so that if a_loadaddr_ref_reg() is used to load the address of a procdef
into a register, the getpointerdef(procdef) won't (wrongly) result in two
*'s
o the previous behaviour was necessary for the tck_simple_procvar2proc
special case that was removed in the previous commit
git-svn-id: branches/hlcgllvm@28477 -
procdef.getcopyas(procvardef,pc_address_only) as the def if we want to
specify that we're interested in the address of a procedure/procvar of
object
git-svn-id: branches/hlcgllvm@28476 -
o results in simpler code, and more importantly is required for some code
that expects a LOC_REGISTER after calling certain system unit helpers
(e.g. in ncgutil.new_exception)
git-svn-id: branches/hlcgllvm@28375 -
o there are no cpu registers on the llvm target, but generic code may still
occasionally allocate them via the generic routines in case of e.g. a
hardcoded use of a function return register
git-svn-id: branches/hlcgllvm@28373 -
have to emit the label we originally allocated for the fallthrough path (otherwise
the branch is not inverted at all, and moreover the caller will also emit the
original target label somewhere so it will be doubly defined)
git-svn-id: branches/hlcgllvm@28367 -
get the offset relative from the returned label from
emit_(ansi|unicode)_string() to the start of the string data
o corrected return value for the offset on LLVM: it's always equal to the
header size, since we can't emit labels in the middle of structured data
there
git-svn-id: branches/hlcgllvm@28324 -
advance (e.g. ansistring constants, as they consist of a header
followed by an array of char equal to the string length, or RTTI data)
o use begin_anonymous_record() to start recording the field types (if
necessary for the current target), and end_anonymous_record() at the
end to generate the recorddef
git-svn-id: branches/hlcgllvm@28323 -
o the generic implementation ignores the need for bitcasts from pointers to
ordinals and vice versa, and also hacks around with offsets
git-svn-id: branches/hlcgllvm@28206 -
o doesn't reuse the generic code generator methods, as the div-to-mul/shift
optimization can be performed by llvm if necessary
git-svn-id: branches/hlcgllvm@28205 -
llvm
* instead, add LLVM-specific name mangling based on the asmsymbol's bind
and typ whenever we write out its name
git-svn-id: branches/hlcgllvm@28166 -
o remove the accompanying tdef operands from tai operands, because such
tai operands should have a tdef already
o check that every queued tai is a tai_simpletypedconst (which have a
type)
o wrap queued bitcast/getelementptr tai instances with a
tai_simpletypedconst
git-svn-id: branches/hlcgllvm@28154 -
ttai_typedconstbuilder so that it can be overridden by other platforms
* override get_string_symofs() for LLVM and always return 0, as you can't
have labels in the middle of a declaration on that platform
git-svn-id: branches/hlcgllvm@28140 -
ttai_lowleveltypedconstbuilder
o set the offset to 0 in the llvm constbuilder before calling
the lowlevel one, because it incorporates this offset already
via getelementptr instructions
git-svn-id: branches/hlcgllvm@28136 -
