fxos: generate defs from insn table + some renaming

CMakeLists.txt

@@ -28,6 +28,22 @@ if(FXOS_CLEAN_LOGS)
     "${PROJECT_BINARY_DIR}/compiler_filter.sh")
 endif()
 
+#---
+# tooling for generating code at compile-time
+#---
+
+add_executable(mkdefs_insn tools/mkdefs_insn.cc)
+target_link_libraries(mkdefs_insn PRIVATE -lfmt)
+
+add_custom_command(
+  OUTPUT "${CMAKE_CURRENT_BINARY_DIR}/gendefs/insn.h"
+  COMMAND mkdir -p "${CMAKE_CURRENT_BINARY_DIR}/gendefs/"
+  COMMAND "$<TARGET_FILE:mkdefs_insn>"
+          "${CMAKE_CURRENT_SOURCE_DIR}/lib/sh3.def"
+          "${CMAKE_CURRENT_SOURCE_DIR}/lib/sh4.def" >
+          "${CMAKE_CURRENT_BINARY_DIR}/gendefs/insn.h"
+  DEPENDS mkdefs_insn lib/sh3.def lib/sh4.def)
+
 #---
 # fxos core
 #---
@@ -69,10 +85,13 @@ set(fxos_core_SOURCES
   lib/util/bson.cpp
   lib/util/Buffer.cpp
   lib/util/log.cpp
-  lib/util/Timer.cpp)
+  lib/util/Timer.cpp
+
+  "${CMAKE_CURRENT_BINARY_DIR}/gendefs/insn.h")
 
 add_library(fxos-core ${fxos_core_SOURCES} ${FLEX_LoadAsm_OUTPUTS})
-target_include_directories(fxos-core PUBLIC include)
+target_include_directories(fxos-core
+  PUBLIC include "${CMAKE_CURRENT_BINARY_DIR}")
 
 #---
 # fxos shell

include/fxos/lang.h

@@ -168,7 +168,7 @@ struct AsmInstruction
 
     /* Original opcode. Initialized to 0 when unset, which is an invalid
        instruction by design. */
-    u32 opcode;
+    u32 encoding;
     /* Operation size (0, 1, 2 or 4) */
     i8 opsize;
     /* Number of arguments */

lib/analysis.cpp

@@ -109,7 +109,7 @@ static ProgramStateDiff interpretInstruction(
     diff.setUnknown();
 
     // TODO: Do this properly
-    u16 opc = ins.opcode().opcode;
+    u16 opc = ins.opcode().encoding;
     if((opc & 0xf000) == 0xe000) {
         int reg = (opc >> 8) & 0xf;
         int val = (int8_t)opc;

lib/disassembly.cpp

@@ -19,13 +19,13 @@ std::array<std::optional<AsmInstruction>, 65536> insmap;
 
 void register_instruction(AsmInstruction const &ins)
 {
-    uint16_t opcode = ins.opcode;
+    uint16_t encoding = ins.encoding;
 
-    if(insmap[opcode])
-        FxOS_log(ERR, "opcode collision between a %s and a %s at %04x",
-            insmap[opcode]->mnemonic, ins.mnemonic, opcode);
+    if(insmap[encoding])
+        FxOS_log(ERR, "encoding collision between a %s and a %s at %04x",
+            insmap[encoding]->mnemonic, ins.mnemonic, encoding);
     else
-        insmap[opcode] = ins;
+        insmap[encoding] = ins;
 }
 
 //---
@@ -40,7 +40,7 @@ Argument::Argument()
 }
 
 OldInstruction::OldInstruction(AsmInstruction const *inst):
-    inst {inst}, args {}, opcode {inst->opcode}, leader {false},
+    inst {inst}, args {}, opcode {inst->encoding}, leader {false},
     delayslot {false}, terminal {false}, jump {false}, condjump {false},
     jmptarget {0xffffffff}
 {

lib/lang.cpp

@@ -196,7 +196,7 @@ u32 AsmArgument::getPCRelativeTarget(u32 pc, int size) const
 //---
 
