[PowerPC] Allow stack frames without parameter save area

The PPCFrameLowering::determineFrameLayout routine currently ensures
that every function that allocates a stack frame provides space for the
parameter save area (via PPCFrameLowering::getMinCallFrameSize).

This is actually not necessary.  There may be functions that never call
another routine but still allocate a frame; those do not require the
parameter save area.  In the future, with the ELFv2 ABI, even some
routines that do call other functions do not need to allocate the
parameter save area.

While it is not a bug to allocate the parameter area when it is not
needed, it is better to avoid it to save stack space.

Note that when any particular function call requires the parameter save
area, this space will already have been included by ABI code in the size
the CALLSEQ_START insn is annotated with, and therefore included in the
size returned by MFI->getMaxCallFrameSize().

This means that determineFrameLayout simply does not need to care about
the parameter save area.  (It still needs to ensure that every frame
provides the linkage area.)  This is implemented by this patch.

Note that this exposed a bug in the new fast-isel code where the parameter
area was *not* included in the CALLSEQ_START size; this is also fixed.

A couple of test cases needed to be adapted for the new (smaller) stack
frame size those tests now see.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211495 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/PowerPC/PPCFastISel.cpp b/lib/Target/PowerPC/PPCFastISel.cpp
index 924a07c6cff388d27d0921d4a28ce92b7572e64b..1b4f12e2be9906b274bca5048c1a8464d8d27d62 100644 (file)
--- a/lib/Target/PowerPC/PPCFastISel.cpp
+++ b/lib/Target/PowerPC/PPCFastISel.cpp
@@ -1197,6 +1197,10 @@ bool PPCFastISel::processCallArgs(SmallVectorImpl<Value*> &Args,
                                   bool IsVarArg) {
   SmallVector<CCValAssign, 16> ArgLocs;
   CCState CCInfo(CC, IsVarArg, *FuncInfo.MF, TM, ArgLocs, *Context);
+
+  // Reserve space for the linkage area on the stack.
+  CCInfo.AllocateStack(PPCFrameLowering::getLinkageSize(true, false), 8);
+
   CCInfo.AnalyzeCallOperands(ArgVTs, ArgFlags, CC_PPC64_ELF_FIS);
 
   // Bail out if we can't handle any of the arguments.
@@ -1218,6 +1222,14 @@ bool PPCFastISel::processCallArgs(SmallVectorImpl<Value*> &Args,
   // Get a count of how many bytes are to be pushed onto the stack.
   NumBytes = CCInfo.getNextStackOffset();
 
+  // The prolog code of the callee may store up to 8 GPR argument registers to
+  // the stack, allowing va_start to index over them in memory if its varargs.
+  // Because we cannot tell if this is needed on the caller side, we have to
+  // conservatively assume that it is needed.  As such, make sure we have at
+  // least enough stack space for the caller to store the 8 GPRs.
+  NumBytes = std::max(NumBytes,
+                      PPCFrameLowering::getMinCallFrameSize(true, false));
+
   // Issue CALLSEQ_START.
   BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
           TII.get(TII.getCallFrameSetupOpcode()))

diff --git a/lib/Target/PowerPC/PPCFrameLowering.cpp b/lib/Target/PowerPC/PPCFrameLowering.cpp
index 9c5e588f2fc2f7b17baba935a391e221b186ae1f..ef4ea77bfca50b5a974c8de2c04c7cf8d691a557 100644 (file)
--- a/lib/Target/PowerPC/PPCFrameLowering.cpp
+++ b/lib/Target/PowerPC/PPCFrameLowering.cpp
@@ -422,9 +422,9 @@ unsigned PPCFrameLowering::determineFrameLayout(MachineFunction &MF,
   // Get the maximum call frame size of all the calls.
   unsigned maxCallFrameSize = MFI->getMaxCallFrameSize();
 
-  // Maximum call frame needs to be at least big enough for linkage and 8 args.
-  unsigned minCallFrameSize = getMinCallFrameSize(Subtarget.isPPC64(),
-                                                  Subtarget.isDarwinABI());
+  // Maximum call frame needs to be at least big enough for linkage area.
+  unsigned minCallFrameSize = getLinkageSize(Subtarget.isPPC64(),
+                                             Subtarget.isDarwinABI());
   maxCallFrameSize = std::max(maxCallFrameSize, minCallFrameSize);
 
