mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-23 19:23:23 +01:00
0c16dcd701
This provides an optimized implementation of SADDO/SSUBO/UADDO/USUBO as well as ADDCARRY/SUBCARRY on top of the new CC implementation. In particular, multi-word arithmetic now uses UADDO/ADDCARRY instead of the old ADDC/ADDE logic, which means we no longer need to use "glue" links for those instructions. This also allows making full use of the memory-based instructions like ALSI, which couldn't be recognized due to limitations in the DAG matcher previously. Also, the llvm.sadd.with.overflow et.al. intrinsincs now expand to directly using the ADD instructions and checking for a CC 3 result. llvm-svn: 331203
99 lines
2.6 KiB
LLVM
99 lines
2.6 KiB
LLVM
; Test 32-bit subtraction in which the second operand is constant.
|
|
;
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
|
|
|
|
declare i32 @foo()
|
|
|
|
; Check subtraction of 1.
|
|
define zeroext i1 @f1(i32 %dummy, i32 %a, i32 *%res) {
|
|
; CHECK-LABEL: f1:
|
|
; CHECK: slfi %r3, 1
|
|
; CHECK-DAG: st %r3, 0(%r4)
|
|
; CHECK-DAG: ipm [[REG:%r[0-5]]]
|
|
; CHECK-DAG: afi [[REG]], -536870912
|
|
; CHECK-DAG: risbg %r2, [[REG]], 63, 191, 33
|
|
; CHECK: br %r14
|
|
%t = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 1)
|
|
%val = extractvalue {i32, i1} %t, 0
|
|
%obit = extractvalue {i32, i1} %t, 1
|
|
store i32 %val, i32 *%res
|
|
ret i1 %obit
|
|
}
|
|
|
|
; Check the high end of the SLFI range.
|
|
define zeroext i1 @f2(i32 %dummy, i32 %a, i32 *%res) {
|
|
; CHECK-LABEL: f2:
|
|
; CHECK: slfi %r3, 4294967295
|
|
; CHECK-DAG: st %r3, 0(%r4)
|
|
; CHECK-DAG: ipm [[REG:%r[0-5]]]
|
|
; CHECK-DAG: afi [[REG]], -536870912
|
|
; CHECK-DAG: risbg %r2, [[REG]], 63, 191, 33
|
|
; CHECK: br %r14
|
|
%t = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 4294967295)
|
|
%val = extractvalue {i32, i1} %t, 0
|
|
%obit = extractvalue {i32, i1} %t, 1
|
|
store i32 %val, i32 *%res
|
|
ret i1 %obit
|
|
}
|
|
|
|
; Check that negative values are treated as unsigned
|
|
define zeroext i1 @f3(i32 %dummy, i32 %a, i32 *%res) {
|
|
; CHECK-LABEL: f3:
|
|
; CHECK: slfi %r3, 4294967295
|
|
; CHECK-DAG: st %r3, 0(%r4)
|
|
; CHECK-DAG: ipm [[REG:%r[0-5]]]
|
|
; CHECK-DAG: afi [[REG]], -536870912
|
|
; CHECK-DAG: risbg %r2, [[REG]], 63, 191, 33
|
|
; CHECK: br %r14
|
|
%t = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 -1)
|
|
%val = extractvalue {i32, i1} %t, 0
|
|
%obit = extractvalue {i32, i1} %t, 1
|
|
store i32 %val, i32 *%res
|
|
ret i1 %obit
|
|
}
|
|
|
|
; Check using the overflow result for a branch.
|
|
define void @f4(i32 %dummy, i32 %a, i32 *%res) {
|
|
; CHECK-LABEL: f4:
|
|
; CHECK: slfi %r3, 1
|
|
; CHECK: st %r3, 0(%r4)
|
|
; CHECK: jgle foo@PLT
|
|
; CHECK: br %r14
|
|
%t = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 1)
|
|
%val = extractvalue {i32, i1} %t, 0
|
|
%obit = extractvalue {i32, i1} %t, 1
|
|
store i32 %val, i32 *%res
|
|
br i1 %obit, label %call, label %exit
|
|
|
|
call:
|
|
tail call i32 @foo()
|
|
br label %exit
|
|
|
|
exit:
|
|
ret void
|
|
}
|
|
|
|
; ... and the same with the inverted direction.
|
|
define void @f5(i32 %dummy, i32 %a, i32 *%res) {
|
|
; CHECK-LABEL: f5:
|
|
; CHECK: slfi %r3, 1
|
|
; CHECK: st %r3, 0(%r4)
|
|
; CHECK: jgnle foo@PLT
|
|
; CHECK: br %r14
|
|
%t = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 1)
|
|
%val = extractvalue {i32, i1} %t, 0
|
|
%obit = extractvalue {i32, i1} %t, 1
|
|
store i32 %val, i32 *%res
|
|
br i1 %obit, label %exit, label %call
|
|
|
|
call:
|
|
tail call i32 @foo()
|
|
br label %exit
|
|
|
|
exit:
|
|
ret void
|
|
}
|
|
|
|
declare {i32, i1} @llvm.usub.with.overflow.i32(i32, i32) nounwind readnone
|
|
|