llvm-svn: 122160

llvm-svn: 122159

llvm-svn: 122158

llvm-svn: 122157

llvm-svn: 122156

It turns out that ppc backend has really weird interdependencies
over different hooks and all stuff is fragile wrt small changes.
This should fix PR8749

llvm-svn: 122155

Just rename the functions, relying on matching a instruction that has the same name as a symbol is way too fragile.

llvm-svn: 122154

llvm-svn: 122153

llvm-svn: 122152

This fixed 8615.

llvm-svn: 122150

the buildbots.

llvm-svn: 122149

llvm-svn: 122148

llvm-svn: 122147

llvm-svn: 122146

of two symbols in the same fragment.

llvm-svn: 122145

llvm-svn: 122144

llvm-svn: 122143

llvm-svn: 122142

llvm-svn: 122141

llvm-svn: 122140

llvm-svn: 122139

I added a note, but suggestions on how to add a test are really welcome.

llvm-svn: 122138

llvm-svn: 122137

don't provide their own explicit visibility attributes should get them
from the template.  Fixes rdar://problem/8778497.

llvm-svn: 122136

llvm-svn: 122135

llvm-svn: 122134

use before rematerializing the load.

This allows us to produce:

    addps	LCPI0_1(%rip), %xmm2

Instead of:

    movaps	LCPI0_1(%rip), %xmm3
    addps	%xmm3, %xmm2

Saving a register and an instruction. The standard spiller already knows how to
do this.

llvm-svn: 122133

llvm-svn: 122132

ARM::tMOVgpr2gpr. But this check didn't change. As a result, we were getting
misaligned references to the jump table from an ADR instruction.

There is a test case, but unfortunately it's sensitive to random code changes.

<rdar://problem/8782223>

llvm-svn: 122131

llvm-svn: 122130

llvm-svn: 122129

llvm-svn: 122128

llvm-svn: 122127

llvm-svn: 122126

llvm-svn: 122125

llvm-svn: 122124

the loop predecessors.

The register can be live-out from a predecessor without being live-in to the
loop header if there is a critical edge from the predecessor.

llvm-svn: 122123

  lib/CodeGen/RegAllocGreedy.cpp:311: error: unused variable 'PhysReg' [-Wunused-variable]

llvm-svn: 122122

llvm-svn: 122121

llvm-svn: 122120