   // If we have dynamic alloca then maxCallFrameSize needs to be aligned so

diff --git a/test/CodeGen/PowerPC/Frames-alloca.ll b/test/CodeGen/PowerPC/Frames-alloca.ll
index 4588bc05352bcb5b503dcb5fb7a0b24b442b9b9a..c701fef8e629be573cb05a369edbd08351c78548 100644 (file)
--- a/test/CodeGen/PowerPC/Frames-alloca.ll
+++ b/test/CodeGen/PowerPC/Frames-alloca.ll
@@ -12,15 +12,15 @@
 ; CHECK-PPC32-NOFP: stw r31, -4(r1)
 ; CHECK-PPC32-NOFP: lwz r1, 0(r1)
 ; CHECK-PPC32-NOFP: lwz r31, -4(r1)
-; CHECK-PPC32-RS: stwu r1, -80(r1)
-; CHECK-PPC32-RS-NOFP: stwu r1, -80(r1)
+; CHECK-PPC32-RS: stwu r1, -48(r1)
+; CHECK-PPC32-RS-NOFP: stwu r1, -48(r1)
 
 ; CHECK-PPC64: std r31, -8(r1)
-; CHECK-PPC64: stdu r1, -128(r1)
+; CHECK-PPC64: stdu r1, -64(r1)
 ; CHECK-PPC64: ld r1, 0(r1)
 ; CHECK-PPC64: ld r31, -8(r1)
 ; CHECK-PPC64-NOFP: std r31, -8(r1)
-; CHECK-PPC64-NOFP: stdu r1, -128(r1)
+; CHECK-PPC64-NOFP: stdu r1, -64(r1)
 ; CHECK-PPC64-NOFP: ld r1, 0(r1)
 ; CHECK-PPC64-NOFP: ld r31, -8(r1)
 

diff --git a/test/CodeGen/PowerPC/Frames-large.ll b/test/CodeGen/PowerPC/Frames-large.ll
index d07fea72677011d4892a0e8629d88a3e4df4b8e3..0ccea42619afeea197bcc89ae15a489b58e5a3dc 100644 (file)
--- a/test/CodeGen/PowerPC/Frames-large.ll
+++ b/test/CodeGen/PowerPC/Frames-large.ll
@@ -15,9 +15,9 @@ define i32* @f1() nounwind {
 
 ; PPC32-NOFP: _f1:
 ; PPC32-NOFP:  lis r0, -1
-; PPC32-NOFP:  ori r0, r0, 32704
+; PPC32-NOFP:  ori r0, r0, 32736
 ; PPC32-NOFP:  stwux r1, r1, r0
-; PPC32-NOFP:  addi r3, r1, 68
+; PPC32-NOFP:  addi r3, r1, 36
 ; PPC32-NOFP:  lwz r1, 0(r1)
 ; PPC32-NOFP:  blr 
 
@@ -25,10 +25,10 @@ define i32* @f1() nounwind {
 ; PPC32-FP: _f1:
 ; PPC32-FP:    lis r0, -1
 ; PPC32-FP:    stw r31, -4(r1)
-; PPC32-FP:    ori r0, r0, 32704
+; PPC32-FP:    ori r0, r0, 32736
 ; PPC32-FP:    stwux r1, r1, r0
 ; PPC32-FP:    mr r31, r1
-; PPC32-FP:    addi r3, r31, 64
+; PPC32-FP:    addi r3, r31, 32
 ; PPC32-FP:    lwz r1, 0(r1)
 ; PPC32-FP:    lwz r31, -4(r1)
 ; PPC32-FP:    blr 
@@ -36,9 +36,9 @@ define i32* @f1() nounwind {
 
 ; PPC64-NOFP: _f1:
 ; PPC64-NOFP:  lis r0, -1
-; PPC64-NOFP:  ori r0, r0, 32656
+; PPC64-NOFP:  ori r0, r0, 32720
 ; PPC64-NOFP:  stdux r1, r1, r0
-; PPC64-NOFP:  addi r3, r1, 116
+; PPC64-NOFP:  addi r3, r1, 52
 ; PPC64-NOFP:  ld r1, 0(r1)
 ; PPC64-NOFP:  blr 
 
