llvm-6502/test/CodeGen/AArch64/arm64-prefetch.ll
David Blaikie 198d8baafb [opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.

This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.

* This doesn't modify gep operators, only instructions (operators will be
  handled separately)

* Textual IR changes only. Bitcode (including upgrade) and changing the
  in-memory representation will be in separate changes.

* geps of vectors are transformed as:
    getelementptr <4 x float*> %x, ...
  ->getelementptr float, <4 x float*> %x, ...
  Then, once the opaque pointer type is introduced, this will ultimately look
  like:
    getelementptr float, <4 x ptr> %x
  with the unambiguous interpretation that it is a vector of pointers to float.

* address spaces remain on the pointer, not the type:
    getelementptr float addrspace(1)* %x
  ->getelementptr float, float addrspace(1)* %x
  Then, eventually:
    getelementptr float, ptr addrspace(1) %x

Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.

update.py:
import fileinput
import sys
import re

ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile(       r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")

def conv(match, line):
  if not match:
    return line
  line = match.groups()[0]
  if len(match.groups()[5]) == 0:
    line += match.groups()[2]
  line += match.groups()[3]
  line += ", "
  line += match.groups()[1]
  line += "\n"
  return line

for line in sys.stdin:
  if line.find("getelementptr ") == line.find("getelementptr inbounds"):
    if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
      line = conv(re.match(ibrep, line), line)
  elif line.find("getelementptr ") != line.find("getelementptr ("):
    line = conv(re.match(normrep, line), line)
  sys.stdout.write(line)

apply.sh:
for name in "$@"
do
  python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
  rm -f "$name.tmp"
done

The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh

After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).

The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.

Reviewers: rafael, dexonsmith, grosser

Differential Revision: http://reviews.llvm.org/D7636

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@230786 91177308-0d34-0410-b5e6-96231b3b80d8
2015-02-27 19:29:02 +00:00

