TargetData to find the pointer size.

llvm-svn: 79490

array member of a struct, it's possible to land in an arbitrary position
inside that struct, such that attempting to find further getelementptr
indices will fail. In such cases, folding cannot be done.

llvm-svn: 79485

static extents of the static array type, it causes GlobalOpt and
other passes to be more conservative. This canonicalization also
allows the constant folder to add "inbounds" to GEPs.

llvm-svn: 79440

delete a newed pointer, and improve readability a little bit.

llvm-svn: 79411

before performing optimizations based on constant string values.

llvm-svn: 79384

TargetData is not present. It still uses TargetData when available.
This generalization also fixed some limitations in the TargetData
case; the attached testcase covers this.

llvm-svn: 79344

unruly indices for arrays that are members of structs.

llvm-svn: 79337

llvm-svn: 79257

llvm-svn: 79252

llvm-svn: 79230

llvm-svn: 79226

unfoldable references to a PHI node in the block being folded, and disable
the transformation in that case.  The correct transformation of such PHI
nodes depends on whether BB dominates Succ, and dominance is expensive 
to compute here.  (Alternatively, it's possible to check whether any 
uses are live, but that's also essentially a dominance calculation. 
Another alternative is to use reg2mem, but it probably isn't a good idea to
use that in simplifycfg.)

Also, remove some incorrect code from CanPropagatePredecessorsForPHIs 
which is made unnecessary with this patch: it didn't consider the case 
where a PHI node in BB has multiple uses.

llvm-svn: 79174

blocks.

llvm-svn: 79132

the new load by the old load instead of by the extract element because
a store could have occurred between the load and extract element.

llvm-svn: 78891

using formatted_raw_ostream's PadToColumn.

Before:

bb1:            ; preds = %bb
  %2 = sext i32 %i.01 to i64            ; <i64> [#uses=1]
  %3 = getelementptr double* %p, i64 %2         ; <double*> [#uses=1]
  %4 = load double* %3, align 8         ; <double> [#uses=1]
  %5 = fmul double %4, 1.100000e+00             ; <double> [#uses=1]
  %6 = sext i32 %i.01 to i64            ; <i64> [#uses=1]
  %7 = getelementptr double* %p, i64 %6         ; <double*> [#uses=1]

After:

bb1:                                        ; preds = %bb
  %2 = sext i32 %i.01 to i64                ; <i64> [#uses=1]
  %3 = getelementptr double* %p, i64 %2     ; <double*> [#uses=1]
  %4 = load double* %3, align 8             ; <double> [#uses=1]
  %5 = fmul double %4, 1.100000e+00         ; <double> [#uses=1]
  %6 = sext i32 %i.01 to i64                ; <i64> [#uses=1]
  %7 = getelementptr double* %p, i64 %6     ; <double*> [#uses=1]

Several tests required whitespace adjustments.

llvm-svn: 78816

llvm-svn: 78812

llvm-svn: 78811

llvm-svn: 78714

alignment. Only the minimum alignment guaranteed by the ABI may be assumed.

llvm-svn: 78668

llvm-svn: 78009

few places in InstCombine to use it, to fix problems handling pointer
types. This fixes the recent llvm-gcc bootstrap error.

llvm-svn: 78005

llvm-svn: 77716

 - Call RAUW to delete all instructions (this is a patch from Nick Lewycky).

llvm-svn: 77512

llvm-svn: 77439

llvm-svn: 77436

into a new BB that has no predecessors.

llvm-svn: 77433

to a few tests where it is required for the expected transformation.

llvm-svn: 77290

llvm-svn: 76864

also apply to vectors.  This allows us to compile this:

#include <emmintrin.h>
__m128i a(__m128 a, __m128 b) { return a==a & b==b; }
__m128i b(__m128 a, __m128 b) { return a!=a | b!=b; }

to:

_a:
	cmpordps	%xmm1, %xmm0
	ret
_b:
	cmpunordps	%xmm1, %xmm0
	ret

with clang instead of to a ton of horrible code.

llvm-svn: 76863

with negative tests: this test wasn't checking what it thought it was
because it was grepping .bc, not .ll.

llvm-svn: 76861

llvm-svn: 76860

llvm-svn: 76527

llvm-svn: 76492

Getelementptrs that are defined to wrap are virtually useless to
optimization, and getelementptrs that are undefined on any kind
of overflow are too restrictive -- it's difficult to ensure that
all intermediate addresses are within bounds. I'm going to take
a different approach.

Remove a few optimizations that depended on this flag.

llvm-svn: 76437

insertelement/extractelement.

I'm not entirely sure this is precisely what we want to do: should we 
prefer bitcast(insertelement) or insertelement(bitcast)?  Similarly. should we 
prefer extractelement(bitcast) or bitcast(extractelement)?

llvm-svn: 76345

way (bitcast -> insert/extractelement).

llvm-svn: 76325

sign bit set.

llvm-svn: 76304

bitcasts.

It would also be possible to canonicalize the other way; does anyone 
have a preference?

llvm-svn: 76300

where int is 32 bits.

llvm-svn: 76293

analysis. This allows indvars to emit a simpler loop trip count
expression.

llvm-svn: 76085