1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-25 12:12:47 +01:00
llvm-mirror/test/CodeGen/SystemZ/vec-intrinsics-02.ll
Ulrich Weigand fc4484e0da [SystemZ] Fix/optimize vec_load_len and related intrinsics
When using vec_load/store_len_r with an immediate length operand
of 16 or larger, LLVM will currently emit an VLRL/VSTRL instruction
with that immediate.  This creates a valid encoding (which should be
supported by the assembler), but always traps at runtime.  This patch
fixes this by not creating VLRL/VSTRL in those cases.

This would result in loading the length into a register and
calling VLRLR/VSTRLR instead.  However, these operations with
a length of 15 or larger are in fact simply equivalent to a
full vector load or store.  And in fact the same holds true for
vec_load/store_len as well.

Therefore, add a DAGCombine rule to replace those operations with
plain vector loads or stores if the length is known at compile
time and equal or larger to 15.
2020-05-06 21:15:58 +02:00

459 lines
14 KiB
LLVM

; Test vector intrinsics added with z14.
;
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z14 | FileCheck %s
declare <2 x i64> @llvm.s390.vbperm(<16 x i8>, <16 x i8>)
declare <16 x i8> @llvm.s390.vmslg(<2 x i64>, <2 x i64>, <16 x i8>, i32)
declare <16 x i8> @llvm.s390.vlrl(i32, i8 *)
declare void @llvm.s390.vstrl(<16 x i8>, i32, i8 *)
declare {<4 x i32>, i32} @llvm.s390.vfcesbs(<4 x float>, <4 x float>)
declare {<4 x i32>, i32} @llvm.s390.vfchsbs(<4 x float>, <4 x float>)
declare {<4 x i32>, i32} @llvm.s390.vfchesbs(<4 x float>, <4 x float>)
declare {<4 x i32>, i32} @llvm.s390.vftcisb(<4 x float>, i32)
declare <4 x float> @llvm.s390.vfisb(<4 x float>, i32, i32)
declare <2 x double> @llvm.s390.vfmaxdb(<2 x double>, <2 x double>, i32)
declare <2 x double> @llvm.s390.vfmindb(<2 x double>, <2 x double>, i32)
declare <4 x float> @llvm.s390.vfmaxsb(<4 x float>, <4 x float>, i32)
declare <4 x float> @llvm.s390.vfminsb(<4 x float>, <4 x float>, i32)
; VBPERM.
define <2 x i64> @test_vbperm(<16 x i8> %a, <16 x i8> %b) {
; CHECK-LABEL: test_vbperm:
; CHECK: vbperm %v24, %v24, %v26
; CHECK: br %r14
%res = call <2 x i64> @llvm.s390.vbperm(<16 x i8> %a, <16 x i8> %b)
ret <2 x i64> %res
}
; VMSLG with no shifts.
define <16 x i8> @test_vmslg1(<2 x i64> %a, <2 x i64> %b, <16 x i8> %c) {
; CHECK-LABEL: test_vmslg1:
; CHECK: vmslg %v24, %v24, %v26, %v28, 0
; CHECK: br %r14
%res = call <16 x i8> @llvm.s390.vmslg(<2 x i64> %a, <2 x i64> %b, <16 x i8> %c, i32 0)
ret <16 x i8> %res
}
; VMSLG with both shifts.
define <16 x i8> @test_vmslg2(<2 x i64> %a, <2 x i64> %b, <16 x i8> %c) {
; CHECK-LABEL: test_vmslg2:
; CHECK: vmslg %v24, %v24, %v26, %v28, 12
; CHECK: br %r14
%res = call <16 x i8> @llvm.s390.vmslg(<2 x i64> %a, <2 x i64> %b, <16 x i8> %c, i32 12)
ret <16 x i8> %res
}
; VLRLR with the lowest in-range displacement.
define <16 x i8> @test_vlrlr1(i8 *%ptr, i32 %length) {
; CHECK-LABEL: test_vlrlr1:
; CHECK: vlrlr %v24, %r3, 0(%r2)
; CHECK: br %r14
%res = call <16 x i8> @llvm.s390.vlrl(i32 %length, i8 *%ptr)
ret <16 x i8> %res
}
; VLRLR with the highest in-range displacement.