 AsmInstruction::AsmInstruction(char const *mn):
-    opcode {0}, opsize {0}, arg_count {0}
+    encoding {0}, opsize {0}, arg_count {0}
 {
     int len = strlen(mn);
     int pos = std::max(0, len - 2);

lib/load-asm.l

@@ -268,7 +268,7 @@ static int instantiate(struct Pattern p, char const *mnemonic, int argtoken1,
         opcode |= (i << p.i_sh);
 
         AsmInstruction ins(mnemonic);
-        ins.opcode = opcode;
+        ins.encoding = opcode;
         ins.tags = insntags;
 
         if(argtoken1) {

lib/view/assembly.cpp

@@ -141,7 +141,7 @@ static void doOldInst(u32 pc, OldInstruction &i,
     }
 
     /* Raw data if instruction cannot be decoded */
-    printf(" %08x:  %04x", pc, (i.inst ? i.inst->opcode : i.opcode));
+    printf(" %08x:  %04x", pc, (i.inst ? i.inst->encoding : i.opcode));
     if(!i.inst) {
         printf("\n");
         m_lastAddress = pc;
@@ -239,7 +239,7 @@ void viewAssemblyInstruction(Instruction const &ins, ViewAssemblyOptions *opts)
     ArgumentOutput argout;
     u32 pc = ins.address();
 
-    printf("    %08x:  %04x", pc, opcode.opcode);
+    printf("    %08x:  %04x", pc, opcode.encoding);
 
     /* Mnemonic */
     static char const *suffixes[5] = {"", ".b", ".w", "", ".l"};

tools/mkdefs_insn.cc

@@ -0,0 +1,67 @@
+#include <fmt/core.h>
+#include <iostream>
+#include <fstream>
+#include <sstream>
+#include <cstring>
+#include <cstdlib>
+#include <cctype>
+#include <cassert>
+#include <map>
+
+static void usage(int rc)
+{
+    fprintf(stderr, "usage: ./mkdefs_insn <insn listing...>\n");
+    exit(rc);
+}
+
+std::string encodeMnemonic(std::string mnemonic)
+{
+    for(size_t i = 0; i < mnemonic.size(); i++) {
+        if(!isalnum(mnemonic[i]))
+            mnemonic[i] = '_';
+    }
+
+    return mnemonic;
+}
+
+static void parseTable(
+    char const *file, std::map<std::string, std::string> &allMnemonics)
+{
+    std::ifstream fin {file};
+    std::string line, pattern, mnemonic;
+
+    while(std::getline(fin, line)) {
+        if(line.size() == 0 || !strchr("01nmdi", line[0]))
+            continue;
+
+        std::istringstream stream {line};
+        stream >> pattern >> mnemonic;
+
+        if(mnemonic.ends_with(".b") || mnemonic.ends_with(".w") ||
+           mnemonic.ends_with(".l"))
+            mnemonic = mnemonic.substr(0, mnemonic.size() - 2);
+
+        std::string enc = encodeMnemonic(mnemonic);
+        assert(!allMnemonics.count(enc) || allMnemonics.at(enc) == mnemonic);
+        allMnemonics.insert({ enc, mnemonic });
+    }
+}
+
+int main(int argc, char **argv)
+{
+    for(int i = 1; i < argc; i++) {
+        if(!strcmp(argv[i], "-h") || !strcmp(argv[i], "--help"))
+            usage(0);
+    }
+    if(argc <= 1)
+        usage(1);
+
+    std::map<std::string, std::string> allMnemonics;
+    for(int i = 1; i < argc; i++)
+        parseTable(argv[i], allMnemonics);
+
+    for(auto const &[e,m]: allMnemonics)
+        fmt::print("GENDEFS_INSN({}, \"{}\")\n", e, m);
+
+    return 0;
+}