[SystemZ] Add LOC and LOCG

As with the stores, these instructions can trap when the condition is false,
so they are only used for things like (cond ? x : *ptr).

llvm-svn: 187112

2 files changed, 260 insertions, 0 deletions

diff --git a/llvm/test/CodeGen/SystemZ/cond-load-01.ll b/llvm/test/CodeGen/SystemZ/cond-load-01.ll
new file mode 100644
index 00000000000..59ed90c99e6
--- /dev/null
+++ b/llvm/test/CodeGen/SystemZ/cond-load-01.ll
@@ -0,0 +1,130 @@
+; Test LOC.
+;
+; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z196 | FileCheck %s
+
+declare i32 @foo(i32 *)
+
+; Test the simple case.
+define i32 @f1(i32 %easy, i32 *%ptr, i32 %limit) {
+; CHECK-LABEL: f1:
+; CHECK: clfi %r4, 42
+; CHECK: locnl %r2, 0(%r3)
+; CHECK: br %r14
+  %cond = icmp ult i32 %limit, 42
+  %other = load i32 *%ptr
+  %res = select i1 %cond, i32 %easy, i32 %other
+  ret i32 %res
+}
+
+; ...and again with the operands swapped.
+define i32 @f2(i32 %easy, i32 *%ptr, i32 %limit) {
+; CHECK-LABEL: f2:
+; CHECK: clfi %r4, 42
+; CHECK: locl %r2, 0(%r3)
+; CHECK: br %r14
+  %cond = icmp ult i32 %limit, 42
+  %other = load i32 *%ptr
+  %res = select i1 %cond, i32 %other, i32 %easy
+  ret i32 %res
+}
+
+; Check the high end of the aligned LOC range.
+define i32 @f3(i32 %easy, i32 *%base, i32 %limit) {
+; CHECK-LABEL: f3:
+; CHECK: clfi %r4, 42
+; CHECK: locnl %r2, 524284(%r3)
+; CHECK: br %r14
+  %ptr = getelementptr i32 *%base, i64 131071
+  %cond = icmp ult i32 %limit, 42
+  %other = load i32 *%ptr
+  %res = select i1 %cond, i32 %easy, i32 %other
+  ret i32 %res
+}
+
+; Check the next word up.  Other sequences besides this one would be OK.
+define i32 @f4(i32 %easy, i32 *%base, i32 %limit) {
+; CHECK-LABEL: f4:
+; CHECK: agfi %r3, 524288
+; CHECK: clfi %r4, 42
+; CHECK: locnl %r2, 0(%r3)
+; CHECK: br %r14
+  %ptr = getelementptr i32 *%base, i64 131072
+  %cond = icmp ult i32 %limit, 42
+  %other = load i32 *%ptr
+  %res = select i1 %cond, i32 %easy, i32 %other
+  ret i32 %res
+}
+
+; Check the low end of the LOC range.
+define i32 @f5(i32 %easy, i32 *%base, i32 %limit) {
+; CHECK-LABEL: f5:
+; CHECK: clfi %r4, 42
+; CHECK: locnl %r2, -524288(%r3)
+; CHECK: br %r14
+  %ptr = getelementptr i32 *%base, i64 -131072
+  %cond = icmp ult i32 %limit, 42
+  %other = load i32 *%ptr
+  %res = select i1 %cond, i32 %easy, i32 %other
+  ret i32 %res
+}
+
+; Check the next word down, with the same comments as f4.
+define i32 @f6(i32 %easy, i32 *%base, i32 %limit) {
+; CHECK-LABEL: f6:
+; CHECK: agfi %r3, -524292
+; CHECK: clfi %r4, 42
+; CHECK: locnl %r2, 0(%r3)
+; CHECK: br %r14
+  %ptr = getelementptr i32 *%base, i64 -131073
+  %cond = icmp ult i32 %limit, 42
+  %other = load i32 *%ptr
+  %res = select i1 %cond, i32 %easy, i32 %other
+  ret i32 %res
+}
+
+; Try a frame index base.
+define i32 @f7(i32 %alt, i32 %limit) {
+; CHECK-LABEL: f7:
+; CHECK: brasl %r14, foo@PLT
+; CHECK: locnl %r2, {{[0-9]+}}(%r15)
+; CHECK: br %r14
+  %ptr = alloca i32
+  %easy = call i32 @foo(i32 *%ptr)
+  %cond = icmp ult i32 %limit, 42
+  %other = load i32 *%ptr
+  %res = select i1 %cond, i32 %easy, i32 %other
+  ret i32 %res
+}
+
+; Try a case when an index is involved.
+define i32 @f8(i32 %easy, i32 %limit, i64 %base, i64 %index) {
+; CHECK-LABEL: f8:
+; CHECK: clfi %r3, 42
+; CHECK: locnl %r2, 0({{%r[1-5]}})
+; CHECK: br %r14
+  %add = add i64 %base, %index
+  %ptr = inttoptr i64 %add to i32 *
+  %cond = icmp ult i32 %limit, 42
+  %other = load i32 *%ptr
+  %res = select i1 %cond, i32 %easy, i32 %other
+  ret i32 %res
+}
+
+; Test that conditionally-executed loads do not use LOC, since it is allowed
+; to trap even when the condition is false.
+define i32 @f9(i32 %easy, i32 %limit, i32 *%ptr) {
+; CHECK-LABEL: f9:
+; CHECK-NOT: loc
+; CHECK: br %r14
+entry:
+  %cmp = icmp ule i32 %easy, %limit
+  br i1 %cmp, label %load, label %exit
+
+load:
+  %other = load i32 *%ptr
