[SystemZ] Test improvements for atomic load/store instructions (NFC). (#75630)

Improve tests for atomic loads and stores, mainly by testing 128-bit atomic load and store instructions both with and w/out natural alignment.

Author: JonPsson1

llvm/test/CodeGen/SystemZ/atomic-load-05.ll

@@ -1,14 +1,22 @@
-; Test 128-bit atomic loads.
+; Test 128-bit integer atomic loads.
 ;
 ; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
 ; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z13 | FileCheck %s
 
 define i128 @f1(ptr %src) {
 ; CHECK-LABEL: f1:
-; CHECK: lpq %r0, 0(%r3)
-; CHECK-DAG: stg %r1, 8(%r2)
-; CHECK-DAG: stg %r0, 0(%r2)
-; CHECK: br %r14
+; CHECK: # %bb.0:
+; CHECK-NEXT: lpq %r0, 0(%r3)
+; CHECK-NEXT: stg %r1, 8(%r2)
+; CHECK-NEXT: stg %r0, 0(%r2)
+; CHECK-NEXT: br %r14
   %val = load atomic i128, ptr %src seq_cst, align 16
   ret i128 %val
 }
+
+define i128 @f2(ptr %src) {
+; CHECK-LABEL: f2:
+; CHECK: brasl %r14, __atomic_load@PLT
+  %val = load atomic i128, ptr %src seq_cst, align 8
+  ret i128 %val
+}

llvm/test/CodeGen/SystemZ/atomic-load-08.ll

@@ -1,19 +1,25 @@
-; Test long double atomic loads. Expect a libcall.
+; Test long double atomic loads. These are emitted by the Clang FE as i128
+; loads with a bitcast, and this test case gets converted into that form as
+; well by the AtomicExpand pass.
 ;
 ; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
+; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z13 | FileCheck %s
 
 define void @f1(ptr %ret, ptr %src) {
 ; CHECK-LABEL: f1:
-; CHECK: lgr [[RET:%r[0-9]+]], %r2
-; CHECK: la %r4, 160(%r15)
-; CHECK: lghi %r2, 16
-; CHECK: lhi %r5, 5
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    lpq %r0, 0(%r3)
+; CHECK-NEXT:    stg %r1, 8(%r2)
+; CHECK-NEXT:    stg %r0, 0(%r2)
+; CHECK-NEXT:    br %r14
+  %val = load atomic fp128, ptr %src seq_cst, align 16
+  store fp128 %val, ptr %ret, align 8
+  ret void
+}
+
+define void @f2(ptr %ret, ptr %src) {
+; CHECK-LABEL: f2:
 ; CHECK: brasl %r14, __atomic_load@PLT
-; CHECK: ld [[FL:%f[0-9]+]], 160(%r15)
-; CHECK: ld [[FH:%f[0-9]+]], 168(%r15)
-; CHECK: std [[FL]], 0([[RET]])
-; CHECK: std [[FH]], 8([[RET]])
-; CHECK: br %r14
   %val = load atomic fp128, ptr %src seq_cst, align 8
   store fp128 %val, ptr %ret, align 8
   ret void

llvm/test/CodeGen/SystemZ/atomic-store-05.ll

@@ -1,4 +1,4 @@
-; Test 128-bit atomic stores.
+; Test 128-bit integer atomic stores.
 ;
 ; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
 ; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z13 | FileCheck %s
@@ -24,3 +24,10 @@ define void @f2(i128 %val, ptr %src) {
   store atomic i128 %val, ptr %src monotonic, align 16
   ret void
 }
+
+define void @f3(i128 %val, ptr %src) {
+; CHECK-LABEL: f3:
+; CHECK: brasl %r14, __atomic_store@PLT
+  store atomic i128 %val, ptr %src seq_cst, align 8
+  ret void
+}

llvm/test/CodeGen/SystemZ/atomic-store-06.ll

@@ -1,13 +1,17 @@
-; Test float atomic loads.
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
+; Test float atomic stores.
 ;
 ; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
 
 define void @f1(ptr %src, float %val) {
 ; CHECK-LABEL: f1:
-; CHECK: lgdr [[R:%r[0-9]+]], %f0
-; CHECK: srlg [[R]], [[R]], 32
-; CHECK: st [[R]], 0(%r2)
-; CHECK: br %r14
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    # kill: def $f0s killed $f0s def $f0d
+; CHECK-NEXT:    lgdr %r0, %f0
+; CHECK-NEXT:    srlg %r0, %r0, 32
+; CHECK-NEXT:    st %r0, 0(%r2)
+; CHECK-NEXT:    bcr 15, %r0
+; CHECK-NEXT:    br %r14
   store atomic float %val, ptr %src seq_cst, align 4
   ret void
 }

llvm/test/CodeGen/SystemZ/atomic-store-07.ll

@@ -1,11 +1,14 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
 ; Test double atomic stores.
 ;
 ; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
 
 define void @f1(ptr %dst, double %val) {
 ; CHECK-LABEL: f1:
-; CHECK: std %f0, 0(%r2)
-; CHECK: br %r14
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    std %f0, 0(%r2)
+; CHECK-NEXT:    bcr 15, %r0
+; CHECK-NEXT:    br %r14
   store atomic double %val, ptr %dst seq_cst, align 8
   ret void
 }

llvm/test/CodeGen/SystemZ/atomic-store-08.ll

@@ -1,19 +1,25 @@
-; Test long double atomic stores. Expect a libcall.
+; Test long double atomic stores. The atomic store is converted to i128 by
+; the AtomicExpand pass.
 ;
 ; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
+; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z13 | FileCheck %s
 
 define void @f1(ptr %dst, ptr %src) {
 ; CHECK-LABEL: f1:
-; CHECK: ld [[FL:%f[0-9]+]], 0(%r3)
-; CHECK: ld [[FH:%f[0-9]+]], 8(%r3)
-; CHECK: lgr %r3, %r2
-; CHECK: std [[FL]], 160(%r15)
-; CHECK: std [[FH]], 168(%r15)
-; CHECK: la %r4, 160(%r15)
-; CHECK: lghi %r2, 16
-; CHECK: lhi %r5, 5
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    lg %r1, 8(%r3)
+; CHECK-NEXT:    lg %r0, 0(%r3)
+; CHECK-NEXT:    stpq %r0, 0(%r2)
+; CHECK-NEXT:    bcr 1{{[45]}}, %r0
+; CHECK-NEXT:    br %r14
+  %val = load fp128, ptr %src, align 8
+  store atomic fp128 %val, ptr %dst seq_cst, align 16
+  ret void
+}
+
+define void @f2(ptr %dst, ptr %src) {
+; CHECK-LABEL: f2:
 ; CHECK: brasl %r14, __atomic_store@PLT
-; CHECK: br %r14
   %val = load fp128, ptr %src, align 8
   store atomic fp128 %val, ptr %dst seq_cst, align 8
   ret void