| ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py |
| ; RUN: llc -mtriple=riscv64 -verify-machineinstrs < %s \ |
| ; RUN: | FileCheck %s -check-prefix=RV64I |
| ; RUN: llc -mtriple=riscv64 -mattr=+zbb -verify-machineinstrs < %s \ |
| ; RUN: | FileCheck %s -check-prefix=RV64ZBB |
| |
| declare i32 @llvm.ctlz.i32(i32, i1) |
| |
| define signext i32 @ctlz_i32(i32 signext %a) nounwind { |
| ; RV64I-LABEL: ctlz_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: beqz a0, .LBB0_2 |
| ; RV64I-NEXT: # %bb.1: # %cond.false |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: srliw a1, a0, 1 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 2 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 4 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 8 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 16 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: not a0, a0 |
| ; RV64I-NEXT: srli a1, a0, 1 |
| ; RV64I-NEXT: lui a2, 349525 |
| ; RV64I-NEXT: addiw a2, a2, 1365 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: sub a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 209715 |
| ; RV64I-NEXT: addiw a1, a1, 819 |
| ; RV64I-NEXT: and a2, a0, a1 |
| ; RV64I-NEXT: srli a0, a0, 2 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: add a0, a2, a0 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 61681 |
| ; RV64I-NEXT: addiw a1, a1, -241 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 4112 |
| ; RV64I-NEXT: addiw a1, a1, 257 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw a0, a0, 24 |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; RV64I-NEXT: .LBB0_2: |
| ; RV64I-NEXT: li a0, 32 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctlz_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: clzw a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i32 @llvm.ctlz.i32(i32 %a, i1 false) |
| ret i32 %1 |
| } |
| |
| define signext i32 @log2_i32(i32 signext %a) nounwind { |
| ; RV64I-LABEL: log2_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: beqz a0, .LBB1_2 |
| ; RV64I-NEXT: # %bb.1: # %cond.false |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: srliw a1, a0, 1 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 2 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 4 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 8 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 16 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: not a0, a0 |
| ; RV64I-NEXT: srli a1, a0, 1 |
| ; RV64I-NEXT: lui a2, 349525 |
| ; RV64I-NEXT: addiw a2, a2, 1365 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: sub a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 209715 |
| ; RV64I-NEXT: addiw a1, a1, 819 |
| ; RV64I-NEXT: and a2, a0, a1 |
| ; RV64I-NEXT: srli a0, a0, 2 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: add a0, a2, a0 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 61681 |
| ; RV64I-NEXT: addiw a1, a1, -241 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 4112 |
| ; RV64I-NEXT: addiw a1, a1, 257 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw a0, a0, 24 |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: j .LBB1_3 |
| ; RV64I-NEXT: .LBB1_2: |
| ; RV64I-NEXT: li a0, 32 |
| ; RV64I-NEXT: .LBB1_3: # %cond.end |
| ; RV64I-NEXT: li a1, 31 |
| ; RV64I-NEXT: sub a0, a1, a0 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: log2_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: clzw a0, a0 |
| ; RV64ZBB-NEXT: li a1, 31 |
| ; RV64ZBB-NEXT: sub a0, a1, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i32 @llvm.ctlz.i32(i32 %a, i1 false) |
| %2 = sub i32 31, %1 |
| ret i32 %2 |
| } |
| |
| define signext i32 @log2_ceil_i32(i32 signext %a) nounwind { |
| ; RV64I-LABEL: log2_ceil_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s0, 0(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: addiw a0, a0, -1 |
| ; RV64I-NEXT: li s0, 32 |
| ; RV64I-NEXT: li a1, 32 |
| ; RV64I-NEXT: beqz a0, .LBB2_2 |
| ; RV64I-NEXT: # %bb.1: # %cond.false |
| ; RV64I-NEXT: srliw a1, a0, 1 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 2 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 4 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 8 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 16 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: not a0, a0 |
| ; RV64I-NEXT: srli a1, a0, 1 |
| ; RV64I-NEXT: lui a2, 349525 |
| ; RV64I-NEXT: addiw a2, a2, 1365 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: sub a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 209715 |
| ; RV64I-NEXT: addiw a1, a1, 819 |
| ; RV64I-NEXT: and a2, a0, a1 |
| ; RV64I-NEXT: srli a0, a0, 2 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: add a0, a2, a0 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 61681 |
| ; RV64I-NEXT: addiw a1, a1, -241 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 4112 |
| ; RV64I-NEXT: addiw a1, a1, 257 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw a1, a0, 24 |
| ; RV64I-NEXT: .LBB2_2: # %cond.end |
| ; RV64I-NEXT: sub a0, s0, a1 |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s0, 0(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: log2_ceil_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: addiw a0, a0, -1 |
| ; RV64ZBB-NEXT: clzw a0, a0 |
| ; RV64ZBB-NEXT: li a1, 32 |
| ; RV64ZBB-NEXT: sub a0, a1, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = sub i32 %a, 1 |
| %2 = call i32 @llvm.ctlz.i32(i32 %1, i1 false) |
| %3 = sub i32 32, %2 |
| ret i32 %3 |
| } |
| |
| define signext i32 @findLastSet_i32(i32 signext %a) nounwind { |
| ; RV64I-LABEL: findLastSet_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s0, 0(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: mv s0, a0 |
| ; RV64I-NEXT: srliw a0, a0, 1 |
| ; RV64I-NEXT: or a0, s0, a0 |
| ; RV64I-NEXT: srliw a1, a0, 2 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 4 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 8 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 16 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: not a0, a0 |
