THU-DSP-LAB
/
llvm-project
mirror of https://github.com/THU-DSP-LAB/llvm-project.git
2
0
Fork 1
Code Issues Packages Projects Releases Wiki Activity
llvm-project/llvm/test/CodeGen/RISCV/rotl-rotr.ll

1838 lines
52 KiB
LLVM
Raw Permalink Blame History

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=riscv32 -verify-machineinstrs < %s \
; RUN: | FileCheck %s -check-prefix=RV32I
; RUN: llc -mtriple=riscv64 -verify-machineinstrs -enable-legalize-types-checking < %s \
; RUN: | FileCheck %s -check-prefix=RV64I
; RUN: llc -mtriple=riscv32 -mattr=+zbb -verify-machineinstrs < %s \
; RUN: | FileCheck %s -check-prefix=RV32ZBB
; RUN: llc -mtriple=riscv64 -mattr=+zbb -verify-machineinstrs < %s \
; RUN: | FileCheck %s -check-prefix=RV64ZBB
; NOTE: -enable-legalize-types-checking is on one command line due to a previous
; assertion failure on an expensive checks build for @rotr_32_mask_multiple.
; These IR sequences are idioms for rotates. If rotate instructions are
; supported, they will be turned into ISD::ROTL or ISD::ROTR.
define i32 @rotl_32(i32 %x, i32 %y) nounwind {
; RV32I-LABEL: rotl_32:
; RV32I: # %bb.0:
; RV32I-NEXT: neg a2, a1
; RV32I-NEXT: sll a1, a0, a1
; RV32I-NEXT: srl a0, a0, a2
; RV32I-NEXT: or a0, a1, a0
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_32:
; RV64I: # %bb.0:
; RV64I-NEXT: negw a2, a1
; RV64I-NEXT: sllw a1, a0, a1
; RV64I-NEXT: srlw a0, a0, a2
; RV64I-NEXT: or a0, a1, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_32:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: rol a0, a0, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_32:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rolw a0, a0, a1
; RV64ZBB-NEXT: ret
%z = sub i32 32, %y
%b = shl i32 %x, %y
%c = lshr i32 %x, %z
%d = or i32 %b, %c
ret i32 %d
}
define i32 @rotr_32(i32 %x, i32 %y) nounwind {
; RV32I-LABEL: rotr_32:
; RV32I: # %bb.0:
; RV32I-NEXT: neg a2, a1
; RV32I-NEXT: srl a1, a0, a1
; RV32I-NEXT: sll a0, a0, a2
; RV32I-NEXT: or a0, a1, a0
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_32:
; RV64I: # %bb.0:
; RV64I-NEXT: negw a2, a1
; RV64I-NEXT: srlw a1, a0, a1
; RV64I-NEXT: sllw a0, a0, a2
; RV64I-NEXT: or a0, a1, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_32:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: ror a0, a0, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_32:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rorw a0, a0, a1
; RV64ZBB-NEXT: ret
%z = sub i32 32, %y
%b = lshr i32 %x, %y
%c = shl i32 %x, %z
%d = or i32 %b, %c
ret i32 %d
}
define i64 @rotl_64(i64 %x, i64 %y) nounwind {
; RV32I-LABEL: rotl_64:
; RV32I: # %bb.0:
; RV32I-NEXT: sll a4, a0, a2
; RV32I-NEXT: addi a3, a2, -32
; RV32I-NEXT: slti a5, a3, 0
; RV32I-NEXT: neg a5, a5
; RV32I-NEXT: bltz a3, .LBB2_2
; RV32I-NEXT: # %bb.1:
; RV32I-NEXT: sll a3, a0, a3
; RV32I-NEXT: j .LBB2_3
; RV32I-NEXT: .LBB2_2:
; RV32I-NEXT: sll a3, a1, a2
; RV32I-NEXT: xori a6, a2, 31
; RV32I-NEXT: srli a7, a0, 1
; RV32I-NEXT: srl a6, a7, a6
; RV32I-NEXT: or a3, a3, a6
; RV32I-NEXT: .LBB2_3:
; RV32I-NEXT: and a4, a5, a4
; RV32I-NEXT: neg a7, a2
; RV32I-NEXT: li a5, 32
; RV32I-NEXT: sub a6, a5, a2
; RV32I-NEXT: srl a5, a1, a7
; RV32I-NEXT: bltz a6, .LBB2_5
; RV32I-NEXT: # %bb.4:
; RV32I-NEXT: mv a0, a5
; RV32I-NEXT: j .LBB2_6
; RV32I-NEXT: .LBB2_5:
; RV32I-NEXT: srl a0, a0, a7
; RV32I-NEXT: li a7, 64
; RV32I-NEXT: sub a2, a7, a2
; RV32I-NEXT: xori a2, a2, 31
; RV32I-NEXT: slli a1, a1, 1
; RV32I-NEXT: sll a1, a1, a2
; RV32I-NEXT: or a0, a0, a1
; RV32I-NEXT: .LBB2_6:
; RV32I-NEXT: slti a1, a6, 0
; RV32I-NEXT: neg a1, a1
; RV32I-NEXT: and a1, a1, a5
; RV32I-NEXT: or a1, a3, a1
; RV32I-NEXT: or a0, a4, a0
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_64:
; RV64I: # %bb.0:
; RV64I-NEXT: negw a2, a1
; RV64I-NEXT: sll a1, a0, a1
; RV64I-NEXT: srl a0, a0, a2
; RV64I-NEXT: or a0, a1, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_64:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: sll a4, a0, a2
; RV32ZBB-NEXT: addi a3, a2, -32
; RV32ZBB-NEXT: slti a5, a3, 0
; RV32ZBB-NEXT: neg a5, a5
; RV32ZBB-NEXT: bltz a3, .LBB2_2
; RV32ZBB-NEXT: # %bb.1:
; RV32ZBB-NEXT: sll a3, a0, a3
; RV32ZBB-NEXT: j .LBB2_3
; RV32ZBB-NEXT: .LBB2_2:
; RV32ZBB-NEXT: sll a3, a1, a2
; RV32ZBB-NEXT: xori a6, a2, 31
; RV32ZBB-NEXT: srli a7, a0, 1
; RV32ZBB-NEXT: srl a6, a7, a6
; RV32ZBB-NEXT: or a3, a3, a6
; RV32ZBB-NEXT: .LBB2_3:
; RV32ZBB-NEXT: and a4, a5, a4
; RV32ZBB-NEXT: neg a7, a2
; RV32ZBB-NEXT: li a5, 32
; RV32ZBB-NEXT: sub a6, a5, a2
; RV32ZBB-NEXT: srl a5, a1, a7
; RV32ZBB-NEXT: bltz a6, .LBB2_5
; RV32ZBB-NEXT: # %bb.4:
; RV32ZBB-NEXT: mv a0, a5
; RV32ZBB-NEXT: j .LBB2_6
; RV32ZBB-NEXT: .LBB2_5:
; RV32ZBB-NEXT: srl a0, a0, a7
; RV32ZBB-NEXT: li a7, 64
; RV32ZBB-NEXT: sub a2, a7, a2
; RV32ZBB-NEXT: xori a2, a2, 31
; RV32ZBB-NEXT: slli a1, a1, 1
; RV32ZBB-NEXT: sll a1, a1, a2
; RV32ZBB-NEXT: or a0, a0, a1
; RV32ZBB-NEXT: .LBB2_6:
; RV32ZBB-NEXT: slti a1, a6, 0
; RV32ZBB-NEXT: neg a1, a1
; RV32ZBB-NEXT: and a1, a1, a5
; RV32ZBB-NEXT: or a1, a3, a1
; RV32ZBB-NEXT: or a0, a4, a0
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_64:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rol a0, a0, a1
; RV64ZBB-NEXT: ret
%z = sub i64 64, %y
%b = shl i64 %x, %y
%c = lshr i64 %x, %z
%d = or i64 %b, %c
ret i64 %d
}
define i64 @rotr_64(i64 %x, i64 %y) nounwind {
; RV32I-LABEL: rotr_64:
; RV32I: # %bb.0:
; RV32I-NEXT: srl a4, a1, a2
; RV32I-NEXT: addi a3, a2, -32
; RV32I-NEXT: slti a5, a3, 0
; RV32I-NEXT: neg a5, a5
; RV32I-NEXT: bltz a3, .LBB3_2
; RV32I-NEXT: # %bb.1:
; RV32I-NEXT: srl a3, a1, a3
; RV32I-NEXT: j .LBB3_3
; RV32I-NEXT: .LBB3_2:
; RV32I-NEXT: srl a3, a0, a2
; RV32I-NEXT: xori a6, a2, 31
; RV32I-NEXT: slli a7, a1, 1
; RV32I-NEXT: sll a6, a7, a6
; RV32I-NEXT: or a3, a3, a6
; RV32I-NEXT: .LBB3_3:
; RV32I-NEXT: and a4, a5, a4
; RV32I-NEXT: neg a7, a2
; RV32I-NEXT: li a5, 32
; RV32I-NEXT: sub a6, a5, a2
; RV32I-NEXT: sll a5, a0, a7
; RV32I-NEXT: bltz a6, .LBB3_5
; RV32I-NEXT: # %bb.4:
; RV32I-NEXT: mv a1, a5
; RV32I-NEXT: j .LBB3_6
; RV32I-NEXT: .LBB3_5:
; RV32I-NEXT: sll a1, a1, a7
; RV32I-NEXT: li a7, 64
; RV32I-NEXT: sub a2, a7, a2
; RV32I-NEXT: xori a2, a2, 31
; RV32I-NEXT: srli a0, a0, 1
; RV32I-NEXT: srl a0, a0, a2
; RV32I-NEXT: or a1, a1, a0
; RV32I-NEXT: .LBB3_6:
; RV32I-NEXT: slti a0, a6, 0
; RV32I-NEXT: neg a0, a0
; RV32I-NEXT: and a0, a0, a5
; RV32I-NEXT: or a0, a3, a0
; RV32I-NEXT: or a1, a4, a1
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_64:
; RV64I: # %bb.0:
; RV64I-NEXT: negw a2, a1
; RV64I-NEXT: srl a1, a0, a1
; RV64I-NEXT: sll a0, a0, a2
; RV64I-NEXT: or a0, a1, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_64:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: srl a4, a1, a2