+  br label %exit
+
+exit:
+  %res = phi i32 [ %easy, %entry ], [ %other, %load ]
+  ret i32 %res
+}
diff --git a/llvm/test/CodeGen/SystemZ/cond-load-02.ll b/llvm/test/CodeGen/SystemZ/cond-load-02.ll
new file mode 100644
index 00000000000..50adb3507dc
--- /dev/null
+++ b/llvm/test/CodeGen/SystemZ/cond-load-02.ll
@@ -0,0 +1,130 @@
+; Test LOCG.
+;
+; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z196 | FileCheck %s
+
+declare i64 @foo(i64 *)
+
+; Test the simple case.
+define i64 @f1(i64 %easy, i64 *%ptr, i64 %limit) {
+; CHECK-LABEL: f1:
+; CHECK: clgfi %r4, 42
+; CHECK: locgnl %r2, 0(%r3)
+; CHECK: br %r14
+  %cond = icmp ult i64 %limit, 42
+  %other = load i64 *%ptr
+  %res = select i1 %cond, i64 %easy, i64 %other
+  ret i64 %res
+}
+
+; ...and again with the operands swapped.
+define i64 @f2(i64 %easy, i64 *%ptr, i64 %limit) {
+; CHECK-LABEL: f2:
+; CHECK: clgfi %r4, 42
+; CHECK: locgl %r2, 0(%r3)
+; CHECK: br %r14
+  %cond = icmp ult i64 %limit, 42
+  %other = load i64 *%ptr
+  %res = select i1 %cond, i64 %other, i64 %easy
+  ret i64 %res
+}
+
+; Check the high end of the aligned LOCG range.
+define i64 @f3(i64 %easy, i64 *%base, i64 %limit) {
+; CHECK-LABEL: f3:
+; CHECK: clgfi %r4, 42
+; CHECK: locgnl %r2, 524280(%r3)
+; CHECK: br %r14
+  %ptr = getelementptr i64 *%base, i64 65535
+  %cond = icmp ult i64 %limit, 42
+  %other = load i64 *%ptr
+  %res = select i1 %cond, i64 %easy, i64 %other
+  ret i64 %res
+}
+
+; Check the next doubleword up.  Other sequences besides this one would be OK.
+define i64 @f4(i64 %easy, i64 *%base, i64 %limit) {
+; CHECK-LABEL: f4:
+; CHECK: agfi %r3, 524288
+; CHECK: clgfi %r4, 42
+; CHECK: locgnl %r2, 0(%r3)
+; CHECK: br %r14
+  %ptr = getelementptr i64 *%base, i64 65536
+  %cond = icmp ult i64 %limit, 42
+  %other = load i64 *%ptr
+  %res = select i1 %cond, i64 %easy, i64 %other
+  ret i64 %res
+}
+
+; Check the low end of the LOCG range.
+define i64 @f5(i64 %easy, i64 *%base, i64 %limit) {
+; CHECK-LABEL: f5:
+; CHECK: clgfi %r4, 42
+; CHECK: locgnl %r2, -524288(%r3)
+; CHECK: br %r14
+  %ptr = getelementptr i64 *%base, i64 -65536
+  %cond = icmp ult i64 %limit, 42
+  %other = load i64 *%ptr
+  %res = select i1 %cond, i64 %easy, i64 %other
+  ret i64 %res
+}
+
+; Check the next doubleword down, with the same comments as f4.
+define i64 @f6(i64 %easy, i64 *%base, i64 %limit) {
+; CHECK-LABEL: f6:
+; CHECK: agfi %r3, -524296
+; CHECK: clgfi %r4, 42
+; CHECK: locgnl %r2, 0(%r3)
+; CHECK: br %r14
+  %ptr = getelementptr i64 *%base, i64 -65537
+  %cond = icmp ult i64 %limit, 42
+  %other = load i64 *%ptr
+  %res = select i1 %cond, i64 %easy, i64 %other
+  ret i64 %res
+}
+
+; Try a frame index base.
+define i64 @f7(i64 %alt, i64 %limit) {
+; CHECK-LABEL: f7:
+; CHECK: brasl %r14, foo@PLT
+; CHECK: locgnl %r2, {{[0-9]+}}(%r15)
+; CHECK: br %r14
+  %ptr = alloca i64
+  %easy = call i64 @foo(i64 *%ptr)
+  %cond = icmp ult i64 %limit, 42
+  %other = load i64 *%ptr
+  %res = select i1 %cond, i64 %easy, i64 %other
+  ret i64 %res
+}
+
+; Try a case when an index is involved.
+define i64 @f8(i64 %easy, i64 %limit, i64 %base, i64 %index) {
+; CHECK-LABEL: f8:
+; CHECK: clgfi %r3, 42
+; CHECK: locgnl %r2, 0({{%r[1-5]}})
+; CHECK: br %r14
+  %add = add i64 %base, %index
+  %ptr = inttoptr i64 %add to i64 *
+  %cond = icmp ult i64 %limit, 42
+  %other = load i64 *%ptr
+  %res = select i1 %cond, i64 %easy, i64 %other
+  ret i64 %res
+}
+
+; Test that conditionally-executed loads do not use LOCG, since it is allowed
+; to trap even when the condition is false.
+define i64 @f9(i64 %easy, i64 %limit, i64 *%ptr) {
+; CHECK-LABEL: f9:
+; CHECK-NOT: locg
+; CHECK: br %r14
+entry:
+  %cmp = icmp ule i64 %easy, %limit
+  br i1 %cmp, label %load, label %exit
+
+load:
+  %other = load i64 *%ptr
+  br label %exit
+
+exit:
+  %res = phi i64 [ %easy, %entry ], [ %other, %load ]
+  ret i64 %res
+}