| ; RV64I-NEXT: srli a1, a0, 1 |
| ; RV64I-NEXT: lui a2, 349525 |
| ; RV64I-NEXT: addiw a2, a2, 1365 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: sub a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 209715 |
| ; RV64I-NEXT: addiw a1, a1, 819 |
| ; RV64I-NEXT: and a2, a0, a1 |
| ; RV64I-NEXT: srli a0, a0, 2 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: add a0, a2, a0 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 61681 |
| ; RV64I-NEXT: addiw a1, a1, -241 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 4112 |
| ; RV64I-NEXT: addiw a1, a1, 257 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw a0, a0, 24 |
| ; RV64I-NEXT: xori a0, a0, 31 |
| ; RV64I-NEXT: snez a1, s0 |
| ; RV64I-NEXT: addi a1, a1, -1 |
| ; RV64I-NEXT: or a0, a1, a0 |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s0, 0(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: findLastSet_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: clzw a1, a0 |
| ; RV64ZBB-NEXT: xori a1, a1, 31 |
| ; RV64ZBB-NEXT: snez a0, a0 |
| ; RV64ZBB-NEXT: addi a0, a0, -1 |
| ; RV64ZBB-NEXT: or a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i32 @llvm.ctlz.i32(i32 %a, i1 true) |
| %2 = xor i32 31, %1 |
| %3 = icmp eq i32 %a, 0 |
| %4 = select i1 %3, i32 -1, i32 %2 |
| ret i32 %4 |
| } |
| |
| define i32 @ctlz_lshr_i32(i32 signext %a) { |
| ; RV64I-LABEL: ctlz_lshr_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: srliw a0, a0, 1 |
| ; RV64I-NEXT: beqz a0, .LBB4_2 |
| ; RV64I-NEXT: # %bb.1: # %cond.false |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: .cfi_def_cfa_offset 16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: .cfi_offset ra, -8 |
| ; RV64I-NEXT: srliw a1, a0, 1 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 2 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 4 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 8 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srliw a1, a0, 16 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: not a0, a0 |
| ; RV64I-NEXT: srli a1, a0, 1 |
| ; RV64I-NEXT: lui a2, 349525 |
| ; RV64I-NEXT: addiw a2, a2, 1365 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: sub a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 209715 |
| ; RV64I-NEXT: addiw a1, a1, 819 |
| ; RV64I-NEXT: and a2, a0, a1 |
| ; RV64I-NEXT: srli a0, a0, 2 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: add a0, a2, a0 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 61681 |
| ; RV64I-NEXT: addiw a1, a1, -241 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 4112 |
| ; RV64I-NEXT: addiw a1, a1, 257 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw a0, a0, 24 |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; RV64I-NEXT: .LBB4_2: |
| ; RV64I-NEXT: li a0, 32 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctlz_lshr_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: srliw a0, a0, 1 |
| ; RV64ZBB-NEXT: clzw a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = lshr i32 %a, 1 |
| %2 = call i32 @llvm.ctlz.i32(i32 %1, i1 false) |
| ret i32 %2 |
| } |
| |
| declare i64 @llvm.ctlz.i64(i64, i1) |
| |
| define i64 @ctlz_i64(i64 %a) nounwind { |
| ; RV64I-LABEL: ctlz_i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: beqz a0, .LBB5_2 |
| ; RV64I-NEXT: # %bb.1: # %cond.false |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: srli a1, a0, 1 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srli a1, a0, 2 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srli a1, a0, 8 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srli a1, a0, 16 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: srli a1, a0, 32 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: not a0, a0 |
| ; RV64I-NEXT: srli a1, a0, 1 |
| ; RV64I-NEXT: lui a2, 349525 |
| ; RV64I-NEXT: addiw a2, a2, 1365 |
| ; RV64I-NEXT: slli a3, a2, 32 |
| ; RV64I-NEXT: add a2, a2, a3 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: sub a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 209715 |
| ; RV64I-NEXT: addiw a1, a1, 819 |
| ; RV64I-NEXT: slli a2, a1, 32 |
| ; RV64I-NEXT: add a1, a1, a2 |
| ; RV64I-NEXT: and a2, a0, a1 |
| ; RV64I-NEXT: srli a0, a0, 2 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: add a0, a2, a0 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 61681 |
| ; RV64I-NEXT: addiw a1, a1, -241 |
| ; RV64I-NEXT: slli a2, a1, 32 |
| ; RV64I-NEXT: add a1, a1, a2 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 4112 |
| ; RV64I-NEXT: addiw a1, a1, 257 |
| ; RV64I-NEXT: slli a2, a1, 32 |
| ; RV64I-NEXT: add a1, a1, a2 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srli a0, a0, 56 |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; RV64I-NEXT: .LBB5_2: |
| ; RV64I-NEXT: li a0, 64 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctlz_i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: clz a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i64 @llvm.ctlz.i64(i64 %a, i1 false) |
| ret i64 %1 |
| } |
| |
| declare i32 @llvm.cttz.i32(i32, i1) |
| |
| define signext i32 @cttz_i32(i32 signext %a) nounwind { |
| ; RV64I-LABEL: cttz_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: beqz a0, .LBB6_2 |
| ; RV64I-NEXT: # %bb.1: # %cond.false |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: neg a1, a0 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 30667 |
| ; RV64I-NEXT: addiw a1, a1, 1329 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw a0, a0, 27 |
| ; RV64I-NEXT: lui a1, %hi(.LCPI6_0) |
| ; RV64I-NEXT: addi a1, a1, %lo(.LCPI6_0) |
| ; RV64I-NEXT: add a0, a1, a0 |
| ; RV64I-NEXT: lbu a0, 0(a0) |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; RV64I-NEXT: .LBB6_2: |
| ; RV64I-NEXT: li a0, 32 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: cttz_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: ctzw a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i32 @llvm.cttz.i32(i32 %a, i1 false) |
