array indexes.  The "complex" case of SRoA still handles them, and correctly.

This fixes a weirdness where we'd correctly avoid transforming A[0][42] if
the 42 was too large, but we'd only do it if it was one gep, not two separate
ones.

llvm-svn: 90007

llvm-svn: 90002

llvm-svn: 90001

fixes a crash analyzing consumer-lame, which had an "%X = add %X, 1"
in unreachable code.

llvm-svn: 90000

where it is not available.  It's unclear how to get this inserted 
computation into GVN's scalar availability sets, Owen, help? :)

llvm-svn: 89997

translation done by memdep, and reenable gep translation 
again.

llvm-svn: 89992

llvm-svn: 89991

llvm-svn: 89990

llvm-svn: 89985

llvm-svn: 89982

This allows us to compile the example from PR5313 into:

LBB1_2:                                                     ## %bb
	incl	%ecx
	movb	%al, (%rsi)
	movslq	%ecx, %rax
	movb	(%rdi,%rax), %al
	testb	%al, %al
	jne	LBB1_2

instead of:

LBB1_2:                                                     ## %bb
	movslq	%eax, %rcx
	incl	%eax
	movb	(%rdi,%rcx), %cl
	movb	%cl, (%rsi)
	movslq	%eax, %rcx
	cmpb	$0, (%rdi,%rcx)
	jne	LBB1_2

llvm-svn: 89981

SimplifyGEPInst method in InstructionSimplify.h.  No functionality change.

llvm-svn: 89980

come.

llvm-svn: 89979

the example in GCC PR16799 to:

LBB1_2:                                                     ## %bb1
	movl	%eax, %eax
	subq	%rax, %rdi
	movq	%rdi, (%rcx)
	movl	(%rdi), %eax
	testl	%eax, %eax
	je	LBB1_2

instead of:

LBB1_2:                                                     ## %bb1
	movl	(%rdi), %ecx
	subq	%rcx, %rdi
	movq	%rdi, (%rax)
	cmpl	$0, (%rdi)
	je	LBB1_2

llvm-svn: 89978

llvm-svn: 89976

llvm-svn: 89975

llvm-svn: 89974

remove in LLVM 3.0

llvm-svn: 89973

llvm-svn: 89972

generates store to undef and some generates store to null as the idiom
for undefined behavior.  Since simplifycfg zaps both, don't remove the
undefined behavior in instcombine.

llvm-svn: 89971

by Alastair Lynn.

llvm-svn: 89970

llvm-svn: 89968

previously already handled it at -m32 because there were no i32->i64 
extensions for addressing.

llvm-svn: 89959

llvm-svn: 89958

llvm-svn: 89957

llvm-svn: 89956

llvm-svn: 89955

to exercise this though.

llvm-svn: 89954

allows us to compile the example in readme.txt into:

LBB1_1:                                                     ## %bb
	movl	4(%rdx,%rax), %ecx
	movl	%ecx, %esi
	imull	(%rdx,%rax), %esi
	imull	%esi, %ecx
	movl	%esi, 8(%rdx,%rax)
	imull	%ecx, %esi
	movl	%ecx, 12(%rdx,%rax)
	movl	%esi, 16(%rdx,%rax)
	imull	%ecx, %esi
	movl	%esi, 20(%rdx,%rax)
	addq	$16, %rax
	cmpq	$4000, %rax
	jne	LBB1_1

instead of:

LBB1_1: 
	movl	(%rdx,%rax), %ecx
	imull	4(%rdx,%rax), %ecx
	movl	%ecx, 8(%rdx,%rax)
	imull	4(%rdx,%rax), %ecx
	movl	%ecx, 12(%rdx,%rax)
	imull	8(%rdx,%rax), %ecx
	movl	%ecx, 16(%rdx,%rax)
	imull	12(%rdx,%rax), %ecx
	movl	%ecx, 20(%rdx,%rax)
	addq	$16, %rax
	cmpq	$4000, %rax
	jne	LBB1_1

GCC (4.2) doesn't seem to be able to eliminate the loads in this 
testcase either, it generates:

L2:
	movl	(%rdx), %eax
	imull	4(%rdx), %eax
	movl	%eax, 8(%rdx)
	imull	4(%rdx), %eax
	movl	%eax, 12(%rdx)
	imull	8(%rdx), %eax
	movl	%eax, 16(%rdx)
	imull	12(%rdx), %eax
	movl	%eax, 20(%rdx)
	addl	$4, %ecx
	addq	$16, %rdx
	cmpl	$1002, %ecx
	jne	L2

llvm-svn: 89952

llvm-svn: 89951

Change the other half of aliasGEP (which handles GEP differencing) to use DecomposeGEPExpression.  This dramatically simplifies and shrinks the code by eliminating the horrible CheckGEPInstructions method, fixes a miscompilation (@test3) and makes the code more aggressive.  In particular, we now handle the @test4 case, which is reduced from the SmallPtrSet constructor.  Missing this caused us to emit a variable length memset instead of a fixed size one.

llvm-svn: 89922

Generalize DecomposeGEPExpression to exactly handle what Value::getUnderlyingObject does (when TD is around).  This allows us to avoid calling DecomposeGEPExpression unless the ultimate alias check we care about passes, speedup up BasicAA a bit.

llvm-svn: 89920

Implement a new DecomposeGEPExpression method, which decomposes a GEP into a list of scaled offsets.  Use this to eliminate some previous ad-hoc code which was subtly broken (it assumed all Constant*'s were non-zero, but strange constant express could be zero).

llvm-svn: 89915

llvm-svn: 89914

llvm-svn: 89913

llvm-svn: 89912

llvm-svn: 89905

running tail duplication when doing branch folding for if-conversion, and
we also want to be able to run tail duplication earlier to fix some
reg alloc problems.  Move the CanFallThrough function from BranchFolding
to MachineBasicBlock so that it can be shared by TailDuplication.

llvm-svn: 89904

Rollback changes r89516: Added two SubtargetFeatures::AddFeatures methods, which accept a comma-separated string or already parsed command line parameters as input, and some code re-factoring to use these new methods.

llvm-svn: 89893

ProcessImplicitDefs should watch out for invalidated iterator and extra implicit operands on copies.

llvm-svn: 89880