arrays. An array element has all qualifiers from itself and from the array
declaration.
git-svn-id: svn://svn.cc65.org/cc65/trunk@4334 b7a2c559-68d2-44c3-8de9-860c34a00d81
function", so dereferencing a function is a legal operation. The result is
immediately converted back to "pointer to function" again. Note: Because of
performance considerations, cc65 does treat functions as "pointer to function"
internally, but nevertheless dereferencing a function must be possible.
git-svn-id: svn://svn.cc65.org/cc65/trunk@4330 b7a2c559-68d2-44c3-8de9-860c34a00d81
expression that yields a constant value may have side effects, and the
expression code must not be removed in this case.
git-svn-id: svn://svn.cc65.org/cc65/trunk@4138 b7a2c559-68d2-44c3-8de9-860c34a00d81
expression that yields a constant value may have side effects, and the
expression code must not be removed in this case.
git-svn-id: svn://svn.cc65.org/cc65/trunk@4137 b7a2c559-68d2-44c3-8de9-860c34a00d81
second operand is necessary. Instead of generating interleaved code with
several jumps, just move the code to the right place.
git-svn-id: svn://svn.cc65.org/cc65/trunk@4105 b7a2c559-68d2-44c3-8de9-860c34a00d81
protect against missing parenthesis like in (si & 0x7F == 0). Hopefully
without too many false positives.
git-svn-id: svn://svn.cc65.org/cc65/trunk@4057 b7a2c559-68d2-44c3-8de9-860c34a00d81
is bad, since information is lost when converting functions and arrays to
pointers and cannot be regained easily. So I've reverted the change and fixed
the problem in a less general manner.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3984 b7a2c559-68d2-44c3-8de9-860c34a00d81
type. Started to add general handling of address size flags in types.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3891 b7a2c559-68d2-44c3-8de9-860c34a00d81
using signed integers as data types. This led to wrong results for mod, div
and compares.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3797 b7a2c559-68d2-44c3-8de9-860c34a00d81
to element. This will make some valid code work but emits now errors for
questionable code.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3775 b7a2c559-68d2-44c3-8de9-860c34a00d81
identifier is not the last token on the line, the compiler will run into
an error.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3744 b7a2c559-68d2-44c3-8de9-860c34a00d81
elements to a string of structs, each representing one type element. This
should fix problems on unusual architectures, since it is no longer necessary
to embedd pointers and other data converted to numbers in the string of
unsigned shorts.
Increased the TypeCode length to unsigned long to make room for more type
bits.
Inline more functions in datatype.h.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3709 b7a2c559-68d2-44c3-8de9-860c34a00d81
The ArrayRef() does really need a rewrite, now since we have better
expression info.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3364 b7a2c559-68d2-44c3-8de9-860c34a00d81
arrays to pointer to first element, and functions to pointer to function.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3169 b7a2c559-68d2-44c3-8de9-860c34a00d81
Changed GetCodePos to also remember the stack pointer at the given location,
this removes the necessity to manipulate the stack when removing code. Since
CodeMark is now a struct, the API for most asmcode functions has changed.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3145 b7a2c559-68d2-44c3-8de9-860c34a00d81
Copy the current optimization settings into a code segment on creation.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3129 b7a2c559-68d2-44c3-8de9-860c34a00d81
Moved LoadExpr + support functions into a separate module.
Removed obsolete files.
Some cleanup and makefile adjustments.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3113 b7a2c559-68d2-44c3-8de9-860c34a00d81
Improved the inlining of standard C functions. Added more standard functions
to inline.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3095 b7a2c559-68d2-44c3-8de9-860c34a00d81
ExprDesc structure, not as separate value.
WARNING: The current code does compile but does not work correctly, because
the lvalue flag is part of ExprDesc.Flags and not masked out in several tests
throughout the code.
git-svn-id: svn://svn.cc65.org/cc65/trunk@3046 b7a2c559-68d2-44c3-8de9-860c34a00d81
argument to the sizeof operator, if the type spec is starting with this
keyword.
git-svn-id: svn://svn.cc65.org/cc65/trunk@2921 b7a2c559-68d2-44c3-8de9-860c34a00d81
The Test() and TestInParens() functions do now return information about the
expression that was tested.
An if... statement where the expression is always true will now print a
warning "Unreachable code" if it has an else clause.
git-svn-id: svn://svn.cc65.org/cc65/trunk@2889 b7a2c559-68d2-44c3-8de9-860c34a00d81
correctly if it did not contain an explicit comparison operator.
Removed an old hack from the test subroutine that did no longer work.
git-svn-id: svn://svn.cc65.org/cc65/trunk@1619 b7a2c559-68d2-44c3-8de9-860c34a00d81
When assigning structures, copy them using the primary register when the
size is 1, 2, or 4.
