llvm-svn: 21639

was contributed by Morten Ofstad, with some minor tweaks and bug fixes added
by me.

llvm-svn: 21636

llvm-svn: 21629

llvm-svn: 21606

llvm-svn: 21605

Implement Value* tracking for loads and stores in the selection DAG.  This enables one to use alias analysis in the backends.

(TRUNK)Stores and (EXT|ZEXT|SEXT)Loads have an extra SDOperand which is a SrcValueSDNode which contains the Value*.  Note that if the operation is introduced by the backend, it will still have the operand, but the value* will be null.

llvm-svn: 21599

llvm-svn: 21552

int foo1(int x, int y) {
  int t1 = x >= 0;
  int t2 = y >= 0;
  return t1 & t2;
}
int foo2(int x, int y) {
  int t1 = x == -1;
  int t2 = y == -1;
  return t1 & t2;
}

produces:

_foo1:
        or r2, r4, r3
        srwi r2, r2, 31
        xori r3, r2, 1
        blr
_foo2:
        and r2, r4, r3
        addic r2, r2, 1
        li r2, 0
        addze r3, r2
        blr

instead of:

_foo1:
        srwi r2, r4, 31
        xori r2, r2, 1
        srwi r3, r3, 31
        xori r3, r3, 1
        and r3, r2, r3
        blr
_foo2:
        addic r2, r4, 1
        li r2, 0
        addze r2, r2
        addic r3, r3, 1
        li r3, 0
        addze r3, r3
        and r3, r2, r3
        blr

llvm-svn: 21547

_foo:
        or r2, r4, r3
        srwi r3, r2, 31
        blr

instead of:

_foo:
        srwi r2, r4, 31
        srwi r3, r3, 31
        or r3, r2, r3
        blr

llvm-svn: 21544

llvm-svn: 21439

llvm-svn: 21420

bool %test(int %X) {
        %Y = and int %X, 8
        %Z = setne int %Y, 0
        ret bool %Z
}

we now generate this:

        rlwinm r2, r3, 0, 28, 28
        srwi r3, r2, 3

instead of this:

        rlwinm r2, r3, 0, 28, 28
        srwi r2, r2, 3
        rlwinm r3, r2, 0, 31, 31

I'll leave it to Nate to get it down to one instruction. :)

---------------------------------------------------------------------

llvm-svn: 21391

This turns this PPC code:

        rlwinm r2, r3, 0, 28, 28
        cmpwi cr7, r2, 8
        mfcr r2
        rlwinm r3, r2, 31, 31, 31

into this:

        rlwinm r2, r3, 0, 28, 28
        srwi r2, r2, 3
        rlwinm r3, r2, 0, 31, 31

Next up, nuking the extra and.

llvm-svn: 21390

llvm-svn: 21319

llvm-svn: 21318

llvm-svn: 21317

        // (X != 0) | (Y != 0) -> (X|Y != 0)
        // (X == 0) & (Y == 0) -> (X|Y == 0)

Compiling this:

int %bar(int %a, int %b) {
        entry:
        %tmp.1 = setne int %a, 0
        %tmp.2 = setne int %b, 0
        %tmp.3 = or bool %tmp.1, %tmp.2
        %retval = cast bool %tmp.3 to int
        ret int %retval
        }

to this:

_bar:
        or r2, r3, r4
        addic r3, r2, -1
        subfe r3, r3, r2
        blr

instead of:

_bar:
        addic r2, r3, -1
        subfe r2, r2, r3
        addic r3, r4, -1
        subfe r3, r3, r4
        or r3, r2, r3
        blr

llvm-svn: 21316

eliminating an and for Nate's testcase:

int %bar(int %a, int %b) {
        entry:
        %tmp.1 = setne int %a, 0
        %tmp.2 = setne int %b, 0
        %tmp.3 = or bool %tmp.1, %tmp.2
        %retval = cast bool %tmp.3 to int
        ret int %retval
        }

generating:

