llvm-svn: 47545

not safe.  This is fixed by more aggressively checking that the return slot is
not used elsewhere in the function.

llvm-svn: 47544

llvm-svn: 47543

llvm-svn: 47542

Fix an issue where GVN would try to use an instruction before its definition when performing return slot optimization.

llvm-svn: 47541

llvm-svn: 47540

llvm-svn: 47539

Move the initialization of SEL/objc_selector from Sema::Sema() to Sema::ActOnTranslationUnitScope() and make sure the type/struct get inserted into the translation unit scope.

Bug submitted by David Chisnall (thanks!).

llvm-svn: 47538

out of illegal elements (BUILD_VECTOR).  Uses and beefs
up BUILD_PAIR, though it didn't really have to.  Like
most of LegalizeTypes, does not support soft-float.
This cures all "make check" vector building failures.

llvm-svn: 47537

- More enum signeness bitfield fixes (MSVC treats enums as signed).

- Fixed in Lex/HeaderSearch.cpp an unsafe copy between two
  HeaderSearch::PerFileInfo entries in a common vector. The copy involved two
  calls to getFileInfo() within the assignment; these calls could have
  side-effects that enlarged the internal vector, and with MSVC this would
  invalidate one of the values in the assignment.
  
Patch by Argiris Kirtzidis!

llvm-svn: 47536

since inodes are bogus abstraction on that platform.

Patch by Argiris Kirtzidis!

llvm-svn: 47535

%r3 on PPC) in their ASM files. However, it's hard for humans to read
during debugging. Adding a new field to the register data that lets you
specify a different name to be printed than the one that goes into the
ASM file -- %x3 instead of %r3, for instance.

llvm-svn: 47534

llvm-svn: 47533

for CellSPU modifications:

- SPUInstrInfo.td refactoring: "multiclass" really is _your_ friend.
- Other improvements based on refactoring effort in SPUISelLowering.cpp,
  esp. in SPUISelLowering::PerformDAGCombine(), where zero amount shifts and
  rotates are now eliminiated, other scalar-to-vector-to-scalar silliness
  is also eliminated.
- 64-bit operations are being implemented, _muldi3.c gcc runtime now
  compiles and generates the right code. More work still needs to be done.

llvm-svn: 47532

llvm-svn: 47531

llvm-svn: 47530

llvm-svn: 47529

llvm-svn: 47527

If remating a machine instr with virtual register operand, make sure the vr is avaliable at all uses regardless of whether it would be folded.

llvm-svn: 47526

llvm-svn: 47525

llvm-svn: 47524

invalid command-line options.

llvm-svn: 47523

No need recognize load from a fixed argument slot as re-materializable. LiveIntervalAnalysis already handles it as a special case.

llvm-svn: 47522

llvm-svn: 47521

Recognize loads of arguments as re-materializable first. Therefore if isReallyTriviallyReMaterializable() returns true it doesn't confuse it as a "normal" re-materializable instruction.

llvm-svn: 47520

llvm-svn: 47519

llvm-svn: 47518

llvm-svn: 47517

llvm-svn: 47516

instead of TokenKind because of signedness issues with MSVC and enums.

Patch from Argiris Kirtzidis.

llvm-svn: 47515

llvm-svn: 47514

llvm-svn: 47513

The patch fixes some debug assertions that the msvcrt throws:

1)
-    if (isprint(value) && value < 256) {
+    if (value < 256 && isprint(value)) {

isprint() does an assertion check of its own for value < 256; check value before calling it to prevent it.


2)
-    Stmt->Names.push_back(std::string(&data[0], data.size()));
+    if (data.size() == 0)
+      Stmt->Names.push_back(std::string());
+    else
+      Stmt->Names.push_back(std::string(&data[0], data.size()));

If data.size() == 0 then data[0] throws "out of range" assertion.

llvm-svn: 47512

llvm-svn: 47511

llvm-svn: 47510

llvm-svn: 47509

To support multiple return values, now ret instruction supports multiple operands instead of one aggregate operand.

llvm-svn: 47508

llvm-svn: 47507

confluence points the liveness information for variables (Decls) and NOT
block-level expressions.

llvm-svn: 47506

stuff into ParamAttrsList.h.  Per feedback from
ParamAttrs changes.

llvm-svn: 47504