llvm-svn: 45415

llvm-svn: 45408

eliminating the llvm.x86.sse2.loadl.pd intrinsic?), one shuffle optzn
may be done (if shufps is better than pinsw, Evan, please review), and
we already know about LICM of simple instructions.

llvm-svn: 45407

comparisons with a constant.  This allows us to compile isnan to:

_foo:
	fcmpu cr7, f1, f1
	mfcr r2
	rlwinm r3, r2, 0, 31, 31
	blr 

instead of:

LCPI1_0:					;  float
	.space	4
_foo:
	lis r2, ha16(LCPI1_0)
	lfs f0, lo16(LCPI1_0)(r2)
	fcmpu cr7, f1, f0
	mfcr r2
	rlwinm r3, r2, 0, 31, 31
	blr 

llvm-svn: 45405

llvm-svn: 45404

Disable null pointer folding transforms for non-generic address spaces. This should probably be a target-specific predicate based on address space. That way for targets where this isn't applicable the predicate can be optimized away.

llvm-svn: 45403

llvm-svn: 45402

as:

_bar:
	pushl	%esi
	subl	$8, %esp
	movl	16(%esp), %esi
	call	L_foo$stub
	fstps	(%esi)
	addl	$8, %esp
	popl	%esi
	#FP_REG_KILL
	ret

instead of:

_bar:
	pushl	%esi
	subl	$8, %esp
	movl	16(%esp), %esi
	call	L_foo$stub
	fstpl	(%esi)
	cvtsd2ss	(%esi), %xmm0
	movss	%xmm0, (%esi)
	addl	$8, %esp
	popl	%esi
	#FP_REG_KILL
	ret

llvm-svn: 45401

llvm-svn: 45400

if we are just going to store it back anyway.  This improves things 
like:
double foo();
void bar(double *P) { *P = foo(); }

llvm-svn: 45399

  x = load p
  store x -> p

llvm-svn: 45398

llvm-svn: 45397

llvm-svn: 45393

have potential side-effects.

llvm-svn: 45392

llvm-svn: 45388

llvm-svn: 45387

Repair a transform that Chris noticed a bug in.  Thanks to Nicholas for pointing out my stupid mistakes when writing this patch. :-)

llvm-svn: 45384

define i32 @main() {
entry:
	%z = alloca i32		; <i32*> [#uses=2]
	store i32 0, i32* %z
	%tmp = load i32* %z		; <i32> [#uses=1]
	%sub = sub i32 %tmp, 1		; <i32> [#uses=1]
	%cmp = icmp ult i32 %sub, 0		; <i1> [#uses=1]
	%retval = select i1 %cmp, i32 1, i32 0		; <i32> [#uses=1]
	ret i32 %retval
}

into ret 1, instead of ret 0.

Christopher, please investigate.

llvm-svn: 45383

llvm-svn: 45377

behaviour of LinkGlobals() function.

llvm-svn: 45375

llvm-svn: 45369

llvm-svn: 45367

when copying functions.

llvm-svn: 45356

it is only a partial fix.  This change is noise for most programs, but
speeds up Shootout-C++/matrix by 20%, Ptrdist/ks by 24%, smg2000 by 8%,
hexxagon by 9%, bzip2 by 9% (not sure I trust this), ackerman by 13%, etc.

OTOH, it slows down Shootout/fib2 by 40% (I'll update PR1877 with this info).

llvm-svn: 45354

    define void @f() {
            ...
            call i32 @g()
            ...
    }

    define void @g() {
            ...
    }

The hazards are:

  - @f and @g have GC, but they differ GC. Inlining is invalid. This
    may never occur.
  - @f has no GC, but @g does. g's GC must be propagated to @f.

The other scenarios are safe:

  - @f and @g have the same GC.
  - @f and @g have no GC.
  - @g has no GC.

This patch adds inliner checks for the former two scenarios.

llvm-svn: 45351

function with GC.

This will catch the error when the inliner inlines a function with
GC into a caller with no GC.

llvm-svn: 45350

LangRef.

llvm-svn: 45349

llvm-svn: 45347

When specified, don't split backedges of single-bb loops.
This helps address PR1877

llvm-svn: 45344

llvm-svn: 45343

llvm-svn: 45342

providing a misleading facility. It's used once in the MIPS backend
and hardcoded as "\t.globl\t" everywhere else.

llvm-svn: 45338

llvm-svn: 45335

llvm-svn: 45334

llvm-svn: 45325

llvm-svn: 45324

legalizer support goes in.

llvm-svn: 45323

llvm-svn: 45322

or after legalize.

llvm-svn: 45321

targets.

llvm-svn: 45320