Dynamic stack frame size support (rust-lang#26)

[SOL] Introduce dynamic stack frames and the SBFv2 flag

Introduce dynamic stack frames, which are currently opt-in and enabled setting cpu=sbfv2.
When sbfv2 is used, ELF files are flagged with e_flags=EF_SBF_V2 so the runtime can detect it
and react accordingly. 

Co-authored-by: Dmitri Makarov <[email protected]>

Author: alessandrod

llvm/include/llvm/BinaryFormat/ELF.h

@@ -807,6 +807,12 @@ enum {
 #include "ELFRelocs/AMDGPU.def"
 };
 
+// SBF specific e_flags
+
+enum : unsigned {
+  EF_SBF_V2 = 0x20,
+};
+
 // ELF Relocation types for BPF
 enum {
 #include "ELFRelocs/BPF.def"

llvm/lib/Target/BPF/BPF.td

@@ -14,15 +14,6 @@ include "BPFInstrInfo.td"
 
 def BPFInstrInfo : InstrInfo;
 
-class Proc<string Name, list<SubtargetFeature> Features>
- : Processor<Name, NoItineraries, Features>;
-
-def : Proc<"generic", []>;
-def : Proc<"v1", []>;
-def : Proc<"v2", []>;
-def : Proc<"v3", []>;
-def : Proc<"probe", []>;
-
 def DummyFeature : SubtargetFeature<"dummy", "isDummyMode",
                                     "true", "unused feature">;
 
@@ -35,6 +26,19 @@ def DwarfRIS: SubtargetFeature<"dwarfris", "UseDwarfRIS", "true",
 def FeatureSolana : SubtargetFeature<"solana", "IsSolana", "true",
                                      "Enable Solana extensions">;
 
+def FeatureDynamicFrames : SubtargetFeature<"dynamic-frames", "HasDynamicFrames", "true",
+                                            "Enable dynamic frames">;
+
+class Proc<string Name, list<SubtargetFeature> Features>
+ : Processor<Name, NoItineraries, Features>;
+
+def : Proc<"generic", []>;
+def : Proc<"v1", []>;
+def : Proc<"v2", []>;
+def : Proc<"v3", []>;
+def : Proc<"probe", []>;
+def : Proc<"sbfv2", [FeatureSolana, FeatureDynamicFrames]>;
+
 def BPFInstPrinter : AsmWriter {
   string AsmWriterClassName  = "InstPrinter";
   bit isMCAsmWriter = 1;

llvm/lib/Target/BPF/BPFFrameLowering.cpp

@@ -20,13 +20,44 @@
 
 using namespace llvm;
 
+namespace {
+
+void adjustStackPointer(MachineFunction &MF, MachineBasicBlock &MBB,
+                        MachineBasicBlock::iterator &MBBI,
+                        unsigned int Opcode) {
+  MachineFrameInfo &MFI = MF.getFrameInfo();
+  int NumBytes = (int)MFI.getStackSize();
+  if (NumBytes) {
+    DebugLoc Dl;
+    const BPFInstrInfo &TII =
+        *static_cast<const BPFInstrInfo *>(MF.getSubtarget().getInstrInfo());
+    BuildMI(MBB, MBBI, Dl, TII.get(Opcode), BPF::R11)
+        .addReg(BPF::R11)
+        .addImm(NumBytes);
+  }
+}
+
+} // namespace
+
 bool BPFFrameLowering::hasFP(const MachineFunction &MF) const { return true; }
 
 void BPFFrameLowering::emitPrologue(MachineFunction &MF,
-                                    MachineBasicBlock &MBB) const {}
+                                    MachineBasicBlock &MBB) const {
+  if (!MF.getSubtarget<BPFSubtarget>().getHasDynamicFrames()) {
+    return;
+  }
+  MachineBasicBlock::iterator MBBI = MBB.begin();
+  adjustStackPointer(MF, MBB, MBBI, BPF::SUB_ri);
+}
 
