llvm-svn: 115649

following amusing sequence:
  - AST writing schedules writing a type X* that it had never seen
  before
  - AST writing starts writing another declaration, ends up
  deserializing X* from a prior AST file. Now we have two type IDs for
  the same type!
  - AST writer tries to write X*. It only has the lower-numbered ID
  from the the prior AST file, so references to the higher-numbered ID
  that was scheduled for writing go off into lalaland.

To fix this, keep the higher-numbered ID so we end up writing the type
twice. Since this issue occurs so rarely, and type records are
generally rather small, I deemed this better than the alternative: to
keep a separate mapping from the higher-numbered IDs to the
lower-numbered IDs, which we would end up having to check whenever we
want to deserialize any type.

Fixes <rdar://problem/8511624>, I think.

llvm-svn: 115647

addressing mode from four to five.

llvm-svn: 115645

block descriptor for outer block was missing even though
the block was importing objects into its inner blocks.
//rdar://84995992

llvm-svn: 115644

isInSymtab and isLocal logic in the two loops don't get easily out of sync.

llvm-svn: 115643

llvm-svn: 115642

llvm-svn: 115641

llvm-svn: 115640

llvm-svn: 115639

Fix handling of the 'Invalid' argument in SourceManager's methods (patch by Dean Sturtevant, reviewed by chandlerc and Sebastian Redl).

llvm-svn: 115638

llvm-svn: 115636

Use a more efficient lowering of uint64_t --> float that can take advantage of hardware signed integer conversion without
having to do a double cast (uint64_t --> double --> float).  This is based on the algorithm from compiler_rt's __floatundisf
for X86-64.

llvm-svn: 115634

llvm-svn: 115633

llvm-svn: 115632

llvm-svn: 115631

llvm-svn: 115629

llvm-svn: 115627

Give every file that ASTReader loads a type: module, PCH, precompiled preamble or main file. Base Decls' PCHLevel on this to make it more sane.

llvm-svn: 115626

llvm-svn: 115625

llvm-svn: 115624

llvm-svn: 115623

llvm-svn: 115622

waiting until we think we need it: we didn't catch all of the places
where we actually needed it, and we probably wouldn't ever. Fixes a
C++ PCH crasher.

llvm-svn: 115621

Graphviz program, print something with a newline, to avoid leaving
the line unfinished.

llvm-svn: 115620

they should be in the symbol table or not. Instead of "guessing", just compute
the symbol table after the relocations are known.

llvm-svn: 115619

llvm-svn: 115618

waiting until we think we need it: we didn't catch all of the places
where we actually needed it, and we probably wouldn't ever. Fixes a
C++ PCH crasher.

llvm-svn: 115617

llvm-svn: 115616

llvm-svn: 115614

llvm-svn: 115612

test/CodeGen/X86/atomic_op.ll: Rename @main to @func. Extra sequences will be inserted to @main as prologue on cygming, to fail.

llvm-svn: 115611

* Fix dump() to make it consistent with the test case.

llvm-svn: 115609

llvm-svn: 115608

llvm-svn: 115607

llvm-svn: 115606

llvm-svn: 115605

bswap32 doesn't read eflags.

llvm-svn: 115604

pseudo instructions.

Move POPCNT to InstrSSE since they are SSE4 instructions.

llvm-svn: 115603

with the 32-bit stuff.

llvm-svn: 115602

llvm-svn: 115601