[ABI] Optimized size of instruction name data (X86, AArch64)

src/asmjit/arm/a64instapi.cpp

@@ -8,7 +8,7 @@
 
 #include "../core/cpuinfo.h"
 #include "../core/misc_p.h"
-#include "../core/support.h"
+#include "../core/support_p.h"
 #include "../arm/a64instapi_p.h"
 #include "../arm/a64instdb_p.h"
 #include "../arm/a64operand.h"
@@ -26,8 +26,11 @@ Error InstInternal::instIdToString(Arch arch, InstId instId, String& output) noe
   if (ASMJIT_UNLIKELY(!Inst::isDefinedId(realId)))
     return DebugUtils::errored(kErrorInvalidInstruction);
 
-  const InstDB::InstInfo& info = InstDB::infoById(realId);
-  return output.append(InstDB::_nameData + info._nameDataIndex);
+
+  char nameData[32];
+  size_t nameSize = Support::decodeInstName(nameData, InstDB::_instNameIndexTable[realId], InstDB::_instNameStringTable);
+
+  return output.append(nameData, nameSize);
 }
 
 InstId InstInternal::stringToInstId(Arch arch, const char* s, size_t len) noexcept {
@@ -46,30 +49,28 @@ InstId InstInternal::stringToInstId(Arch arch, const char* s, size_t len) noexce
   if (ASMJIT_UNLIKELY(prefix > 'z' - 'a'))
     return Inst::kIdNone;
 
-  uint32_t index = InstDB::instNameIndex[prefix].start;
-  if (ASMJIT_UNLIKELY(!index))
+  size_t base = InstDB::instNameIndex[prefix].start;
+  size_t end = InstDB::instNameIndex[prefix].end;
+
+  if (ASMJIT_UNLIKELY(!base))
     return Inst::kIdNone;
 
-  const char* nameData = InstDB::_nameData;
-  const InstDB::InstInfo* table = InstDB::_instInfoTable;
+  char nameData[32];
+  for (size_t lim = end - base; lim != 0; lim >>= 1) {
+    size_t instId = base + (lim >> 1);
+    size_t nameSize = Support::decodeInstName(nameData, InstDB::_instNameIndexTable[instId], InstDB::_instNameStringTable);
 
-  const InstDB::InstInfo* base = table + index;
-  const InstDB::InstInfo* end  = table + InstDB::instNameIndex[prefix].end;
+    int result = Support::compareStringViews(s, len, nameData, nameSize);
+    if (result < 0)
+      continue;
 
-  for (size_t lim = (size_t)(end - base); lim != 0; lim >>= 1) {
-    const InstDB::InstInfo* cur = base + (lim >> 1);
-    int result = Support::cmpInstName(nameData + cur[0]._nameDataIndex, s, len);
-
-    if (result < 0) {
-      base = cur + 1;
+    if (result > 0) {
+      base = instId + 1;
       lim--;
       continue;
     }
 
-    if (result > 0)
-      continue;
-
-    return uint32_t((size_t)(cur - table));
+    return InstId(instId);
   }
 
   return Inst::kIdNone;

src/asmjit/arm/a64instdb.h

@@ -39,9 +39,7 @@ struct InstInfo {
   uint32_t _encoding : 8;
   //! Index to data specific to each encoding type.
   uint32_t _encodingDataIndex : 8;
-  uint32_t _reserved : 2;
-  //! Index to \ref _nameData.
-  uint32_t _nameDataIndex : 14;
+  uint32_t _reserved : 16;
 
   uint16_t _rwInfoIndex;
   uint16_t _flags;

src/asmjit/arm/a64instdb_p.h

@@ -861,7 +861,8 @@ struct InstNameIndex {
 // ====================
 
 #ifndef ASMJIT_NO_TEXT
-extern const char _nameData[];
+extern const uint32_t _instNameIndexTable[];
+extern const char _instNameStringTable[];
 extern const InstNameIndex instNameIndex[26];
 #endif // !ASMJIT_NO_TEXT
 

src/asmjit/core/support.h

@@ -939,6 +939,18 @@ static ASMJIT_FORCE_INLINE int cmpInstName(const char* a, const char* b, size_t
   return int(uint8_t(a[size]));
 }
 
