compiler's internal assembly representation. It is not yet generated by the
compiler and is only implemented properly in the NASM writer (but will
eventually be implemented in the other assembler writers as well, including
the internal assembler). In the future, it is intended to allow:
1) catching instructions, generated by the compiler and not supported by the
target CPU, due to a bug in the code generator
2) allowing a different instruction set to be used in inline asm blocks (I'm
planning a new directive for that - {$asmcpu XXX}), because the code
generator will be able to switch target CPU in the middle of the asm output
stream
git-svn-id: trunk@33121 -
* generate code to load the TOC register if it's required, and
emit the the ".localentry" directive to indicate to the linker
where the actual function body starts
* support the ".localentry" directive in the ppc64 assembler reader
o disable the fpc_qword_to_double() assembler implementation for
ELFv2, because we can't manually insert the .localentry directive
before the stack allocation code
* perform indirect calls on ppc64 via R12 in ncgcal, as R12 needs to
contain the function address on entry on ppc64/ELFv2 (and it's
a volatile register, so there's no problem with always using
it)
git-svn-id: trunk@30198 -
backend (required for inserting bitcasts/ptrtoint/... operations inside
typed constant declarations)
o taillvm constructors for creating bitcast-style and getlementptr
operations that accept a tai as their source/base operand
git-svn-id: branches/hlcgllvm@28112 -
to free up space for more ait_* types in taitype (can't have
more than 32 because they have to fit in a small set)
o factored out writing of floating point numbers as an array of
byte in the external assemblers
git-svn-id: branches/hlcgllvm@28105 -
o make use of the fact that callparanodes always first load everything
in temporary parameters (allocated via paramanager.createtempparaloc)
-> we pass those temporary paralocs to the llvm call nodes and ignore
the "real" locations with physical registers
o all function results are forced to memory before handing them back
to the generic code generator to avoid having to deal with special
result locations and llvm<->pascal type differences (llvm removes
the extra temp stores/loads while optimising)
o ideally, the llvm code generator should share the common code with
the generic code generator for a_load_ref_cgpara() at some point in
the future
git-svn-id: branches/hlcgllvm@27013 -
generate declarations when external symbols are referred in a taillvm
instruction
* added "declared" field to tasmsymbol for llvm to make avoid declaring a
symbol multiple times (harmless, but unnecessary)
* all kinds of declarations/definitions are now handled using the new
taillvmdecl, the old separate types have been removed
git-svn-id: branches/hlcgllvm@27010 -
* Change MIPS jump tables in PIC mode to use gp-relative constants, making them ABI-compliant and not requiring dynamic relocations.
git-svn-id: trunk@26886 -