; RV32ZBB-NEXT: addi a3, a2, -32
; RV32ZBB-NEXT: slti a5, a3, 0
; RV32ZBB-NEXT: neg a5, a5
; RV32ZBB-NEXT: bltz a3, .LBB3_2
; RV32ZBB-NEXT: # %bb.1:
; RV32ZBB-NEXT: srl a3, a1, a3
; RV32ZBB-NEXT: j .LBB3_3
; RV32ZBB-NEXT: .LBB3_2:
; RV32ZBB-NEXT: srl a3, a0, a2
; RV32ZBB-NEXT: xori a6, a2, 31
; RV32ZBB-NEXT: slli a7, a1, 1
; RV32ZBB-NEXT: sll a6, a7, a6
; RV32ZBB-NEXT: or a3, a3, a6
; RV32ZBB-NEXT: .LBB3_3:
; RV32ZBB-NEXT: and a4, a5, a4
; RV32ZBB-NEXT: neg a7, a2
; RV32ZBB-NEXT: li a5, 32
; RV32ZBB-NEXT: sub a6, a5, a2
; RV32ZBB-NEXT: sll a5, a0, a7
; RV32ZBB-NEXT: bltz a6, .LBB3_5
; RV32ZBB-NEXT: # %bb.4:
; RV32ZBB-NEXT: mv a1, a5
; RV32ZBB-NEXT: j .LBB3_6
; RV32ZBB-NEXT: .LBB3_5:
; RV32ZBB-NEXT: sll a1, a1, a7
; RV32ZBB-NEXT: li a7, 64
; RV32ZBB-NEXT: sub a2, a7, a2
; RV32ZBB-NEXT: xori a2, a2, 31
; RV32ZBB-NEXT: srli a0, a0, 1
; RV32ZBB-NEXT: srl a0, a0, a2
; RV32ZBB-NEXT: or a1, a1, a0
; RV32ZBB-NEXT: .LBB3_6:
; RV32ZBB-NEXT: slti a0, a6, 0
; RV32ZBB-NEXT: neg a0, a0
; RV32ZBB-NEXT: and a0, a0, a5
; RV32ZBB-NEXT: or a0, a3, a0
; RV32ZBB-NEXT: or a1, a4, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_64:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: ror a0, a0, a1
; RV64ZBB-NEXT: ret
%z = sub i64 64, %y
%b = lshr i64 %x, %y
%c = shl i64 %x, %z
%d = or i64 %b, %c
ret i64 %d
}
define i32 @rotl_32_mask(i32 %x, i32 %y) nounwind {
; RV32I-LABEL: rotl_32_mask:
; RV32I: # %bb.0:
; RV32I-NEXT: neg a2, a1
; RV32I-NEXT: sll a1, a0, a1
; RV32I-NEXT: srl a0, a0, a2
; RV32I-NEXT: or a0, a1, a0
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_32_mask:
; RV64I: # %bb.0:
; RV64I-NEXT: negw a2, a1
; RV64I-NEXT: sllw a1, a0, a1
; RV64I-NEXT: srlw a0, a0, a2
; RV64I-NEXT: or a0, a1, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_32_mask:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: rol a0, a0, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_32_mask:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rolw a0, a0, a1
; RV64ZBB-NEXT: ret
%z = sub i32 0, %y
%and = and i32 %z, 31
%b = shl i32 %x, %y
%c = lshr i32 %x, %and
%d = or i32 %b, %c
ret i32 %d
}
define i32 @rotl_32_mask_and_63_and_31(i32 %x, i32 %y) nounwind {
; RV32I-LABEL: rotl_32_mask_and_63_and_31:
; RV32I: # %bb.0:
; RV32I-NEXT: sll a2, a0, a1
; RV32I-NEXT: neg a1, a1
; RV32I-NEXT: srl a0, a0, a1
; RV32I-NEXT: or a0, a2, a0
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_32_mask_and_63_and_31:
; RV64I: # %bb.0:
; RV64I-NEXT: sllw a2, a0, a1
; RV64I-NEXT: negw a1, a1
; RV64I-NEXT: srlw a0, a0, a1
; RV64I-NEXT: or a0, a2, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_32_mask_and_63_and_31:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: rol a0, a0, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_32_mask_and_63_and_31:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rolw a0, a0, a1
; RV64ZBB-NEXT: ret
%a = and i32 %y, 63
%b = shl i32 %x, %a
%c = sub i32 0, %y
%d = and i32 %c, 31
%e = lshr i32 %x, %d
%f = or i32 %b, %e
ret i32 %f
}
define i32 @rotl_32_mask_or_64_or_32(i32 %x, i32 %y) nounwind {
; RV32I-LABEL: rotl_32_mask_or_64_or_32:
; RV32I: # %bb.0:
; RV32I-NEXT: ori a2, a1, 64
; RV32I-NEXT: sll a2, a0, a2
; RV32I-NEXT: neg a1, a1
; RV32I-NEXT: ori a1, a1, 32
; RV32I-NEXT: srl a0, a0, a1
; RV32I-NEXT: or a0, a2, a0
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_32_mask_or_64_or_32:
; RV64I: # %bb.0:
; RV64I-NEXT: sllw a2, a0, a1
; RV64I-NEXT: negw a1, a1
; RV64I-NEXT: srlw a0, a0, a1
; RV64I-NEXT: or a0, a2, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_32_mask_or_64_or_32:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: rol a0, a0, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_32_mask_or_64_or_32:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rolw a0, a0, a1
; RV64ZBB-NEXT: ret
%a = or i32 %y, 64
%b = shl i32 %x, %a
%c = sub i32 0, %y
%d = or i32 %c, 32
%e = lshr i32 %x, %d
%f = or i32 %b, %e
ret i32 %f
}
define i32 @rotr_32_mask(i32 %x, i32 %y) nounwind {
; RV32I-LABEL: rotr_32_mask:
; RV32I: # %bb.0:
; RV32I-NEXT: neg a2, a1
; RV32I-NEXT: srl a1, a0, a1
; RV32I-NEXT: sll a0, a0, a2
; RV32I-NEXT: or a0, a1, a0
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_32_mask:
; RV64I: # %bb.0:
; RV64I-NEXT: negw a2, a1
; RV64I-NEXT: srlw a1, a0, a1
; RV64I-NEXT: sllw a0, a0, a2
; RV64I-NEXT: or a0, a1, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_32_mask:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: ror a0, a0, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_32_mask:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rorw a0, a0, a1
; RV64ZBB-NEXT: ret
%z = sub i32 0, %y
%and = and i32 %z, 31
%b = lshr i32 %x, %y
%c = shl i32 %x, %and
%d = or i32 %b, %c
ret i32 %d
}
define i32 @rotr_32_mask_and_63_and_31(i32 %x, i32 %y) nounwind {
; RV32I-LABEL: rotr_32_mask_and_63_and_31:
; RV32I: # %bb.0:
; RV32I-NEXT: srl a2, a0, a1
; RV32I-NEXT: neg a1, a1
; RV32I-NEXT: sll a0, a0, a1
; RV32I-NEXT: or a0, a2, a0
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_32_mask_and_63_and_31:
; RV64I: # %bb.0:
; RV64I-NEXT: srlw a2, a0, a1
; RV64I-NEXT: negw a1, a1
; RV64I-NEXT: sllw a0, a0, a1
; RV64I-NEXT: or a0, a2, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_32_mask_and_63_and_31:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: ror a0, a0, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_32_mask_and_63_and_31:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rorw a0, a0, a1
; RV64ZBB-NEXT: ret
%a = and i32 %y, 63
%b = lshr i32 %x, %a
%c = sub i32 0, %y
%d = and i32 %c, 31
%e = shl i32 %x, %d
%f = or i32 %b, %e
ret i32 %f
}
define i32 @rotr_32_mask_or_64_or_32(i32 %x, i32 %y) nounwind {
; RV32I-LABEL: rotr_32_mask_or_64_or_32:
; RV32I: # %bb.0:
; RV32I-NEXT: ori a2, a1, 64
; RV32I-NEXT: srl a2, a0, a2
; RV32I-NEXT: neg a1, a1
; RV32I-NEXT: ori a1, a1, 32
; RV32I-NEXT: sll a0, a0, a1
; RV32I-NEXT: or a0, a2, a0
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_32_mask_or_64_or_32:
; RV64I: # %bb.0:
; RV64I-NEXT: srlw a2, a0, a1
; RV64I-NEXT: negw a1, a1
; RV64I-NEXT: sllw a0, a0, a1
; RV64I-NEXT: or a0, a2, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_32_mask_or_64_or_32:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: ror a0, a0, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_32_mask_or_64_or_32:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rorw a0, a0, a1
; RV64ZBB-NEXT: ret
%a = or i32 %y, 64
%b = lshr i32 %x, %a
%c = sub i32 0, %y
%d = or i32 %c, 32
%e = shl i32 %x, %d
%f = or i32 %b, %e
ret i32 %f
}
define i64 @rotl_64_mask(i64 %x, i64 %y) nounwind {