@@ -46,10 +46,10 @@ define i32* @f1() nounwind {
 ; PPC64-FP: _f1:
 ; PPC64-FP:    lis r0, -1
 ; PPC64-FP:    std r31, -8(r1)
-; PPC64-FP:    ori r0, r0, 32640
+; PPC64-FP:    ori r0, r0, 32704
 ; PPC64-FP:    stdux r1, r1, r0
 ; PPC64-FP:    mr r31, r1
-; PPC64-FP:    addi r3, r31, 124
+; PPC64-FP:    addi r3, r31, 60
 ; PPC64-FP:    ld r1, 0(r1)
 ; PPC64-FP:    ld r31, -8(r1)
 ; PPC64-FP:    blr 

diff --git a/test/CodeGen/PowerPC/Frames-small.ll b/test/CodeGen/PowerPC/Frames-small.ll
index 0f6bd1021f800758be9b54afb5180d649f686bfa..28c1a5b54dd2d5bee06caab0eee0a960a2ede659 100644 (file)
--- a/test/CodeGen/PowerPC/Frames-small.ll
+++ b/test/CodeGen/PowerPC/Frames-small.ll
@@ -1,25 +1,25 @@
 ; RUN: llc < %s -march=ppc32 -mtriple=powerpc-apple-darwin8 -o %t1
 ; RUN: not grep "stw r31, -4(r1)" %t1
-; RUN: grep "stwu r1, -16448(r1)" %t1
-; RUN: grep "addi r1, r1, 16448" %t1
+; RUN: grep "stwu r1, -16416(r1)" %t1
+; RUN: grep "addi r1, r1, 16416" %t1
 ; RUN: llc < %s -march=ppc32 | \
 ; RUN: not grep "lwz r31, -4(r1)"
 ; RUN: llc < %s -march=ppc32 -mtriple=powerpc-apple-darwin8 -disable-fp-elim \
 ; RUN:   -o %t2
 ; RUN: grep "stw r31, -4(r1)" %t2
-; RUN: grep "stwu r1, -16448(r1)" %t2
-; RUN: grep "addi r1, r1, 16448" %t2
+; RUN: grep "stwu r1, -16416(r1)" %t2
+; RUN: grep "addi r1, r1, 16416" %t2
 ; RUN: grep "lwz r31, -4(r1)" %t2
 ; RUN: llc < %s -march=ppc64 -mtriple=powerpc-apple-darwin8 -o %t3
 ; RUN: not grep "std r31, -8(r1)" %t3
-; RUN: grep "stdu r1, -16496(r1)" %t3
-; RUN: grep "addi r1, r1, 16496" %t3
+; RUN: grep "stdu r1, -16432(r1)" %t3
+; RUN: grep "addi r1, r1, 16432" %t3
 ; RUN: not grep "ld r31, -8(r1)" %t3
 ; RUN: llc < %s -march=ppc64 -mtriple=powerpc-apple-darwin8 -disable-fp-elim \
 ; RUN:   -o %t4
 ; RUN: grep "std r31, -8(r1)" %t4
-; RUN: grep "stdu r1, -16512(r1)" %t4
-; RUN: grep "addi r1, r1, 16512" %t4
+; RUN: grep "stdu r1, -16448(r1)" %t4
+; RUN: grep "addi r1, r1, 16448" %t4
 ; RUN: grep "ld r31, -8(r1)" %t4
 
 define i32* @f1() {

diff --git a/test/CodeGen/PowerPC/svr4-redzone.ll b/test/CodeGen/PowerPC/svr4-redzone.ll
index 7c51b67aeecbed0c0826ab7b9ac2f47677d2ef2f..bee3ac32b6488c15f6efd174802d2376f7ce1bcb 100644 (file)
--- a/test/CodeGen/PowerPC/svr4-redzone.ll
+++ b/test/CodeGen/PowerPC/svr4-redzone.ll
@@ -36,4 +36,4 @@ entry:
 ; PPC32: stwu 1, -240(1)
 
 ; PPC64-LABEL: bigstack:
-; PPC64: stdu 1, -352(1)
+; PPC64: stdu 1, -288(1)