llvm-svn: 30959

being matched and ensure there isn't a non-direct path to the load (i.e. a
path that goes out of the sub-dag.)

llvm-svn: 30958

llvm-svn: 30957

way to reach the load via any nodes that would be folded. Start from the
root of the matched sub-tree.

llvm-svn: 30956

so that it can be deleted if unused.

llvm-svn: 30955

_test:
        stwu r1, -80(r1)
        stw r1, 76(r1)
        mflr r11
        stw r11, 88(r1)
        rlwinm r2, r4, 0, 30, 31
        cmplwi cr7, r2, 2
        mfcr r2
        rlwinm r2, r2, 30, 31, 31
        cmplwi cr0, r2, 0
        bne cr0, LBB1_2 ;UnifiedReturnBlock

after it is:

_test:
        stwu r1, -80(r1)
        stw r1, 76(r1)
        mflr r11
        stw r11, 88(r1)
        rlwinm r2, r4, 0, 30, 31
        cmplwi cr0, r2, 2
        bgt cr0, LBB1_2 ;UnifiedReturnBlock

llvm-svn: 30954

llvm-svn: 30953

SelectionDAG and it has since bitrotted.  Remove the copy from SelectionDAG.
Next, remove the constant folding piece of DAGCombiner::SimplifySetCC into
a new FoldSetCC method which can be used by getNode() and SimplifySetCC.

This fixes obscure bugs.

llvm-svn: 30952

llvm-svn: 30951

llvm-svn: 30950

llvm-svn: 30949

llvm-svn: 30948

which is undefined.  "0" isn't a power of 2.

llvm-svn: 30947

llvm-svn: 30946

llvm-svn: 30945

llvm-svn: 30944

llvm-svn: 30943

llvm-svn: 30942

it just deletes empty MBB's.  Soon it will do more :)

llvm-svn: 30941

useful and general.

llvm-svn: 30940

llvm-svn: 30939

llvm-svn: 30938

llvm-svn: 30937

llvm-svn: 30936

llvm-svn: 30935

llvm-svn: 30934

llvm-svn: 30933

llvm-svn: 30932

being destroyed at inconvenient times.  Switch to using non-local ManagedStatic
objects, which actually also speeds up ConstRules::get.

llvm-svn: 30931

llvm-svn: 30930

llvm-svn: 30929

llvm-svn: 30928

llvm-svn: 30927

llvm-svn: 30926

llvm-svn: 30925

llvm-svn: 30924

llvm-svn: 30923

llvm-svn: 30922

llvm-svn: 30921

apply to rems as well as divs.  This fixes PR945 and speeds up ReedSolomon
from 14.57s to 10.90s (which is now faster than gcc).

It compiles CodeGen/X86/rem.ll into:

_test1:
        subl $4, %esp
        movl %esi, (%esp)
        movl $2155905153, %ecx
        movl 8(%esp), %esi
        movl %esi, %eax
        imull %ecx
        addl %esi, %edx
        movl %edx, %eax
        shrl $31, %eax
        sarl $7, %edx
        addl %eax, %edx
        imull $255, %edx, %eax
        subl %eax, %esi
        movl %esi, %eax
        movl (%esp), %esi
        addl $4, %esp
        ret
_test2:
        movl 4(%esp), %eax
        movl %eax, %ecx
        sarl $31, %ecx
        shrl $24, %ecx
        addl %eax, %ecx
        andl $4294967040, %ecx
        subl %ecx, %eax
        ret
_test3:
        subl $4, %esp
        movl %esi, (%esp)
        movl $2155905153, %ecx
        movl 8(%esp), %esi
        movl %esi, %eax
        mull %ecx
        shrl $7, %edx
        imull $255, %edx, %eax
        subl %eax, %esi
        movl %esi, %eax
        movl (%esp), %esi
        addl $4, %esp
        ret

instead of div/idiv instructions.

llvm-svn: 30920