| ret i32 %1 |
| } |
| |
| define signext i32 @cttz_zero_undef_i32(i32 signext %a) nounwind { |
| ; RV64I-LABEL: cttz_zero_undef_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: neg a1, a0 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 30667 |
| ; RV64I-NEXT: addiw a1, a1, 1329 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw a0, a0, 27 |
| ; RV64I-NEXT: lui a1, %hi(.LCPI7_0) |
| ; RV64I-NEXT: addi a1, a1, %lo(.LCPI7_0) |
| ; RV64I-NEXT: add a0, a1, a0 |
| ; RV64I-NEXT: lbu a0, 0(a0) |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: cttz_zero_undef_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: ctzw a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i32 @llvm.cttz.i32(i32 %a, i1 true) |
| ret i32 %1 |
| } |
| |
| define signext i32 @findFirstSet_i32(i32 signext %a) nounwind { |
| ; RV64I-LABEL: findFirstSet_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s0, 0(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: mv s0, a0 |
| ; RV64I-NEXT: neg a0, a0 |
| ; RV64I-NEXT: and a0, s0, a0 |
| ; RV64I-NEXT: lui a1, 30667 |
| ; RV64I-NEXT: addiw a1, a1, 1329 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw a0, a0, 27 |
| ; RV64I-NEXT: lui a1, %hi(.LCPI8_0) |
| ; RV64I-NEXT: addi a1, a1, %lo(.LCPI8_0) |
| ; RV64I-NEXT: add a0, a1, a0 |
| ; RV64I-NEXT: lbu a0, 0(a0) |
| ; RV64I-NEXT: snez a1, s0 |
| ; RV64I-NEXT: addi a1, a1, -1 |
| ; RV64I-NEXT: or a0, a1, a0 |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s0, 0(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: findFirstSet_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: ctzw a1, a0 |
| ; RV64ZBB-NEXT: snez a0, a0 |
| ; RV64ZBB-NEXT: addi a0, a0, -1 |
| ; RV64ZBB-NEXT: or a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i32 @llvm.cttz.i32(i32 %a, i1 true) |
| %2 = icmp eq i32 %a, 0 |
| %3 = select i1 %2, i32 -1, i32 %1 |
| ret i32 %3 |
| } |
| |
| define signext i32 @ffs_i32(i32 signext %a) nounwind { |
| ; RV64I-LABEL: ffs_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s0, 0(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: mv s0, a0 |
| ; RV64I-NEXT: neg a0, a0 |
| ; RV64I-NEXT: and a0, s0, a0 |
| ; RV64I-NEXT: lui a1, 30667 |
| ; RV64I-NEXT: addiw a1, a1, 1329 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw a0, a0, 27 |
| ; RV64I-NEXT: lui a1, %hi(.LCPI9_0) |
| ; RV64I-NEXT: addi a1, a1, %lo(.LCPI9_0) |
| ; RV64I-NEXT: add a0, a1, a0 |
| ; RV64I-NEXT: lbu a0, 0(a0) |
| ; RV64I-NEXT: addi a0, a0, 1 |
| ; RV64I-NEXT: seqz a1, s0 |
| ; RV64I-NEXT: addi a1, a1, -1 |
| ; RV64I-NEXT: and a0, a1, a0 |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s0, 0(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ffs_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: ctzw a1, a0 |
| ; RV64ZBB-NEXT: addi a1, a1, 1 |
| ; RV64ZBB-NEXT: seqz a0, a0 |
| ; RV64ZBB-NEXT: addi a0, a0, -1 |
| ; RV64ZBB-NEXT: and a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i32 @llvm.cttz.i32(i32 %a, i1 true) |
| %2 = add i32 %1, 1 |
| %3 = icmp eq i32 %a, 0 |
| %4 = select i1 %3, i32 0, i32 %2 |
| ret i32 %4 |
| } |
| |
| declare i64 @llvm.cttz.i64(i64, i1) |
| |
| define i64 @cttz_i64(i64 %a) nounwind { |
| ; RV64I-LABEL: cttz_i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: beqz a0, .LBB10_2 |
| ; RV64I-NEXT: # %bb.1: # %cond.false |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: neg a1, a0 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: lui a1, %hi(.LCPI10_0) |
| ; RV64I-NEXT: ld a1, %lo(.LCPI10_0)(a1) |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srli a0, a0, 58 |
| ; RV64I-NEXT: lui a1, %hi(.LCPI10_1) |
| ; RV64I-NEXT: addi a1, a1, %lo(.LCPI10_1) |
| ; RV64I-NEXT: add a0, a1, a0 |
| ; RV64I-NEXT: lbu a0, 0(a0) |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; RV64I-NEXT: .LBB10_2: |
| ; RV64I-NEXT: li a0, 64 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: cttz_i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: ctz a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i64 @llvm.cttz.i64(i64 %a, i1 false) |
| ret i64 %1 |
| } |
| |
| declare i32 @llvm.ctpop.i32(i32) |
| |
| define signext i32 @ctpop_i32(i32 signext %a) nounwind { |
| ; RV64I-LABEL: ctpop_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: srli a1, a0, 1 |
| ; RV64I-NEXT: lui a2, 349525 |
| ; RV64I-NEXT: addiw a2, a2, 1365 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: sub a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 209715 |
| ; RV64I-NEXT: addiw a1, a1, 819 |
| ; RV64I-NEXT: and a2, a0, a1 |
| ; RV64I-NEXT: srli a0, a0, 2 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: add a0, a2, a0 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 61681 |
| ; RV64I-NEXT: addiw a1, a1, -241 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 4112 |
| ; RV64I-NEXT: addiw a1, a1, 257 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw a0, a0, 24 |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpopw a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i32 @llvm.ctpop.i32(i32 %a) |
| ret i32 %1 |
| } |
| |
| define i1 @ctpop_i32_ult_two(i32 signext %a) nounwind { |
| ; RV64I-LABEL: ctpop_i32_ult_two: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addiw a1, a0, -1 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: seqz a0, a0 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_i32_ult_two: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpopw a0, a0 |
| ; RV64ZBB-NEXT: sltiu a0, a0, 2 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i32 @llvm.ctpop.i32(i32 %a) |
| %2 = icmp ult i32 %1, 2 |
| ret i1 %2 |
| } |
| |
| define i1 @ctpop_i32_ugt_one(i32 signext %a) nounwind { |
| ; RV64I-LABEL: ctpop_i32_ugt_one: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addiw a1, a0, -1 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: snez a0, a0 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_i32_ugt_one: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpopw a0, a0 |
