llvm-svn: 76280

Revert 76177 for now. It's messing up ARM asm printing. Also this significant debate about its efficiency.

llvm-svn: 76279

llvm-svn: 76278

Fix some fallout from CFGBuilder restructuring: all expressions that we explicitly handle have the possibility to be block-level expressions.

llvm-svn: 76277

Also a test commit.

llvm-svn: 76276

Their usefulness is questionable since redecl_iterator was introduced.

llvm-svn: 76275

llvm-svn: 76274

Move the functionality of ASTContext::getCanonicalDecl(), into a virtual method Decl::getCanonicalDecl().

llvm-svn: 76273

llvm-svn: 76272

llvm-svn: 76271

llvm-svn: 76270

llvm-svn: 76269

llvm-svn: 76268

llvm-svn: 76267

populate CGM's list directly.

llvm-svn: 76266

ConstantExpr and Instruction. This involves duplicating some code
between GetElementPtrInst and GEPOperator, but it's not a lot.

llvm-svn: 76265

llvm-svn: 76264

llvm-svn: 76263

llvm-svn: 76262

llvm-svn: 76261

llvm-svn: 76260

llvm-svn: 76259

llvm-svn: 76258

llvm-svn: 76257

llvm-svn: 76256

llvm-svn: 76253

GEPOperator's hasNoPointer0verflow(), and make a few places in instcombine
that create GEPs that may overflow clear the NoOverflow value. Among
other things, this partially addresses PR2831.

llvm-svn: 76252

is an algorithm that is much easier to understand and slightly more efficient.
Thanks to Mike Stump for our discussions on the CFGBuilder and his comments that
helped prompt this long needed cleanup.

llvm-svn: 76250

llvm-svn: 76249

Fix x86 inline ams 'q' constraint support. In 32-bit mode, it's just like 'Q', i.e. EAX, EDX, ECX, EBX. In 64-bit mode, it just means all the i64r registers. Yeah, that makes sense.

llvm-svn: 76248

llvm-svn: 76247

llvm-svn: 76246

accessors into it.

llvm-svn: 76245

llvm-svn: 76244

llvm-svn: 76243

llvm-svn: 76242

only do this if you are a crazy russian hacker. ;-)

llvm-svn: 76241

llvm-svn: 76240

llvm-svn: 76239

llvm-svn: 76238