From: Tim Northover <tnorthover@apple.com>
Date: Wed, 5 Nov 2014 00:27:20 +0000 (+0000)
Subject: ARM: try to add extra CS-register whenever stack alignment >= 8.
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=dc0d9e46a5efbcd496fe60070f569185a9418397;p=platform%2Fupstream%2Fllvm.git

ARM: try to add extra CS-register whenever stack alignment >= 8.

We currently try to push an even number of registers to preserve 8-byte
alignment during a function's prologue, but only when the stack alignment is
prcisely 8. Many of the reasons for doing it apply also when that alignment > 8
(the extra store is often free, and can save another stack adjustment, though
less frequently for 16-byte stack alignment).

llvm-svn: 221321
---

diff --git a/llvm/lib/Target/ARM/ARMFrameLowering.cpp b/llvm/lib/Target/ARM/ARMFrameLowering.cpp
index 4589799..6ebc3aa 100644
--- a/llvm/lib/Target/ARM/ARMFrameLowering.cpp
+++ b/llvm/lib/Target/ARM/ARMFrameLowering.cpp
@@ -1575,7 +1575,7 @@ ARMFrameLowering::processFunctionBeforeCalleeSavedScan(MachineFunction &MF,
     // of GPRs, spill one extra callee save GPR so we won't have to pad between
     // the integer and double callee save areas.
     unsigned TargetAlign = getStackAlignment();
-    if (TargetAlign == 8 && (NumGPRSpills & 1)) {
+    if (TargetAlign >= 8 && (NumGPRSpills & 1)) {
       if (CS1Spilled && !UnspilledCS1GPRs.empty()) {
         for (unsigned i = 0, e = UnspilledCS1GPRs.size(); i != e; ++i) {
           unsigned Reg = UnspilledCS1GPRs[i];
diff --git a/llvm/test/CodeGen/ARM/varargs-spill-stack-align-nacl.ll b/llvm/test/CodeGen/ARM/varargs-spill-stack-align-nacl.ll
index 19d6cbe..148a79d 100644
--- a/llvm/test/CodeGen/ARM/varargs-spill-stack-align-nacl.ll
+++ b/llvm/test/CodeGen/ARM/varargs-spill-stack-align-nacl.ll
@@ -22,9 +22,9 @@ define void @varargs_func(i32 %arg1, ...) {
 ; Reserve space for the varargs save area.  This currently reserves
 ; more than enough (16 bytes rather than the 12 bytes needed).
 ; CHECK: sub sp, sp, #16
-; CHECK: push {lr}
+; CHECK: push {r11, lr}
 ; Align the stack pointer to a multiple of 16.
-; CHECK: sub sp, sp, #12
+; CHECK: sub sp, sp, #8
 ; Calculate the address of the varargs save area and save varargs
 ; arguments into it.
 ; CHECK-NEXT: add r0, sp, #20