StrongPHIElimination.

llvm-svn: 120961

difficult on current ARM implementations for a few reasons.
1. Even though a single vmla has latency that is one cycle shorter than a pair
   of vmul + vadd, a RAW hazard during the first (4? on Cortex-a8) can cause
   additional pipeline stall. So it's frequently better to single codegen
   vmul + vadd.
2. A vmla folowed by a vmul, vmadd, or vsub causes the second fp instruction to
   stall for 4 cycles. We need to schedule them apart.
3. A vmla followed vmla is a special case. Obvious issuing back to back RAW
   vmla + vmla is very bad. But this isn't ideal either:
     vmul
     vadd
     vmla
   Instead, we want to expand the second vmla:
     vmla
     vmul
     vadd
   Even with the 4 cycle vmul stall, the second sequence is still 2 cycles
   faster.

Up to now, isel simply avoid codegen'ing fp vmla / vmls. This works well enough
but it isn't the optimial solution. This patch attempts to make it possible to
use vmla / vmls in cases where it is profitable.

A. Add missing isel predicates which cause vmla to be codegen'ed.
B. Make sure the fmul in (fadd (fmul)) has a single use. We don't want to
   compute a fmul and a fmla.
C. Add additional isel checks for vmla, avoid cases where vmla is feeding into
   fp instructions (except for the #3 exceptional case).
D. Add ARM hazard recognizer to model the vmla / vmls hazards.
E. Add a special pre-regalloc case to expand vmla / vmls when it's likely the
   vmla / vmls will trigger one of the special hazards.

Work in progress, only A+B are enabled.

llvm-svn: 120960

llvm-svn: 120959

Clarify some of the differences between indexing with getelementptr and indexing with insertvalue/extractvalue.

llvm-svn: 120957

Also add asserts that the indices are valid in InsertValueInst::init(). ExtractValueInst already asserts when constructed with invalid indices.

llvm-svn: 120956

PHIElimination.h.

llvm-svn: 120953

time, this method existed, but now PHIElimination uses the method of the same
name on MachineBasicBlock.

llvm-svn: 120952

function so that it can be shared with StrongPHIElimination.

llvm-svn: 120951

Should have no functional change other than the order of two transformations that are mutually-exclusive and the exact formatting of debug output.
Internally, it now stores the ConstantInt*s as Constant*s, and actual undef values instead of nulls.

llvm-svn: 120946

llvm-svn: 120945

  (indirectbr (select cond, blockaddress(@fn, BlockA),
                            blockaddress(@fn, BlockB)))
into
  (br cond, BlockA, BlockB).

llvm-svn: 120943

result.  This allows us to compile:

void *test12(long count) {
      return new int[count];
}

into:

test12:
	movl	$4, %ecx
	movq	%rdi, %rax
	mulq	%rcx
	movq	$-1, %rdi
	cmovnoq	%rax, %rdi
	jmp	__Znam                  ## TAILCALL

instead of:

test12:
	movl	$4, %ecx
	movq	%rdi, %rax
	mulq	%rcx
	seto	%cl
	testb	%cl, %cl
	movq	$-1, %rdi
	cmoveq	%rax, %rdi
	jmp	__Znam

Of course it would be even better if the regalloc inverted the cmov to 'cmovoq',
which would eliminate the need for the 'movq %rdi, %rax'.

llvm-svn: 120936

backend that they were all implemented except umul.  This one fell back
to the default implementation that did a hi/lo multiply and compared the
top.  Fix this to check the overflow flag that the 'mul' instruction
sets, so we can avoid an explicit test.  Now we compile:

void *func(long count) {
      return new int[count];
}

into:

__Z4funcl:                              ## @_Z4funcl
	movl	$4, %ecx                ## encoding: [0xb9,0x04,0x00,0x00,0x00]
	movq	%rdi, %rax              ## encoding: [0x48,0x89,0xf8]
	mulq	%rcx                    ## encoding: [0x48,0xf7,0xe1]
	seto	%cl                     ## encoding: [0x0f,0x90,0xc1]
	testb	%cl, %cl                ## encoding: [0x84,0xc9]
	movq	$-1, %rdi               ## encoding: [0x48,0xc7,0xc7,0xff,0xff,0xff,0xff]
	cmoveq	%rax, %rdi              ## encoding: [0x48,0x0f,0x44,0xf8]
	jmp	__Znam                  ## TAILCALL

instead of:

__Z4funcl:                              ## @_Z4funcl
	movl	$4, %ecx                ## encoding: [0xb9,0x04,0x00,0x00,0x00]
	movq	%rdi, %rax              ## encoding: [0x48,0x89,0xf8]
	mulq	%rcx                    ## encoding: [0x48,0xf7,0xe1]
	testq	%rdx, %rdx              ## encoding: [0x48,0x85,0xd2]
	movq	$-1, %rdi               ## encoding: [0x48,0xc7,0xc7,0xff,0xff,0xff,0xff]
	cmoveq	%rax, %rdi              ## encoding: [0x48,0x0f,0x44,0xf8]
	jmp	__Znam                  ## TAILCALL

Other than the silly seto+test, this is using the o bit directly, so it's going in the right
direction.

llvm-svn: 120935

llvm-svn: 120933

select, inserting a not to compensate.  Add a missing isZero check
that I lost somehow.

