keeping them around until the pass is destroyed, keep them
around a) just when useful (not for outer loops) and b) destroy
them right after we use them.  This should reduce memory use
and fixes potential bugs where a loop is deleted and another
loop gets allocated to the same address.

llvm-svn: 112446

claims that it preserves domfrontier if it doesn't really.

llvm-svn: 112445

for the unroller to pretend it supports updating it.  It still
has a horrible hack for DomTree.

llvm-svn: 112444

other filtering techniques, as those may allow it to filter
out more obviously unprofitable candidates.

llvm-svn: 112441

LSRInstance data structures up to date. This fixes some
pessimizations caused by stale data which will be exposed
in an upcoming change.

llvm-svn: 112440

NarrowSearchSpaceUsingHeuristics into separate functions.

llvm-svn: 112439

llvm-svn: 112438

llvm-svn: 112437

the high-level logic.

llvm-svn: 112436

llvm-svn: 112435

llvm-svn: 112434

preserves domfrontier.  It does preserve AA though.

llvm-svn: 112419

require DomFrontier.  Dropping this doesn't actually save any runs
of the pass though.

llvm-svn: 112418

Among other things, this uses SSAUpdater instead of 
PromoteMemToReg.

llvm-svn: 112417

llvm-svn: 112412

This leads to much simpler code.

llvm-svn: 112410

llvm-svn: 112409

llvm-svn: 112408

getUniqueExitBlocks instead of getExitBlocks and a manual
set to eliminate dupes.

llvm-svn: 112405

llvm-svn: 112404

being actively maintained, improved, or extended.

llvm-svn: 112356

I'm aware of, aren't maintained, and LVI will be replacing their value.
nlewycky approved this on irc.

llvm-svn: 112355

llvm-svn: 112351

llvm-svn: 112350

like this:

struct S { float A, B, C, D; };

struct S g;
struct S bar() { 
  struct S A = g;
  ++A.B;
  A.A = 42;
  return A;
}

we now generate:

_bar:                                   ## @bar
## BB#0:                                ## %entry
	movq	_g@GOTPCREL(%rip), %rax
	movss	12(%rax), %xmm0
	pshufd	$16, %xmm0, %xmm0
	movss	4(%rax), %xmm2
	movss	8(%rax), %xmm1
	pshufd	$16, %xmm1, %xmm1
	unpcklps	%xmm0, %xmm1
	addss	LCPI1_0(%rip), %xmm2
	pshufd	$16, %xmm2, %xmm2
	movss	LCPI1_1(%rip), %xmm0
	pshufd	$16, %xmm0, %xmm0
	unpcklps	%xmm2, %xmm0
	ret

instead of:

_bar:                                   ## @bar
## BB#0:                                ## %entry
	movq	_g@GOTPCREL(%rip), %rax
	movss	12(%rax), %xmm0
	pshufd	$16, %xmm0, %xmm0
	movss	4(%rax), %xmm2
	movss	8(%rax), %xmm1
	pshufd	$16, %xmm1, %xmm1
	unpcklps	%xmm0, %xmm1
	addss	LCPI1_0(%rip), %xmm2
	movd	%xmm2, %eax
	shlq	$32, %rax
	addq	$1109917696, %rax       ## imm = 0x42280000
	movd	%rax, %xmm0
	ret

llvm-svn: 112345

element insertion from the pieces that feed into the vector.
This handles a pattern that occurs frequently due to code
generated for the x86-64 abi.  We now compile something like
this:

struct S { float A, B, C, D; };
struct S g;
struct S bar() { 
  struct S A = g;
  ++A.A;
  ++A.C;
  return A;
}

into all nice vector operations:

_bar:                                   ## @bar
## BB#0:                                ## %entry
	movq	_g@GOTPCREL(%rip), %rax
	movss	LCPI1_0(%rip), %xmm1
	movss	(%rax), %xmm0
	addss	%xmm1, %xmm0
	pshufd	$16, %xmm0, %xmm0
	movss	4(%rax), %xmm2
	movss	12(%rax), %xmm3
	pshufd	$16, %xmm2, %xmm2
	unpcklps	%xmm2, %xmm0
	addss	8(%rax), %xmm1
	pshufd	$16, %xmm1, %xmm1
	pshufd	$16, %xmm3, %xmm2
	unpcklps	%xmm2, %xmm1
	ret

instead of icky integer operations:

_bar:                                   ## @bar
	movq	_g@GOTPCREL(%rip), %rax
	movss	LCPI1_0(%rip), %xmm1
	movss	(%rax), %xmm0
	addss	%xmm1, %xmm0
	movd	%xmm0, %ecx
	movl	4(%rax), %edx
	movl	12(%rax), %esi
	shlq	$32, %rdx
	addq	%rcx, %rdx
	movd	%rdx, %xmm0
	addss	8(%rax), %xmm1
	movd	%xmm1, %eax
	shlq	$32, %rsi
	addq	%rax, %rsi
	movd	%rsi, %xmm1
	ret