; RV32I-LABEL: rotl_64_mask:
; RV32I: # %bb.0:
; RV32I-NEXT: addi a4, a2, -32
; RV32I-NEXT: bltz a4, .LBB10_2
; RV32I-NEXT: # %bb.1:
; RV32I-NEXT: sll a3, a0, a4
; RV32I-NEXT: j .LBB10_3
; RV32I-NEXT: .LBB10_2:
; RV32I-NEXT: sll a3, a1, a2
; RV32I-NEXT: xori a5, a2, 31
; RV32I-NEXT: srli a6, a0, 1
; RV32I-NEXT: srl a5, a6, a5
; RV32I-NEXT: or a3, a3, a5
; RV32I-NEXT: .LBB10_3:
; RV32I-NEXT: sll a5, a0, a2
; RV32I-NEXT: slti a4, a4, 0
; RV32I-NEXT: neg a4, a4
; RV32I-NEXT: and a4, a4, a5
; RV32I-NEXT: neg a5, a2
; RV32I-NEXT: srl a2, a1, a5
; RV32I-NEXT: andi a6, a5, 63
; RV32I-NEXT: addi a7, a6, -32
; RV32I-NEXT: slti t0, a7, 0
; RV32I-NEXT: neg t0, t0
; RV32I-NEXT: and a2, t0, a2
; RV32I-NEXT: bltz a7, .LBB10_5
; RV32I-NEXT: # %bb.4:
; RV32I-NEXT: srl a0, a1, a7
; RV32I-NEXT: j .LBB10_6
; RV32I-NEXT: .LBB10_5:
; RV32I-NEXT: srl a0, a0, a5
; RV32I-NEXT: xori a5, a6, 31
; RV32I-NEXT: slli a1, a1, 1
; RV32I-NEXT: sll a1, a1, a5
; RV32I-NEXT: or a0, a0, a1
; RV32I-NEXT: .LBB10_6:
; RV32I-NEXT: or a0, a4, a0
; RV32I-NEXT: or a1, a3, a2
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_64_mask:
; RV64I: # %bb.0:
; RV64I-NEXT: negw a2, a1
; RV64I-NEXT: sll a1, a0, a1
; RV64I-NEXT: srl a0, a0, a2
; RV64I-NEXT: or a0, a1, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_64_mask:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: addi a4, a2, -32
; RV32ZBB-NEXT: bltz a4, .LBB10_2
; RV32ZBB-NEXT: # %bb.1:
; RV32ZBB-NEXT: sll a3, a0, a4
; RV32ZBB-NEXT: j .LBB10_3
; RV32ZBB-NEXT: .LBB10_2:
; RV32ZBB-NEXT: sll a3, a1, a2
; RV32ZBB-NEXT: xori a5, a2, 31
; RV32ZBB-NEXT: srli a6, a0, 1
; RV32ZBB-NEXT: srl a5, a6, a5
; RV32ZBB-NEXT: or a3, a3, a5
; RV32ZBB-NEXT: .LBB10_3:
; RV32ZBB-NEXT: sll a5, a0, a2
; RV32ZBB-NEXT: slti a4, a4, 0
; RV32ZBB-NEXT: neg a4, a4
; RV32ZBB-NEXT: and a4, a4, a5
; RV32ZBB-NEXT: neg a5, a2
; RV32ZBB-NEXT: srl a2, a1, a5
; RV32ZBB-NEXT: andi a6, a5, 63
; RV32ZBB-NEXT: addi a7, a6, -32
; RV32ZBB-NEXT: slti t0, a7, 0
; RV32ZBB-NEXT: neg t0, t0
; RV32ZBB-NEXT: and a2, t0, a2
; RV32ZBB-NEXT: bltz a7, .LBB10_5
; RV32ZBB-NEXT: # %bb.4:
; RV32ZBB-NEXT: srl a0, a1, a7
; RV32ZBB-NEXT: j .LBB10_6
; RV32ZBB-NEXT: .LBB10_5:
; RV32ZBB-NEXT: srl a0, a0, a5
; RV32ZBB-NEXT: xori a5, a6, 31
; RV32ZBB-NEXT: slli a1, a1, 1
; RV32ZBB-NEXT: sll a1, a1, a5
; RV32ZBB-NEXT: or a0, a0, a1
; RV32ZBB-NEXT: .LBB10_6:
; RV32ZBB-NEXT: or a0, a4, a0
; RV32ZBB-NEXT: or a1, a3, a2
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_64_mask:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rol a0, a0, a1
; RV64ZBB-NEXT: ret
%z = sub i64 0, %y
%and = and i64 %z, 63
%b = shl i64 %x, %y
%c = lshr i64 %x, %and
%d = or i64 %b, %c
ret i64 %d
}
define i64 @rotl_64_mask_and_127_and_63(i64 %x, i64 %y) nounwind {
; RV32I-LABEL: rotl_64_mask_and_127_and_63:
; RV32I: # %bb.0:
; RV32I-NEXT: andi a3, a2, 127
; RV32I-NEXT: addi a4, a3, -32
; RV32I-NEXT: bltz a4, .LBB11_2
; RV32I-NEXT: # %bb.1:
; RV32I-NEXT: sll a3, a0, a4
; RV32I-NEXT: j .LBB11_3
; RV32I-NEXT: .LBB11_2:
; RV32I-NEXT: sll a5, a1, a2
; RV32I-NEXT: srli a6, a0, 1
; RV32I-NEXT: xori a3, a3, 31
; RV32I-NEXT: srl a3, a6, a3
; RV32I-NEXT: or a3, a5, a3
; RV32I-NEXT: .LBB11_3:
; RV32I-NEXT: sll a5, a0, a2
; RV32I-NEXT: slti a4, a4, 0
; RV32I-NEXT: neg a4, a4
; RV32I-NEXT: and a4, a4, a5
; RV32I-NEXT: neg a5, a2
; RV32I-NEXT: srl a2, a1, a5
; RV32I-NEXT: andi a6, a5, 63
; RV32I-NEXT: addi a7, a6, -32
; RV32I-NEXT: slti t0, a7, 0
; RV32I-NEXT: neg t0, t0
; RV32I-NEXT: and a2, t0, a2
; RV32I-NEXT: bltz a7, .LBB11_5
; RV32I-NEXT: # %bb.4:
; RV32I-NEXT: srl a0, a1, a7
; RV32I-NEXT: j .LBB11_6
; RV32I-NEXT: .LBB11_5:
; RV32I-NEXT: srl a0, a0, a5
; RV32I-NEXT: xori a5, a6, 31
; RV32I-NEXT: slli a1, a1, 1
; RV32I-NEXT: sll a1, a1, a5
; RV32I-NEXT: or a0, a0, a1
; RV32I-NEXT: .LBB11_6:
; RV32I-NEXT: or a0, a4, a0
; RV32I-NEXT: or a1, a3, a2
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_64_mask_and_127_and_63:
; RV64I: # %bb.0:
; RV64I-NEXT: sll a2, a0, a1
; RV64I-NEXT: negw a1, a1
; RV64I-NEXT: srl a0, a0, a1
; RV64I-NEXT: or a0, a2, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_64_mask_and_127_and_63:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: andi a3, a2, 127
; RV32ZBB-NEXT: addi a4, a3, -32
; RV32ZBB-NEXT: bltz a4, .LBB11_2
; RV32ZBB-NEXT: # %bb.1:
; RV32ZBB-NEXT: sll a3, a0, a4
; RV32ZBB-NEXT: j .LBB11_3
; RV32ZBB-NEXT: .LBB11_2:
; RV32ZBB-NEXT: sll a5, a1, a2
; RV32ZBB-NEXT: srli a6, a0, 1
; RV32ZBB-NEXT: xori a3, a3, 31
; RV32ZBB-NEXT: srl a3, a6, a3
; RV32ZBB-NEXT: or a3, a5, a3
; RV32ZBB-NEXT: .LBB11_3:
; RV32ZBB-NEXT: sll a5, a0, a2
; RV32ZBB-NEXT: slti a4, a4, 0
; RV32ZBB-NEXT: neg a4, a4
; RV32ZBB-NEXT: and a4, a4, a5
; RV32ZBB-NEXT: neg a5, a2
; RV32ZBB-NEXT: srl a2, a1, a5
; RV32ZBB-NEXT: andi a6, a5, 63
; RV32ZBB-NEXT: addi a7, a6, -32
; RV32ZBB-NEXT: slti t0, a7, 0
; RV32ZBB-NEXT: neg t0, t0
; RV32ZBB-NEXT: and a2, t0, a2
; RV32ZBB-NEXT: bltz a7, .LBB11_5
; RV32ZBB-NEXT: # %bb.4:
; RV32ZBB-NEXT: srl a0, a1, a7
; RV32ZBB-NEXT: j .LBB11_6
; RV32ZBB-NEXT: .LBB11_5:
; RV32ZBB-NEXT: srl a0, a0, a5
; RV32ZBB-NEXT: xori a5, a6, 31
; RV32ZBB-NEXT: slli a1, a1, 1
; RV32ZBB-NEXT: sll a1, a1, a5
; RV32ZBB-NEXT: or a0, a0, a1
; RV32ZBB-NEXT: .LBB11_6:
; RV32ZBB-NEXT: or a0, a4, a0
; RV32ZBB-NEXT: or a1, a3, a2
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_64_mask_and_127_and_63:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rol a0, a0, a1
; RV64ZBB-NEXT: ret
%a = and i64 %y, 127
%b = shl i64 %x, %a
%c = sub i64 0, %y
%d = and i64 %c, 63
%e = lshr i64 %x, %d
%f = or i64 %b, %e
ret i64 %f
}
define i64 @rotl_64_mask_or_128_or_64(i64 %x, i64 %y) nounwind {
; RV32I-LABEL: rotl_64_mask_or_128_or_64:
; RV32I: # %bb.0:
; RV32I-NEXT: sll a3, a0, a2
; RV32I-NEXT: neg a0, a2
; RV32I-NEXT: srl a0, a1, a0
; RV32I-NEXT: mv a1, a3
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_64_mask_or_128_or_64:
; RV64I: # %bb.0:
; RV64I-NEXT: ori a2, a1, 128
; RV64I-NEXT: sll a2, a0, a2
; RV64I-NEXT: neg a1, a1
; RV64I-NEXT: ori a1, a1, 64
; RV64I-NEXT: srl a0, a0, a1
; RV64I-NEXT: or a0, a2, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_64_mask_or_128_or_64:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: sll a3, a0, a2
; RV32ZBB-NEXT: neg a0, a2
; RV32ZBB-NEXT: srl a0, a1, a0
; RV32ZBB-NEXT: mv a1, a3
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_64_mask_or_128_or_64:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rol a0, a0, a1
; RV64ZBB-NEXT: ret
%a = or i64 %y, 128
%b = shl i64 %x, %a
%c = sub i64 0, %y
%d = or i64 %c, 64
%e = lshr i64 %x, %d
%f = or i64 %b, %e
ret i64 %f
}
define i64 @rotr_64_mask(i64 %x, i64 %y) nounwind {
; RV32I-LABEL: rotr_64_mask:
; RV32I: # %bb.0:
; RV32I-NEXT: srl a4, a1, a2
; RV32I-NEXT: addi a3, a2, -32
; RV32I-NEXT: slti a5, a3, 0
; RV32I-NEXT: neg a5, a5
; RV32I-NEXT: bltz a3, .LBB13_2
; RV32I-NEXT: # %bb.1:
; RV32I-NEXT: srl a3, a1, a3
; RV32I-NEXT: j .LBB13_3
; RV32I-NEXT: .LBB13_2:
; RV32I-NEXT: srl a3, a0, a2
; RV32I-NEXT: xori a6, a2, 31
; RV32I-NEXT: slli a7, a1, 1
; RV32I-NEXT: sll a6, a7, a6
; RV32I-NEXT: or a3, a3, a6
; RV32I-NEXT: .LBB13_3:
; RV32I-NEXT: neg a6, a2
; RV32I-NEXT: andi t0, a6, 63
; RV32I-NEXT: addi a7, t0, -32
; RV32I-NEXT: and a2, a5, a4
; RV32I-NEXT: bltz a7, .LBB13_5
; RV32I-NEXT: # %bb.4:
; RV32I-NEXT: sll a1, a0, a7
; RV32I-NEXT: j .LBB13_6
; RV32I-NEXT: .LBB13_5:
; RV32I-NEXT: sll a1, a1, a6
; RV32I-NEXT: xori a4, t0, 31
; RV32I-NEXT: srli a5, a0, 1
; RV32I-NEXT: srl a4, a5, a4
; RV32I-NEXT: or a1, a1, a4
; RV32I-NEXT: .LBB13_6:
; RV32I-NEXT: sll a0, a0, a6
; RV32I-NEXT: slti a4, a7, 0
; RV32I-NEXT: neg a4, a4
; RV32I-NEXT: and a0, a4, a0
; RV32I-NEXT: or a0, a3, a0
; RV32I-NEXT: or a1, a2, a1
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_64_mask:
; RV64I: # %bb.0:
; RV64I-NEXT: negw a2, a1
; RV64I-NEXT: srl a1, a0, a1
; RV64I-NEXT: sll a0, a0, a2
; RV64I-NEXT: or a0, a1, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_64_mask:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: srl a4, a1, a2
; RV32ZBB-NEXT: addi a3, a2, -32
; RV32ZBB-NEXT: slti a5, a3, 0
; RV32ZBB-NEXT: neg a5, a5
; RV32ZBB-NEXT: bltz a3, .LBB13_2
; RV32ZBB-NEXT: # %bb.1:
; RV32ZBB-NEXT: srl a3, a1, a3
; RV32ZBB-NEXT: j .LBB13_3
; RV32ZBB-NEXT: .LBB13_2:
; RV32ZBB-NEXT: srl a3, a0, a2
; RV32ZBB-NEXT: xori a6, a2, 31
; RV32ZBB-NEXT: slli a7, a1, 1
; RV32ZBB-NEXT: sll a6, a7, a6
; RV32ZBB-NEXT: or a3, a3, a6
; RV32ZBB-NEXT: .LBB13_3:
; RV32ZBB-NEXT: neg a6, a2
; RV32ZBB-NEXT: andi t0, a6, 63
; RV32ZBB-NEXT: addi a7, t0, -32
; RV32ZBB-NEXT: and a2, a5, a4
; RV32ZBB-NEXT: bltz a7, .LBB13_5
; RV32ZBB-NEXT: # %bb.4:
; RV32ZBB-NEXT: sll a1, a0, a7
; RV32ZBB-NEXT: j .LBB13_6
; RV32ZBB-NEXT: .LBB13_5:
; RV32ZBB-NEXT: sll a1, a1, a6
; RV32ZBB-NEXT: xori a4, t0, 31
; RV32ZBB-NEXT: srli a5, a0, 1
; RV32ZBB-NEXT: srl a4, a5, a4
; RV32ZBB-NEXT: or a1, a1, a4
; RV32ZBB-NEXT: .LBB13_6:
; RV32ZBB-NEXT: sll a0, a0, a6
; RV32ZBB-NEXT: slti a4, a7, 0
; RV32ZBB-NEXT: neg a4, a4
; RV32ZBB-NEXT: and a0, a4, a0
; RV32ZBB-NEXT: or a0, a3, a0
; RV32ZBB-NEXT: or a1, a2, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_64_mask:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: ror a0, a0, a1
; RV64ZBB-NEXT: ret
%z = sub i64 0, %y
%and = and i64 %z, 63
%b = lshr i64 %x, %y
%c = shl i64 %x, %and
%d = or i64 %b, %c
ret i64 %d
}
define i64 @rotr_64_mask_and_127_and_63(i64 %x, i64 %y) nounwind {
; RV32I-LABEL: rotr_64_mask_and_127_and_63:
; RV32I: # %bb.0:
; RV32I-NEXT: srl a4, a1, a2
; RV32I-NEXT: andi a3, a2, 127
; RV32I-NEXT: addi a6, a3, -32
; RV32I-NEXT: slti a5, a6, 0
; RV32I-NEXT: neg a5, a5
; RV32I-NEXT: bltz a6, .LBB14_2
; RV32I-NEXT: # %bb.1:
; RV32I-NEXT: srl a3, a1, a6
; RV32I-NEXT: j .LBB14_3
; RV32I-NEXT: .LBB14_2:
; RV32I-NEXT: srl a6, a0, a2
; RV32I-NEXT: slli a7, a1, 1
; RV32I-NEXT: xori a3, a3, 31
; RV32I-NEXT: sll a3, a7, a3
; RV32I-NEXT: or a3, a6, a3
; RV32I-NEXT: .LBB14_3:
; RV32I-NEXT: neg a6, a2
; RV32I-NEXT: andi t0, a6, 63
; RV32I-NEXT: addi a7, t0, -32
; RV32I-NEXT: and a2, a5, a4
; RV32I-NEXT: bltz a7, .LBB14_5
; RV32I-NEXT: # %bb.4:
; RV32I-NEXT: sll a1, a0, a7
; RV32I-NEXT: j .LBB14_6
; RV32I-NEXT: .LBB14_5:
; RV32I-NEXT: sll a1, a1, a6
; RV32I-NEXT: xori a4, t0, 31
; RV32I-NEXT: srli a5, a0, 1
; RV32I-NEXT: srl a4, a5, a4
; RV32I-NEXT: or a1, a1, a4
; RV32I-NEXT: .LBB14_6:
; RV32I-NEXT: sll a0, a0, a6
; RV32I-NEXT: slti a4, a7, 0
; RV32I-NEXT: neg a4, a4
; RV32I-NEXT: and a0, a4, a0
; RV32I-NEXT: or a0, a3, a0
; RV32I-NEXT: or a1, a2, a1
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_64_mask_and_127_and_63:
; RV64I: # %bb.0:
; RV64I-NEXT: srl a2, a0, a1
; RV64I-NEXT: negw a1, a1
; RV64I-NEXT: sll a0, a0, a1
; RV64I-NEXT: or a0, a2, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_64_mask_and_127_and_63:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: srl a4, a1, a2
; RV32ZBB-NEXT: andi a3, a2, 127
; RV32ZBB-NEXT: addi a6, a3, -32
; RV32ZBB-NEXT: slti a5, a6, 0
; RV32ZBB-NEXT: neg a5, a5
; RV32ZBB-NEXT: bltz a6, .LBB14_2
; RV32ZBB-NEXT: # %bb.1:
; RV32ZBB-NEXT: srl a3, a1, a6
; RV32ZBB-NEXT: j .LBB14_3
; RV32ZBB-NEXT: .LBB14_2:
; RV32ZBB-NEXT: srl a6, a0, a2
; RV32ZBB-NEXT: slli a7, a1, 1
; RV32ZBB-NEXT: xori a3, a3, 31
; RV32ZBB-NEXT: sll a3, a7, a3
; RV32ZBB-NEXT: or a3, a6, a3
; RV32ZBB-NEXT: .LBB14_3:
; RV32ZBB-NEXT: neg a6, a2
; RV32ZBB-NEXT: andi t0, a6, 63
; RV32ZBB-NEXT: addi a7, t0, -32
; RV32ZBB-NEXT: and a2, a5, a4
; RV32ZBB-NEXT: bltz a7, .LBB14_5
; RV32ZBB-NEXT: # %bb.4:
; RV32ZBB-NEXT: sll a1, a0, a7
; RV32ZBB-NEXT: j .LBB14_6
; RV32ZBB-NEXT: .LBB14_5:
; RV32ZBB-NEXT: sll a1, a1, a6
; RV32ZBB-NEXT: xori a4, t0, 31
; RV32ZBB-NEXT: srli a5, a0, 1
; RV32ZBB-NEXT: srl a4, a5, a4
; RV32ZBB-NEXT: or a1, a1, a4
; RV32ZBB-NEXT: .LBB14_6:
; RV32ZBB-NEXT: sll a0, a0, a6
; RV32ZBB-NEXT: slti a4, a7, 0
; RV32ZBB-NEXT: neg a4, a4
; RV32ZBB-NEXT: and a0, a4, a0
; RV32ZBB-NEXT: or a0, a3, a0
; RV32ZBB-NEXT: or a1, a2, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_64_mask_and_127_and_63:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: ror a0, a0, a1
; RV64ZBB-NEXT: ret
%a = and i64 %y, 127
%b = lshr i64 %x, %a
%c = sub i64 0, %y
%d = and i64 %c, 63
%e = shl i64 %x, %d
%f = or i64 %b, %e
ret i64 %f
}
define i64 @rotr_64_mask_or_128_or_64(i64 %x, i64 %y) nounwind {
; RV32I-LABEL: rotr_64_mask_or_128_or_64:
; RV32I: # %bb.0:
; RV32I-NEXT: srl a3, a1, a2
; RV32I-NEXT: neg a1, a2
; RV32I-NEXT: sll a1, a0, a1
; RV32I-NEXT: mv a0, a3
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_64_mask_or_128_or_64:
; RV64I: # %bb.0:
; RV64I-NEXT: ori a2, a1, 128
; RV64I-NEXT: srl a2, a0, a2
; RV64I-NEXT: neg a1, a1
; RV64I-NEXT: ori a1, a1, 64
; RV64I-NEXT: sll a0, a0, a1
; RV64I-NEXT: or a0, a2, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_64_mask_or_128_or_64:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: srl a3, a1, a2
; RV32ZBB-NEXT: neg a1, a2
; RV32ZBB-NEXT: sll a1, a0, a1
; RV32ZBB-NEXT: mv a0, a3
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_64_mask_or_128_or_64:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: ror a0, a0, a1
; RV64ZBB-NEXT: ret
%a = or i64 %y, 128
%b = lshr i64 %x, %a
%c = sub i64 0, %y
%d = or i64 %c, 64
%e = shl i64 %x, %d
%f = or i64 %b, %e
ret i64 %f
}
; Test that we're able to remove a mask on the rotate amount that has more than
; one use.
