llvm-svn: 24887

llvm-svn: 24877

last night, breaking crafty and twolf.  Make sure that the newly found
legal nodes are themselves not re-legalized until the next iteration.

Also, since this functionality exists now, we can reduce number of legalizer
iterations by depending on this behavior instead of having to misuse 'do
another iteration' to get the same effect.

llvm-svn: 24875

llvm-svn: 24870

llvm-svn: 24869

creating graphs that cannot be scheduled.

llvm-svn: 24866

llvm-svn: 24861

latency period between the two.

llvm-svn: 24860

llvm-svn: 24843

llvm-svn: 24835

llvm-svn: 24828

llvm-svn: 24826

needed to be added to the ordering list, not the first member of the group.

llvm-svn: 24816

were not the same node.  Ultimately the test was bogus.

llvm-svn: 24815

llvm-svn: 24806

llvm-svn: 24804

llvm-svn: 24754

llvm-svn: 24748

if after legalize.  This fixes IA64 failures.

llvm-svn: 24725

load.  This reduces number of worklist iterations and avoid missing optimizations
depending on folding of things into sext_inreg nodes (which aren't supported by
all targets).
Tested by Regression/CodeGen/X86/extend.ll:test2

llvm-svn: 24712

Allow (zext (truncate)) to apply after legalize if the target supports
AND (which all do).

This compiles
short %foo() {
        %tmp.0 = load ubyte* %X         ; <ubyte> [#uses=1]
        %tmp.3 = cast ubyte %tmp.0 to short             ; <short> [#uses=1]
        ret short %tmp.3
}

to:
_foo:
        movzbl _X, %eax
        ret

instead of:

_foo:
        movzbl _X, %eax
        movzbl %al, %eax
        ret

thanks to Evan for pointing this out.

llvm-svn: 24709

llvm-svn: 24706

llvm-svn: 24702

llvm-svn: 24697

for emitting the ctor/dtor list for common targets.

llvm-svn: 24694

ConstantVec legalizing code, which would return constantpool nodes that
were not of the target's pointer type.

llvm-svn: 24691

llvm-svn: 24678

llvm-svn: 24677

llvm-svn: 24663

them in the PPC backend, to simplify some logic out of Select and
SelectAddr.

llvm-svn: 24657

llvm-svn: 24653

llvm-svn: 24646

out to me.

llvm-svn: 24644

1. Only forward subst offsets into loads and stores, not into arbitrary
   things, where it will likely become a load.
2. If the source is a cast from pointer, forward subst the cast as well,
   allowing us to fold the cast away (improving cases when the cast is
   from an alloca or global).

This hasn't been fully tested, but does appear to further reduce register
pressure and improve code.  Lets let the testers grind on it a bit. :)

llvm-svn: 24640

type when the target did not support them.  Also teach Legalize how to
expand ConstantVecs.

This allows us to generate

_test:
        lwz r2, 12(r3)
        lwz r4, 8(r3)
        lwz r5, 4(r3)
        lwz r6, 0(r3)
        addi r2, r2, 4
        addi r4, r4, 3
        addi r5, r5, 2
        addi r6, r6, 1
        stw r2, 12(r3)
        stw r4, 8(r3)
        stw r5, 4(r3)
        stw r6, 0(r3)
        blr

For:

void %test(%v4i *%P) {
        %T = load %v4i* %P
        %S = add %v4i %T, <int 1, int 2, int 3, int 4>
        store %v4i %S, %v4i * %P
        ret void
}

On PowerPC.

llvm-svn: 24633

if the target supports the resultant sextinreg

llvm-svn: 24632

when the types match up.  This allows the X86 backend to compile:

sbyte %toggle_value(sbyte* %tmp.1) {
        %tmp.2 = load sbyte* %tmp.1
        ret sbyte %tmp.2
}

to this:

_toggle_value:
        mov %EAX, DWORD PTR [%ESP + 4]
        movsx %EAX, BYTE PTR [%EAX]
        ret

instead of this:

_toggle_value:
        mov %EAX, DWORD PTR [%ESP + 4]
        movsx %EAX, BYTE PTR [%EAX]
        movsx %EAX, %AL
        ret

noticed in Shootout/objinst.

-Chris

llvm-svn: 24630

constant nodes with vector types.  Also teach the asm printer how to print
ConstantPacked constant pool entries.  This allows us to generate altivec
code such as the following, which adds a vector constantto a packed float.

