[CHERIoT] Make it possible to use captable relocations for heavily used globals, for which it is a net improvement in code size.

Author: resistor

llvm/lib/Target/RISCV/RISCVExpandPseudoInsts.cpp

@@ -707,7 +707,8 @@ bool RISCVExpandPseudo::expandAuipccInstPair(
       MF->getSubtarget<RISCVSubtarget>().isRV32E() && Symbol.isGlobal() &&
       isa<GlobalVariable>(Symbol.getGlobal()) &&
       (cast<GlobalVariable>(Symbol.getGlobal())->getSection() !=
-       ".compartment_imports"))
+       ".compartment_imports") &&
+      FlagsHi != RISCVII::MO_CAPTAB_PCREL_HI)
     BuildMI(NewMBB, DL, TII->get(RISCV::CSetBoundsImm), DestReg)
         .addReg(DestReg)
         .addDisp(Symbol, 0, RISCVII::MO_CHERIOT_COMPARTMENT_SIZE);

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

@@ -94,6 +94,11 @@ static cl::opt<bool>
                                "be combined with a shift"),
                       cl::init(true));
 
+static cl::opt<bool>
+    CheriotUseCapTable("cheriot-use-cap-table", cl::Hidden,
+                      cl::desc("Use cap tables to access globals with many static references (CHERIoT)"),
+                      cl::init(false));
+
 RISCVTargetLowering::RISCVTargetLowering(const TargetMachine &TM,
                                          const RISCVSubtarget &STI)
     : TargetLowering(TM), Subtarget(STI) {
@@ -8977,14 +8982,17 @@ static SDValue getLargeExternalSymbol(ExternalSymbolSDNode *N, const SDLoc &DL,
 template <class NodeTy>
 SDValue RISCVTargetLowering::getAddr(NodeTy *N, EVT Ty, SelectionDAG &DAG,
                                      bool IsLocal, bool CanDeriveFromPcc,
-                                     bool IsExternWeak) const {
+                                     bool IsExternWeak,
+                                     const GlobalValue *GV) const {
   SDLoc DL(N);
 
   if (RISCVABI::isCheriPureCapABI(Subtarget.getTargetABI())) {
     bool IsCheriot = Subtarget.getTargetABI() == RISCVABI::ABI_CHERIOT ||
                      Subtarget.getTargetABI() == RISCVABI::ABI_CHERIOT_BAREMETAL;
     SDValue Addr = getTargetNode(N, DL, Ty, DAG, 0);
-    if ((IsLocal && CanDeriveFromPcc) || IsCheriot) {
+    bool NotHighUseGV = !GV || GV->getNumUses() < 4;
+    if ((IsLocal && CanDeriveFromPcc) ||
+        (IsCheriot && CheriotUseCapTable && NotHighUseGV)) {
       // Use PC-relative addressing to access the symbol. This generates the
       // pattern (PseudoCLLC sym), which expands to
       // (cincoffsetimm (auipcc %pcrel_hi(sym)) %pcrel_lo(auipc)).
@@ -9121,7 +9129,7 @@ SDValue RISCVTargetLowering::lowerGlobalAddress(SDValue Op,
   }
 
   return getAddr(N, Ty, DAG, GV->isDSOLocal(), /*CanDeriveFromPcc=*/false,
-                 GV->hasExternalWeakLinkage());
+                 GV->hasExternalWeakLinkage(), GV);
 }
 
 SDValue RISCVTargetLowering::lowerBlockAddress(SDValue Op,

llvm/lib/Target/RISCV/RISCVISelLowering.h

@@ -500,7 +500,8 @@ class RISCVTargetLowering : public TargetLowering {
 
   template <class NodeTy>
   SDValue getAddr(NodeTy *N, EVT Ty, SelectionDAG &DAG, bool IsLocal,
-                  bool CanDeriveFromPcc, bool IsExternWeak = false) const;
+                  bool CanDeriveFromPcc, bool IsExternWeak = false,
+                  const GlobalValue *GV = nullptr) const;
   SDValue getStaticTLSAddr(GlobalAddressSDNode *N, EVT Ty, SelectionDAG &DAG,
                            bool NotLocal) const;
   SDValue getDynamicTLSAddr(GlobalAddressSDNode *N, EVT Ty,