define <16 x i8> @test_vlrlr2(i8 *%base, i32 %length) {
; CHECK-LABEL: test_vlrlr2:
; CHECK: vlrlr %v24, %r3, 4095(%r2)
; CHECK: br %r14
%ptr = getelementptr i8, i8 *%base, i64 4095
%res = call <16 x i8> @llvm.s390.vlrl(i32 %length, i8 *%ptr)
ret <16 x i8> %res
}
; VLRLR with an out-of-range displacement.
define <16 x i8> @test_vlrlr3(i8 *%base, i32 %length) {
; CHECK-LABEL: test_vlrlr3:
; CHECK: vlrlr %v24, %r3, 0({{%r[1-5]}})
; CHECK: br %r14
%ptr = getelementptr i8, i8 *%base, i64 4096
%res = call <16 x i8> @llvm.s390.vlrl(i32 %length, i8 *%ptr)
ret <16 x i8> %res
}
; Check that VLRLR doesn't allow an index.
define <16 x i8> @test_vlrlr4(i8 *%base, i64 %index, i32 %length) {
; CHECK-LABEL: test_vlrlr4:
; CHECK: vlrlr %v24, %r4, 0({{%r[1-5]}})
; CHECK: br %r14
%ptr = getelementptr i8, i8 *%base, i64 %index
%res = call <16 x i8> @llvm.s390.vlrl(i32 %length, i8 *%ptr)
ret <16 x i8> %res
}
; VLRL with the lowest in-range displacement.
define <16 x i8> @test_vlrl1(i8 *%ptr) {
; CHECK-LABEL: test_vlrl1:
; CHECK: vlrl %v24, 0(%r2), 0
; CHECK: br %r14
%res = call <16 x i8> @llvm.s390.vlrl(i32 0, i8 *%ptr)
ret <16 x i8> %res
}
; VLRL with the highest in-range displacement.
define <16 x i8> @test_vlrl2(i8 *%base) {
; CHECK-LABEL: test_vlrl2:
; CHECK: vlrl %v24, 4095(%r2), 0
; CHECK: br %r14
%ptr = getelementptr i8, i8 *%base, i64 4095
%res = call <16 x i8> @llvm.s390.vlrl(i32 0, i8 *%ptr)
ret <16 x i8> %res
}
; VLRL with an out-of-range displacement.
define <16 x i8> @test_vlrl3(i8 *%base) {
; CHECK-LABEL: test_vlrl3:
; CHECK: vlrl %v24, 0({{%r[1-5]}}), 0
; CHECK: br %r14
%ptr = getelementptr i8, i8 *%base, i64 4096
%res = call <16 x i8> @llvm.s390.vlrl(i32 0, i8 *%ptr)
ret <16 x i8> %res
}
; Check that VLRL doesn't allow an index.
define <16 x i8> @test_vlrl4(i8 *%base, i64 %index) {
; CHECK-LABEL: test_vlrl4:
; CHECK: vlrl %v24, 0({{%r[1-5]}}), 0
; CHECK: br %r14
%ptr = getelementptr i8, i8 *%base, i64 %index
%res = call <16 x i8> @llvm.s390.vlrl(i32 0, i8 *%ptr)
ret <16 x i8> %res
}
; VLRL with length >= 15 should become VL.
define <16 x i8> @test_vlrl5(i8 *%ptr) {
; CHECK-LABEL: test_vlrl5:
; CHECK: vl %v24, 0({{%r[1-5]}})
; CHECK: br %r14
%res = call <16 x i8> @llvm.s390.vlrl(i32 15, i8 *%ptr)
ret <16 x i8> %res
}
; VSTRLR with the lowest in-range displacement.
define void @test_vstrlr1(<16 x i8> %vec, i8 *%ptr, i32 %length) {
; CHECK-LABEL: test_vstrlr1:
; CHECK: vstrlr %v24, %r3, 0(%r2)
; CHECK: br %r14
call void @llvm.s390.vstrl(<16 x i8> %vec, i32 %length, i8 *%ptr)
ret void
}
; VSTRLR with the highest in-range displacement.
define void @test_vstrlr2(<16 x i8> %vec, i8 *%base, i32 %length) {
; CHECK-LABEL: test_vstrlr2:
; CHECK: vstrlr %v24, %r3, 4095(%r2)
; CHECK: br %r14
%ptr = getelementptr i8, i8 *%base, i64 4095
call void @llvm.s390.vstrl(<16 x i8> %vec, i32 %length, i8 *%ptr)
ret void
}
; VSTRLR with an out-of-range displacement.