| ; RV64ZBB-NEXT: sltiu a0, a0, 2 |
| ; RV64ZBB-NEXT: xori a0, a0, 1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i32 @llvm.ctpop.i32(i32 %a) |
| %2 = icmp ugt i32 %1, 1 |
| ret i1 %2 |
| } |
| |
| define i1 @ctpop_i32_eq_one(i32 signext %a) nounwind { |
| ; RV64I-LABEL: ctpop_i32_eq_one: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addiw a1, a0, -1 |
| ; RV64I-NEXT: and a1, a0, a1 |
| ; RV64I-NEXT: seqz a1, a1 |
| ; RV64I-NEXT: snez a0, a0 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_i32_eq_one: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpopw a0, a0 |
| ; RV64ZBB-NEXT: addi a0, a0, -1 |
| ; RV64ZBB-NEXT: seqz a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i32 @llvm.ctpop.i32(i32 %a) |
| %2 = icmp eq i32 %1, 1 |
| ret i1 %2 |
| } |
| |
| define i1 @ctpop_i32_ne_one(i32 signext %a) nounwind { |
| ; RV64I-LABEL: ctpop_i32_ne_one: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addiw a1, a0, -1 |
| ; RV64I-NEXT: and a1, a0, a1 |
| ; RV64I-NEXT: snez a1, a1 |
| ; RV64I-NEXT: seqz a0, a0 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_i32_ne_one: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpopw a0, a0 |
| ; RV64ZBB-NEXT: addi a0, a0, -1 |
| ; RV64ZBB-NEXT: snez a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i32 @llvm.ctpop.i32(i32 %a) |
| %2 = icmp ne i32 %1, 1 |
| ret i1 %2 |
| } |
| |
| define signext i32 @ctpop_i32_load(ptr %p) nounwind { |
| ; RV64I-LABEL: ctpop_i32_load: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: lw a0, 0(a0) |
| ; RV64I-NEXT: srli a1, a0, 1 |
| ; RV64I-NEXT: lui a2, 349525 |
| ; RV64I-NEXT: addiw a2, a2, 1365 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: sub a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 209715 |
| ; RV64I-NEXT: addiw a1, a1, 819 |
| ; RV64I-NEXT: and a2, a0, a1 |
| ; RV64I-NEXT: srli a0, a0, 2 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: add a0, a2, a0 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 61681 |
| ; RV64I-NEXT: addiw a1, a1, -241 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 4112 |
| ; RV64I-NEXT: addiw a1, a1, 257 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw a0, a0, 24 |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_i32_load: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: lwu a0, 0(a0) |
| ; RV64ZBB-NEXT: cpopw a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %a = load i32, ptr %p |
| %1 = call i32 @llvm.ctpop.i32(i32 %a) |
| ret i32 %1 |
| } |
| |
| declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32>) |
| |
| define <2 x i32> @ctpop_v2i32(<2 x i32> %a) nounwind { |
| ; RV64I-LABEL: ctpop_v2i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi sp, sp, -64 |
| ; RV64I-NEXT: sd ra, 56(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s0, 48(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s1, 40(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s2, 32(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s3, 24(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s4, 16(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s5, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: mv s0, a1 |
| ; RV64I-NEXT: srli a1, a0, 1 |
| ; RV64I-NEXT: lui a2, 349525 |
| ; RV64I-NEXT: addiw s3, a2, 1365 |
| ; RV64I-NEXT: and a1, a1, s3 |
| ; RV64I-NEXT: sub a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 209715 |
| ; RV64I-NEXT: addiw s4, a1, 819 |
| ; RV64I-NEXT: and a1, a0, s4 |
| ; RV64I-NEXT: srli a0, a0, 2 |
| ; RV64I-NEXT: and a0, a0, s4 |
| ; RV64I-NEXT: add a0, a1, a0 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 61681 |
| ; RV64I-NEXT: addiw s5, a1, -241 |
| ; RV64I-NEXT: and a0, a0, s5 |
| ; RV64I-NEXT: lui a1, 4112 |
| ; RV64I-NEXT: addiw s1, a1, 257 |
| ; RV64I-NEXT: mv a1, s1 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw s2, a0, 24 |
| ; RV64I-NEXT: srli a0, s0, 1 |
| ; RV64I-NEXT: and a0, a0, s3 |
| ; RV64I-NEXT: sub s0, s0, a0 |
| ; RV64I-NEXT: and a0, s0, s4 |
| ; RV64I-NEXT: srli s0, s0, 2 |
| ; RV64I-NEXT: and a1, s0, s4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: and a0, a0, s5 |
| ; RV64I-NEXT: mv a1, s1 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srliw a1, a0, 24 |
| ; RV64I-NEXT: mv a0, s2 |
| ; RV64I-NEXT: ld ra, 56(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s0, 48(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s1, 40(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s2, 32(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s3, 24(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s4, 16(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s5, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 64 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_v2i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpopw a0, a0 |
| ; RV64ZBB-NEXT: cpopw a1, a1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call <2 x i32> @llvm.ctpop.v2i32(<2 x i32> %a) |
| ret <2 x i32> %1 |
| } |
| |
| define <2 x i1> @ctpop_v2i32_ult_two(<2 x i32> %a) nounwind { |
| ; RV64I-LABEL: ctpop_v2i32_ult_two: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addiw a2, a0, -1 |
| ; RV64I-NEXT: and a0, a0, a2 |
| ; RV64I-NEXT: sext.w a0, a0 |
| ; RV64I-NEXT: seqz a0, a0 |
| ; RV64I-NEXT: addiw a2, a1, -1 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: sext.w a1, a1 |
| ; RV64I-NEXT: seqz a1, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_v2i32_ult_two: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpopw a1, a1 |
| ; RV64ZBB-NEXT: cpopw a0, a0 |
| ; RV64ZBB-NEXT: sltiu a0, a0, 2 |
| ; RV64ZBB-NEXT: sltiu a1, a1, 2 |
| ; RV64ZBB-NEXT: ret |
| %1 = call <2 x i32> @llvm.ctpop.v2i32(<2 x i32> %a) |
| %2 = icmp ult <2 x i32> %1, <i32 2, i32 2> |
| ret <2 x i1> %2 |
| } |
| |
| define <2 x i1> @ctpop_v2i32_ugt_one(<2 x i32> %a) nounwind { |