When assigning structures, allow structures by value when the size is 1,
2, or 4. These structures are expected to be in the primary register. The
only case when this can happen is by return from function, so this change
makes div() work.
git-svn-id: svn://svn.cc65.org/cc65/trunk@1477 b7a2c559-68d2-44c3-8de9-860c34a00d81
Allow the second and third operand of the :? operator to be of type int.
git-svn-id: svn://svn.cc65.org/cc65/trunk@1395 b7a2c559-68d2-44c3-8de9-860c34a00d81
Fixed a problem with K&R functions: In a function with no return type
specified, the compiler did not allow a simple "return" statement. However,
there was no "void" type at that time, so it was not possible to specify
something else. The solution is to allow omission of a return value in a
K&R function with an implicit int type. Other types or an explicit int is
still checked.
git-svn-id: svn://svn.cc65.org/cc65/trunk@1302 b7a2c559-68d2-44c3-8de9-860c34a00d81
cdecl functions, this generates better code.
Remove a small error that caused indirect non fastcall function to always be
handled as variadic, which meant a useless load of the Y register was added.
git-svn-id: svn://svn.cc65.org/cc65/trunk@1292 b7a2c559-68d2-44c3-8de9-860c34a00d81
internal compiler errors in several places.
An old cc65 extension (variables of type void) was broken and did no longer
work. Fixed that.
git-svn-id: svn://svn.cc65.org/cc65/trunk@1184 b7a2c559-68d2-44c3-8de9-860c34a00d81
Remove the -Wno-unused-parameter in favour of __atribute__((unused)).
git-svn-id: svn://svn.cc65.org/cc65/trunk@986 b7a2c559-68d2-44c3-8de9-860c34a00d81
unused stores into these registers.
Because of this, the old code using regsave does no longer work. Started
to rewrite it.
git-svn-id: svn://svn.cc65.org/cc65/trunk@966 b7a2c559-68d2-44c3-8de9-860c34a00d81
preprocessor mode to work around limitations in the existing ones (constant
expressions are not detected).
git-svn-id: svn://svn.cc65.org/cc65/trunk@864 b7a2c559-68d2-44c3-8de9-860c34a00d81
supplied data structures. Re-added the -T option which is much more exact
now because of the better line info stuff.
Cleanups in the scanner (remove old #defines).
git-svn-id: svn://svn.cc65.org/cc65/trunk@740 b7a2c559-68d2-44c3-8de9-860c34a00d81
* __fixargs__ is now an actual function symbol and no longer handled in the
scanner.
* Additional symbol __argsize__ that is a constant in normal functions and
a const local variable for variadic functions. Using this symbol, the
va_arg macro gets a lot simpler and smaller.
* Added special code to handle the fixed parameters of a variadic function.
The code has some overhead, but the va_fix macro is no longer needed (and
the compiler generated code is better than va_fix anyway).
git-svn-id: svn://svn.cc65.org/cc65/trunk@652 b7a2c559-68d2-44c3-8de9-860c34a00d81
more identical parameter names. The input is of course wrong, but the
compiler shouldn't crash.
git-svn-id: svn://svn.cc65.org/cc65/trunk@644 b7a2c559-68d2-44c3-8de9-860c34a00d81
pointer-to-function type) was not possible. It is redundant but allowed.
A cast with a function on the right side resulted in an internal compiler
error, since the implicit cast from function to pointer-to-function was
missing in this case.
An assignment of a function to a void pointer was not possible because the
same implicit conversion was missing.
git-svn-id: svn://svn.cc65.org/cc65/trunk@482 b7a2c559-68d2-44c3-8de9-860c34a00d81
were harder to manage, made the code harder to read, and lead no less
verbose messages, since it was effort to add a new message. The new approach
has the drawback that it's more difficult to translate, but this won't
happen anyway.
git-svn-id: svn://svn.cc65.org/cc65/trunk@428 b7a2c559-68d2-44c3-8de9-860c34a00d81
Check for and honour const in several places: Assignment to const is flagged
with an error. Const data is placed in the rodata segment.
git-svn-id: svn://svn.cc65.org/cc65/trunk@252 b7a2c559-68d2-44c3-8de9-860c34a00d81
Rewrote the include file management, keeping a list of all include files
and cleaning up the code while going.
git-svn-id: svn://svn.cc65.org/cc65/trunk@82 b7a2c559-68d2-44c3-8de9-860c34a00d81
Several renames for better readibility.
Removed separate struct and enum symbol tables in favour of one tag table.
Check for some more error conditions or dubious constructs.
git-svn-id: svn://svn.cc65.org/cc65/trunk@62 b7a2c559-68d2-44c3-8de9-860c34a00d81