llvm-project/llvm/test/CodeGen/X86/liveness-local-regalloc.ll

93 lines
3.2 KiB
LLVM
Raw Normal View History

; RUN: llc < %s -regalloc=fast -optimize-regalloc=0 -verify-machineinstrs -mtriple=x86_64-apple-darwin10
; <rdar://problem/7755473>
; PR12821
%0 = type { i32, i8*, i8*, %1*, i8*, i64, i64, i32, i32, i32, i32, [1024 x i8] }
%1 = type { i8*, i32, i32, i16, i16, %2, i32, i8*, i32 (i8*)*, i32 (i8*, i8*, i32)*, i64 (i8*, i64, i32)*, i32 (i8*, i8*, i32)*, %2, %3*, i32, [3 x i8], [1 x i8], %2, i32, i64 }
%2 = type { i8*, i32 }
%3 = type opaque
declare fastcc i32 @func(%0*, i32, i32) nounwind ssp
define fastcc void @func2(%0* %arg, i32 %arg1) nounwind ssp {
bb:
br label %.exit3
.exit3: ; preds = %.exit3, %bb
switch i32 undef, label %.exit3 [
i32 -1, label %.loopexit
i32 37, label %bb2
]
bb2: ; preds = %bb5, %bb3, %.exit3
br i1 undef, label %bb3, label %bb5
bb3: ; preds = %bb2
switch i32 undef, label %infloop [
i32 125, label %.loopexit
i32 -1, label %bb4
i32 37, label %bb2
]
bb4: ; preds = %bb3
%tmp = add nsw i32 undef, 1 ; <i32> [#uses=1]
br label %.loopexit
bb5: ; preds = %bb2
switch i32 undef, label %infloop1 [
i32 -1, label %.loopexit
i32 37, label %bb2
]
.loopexit: ; preds = %bb5, %bb4, %bb3, %.exit3
%.04 = phi i32 [ %tmp, %bb4 ], [ undef, %bb3 ], [ undef, %.exit3 ], [ undef, %bb5 ] ; <i32> [#uses=2]
br i1 undef, label %bb8, label %bb6
bb6: ; preds = %.loopexit
%tmp7 = tail call fastcc i32 @func(%0* %arg, i32 %.04, i32 undef) nounwind ssp ; <i32> [#uses=0]
ret void
bb8: ; preds = %.loopexit
%tmp9 = sext i32 %.04 to i64 ; <i64> [#uses=1]
[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 llvm-svn: 230786
2015-02-28 03:29:02 +08:00
%tmp10 = getelementptr inbounds %0, %0* %arg, i64 0, i32 11, i64 %tmp9 ; <i8*> [#uses=1]
store i8 0, i8* %tmp10, align 1
ret void
infloop: ; preds = %infloop, %bb3
br label %infloop
infloop1: ; preds = %infloop1, %bb5
br label %infloop1
}
; RAFast would forget to add a super-register <imp-def> when rewriting:
; %vreg10:sub_32bit<def,read-undef> = COPY %R9D<kill>
; This trips up the machine code verifier.
define void @autogen_SD24657(i8*, i32*, i64*, i32, i64, i8) {
BB:
%A4 = alloca <16 x i16>
%A3 = alloca double
%A2 = alloca <2 x i8>
%A1 = alloca i1
%A = alloca i32
%L = load i8, i8* %0
store i8 -37, i8* %0
%E = extractelement <4 x i64> zeroinitializer, i32 2
%Shuff = shufflevector <4 x i64> zeroinitializer, <4 x i64> zeroinitializer, <4 x i32> <i32 5, i32 7, i32 1, i32 3>
%I = insertelement <2 x i8> <i8 -1, i8 -1>, i8 %5, i32 1
%B = fadd float 0x45CDF5B1C0000000, 0x45CDF5B1C0000000
%FC = uitofp i32 275048 to double
%Sl = select i1 true, <2 x i8> %I, <2 x i8> <i8 -1, i8 -1>
%Cmp = icmp slt i64 0, %E
br label %CF
CF: ; preds = %BB
store i8 %5, i8* %0
store <2 x i8> %I, <2 x i8>* %A2
store i8 %5, i8* %0
store i8 %5, i8* %0
store i8 %5, i8* %0
ret void
}