[ARM][NEON] Use address space in vld([1234]|[234]lane) and vst([1234]|[234]lane) instructions
This commit changes the interface of the vld[1234], vld[234]lane, and vst[1234],
vst[234]lane ARM neon intrinsics and associates an address space with the
pointer that these intrinsics take. This changes, e.g.,
<2 x i32> @llvm.arm.neon.vld1.v2i32(i8*, i32)
to
<2 x i32> @llvm.arm.neon.vld1.v2i32.p0i8(i8*, i32)
This change ensures that address spaces are fully taken into account in the ARM
target during lowering of interleaved loads and stores.
Differential Revision: http://reviews.llvm.org/D12985
llvm-svn: 248887
diff --git a/llvm/test/CodeGen/Thumb2/crash.ll b/llvm/test/CodeGen/Thumb2/crash.ll
index 893a45d..fb32a2c 100644
--- a/llvm/test/CodeGen/Thumb2/crash.ll
+++ b/llvm/test/CodeGen/Thumb2/crash.ll
@@ -15,11 +15,11 @@
%6 = bitcast i32* %sp3 to <4 x i32>* ; <<4 x i32>*> [#uses=1]
%7 = load <4 x i32>, <4 x i32>* %6, align 16 ; <<4 x i32>> [#uses=1]
%8 = bitcast i32* %dp to i8* ; <i8*> [#uses=1]
- tail call void @llvm.arm.neon.vst4.v4i32(i8* %8, <4 x i32> %1, <4 x i32> %3, <4 x i32> %5, <4 x i32> %7, i32 1)
+ tail call void @llvm.arm.neon.vst4.p0i8.v4i32(i8* %8, <4 x i32> %1, <4 x i32> %3, <4 x i32> %5, <4 x i32> %7, i32 1)
ret void
}
-declare void @llvm.arm.neon.vst4.v4i32(i8*, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, i32) nounwind
+declare void @llvm.arm.neon.vst4.p0i8.v4i32(i8*, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, i32) nounwind
@sbuf = common global [16 x i32] zeroinitializer, align 16 ; <[16 x i32]*> [#uses=5]
@dbuf = common global [16 x i32] zeroinitializer ; <[16 x i32]*> [#uses=2]
@@ -45,7 +45,7 @@
%3 = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @sbuf, i32 0, i32 4) to <4 x i32>*), align 16 ; <<4 x i32>> [#uses=1]
%4 = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @sbuf, i32 0, i32 8) to <4 x i32>*), align 16 ; <<4 x i32>> [#uses=1]
%5 = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @sbuf, i32 0, i32 12) to <4 x i32>*), align 16 ; <<4 x i32>> [#uses=1]
- tail call void @llvm.arm.neon.vst4.v4i32(i8* bitcast ([16 x i32]* @dbuf to i8*), <4 x i32> %2, <4 x i32> %3, <4 x i32> %4, <4 x i32> %5, i32 1) nounwind
+ tail call void @llvm.arm.neon.vst4.p0i8.v4i32(i8* bitcast ([16 x i32]* @dbuf to i8*), <4 x i32> %2, <4 x i32> %3, <4 x i32> %4, <4 x i32> %5, i32 1) nounwind
ret i32 0
}
@@ -53,15 +53,15 @@
; Make sure the DPair register class can spill.
define void @pr12389(i8* %p) nounwind ssp {
entry:
- %vld1 = tail call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* %p, i32 1)
+ %vld1 = tail call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* %p, i32 1)
tail call void asm sideeffect "", "~{q0},~{q1},~{q2},~{q3},~{q4},~{q5},~{q6},~{q7},~{q8},~{q9},~{q10},~{q11},~{q12},~{q13},~{q14},~{q15}"() nounwind
- tail call void @llvm.arm.neon.vst1.v4f32(i8* %p, <4 x float> %vld1, i32 1)
+ tail call void @llvm.arm.neon.vst1.p0i8.v4f32(i8* %p, <4 x float> %vld1, i32 1)
ret void
}
-declare <4 x float> @llvm.arm.neon.vld1.v4f32(i8*, i32) nounwind readonly
+declare <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8*, i32) nounwind readonly
-declare void @llvm.arm.neon.vst1.v4f32(i8*, <4 x float>, i32) nounwind
+declare void @llvm.arm.neon.vst1.p0i8.v4f32(i8*, <4 x float>, i32) nounwind
; <rdar://problem/11101911>
; When an strd is expanded into two str instructions, make sure the first str
diff --git a/llvm/test/CodeGen/Thumb2/machine-licm.ll b/llvm/test/CodeGen/Thumb2/machine-licm.ll
index 2b1caa3..c57274e 100644
--- a/llvm/test/CodeGen/Thumb2/machine-licm.ll
+++ b/llvm/test/CodeGen/Thumb2/machine-licm.ll
@@ -59,10 +59,10 @@
%indvar = phi i32 [ %indvar.next, %bb1 ], [ 0, %entry ]
%tmp1 = shl i32 %indvar, 2
%gep1 = getelementptr i8, i8* %ptr1, i32 %tmp1
- %tmp2 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* %gep1, i32 1)
+ %tmp2 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* %gep1, i32 1)
%tmp3 = call <4 x float> @llvm.arm.neon.vmaxs.v4f32(<4 x float> <float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00>, <4 x float> %tmp2)
%gep2 = getelementptr i8, i8* %ptr2, i32 %tmp1
- call void @llvm.arm.neon.vst1.v4f32(i8* %gep2, <4 x float> %tmp3, i32 1)
+ call void @llvm.arm.neon.vst1.p0i8.v4f32(i8* %gep2, <4 x float> %tmp3, i32 1)
%indvar.next = add i32 %indvar, 1
%cond = icmp eq i32 %indvar.next, 10
br i1 %cond, label %bb2, label %bb1
@@ -73,9 +73,9 @@
; CHECK-NOT: LCPI1_0:
-declare <4 x float> @llvm.arm.neon.vld1.v4f32(i8*, i32) nounwind readonly
+declare <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8*, i32) nounwind readonly
-declare void @llvm.arm.neon.vst1.v4f32(i8*, <4 x float>, i32) nounwind
