llvm-svn: 41051

llvm-svn: 41050

see if the base register is already occupied before assuming it can be
used. This fixes bogus code generation in the accompanying testcase.

llvm-svn: 41049

llvm-svn: 41048

Into: inttoptr (i64 0 to i8*)  -> null

This occurs in the example in PR1602.  With this fixed, we now compile
the example in PR1602 into fully "devirtualized" code:

define void @_Z1g1S(%struct.S* noalias  %s) {
entry:        %tmp131415 = getelementptr %struct.S* %s, i32 0, i32 0          ; <i32 (...)***> [#uses=1]        %tmp16 = load i32 (...)*** %tmp131415, align 4          ; <i32 (...)**> [#uses=1]
        %tmp26277 = load i32 (...)** %tmp16             ; <i32 (...)*> [#uses=1]
        %tmp2829 = bitcast i32 (...)* %tmp26277 to void (%struct.S*)*           ; <void (%struct.S*)*> [#uses=1]
        tail call void %tmp2829( %struct.S* %s )
        ret void
}

This still has the vtable dispatch (as required) but does not have any pointer
to method cruft left.

llvm-svn: 41046

llvm-svn: 41043

llvm-svn: 41029

llvm-svn: 41028

llvm-svn: 41024

us to fold the entry block of PR1602 to false instead of:

br i1 icmp eq (i32 and (i32 ptrtoint (void (%struct.S*)* inttoptr (i64
1 to void (%struct.S*)*) to i32), i32 1), i32 0), label %cond_next, label
%cond_true

llvm-svn: 41023

have complex pointer manipulation going on.  This allows us to compile
stuff like this:

__m128i foo(__m128i x){
                static const unsigned int c_0[4] = { 0, 0, 0, 0 };
                __m128i v_Zero = _mm_loadu_si128((__m128i*)c_0);
                x  = _mm_unpacklo_epi8(x,  v_Zero);
                return x;
}

into:

_foo:
        xorps   %xmm1, %xmm1
        punpcklbw       %xmm1, %xmm0
        ret

llvm-svn: 41022

llvm-svn: 41021

able to 3-addressify away stuff like this:

        movl    %ecx, %eax
        decl    %eax

llvm-svn: 41020

llvm-svn: 41018

Make a 'memop' pattern just for them.

llvm-svn: 41017

llvm-svn: 41016

llvm-svn: 41013

Increase efficiency of sign_extend_inreg by using subregisters for truncation. As the README suggests sign_extend_subreg is selected to (sext(trunc)).

llvm-svn: 41010

llvm-svn: 41009

Add 2-addr to 3-addr promotion code that allows 32-bit LEA to be used via subregisters when 16-bit LEA is disabled.

llvm-svn: 41007

Move isSubRegOf into MRegisterInfo. Fix a missed move elimination in LowerSubregs and add more debugging output there.

llvm-svn: 41005

llvm-svn: 41001

llvm-svn: 40998

llvm-svn: 40997

llvm-svn: 40989

use an intptr ValueType instead of i32 for the index operand in
getCopyToParts.

llvm-svn: 40987

llvm-svn: 40986

llvm-svn: 40985

llvm-svn: 40982

llvm-svn: 40981

llvm-svn: 40980

llvm-svn: 40979

llvm-svn: 40978

not split condition constraints.

llvm-svn: 40977

llvm-svn: 40976

llvm-svn: 40975

to handle values bigger than double.  If we assume host==target and host
long double works correctly, this is not too bad, but we don't want to
have that limitation longterm.  I could implement accepting double
constants as long double or something like that, which would lead to
incorrect codegen with no errors; the more I think about that the worse
it seems.  Rather than do such a hack that would be backed out later,
I'm settling for giving reasonable error messages, for now.

llvm-svn: 40974

llvm-svn: 40973

divb / mulb outputs to ah. Under x86-64 it's not legal to read ah if the instruction requires a rex prefix (i.e. outputs to r8b, etc.). So issue shift right by 8 on AX and then truncate it to 8 bits instead.

llvm-svn: 40972

alloca, increase the alignment of the load, turning it into an aligned load.

This allows us to compile:

#include <xmmintrin.h>
__m128i foo(__m128i x){
 static const unsigned int c_0[4] = { 0, 0, 0, 0 };
	  __m128i v_Zero = _mm_loadu_si128((__m128i*)c_0);
  x  = _mm_unpacklo_epi8(x,  v_Zero);
  return x;
}

into:

_foo:
	punpcklbw	_c_0.5944, %xmm0
	ret
	.data
	.lcomm	_c_0.5944,16,4		# c_0.5944

instead of:

_foo:
	movdqu	_c_0.5944, %xmm1
	punpcklbw	%xmm1, %xmm0
	ret
	.data
	.lcomm	_c_0.5944,16,2		# c_0.5944

llvm-svn: 40971