[PowerPC] Recurse through constants when looking for TLS globals

We cannot form ctr-based loops around function calls, including calls to
__tls_get_addr used for PIC TLS variables. References to such TLS variables,
however, might be buried within constant expressions, and so we need to search
the entire constant expression to be sure that no references to such TLS
variables exist.

Fixes PR25256, reported by Eric Schweitz. This is a slightly-modified version
of the patch suggested by Eric in the bug report, and a test case I created.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@251582 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/PowerPC/PPCCTRLoops.cpp b/lib/Target/PowerPC/PPCCTRLoops.cpp
index a9687adfd379bb0ac95d8ff153b3b4e6cec77a2a..58ccb4a4cdcb084ec915a3ca2ef43c8ce7d94ee5 100644 (file)
--- a/lib/Target/PowerPC/PPCCTRLoops.cpp
+++ b/lib/Target/PowerPC/PPCCTRLoops.cpp
@@ -197,10 +197,18 @@ static bool isLargeIntegerTy(bool Is32Bit, Type *Ty) {
 // Determining the address of a TLS variable results in a function call in
 // certain TLS models.
 static bool memAddrUsesCTR(const PPCTargetMachine *TM,
-                           const llvm::Value *MemAddr) {
+                           const Value *MemAddr) {
   const auto *GV = dyn_cast<GlobalValue>(MemAddr);
-  if (!GV)
+  if (!GV) {
+    // Recurse to check for constants that refer to TLS global variables.
+    if (const auto *CV = dyn_cast<Constant>(MemAddr))
+      for (const auto &CO : CV->operands())
+        if (memAddrUsesCTR(TM, CO))
+          return true;
+
     return false;
+  }
+
   if (!GV->isThreadLocal())
     return false;
   if (!TM)

diff --git a/test/CodeGen/PowerPC/ctr-loop-tls-const.ll b/test/CodeGen/PowerPC/ctr-loop-tls-const.ll
new file mode 100644 (file)
index 0000000..01f837c
--- /dev/null
+++ b/test/CodeGen/PowerPC/ctr-loop-tls-const.ll
@@ -0,0 +1,40 @@
+; RUN: llc -mcpu=pwr7 -relocation-model=pic < %s | FileCheck %s
+target datalayout = "E-m:e-i64:64-n32:64"
+target triple = "powerpc64-unknown-linux-gnu"
+
+@x = thread_local global [1600 x i32] zeroinitializer, align 4
+
+; Function Attrs: nounwind
+define void @foo(i32 signext %v) #0 {
+entry:
+  br label %vector.body
+
+vector.body:                                      ; preds = %vector.body, %entry
+  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
+  %induction5 = or i64 %index, 1
+  %0 = getelementptr inbounds [1600 x i32], [1600 x i32]* @x, i64 0, i64 %index
+  %1 = getelementptr inbounds [1600 x i32], [1600 x i32]* @x, i64 0, i64 %induction5
+  %2 = load i32, i32* %0, align 4
+  %3 = load i32, i32* %1, align 4
+  %4 = add nsw i32 %2, %v
+  %5 = add nsw i32 %3, %v
+  store i32 %4, i32* %0, align 4
+  store i32 %5, i32* %1, align 4
+  %index.next = add i64 %index, 2
+  %6 = icmp eq i64 %index.next, 1600
+  br i1 %6, label %for.cond.cleanup, label %vector.body
+
+for.cond.cleanup:                                 ; preds = %vector.body
+  ret void
+}
+
+; CHECK-LABEL: @foo
+; CHECK-NOT: mtctr
+; CHECK: __tls_get_addr
+
+attributes #0 = { nounwind }
+
+!llvm.module.flags = !{!0}
+
+!0 = !{i32 1, !"PIC Level", i32 2}
+