| ; RV64I-LABEL: ctpop_v2i32_ugt_one: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addiw a2, a0, -1 |
| ; RV64I-NEXT: and a0, a0, a2 |
| ; RV64I-NEXT: sext.w a0, a0 |
| ; RV64I-NEXT: snez a0, a0 |
| ; RV64I-NEXT: addiw a2, a1, -1 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: sext.w a1, a1 |
| ; RV64I-NEXT: snez a1, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_v2i32_ugt_one: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpopw a1, a1 |
| ; RV64ZBB-NEXT: cpopw a0, a0 |
| ; RV64ZBB-NEXT: sltiu a0, a0, 2 |
| ; RV64ZBB-NEXT: xori a0, a0, 1 |
| ; RV64ZBB-NEXT: sltiu a1, a1, 2 |
| ; RV64ZBB-NEXT: xori a1, a1, 1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call <2 x i32> @llvm.ctpop.v2i32(<2 x i32> %a) |
| %2 = icmp ugt <2 x i32> %1, <i32 1, i32 1> |
| ret <2 x i1> %2 |
| } |
| |
| define <2 x i1> @ctpop_v2i32_eq_one(<2 x i32> %a) nounwind { |
| ; RV64I-LABEL: ctpop_v2i32_eq_one: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: sext.w a2, a1 |
| ; RV64I-NEXT: sext.w a3, a0 |
| ; RV64I-NEXT: addiw a4, a0, -1 |
| ; RV64I-NEXT: and a0, a0, a4 |
| ; RV64I-NEXT: sext.w a0, a0 |
| ; RV64I-NEXT: seqz a0, a0 |
| ; RV64I-NEXT: snez a3, a3 |
| ; RV64I-NEXT: and a0, a3, a0 |
| ; RV64I-NEXT: addiw a3, a1, -1 |
| ; RV64I-NEXT: and a1, a1, a3 |
| ; RV64I-NEXT: sext.w a1, a1 |
| ; RV64I-NEXT: seqz a1, a1 |
| ; RV64I-NEXT: snez a2, a2 |
| ; RV64I-NEXT: and a1, a2, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_v2i32_eq_one: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpopw a1, a1 |
| ; RV64ZBB-NEXT: cpopw a0, a0 |
| ; RV64ZBB-NEXT: addi a0, a0, -1 |
| ; RV64ZBB-NEXT: seqz a0, a0 |
| ; RV64ZBB-NEXT: addi a1, a1, -1 |
| ; RV64ZBB-NEXT: seqz a1, a1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call <2 x i32> @llvm.ctpop.v2i32(<2 x i32> %a) |
| %2 = icmp eq <2 x i32> %1, <i32 1, i32 1> |
| ret <2 x i1> %2 |
| } |
| |
| define <2 x i1> @ctpop_v2i32_ne_one(<2 x i32> %a) nounwind { |
| ; RV64I-LABEL: ctpop_v2i32_ne_one: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: sext.w a2, a1 |
| ; RV64I-NEXT: sext.w a3, a0 |
| ; RV64I-NEXT: addiw a4, a0, -1 |
| ; RV64I-NEXT: and a0, a0, a4 |
| ; RV64I-NEXT: sext.w a0, a0 |
| ; RV64I-NEXT: snez a0, a0 |
| ; RV64I-NEXT: seqz a3, a3 |
| ; RV64I-NEXT: or a0, a3, a0 |
| ; RV64I-NEXT: addiw a3, a1, -1 |
| ; RV64I-NEXT: and a1, a1, a3 |
| ; RV64I-NEXT: sext.w a1, a1 |
| ; RV64I-NEXT: snez a1, a1 |
| ; RV64I-NEXT: seqz a2, a2 |
| ; RV64I-NEXT: or a1, a2, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_v2i32_ne_one: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpopw a1, a1 |
| ; RV64ZBB-NEXT: cpopw a0, a0 |
| ; RV64ZBB-NEXT: addi a0, a0, -1 |
| ; RV64ZBB-NEXT: snez a0, a0 |
| ; RV64ZBB-NEXT: addi a1, a1, -1 |
| ; RV64ZBB-NEXT: snez a1, a1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call <2 x i32> @llvm.ctpop.v2i32(<2 x i32> %a) |
| %2 = icmp ne <2 x i32> %1, <i32 1, i32 1> |
| ret <2 x i1> %2 |
| } |
| |
| declare i64 @llvm.ctpop.i64(i64) |
| |
| define i64 @ctpop_i64(i64 %a) nounwind { |
| ; RV64I-LABEL: ctpop_i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi sp, sp, -16 |
| ; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: srli a1, a0, 1 |
| ; RV64I-NEXT: lui a2, 349525 |
| ; RV64I-NEXT: addiw a2, a2, 1365 |
| ; RV64I-NEXT: slli a3, a2, 32 |
| ; RV64I-NEXT: add a2, a2, a3 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: sub a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 209715 |
| ; RV64I-NEXT: addiw a1, a1, 819 |
| ; RV64I-NEXT: slli a2, a1, 32 |
| ; RV64I-NEXT: add a1, a1, a2 |
| ; RV64I-NEXT: and a2, a0, a1 |
| ; RV64I-NEXT: srli a0, a0, 2 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: add a0, a2, a0 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 61681 |
| ; RV64I-NEXT: addiw a1, a1, -241 |
| ; RV64I-NEXT: slli a2, a1, 32 |
| ; RV64I-NEXT: add a1, a1, a2 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 4112 |
| ; RV64I-NEXT: addiw a1, a1, 257 |
| ; RV64I-NEXT: slli a2, a1, 32 |
| ; RV64I-NEXT: add a1, a1, a2 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srli a0, a0, 56 |
| ; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 16 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpop a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i64 @llvm.ctpop.i64(i64 %a) |
| ret i64 %1 |
| } |
| |
| define i1 @ctpop_i64_ugt_two(i64 %a) nounwind { |
| ; RV64I-LABEL: ctpop_i64_ugt_two: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi a1, a0, -1 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: seqz a0, a0 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_i64_ugt_two: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpop a0, a0 |
| ; RV64ZBB-NEXT: sltiu a0, a0, 2 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i64 @llvm.ctpop.i64(i64 %a) |
| %2 = icmp ult i64 %1, 2 |
| ret i1 %2 |
| } |
| |
| define i1 @ctpop_i64_ugt_one(i64 %a) nounwind { |
| ; RV64I-LABEL: ctpop_i64_ugt_one: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi a1, a0, -1 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: snez a0, a0 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_i64_ugt_one: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpop a0, a0 |
| ; RV64ZBB-NEXT: sltiu a0, a0, 2 |
| ; RV64ZBB-NEXT: xori a0, a0, 1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i64 @llvm.ctpop.i64(i64 %a) |
| %2 = icmp ugt i64 %1, 1 |
| ret i1 %2 |
| } |
| |
| define i1 @ctpop_i64_eq_one(i64 %a) nounwind { |
| ; RV64I-LABEL: ctpop_i64_eq_one: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi a1, a0, -1 |
| ; RV64I-NEXT: and a1, a0, a1 |
| ; RV64I-NEXT: seqz a1, a1 |
| ; RV64I-NEXT: snez a0, a0 |
| ; RV64I-NEXT: and a0, a0, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_i64_eq_one: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpop a0, a0 |
| ; RV64ZBB-NEXT: addi a0, a0, -1 |
| ; RV64ZBB-NEXT: seqz a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i64 @llvm.ctpop.i64(i64 %a) |
| %2 = icmp eq i64 %1, 1 |