LCPI1_0:  <4 x float> < float 0.0e+0, float 0.0e+0, float 0.0e+0, float 1.0e+0 >
        .space  4
        .space  4
        .space  4
        .long   1065353216      ; float 1
        .text
        .align  4
        .globl  _foo
_foo:
        lis r2, ha16(LCPI1_0)
        la r2, lo16(LCPI1_0)(r2)
        li r4, 0
        lvx v0, r4, r2
        lvx v1, r4, r3
        vaddfp v0, v1, v0
        stvx v0, r4, r3
        blr

For the llvm code:

void %foo(<4 x float> * %a) {
entry:
  %tmp1 = load <4 x float> * %a;
  %tmp2 = add <4 x float> %tmp1, < float 0.0, float 0.0, float 0.0, float 1.0 >
  store <4 x float> %tmp2, <4 x float> *%a
  ret void
}

llvm-svn: 24616

PPC and other targets).  In a particular, consider code like this:

struct Vector3 { double x, y, z; };
struct Matrix3 { Vector3 a, b, c; };
double dot(Vector3 &a, Vector3 &b) {
   return a.x * b.x  +  a.y * b.y  +  a.z * b.z;
}
Vector3 mul(Vector3 &a, Matrix3 &b) {
   Vector3 r;
   r.x = dot( a, b.a );
   r.y = dot( a, b.b );
   r.z = dot( a, b.c );
   return r;
}
void transform(Matrix3 &m, Vector3 *x, int n) {
   for (int i = 0; i < n; i++)
      x[i] = mul( x[i], m );
}

we compile transform to a loop with all of the GEP instructions for indexing
into 'm' pulled out of the loop (9 of them).  Because isel occurs a bb at a time
we are unable to fold the constant index into the loads in the loop, leading to
PPC code that looks like this:

LBB3_1: ; no_exit.preheader
        li r2, 0
        addi r6, r3, 64        ;; 9 values live across the loop body!
        addi r7, r3, 56
        addi r8, r3, 48
        addi r9, r3, 40
        addi r10, r3, 32
        addi r11, r3, 24
        addi r12, r3, 16
        addi r30, r3, 8
LBB3_2: ; no_exit
        lfd f0, 0(r30)
        lfd f1, 8(r4)
        fmul f0, f1, f0
        lfd f2, 0(r3)        ;; no constant indices folded into the loads!
        lfd f3, 0(r4)
        lfd f4, 0(r10)
        lfd f5, 0(r6)
        lfd f6, 0(r7)
        lfd f7, 0(r8)
        lfd f8, 0(r9)
        lfd f9, 0(r11)
        lfd f10, 0(r12)
        lfd f11, 16(r4)
        fmadd f0, f3, f2, f0
        fmul f2, f1, f4
        fmadd f0, f11, f10, f0
        fmadd f2, f3, f9, f2
        fmul f1, f1, f6
        stfd f0, 0(r4)
        fmadd f0, f11, f8, f2
        fmadd f1, f3, f7, f1
        stfd f0, 8(r4)
        fmadd f0, f11, f5, f1
        addi r29, r4, 24
        stfd f0, 16(r4)
        addi r2, r2, 1
        cmpw cr0, r2, r5
        or r4, r29, r29
        bne cr0, LBB3_2 ; no_exit

uh, yuck.  With this patch, we now sink the constant offsets into the loop, producing
this code:

LBB3_1: ; no_exit.preheader
        li r2, 0
LBB3_2: ; no_exit
        lfd f0, 8(r3)
        lfd f1, 8(r4)
        fmul f0, f1, f0
        lfd f2, 0(r3)
        lfd f3, 0(r4)
        lfd f4, 32(r3)       ;; much nicer.
        lfd f5, 64(r3)
        lfd f6, 56(r3)
        lfd f7, 48(r3)
        lfd f8, 40(r3)
        lfd f9, 24(r3)
        lfd f10, 16(r3)
        lfd f11, 16(r4)
        fmadd f0, f3, f2, f0
        fmul f2, f1, f4
        fmadd f0, f11, f10, f0
        fmadd f2, f3, f9, f2
        fmul f1, f1, f6
        stfd f0, 0(r4)
        fmadd f0, f11, f8, f2
        fmadd f1, f3, f7, f1
        stfd f0, 8(r4)
        fmadd f0, f11, f5, f1
        addi r6, r4, 24
        stfd f0, 16(r4)
        addi r2, r2, 1
        cmpw cr0, r2, r5
        or r4, r6, r6
        bne cr0, LBB3_2 ; no_exit

This is much nicer as it reduces register pressure in the loop a lot.  On X86,
this takes the function from having 9 spilled registers to 2.  This should help
some spec programs on X86 (gzip?)

This is currently only enabled with -enable-gep-isel-opt to allow perf testing
tonight.

llvm-svn: 24606

llvm-svn: 24583