llvm-6502/test/CodeGen/ARM/fabss.ll
Jakob Stoklund Olesen 4a74b3b933 Inflate register classes after coalescing.
Coalescing can remove copy-like instructions with sub-register operands
that constrained the register class.  Examples are:

  x86: GR32_ABCD:sub_8bit_hi -> GR32
  arm: DPR_VFP2:ssub0 -> DPR

Recompute the register class of any virtual registers that are used by
less instructions after coalescing.

This affects code generation for the Cortex-A8 where we use NEON
instructions for f32 operations, c.f. fp_convert.ll:

  vadd.f32  d16, d1, d0
  vcvt.s32.f32  d0, d16

The register allocator is now free to use d16 for the temporary, and
that comes first in the allocation order because it doesn't interfere
with any s-registers.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137133 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-09 18:19:41 +00:00

30 lines
798 B
LLVM

; RUN: llc < %s -march=arm -mattr=+vfp2 | FileCheck %s -check-prefix=VFP2
; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s -check-prefix=NFP0
; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s -check-prefix=CORTEXA8
; RUN: llc < %s -march=arm -mcpu=cortex-a9 | FileCheck %s -check-prefix=CORTEXA9
define float @test(float %a, float %b) {
entry:
%dum = fadd float %a, %b
%0 = tail call float @fabsf(float %dum)
%dum1 = fadd float %0, %b
ret float %dum1
}
declare float @fabsf(float)
; VFP2: test:
; VFP2: vabs.f32 s1, s1
; NFP1: test:
; NFP1: vabs.f32 d1, d1
; NFP0: test:
; NFP0: vabs.f32 s1, s1
; CORTEXA8: test:
; CORTEXA8: vadd.f32 [[D1:d[0-9]+]]
; CORTEXA8: vabs.f32 {{d[0-9]+}}, [[D1]]
; CORTEXA9: test:
; CORTEXA9: vabs.f32 s{{.}}, s{{.}}