define signext i32 @rotl_32_mask_shared(i32 signext %a, i32 signext %b, i32 signext %amt) nounwind {
; RV32I-LABEL: rotl_32_mask_shared:
; RV32I: # %bb.0:
; RV32I-NEXT: sll a3, a0, a2
; RV32I-NEXT: neg a4, a2
; RV32I-NEXT: srl a0, a0, a4
; RV32I-NEXT: or a0, a3, a0
; RV32I-NEXT: sll a1, a1, a2
; RV32I-NEXT: add a0, a0, a1
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_32_mask_shared:
; RV64I: # %bb.0:
; RV64I-NEXT: sllw a3, a0, a2
; RV64I-NEXT: negw a4, a2
; RV64I-NEXT: srlw a0, a0, a4
; RV64I-NEXT: or a0, a3, a0
; RV64I-NEXT: sllw a1, a1, a2
; RV64I-NEXT: addw a0, a0, a1
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_32_mask_shared:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: rol a0, a0, a2
; RV32ZBB-NEXT: sll a1, a1, a2
; RV32ZBB-NEXT: add a0, a0, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_32_mask_shared:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rolw a0, a0, a2
; RV64ZBB-NEXT: sllw a1, a1, a2
; RV64ZBB-NEXT: addw a0, a0, a1
; RV64ZBB-NEXT: ret
%maskedamt = and i32 %amt, 31
%1 = tail call i32 @llvm.fshl.i32(i32 %a, i32 %a, i32 %maskedamt)
%2 = shl i32 %b, %maskedamt
%3 = add i32 %1, %2
ret i32 %3
}
declare i32 @llvm.fshl.i32(i32, i32, i32)
define signext i64 @rotl_64_mask_shared(i64 signext %a, i64 signext %b, i64 signext %amt) nounwind {
; RV32I-LABEL: rotl_64_mask_shared:
; RV32I: # %bb.0:
; RV32I-NEXT: slli a5, a4, 26
; RV32I-NEXT: srli a5, a5, 31
; RV32I-NEXT: mv a7, a0
; RV32I-NEXT: bnez a5, .LBB17_2
; RV32I-NEXT: # %bb.1:
; RV32I-NEXT: mv a7, a1
; RV32I-NEXT: .LBB17_2:
; RV32I-NEXT: andi a6, a4, 63
; RV32I-NEXT: sll t0, a7, a4
; RV32I-NEXT: bnez a5, .LBB17_4
; RV32I-NEXT: # %bb.3:
; RV32I-NEXT: mv a1, a0
; RV32I-NEXT: .LBB17_4:
; RV32I-NEXT: srli a0, a1, 1
; RV32I-NEXT: not t1, a4
; RV32I-NEXT: srl a0, a0, t1
; RV32I-NEXT: or a5, t0, a0
; RV32I-NEXT: sll a1, a1, a4
; RV32I-NEXT: srli a0, a7, 1
; RV32I-NEXT: srl a7, a0, t1
; RV32I-NEXT: addi a0, a6, -32
; RV32I-NEXT: or a1, a1, a7
; RV32I-NEXT: bltz a0, .LBB17_6
; RV32I-NEXT: # %bb.5:
; RV32I-NEXT: sll a3, a2, a0
; RV32I-NEXT: j .LBB17_7
; RV32I-NEXT: .LBB17_6:
; RV32I-NEXT: sll a3, a3, a4
; RV32I-NEXT: srli a7, a2, 1
; RV32I-NEXT: xori a6, a6, 31
; RV32I-NEXT: srl a6, a7, a6
; RV32I-NEXT: or a3, a3, a6
; RV32I-NEXT: .LBB17_7:
; RV32I-NEXT: sll a2, a2, a4
; RV32I-NEXT: slti a0, a0, 0
; RV32I-NEXT: neg a0, a0
; RV32I-NEXT: and a0, a0, a2
; RV32I-NEXT: add a0, a1, a0
; RV32I-NEXT: sltu a1, a0, a1
; RV32I-NEXT: add a3, a5, a3
; RV32I-NEXT: add a1, a3, a1
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_64_mask_shared:
; RV64I: # %bb.0:
; RV64I-NEXT: sll a3, a0, a2
; RV64I-NEXT: negw a4, a2
; RV64I-NEXT: srl a0, a0, a4
; RV64I-NEXT: or a0, a3, a0
; RV64I-NEXT: sll a1, a1, a2
; RV64I-NEXT: add a0, a0, a1
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_64_mask_shared:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: slli a5, a4, 26
; RV32ZBB-NEXT: srli a5, a5, 31
; RV32ZBB-NEXT: mv a7, a0
; RV32ZBB-NEXT: bnez a5, .LBB17_2
; RV32ZBB-NEXT: # %bb.1:
; RV32ZBB-NEXT: mv a7, a1
; RV32ZBB-NEXT: .LBB17_2:
; RV32ZBB-NEXT: andi a6, a4, 63
; RV32ZBB-NEXT: sll t0, a7, a4
; RV32ZBB-NEXT: bnez a5, .LBB17_4
; RV32ZBB-NEXT: # %bb.3:
; RV32ZBB-NEXT: mv a1, a0
; RV32ZBB-NEXT: .LBB17_4:
; RV32ZBB-NEXT: srli a0, a1, 1
; RV32ZBB-NEXT: not t1, a4
; RV32ZBB-NEXT: srl a0, a0, t1
; RV32ZBB-NEXT: or a5, t0, a0
; RV32ZBB-NEXT: sll a1, a1, a4
; RV32ZBB-NEXT: srli a0, a7, 1
; RV32ZBB-NEXT: srl a7, a0, t1
; RV32ZBB-NEXT: addi a0, a6, -32
; RV32ZBB-NEXT: or a1, a1, a7
; RV32ZBB-NEXT: bltz a0, .LBB17_6
; RV32ZBB-NEXT: # %bb.5:
; RV32ZBB-NEXT: sll a3, a2, a0
; RV32ZBB-NEXT: j .LBB17_7
; RV32ZBB-NEXT: .LBB17_6:
; RV32ZBB-NEXT: sll a3, a3, a4
; RV32ZBB-NEXT: srli a7, a2, 1
; RV32ZBB-NEXT: xori a6, a6, 31
; RV32ZBB-NEXT: srl a6, a7, a6
; RV32ZBB-NEXT: or a3, a3, a6
; RV32ZBB-NEXT: .LBB17_7:
; RV32ZBB-NEXT: sll a2, a2, a4
; RV32ZBB-NEXT: slti a0, a0, 0
; RV32ZBB-NEXT: neg a0, a0
; RV32ZBB-NEXT: and a0, a0, a2
; RV32ZBB-NEXT: add a0, a1, a0
; RV32ZBB-NEXT: sltu a1, a0, a1
; RV32ZBB-NEXT: add a3, a5, a3
; RV32ZBB-NEXT: add a1, a3, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_64_mask_shared:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rol a0, a0, a2
; RV64ZBB-NEXT: sll a1, a1, a2
; RV64ZBB-NEXT: add a0, a0, a1
; RV64ZBB-NEXT: ret
%maskedamt = and i64 %amt, 63
%1 = tail call i64 @llvm.fshl.i64(i64 %a, i64 %a, i64 %maskedamt)
%2 = shl i64 %b, %maskedamt
%3 = add i64 %1, %2
ret i64 %3
}
declare i64 @llvm.fshl.i64(i64, i64, i64)
define signext i32 @rotr_32_mask_shared(i32 signext %a, i32 signext %b, i32 signext %amt) nounwind {
; RV32I-LABEL: rotr_32_mask_shared:
; RV32I: # %bb.0:
; RV32I-NEXT: srl a3, a0, a2
; RV32I-NEXT: neg a4, a2
; RV32I-NEXT: sll a0, a0, a4
; RV32I-NEXT: or a0, a3, a0
; RV32I-NEXT: sll a1, a1, a2
; RV32I-NEXT: add a0, a0, a1
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_32_mask_shared:
; RV64I: # %bb.0:
; RV64I-NEXT: srlw a3, a0, a2
; RV64I-NEXT: negw a4, a2
; RV64I-NEXT: sllw a0, a0, a4
; RV64I-NEXT: or a0, a3, a0
; RV64I-NEXT: sllw a1, a1, a2
; RV64I-NEXT: addw a0, a0, a1
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_32_mask_shared:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: ror a0, a0, a2
; RV32ZBB-NEXT: sll a1, a1, a2
; RV32ZBB-NEXT: add a0, a0, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_32_mask_shared:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rorw a0, a0, a2
; RV64ZBB-NEXT: sllw a1, a1, a2
; RV64ZBB-NEXT: addw a0, a0, a1
; RV64ZBB-NEXT: ret
%maskedamt = and i32 %amt, 31
%1 = tail call i32 @llvm.fshr.i32(i32 %a, i32 %a, i32 %maskedamt)
%2 = shl i32 %b, %maskedamt
%3 = add i32 %1, %2
ret i32 %3
}
declare i32 @llvm.fshr.i32(i32, i32, i32)
define signext i64 @rotr_64_mask_shared(i64 signext %a, i64 signext %b, i64 signext %amt) nounwind {
; RV32I-LABEL: rotr_64_mask_shared:
; RV32I: # %bb.0:
; RV32I-NEXT: andi a7, a4, 32
; RV32I-NEXT: mv a6, a1
; RV32I-NEXT: beqz a7, .LBB19_2
; RV32I-NEXT: # %bb.1:
; RV32I-NEXT: mv a6, a0
; RV32I-NEXT: .LBB19_2:
; RV32I-NEXT: andi a5, a4, 63
; RV32I-NEXT: srl t0, a6, a4
; RV32I-NEXT: beqz a7, .LBB19_4
; RV32I-NEXT: # %bb.3:
; RV32I-NEXT: mv a0, a1
; RV32I-NEXT: .LBB19_4:
; RV32I-NEXT: slli a1, a0, 1
; RV32I-NEXT: not a7, a4
; RV32I-NEXT: sll a1, a1, a7
; RV32I-NEXT: or a1, a1, t0
; RV32I-NEXT: srl t0, a0, a4
; RV32I-NEXT: slli a6, a6, 1
; RV32I-NEXT: sll a6, a6, a7
; RV32I-NEXT: addi a0, a5, -32
; RV32I-NEXT: or a6, a6, t0
; RV32I-NEXT: bltz a0, .LBB19_6
; RV32I-NEXT: # %bb.5:
; RV32I-NEXT: sll a3, a2, a0
; RV32I-NEXT: j .LBB19_7
; RV32I-NEXT: .LBB19_6:
; RV32I-NEXT: sll a3, a3, a4
; RV32I-NEXT: srli a7, a2, 1
; RV32I-NEXT: xori a5, a5, 31
; RV32I-NEXT: srl a5, a7, a5
; RV32I-NEXT: or a3, a3, a5
; RV32I-NEXT: .LBB19_7:
; RV32I-NEXT: sll a2, a2, a4
; RV32I-NEXT: slti a0, a0, 0
; RV32I-NEXT: neg a0, a0
; RV32I-NEXT: and a0, a0, a2
; RV32I-NEXT: add a0, a6, a0
; RV32I-NEXT: sltu a2, a0, a6
; RV32I-NEXT: add a1, a1, a3
; RV32I-NEXT: add a1, a1, a2
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_64_mask_shared:
; RV64I: # %bb.0:
; RV64I-NEXT: srl a3, a0, a2
; RV64I-NEXT: negw a4, a2
; RV64I-NEXT: sll a0, a0, a4
; RV64I-NEXT: or a0, a3, a0
; RV64I-NEXT: sll a1, a1, a2
; RV64I-NEXT: add a0, a0, a1
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_64_mask_shared:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: andi a7, a4, 32
; RV32ZBB-NEXT: mv a6, a1
; RV32ZBB-NEXT: beqz a7, .LBB19_2
; RV32ZBB-NEXT: # %bb.1:
; RV32ZBB-NEXT: mv a6, a0
; RV32ZBB-NEXT: .LBB19_2:
; RV32ZBB-NEXT: andi a5, a4, 63
; RV32ZBB-NEXT: srl t0, a6, a4
; RV32ZBB-NEXT: beqz a7, .LBB19_4
; RV32ZBB-NEXT: # %bb.3:
; RV32ZBB-NEXT: mv a0, a1
; RV32ZBB-NEXT: .LBB19_4:
; RV32ZBB-NEXT: slli a1, a0, 1
; RV32ZBB-NEXT: not a7, a4
; RV32ZBB-NEXT: sll a1, a1, a7
; RV32ZBB-NEXT: or a1, a1, t0
; RV32ZBB-NEXT: srl t0, a0, a4
; RV32ZBB-NEXT: slli a6, a6, 1
; RV32ZBB-NEXT: sll a6, a6, a7
; RV32ZBB-NEXT: addi a0, a5, -32
; RV32ZBB-NEXT: or a6, a6, t0
; RV32ZBB-NEXT: bltz a0, .LBB19_6
; RV32ZBB-NEXT: # %bb.5:
; RV32ZBB-NEXT: sll a3, a2, a0
; RV32ZBB-NEXT: j .LBB19_7
; RV32ZBB-NEXT: .LBB19_6:
; RV32ZBB-NEXT: sll a3, a3, a4
; RV32ZBB-NEXT: srli a7, a2, 1
; RV32ZBB-NEXT: xori a5, a5, 31
; RV32ZBB-NEXT: srl a5, a7, a5
; RV32ZBB-NEXT: or a3, a3, a5
; RV32ZBB-NEXT: .LBB19_7:
; RV32ZBB-NEXT: sll a2, a2, a4
; RV32ZBB-NEXT: slti a0, a0, 0
; RV32ZBB-NEXT: neg a0, a0
; RV32ZBB-NEXT: and a0, a0, a2
; RV32ZBB-NEXT: add a0, a6, a0
; RV32ZBB-NEXT: sltu a2, a0, a6
; RV32ZBB-NEXT: add a1, a1, a3
; RV32ZBB-NEXT: add a1, a1, a2
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_64_mask_shared:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: ror a0, a0, a2
; RV64ZBB-NEXT: sll a1, a1, a2
; RV64ZBB-NEXT: add a0, a0, a1
; RV64ZBB-NEXT: ret
%maskedamt = and i64 %amt, 63
%1 = tail call i64 @llvm.fshr.i64(i64 %a, i64 %a, i64 %maskedamt)
%2 = shl i64 %b, %maskedamt
%3 = add i64 %1, %2
ret i64 %3
}
declare i64 @llvm.fshr.i64(i64, i64, i64)
define signext i32 @rotl_32_mask_multiple(i32 signext %a, i32 signext %b, i32 signext %amt) nounwind {
; RV32I-LABEL: rotl_32_mask_multiple:
; RV32I: # %bb.0:
; RV32I-NEXT: sll a3, a0, a2
; RV32I-NEXT: neg a4, a2
; RV32I-NEXT: srl a0, a0, a4
; RV32I-NEXT: or a0, a3, a0
; RV32I-NEXT: sll a2, a1, a2
; RV32I-NEXT: srl a1, a1, a4
; RV32I-NEXT: or a1, a2, a1
; RV32I-NEXT: add a0, a0, a1
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_32_mask_multiple:
; RV64I: # %bb.0:
; RV64I-NEXT: sllw a3, a0, a2
; RV64I-NEXT: negw a4, a2
; RV64I-NEXT: srlw a0, a0, a4
; RV64I-NEXT: or a0, a3, a0
; RV64I-NEXT: sllw a2, a1, a2
; RV64I-NEXT: srlw a1, a1, a4
; RV64I-NEXT: or a1, a2, a1
; RV64I-NEXT: addw a0, a0, a1
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_32_mask_multiple:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: rol a0, a0, a2
; RV32ZBB-NEXT: rol a1, a1, a2
; RV32ZBB-NEXT: add a0, a0, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_32_mask_multiple:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rolw a0, a0, a2
; RV64ZBB-NEXT: rolw a1, a1, a2
; RV64ZBB-NEXT: addw a0, a0, a1
; RV64ZBB-NEXT: ret
%maskedamt = and i32 %amt, 31
%1 = tail call i32 @llvm.fshl.i32(i32 %a, i32 %a, i32 %maskedamt)
%2 = tail call i32 @llvm.fshl.i32(i32 %b, i32 %b, i32 %maskedamt)
%3 = add i32 %1, %2
ret i32 %3
}
define i64 @rotl_64_mask_multiple(i64 %a, i64 %b, i64 %amt) nounwind {
; RV32I-LABEL: rotl_64_mask_multiple:
; RV32I: # %bb.0:
; RV32I-NEXT: slli a5, a4, 26
; RV32I-NEXT: srli a5, a5, 31
; RV32I-NEXT: mv a6, a1
; RV32I-NEXT: bnez a5, .LBB21_2
; RV32I-NEXT: # %bb.1:
; RV32I-NEXT: mv a6, a0
; RV32I-NEXT: .LBB21_2:
; RV32I-NEXT: bnez a5, .LBB21_4
; RV32I-NEXT: # %bb.3:
; RV32I-NEXT: mv a0, a1
; RV32I-NEXT: .LBB21_4:
; RV32I-NEXT: sll a7, a6, a4
; RV32I-NEXT: srli t0, a0, 1
; RV32I-NEXT: not a1, a4
; RV32I-NEXT: srl t0, t0, a1
; RV32I-NEXT: sll t1, a0, a4
; RV32I-NEXT: srli a0, a6, 1
; RV32I-NEXT: srl t2, a0, a1
; RV32I-NEXT: mv a0, a3
; RV32I-NEXT: bnez a5, .LBB21_6
; RV32I-NEXT: # %bb.5:
; RV32I-NEXT: mv a0, a2
; RV32I-NEXT: .LBB21_6:
; RV32I-NEXT: or a6, a7, t0
; RV32I-NEXT: or a7, t1, t2
; RV32I-NEXT: sll t0, a0, a4
; RV32I-NEXT: bnez a5, .LBB21_8
; RV32I-NEXT: # %bb.7:
; RV32I-NEXT: mv a2, a3
; RV32I-NEXT: .LBB21_8:
; RV32I-NEXT: srli a3, a2, 1
; RV32I-NEXT: srl a3, a3, a1
; RV32I-NEXT: or a3, t0, a3
; RV32I-NEXT: sll a2, a2, a4
; RV32I-NEXT: srli a0, a0, 1
; RV32I-NEXT: srl a0, a0, a1
; RV32I-NEXT: or a0, a2, a0
; RV32I-NEXT: add a1, a7, a0
; RV32I-NEXT: add a0, a6, a3
; RV32I-NEXT: sltu a2, a0, a6
; RV32I-NEXT: add a1, a1, a2
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_64_mask_multiple:
; RV64I: # %bb.0:
; RV64I-NEXT: sll a3, a0, a2
; RV64I-NEXT: negw a4, a2
; RV64I-NEXT: srl a0, a0, a4
; RV64I-NEXT: or a0, a3, a0
; RV64I-NEXT: sll a2, a1, a2
; RV64I-NEXT: srl a1, a1, a4
; RV64I-NEXT: or a1, a2, a1
; RV64I-NEXT: add a0, a0, a1
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_64_mask_multiple:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: slli a5, a4, 26
