llvm-svn: 89404

assembly can confuse things utterly, as it's assumed that instructions in
inline assembly are 4 bytes wide. For Thumb mode, that's often not true,
so the calculations for when alignment padding will be present get thrown off,
ultimately leading to out of range constant pool entry references. Making
more conservative assumptions that padding may be necessary when inline asm
is present avoids this situation.

llvm-svn: 89403

appropriate lookup and simply can't resolve the referrent yet, and
"dependent scope" expressions, where we can't do the lookup yet because the
entity we need to look into is a dependent type.

llvm-svn: 89402

(radar 7409165).

llvm-svn: 89400

llvm-svn: 89399

if it is not ultimately captured. Teach BasicAliasAnalysis that a 
local object address which does not escape and is never stored does
not alias with a value resulting from a load.

llvm-svn: 89398

llvm-svn: 89396

llvm-svn: 89389

errors.

llvm-svn: 89388

llvm-svn: 89387

llvm-svn: 89386

fixes PR5395.

llvm-svn: 89385

Fix crash when using --analyzer-store=region when handling initializers with nested arrays/structs whose values are not explicitly specified.  Fixes <rdar://problem/7403269>.

llvm-svn: 89384

llvm-svn: 89383

llvm-svn: 89382

critical edges in PHIElimination.

This has a huge impact on regalloc performance, and we recover almost all of
the 10% compile time regression that edge splitting introduced.

llvm-svn: 89381

$ svn merge -c -89279 https://llvm.org/svn/llvm-project/llvm/trunk
--- Reverse-merging r89279 into '.':
U    lib/CodeGen/AsmPrinter/DwarfException.cpp
U    lib/Target/TargetLoweringObjectFile.cpp
$ svn merge -c -89270 https://llvm.org/svn/llvm-project/llvm/trunk
--- Reverse-merging r89270 into '.':
G    lib/CodeGen/AsmPrinter/DwarfException.cpp
G    lib/Target/TargetLoweringObjectFile.cpp

llvm-svn: 89379

fully specified at this level.  Subclasses of NLdStLN can specify selective
bit(s) for Inst{7-4}, as is done for VLD[234]LN* and VST[234]LN* inside
ARMInstrNEON.td.

llvm-svn: 89377

Fix a small bug.

Fix one case we missed to make sure we reserve registers from
allocation.

llvm-svn: 89376

llvm-svn: 89375

they are lowered to instruction sequences more complex than a simple
load, such that CodeGen cannot rematerialize them, a reload from a
spill slot is likely to be cheaper than the complex sequence.

llvm-svn: 89374

llvm-svn: 89373

llvm-svn: 89372

llvm-svn: 89371

Fixes pr5547.

llvm-svn: 89370

llvm-svn: 89369

everything else.

llvm-svn: 89368

llvm-svn: 89367

incomplete array initialization, where we have the following in a
template:

  int a[] = { 1, 2, something-value-dependent };
  // ...
  sizeof(a);

The type of "a" appears to be a non-dependent IncompleteArrayType, but
treating it as such makes the sizeof(a) fail at template definition
time. We now correctly handle this by morphing the IncompleteArrayType
into a DependentSizedArrayType with a NULL expression, indicating that
its size has no corresponding expression (and, therefore, the type is
distinct from others).

llvm-svn: 89366

llvm-svn: 89364

llvm-svn: 89363

llvm-svn: 89362

llvm-svn: 89361

llvm-svn: 89360

llvm-svn: 89359

llvm-svn: 89357

Add support for spreading register allocation.

Add a -linearscan-skip-count argument (default to 0) that tells the
allocator to remember the last N registers it allocated and skip them
when looking for a register candidate.  This tends to spread out
register usage and free up post-allocation scheduling at the cost of
slightly more register pressure.  The primary benefit is the ability
to backschedule reloads.

This is turned off by default.

llvm-svn: 89356

other limit macros.

llvm-svn: 89355

llvm-svn: 89354

llvm-svn: 89353