eb7d16ea25
mflr is kind of expensive on Power version smaller than 10, so we should schedule the store for the mflr's def away from mflr. In epilogue, the expensive mtlr has no user for its def, so it doesn't matter that the load and the mtlr are back-to-back. Reviewed By: RolandF Differential Revision: https://reviews.llvm.org/D137423
107 lines
2.8 KiB
LLVM
107 lines
2.8 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -verify-machineinstrs -mtriple=powerpc-unknown-linux-gnu -mcpu=ppc64 < %s | FileCheck --check-prefix=CHECK %s
|
|
; RUN: llc -verify-machineinstrs -mtriple=powerpc-unknown-linux-gnu -mcpu=ppc < %s | FileCheck --check-prefix=CHECK %s
|
|
|
|
define i32 @fold_urem_positive_odd(i32 %x) {
|
|
; CHECK-LABEL: fold_urem_positive_odd:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lis 4, 22765
|
|
; CHECK-NEXT: ori 4, 4, 8969
|
|
; CHECK-NEXT: mulhwu 4, 3, 4
|
|
; CHECK-NEXT: sub 5, 3, 4
|
|
; CHECK-NEXT: srwi 5, 5, 1
|
|
; CHECK-NEXT: add 4, 5, 4
|
|
; CHECK-NEXT: srwi 4, 4, 6
|
|
; CHECK-NEXT: mulli 4, 4, 95
|
|
; CHECK-NEXT: sub 3, 3, 4
|
|
; CHECK-NEXT: blr
|
|
%1 = urem i32 %x, 95
|
|
ret i32 %1
|
|
}
|
|
|
|
|
|
define i32 @fold_urem_positive_even(i32 %x) {
|
|
; CHECK-LABEL: fold_urem_positive_even:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lis 4, -2226
|
|
; CHECK-NEXT: ori 4, 4, 16323
|
|
; CHECK-NEXT: mulhwu 4, 3, 4
|
|
; CHECK-NEXT: srwi 4, 4, 10
|
|
; CHECK-NEXT: mulli 4, 4, 1060
|
|
; CHECK-NEXT: sub 3, 3, 4
|
|
; CHECK-NEXT: blr
|
|
%1 = urem i32 %x, 1060
|
|
ret i32 %1
|
|
}
|
|
|
|
|
|
; Don't fold if we can combine urem with udiv.
|
|
define i32 @combine_urem_udiv(i32 %x) {
|
|
; CHECK-LABEL: combine_urem_udiv:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lis 4, 22765
|
|
; CHECK-NEXT: ori 4, 4, 8969
|
|
; CHECK-NEXT: mulhwu 4, 3, 4
|
|
; CHECK-NEXT: sub 5, 3, 4
|
|
; CHECK-NEXT: srwi 5, 5, 1
|
|
; CHECK-NEXT: add 4, 5, 4
|
|
; CHECK-NEXT: srwi 4, 4, 6
|
|
; CHECK-NEXT: mulli 5, 4, 95
|
|
; CHECK-NEXT: sub 3, 3, 5
|
|
; CHECK-NEXT: add 3, 3, 4
|
|
; CHECK-NEXT: blr
|
|
%1 = urem i32 %x, 95
|
|
%2 = udiv i32 %x, 95
|
|
%3 = add i32 %1, %2
|
|
ret i32 %3
|
|
}
|
|
|
|
; Don't fold for divisors that are a power of two.
|
|
define i32 @dont_fold_urem_power_of_two(i32 %x) {
|
|
; CHECK-LABEL: dont_fold_urem_power_of_two:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: clrlwi 3, 3, 26
|
|
; CHECK-NEXT: blr
|
|
%1 = urem i32 %x, 64
|
|
ret i32 %1
|
|
}
|
|
|
|
; Don't fold if the divisor is one.
|
|
define i32 @dont_fold_urem_one(i32 %x) {
|
|
; CHECK-LABEL: dont_fold_urem_one:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: li 3, 0
|
|
; CHECK-NEXT: blr
|
|
%1 = urem i32 %x, 1
|
|
ret i32 %1
|
|
}
|
|
|
|
; Don't fold if the divisor is 2^32.
|
|
define i32 @dont_fold_urem_i32_umax(i32 %x) {
|
|
; CHECK-LABEL: dont_fold_urem_i32_umax:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: blr
|
|
%1 = urem i32 %x, 4294967296
|
|
ret i32 %1
|
|
}
|
|
|
|
; Don't fold i64 urem
|
|
define i64 @dont_fold_urem_i64(i64 %x) {
|
|
; CHECK-LABEL: dont_fold_urem_i64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: mflr 0
|
|
; CHECK-NEXT: stwu 1, -16(1)
|
|
; CHECK-NEXT: stw 0, 20(1)
|
|
; CHECK-NEXT: .cfi_def_cfa_offset 16
|
|
; CHECK-NEXT: .cfi_offset lr, 4
|
|
; CHECK-NEXT: li 5, 0
|
|
; CHECK-NEXT: li 6, 98
|
|
; CHECK-NEXT: bl __umoddi3
|
|
; CHECK-NEXT: lwz 0, 20(1)
|
|
; CHECK-NEXT: addi 1, 1, 16
|
|
; CHECK-NEXT: mtlr 0
|
|
; CHECK-NEXT: blr
|
|
%1 = urem i64 %x, 98
|
|
ret i64 %1
|
|
}
|