Hash all directives, even the ones that are backend-specific,
and instead pass the backend an already-parsed directive number.
Furthermore, unify null functions across various backends.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
We fopen() the output file in common code but fclose() it in the
backend. This is bad for a variety of reasons:
1. it is generally an awkward interface to change ownership.
2. we should use ferror() to test for write errors, and that is
better done in common code.
3. it requires more code.
4. we still need to fclose() in common code during error handing.
Thus, move the fclose() of the output out of the backends, and add
fflush() so we can test ferror() on output.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Instead of removing the list file on error, keep the list file and
include the errors in the list file. This makes it actually possible
to debug things that involve deep macro recursion, where the line
number is pretty much meaningless.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
*To the best of my knowledge*, we now have authorization from everyone
who has significantly contributed to NASM in the past. As such,
change the license to the 2-clause BSD license.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Allow the backend to specify that an output format is either text or
binary. For future uses, define this as a flag word so we can define
other flags in the future if it would make sense.
Currently, the ieee and dbg formats are text; all the others are
binary.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Fix report_error() to (hopefully) not fault if used without ERR_NOFILE
if no filename is available.
Avoid nuisance phase error between passes warnings if we have detected
other errors. In those case, the phase error is almost certainly
spurious.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Add copyright headers to the *.c/*.h files in the main directory. For
files where I'm sure enough that we have all the approvals, I have
given them the 2-BSD license, the others have been given the "LGPL for
now" license header. Most of them can probably be changed after
auditing.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Move backend-specific code into the output/ directory, and make the
null debugging backend a separate file (it certainly isn't needed for
ndisasm...)
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
We already call current_dfmt->init in the same place (at the very end
of ofmt->init) in all the backends that do it; instead call it
centrally in nasm.c after ofmt->init.
This fixes invalid ELF files with when compiling with -F dwarf, since
the dwarf initialization routine never got called.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
When optimizing, we have to keep track of common labels, since a
common symbol cannot be optimized -- only the linker will know where
it will end up. In that sense it is similar to an EXTERN symbol.
Thus, allow them to be entered in the symbol table but make sure we
don't holler too hard on redefinition.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Clean up the pass exit logic. In particular, we had a whole routine
to do the same thing that the normal exit logic would do anyway if we
only would actually get there.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Checkin 4b9358928b changed the version
message from stdout to stderr, but:
a) doesn't motivate the change in the commit log;
b) bundles that with other changes;
c) is inconsistent with other programs;
d) was done by me and I can't remember any reason for it.
Hence conclude it was unintentional and therefore a bug. Since this
commit was done after 2.05.01 no stable release has been affected.
nasm.c should respect the default debug format of the output format,
instead of replacing it with the first format in the list.
This is cleaner and allows the list to be sorted normally.
This commit rewrites commit 116994111b which was very fragile.
Move all the version strings to a single compilation unit, ver.c; this
does not include the version macros, which are fed into macros.c.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Better grammar and fix incorrect description of the number-overflow
warning (it is not just limited to 64-bit arithmetic overflow, it also
triggers when trying to squeeze in a value which is too large into an
immediate.)
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
The code to parse the -w/-W options was updating warning_on[], not
warning_on_global[], but warning_on[] is reset at the beginning of
each pass (to let the warning directive work); as a result the -w/-W
options don't actually do anything at all.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Issue better warnings for out-of-range values. This is not yet
complete.
In particular, note we may have out-of-range for values that end up
being subject to optimization. That is because the optimization takes
place on the *truncated* value, not the pre-truncated value.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Put the static information about warnings in a structure, so one can
see what goes with what. Also, change the sense so "true" means
enabled.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Change global_offset_changed from bool to int so that
progress of convergence can be monitored. If change count
does not decrease from previous pass, increment stall counter.
If stall count reaches threshold, terminate assembly
with error message.
Now that there is proper forward reference resolution,
we can get rid of this junk. Wiping the flags also
removed the SBYTEnn flags, causing
cmp eax, a-b
a: nop
b:
to assemble with -Ox like
cmp eax, strict dword -1
This is now fixed.
Users who wish to control the level of optimization can
continue to specify -O0, -O1, or -Ox,
where x can be the letter itself, or any number > 1.
However, even with optimization turned off,
NASM will always make enough passes to resolve
forward references. As a result, INCBIN is now the only
item left in the critical expressions list, although TIMES
still has its own constant value check.
ctype functions take an *int*, which the user is expected to have
taken the input character from getc() and friends, or taken a
character and cast it to (unsigned char).
We don't care about EOF (-1), so use macros that cast to (unsigned
char) for us.
On some platforms, tolower() is implemented as a function call, in
order to handle locale support. We never change locales, so can the
result of tolower() into a table, so we don't have to sit through the
function call every time.
~1.3% overall performance improvement on a macro-heavy benchmark under
Linux x86-64.
- Add %warning directive
- Only unquote an %error or %warning string if it is the only thing on
the directive line.
- Don't expand macros inside a quoted string, even for %error.
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.
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.
First cut at AVX machinery support. The only instruction implemented
is VPERMIL2PS, and it's probably buggy. I'm checking this in with the
hope that other people can start helping out with (a) testing this,
and (b) adding instructions.
NDISASM support is not there yet.
This checkin creates the following date and time macros:
__DATE__, __TIME__, __UTC_DATE__, __UTC_TIME__: strings
__DATE_NUM__, __TIME_NUM__, __UTC_DATE_NUM__, __UTC_TIME_NUM__:
civil dates in digit-string formats
__POSIX_TIME__: time in POSIX time_t format
is_suppressed_warning() should never return true unless we're actually
dealing with a warning. There is a handful of cases where we pass
ERR_PASS1 down together with errors, but that's mostly because it fits
into an overall pattern. Thus, ignore it.
For PASS1 warnings, only do them when pass0 == 1. The prior passes
are to be considered training passes. This is a bit awkward if we
then hit an error, but it's better than n repeated warnings.
The five-pass-minimum was a hack for a bug which I think is identified
now. Doesn't really change the fact that if you want the optimizer,
you probably want -Ox.
We have a number of bug reports about things not working properly when
the optimizer is running out of passes. I suspect the reason is
simply that we don't properly execute the final passes (pass0 = 1, 2)
when hitting the limit. Make sure we advance pass0 the last few
times.