 void BPFFrameLowering::emitEpilogue(MachineFunction &MF,
-                                    MachineBasicBlock &MBB) const {}
+                                    MachineBasicBlock &MBB) const {
+  if (!MF.getSubtarget<BPFSubtarget>().getHasDynamicFrames()) {
+    return;
+  }
+  MachineBasicBlock::iterator MBBI = MBB.getLastNonDebugInstr();
+  adjustStackPointer(MF, MBB, MBBI, BPF::ADD_ri);
+}
 
 void BPFFrameLowering::determineCalleeSaves(MachineFunction &MF,
                                             BitVector &SavedRegs,

llvm/lib/Target/BPF/BPFRegisterInfo.cpp

@@ -52,22 +52,11 @@ static void WarnSize(int Offset, MachineFunction &MF, DebugLoc& DL)
   OldMF = &(MF.getFunction());
   int MaxOffset = -1 * BPFRegisterInfo::FrameLength;
   if (Offset <= MaxOffset) {
-    if (MF.getSubtarget<BPFSubtarget>().isSolana()) {
-      dbgs() << "Error:";
-      if (DL) {
-        dbgs() << " ";
-        DL.print(dbgs());
-      }
-      dbgs() << " Function " << MF.getFunction().getName() << " Stack offset of " << -Offset
-             << " exceeded max offset of " <<  -MaxOffset << " by "
-             << MaxOffset - Offset << " bytes, please minimize large stack variables\n";
-    } else {
-      DiagnosticInfoUnsupported DiagStackSize(MF.getFunction(),
-          "BPF stack limit of 512 bytes is exceeded. "
-          "Please move large on stack variables into BPF per-cpu array map.\n",
-          DL);
-      MF.getFunction().getContext().diagnose(DiagStackSize);
-    }
+    DiagnosticInfoUnsupported DiagStackSize(MF.getFunction(),
+        "BPF stack limit of 512 bytes is exceeded. "
+        "Please move large on stack variables into BPF per-cpu array map.\n",
+        DL);
+    MF.getFunction().getContext().diagnose(DiagStackSize);
   }
 }
 
@@ -102,7 +91,9 @@ void BPFRegisterInfo::eliminateFrameIndex(MachineBasicBlock::iterator II,
   if (MI.getOpcode() == BPF::MOV_rr) {
     int Offset = MF.getFrameInfo().getObjectOffset(FrameIndex);
 
-    WarnSize(Offset, MF, DL);
+    if (!MF.getSubtarget<BPFSubtarget>().getHasDynamicFrames()) {
+      WarnSize(Offset, MF, DL);
+    }
     MI.getOperand(i).ChangeToRegister(FrameReg, false);
     Register reg = MI.getOperand(i - 1).getReg();
     BuildMI(MBB, ++II, DL, TII.get(BPF::ADD_ri), reg)
@@ -117,7 +108,9 @@ void BPFRegisterInfo::eliminateFrameIndex(MachineBasicBlock::iterator II,
   if (!isInt<32>(Offset))
     llvm_unreachable("bug in frame offset");
 
-  WarnSize(Offset, MF, DL);
+  if (!MF.getSubtarget<BPFSubtarget>().getHasDynamicFrames()) {
+    WarnSize(Offset, MF, DL);
+  }
 
   if (MI.getOpcode() == BPF::FI_ri) {
     // architecture does not really support FI_ri, replace it with

llvm/lib/Target/BPF/BPFSubtarget.cpp

@@ -30,7 +30,6 @@ BPFSubtarget &BPFSubtarget::initializeSubtargetDependencies(const Triple &TT,
                                                             StringRef FS) {
   initializeEnvironment(TT);
   initSubtargetFeatures(CPU, FS);
-  ParseSubtargetFeatures(CPU, /*TuneCPU*/ CPU, FS);
   return *this;
 }
 