124 lines
4.4 KiB
LLVM

; RUN: llc %s -march arm64 -o - | FileCheck %s
@a = common global i32* null, align 8
define void @test(i32 %i, i32 %j) nounwind ssp {
entry:
; CHECK: @test
%j.addr = alloca i32, align 4
store i32 %j, i32* %j.addr, align 4, !tbaa !0
%tmp = bitcast i32* %j.addr to i8*
; CHECK: prfum pldl1strm
call void @llvm.prefetch(i8* %tmp, i32 0, i32 0, i32 1)
; CHECK: prfum pldl3keep
call void @llvm.prefetch(i8* %tmp, i32 0, i32 1, i32 1)
; CHECK: prfum pldl2keep
call void @llvm.prefetch(i8* %tmp, i32 0, i32 2, i32 1)
; CHECK: prfum pldl1keep
call void @llvm.prefetch(i8* %tmp, i32 0, i32 3, i32 1)
; CHECK: prfum plil1strm
call void @llvm.prefetch(i8* %tmp, i32 0, i32 0, i32 0)
; CHECK: prfum plil3keep
call void @llvm.prefetch(i8* %tmp, i32 0, i32 1, i32 0)
; CHECK: prfum plil2keep
call void @llvm.prefetch(i8* %tmp, i32 0, i32 2, i32 0)
; CHECK: prfum plil1keep
call void @llvm.prefetch(i8* %tmp, i32 0, i32 3, i32 0)
; CHECK: prfum pstl1strm
call void @llvm.prefetch(i8* %tmp, i32 1, i32 0, i32 1)
; CHECK: prfum pstl3keep
call void @llvm.prefetch(i8* %tmp, i32 1, i32 1, i32 1)
; CHECK: prfum pstl2keep
call void @llvm.prefetch(i8* %tmp, i32 1, i32 2, i32 1)
; CHECK: prfum pstl1keep
call void @llvm.prefetch(i8* %tmp, i32 1, i32 3, i32 1)
%tmp1 = load i32* %j.addr, align 4, !tbaa !0
%add = add nsw i32 %tmp1, %i
%idxprom = sext i32 %add to i64
%tmp2 = load i32** @a, align 8, !tbaa !3
%arrayidx = getelementptr inbounds i32, i32* %tmp2, i64 %idxprom
%tmp3 = bitcast i32* %arrayidx to i8*
; CHECK: prfm pldl1strm
call void @llvm.prefetch(i8* %tmp3, i32 0, i32 0, i32 1)
%tmp4 = load i32** @a, align 8, !tbaa !3
%arrayidx3 = getelementptr inbounds i32, i32* %tmp4, i64 %idxprom
%tmp5 = bitcast i32* %arrayidx3 to i8*
; CHECK: prfm pldl3keep
call void @llvm.prefetch(i8* %tmp5, i32 0, i32 1, i32 1)
%tmp6 = load i32** @a, align 8, !tbaa !3
%arrayidx6 = getelementptr inbounds i32, i32* %tmp6, i64 %idxprom
%tmp7 = bitcast i32* %arrayidx6 to i8*
; CHECK: prfm pldl2keep
call void @llvm.prefetch(i8* %tmp7, i32 0, i32 2, i32 1)
%tmp8 = load i32** @a, align 8, !tbaa !3
%arrayidx9 = getelementptr inbounds i32, i32* %tmp8, i64 %idxprom
%tmp9 = bitcast i32* %arrayidx9 to i8*
; CHECK: prfm pldl1keep
call void @llvm.prefetch(i8* %tmp9, i32 0, i32 3, i32 1)
%tmp10 = load i32** @a, align 8, !tbaa !3
%arrayidx12 = getelementptr inbounds i32, i32* %tmp10, i64 %idxprom
%tmp11 = bitcast i32* %arrayidx12 to i8*
; CHECK: prfm plil1strm
call void @llvm.prefetch(i8* %tmp11, i32 0, i32 0, i32 0)
%tmp12 = load i32** @a, align 8, !tbaa !3
%arrayidx15 = getelementptr inbounds i32, i32* %tmp12, i64 %idxprom
%tmp13 = bitcast i32* %arrayidx3 to i8*
; CHECK: prfm plil3keep
call void @llvm.prefetch(i8* %tmp13, i32 0, i32 1, i32 0)
%tmp14 = load i32** @a, align 8, !tbaa !3
%arrayidx18 = getelementptr inbounds i32, i32* %tmp14, i64 %idxprom
%tmp15 = bitcast i32* %arrayidx6 to i8*
; CHECK: prfm plil2keep
call void @llvm.prefetch(i8* %tmp15, i32 0, i32 2, i32 0)
%tmp16 = load i32** @a, align 8, !tbaa !3
%arrayidx21 = getelementptr inbounds i32, i32* %tmp16, i64 %idxprom
%tmp17 = bitcast i32* %arrayidx9 to i8*
; CHECK: prfm plil1keep
call void @llvm.prefetch(i8* %tmp17, i32 0, i32 3, i32 0)
%tmp18 = load i32** @a, align 8, !tbaa !3
%arrayidx24 = getelementptr inbounds i32, i32* %tmp18, i64 %idxprom
%tmp19 = bitcast i32* %arrayidx12 to i8*
; CHECK: prfm pstl1strm
call void @llvm.prefetch(i8* %tmp19, i32 1, i32 0, i32 1)
%tmp20 = load i32** @a, align 8, !tbaa !3
%arrayidx27 = getelementptr inbounds i32, i32* %tmp20, i64 %idxprom
%tmp21 = bitcast i32* %arrayidx15 to i8*
; CHECK: prfm pstl3keep
call void @llvm.prefetch(i8* %tmp21, i32 1, i32 1, i32 1)
%tmp22 = load i32** @a, align 8, !tbaa !3
%arrayidx30 = getelementptr inbounds i32, i32* %tmp22, i64 %idxprom
%tmp23 = bitcast i32* %arrayidx18 to i8*
; CHECK: prfm pstl2keep
call void @llvm.prefetch(i8* %tmp23, i32 1, i32 2, i32 1)
%tmp24 = load i32** @a, align 8, !tbaa !3
%arrayidx33 = getelementptr inbounds i32, i32* %tmp24, i64 %idxprom
%tmp25 = bitcast i32* %arrayidx21 to i8*
; CHECK: prfm pstl1keep
call void @llvm.prefetch(i8* %tmp25, i32 1, i32 3, i32 1)
ret void
}
declare void @llvm.prefetch(i8* nocapture, i32, i32, i32) nounwind
!0 = !{!"int", !1}
!1 = !{!"omnipotent char", !2}
!2 = !{!"Simple C/C++ TBAA"}
!3 = !{!"any pointer", !1}