Automate architecture synchronization with latest LLVM

.clang-format

@@ -1,6 +1 @@
-BraceWrapping:
-  AfterClass:      true
-  AfterControlStatement: false
-  AfterFunction:   true
-BreakBeforeBraces: Custom
-UseTab: Always
+BasedOnStyle: LLVM

.gitignore

@@ -129,3 +129,6 @@ cstool/cstool
 
 # android
 android-ndk-*
+
+# cmake
+cmake-build-debug

CMakeLists.txt

@@ -29,7 +29,7 @@ option(CAPSTONE_BUILD_CSTOOL "Build cstool" ON)
 option(CAPSTONE_USE_DEFAULT_ALLOC "Use default memory allocation functions" ON)
 option(CAPSTONE_ARCHITECTURE_DEFAULT "Whether architectures are enabled by default" ON)
 option(CAPSTONE_DEBUG "Whether to enable extra debug assertions" OFF)
-option(CAPSTONE_INSTALL "Generate install target" OFF)
+option(CAPSTONE_INSTALL "Generate install target" ON)
 
 set(SUPPORTED_ARCHITECTURES ARM ARM64 M68K MIPS PPC SPARC SYSZ XCORE X86 TMS320C64X M680X EVM MOS65XX WASM BPF RISCV)
 set(SUPPORTED_ARCHITECTURE_LABELS ARM ARM64 M68K MIPS PowerPC Sparc SystemZ XCore x86 TMS320C64x M680x EVM MOS65XX WASM BPF RISCV)
@@ -89,6 +89,7 @@ set(SOURCES_ENGINE
     MCInst.c
     MCInstrDesc.c
     MCRegisterInfo.c
+    MCInstPrinter.c
     SStream.c
     utils.c
 )
@@ -101,6 +102,7 @@ set(HEADERS_ENGINE
     MCInst.h
     MCInstrDesc.h
     MCRegisterInfo.h
+    MCInstPrinter.h
     SStream.h
     utils.h
     )
@@ -201,21 +203,14 @@ if (CAPSTONE_MIPS_SUPPORT)
         arch/Mips/MipsDisassembler.h
         arch/Mips/MipsGenAsmWriter.inc
         arch/Mips/MipsGenDisassemblerTables.inc
-        arch/Mips/MipsGenInstrInfo.inc
-        arch/Mips/MipsGenRegisterInfo.inc
-        arch/Mips/MipsGenSubtargetInfo.inc
-        arch/Mips/MipsInstPrinter.h
+            arch/Mips/MipsInstPrinter.h
         arch/Mips/MipsMapping.h
-        arch/Mips/MipsMappingInsn.inc
-        )
+            )
     set(HEADERS_MIPS
         arch/Mips/MipsDisassembler.h
         arch/Mips/MipsGenAsmWriter.inc
         arch/Mips/MipsGenDisassemblerTables.inc
-        arch/Mips/MipsGenInstrInfo.inc
-        arch/Mips/MipsGenRegisterInfo.inc
-        arch/Mips/MipsGenSubtargetInfo.inc
-        arch/Mips/MipsInstPrinter.h
+            arch/Mips/MipsInstPrinter.h
         arch/Mips/MipsMapping.h
         )
     set(TEST_SOURCES ${TEST_SOURCES} test_mips.c)
@@ -606,7 +601,7 @@ if (CAPSTONE_BUILD_TESTS)
     if (CAPSTONE_ARM_SUPPORT)
         set(ARM_REGRESS_TEST test_arm_regression.c)
         STRING(REGEX REPLACE ".c$" "" ARM_REGRESS_BIN ${ARM_REGRESS_TEST})
-        add_executable(${ARM_REGRESS_BIN} "suite/arm/${ARM_REGRESS_TEST}")
+        add_executable(${ARM_REGRESS_BIN} "suite/arm/${ARM_REGRESS_TEST}" sync/logger.h)
         target_link_libraries(${ARM_REGRESS_BIN} ${default-target})
         add_test(NAME "capstone_${ARM_REGRESS_BIN}" COMMAND ${ARM_REGRESS_BIN})
     endif()

