The NASM 2.09 preprocessor allows some illogical constructs, but which
unfortunately has been found in real code in the field. We need
a compatibility solution or a pragma before we can avoid that.
However, we need the other features in NASM 2.10 to come out, so
revert the preprocessor changes for now.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
In case if there a production
{tok},{%+},{whitespace}*
the preprocessor does not delete
ending paste+spaces tokens. Fix it.
http://bugzilla.nasm.us/show_bug.cgi?id=3392200
Reported-by: KO Myung-Hun <komh@chollian.net>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
It is a potential place for a bug if list_for_each
helper is used with multiple lines of code without
a proper bracing.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
In case if there a trailing spaces after expansion
%if[n]def might produce a false alarms on macro id
being expected.
Note it also makes the constructions like bare
%ifdef
%endif
invalid, while before they were passed well without
a notice.
Reported-by: KO Myung-Hun
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
new_ExpDef does zalloc'ate memory so no need
to assign zeros again.
Same time it should fix MSC complains on NULL assignment
on boolean variable.
Reported-by: Jasper Neuman <jasper.neumann@web.de>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
There is no need to hide this structure into a type.
The former preproc_ops is a way more descriptive.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
2.09 series was the last one we support context-thru search
(and we were issuing a warning about that) so drop all-context
from get_ctx() routine.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
This reverts commit cb00cd1ba7f07f60f0a94c43cfc4fa7b01e2ff95.
As Victor pointed out some tests do not pass with this commit
so revert it until things get fixed.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
It's a bit more than that, also TRACE=1 make
flag added to run this facility on and off
at compiling time.
Debug feature only, doesn't affect regular users.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
We simply allow the following terminals to be concat'ed
if they are written without space or any other separator
inbetween.
a := id | preproc-id | number | float | other
b := id | preproc-id | number | float | other
if match(a,b):
s := concat(a,b)
re-tokenize(s)
Basically it means it's up to code author to write
preproc code a way the sane production appears.
Some notes.
1) We don't concat strings.
2) The 'weirdpaste' test fails now because with relaxed
rules it works as needed and was borken before.
The lacmus snippet is
%define N 1e%++%+ 5
dd N, 1e+5
Previously the output was
dd 1e+%+ 5, 1e+5
which is wrong since we have explicit concat here
with %+ operator. The new code production is correct
and looks like
dd 1e+5, 1e+5
as expected.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
src_set_fname simply gets copy of pointer (ideally
we need refcounting here) so don't pass the name
which will be freed soon but rather pass a copy.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
While being debugging some nifty problem I found
that it might be useful to produce a full dump of
tokens, in particular text of tokens.
For this reason dump_token is here just to not loose
it. It doesn't affect normal build procedure since it
requires a special -DNASM_TRACE to be passed to the
compiler. Which of course we don't in a regular
compilations.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
If not all members of structure being allocated from
heap get initialized we better to use nasm_zalloc instead
of nasm_malloc.
For example inc gets allocated in do_directive being parially
initialized and we erroniously get mmac_depth set to some
crappy value leading to SIGSEV in result.
[ http://forum.nasm.us/index.php?topic=921.msg3257#msg3257 ]
nb: I've cleaned verror from tab/space mess while were at it
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
