiting relocations number to 16 bits even if
in real there were a way more relocations.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
The implicit operand size override code didn't set the operand size
prefix, which confused the size calculation code for the range check.
The BITS 64 operand size calculation is still off, but "fixing" it by
making it 32-bit unless REX.W is set breaks PUSH and maybe others.
reloc_value returns 64bit numbers but we strip it down
to 'int' which causes problems if the former value is
big enough to overflow 'int'. Fix it.
[ BR3104312 ]
Reported-by: Christian Masloch
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
The backport of
4db724fdd7359b63f89701102ee8e62672af7379
so coff output target to be able to handle
massive relocations.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
calcsize() had the wrong criterion for when C5 prefixes are permitted
(REX.R is permitted, REX.X is forbidden.) assemble() had the right
test already. This caused symbol value errors.
As being pointed by "matching braces" topic on
[ http://forum.nasm.us/index.php?topic=905.0 ]
we don't issue warning on missed match for "{"
brace opened.
Strictly speaking we should issue error instead and
force user to fix asm source code but since it's
here for a long time already -- lets be "admissive".
Reported-by: Klod
CC: Frank Kotler <fbkotler@zytor.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
We could have accessed malloc'ed data on external symbols
in obj and ieee output formats. Fix it by using nasm_zalloc.
Reported-by: Jiri Malak
Patch-by: Jiri Malak
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
If we're to print inside %rep block we should find
out which %macro it belongs.
Reported-by: Rob Neff
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
It seems to be a bit long story for the reason if this bug. But
lets be verbose and describe all byte-to-byte. And it is all about
preprocessor code, in particular paste_tokens and expand_mmac_params.
Initially the problem (not the same but similar) was noticed and
fixed in commit ec88c1be. The problem reveals itself with code snippets
like
| %macro m 1
| %push
| %define %$arg %1
| %%top_%$arg:
| resb ($ - %%top_%$arg)
| %pop
| %endmacro
So with commits ec88c1be, 51fd86e0, 1f6741fc, 985d880c we did expand
local single macro before processing tokens pasting unconditionally.
But then it being found that such approach breaks %assign directive.
The snippets like below didn't work
| %macro m 1
| %push
| %assign %$arg %1
| %assign %$arg %1+%$arg
| %pop
| %endmacro
So all these commits were reverted and we just stop pasting tokens
in paste_tokens() after TOK_PREPROC_ID (commit 20a94ad7). Unfortunately
this breaks %assign with compound preproc id
| %macro m3 1
| %push
| %assign %$_uses 0
| %rep 4
| %assign %$_ur%$_uses %$_uses
| mov ecx, %$_ur%$_uses
| %assign %$_uses %$_uses+1
| %endrep
| %pop
| %endmacro
To fix this bug we have to combine two approaches at once,
we should continue pasting after TOK_PREPROC_ID and expand
sequential TOK_PREPROC_IDs except first one.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
error() routine is conditional dependent so we should
use nasm_error instead to yield message unconditionally.
Reported-by: Christian Masloch
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Smacros are apparently stored with the token stream reversed, so make
sure %deftok matches that sense of relatity.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Make %substr robust to handle -1,-1 parameters
and restore old behavior when number of characters
in substring is greater then length of string itself.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
In case if a second parameter of %deftok is missed we hit
NULL dereference. Fix it.
Reported-by: Christian Masloch
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
A lot of instructions ignore the L bit in the VEX prefix, just like
a lot of instructions ignore the W bit, so don't use them in the
sub-table select.
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Fix handling of byte codes 250-253 (sign-extended 32-bit immediate,
extended to 64 bits) and 324 (instruction must have osize 64).
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Allow implicit operands for VBLENDVP, just as for other instructions,
since the semi-legacy forms now are removed.
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Version 7 of the AVX spec specifically forbids (#UD) using the
66 0F 38 14/15 forms of the BLENDV instructions with a VEX prefix;
those encodings are strictly legacy SSE 4.1.
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Updates from the AVX version 7 specification: mostly tightening of the
rules for VEX.L and VEX.W, but remove the VPERMIL2 instructions.
Also encode all the full-length forms of the VCMP instructions and
prefer those for the disassembly.
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Change the .wx (ignore the W field) to .wig, to match the latest
version of the AVX specification. This is not a functional change,
but just makes instruction patterns a little easier to write.
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
We use -Ox by default since Nasm-2.09 so "-O0" is not
longer "default" option. But first paragraph still has
the reference which confuses people. Get rid of it.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
