llvm-project/llvm/test/CodeGen/SystemZ/vec-bswap-05.ll

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

137 lines
3.9 KiB
LLVM
Raw Normal View History

; Test vector insertions of byte-swapped memory values into 0.
;
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z15 | FileCheck %s
declare i16 @llvm.bswap.i16(i16)
declare i32 @llvm.bswap.i32(i32)
declare i64 @llvm.bswap.i64(i64)
declare <8 x i16> @llvm.bswap.v8i16(<8 x i16>)
declare <4 x i32> @llvm.bswap.v4i32(<4 x i32>)
declare <2 x i64> @llvm.bswap.v2i64(<2 x i64>)
; Test VLLEBRZH.
define <8 x i16> @f1(i16 *%ptr) {
; CHECK-LABEL: f1:
; CHECK: vllebrzh %v24, 0(%r2)
; CHECK: br %r14
%val = load i16, i16 *%ptr
%swap = call i16 @llvm.bswap.i16(i16 %val)
%ret = insertelement <8 x i16> zeroinitializer, i16 %swap, i32 3
ret <8 x i16> %ret
}
; Test VLLEBRZH using a vector bswap.
define <8 x i16> @f2(i16 *%ptr) {
; CHECK-LABEL: f2:
; CHECK: vllebrzh %v24, 0(%r2)
; CHECK: br %r14
%val = load i16, i16 *%ptr
%insert = insertelement <8 x i16> zeroinitializer, i16 %val, i32 3
%ret = call <8 x i16> @llvm.bswap.v8i16(<8 x i16> %insert)
ret <8 x i16> %ret
}
; Test VLLEBRZF.
define <4 x i32> @f3(i32 *%ptr) {
; CHECK-LABEL: f3:
; CHECK: vllebrzf %v24, 0(%r2)
; CHECK: br %r14
%val = load i32, i32 *%ptr
%swap = call i32 @llvm.bswap.i32(i32 %val)
%ret = insertelement <4 x i32> zeroinitializer, i32 %swap, i32 1
ret <4 x i32> %ret
}
; Test VLLEBRZF using a vector bswap.
define <4 x i32> @f4(i32 *%ptr) {
; CHECK-LABEL: f4:
; CHECK: vllebrzf %v24, 0(%r2)
; CHECK: br %r14
%val = load i32, i32 *%ptr
%insert = insertelement <4 x i32> zeroinitializer, i32 %val, i32 1
%ret = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %insert)
ret <4 x i32> %ret
}
; Test VLLEBRZG.
define <2 x i64> @f5(i64 *%ptr) {
; CHECK-LABEL: f5:
; CHECK: vllebrzg %v24, 0(%r2)
; CHECK: br %r14
%val = load i64, i64 *%ptr
%swap = call i64 @llvm.bswap.i64(i64 %val)
%ret = insertelement <2 x i64> zeroinitializer, i64 %swap, i32 0
ret <2 x i64> %ret
}
; Test VLLEBRZG using a vector bswap.
define <2 x i64> @f6(i64 *%ptr) {
; CHECK-LABEL: f6:
; CHECK: vllebrzg %v24, 0(%r2)
; CHECK: br %r14
%val = load i64, i64 *%ptr
%insert = insertelement <2 x i64> zeroinitializer, i64 %val, i32 0
%ret = call <2 x i64> @llvm.bswap.v2i64(<2 x i64> %insert)
ret <2 x i64> %ret
}
; Test VLLEBRZE.
define <4 x i32> @f7(i32 *%ptr) {
; CHECK-LABEL: f7:
; CHECK: vllebrze %v24, 0(%r2)
; CHECK: br %r14
%val = load i32, i32 *%ptr
%swap = call i32 @llvm.bswap.i32(i32 %val)
%ret = insertelement <4 x i32> zeroinitializer, i32 %swap, i32 0
ret <4 x i32> %ret
}
; Test VLLEBRZE using a vector bswap.
define <4 x i32> @f8(i32 *%ptr) {
; CHECK-LABEL: f8:
; CHECK: vllebrze %v24, 0(%r2)
; CHECK: br %r14
%val = load i32, i32 *%ptr
%insert = insertelement <4 x i32> zeroinitializer, i32 %val, i32 0
%ret = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %insert)
ret <4 x i32> %ret
}
; Test VLLEBRZH with the highest in-range offset.
define <8 x i16> @f9(i16 *%base) {
; CHECK-LABEL: f9:
; CHECK: vllebrzh %v24, 4094(%r2)
; CHECK: br %r14
%ptr = getelementptr i16, i16 *%base, i64 2047
%val = load i16, i16 *%ptr
%swap = call i16 @llvm.bswap.i16(i16 %val)
%ret = insertelement <8 x i16> zeroinitializer, i16 %swap, i32 3
ret <8 x i16> %ret
}
; Test VLLEBRZH with the next highest offset.
define <8 x i16> @f10(i16 *%base) {
; CHECK-LABEL: f10:
; CHECK-NOT: vllebrzh %v24, 4096(%r2)
; CHECK: br %r14
%ptr = getelementptr i16, i16 *%base, i64 2048
%val = load i16, i16 *%ptr
%swap = call i16 @llvm.bswap.i16(i16 %val)
%ret = insertelement <8 x i16> zeroinitializer, i16 %swap, i32 3
ret <8 x i16> %ret
}
; Test that VLLEBRZH allows an index.
define <8 x i16> @f11(i16 *%base, i64 %index) {
; CHECK-LABEL: f11:
; CHECK: sllg [[REG:%r[1-5]]], %r3, 1
; CHECK: vllebrzh %v24, 0([[REG]],%r2)
; CHECK: br %r14
%ptr = getelementptr i16, i16 *%base, i64 %index
%val = load i16, i16 *%ptr
%swap = call i16 @llvm.bswap.i16(i16 %val)
%ret = insertelement <8 x i16> zeroinitializer, i16 %swap, i32 3
ret <8 x i16> %ret
}