| ret i1 %2 |
| } |
| |
| define i1 @ctpop_i64_ne_one(i64 %a) nounwind { |
| ; RV64I-LABEL: ctpop_i64_ne_one: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi a1, a0, -1 |
| ; RV64I-NEXT: and a1, a0, a1 |
| ; RV64I-NEXT: snez a1, a1 |
| ; RV64I-NEXT: seqz a0, a0 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_i64_ne_one: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpop a0, a0 |
| ; RV64ZBB-NEXT: addi a0, a0, -1 |
| ; RV64ZBB-NEXT: snez a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i64 @llvm.ctpop.i64(i64 %a) |
| %2 = icmp ne i64 %1, 1 |
| ret i1 %2 |
| } |
| |
| declare <2 x i64> @llvm.ctpop.v2i64(<2 x i64>) |
| |
| define <2 x i64> @ctpop_v2i64(<2 x i64> %a) nounwind { |
| ; RV64I-LABEL: ctpop_v2i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi sp, sp, -64 |
| ; RV64I-NEXT: sd ra, 56(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s0, 48(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s1, 40(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s2, 32(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s3, 24(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s4, 16(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: sd s5, 8(sp) # 8-byte Folded Spill |
| ; RV64I-NEXT: mv s0, a1 |
| ; RV64I-NEXT: srli a1, a0, 1 |
| ; RV64I-NEXT: lui a2, 349525 |
| ; RV64I-NEXT: addiw a2, a2, 1365 |
| ; RV64I-NEXT: slli a3, a2, 32 |
| ; RV64I-NEXT: add s3, a2, a3 |
| ; RV64I-NEXT: and a1, a1, s3 |
| ; RV64I-NEXT: sub a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 209715 |
| ; RV64I-NEXT: addiw a1, a1, 819 |
| ; RV64I-NEXT: slli a2, a1, 32 |
| ; RV64I-NEXT: add s4, a1, a2 |
| ; RV64I-NEXT: and a1, a0, s4 |
| ; RV64I-NEXT: srli a0, a0, 2 |
| ; RV64I-NEXT: and a0, a0, s4 |
| ; RV64I-NEXT: add a0, a1, a0 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: lui a1, 61681 |
| ; RV64I-NEXT: addiw a1, a1, -241 |
| ; RV64I-NEXT: slli a2, a1, 32 |
| ; RV64I-NEXT: add s5, a1, a2 |
| ; RV64I-NEXT: and a0, a0, s5 |
| ; RV64I-NEXT: lui a1, 4112 |
| ; RV64I-NEXT: addiw s1, a1, 257 |
| ; RV64I-NEXT: slli a1, s1, 32 |
| ; RV64I-NEXT: add s1, s1, a1 |
| ; RV64I-NEXT: mv a1, s1 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srli s2, a0, 56 |
| ; RV64I-NEXT: srli a0, s0, 1 |
| ; RV64I-NEXT: and a0, a0, s3 |
| ; RV64I-NEXT: sub s0, s0, a0 |
| ; RV64I-NEXT: and a0, s0, s4 |
| ; RV64I-NEXT: srli s0, s0, 2 |
| ; RV64I-NEXT: and a1, s0, s4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: srli a1, a0, 4 |
| ; RV64I-NEXT: add a0, a0, a1 |
| ; RV64I-NEXT: and a0, a0, s5 |
| ; RV64I-NEXT: mv a1, s1 |
| ; RV64I-NEXT: call __muldi3@plt |
| ; RV64I-NEXT: srli a1, a0, 56 |
| ; RV64I-NEXT: mv a0, s2 |
| ; RV64I-NEXT: ld ra, 56(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s0, 48(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s1, 40(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s2, 32(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s3, 24(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s4, 16(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: ld s5, 8(sp) # 8-byte Folded Reload |
| ; RV64I-NEXT: addi sp, sp, 64 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_v2i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpop a0, a0 |
| ; RV64ZBB-NEXT: cpop a1, a1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %a) |
| ret <2 x i64> %1 |
| } |
| |
| define <2 x i1> @ctpop_v2i64_ult_two(<2 x i64> %a) nounwind { |
| ; RV64I-LABEL: ctpop_v2i64_ult_two: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi a2, a0, -1 |
| ; RV64I-NEXT: and a0, a0, a2 |
| ; RV64I-NEXT: seqz a0, a0 |
| ; RV64I-NEXT: addi a2, a1, -1 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: seqz a1, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_v2i64_ult_two: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpop a1, a1 |
| ; RV64ZBB-NEXT: cpop a0, a0 |
| ; RV64ZBB-NEXT: sltiu a0, a0, 2 |
| ; RV64ZBB-NEXT: sltiu a1, a1, 2 |
| ; RV64ZBB-NEXT: ret |
| %1 = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %a) |
| %2 = icmp ult <2 x i64> %1, <i64 2, i64 2> |
| ret <2 x i1> %2 |
| } |
| |
| define <2 x i1> @ctpop_v2i64_ugt_one(<2 x i64> %a) nounwind { |
| ; RV64I-LABEL: ctpop_v2i64_ugt_one: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi a2, a0, -1 |
| ; RV64I-NEXT: and a0, a0, a2 |
| ; RV64I-NEXT: snez a0, a0 |
| ; RV64I-NEXT: addi a2, a1, -1 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: snez a1, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_v2i64_ugt_one: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpop a1, a1 |
| ; RV64ZBB-NEXT: cpop a0, a0 |
| ; RV64ZBB-NEXT: sltiu a0, a0, 2 |
| ; RV64ZBB-NEXT: xori a0, a0, 1 |
| ; RV64ZBB-NEXT: sltiu a1, a1, 2 |
| ; RV64ZBB-NEXT: xori a1, a1, 1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %a) |
| %2 = icmp ugt <2 x i64> %1, <i64 1, i64 1> |
| ret <2 x i1> %2 |
| } |
| |
| define <2 x i1> @ctpop_v2i64_eq_one(<2 x i64> %a) nounwind { |
| ; RV64I-LABEL: ctpop_v2i64_eq_one: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi a2, a0, -1 |
| ; RV64I-NEXT: and a2, a0, a2 |
| ; RV64I-NEXT: seqz a2, a2 |
| ; RV64I-NEXT: snez a0, a0 |
| ; RV64I-NEXT: and a0, a0, a2 |
| ; RV64I-NEXT: addi a2, a1, -1 |
| ; RV64I-NEXT: and a2, a1, a2 |
| ; RV64I-NEXT: seqz a2, a2 |
| ; RV64I-NEXT: snez a1, a1 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_v2i64_eq_one: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpop a1, a1 |
| ; RV64ZBB-NEXT: cpop a0, a0 |
| ; RV64ZBB-NEXT: addi a0, a0, -1 |
| ; RV64ZBB-NEXT: seqz a0, a0 |
| ; RV64ZBB-NEXT: addi a1, a1, -1 |
| ; RV64ZBB-NEXT: seqz a1, a1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %a) |
| %2 = icmp eq <2 x i64> %1, <i64 1, i64 1> |
| ret <2 x i1> %2 |
| } |
| |
| define <2 x i1> @ctpop_v2i64_ne_one(<2 x i64> %a) nounwind { |
| ; RV64I-LABEL: ctpop_v2i64_ne_one: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: addi a2, a0, -1 |
| ; RV64I-NEXT: and a2, a0, a2 |
| ; RV64I-NEXT: snez a2, a2 |