_bar:
        addic r2, r3, -1
        subfe r2, r2, r3
        addic r3, r4, -1
        subfe r3, r3, r4
        or r3, r2, r3
        blr

instead of:

_bar:
        addic r2, r3, -1
        subfe r2, r2, r3
        addic r3, r4, -1
        subfe r3, r3, r4
        or r2, r2, r3
        rlwinm r3, r2, 0, 31, 31
        blr

llvm-svn: 21315

in the PPC Pattern ISel

llvm-svn: 21297

Move the transform for select (a < 0) ? b : 0 into the dag from ppc isel
Enable the dag to fold and (setcc, 1) -> setcc for targets where setcc
  always produces zero or one.

llvm-svn: 21291

llvm-svn: 21289

llvm-svn: 21288

with != 0 comparisons vanishing.

llvm-svn: 21287

Note that this probably only works for little endian targets, but is enough
to get siod working :)

llvm-svn: 21280

llvm-svn: 21275

llvm-svn: 21273

llvm-svn: 21272

SelectionDAG to do the job with AND.  Don't legalize Z_E_I anymore as
it is gone

llvm-svn: 21266

instead.  OVerall, this increases the amount of folding we can do.

llvm-svn: 21265

llvm-svn: 21262

Make llvm undef values generate ISD::UNDEF nodes

llvm-svn: 21261

llvm-svn: 21258

compile this:

int foo (unsigned long a, unsigned long long g) {
  return a >= g;
}

To:

foo:
        movl 8(%esp), %eax
        cmpl %eax, 4(%esp)
        setae %al
        cmpl $0, 12(%esp)
        sete %cl
        andb %al, %cl
        movzbl %cl, %eax
        ret

instead of:

foo:
        movl 8(%esp), %eax
        cmpl %eax, 4(%esp)
        setae %al
        movzbw %al, %cx
        movl 12(%esp), %edx
        cmpl $0, %edx
        sete %al
        movzbw %al, %ax
        cmpl $0, %edx
        cmove %cx, %ax
        movzbl %al, %eax
        ret

llvm-svn: 21244

bool %test1(long %X) {
        %A = setlt long %X, 0
        ret bool %A
}

like this:

test1:
        cmpl $0, 8(%esp)
        setl %al
        movzbl %al, %eax
        ret

instead of:

test1:
        movl 8(%esp), %ecx
        cmpl $0, %ecx
        setl %al
        movzbw %al, %ax
        cmpl $0, 4(%esp)
        setb %dl
        movzbw %dl, %dx
        cmpl $0, %ecx
        cmove %dx, %ax
        movzbl %al, %eax
        ret

llvm-svn: 21243

test2:
        movl 8(%esp), %eax
        notl %eax
        movl 4(%esp), %ecx
        notl %ecx
        orl %eax, %ecx
        cmpl $0, %ecx
        sete %al
        movzbl %al, %eax
        ret

or this (PPC):

_test2:
        nor r2, r4, r4
        nor r3, r3, r3
        or r2, r2, r3
        cntlzw r2, r2
        srwi r3, r2, 5
        blr

Emit this:

test2:
        movl 8(%esp), %eax
        andl 4(%esp), %eax
        cmpl $-1, %eax
        sete %al
        movzbl %al, %eax
        ret

or this:

_test2:
.LBB_test2_0:   ;
        and r2, r4, r3
        cmpwi cr0, r2, -1
        li r3, 1
        li r2, 0
        beq .LBB_test2_2        ;
.LBB_test2_1:   ;
        or r3, r2, r2
.LBB_test2_2:   ;
        blr

it seems like the PPC isel could do better for R32 == -1 case.

llvm-svn: 21242

unsigned long long g;
unsigned long foo (unsigned long a) {
  return (a >= g) ? 1 : 0;
}

It changes the ppc code from:

_foo:
.LBB_foo_0:     ; entry
        mflr r11
        stw r11, 8(r1)
        bl "L00000$pb"
"L00000$pb":
        mflr r2
        addis r2, r2, ha16(L_g$non_lazy_ptr-"L00000$pb")
        lwz r2, lo16(L_g$non_lazy_ptr-"L00000$pb")(r2)
        lwz r4, 0(r2)
        lwz r2, 4(r2)
        cmplw cr0, r3, r2
        li r2, 1
        li r3, 0
        bge .LBB_foo_2  ; entry
.LBB_foo_1:     ; entry
        or r2, r3, r3
.LBB_foo_2:     ; entry
        cmplwi cr0, r4, 1
        li r3, 1
        li r5, 0
        blt .LBB_foo_4  ; entry
.LBB_foo_3:     ; entry
        or r3, r5, r5
.LBB_foo_4:     ; entry
        cmpwi cr0, r4, 0
        beq .LBB_foo_6  ; entry
.LBB_foo_5:     ; entry
        or r2, r3, r3
.LBB_foo_6:     ; entry
        rlwinm r3, r2, 0, 31, 31
        lwz r11, 8(r1)
        mtlr r11
        blr


to:

_foo:
.LBB_foo_0:     ; entry
        mflr r11
        stw r11, 8(r1)
        bl "L00000$pb"
"L00000$pb":
        mflr r2
        addis r2, r2, ha16(L_g$non_lazy_ptr-"L00000$pb")
        lwz r2, lo16(L_g$non_lazy_ptr-"L00000$pb")(r2)
        lwz r4, 0(r2)
        lwz r2, 4(r2)
        cmplw cr0, r3, r2
        li r2, 1
        li r3, 0
        bge .LBB_foo_2  ; entry
.LBB_foo_1:     ; entry
        or r2, r3, r3
.LBB_foo_2:     ; entry
        cntlzw r3, r4
        srwi r3, r3, 5
        cmpwi cr0, r4, 0
        beq .LBB_foo_4  ; entry
.LBB_foo_3:     ; entry
        or r2, r3, r3
.LBB_foo_4:     ; entry
        rlwinm r3, r2, 0, 31, 31
        lwz r11, 8(r1)
        mtlr r11
        blr

llvm-svn: 21241

long long test2(unsigned A, unsigned B) {
        return ((unsigned long long)A << 32) + B;
}

is equivalent to this:

long long test1(unsigned A, unsigned B) {
        return ((unsigned long long)A << 32) | B;
}

Now they are both codegen'd to this on ppc:

_test2:
        blr

or this on x86:

test2:
        movl 4(%esp), %edx
        movl 8(%esp), %eax
        ret

llvm-svn: 21231

masking shifts.

This fixes the miscompilation of this:

long long test1(unsigned A, unsigned B) {
        return ((unsigned long long)A << 32) | B;
}

into this:

test1:
        movl 4(%esp), %edx
        movl %edx, %eax
        orl 8(%esp), %eax
        ret

allowing us to generate this instead:

test1:
        movl 4(%esp), %edx
        movl 8(%esp), %eax
        ret

llvm-svn: 21230

Fix libcall code to not crash or assert looking for an ADJCALLSTACKUP node
  when it is known that there is no ADJCALLSTACKDOWN to match.
Expand i64 multiply when ISD::MULHU is legal for the target.

llvm-svn: 21214

the result does change as a result of the extend.

This improves codegen for Alpha on this testcase:

int %a(ushort* %i) {
        %tmp.1 = load ushort* %i
        %tmp.2 = cast ushort %tmp.1 to int
        %tmp.4 = and int %tmp.2, 1
        ret int %tmp.4
}

Generating:

a:
        ldgp $29, 0($27)
        ldwu $0,0($16)
        and $0,1,$0
        ret $31,($26),1

instead of:

a:
        ldgp $29, 0($27)
        ldwu $0,0($16)
        and $0,1,$0
        addl $0,0,$0
        ret $31,($26),1

btw, alpha really should switch to livein/outs for args :)

llvm-svn: 21213