+declare void @llvm.arm.neon.vst1.p0i8.v4f32(i8*, <4 x float>, i32) nounwind
declare <4 x float> @llvm.arm.neon.vmaxs.v4f32(<4 x float>, <4 x float>) nounwind readnone
diff --git a/llvm/test/CodeGen/Thumb2/thumb2-spill-q.ll b/llvm/test/CodeGen/Thumb2/thumb2-spill-q.ll
index e0f7b5b..f408242 100644
--- a/llvm/test/CodeGen/Thumb2/thumb2-spill-q.ll
+++ b/llvm/test/CodeGen/Thumb2/thumb2-spill-q.ll
@@ -7,7 +7,7 @@
%quux = type { i32 (...)**, %baz*, i32 }
%quuz = type { %quux, i32, %bar, [128 x i8], [16 x %foo], %foo, %foo, %foo }
-declare <4 x float> @llvm.arm.neon.vld1.v4f32(i8*, i32) nounwind readonly
+declare <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8*, i32) nounwind readonly
define void @aaa(%quuz* %this, i8* %block) {
; CHECK-LABEL: aaa:
@@ -18,30 +18,30 @@
%aligned_vec = alloca <4 x float>, align 16
%"alloca point" = bitcast i32 0 to i32
%vecptr = bitcast <4 x float>* %aligned_vec to i8*
- %0 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* %vecptr, i32 1) nounwind
+ %0 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* %vecptr, i32 1) nounwind
store float 6.300000e+01, float* undef, align 4
- %1 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* undef, i32 1) nounwind ; <<4 x float>> [#uses=1]
+ %1 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* undef, i32 1) nounwind ; <<4 x float>> [#uses=1]
store float 0.000000e+00, float* undef, align 4
- %2 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* undef, i32 1) nounwind ; <<4 x float>> [#uses=1]
- %ld3 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* undef, i32 1) nounwind
+ %2 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* undef, i32 1) nounwind ; <<4 x float>> [#uses=1]
+ %ld3 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* undef, i32 1) nounwind
store float 0.000000e+00, float* undef, align 4
- %ld4 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* undef, i32 1) nounwind
+ %ld4 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* undef, i32 1) nounwind
store float 0.000000e+00, float* undef, align 4
- %ld5 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* undef, i32 1) nounwind
+ %ld5 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* undef, i32 1) nounwind
store float 0.000000e+00, float* undef, align 4
- %ld6 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* undef, i32 1) nounwind
+ %ld6 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* undef, i32 1) nounwind
store float 0.000000e+00, float* undef, align 4
- %ld7 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* undef, i32 1) nounwind
+ %ld7 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* undef, i32 1) nounwind
store float 0.000000e+00, float* undef, align 4
- %ld8 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* undef, i32 1) nounwind
+ %ld8 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* undef, i32 1) nounwind
store float 0.000000e+00, float* undef, align 4
- %ld9 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* undef, i32 1) nounwind
+ %ld9 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* undef, i32 1) nounwind
store float 0.000000e+00, float* undef, align 4
- %ld10 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* undef, i32 1) nounwind
+ %ld10 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* undef, i32 1) nounwind
store float 0.000000e+00, float* undef, align 4
- %ld11 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* undef, i32 1) nounwind
+ %ld11 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* undef, i32 1) nounwind
store float 0.000000e+00, float* undef, align 4
- %ld12 = call <4 x float> @llvm.arm.neon.vld1.v4f32(i8* undef, i32 1) nounwind
+ %ld12 = call <4 x float> @llvm.arm.neon.vld1.v4f32.p0i8(i8* undef, i32 1) nounwind
store float 0.000000e+00, float* undef, align 4
%val173 = load <4 x float>, <4 x float>* undef ; <<4 x float>> [#uses=1]
br label %bb4
diff --git a/llvm/test/CodeGen/Thumb2/v8_IT_1.ll b/llvm/test/CodeGen/Thumb2/v8_IT_1.ll
index 30dbb48..948f159 100644
--- a/llvm/test/CodeGen/Thumb2/v8_IT_1.ll
+++ b/llvm/test/CodeGen/Thumb2/v8_IT_1.ll
@@ -6,12 +6,12 @@
;CHECK: bx
define <16 x i8> @select_s_v_v(i32 %avail, i8* %bar) {
entry:
- %vld1 = call <16 x i8> @llvm.arm.neon.vld1.v16i8(i8* %bar, i32 1)
+ %vld1 = call <16 x i8> @llvm.arm.neon.vld1.v16i8.p0i8(i8* %bar, i32 1)
%and = and i32 %avail, 1
%tobool = icmp eq i32 %and, 0
%vld1. = select i1 %tobool, <16 x i8> %vld1, <16 x i8> zeroinitializer
ret <16 x i8> %vld1.
}
-declare <16 x i8> @llvm.arm.neon.vld1.v16i8(i8* , i32 )
+declare <16 x i8> @llvm.arm.neon.vld1.v16i8.p0i8(i8* , i32 )