define void @test_vstrlr3(<16 x i8> %vec, i8 *%base, i32 %length) {
; CHECK-LABEL: test_vstrlr3:
; CHECK: vstrlr %v24, %r3, 0({{%r[1-5]}})
; CHECK: br %r14
%ptr = getelementptr i8, i8 *%base, i64 4096
call void @llvm.s390.vstrl(<16 x i8> %vec, i32 %length, i8 *%ptr)
ret void
}
; Check that VSTRLR doesn't allow an index.
define void @test_vstrlr4(<16 x i8> %vec, i8 *%base, i64 %index, i32 %length) {
; CHECK-LABEL: test_vstrlr4:
; CHECK: vstrlr %v24, %r4, 0({{%r[1-5]}})
; CHECK: br %r14
%ptr = getelementptr i8, i8 *%base, i64 %index
call void @llvm.s390.vstrl(<16 x i8> %vec, i32 %length, i8 *%ptr)
ret void
}
; VSTRL with the lowest in-range displacement.
define void @test_vstrl1(<16 x i8> %vec, i8 *%ptr) {
; CHECK-LABEL: test_vstrl1:
; CHECK: vstrl %v24, 0(%r2), 8
; CHECK: br %r14
call void @llvm.s390.vstrl(<16 x i8> %vec, i32 8, i8 *%ptr)
ret void
}
; VSTRL with the highest in-range displacement.
define void @test_vstrl2(<16 x i8> %vec, i8 *%base) {
; CHECK-LABEL: test_vstrl2:
; CHECK: vstrl %v24, 4095(%r2), 8
; CHECK: br %r14
%ptr = getelementptr i8, i8 *%base, i64 4095
call void @llvm.s390.vstrl(<16 x i8> %vec, i32 8, i8 *%ptr)
ret void
}
; VSTRL with an out-of-range displacement.
define void @test_vstrl3(<16 x i8> %vec, i8 *%base) {
; CHECK-LABEL: test_vstrl3:
; CHECK: vstrl %v24, 0({{%r[1-5]}}), 8
; CHECK: br %r14
%ptr = getelementptr i8, i8 *%base, i64 4096
call void @llvm.s390.vstrl(<16 x i8> %vec, i32 8, i8 *%ptr)
ret void
}
; Check that VSTRL doesn't allow an index.
define void @test_vstrl4(<16 x i8> %vec, i8 *%base, i64 %index) {
; CHECK-LABEL: test_vstrl4:
; CHECK: vstrl %v24, 0({{%r[1-5]}}), 8
; CHECK: br %r14
%ptr = getelementptr i8, i8 *%base, i64 %index
call void @llvm.s390.vstrl(<16 x i8> %vec, i32 8, i8 *%ptr)
ret void
}
; VSTRL with length >= 15 should become VST.
define void @test_vstrl5(<16 x i8> %vec, i8 *%ptr) {
; CHECK-LABEL: test_vstrl5:
; CHECK: vst %v24, 0({{%r[1-5]}})
; CHECK: br %r14
call void @llvm.s390.vstrl(<16 x i8> %vec, i32 15, i8 *%ptr)
ret void
}
; VFCESBS with no processing of the result.
define i32 @test_vfcesbs(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: test_vfcesbs:
; CHECK: vfcesbs {{%v[0-9]+}}, %v24, %v26
; CHECK: ipm %r2
; CHECK: srl %r2, 28
; CHECK: br %r14
%call = call {<4 x i32>, i32} @llvm.s390.vfcesbs(<4 x float> %a,
<4 x float> %b)
%res = extractvalue {<4 x i32>, i32} %call, 1
ret i32 %res
}
; VFCESBS, returning 1 if any elements are equal (CC != 3).
define i32 @test_vfcesbs_any_bool(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: test_vfcesbs_any_bool:
; CHECK: vfcesbs {{%v[0-9]+}}, %v24, %v26
; CHECK: lhi %r2, 0
; CHECK: lochile %r2, 1
; CHECK: br %r14
%call = call {<4 x i32>, i32} @llvm.s390.vfcesbs(<4 x float> %a,
<4 x float> %b)
%res = extractvalue {<4 x i32>, i32} %call, 1
%cmp = icmp ne i32 %res, 3
%ext = zext i1 %cmp to i32
ret i32 %ext
}
; VFCESBS, storing to %ptr if any elements are equal.
