Document CPU LATEVEX and the associated prefixes; add CPU EVEX and CPU
VEX flags to further control encodings.
Fix the error message for invalid encodings due to flags.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
For VEX instructions created *after* the corresponding EVEX
instructions, we need the user to either explicitly declare them {vex}
or specifying "cpu latevex".
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
"Why dup_tlist() here? We should own it."
Yes, we own it, but we still need to advance the tail pointer. Create
steal_tlist() for this purpose.
Fixes: https://bugzilla.nasm.us/show_bug.cgi?id=3392774
Reported-and-Debugged-by: C. Masloch <pushbx@ulukai.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Broken %if, %rep and %macro nesting can result in the %exitmacro
unwind overrunning the condition stack. Fix.
Fixes: https://bugzilla.nasm.us/show_bug.cgi?id=3392796
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
The argument to nasm_new() is the pointer, not the indirection from
the pointer. This code is only relevant when compiled without token
recycling (TOKEN_BLOCKSIZE not set), but it is still wrong...
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Fix the expansion of the %00 mmacro parameter; broken due to a missing
change of the token type.
Fixes: https://bugzilla.nasm.us/show_bug.cgi?id=3392803
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Add the option of having strings only conditionally quoted (&&) -- do
not quote an already quoted string again -- as opposed to always
quoting a string.
This makes a lot of the string functions way simpler to implement, and
removes the need to share ad hoc parsing code with directives.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Add the %abs() function, to produce the absolute value as an
always-positive decimal constant.
Change the order of the arguments for %num().
Refactor the handling of optional arguments, to reduce the amount of
redundant code. This is currently only used for builtin functions, but
might be extended in the future.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Add the %num() preprocessor function, which returns a quoted string
with a number formatted in any base between 2 and 64 (using bash
encoding with '@' for 62 and '_' for 63.)
It can specify a fixed number of digits with or without truncation.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Classify all remaining WARN_OTHER warnings in the preprocessor. Move
all preprocessor warnings except "user" under a common pp-* prefix.
Warn for an out-of-range argument to the %sel() function.
Finally, use "dname" in additional places for consistency and future
ease of use.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Implement the %cond() and %sel() functions that expand to a specific
one of the arguments. %cond(x,y,z) is basically a shorthand for
%sel(2-!(x),y,z) used when x is a boolean condition.
Fix a memory leak in %strcat and %strlen.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Allow preprocessor function expansion to recurse. Nearly all the
machinery for recursive smacros was already in place; this merely
activates it for the specific case of preprocessor functions. Making
it a general facility should be deferred to a later relese, though.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Instead of handling conditional instructions ad hoc, generate
individual instruction patterns as normal. This simplifies the code
and makes CMPccXADD support simpler (otherwise it would be necessary
to hack in the handling of a condition code in the middle of an
instruction.)
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Some preprocessor functions have the same name as directives. In those
cases, they should be expanded as functions if and only if they are
followed by a left parenthesis. Although it is not inherently true that
either preprocessor functions require a paren nor that directives
cannot start with one, but it is true and will remain true for all
cases where there is a namespace collision.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
%substr contained a token skip to "skip expanded ID", which is
incorrect, as that has already been skipped at that point. It worked
anyway, accidentally, as this token would always be a whitespace token
-- but we then do skip_white() immediately thereafter.
Delete this to allow this code to be factored.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
This commit adds a check to see if the macro that we want to unmacro exists.
A previous commit, introduced a check to see if the unmacro was undefining a macro being expanded, but that same check included a null pointer dereference if the macro to undefine did not exist.
The following code reproduced the issue:
```asm
%macro baz 0
%unmacro F 0
%endmacro
baz
```
Compile with:
```shell
$ nasm -f elf64 -g -FDWARF -o tmp.o -werror file.asm
```
[hpa: adjusted code to match NASM style]
Fixes bug 3392761
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
The FP16 patch had a case of bit overlap. Clean up the handling of
broadcast flags a little in the process.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Warnings with explicit [rel] would pretty much *always* warn after
checkin f4e7a636a85bab02e7ac0067c5c58768779900a0. Fix this.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
This reverts commit 8fcc785f95b842694015e03d909a3131cbadbeb3.