This improves codegen of:

void *func(long count) {
      return new int[count];
}

from:

__Z4funcl:                              ## @_Z4funcl
	movl	$4, %ecx                ## encoding: [0xb9,0x04,0x00,0x00,0x00]
	movq	%rdi, %rax              ## encoding: [0x48,0x89,0xf8]
	mulq	%rcx                    ## encoding: [0x48,0xf7,0xe1]
	testq	%rdx, %rdx              ## encoding: [0x48,0x85,0xd2]
	movq	$-1, %rdi               ## encoding: [0x48,0xc7,0xc7,0xff,0xff,0xff,0xff]
	cmoveq	%rax, %rdi              ## encoding: [0x48,0x0f,0x44,0xf8]
	jmp	__Znam                  ## TAILCALL
                                        ## encoding: [0xeb,A]

to:

__Z4funcl:                              ## @_Z4funcl
	movl	$4, %ecx                ## encoding: [0xb9,0x04,0x00,0x00,0x00]
	movq	%rdi, %rax              ## encoding: [0x48,0x89,0xf8]
	mulq	%rcx                    ## encoding: [0x48,0xf7,0xe1]
	cmpq	$1, %rdx                ## encoding: [0x48,0x83,0xfa,0x01]
	sbbq	%rdi, %rdi              ## encoding: [0x48,0x19,0xff]
	notq	%rdi                    ## encoding: [0x48,0xf7,0xd7]
	orq	%rax, %rdi              ## encoding: [0x48,0x09,0xc7]
	jmp	__Znam                  ## TAILCALL
                                        ## encoding: [0xeb,A]

llvm-svn: 120932

llvm-svn: 120930

1. generalize 
    (select (x == 0), -1, 0) -> (sign_bit (x - 1))
to:
    (select (x == 0), -1, y) -> (sign_bit (x - 1)) | y

2. Handle the identical pattern that happens with !=:
   (select (x != 0), y, -1) -> (sign_bit (x - 1)) | y

cmov is often high latency and can't fold immediates or
memory operands.  For example for (x == 0) ? -1 : 1, before 
we got:

< 	testb	%sil, %sil
< 	movl	$-1, %ecx
< 	movl	$1, %eax
< 	cmovel	%ecx, %eax

now we get:

> 	cmpb	$1, %sil
> 	sbbl	%eax, %eax
> 	orl	$1, %eax

llvm-svn: 120929

llvm-svn: 120928

llvm-svn: 120927

llvm-svn: 120926

llvm-svn: 120923

valid. Addresses will not change.

llvm-svn: 120921

having to evaluate the expression again when writing.

llvm-svn: 120920

llvm-svn: 120918

- Also adds a new POPCNT subtarget feature that is currently enabled if the target
  supports SSE4.2 (nehalem) or SSE4A (barcelona).

llvm-svn: 120917

may determine that they cannot be used uninitialized. But that might be a bit
too much for the compiler to determine.

llvm-svn: 120916

llvm-svn: 120913

llvm-svn: 120912

llvm-svn: 120911

llvm-svn: 120910

llvm-svn: 120909

Total size of bin\Release on disk goes from 82.9 MB to 74.2 MB. (~10% saving)

llvm-svn: 120908

llvm-svn: 120907

Total size of bin\Release on disk goes from 83.6 MB to 81.8MB. (~2% saving)

llvm-svn: 120901

Check for that and try narrowing it to tADDspi instead.  Radar 8724703.

llvm-svn: 120892

llvm-svn: 120891

foo = a - b
.long foo
instead of just
.long a - b

First, on darwin9 64 bits the assembler produces the wrong result. Second,
if "a" is the end of the section all darwin assemblers (9, 10 and mc) will not
consider a - b to be a constant but will if the dummy foo is created.

Split how we handle these cases. The first one is something MC should take care
of. The second one has to be handled by the caller.

llvm-svn: 120889

llvm-svn: 120888

llvm-svn: 120870

llvm-svn: 120869