define <4 x i32> @test_vfcesbs_any_store(<4 x float> %a, <4 x float> %b,
i32 *%ptr) {
; CHECK-LABEL: test_vfcesbs_any_store:
; CHECK-NOT: %r
; CHECK: vfcesbs %v24, %v24, %v26
; CHECK-NEXT: {{bor|bnler}} %r14
; CHECK: mvhi 0(%r2), 0
; CHECK: br %r14
%call = call {<4 x i32>, i32} @llvm.s390.vfcesbs(<4 x float> %a,
<4 x float> %b)
%res = extractvalue {<4 x i32>, i32} %call, 0
%cc = extractvalue {<4 x i32>, i32} %call, 1
%cmp = icmp ule i32 %cc, 2
br i1 %cmp, label %store, label %exit
store:
store i32 0, i32 *%ptr
br label %exit
exit:
ret <4 x i32> %res
}
; VFCHSBS with no processing of the result.
define i32 @test_vfchsbs(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: test_vfchsbs:
; CHECK: vfchsbs {{%v[0-9]+}}, %v24, %v26
; CHECK: ipm %r2
; CHECK: srl %r2, 28
; CHECK: br %r14
%call = call {<4 x i32>, i32} @llvm.s390.vfchsbs(<4 x float> %a,
<4 x float> %b)
%res = extractvalue {<4 x i32>, i32} %call, 1
ret i32 %res
}
; VFCHSBS, returning 1 if not all elements are higher.
define i32 @test_vfchsbs_notall_bool(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: test_vfchsbs_notall_bool:
; CHECK: vfchsbs {{%v[0-9]+}}, %v24, %v26
; CHECK: lhi %r2, 0
; CHECK: lochinhe %r2, 1
; CHECK: br %r14
%call = call {<4 x i32>, i32} @llvm.s390.vfchsbs(<4 x float> %a,
<4 x float> %b)
%res = extractvalue {<4 x i32>, i32} %call, 1
%cmp = icmp sge i32 %res, 1
%ext = zext i1 %cmp to i32
ret i32 %ext
}
; VFCHSBS, storing to %ptr if not all elements are higher.
define <4 x i32> @test_vfchsbs_notall_store(<4 x float> %a, <4 x float> %b,
i32 *%ptr) {
; CHECK-LABEL: test_vfchsbs_notall_store:
; CHECK-NOT: %r
; CHECK: vfchsbs %v24, %v24, %v26
; CHECK-NEXT: {{bher|ber}} %r14
; CHECK: mvhi 0(%r2), 0
; CHECK: br %r14
%call = call {<4 x i32>, i32} @llvm.s390.vfchsbs(<4 x float> %a,
<4 x float> %b)
%res = extractvalue {<4 x i32>, i32} %call, 0
%cc = extractvalue {<4 x i32>, i32} %call, 1
%cmp = icmp ugt i32 %cc, 0
br i1 %cmp, label %store, label %exit
store:
store i32 0, i32 *%ptr
br label %exit
exit:
ret <4 x i32> %res
}
; VFCHESBS with no processing of the result.
define i32 @test_vfchesbs(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: test_vfchesbs:
; CHECK: vfchesbs {{%v[0-9]+}}, %v24, %v26
; CHECK: ipm %r2
; CHECK: srl %r2, 28
; CHECK: br %r14
%call = call {<4 x i32>, i32} @llvm.s390.vfchesbs(<4 x float> %a,
<4 x float> %b)
%res = extractvalue {<4 x i32>, i32} %call, 1
ret i32 %res
}
; VFCHESBS, returning 1 if neither element is higher or equal.
define i32 @test_vfchesbs_none_bool(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: test_vfchesbs_none_bool:
; CHECK: vfchesbs {{%v[0-9]+}}, %v24, %v26
; CHECK: lhi %r2, 0
; CHECK: lochio %r2, 1
; CHECK: br %r14
%call = call {<4 x i32>, i32} @llvm.s390.vfchesbs(<4 x float> %a,
<4 x float> %b)
%res = extractvalue {<4 x i32>, i32} %call, 1
%cmp = icmp eq i32 %res, 3
%ext = zext i1 %cmp to i32
ret i32 %ext
}
; VFCHESBS, storing to %ptr if neither element is higher or equal.