| ; RV64I-NEXT: seqz a0, a0 |
| ; RV64I-NEXT: or a0, a0, a2 |
| ; RV64I-NEXT: addi a2, a1, -1 |
| ; RV64I-NEXT: and a2, a1, a2 |
| ; RV64I-NEXT: snez a2, a2 |
| ; RV64I-NEXT: seqz a1, a1 |
| ; RV64I-NEXT: or a1, a1, a2 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: ctpop_v2i64_ne_one: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: cpop a1, a1 |
| ; RV64ZBB-NEXT: cpop a0, a0 |
| ; RV64ZBB-NEXT: addi a0, a0, -1 |
| ; RV64ZBB-NEXT: snez a0, a0 |
| ; RV64ZBB-NEXT: addi a1, a1, -1 |
| ; RV64ZBB-NEXT: snez a1, a1 |
| ; RV64ZBB-NEXT: ret |
| %1 = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %a) |
| %2 = icmp ne <2 x i64> %1, <i64 1, i64 1> |
| ret <2 x i1> %2 |
| } |
| |
| define signext i32 @sextb_i32(i32 signext %a) nounwind { |
| ; RV64I-LABEL: sextb_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: slli a0, a0, 56 |
| ; RV64I-NEXT: srai a0, a0, 56 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: sextb_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: sext.b a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %shl = shl i32 %a, 24 |
| %shr = ashr exact i32 %shl, 24 |
| ret i32 %shr |
| } |
| |
| define i64 @sextb_i64(i64 %a) nounwind { |
| ; RV64I-LABEL: sextb_i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: slli a0, a0, 56 |
| ; RV64I-NEXT: srai a0, a0, 56 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: sextb_i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: sext.b a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %shl = shl i64 %a, 56 |
| %shr = ashr exact i64 %shl, 56 |
| ret i64 %shr |
| } |
| |
| define signext i32 @sexth_i32(i32 signext %a) nounwind { |
| ; RV64I-LABEL: sexth_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: slli a0, a0, 48 |
| ; RV64I-NEXT: srai a0, a0, 48 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: sexth_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: sext.h a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %shl = shl i32 %a, 16 |
| %shr = ashr exact i32 %shl, 16 |
| ret i32 %shr |
| } |
| |
| define i64 @sexth_i64(i64 %a) nounwind { |
| ; RV64I-LABEL: sexth_i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: slli a0, a0, 48 |
| ; RV64I-NEXT: srai a0, a0, 48 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: sexth_i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: sext.h a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %shl = shl i64 %a, 48 |
| %shr = ashr exact i64 %shl, 48 |
| ret i64 %shr |
| } |
| |
| define signext i32 @min_i32(i32 signext %a, i32 signext %b) nounwind { |
| ; RV64I-LABEL: min_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: blt a0, a1, .LBB36_2 |
| ; RV64I-NEXT: # %bb.1: |
| ; RV64I-NEXT: mv a0, a1 |
| ; RV64I-NEXT: .LBB36_2: |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: min_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: min a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %cmp = icmp slt i32 %a, %b |
| %cond = select i1 %cmp, i32 %a, i32 %b |
| ret i32 %cond |
| } |
| |
| define i64 @min_i64(i64 %a, i64 %b) nounwind { |
| ; RV64I-LABEL: min_i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: blt a0, a1, .LBB37_2 |
| ; RV64I-NEXT: # %bb.1: |
| ; RV64I-NEXT: mv a0, a1 |
| ; RV64I-NEXT: .LBB37_2: |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: min_i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: min a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %cmp = icmp slt i64 %a, %b |
| %cond = select i1 %cmp, i64 %a, i64 %b |
| ret i64 %cond |
| } |
| |
| define signext i32 @max_i32(i32 signext %a, i32 signext %b) nounwind { |
| ; RV64I-LABEL: max_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: blt a1, a0, .LBB38_2 |
| ; RV64I-NEXT: # %bb.1: |
| ; RV64I-NEXT: mv a0, a1 |
| ; RV64I-NEXT: .LBB38_2: |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: max_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: max a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %cmp = icmp sgt i32 %a, %b |
| %cond = select i1 %cmp, i32 %a, i32 %b |
| ret i32 %cond |
| } |
| |
| define i64 @max_i64(i64 %a, i64 %b) nounwind { |
| ; RV64I-LABEL: max_i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: blt a1, a0, .LBB39_2 |
| ; RV64I-NEXT: # %bb.1: |
| ; RV64I-NEXT: mv a0, a1 |
| ; RV64I-NEXT: .LBB39_2: |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: max_i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: max a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %cmp = icmp sgt i64 %a, %b |
| %cond = select i1 %cmp, i64 %a, i64 %b |
| ret i64 %cond |
| } |
| |
| define signext i32 @minu_i32(i32 signext %a, i32 signext %b) nounwind { |
| ; RV64I-LABEL: minu_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: bltu a0, a1, .LBB40_2 |
| ; RV64I-NEXT: # %bb.1: |
| ; RV64I-NEXT: mv a0, a1 |
| ; RV64I-NEXT: .LBB40_2: |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: minu_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: minu a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %cmp = icmp ult i32 %a, %b |
| %cond = select i1 %cmp, i32 %a, i32 %b |
| ret i32 %cond |
| } |
| |
| define i64 @minu_i64(i64 %a, i64 %b) nounwind { |
| ; RV64I-LABEL: minu_i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: bltu a0, a1, .LBB41_2 |
| ; RV64I-NEXT: # %bb.1: |
| ; RV64I-NEXT: mv a0, a1 |
| ; RV64I-NEXT: .LBB41_2: |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: minu_i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: minu a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %cmp = icmp ult i64 %a, %b |
| %cond = select i1 %cmp, i64 %a, i64 %b |
| ret i64 %cond |
| } |
| |
| define signext i32 @maxu_i32(i32 signext %a, i32 signext %b) nounwind { |
| ; RV64I-LABEL: maxu_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: bltu a1, a0, .LBB42_2 |
| ; RV64I-NEXT: # %bb.1: |
| ; RV64I-NEXT: mv a0, a1 |