@@ -39,23 +38,28 @@ void BPFSubtarget::initializeEnvironment(const Triple &TT) {
   HasJmpExt = false;
   HasJmp32 = false;
   HasAlu32 = false;
+  HasDynamicFrames = false;
   UseDwarfRIS = false;
 }
 
 void BPFSubtarget::initSubtargetFeatures(StringRef CPU, StringRef FS) {
   if (CPU == "probe")
     CPU = sys::detail::getHostCPUNameForBPF();
-  if (CPU == "generic" || CPU == "v1")
-    return;
+
+  ParseSubtargetFeatures(CPU, /*TuneCPU*/ CPU, FS);
+
   if (CPU == "v2") {
     HasJmpExt = true;
-    return;
   }
+
   if (CPU == "v3") {
     HasJmpExt = true;
     HasJmp32 = true;
     HasAlu32 = true;
-    return;
+  }
+
+  if (CPU == "sbfv2" && !HasDynamicFrames) {
+    report_fatal_error("sbfv2 requires dynamic-frames\n", false);
   }
 }
 

llvm/lib/Target/BPF/BPFSubtarget.h

@@ -57,6 +57,9 @@ class BPFSubtarget : public BPFGenSubtargetInfo {
   // whether the cpu supports alu32 instructions.
   bool HasAlu32;
 
+  // whether we should use fixed or dynamic frames
+  bool HasDynamicFrames;
+
   // whether we should enable MCAsmInfo DwarfUsesRelocationsAcrossSections
   bool UseDwarfRIS;
 
@@ -75,6 +78,7 @@ class BPFSubtarget : public BPFGenSubtargetInfo {
   bool getHasJmpExt() const { return HasJmpExt; }
   bool getHasJmp32() const { return HasJmp32; }
   bool getHasAlu32() const { return HasAlu32; }
+  bool getHasDynamicFrames() const { return HasDynamicFrames; }
   bool getUseDwarfRIS() const { return UseDwarfRIS; }
 
   const BPFInstrInfo *getInstrInfo() const override { return &InstrInfo; }

llvm/lib/Target/BPF/MCTargetDesc/BPFMCTargetDesc.cpp

@@ -14,8 +14,14 @@
 #include "MCTargetDesc/BPFInstPrinter.h"
 #include "MCTargetDesc/BPFMCAsmInfo.h"
 #include "TargetInfo/BPFTargetInfo.h"
+#include "llvm/BinaryFormat/ELF.h"
+#include "llvm/MC/MCAsmBackend.h"
+#include "llvm/MC/MCCodeEmitter.h"
+#include "llvm/MC/MCContext.h"
+#include "llvm/MC/MCELFStreamer.h"
 #include "llvm/MC/MCInstrAnalysis.h"
 #include "llvm/MC/MCInstrInfo.h"
+#include "llvm/MC/MCObjectWriter.h"
 #include "llvm/MC/MCRegisterInfo.h"
 #include "llvm/MC/MCSubtargetInfo.h"
 #include "llvm/Support/Host.h"
@@ -54,8 +60,18 @@ static MCStreamer *createBPFMCStreamer(const Triple &T, MCContext &Ctx,
                                        std::unique_ptr<MCObjectWriter> &&OW,
                                        std::unique_ptr<MCCodeEmitter> &&Emitter,
                                        bool RelaxAll) {
-  return createELFStreamer(Ctx, std::move(MAB), std::move(OW), std::move(Emitter),
-                           RelaxAll);
+  MCELFStreamer *S =
+      new MCELFStreamer(Ctx, std::move(MAB), std::move(OW), std::move(Emitter));
+  MCAssembler &A = S->getAssembler();
+  if (RelaxAll)
+    A.setRelaxAll(true);
+
+  const MCSubtargetInfo *STI = Ctx.getSubtargetInfo();
+  if (STI->getCPU() == "sbfv2") {
+    A.setELFHeaderEFlags(llvm::ELF::EF_SBF_V2);
+  }
+
+  return S;
 }
 
 static MCInstPrinter *createBPFMCInstPrinter(const Triple &T,