; RV32ZBB-NEXT: srli a5, a5, 31
; RV32ZBB-NEXT: mv a6, a1
; RV32ZBB-NEXT: bnez a5, .LBB21_2
; RV32ZBB-NEXT: # %bb.1:
; RV32ZBB-NEXT: mv a6, a0
; RV32ZBB-NEXT: .LBB21_2:
; RV32ZBB-NEXT: bnez a5, .LBB21_4
; RV32ZBB-NEXT: # %bb.3:
; RV32ZBB-NEXT: mv a0, a1
; RV32ZBB-NEXT: .LBB21_4:
; RV32ZBB-NEXT: sll a7, a6, a4
; RV32ZBB-NEXT: srli t0, a0, 1
; RV32ZBB-NEXT: not a1, a4
; RV32ZBB-NEXT: srl t0, t0, a1
; RV32ZBB-NEXT: sll t1, a0, a4
; RV32ZBB-NEXT: srli a0, a6, 1
; RV32ZBB-NEXT: srl t2, a0, a1
; RV32ZBB-NEXT: mv a0, a3
; RV32ZBB-NEXT: bnez a5, .LBB21_6
; RV32ZBB-NEXT: # %bb.5:
; RV32ZBB-NEXT: mv a0, a2
; RV32ZBB-NEXT: .LBB21_6:
; RV32ZBB-NEXT: or a6, a7, t0
; RV32ZBB-NEXT: or a7, t1, t2
; RV32ZBB-NEXT: sll t0, a0, a4
; RV32ZBB-NEXT: bnez a5, .LBB21_8
; RV32ZBB-NEXT: # %bb.7:
; RV32ZBB-NEXT: mv a2, a3
; RV32ZBB-NEXT: .LBB21_8:
; RV32ZBB-NEXT: srli a3, a2, 1
; RV32ZBB-NEXT: srl a3, a3, a1
; RV32ZBB-NEXT: or a3, t0, a3
; RV32ZBB-NEXT: sll a2, a2, a4
; RV32ZBB-NEXT: srli a0, a0, 1
; RV32ZBB-NEXT: srl a0, a0, a1
; RV32ZBB-NEXT: or a0, a2, a0
; RV32ZBB-NEXT: add a1, a7, a0
; RV32ZBB-NEXT: add a0, a6, a3
; RV32ZBB-NEXT: sltu a2, a0, a6
; RV32ZBB-NEXT: add a1, a1, a2
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_64_mask_multiple:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rol a0, a0, a2
; RV64ZBB-NEXT: rol a1, a1, a2
; RV64ZBB-NEXT: add a0, a0, a1
; RV64ZBB-NEXT: ret
%maskedamt = and i64 %amt, 63
%1 = tail call i64 @llvm.fshl.i64(i64 %a, i64 %a, i64 %maskedamt)
%2 = tail call i64 @llvm.fshl.i64(i64 %b, i64 %b, i64 %maskedamt)
%3 = add i64 %1, %2
ret i64 %3
}
define signext i32 @rotr_32_mask_multiple(i32 signext %a, i32 signext %b, i32 signext %amt) nounwind {
; RV32I-LABEL: rotr_32_mask_multiple:
; RV32I: # %bb.0:
; RV32I-NEXT: srl a3, a0, a2
; RV32I-NEXT: neg a4, a2
; RV32I-NEXT: sll a0, a0, a4
; RV32I-NEXT: or a0, a3, a0
; RV32I-NEXT: srl a2, a1, a2
; RV32I-NEXT: sll a1, a1, a4
; RV32I-NEXT: or a1, a2, a1
; RV32I-NEXT: add a0, a0, a1
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_32_mask_multiple:
; RV64I: # %bb.0:
; RV64I-NEXT: srlw a3, a0, a2
; RV64I-NEXT: negw a4, a2
; RV64I-NEXT: sllw a0, a0, a4
; RV64I-NEXT: or a0, a3, a0
; RV64I-NEXT: srlw a2, a1, a2
; RV64I-NEXT: sllw a1, a1, a4
; RV64I-NEXT: or a1, a2, a1
; RV64I-NEXT: addw a0, a0, a1
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_32_mask_multiple:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: ror a0, a0, a2
; RV32ZBB-NEXT: ror a1, a1, a2
; RV32ZBB-NEXT: add a0, a0, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_32_mask_multiple:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rorw a0, a0, a2
; RV64ZBB-NEXT: rorw a1, a1, a2
; RV64ZBB-NEXT: addw a0, a0, a1
; RV64ZBB-NEXT: ret
%maskedamt = and i32 %amt, 31
%1 = tail call i32 @llvm.fshr.i32(i32 %a, i32 %a, i32 %maskedamt)
%2 = tail call i32 @llvm.fshr.i32(i32 %b, i32 %b, i32 %maskedamt)
%3 = add i32 %1, %2
ret i32 %3
}
define i64 @rotr_64_mask_multiple(i64 %a, i64 %b, i64 %amt) nounwind {
; RV32I-LABEL: rotr_64_mask_multiple:
; RV32I: # %bb.0:
; RV32I-NEXT: andi a5, a4, 32
; RV32I-NEXT: mv a6, a0
; RV32I-NEXT: beqz a5, .LBB23_2
; RV32I-NEXT: # %bb.1:
; RV32I-NEXT: mv a6, a1
; RV32I-NEXT: .LBB23_2:
; RV32I-NEXT: beqz a5, .LBB23_4
; RV32I-NEXT: # %bb.3:
; RV32I-NEXT: mv a1, a0
; RV32I-NEXT: .LBB23_4:
; RV32I-NEXT: srl a7, a6, a4
; RV32I-NEXT: slli t0, a1, 1
; RV32I-NEXT: not a0, a4
; RV32I-NEXT: sll t0, t0, a0
; RV32I-NEXT: srl t1, a1, a4
; RV32I-NEXT: slli a6, a6, 1
; RV32I-NEXT: sll t2, a6, a0
; RV32I-NEXT: mv a6, a2
; RV32I-NEXT: beqz a5, .LBB23_6
; RV32I-NEXT: # %bb.5:
; RV32I-NEXT: mv a6, a3
; RV32I-NEXT: .LBB23_6:
; RV32I-NEXT: or a1, t0, a7
; RV32I-NEXT: or a7, t2, t1
; RV32I-NEXT: srl t0, a6, a4
; RV32I-NEXT: beqz a5, .LBB23_8
; RV32I-NEXT: # %bb.7:
; RV32I-NEXT: mv a3, a2
; RV32I-NEXT: .LBB23_8:
; RV32I-NEXT: slli a2, a3, 1
; RV32I-NEXT: sll a2, a2, a0
; RV32I-NEXT: or a2, a2, t0
; RV32I-NEXT: srl a3, a3, a4
; RV32I-NEXT: slli a6, a6, 1
; RV32I-NEXT: sll a0, a6, a0
; RV32I-NEXT: or a0, a0, a3
; RV32I-NEXT: add a7, a7, a0
; RV32I-NEXT: add a0, a1, a2
; RV32I-NEXT: sltu a1, a0, a1
; RV32I-NEXT: add a1, a7, a1
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_64_mask_multiple:
; RV64I: # %bb.0:
; RV64I-NEXT: srl a3, a0, a2
; RV64I-NEXT: negw a4, a2
; RV64I-NEXT: sll a0, a0, a4
; RV64I-NEXT: or a0, a3, a0
; RV64I-NEXT: srl a2, a1, a2
; RV64I-NEXT: sll a1, a1, a4
; RV64I-NEXT: or a1, a2, a1
; RV64I-NEXT: add a0, a0, a1
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_64_mask_multiple:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: andi a5, a4, 32
; RV32ZBB-NEXT: mv a6, a0
; RV32ZBB-NEXT: beqz a5, .LBB23_2
; RV32ZBB-NEXT: # %bb.1:
; RV32ZBB-NEXT: mv a6, a1
; RV32ZBB-NEXT: .LBB23_2:
; RV32ZBB-NEXT: beqz a5, .LBB23_4
; RV32ZBB-NEXT: # %bb.3:
; RV32ZBB-NEXT: mv a1, a0
; RV32ZBB-NEXT: .LBB23_4:
; RV32ZBB-NEXT: srl a7, a6, a4
; RV32ZBB-NEXT: slli t0, a1, 1
; RV32ZBB-NEXT: not a0, a4
; RV32ZBB-NEXT: sll t0, t0, a0
; RV32ZBB-NEXT: srl t1, a1, a4
; RV32ZBB-NEXT: slli a6, a6, 1
; RV32ZBB-NEXT: sll t2, a6, a0
; RV32ZBB-NEXT: mv a6, a2
; RV32ZBB-NEXT: beqz a5, .LBB23_6
; RV32ZBB-NEXT: # %bb.5:
; RV32ZBB-NEXT: mv a6, a3
; RV32ZBB-NEXT: .LBB23_6:
; RV32ZBB-NEXT: or a1, t0, a7
; RV32ZBB-NEXT: or a7, t2, t1
; RV32ZBB-NEXT: srl t0, a6, a4
; RV32ZBB-NEXT: beqz a5, .LBB23_8
; RV32ZBB-NEXT: # %bb.7:
; RV32ZBB-NEXT: mv a3, a2
; RV32ZBB-NEXT: .LBB23_8:
; RV32ZBB-NEXT: slli a2, a3, 1
; RV32ZBB-NEXT: sll a2, a2, a0
; RV32ZBB-NEXT: or a2, a2, t0
; RV32ZBB-NEXT: srl a3, a3, a4
; RV32ZBB-NEXT: slli a6, a6, 1
; RV32ZBB-NEXT: sll a0, a6, a0
; RV32ZBB-NEXT: or a0, a0, a3
; RV32ZBB-NEXT: add a7, a7, a0
; RV32ZBB-NEXT: add a0, a1, a2
; RV32ZBB-NEXT: sltu a1, a0, a1
; RV32ZBB-NEXT: add a1, a7, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_64_mask_multiple:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: ror a0, a0, a2
; RV64ZBB-NEXT: ror a1, a1, a2