This patch causes test a32offs.asm, and in general *any* use of the
"bits" macro, to totally fail.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
case PP_ENDM:
case PP_ENDMACRO:
if (!(defining && defining->name)) {
nasm_nonfatal("`%s': not defining a macro", tok_text(tline));
goto done;
}
mmhead = (MMacro **) hash_findi_add(&mmacros, defining->name);
defining->next = *mmhead;
*mmhead = defining;
defining = NULL;
break;
The variable: mmacros has not been released, which will cause a memory
leak. Repair cve-2021-33450 cve-2021-33452 synchronously
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Add support for AVX512-FP16 instructions and the associated
handling. Allow "mapN" syntax as well as "mN" syntax to match the
documentation.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
free() and nasm_free() are required to be compatible (as we may end up
having memory allocated on the heap by the C library), but that
doesn't mean we shouldn't use it whereever possible to allow for
better debugging.
Fixes: https://bugzilla.nasm.us/show_bug.cgi?id=3392804
Reported-by: C. Masloch <pushbx@ulukai.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
An empty %{} becomes % which is simply the arithmetic
operator. Although that is consistent, it might be surprising for
users, to issue a warning.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
%[] amounts to an empty token; this needs to be handled specially so
that it gets properly dropped.
Fixes: https://bugzilla.nasm.us/show_bug.cgi?id=3392806
Reported-by: C. Masloch <pushbx@ulukai.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
while (nbs--) { ... } ends with nbs == -1. Rather than a minimal fix,
introduce mempset() to make these kinds of errors less likely in the
future.
Fixes: https://bugzilla.nasm.us/show_bug.cgi?id=3392815
Reported-by: <13579and24680@gmail.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
The size calculation done in len_extops() (called by insn_size()) for
EOT_DB_RESERVE (i.e. uninitialized storage "?" token) does not take
into account the element size (e->elem), thus calculating a wrong
size for any Dx larger than DB (DW, DQ, etc).
The bug is silent, but it makes NASM error out if a "Dx ?" (larger
than DB) is followed by any label because the label offset gets
mismatched in the final code generation stage:
$ cat test.asm
[section .bss]
DW ?
x:
$ nasm test.asm
test.asm:3: error: label `x' changed during code generation [-w+error=label-redef-late]
See also: https://stackoverflow.com/q/70012188/3889449
Signed-off-by: Marco Bonelli <marco@mebeim.net>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Andrew reported that we may access unitialized memory
> SUMMARY: MemorySanitizer: use-of-uninitialized-value nasm/asm/parser.c:982:41 in parse_line
It turns out that in case of malformed data the expression is terminator
itself so we should not "lookup ahead" for next one. Thus test for first
expression initially and if test passes check for terminator.
Reported-by: Andrew Bao <xiaobaozidi@gmail.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Regression in commit 20e0d616dc954d567c8bf2c7e11cc5d6c10ac544.
Independently discovered and fixed by C. Masloch:
https://bugzilla.nasm.us/show_bug.cgi?id=3392747
Signed-off-by: Oleg Oshmyan <chortos@inbox.lv>
In commit 2469b8b6 we occasionally bring the ability
to read unitialized memory due to refactoring. Fix it
doing needed test inside the function and setting up
an error message if needed.
Side note: passing 7 arguments into the function means
we have to decompose this helper somehow, such number
of arguments is a way over the top.
Bugzilla: https://bugzilla.nasm.us/show_bug.cgi?id=3392751
Reported-by: Marco <mvanotti@protonmail.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Add a {rex} prefix to force REX encoding (typically a redundant 40h
prefix).
For prefix parsing, we can use t_inttwo to encode the prefix slot
number.
Give more verbose error messages for encoding mismatches.
Make the pasting behavior of TOKEN_QMARK, TOKEN_HERE and TOKEN_BASE
match the NASM 2.15 behavior: ? is a keyword and pastes as an ID, $
and $$ are treated as operators (which doesn't seem to make much
sense, but it is the current legacy behavior.)
Reported-by: C. Masloch <pushbx@ulukai.org>
Bugzilla: https://bugzilla.nasm.us/show_bug.cgi?id=3392733
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
If macro is undefined while it's being expanded, use after free occurs,
since the MMacro instance is released, but it is still used to proceed
the expansion.
This change forbids macro undefinition: non-fatal error is raised and
the MMacro instance is not released if it is being processed by NASM
preprocessor.
Consider the following example:
| $ cat test.asm
| %macro m 0
| %unmacro m 0
| %endmacro
| m
| $ ./nasm test.asm
| test.asm:4: error: `%unmacro' can't undefine the macro being expanded
| test.asm:2: ... from macro `m' defined here
Fixes BR3392531 and BR3392716.
Signed-off-by: Igor Munkin <imun@cpan.org>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Better point out explicitly that SMacro::next member
is untouched, thus do not use SMacro::next and an array.
CID 1432925
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
When we process a TOKEN_QMARK we also need to advance p, in order to
get the proper start for the next token.
This fixes travis test br3392707.
Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
