diff options
| author | Daniel Berlin <dberlin@dberlin.org> | 2016-06-01 18:55:32 +0000 |
|---|---|---|
| committer | Daniel Berlin <dberlin@dberlin.org> | 2016-06-01 18:55:32 +0000 |
| commit | 73694bb92bfbe55ee2c7afc0e3cf4a186741aeb9 (patch) | |
| tree | 25588ba85835d0b4ee91f773f570edbe08e48d04 /llvm/test/Transforms | |
| parent | 18b83fe6cf7848d66f2797c231a81e96c2f61d2e (diff) | |
| download | bcm5719-llvm-73694bb92bfbe55ee2c7afc0e3cf4a186741aeb9.tar.gz bcm5719-llvm-73694bb92bfbe55ee2c7afc0e3cf4a186741aeb9.zip | |
Revert "Claim NoAlias if two GEPs index different fields of the same struct"
This reverts commit 2d5d6493f43eb68493a3852b8c226ac9fafdc7eb.
llvm-svn: 271422
Diffstat (limited to 'llvm/test/Transforms')
| -rw-r--r-- | llvm/test/Transforms/LoopVectorize/global_alias.ll | 158 |
1 files changed, 81 insertions, 77 deletions
diff --git a/llvm/test/Transforms/LoopVectorize/global_alias.ll b/llvm/test/Transforms/LoopVectorize/global_alias.ll index ae4e121cedf..0da841bcbbd 100644 --- a/llvm/test/Transforms/LoopVectorize/global_alias.ll +++ b/llvm/test/Transforms/LoopVectorize/global_alias.ll @@ -765,18 +765,22 @@ for.end: ; preds = %for.cond ret i32 %7 } -; /// Different objects, swapped induction -; int noAlias15(int a) { + +;; === Now, the tests that we could vectorize with induction changes or run-time checks === + + +; /// Different objects, swapped induction, alias at the end +; int mayAlias01 (int a) { ; int i; ; for (i=0; i<SIZE; i++) ; Foo.A[i] = Foo.B[SIZE-i-1] + a; ; return Foo.A[a]; ; } -; CHECK-LABEL: define i32 @noAlias15( -; CHECK: add nsw <4 x i32> +; CHECK-LABEL: define i32 @mayAlias01( +; CHECK-NOT: add nsw <4 x i32> ; CHECK: ret -define i32 @noAlias15(i32 %a) nounwind { +define i32 @mayAlias01(i32 %a) nounwind { entry: %a.addr = alloca i32, align 4 %i = alloca i32, align 4 @@ -815,18 +819,18 @@ for.end: ; preds = %for.cond ret i32 %7 } -; /// Different objects, swapped induction -; int noAlias16 (int a) { +; /// Different objects, swapped induction, alias at the beginning +; int mayAlias02 (int a) { ; int i; ; for (i=0; i<SIZE; i++) ; Foo.A[SIZE-i-1] = Foo.B[i] + a; ; return Foo.A[a]; ; } -; CHECK-LABEL: define i32 @noAlias16( -; CHECK: add nsw <4 x i32> +; CHECK-LABEL: define i32 @mayAlias02( +; CHECK-NOT: add nsw <4 x i32> ; CHECK: ret -define i32 @noAlias16(i32 %a) nounwind { +define i32 @mayAlias02(i32 %a) nounwind { entry: %a.addr = alloca i32, align 4 %i = alloca i32, align 4 @@ -865,21 +869,75 @@ for.end: ; preds = %for.cond ret i32 %7 } +; /// Pointer access, run-time check added +; int mayAlias03 (int a) { +; int i; +; for (i=0; i<SIZE; i++) +; *(PA+i) = *(PB+SIZE-i-1) + a; +; return *(PA+a); +; } +; CHECK-LABEL: define i32 @mayAlias03( +; CHECK-NOT: add nsw <4 x i32> +; CHECK: ret + +define i32 @mayAlias03(i32 %a) nounwind { +entry: + %a.addr = alloca i32, align 4 + %i = alloca i32, align 4 + store i32 %a, i32* %a.addr, align 4 + store i32 0, i32* %i, align 4 + br label %for.cond + +for.cond: ; preds = %for.inc, %entry + %0 = load i32, i32* %i, align 4 + %cmp = icmp slt i32 %0, 100 + br i1 %cmp, label %for.body, label %for.end + +for.body: ; preds = %for.cond + %1 = load i32*, i32** @PB, align 4 + %add.ptr = getelementptr inbounds i32, i32* %1, i32 100 + %2 = load i32, i32* %i, align 4 + %idx.neg = sub i32 0, %2 + %add.ptr1 = getelementptr inbounds i32, i32* %add.ptr, i32 %idx.neg + %add.ptr2 = getelementptr inbounds i32, i32* %add.ptr1, i32 -1 + %3 = load i32, i32* %add.ptr2, align 4 + %4 = load i32, i32* %a.addr, align 4 + %add = add nsw i32 %3, %4 + %5 = load i32*, i32** @PA, align 4 + %6 = load i32, i32* %i, align 4 + %add.ptr3 = getelementptr inbounds i32, i32* %5, i32 %6 + store i32 %add, i32* %add.ptr3, align 4 + br label %for.inc + +for.inc: ; preds = %for.body + %7 = load i32, i32* %i, align 4 + %inc = add nsw i32 %7, 1 + store i32 %inc, i32* %i, align 4 + br label %for.cond + +for.end: ; preds = %for.cond + %8 = load i32*, i32** @PA, align 4 + %9 = load i32, i32* %a.addr, align 4 + %add.ptr4 = getelementptr inbounds i32, i32* %8, i32 %9 + %10 = load i32, i32* %add.ptr4, align 4 + ret i32 %10 +} + -;; === Ignore undefined behavior === +;; === Finally, the tests that should only vectorize with care (or if we ignore undefined behaviour at all) === -; int noAlias17(int a) { +; int mustAlias01 (int a) { ; int i; ; for (i=0; i<SIZE; i++) ; Foo.A[i+10] = Foo.B[SIZE-i-1] + a; ; return Foo.A[a]; ; } -; CHECK-LABEL: define i32 @noAlias17( -; CHECK: add nsw <4 x i32> +; CHECK-LABEL: define i32 @mustAlias01( +; CHECK-NOT: add nsw <4 x i32> ; CHECK: ret -define i32 @noAlias17(i32 %a) nounwind { +define i32 @mustAlias01(i32 %a) nounwind { entry: %a.addr = alloca i32, align 4 %i = alloca i32, align 4 @@ -919,17 +977,17 @@ for.end: ; preds = %for.cond ret i32 %7 } -; int noAlias18(int a) { +; int mustAlias02 (int a) { ; int i; ; for (i=0; i<SIZE; i++) ; Foo.A[i] = Foo.B[SIZE-i-10] + a; ; return Foo.A[a]; ; } -; CHECK-LABEL: define i32 @noAlias18( -; CHECK: add nsw <4 x i32> +; CHECK-LABEL: define i32 @mustAlias02( +; CHECK-NOT: add nsw <4 x i32> ; CHECK: ret -define i32 @noAlias18(i32 %a) nounwind { +define i32 @mustAlias02(i32 %a) nounwind { entry: %a.addr = alloca i32, align 4 %i = alloca i32, align 4 @@ -968,17 +1026,17 @@ for.end: ; preds = %for.cond ret i32 %7 } -; int noAlias19(int a) { +; int mustAlias03 (int a) { ; int i; ; for (i=0; i<SIZE; i++) ; Foo.A[i+10] = Foo.B[SIZE-i-10] + a; ; return Foo.A[a]; ; } -; CHECK-LABEL: define i32 @noAlias19( -; CHECK: add nsw <4 x i32> +; CHECK-LABEL: define i32 @mustAlias03( +; CHECK-NOT: add nsw <4 x i32> ; CHECK: ret -define i32 @noAlias19(i32 %a) nounwind { +define i32 @mustAlias03(i32 %a) nounwind { entry: %a.addr = alloca i32, align 4 %i = alloca i32, align 4 @@ -1017,57 +1075,3 @@ for.end: ; preds = %for.cond %7 = load i32, i32* %arrayidx4, align 4 ret i32 %7 } - -; /// Pointer access, run-time check added -; int mayAlias01 (int a) { -; int i; -; for (i=0; i<SIZE; i++) -; *(PA+i) = *(PB+SIZE-i-1) + a; -; return *(PA+a); -; } -; CHECK-LABEL: define i32 @mayAlias01( -; CHECK-NOT: add nsw <4 x i32> -; CHECK: ret - -define i32 @mayAlias01(i32 %a) nounwind { -entry: - %a.addr = alloca i32, align 4 - %i = alloca i32, align 4 - store i32 %a, i32* %a.addr, align 4 - store i32 0, i32* %i, align 4 - br label %for.cond - -for.cond: ; preds = %for.inc, %entry - %0 = load i32, i32* %i, align 4 - %cmp = icmp slt i32 %0, 100 - br i1 %cmp, label %for.body, label %for.end - -for.body: ; preds = %for.cond - %1 = load i32*, i32** @PB, align 4 - %add.ptr = getelementptr inbounds i32, i32* %1, i32 100 - %2 = load i32, i32* %i, align 4 - %idx.neg = sub i32 0, %2 - %add.ptr1 = getelementptr inbounds i32, i32* %add.ptr, i32 %idx.neg - %add.ptr2 = getelementptr inbounds i32, i32* %add.ptr1, i32 -1 - %3 = load i32, i32* %add.ptr2, align 4 - %4 = load i32, i32* %a.addr, align 4 - %add = add nsw i32 %3, %4 - %5 = load i32*, i32** @PA, align 4 - %6 = load i32, i32* %i, align 4 - %add.ptr3 = getelementptr inbounds i32, i32* %5, i32 %6 - store i32 %add, i32* %add.ptr3, align 4 - br label %for.inc - -for.inc: ; preds = %for.body - %7 = load i32, i32* %i, align 4 - %inc = add nsw i32 %7, 1 - store i32 %inc, i32* %i, align 4 - br label %for.cond - -for.end: ; preds = %for.cond - %8 = load i32*, i32** @PA, align 4 - %9 = load i32, i32* %a.addr, align 4 - %add.ptr4 = getelementptr inbounds i32, i32* %8, i32 %9 - %10 = load i32, i32* %add.ptr4, align 4 - ret i32 %10 -} |