MCInstPrinter.c

@@ -0,0 +1,115 @@
+//
+// Created by Phosphorus15 on 2021/5/14.
+//
+#include "MCInstPrinter.h"
+#include "MCInst.h"
+#include "sync/logger.h"
+
+static bool MCInstPrinter_matchAliasCondition(
+    const MCInst *MI, unsigned *OpIdx, const PatternsForOpcode *OpToPatterns,
+    const AliasPattern *Patterns, const AliasPatternCond *Conds,
+    const AliasPatternCond *Cond, bool *OrPredicateResult) {
+  // FIXME so here's on problem we ought to detect feature bits here
+  if (Cond->Kind == AliasPatternCond_K_Feature ||
+      Cond->Kind == AliasPatternCond_K_NegFeature)
+    return true; // STI->getFeatureBits().test(C.Value);
+  // For feature tests where just one feature is required in a list, set the
+  // predicate result bit to whether the expression will return true, and only
+  // return the real result at the end of list marker.
+  if (Cond->Kind == AliasPatternCond_K_OrFeature) {
+    //        *OrPredicateResult |= STI->getFeatureBits().test(C.Value);
+    return true;
+  }
+  if (Cond->Kind == AliasPatternCond_K_OrNegFeature) {
+    //        *OrPredicateResult |= !(STI->getFeatureBits().test(C.Value));
+    return true;
+  }
+  if (Cond->Kind == AliasPatternCond_K_EndOrFeatures) {
+    bool Res = *OrPredicateResult;
+    *OrPredicateResult = false;
+    return Res;
+  }
+
+  const MCOperand *Opnd = MCInst_getOperand(MI, *OpIdx);
+  *OpIdx = *OpIdx + 1;
+  switch (Cond->Kind) {
+  case AliasPatternCond_K_Imm:
+    // Operand must be a specific immediate.
+    return MCOperand_isImm(Opnd) && MCOperand_getImm(Opnd) == Cond->Value;
+  case AliasPatternCond_K_Reg:
+    // Operand must be a specific register.
+    return MCOperand_isReg(Opnd) && MCOperand_getReg(Opnd) == Cond->Value;
+  case AliasPatternCond_K_TiedReg:
+    // Operand must match the register of another operand.
+    return MCOperand_isReg(Opnd) &&
+           MCOperand_getReg(Opnd) ==
+               MCOperand_getReg(MCInst_getOperand(MI, Cond->Value));
+  case AliasPatternCond_K_RegClass:
+    // Operand must be a register in this class. Value is a register class id.
+    return MCOperand_isReg(Opnd) &&
+           MCRegisterClass_contains(
+               MCRegisterInfo_getRegClass(MRI, Cond->Value),
+               MCOperand_getReg(Opnd));
+  case AliasPatternCond_K_Custom:
+    // Operand must match some custom criteria.
+    // TODO might affect something            return M.ValidateMCOperand(Opnd,
+    // *STI, C.Value);
+    return false;
+  case AliasPatternCond_K_Ignore:
+    // Operand can be anything.
+    return true;
+  case AliasPatternCond_K_Feature:
+  case AliasPatternCond_K_NegFeature:
+  case AliasPatternCond_K_OrFeature:
+  case AliasPatternCond_K_OrNegFeature:
+  case AliasPatternCond_K_EndOrFeatures:
+    0x0;
+    // llvm_unreachable("handled earlier");
+  }
+}
+
+const char *MCInstPrinter_matchAliasPatterns(
+    const MCInst *MI, const PatternsForOpcode *OpToPatterns,
+    const AliasPattern *Patterns, const AliasPatternCond *Conds,
+    const char *AsmStrings[], unsigned len) {
+  // Binary search by opcode. Return false if there are no aliases for this
+  // opcode.
+  PatternsForOpcode *It =
+      Binary_Search(OpToPatterns, MCInst_getOpcode(MI), len);
+  debugln("binary search result %p, with opcode %d", It, MCInst_getOpcode(MI));
+  if (It == NULL || It->Opcode != MCInst_getOpcode(MI))
+    return NULL;
+
+  // Try all patterns for this opcode.
+  uint32_t AsmStrOffset = ~0U;
+
+  for (unsigned i = It->PatternStart; i < It->PatternStart + It->NumPatterns;
+       i++) {
+    const AliasPattern Pattern = Patterns[i];
+    if (MCInst_getNumOperands(MI) != Pattern.NumOperands)
+      return NULL;
+    unsigned OpIdx = 0;
+    bool OrPredicateResult = false;
+    bool fallThrough = true;
+    for (unsigned j = Pattern.AliasCondStart;
+         j < Pattern.AliasCondStart + Pattern.NumConds; j++) {
+      fallThrough &= MCInstPrinter_matchAliasCondition(
+          MI, &OpIdx, OpToPatterns, Patterns, Conds, &Conds[j],
+          &OrPredicateResult);
+      if (!fallThrough)
+        break;
+    }
+    if (fallThrough) {
+      AsmStrOffset = Pattern.AsmStrOffset;
+      break;
+    }
+  }
+
+  debugln("end matching with offset %d", AsmStrOffset);
+  // If no alias matched, don't print an alias.
+  if (AsmStrOffset == ~0U)
+    return NULL;
+  debugln("string with final offset %s",
+          (const char *)((*AsmStrings) + AsmStrOffset));
+  return (const char *)((*AsmStrings) + AsmStrOffset);
+}