This resolves rdar://8360454

llvm-svn: 112343

llvm-svn: 112332

Add a prototype of a new peephole optimizing pass that uses LazyValue info to simplify PHIs and select's.
This pass addresses the missed optimizations from PR2581 and PR4420.

llvm-svn: 112325

A = shl x, 42
...
B = lshr ..., 38

which can be transformed into:
A = shl x, 4
...

iff we can prove that the would-be-shifted-in bits
are already zero.  This eliminates two shifts in the testcase
and allows eliminate of the whole i128 chain in the real example.

llvm-svn: 112314

framework, which is good at ripping through bitfield
operations.  This generalize a bunch of the existing
xforms that instcombine does, such as 
  (x << c) >> c -> and
to handle intermediate logical nodes.  This is useful for
ripping up the "promote to large integer" code produced by
SRoA.

llvm-svn: 112304

more general simplify demanded bits logic.

llvm-svn: 112291

llvm-svn: 112286

computation can be truncated if it is fed by a sext/zext that doesn't
have to be exactly equal to the truncation result type.

llvm-svn: 112285

by the SRoA "promote to large integer" code, eliminating
some type conversions like this:

   %94 = zext i16 %93 to i32                       ; <i32> [#uses=2]
   %96 = lshr i32 %94, 8                           ; <i32> [#uses=1]
   %101 = trunc i32 %96 to i8                      ; <i8> [#uses=1]

This also unblocks other xforms from happening, now clang is able to compile:

struct S { float A, B, C, D; };
float foo(struct S A) { return A.A + A.B+A.C+A.D; }

into:

_foo:                                   ## @foo
## BB#0:                                ## %entry
	pshufd	$1, %xmm0, %xmm2
	addss	%xmm0, %xmm2
	movdqa	%xmm1, %xmm3
	addss	%xmm2, %xmm3
	pshufd	$1, %xmm1, %xmm0
	addss	%xmm3, %xmm0
	ret

on x86-64, instead of:

_foo:                                   ## @foo
## BB#0:                                ## %entry
	movd	%xmm0, %rax
	shrq	$32, %rax
	movd	%eax, %xmm2
	addss	%xmm0, %xmm2
	movapd	%xmm1, %xmm3
	addss	%xmm2, %xmm3
	movd	%xmm1, %rax
	shrq	$32, %rax
	movd	%eax, %xmm0
	addss	%xmm3, %xmm0
	ret

This seems pretty close to optimal to me, at least without
using horizontal adds.  This also triggers in lots of other
code, including SPEC.

llvm-svn: 112278

Use LVI to eliminate conditional branches where we've tested a related condition previously.  Update tests for this change.
This fixes PR5652.

llvm-svn: 112270

by SRoA.  This is part of rdar://7892780, but needs another xform to
expose this.

llvm-svn: 112232

is a vector to be a vector element extraction.  This allows clang to
compile:

struct S { float A, B, C, D; };
float foo(struct S A) { return A.A + A.B+A.C+A.D; }

into:

_foo:                                   ## @foo
## BB#0:                                ## %entry
	movd	%xmm0, %rax
	shrq	$32, %rax
	movd	%eax, %xmm2
	addss	%xmm0, %xmm2
	movapd	%xmm1, %xmm3
	addss	%xmm2, %xmm3
	movd	%xmm1, %rax
	shrq	$32, %rax
	movd	%eax, %xmm0
	addss	%xmm3, %xmm0
	ret

instead of:

_foo:                                   ## @foo
## BB#0:                                ## %entry
	movd	%xmm0, %rax
	movd	%eax, %xmm0
	shrq	$32, %rax
	movd	%eax, %xmm2
	addss	%xmm0, %xmm2
	movd	%xmm1, %rax
	movd	%eax, %xmm1
	addss	%xmm2, %xmm1
	shrq	$32, %rax
	movd	%eax, %xmm0
	addss	%xmm1, %xmm0
	ret

... eliminating half of the horribleness.

llvm-svn: 112227

llvm-svn: 112198

fix: add a flag to MapValue and friends which indicates whether
any module-level mappings are being made. In the common case of
inlining, no module-level mappings are needed, so MapValue doesn't
need to examine non-function-local metadata, which can be very
expensive in the case of a large module with really deep metadata
(e.g. a large C++ program compiled with -g).

This flag is a little awkward; perhaps eventually it can be moved
into the ClonedCodeInfo class.

llvm-svn: 112190

except ...", it is causing *massive* performance regressions when building Clang
with itself (-O3 -g).

llvm-svn: 112158