define <4 x i32> @test_vfchesbs_none_store(<4 x float> %a, <4 x float> %b,
i32 *%ptr) {
; CHECK-LABEL: test_vfchesbs_none_store:
; CHECK-NOT: %r
; CHECK: vfchesbs %v24, %v24, %v26
; CHECK-NEXT: {{bnor|bler}} %r14
; CHECK: mvhi 0(%r2), 0
; CHECK: br %r14
%call = call {<4 x i32>, i32} @llvm.s390.vfchesbs(<4 x float> %a,
<4 x float> %b)
%res = extractvalue {<4 x i32>, i32} %call, 0
%cc = extractvalue {<4 x i32>, i32} %call, 1
%cmp = icmp uge i32 %cc, 3
br i1 %cmp, label %store, label %exit
store:
store i32 0, i32 *%ptr
br label %exit
exit:
ret <4 x i32> %res
}
; VFTCISB with the lowest useful class selector and no processing of the result.
define i32 @test_vftcisb(<4 x float> %a) {
; CHECK-LABEL: test_vftcisb:
; CHECK: vftcisb {{%v[0-9]+}}, %v24, 1
; CHECK: ipm %r2
; CHECK: srl %r2, 28
; CHECK: br %r14
%call = call {<4 x i32>, i32} @llvm.s390.vftcisb(<4 x float> %a, i32 1)
%res = extractvalue {<4 x i32>, i32} %call, 1
ret i32 %res
}
; VFTCISB with the highest useful class selector, returning 1 if all elements
; have the right class (CC == 0).
define i32 @test_vftcisb_all_bool(<4 x float> %a) {
; CHECK-LABEL: test_vftcisb_all_bool:
; CHECK: vftcisb {{%v[0-9]+}}, %v24, 4094
; CHECK: lhi %r2, 0
; CHECK: lochie %r2, 1
; CHECK: br %r14
%call = call {<4 x i32>, i32} @llvm.s390.vftcisb(<4 x float> %a, i32 4094)
%res = extractvalue {<4 x i32>, i32} %call, 1
%cmp = icmp eq i32 %res, 0
%ext = zext i1 %cmp to i32
ret i32 %ext
}
; VFISB with a rounding mode not usable via standard intrinsics.
define <4 x float> @test_vfisb_0_4(<4 x float> %a) {
; CHECK-LABEL: test_vfisb_0_4:
; CHECK: vfisb %v24, %v24, 0, 4
; CHECK: br %r14
%res = call <4 x float> @llvm.s390.vfisb(<4 x float> %a, i32 0, i32 4)
ret <4 x float> %res
}
; VFISB with IEEE-inexact exception suppressed.
define <4 x float> @test_vfisb_4_0(<4 x float> %a) {
; CHECK-LABEL: test_vfisb_4_0:
; CHECK: vfisb %v24, %v24, 4, 0
; CHECK: br %r14
%res = call <4 x float> @llvm.s390.vfisb(<4 x float> %a, i32 4, i32 0)
ret <4 x float> %res
}
; VFMAXDB.
define <2 x double> @test_vfmaxdb(<2 x double> %a, <2 x double> %b) {
; CHECK-LABEL: test_vfmaxdb:
; CHECK: vfmaxdb %v24, %v24, %v26, 4
; CHECK: br %r14
%res = call <2 x double> @llvm.s390.vfmaxdb(<2 x double> %a, <2 x double> %b, i32 4)
ret <2 x double> %res
}
; VFMINDB.
define <2 x double> @test_vfmindb(<2 x double> %a, <2 x double> %b) {
; CHECK-LABEL: test_vfmindb:
; CHECK: vfmindb %v24, %v24, %v26, 4
; CHECK: br %r14
%res = call <2 x double> @llvm.s390.vfmindb(<2 x double> %a, <2 x double> %b, i32 4)
ret <2 x double> %res
}
; VFMAXSB.
define <4 x float> @test_vfmaxsb(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: test_vfmaxsb:
; CHECK: vfmaxsb %v24, %v24, %v26, 4
; CHECK: br %r14
%res = call <4 x float> @llvm.s390.vfmaxsb(<4 x float> %a, <4 x float> %b, i32 4)
ret <4 x float> %res
}
; VFMINSB.
define <4 x float> @test_vfminsb(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: test_vfminsb:
; CHECK: vfminsb %v24, %v24, %v26, 4
; CHECK: br %r14
%res = call <4 x float> @llvm.s390.vfminsb(<4 x float> %a, <4 x float> %b, i32 4)
ret <4 x float> %res
}