llvm-svn: 24611

llvm-svn: 24609

know that small negative values fit into the immediate field of addressing
modes.

llvm-svn: 24608

llvm-svn: 24607

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

internal linkage.

Patch provided by Evan Jones, thanks!

llvm-svn: 24604

llvm-svn: 24602

1. Remove redundant type casts now that PR673 is implemented.
2. Implement the OUT*ir instructions correctly.  The port number really
   *is* a 16-bit value, but the patterns should only match if the number
   is 0-255.  Update the patterns so they now match.
3. Fix patterns for shifts to reflect that the shift amount is always an
   i8, not an i16 as they were believed to be before.  This previous fib
   stopped working when we started knowing that CL has type i8.
4. Change use of i16i8imm in SH*ri patterns to all be imm.

llvm-svn: 24599

being shifted.  Don't assume they are.

llvm-svn: 24598

Add some explicit type casts so that tblgen knows the type of the shiftamount, which is not necessarily the same as the type being shifted.

llvm-svn: 24595

amount, which is not necessarily the same as the type being shifted.

llvm-svn: 24594

llvm-svn: 24592

improvements.

llvm-svn: 24591

llvm-svn: 24590

llvm-svn: 24589

llvm-svn: 24588

  chains.
* Added DAG node property SDNPHasChain for nodes which r/w control-flow
  chains.
* Renamed SDTVT to SDTOther.
* Added several new SDTypeProfiles for BR, BRCOND, RET, and WRITEPORT.
* Added SDNode definitions for BR, BRCOND, RET, and WRITEPORT.

llvm-svn: 24586

llvm-svn: 24585

llvm-svn: 24583

llvm-svn: 24581

This also fixes 177.mesa, the only program that fails with --enable-x86-fastcc
turned on.  Given a clean nightly tester run, we should be able to turn it
on by default!

llvm-svn: 24578

llvm-svn: 24574

llvm-svn: 24573

llvm-svn: 24572

should come from the arbitrary ops map.

This fixes Regression/CodeGen/PowerPC/2005-12-01-Crash.ll

llvm-svn: 24571

llvm-svn: 24568

stuff isn't using ISelLowering.cpp

llvm-svn: 24567

llvm-svn: 24566

selecting a node and use a mix of getTargetNode() and SelectNodeTo.  Because
SelectNodeTo didn't check the CSE maps for a preexisting node and didn't insert
its result into the CSE maps, we would sometimes miss a CSE opportunity.

This is extremely rare, but worth fixing for completeness.

llvm-svn: 24565

llvm-svn: 24564

work.  This change has no effect on generated code.

llvm-svn: 24563

llvm-svn: 24562

llvm-svn: 24561

llvm-svn: 24560

llvm-svn: 24559

llvm-svn: 24558

llvm-svn: 24552

llvm-svn: 24551

llvm-svn: 24550

llvm-svn: 24549