| ; RV64I-NEXT: .LBB42_2: |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: maxu_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: maxu a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %cmp = icmp ugt i32 %a, %b |
| %cond = select i1 %cmp, i32 %a, i32 %b |
| ret i32 %cond |
| } |
| |
| define i64 @maxu_i64(i64 %a, i64 %b) nounwind { |
| ; RV64I-LABEL: maxu_i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: bltu a1, a0, .LBB43_2 |
| ; RV64I-NEXT: # %bb.1: |
| ; RV64I-NEXT: mv a0, a1 |
| ; RV64I-NEXT: .LBB43_2: |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: maxu_i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: maxu a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %cmp = icmp ugt i64 %a, %b |
| %cond = select i1 %cmp, i64 %a, i64 %b |
| ret i64 %cond |
| } |
| |
| declare i32 @llvm.abs.i32(i32, i1 immarg) |
| |
| define i32 @abs_i32(i32 %x) { |
| ; RV64I-LABEL: abs_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: sraiw a1, a0, 31 |
| ; RV64I-NEXT: xor a0, a0, a1 |
| ; RV64I-NEXT: subw a0, a0, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: abs_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: sext.w a0, a0 |
| ; RV64ZBB-NEXT: negw a1, a0 |
| ; RV64ZBB-NEXT: max a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %abs = tail call i32 @llvm.abs.i32(i32 %x, i1 true) |
| ret i32 %abs |
| } |
| |
| define signext i32 @abs_i32_sext(i32 signext %x) { |
| ; RV64I-LABEL: abs_i32_sext: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: srai a1, a0, 31 |
| ; RV64I-NEXT: xor a0, a0, a1 |
| ; RV64I-NEXT: subw a0, a0, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: abs_i32_sext: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: negw a1, a0 |
| ; RV64ZBB-NEXT: max a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %abs = tail call i32 @llvm.abs.i32(i32 %x, i1 true) |
| ret i32 %abs |
| } |
| |
| declare i64 @llvm.abs.i64(i64, i1 immarg) |
| |
| define i64 @abs_i64(i64 %x) { |
| ; RV64I-LABEL: abs_i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: srai a1, a0, 63 |
| ; RV64I-NEXT: xor a0, a0, a1 |
| ; RV64I-NEXT: sub a0, a0, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: abs_i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: neg a1, a0 |
| ; RV64ZBB-NEXT: max a0, a0, a1 |
| ; RV64ZBB-NEXT: ret |
| %abs = tail call i64 @llvm.abs.i64(i64 %x, i1 true) |
| ret i64 %abs |
| } |
| |
| define i32 @zexth_i32(i32 %a) nounwind { |
| ; RV64I-LABEL: zexth_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: slli a0, a0, 48 |
| ; RV64I-NEXT: srli a0, a0, 48 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: zexth_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: zext.h a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %and = and i32 %a, 65535 |
| ret i32 %and |
| } |
| |
| define i64 @zexth_i64(i64 %a) nounwind { |
| ; RV64I-LABEL: zexth_i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: slli a0, a0, 48 |
| ; RV64I-NEXT: srli a0, a0, 48 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: zexth_i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: zext.h a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %and = and i64 %a, 65535 |
| ret i64 %and |
| } |
| |
| declare i32 @llvm.bswap.i32(i32) |
| |
| define signext i32 @bswap_i32(i32 signext %a) nounwind { |
| ; RV64I-LABEL: bswap_i32: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: srli a1, a0, 8 |
| ; RV64I-NEXT: lui a2, 16 |
| ; RV64I-NEXT: addiw a2, a2, -256 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: srliw a3, a0, 24 |
| ; RV64I-NEXT: or a1, a1, a3 |
| ; RV64I-NEXT: and a2, a0, a2 |
| ; RV64I-NEXT: slli a2, a2, 8 |
| ; RV64I-NEXT: slliw a0, a0, 24 |
| ; RV64I-NEXT: or a0, a0, a2 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: bswap_i32: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: rev8 a0, a0 |
| ; RV64ZBB-NEXT: srai a0, a0, 32 |
| ; RV64ZBB-NEXT: ret |
| %1 = tail call i32 @llvm.bswap.i32(i32 %a) |
| ret i32 %1 |
| } |
| |
| ; Similar to bswap_i32 but the result is not sign extended. |
| define void @bswap_i32_nosext(i32 signext %a, ptr %x) nounwind { |
| ; RV64I-LABEL: bswap_i32_nosext: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: srli a2, a0, 8 |
| ; RV64I-NEXT: lui a3, 16 |
| ; RV64I-NEXT: addiw a3, a3, -256 |
| ; RV64I-NEXT: and a2, a2, a3 |
| ; RV64I-NEXT: srliw a4, a0, 24 |
| ; RV64I-NEXT: or a2, a2, a4 |
| ; RV64I-NEXT: and a3, a0, a3 |
| ; RV64I-NEXT: slli a3, a3, 8 |
| ; RV64I-NEXT: slli a0, a0, 24 |
| ; RV64I-NEXT: or a0, a0, a3 |
| ; RV64I-NEXT: or a0, a0, a2 |
| ; RV64I-NEXT: sw a0, 0(a1) |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: bswap_i32_nosext: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: rev8 a0, a0 |
| ; RV64ZBB-NEXT: srli a0, a0, 32 |
| ; RV64ZBB-NEXT: sw a0, 0(a1) |
| ; RV64ZBB-NEXT: ret |
| %1 = tail call i32 @llvm.bswap.i32(i32 %a) |
| store i32 %1, ptr %x |
| ret void |
| } |
| |
| declare i64 @llvm.bswap.i64(i64) |
| |
| define i64 @bswap_i64(i64 %a) { |
| ; RV64I-LABEL: bswap_i64: |
| ; RV64I: # %bb.0: |
| ; RV64I-NEXT: srli a1, a0, 40 |
| ; RV64I-NEXT: lui a2, 16 |
| ; RV64I-NEXT: addiw a2, a2, -256 |
| ; RV64I-NEXT: and a1, a1, a2 |
| ; RV64I-NEXT: srli a3, a0, 56 |
| ; RV64I-NEXT: or a1, a1, a3 |
| ; RV64I-NEXT: srli a3, a0, 24 |
| ; RV64I-NEXT: lui a4, 4080 |
| ; RV64I-NEXT: and a3, a3, a4 |
| ; RV64I-NEXT: srli a5, a0, 8 |
| ; RV64I-NEXT: srliw a5, a5, 24 |
| ; RV64I-NEXT: slli a5, a5, 24 |
| ; RV64I-NEXT: or a3, a5, a3 |
| ; RV64I-NEXT: or a1, a3, a1 |
| ; RV64I-NEXT: and a4, a0, a4 |
| ; RV64I-NEXT: slli a4, a4, 24 |
| ; RV64I-NEXT: srliw a3, a0, 24 |
| ; RV64I-NEXT: slli a3, a3, 32 |
| ; RV64I-NEXT: or a3, a4, a3 |
| ; RV64I-NEXT: and a2, a0, a2 |
| ; RV64I-NEXT: slli a2, a2, 40 |
| ; RV64I-NEXT: slli a0, a0, 56 |
| ; RV64I-NEXT: or a0, a0, a2 |
| ; RV64I-NEXT: or a0, a0, a3 |
| ; RV64I-NEXT: or a0, a0, a1 |
| ; RV64I-NEXT: ret |
| ; |
| ; RV64ZBB-LABEL: bswap_i64: |
| ; RV64ZBB: # %bb.0: |
| ; RV64ZBB-NEXT: rev8 a0, a0 |
| ; RV64ZBB-NEXT: ret |
| %1 = call i64 @llvm.bswap.i64(i64 %a) |
| ret i64 %1 |
| } |