MCInstPrinter.h

@@ -0,0 +1,73 @@
+//
+// Created by Phosphorus15 on 2021/5/14.
+//
+
+#ifndef CAPSTONE_MCINSTPRINTER_H
+#define CAPSTONE_MCINSTPRINTER_H
+
+#include "MCInst.h"
+#include "capstone/platform.h"
+#include "stdlib.h"
+
+// TODO we'll need this later
+const MCRegisterInfo *MRI;
+
+typedef enum CondKind {
+  AliasPatternCond_K_Feature,       // Match only if a feature is enabled.
+  AliasPatternCond_K_NegFeature,    // Match only if a feature is disabled.
+  AliasPatternCond_K_OrFeature,     // Match only if one of a set of features is
+                                    // enabled.
+  AliasPatternCond_K_OrNegFeature,  // Match only if one of a set of features is
+                                    // disabled.
+  AliasPatternCond_K_EndOrFeatures, // Note end of list of K_Or(Neg)?Features.
+  AliasPatternCond_K_Ignore,        // Match any operand.
+  AliasPatternCond_K_Reg,           // Match a specific register.
+  AliasPatternCond_K_TiedReg,       // Match another already matched register.
+  AliasPatternCond_K_Imm,           // Match a specific immediate.
+  AliasPatternCond_K_RegClass,      // Match registers in a class.
+  AliasPatternCond_K_Custom,        // Call custom matcher by index.
+} CondKind;
+
+typedef struct PatternsForOpcode {
+  uint32_t Opcode;
+  uint16_t PatternStart;
+  uint16_t NumPatterns;
+} PatternsForOpcode;
+
+typedef struct AliasPattern {
+  uint32_t AsmStrOffset;
+  uint32_t AliasCondStart;
+  uint8_t NumOperands;
+  uint8_t NumConds;
+} AliasPattern;
+
+typedef struct AliasPatternCond {
+  CondKind Kind;
+  uint32_t Value;
+} AliasPatternCond;
+
+static int cmp_less(const void *l, const void *r) {
+  return ((signed)((const PatternsForOpcode *)l)->Opcode) -
+         ((signed)((const PatternsForOpcode *)r)->Opcode);
+}
+
+// Binary Search Implementation - let's use bsearch for now
+static PatternsForOpcode *Binary_Search(const PatternsForOpcode *OpToPatterns,
+                                        const unsigned opcode, unsigned len) {
+  return bsearch((void *)&opcode, (void *)OpToPatterns, len,
+                 sizeof(PatternsForOpcode), cmp_less);
+}
+
+// TODO I'm not sure if this is complete, refer to lib/MC/MCInstPrinter.cpp in
+// llvm-project
+static bool MCInstPrinter_matchAliasCondition(
+    const MCInst *MI, unsigned *OpIdx, const PatternsForOpcode *OpToPatterns,
+    const AliasPattern *Patterns, const AliasPatternCond *Conds,
+    const AliasPatternCond *Cond, bool *OrPredicateResult);
+
+const char *MCInstPrinter_matchAliasPatterns(
+    const MCInst *MI, const PatternsForOpcode *OpToPatterns,
+    const AliasPattern *Patterns, const AliasPatternCond *Conds,
+    const char *AsmStrings[], unsigned len);
+
+#endif // CAPSTONE_MCINSTPRINTER_H

Makefile

@@ -305,7 +305,7 @@ endif
 
 
 LIBOBJ =
-LIBOBJ += $(OBJDIR)/cs.o $(OBJDIR)/utils.o $(OBJDIR)/SStream.o $(OBJDIR)/MCInstrDesc.o $(OBJDIR)/MCRegisterInfo.o
+LIBOBJ += $(OBJDIR)/cs.o $(OBJDIR)/utils.o $(OBJDIR)/SStream.o $(OBJDIR)/MCInstrDesc.o $(OBJDIR)/MCRegisterInfo.o $(OBJDIR)/MCInstPrinter.o
 LIBOBJ += $(LIBOBJ_ARM) $(LIBOBJ_ARM64) $(LIBOBJ_M68K) $(LIBOBJ_MIPS) $(LIBOBJ_PPC) $(LIBOBJ_RISCV) $(LIBOBJ_SPARC) $(LIBOBJ_SYSZ)
 LIBOBJ += $(LIBOBJ_X86) $(LIBOBJ_XCORE) $(LIBOBJ_TMS320C64X) $(LIBOBJ_M680X) $(LIBOBJ_EVM) $(LIBOBJ_MOS65XX) $(LIBOBJ_WASM) $(LIBOBJ_BPF)
 LIBOBJ += $(OBJDIR)/MCInst.o