; RV64ZBB-NEXT: add a0, a0, a1
; RV64ZBB-NEXT: ret
%maskedamt = and i64 %amt, 63
%1 = tail call i64 @llvm.fshr.i64(i64 %a, i64 %a, i64 %maskedamt)
%2 = tail call i64 @llvm.fshr.i64(i64 %b, i64 %b, i64 %maskedamt)
%3 = add i64 %1, %2
ret i64 %3
}
define i64 @rotl_64_zext(i64 %x, i32 %y) nounwind {
; RV32I-LABEL: rotl_64_zext:
; RV32I: # %bb.0:
; RV32I-NEXT: neg a4, a2
; RV32I-NEXT: sll a5, a0, a2
; RV32I-NEXT: addi a3, a2, -32
; RV32I-NEXT: slti a6, a3, 0
; RV32I-NEXT: neg a6, a6
; RV32I-NEXT: bltz a3, .LBB24_2
; RV32I-NEXT: # %bb.1:
; RV32I-NEXT: sll a3, a0, a3
; RV32I-NEXT: j .LBB24_3
; RV32I-NEXT: .LBB24_2:
; RV32I-NEXT: sll a3, a1, a2
; RV32I-NEXT: xori a7, a2, 31
; RV32I-NEXT: srli t0, a0, 1
; RV32I-NEXT: srl a7, t0, a7
; RV32I-NEXT: or a3, a3, a7
; RV32I-NEXT: .LBB24_3:
; RV32I-NEXT: and a5, a6, a5
; RV32I-NEXT: li a6, 32
; RV32I-NEXT: sub a7, a6, a2
; RV32I-NEXT: srl a6, a1, a4
; RV32I-NEXT: bltz a7, .LBB24_5
; RV32I-NEXT: # %bb.4:
; RV32I-NEXT: mv a0, a6
; RV32I-NEXT: j .LBB24_6
; RV32I-NEXT: .LBB24_5:
; RV32I-NEXT: li t0, 64
; RV32I-NEXT: sub a2, t0, a2
; RV32I-NEXT: srl a0, a0, a4
; RV32I-NEXT: xori a2, a2, 31
; RV32I-NEXT: slli a1, a1, 1
; RV32I-NEXT: sll a1, a1, a2
; RV32I-NEXT: or a0, a0, a1
; RV32I-NEXT: .LBB24_6:
; RV32I-NEXT: slti a1, a7, 0
; RV32I-NEXT: neg a1, a1
; RV32I-NEXT: and a1, a1, a6
; RV32I-NEXT: or a1, a3, a1
; RV32I-NEXT: or a0, a5, a0
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotl_64_zext:
; RV64I: # %bb.0:
; RV64I-NEXT: negw a2, a1
; RV64I-NEXT: sll a1, a0, a1
; RV64I-NEXT: srl a0, a0, a2
; RV64I-NEXT: or a0, a1, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotl_64_zext:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: neg a4, a2
; RV32ZBB-NEXT: sll a5, a0, a2
; RV32ZBB-NEXT: addi a3, a2, -32
; RV32ZBB-NEXT: slti a6, a3, 0
; RV32ZBB-NEXT: neg a6, a6
; RV32ZBB-NEXT: bltz a3, .LBB24_2
; RV32ZBB-NEXT: # %bb.1:
; RV32ZBB-NEXT: sll a3, a0, a3
; RV32ZBB-NEXT: j .LBB24_3
; RV32ZBB-NEXT: .LBB24_2:
; RV32ZBB-NEXT: sll a3, a1, a2
; RV32ZBB-NEXT: xori a7, a2, 31
; RV32ZBB-NEXT: srli t0, a0, 1
; RV32ZBB-NEXT: srl a7, t0, a7
; RV32ZBB-NEXT: or a3, a3, a7
; RV32ZBB-NEXT: .LBB24_3:
; RV32ZBB-NEXT: and a5, a6, a5
; RV32ZBB-NEXT: li a6, 32
; RV32ZBB-NEXT: sub a7, a6, a2
; RV32ZBB-NEXT: srl a6, a1, a4
; RV32ZBB-NEXT: bltz a7, .LBB24_5
; RV32ZBB-NEXT: # %bb.4:
; RV32ZBB-NEXT: mv a0, a6
; RV32ZBB-NEXT: j .LBB24_6
; RV32ZBB-NEXT: .LBB24_5:
; RV32ZBB-NEXT: li t0, 64
; RV32ZBB-NEXT: sub a2, t0, a2
; RV32ZBB-NEXT: srl a0, a0, a4
; RV32ZBB-NEXT: xori a2, a2, 31
; RV32ZBB-NEXT: slli a1, a1, 1
; RV32ZBB-NEXT: sll a1, a1, a2
; RV32ZBB-NEXT: or a0, a0, a1
; RV32ZBB-NEXT: .LBB24_6:
; RV32ZBB-NEXT: slti a1, a7, 0
; RV32ZBB-NEXT: neg a1, a1
; RV32ZBB-NEXT: and a1, a1, a6
; RV32ZBB-NEXT: or a1, a3, a1
; RV32ZBB-NEXT: or a0, a5, a0
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotl_64_zext:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: rol a0, a0, a1
; RV64ZBB-NEXT: ret
%z = sub i32 64, %y
%zext = zext i32 %z to i64
%zexty = zext i32 %y to i64
%b = shl i64 %x, %zexty
%c = lshr i64 %x, %zext
%d = or i64 %b, %c
ret i64 %d
}
define i64 @rotr_64_zext(i64 %x, i32 %y) nounwind {
; RV32I-LABEL: rotr_64_zext:
; RV32I: # %bb.0:
; RV32I-NEXT: neg a4, a2
; RV32I-NEXT: srl a5, a1, a2
; RV32I-NEXT: addi a3, a2, -32
; RV32I-NEXT: slti a6, a3, 0
; RV32I-NEXT: neg a6, a6
; RV32I-NEXT: bltz a3, .LBB25_2
; RV32I-NEXT: # %bb.1:
; RV32I-NEXT: srl a3, a1, a3
; RV32I-NEXT: j .LBB25_3
; RV32I-NEXT: .LBB25_2:
; RV32I-NEXT: srl a3, a0, a2
; RV32I-NEXT: xori a7, a2, 31
; RV32I-NEXT: slli t0, a1, 1
; RV32I-NEXT: sll a7, t0, a7
; RV32I-NEXT: or a3, a3, a7
; RV32I-NEXT: .LBB25_3:
; RV32I-NEXT: and a5, a6, a5
; RV32I-NEXT: li a6, 32
; RV32I-NEXT: sub a7, a6, a2
; RV32I-NEXT: sll a6, a0, a4
; RV32I-NEXT: bltz a7, .LBB25_5
; RV32I-NEXT: # %bb.4:
; RV32I-NEXT: mv a1, a6
; RV32I-NEXT: j .LBB25_6
; RV32I-NEXT: .LBB25_5:
; RV32I-NEXT: li t0, 64
; RV32I-NEXT: sub a2, t0, a2
; RV32I-NEXT: sll a1, a1, a4
; RV32I-NEXT: xori a2, a2, 31
; RV32I-NEXT: srli a0, a0, 1
; RV32I-NEXT: srl a0, a0, a2
; RV32I-NEXT: or a1, a1, a0
; RV32I-NEXT: .LBB25_6:
; RV32I-NEXT: slti a0, a7, 0
; RV32I-NEXT: neg a0, a0
; RV32I-NEXT: and a0, a0, a6
; RV32I-NEXT: or a0, a3, a0
; RV32I-NEXT: or a1, a5, a1
; RV32I-NEXT: ret
;
; RV64I-LABEL: rotr_64_zext:
; RV64I: # %bb.0:
; RV64I-NEXT: negw a2, a1
; RV64I-NEXT: srl a1, a0, a1
; RV64I-NEXT: sll a0, a0, a2
; RV64I-NEXT: or a0, a1, a0
; RV64I-NEXT: ret
;
; RV32ZBB-LABEL: rotr_64_zext:
; RV32ZBB: # %bb.0:
; RV32ZBB-NEXT: neg a4, a2
; RV32ZBB-NEXT: srl a5, a1, a2
; RV32ZBB-NEXT: addi a3, a2, -32
; RV32ZBB-NEXT: slti a6, a3, 0
; RV32ZBB-NEXT: neg a6, a6
; RV32ZBB-NEXT: bltz a3, .LBB25_2
; RV32ZBB-NEXT: # %bb.1:
; RV32ZBB-NEXT: srl a3, a1, a3
; RV32ZBB-NEXT: j .LBB25_3
; RV32ZBB-NEXT: .LBB25_2:
; RV32ZBB-NEXT: srl a3, a0, a2
; RV32ZBB-NEXT: xori a7, a2, 31
; RV32ZBB-NEXT: slli t0, a1, 1
; RV32ZBB-NEXT: sll a7, t0, a7
; RV32ZBB-NEXT: or a3, a3, a7
; RV32ZBB-NEXT: .LBB25_3:
; RV32ZBB-NEXT: and a5, a6, a5
; RV32ZBB-NEXT: li a6, 32
; RV32ZBB-NEXT: sub a7, a6, a2
; RV32ZBB-NEXT: sll a6, a0, a4
; RV32ZBB-NEXT: bltz a7, .LBB25_5
; RV32ZBB-NEXT: # %bb.4:
; RV32ZBB-NEXT: mv a1, a6
; RV32ZBB-NEXT: j .LBB25_6
; RV32ZBB-NEXT: .LBB25_5:
; RV32ZBB-NEXT: li t0, 64
; RV32ZBB-NEXT: sub a2, t0, a2
; RV32ZBB-NEXT: sll a1, a1, a4
; RV32ZBB-NEXT: xori a2, a2, 31
; RV32ZBB-NEXT: srli a0, a0, 1
; RV32ZBB-NEXT: srl a0, a0, a2
; RV32ZBB-NEXT: or a1, a1, a0
; RV32ZBB-NEXT: .LBB25_6:
; RV32ZBB-NEXT: slti a0, a7, 0
; RV32ZBB-NEXT: neg a0, a0
; RV32ZBB-NEXT: and a0, a0, a6
; RV32ZBB-NEXT: or a0, a3, a0
; RV32ZBB-NEXT: or a1, a5, a1
; RV32ZBB-NEXT: ret
;
; RV64ZBB-LABEL: rotr_64_zext:
; RV64ZBB: # %bb.0:
; RV64ZBB-NEXT: ror a0, a0, a1
; RV64ZBB-NEXT: ret
%z = sub i32 64, %y
%zext = zext i32 %z to i64
%zexty = zext i32 %y to i64
%b = lshr i64 %x, %zexty
%c = shl i64 %x, %zext
%d = or i64 %b, %c
ret i64 %d
}
Reference in New Issue View Git Blame Copy Permalink