ARM64: separate load/store operands to simplify assembler

This changes ARM64 to use separate operands for each component of an
address, and look for separate '[', '$Rn, ..., ']' tokens when
parsing.

This allows us to do away with quite a bit of special C++ code to
handle monolithic "addressing modes" in the MC components. The more
incremental matching of the assembler operands also allows for better
diagnostics when LLVM is presented with invalid input.

Most of the complexity here is with the register-offset instructions,
which were extremely dodgy beforehand: even when the instruction used
wM, LLVM's model had xM as an operand. We papered over this
discrepancy before, but that approach doesn't work now so I split them
into separate X and W variants.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209425 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Tim Northover
2014-05-22 11:56:09 +00:00
parent b08e03806f
commit e072ed71c8
15 changed files with 2529 additions and 2694 deletions

View File

@ -1718,9 +1718,9 @@ bool TreePatternNode::ApplyTypeConstraints(TreePattern &TP, bool NotRegisters) {
DagInit *MIOpInfo = OperandNode->getValueAsDag("MIOperandInfo");
if (unsigned NumArgs = MIOpInfo->getNumArgs()) {
// But don't do that if the whole operand is being provided by
// a single ComplexPattern.
const ComplexPattern *AM = Child->getComplexPatternInfo(CDP);
if (!AM || AM->getNumOperands() < NumArgs) {
// a single ComplexPattern-related Operand.
if (Child->getNumMIResults(CDP) < NumArgs) {
// Match first sub-operand against the child we already have.
Record *SubRec = cast<DefInit>(MIOpInfo->getArg(0))->getDef();
MadeChange |=