298127dcbe
Relands 7ff3a9acd84654c9ec2939f45ba27f162ae7fbc3 after regenerating the test case. Supersedes the draft PR #94992, taking a different approach following feedback: * Lower in PreISelIntrinsicLowering * Don't require that the number of bytes to set is a compile-time constant * Define llvm.memset_pattern rather than llvm.memset_pattern.inline As discussed in the [RFC thread](https://discourse.llvm.org/t/rfc-introducing-an-llvm-memset-pattern-inline-intrinsic/79496), the intent is that the intrinsic will be lowered to loops, a sequence of stores, or libcalls depending on the expected cost and availability of libcalls on the target. Right now, there's just a single lowering path that aims to handle all cases. My intent would be to follow up with additional PRs that add additional optimisations when possible (e.g. when libcalls are available, when arguments are known to be constant etc).
25 lines
1.2 KiB
LLVM
25 lines
1.2 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: opt -mtriple=powerpc64 -passes=pre-isel-intrinsic-lowering -S -o - %s 2>&1 | FileCheck %s
|
|
|
|
; Simple smoke test that memset.pattern is still expanded on big endian
|
|
; targets.
|
|
|
|
define void @memset.pattern(ptr %a, i128 %value, i64 %x) nounwind {
|
|
; CHECK-LABEL: define void @memset.pattern(
|
|
; CHECK-SAME: ptr [[A:%.*]], i128 [[VALUE:%.*]], i64 [[X:%.*]]) #[[ATTR0:[0-9]+]] {
|
|
; CHECK-NEXT: [[TMP1:%.*]] = icmp eq i64 0, [[X]]
|
|
; CHECK-NEXT: br i1 [[TMP1]], label %[[SPLIT:.*]], label %[[LOADSTORELOOP:.*]]
|
|
; CHECK: [[LOADSTORELOOP]]:
|
|
; CHECK-NEXT: [[TMP3:%.*]] = phi i64 [ 0, [[TMP0:%.*]] ], [ [[TMP4:%.*]], %[[LOADSTORELOOP]] ]
|
|
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i128, ptr [[A]], i64 [[TMP3]]
|
|
; CHECK-NEXT: store i128 [[VALUE]], ptr [[TMP2]], align 1
|
|
; CHECK-NEXT: [[TMP4]] = add i64 [[TMP3]], 1
|
|
; CHECK-NEXT: [[TMP5:%.*]] = icmp ult i64 [[TMP4]], [[X]]
|
|
; CHECK-NEXT: br i1 [[TMP5]], label %[[LOADSTORELOOP]], label %[[SPLIT]]
|
|
; CHECK: [[SPLIT]]:
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
tail call void @llvm.experimental.memset.pattern(ptr %a, i128 %value, i64 %x, i1 0)
|
|
ret void
|
|
}
|