+//! Compares two string views.
+static ASMJIT_FORCE_INLINE int compareStringViews(const char* aData, size_t aSize, const char* bData, size_t bSize) noexcept {
+  size_t size = Support::min(aSize, bSize);
+
+  for (size_t i = 0; i < size; i++) {
+    int c = int(uint8_t(aData[i])) - int(uint8_t(bData[i]));
+    if (c != 0)
+      return c;
+  }
+
+  return int(aSize) - int(bSize);
+}
 // Support - Memory Read Access - 8 Bits
 // =====================================
 

src/asmjit/core/support_p.h

@@ -0,0 +1,64 @@
+// This file is part of AsmJit project <https://asmjit.com>
+//
+// See asmjit.h or LICENSE.md for license and copyright information
+// SPDX-License-Identifier: Zlib
+
+#ifndef ASMJIT_CORE_SUPPORT_P_H_INCLUDED
+#define ASMJIT_CORE_SUPPORT_P_H_INCLUDED
+
+#include "../core/support.h"
+
+ASMJIT_BEGIN_NAMESPACE
+
+//! \addtogroup asmjit_utilities
+//! \{
+
+namespace Support {
+
+//! \cond INTERNAL
+
+static ASMJIT_FORCE_INLINE char decode5BitChar(uint32_t c) noexcept {
+  uint32_t base = c <= 26 ? uint32_t('a') - 1u : uint32_t('0') - 27u;
+  return char(base + c);
+}
+
+static ASMJIT_FORCE_INLINE size_t decodeInstName(char nameOut[32], uint32_t index, const char* stringTable) noexcept {
+  size_t i;
+
+  if (index & 0x80000000u) {
+    // Small string of 5-bit characters.
+    for (i = 0; i < 6; i++, index >>= 5) {
+      uint32_t c = index & 0x1F;
+      if (c == 0)
+        break;
+      nameOut[i] = decode5BitChar(c);
+    }
+    return i;
+  }
+  else {
+    size_t prefixBase = index & 0xFFFu;
+    size_t prefixSize = (index >> 12) & 0xFu;
+
+    size_t suffixBase = (index >> 16) & 0xFFFu;
+    size_t suffixSize = (index >> 28) & 0x7u;
+
+    for (i = 0; i < prefixSize; i++)
+      nameOut[i] = stringTable[prefixBase + i];
+
+    char* suffixOut = nameOut + prefixSize;
+    for (i = 0; i < suffixSize; i++)
+      suffixOut[i] = stringTable[suffixBase + i];
+
+    return prefixSize + suffixSize;
+  }
+}
+
+//! \endcond
+
+} // {Support}
+
+//! \}
+
+ASMJIT_END_NAMESPACE
+
+#endif // ASMJIT_CORE_SUPPORT_P_H_INCLUDED

src/asmjit/x86/x86instapi.cpp

@@ -26,7 +26,7 @@
 
 #include "../core/cpuinfo.h"
 #include "../core/misc_p.h"
-#include "../core/support.h"
+#include "../core/support_p.h"
 #include "../x86/x86instapi_p.h"
 #include "../x86/x86instdb_p.h"
 #include "../x86/x86opcode_p.h"
@@ -44,8 +44,10 @@ Error InstInternal::instIdToString(Arch arch, InstId instId, String& output) noe
   if (ASMJIT_UNLIKELY(!Inst::isDefinedId(instId)))
     return DebugUtils::errored(kErrorInvalidInstruction);
 
-  const InstDB::InstInfo& info = InstDB::infoById(instId);
-  return output.append(InstDB::_nameData + info._nameDataIndex);
+  char nameData[32];
+  size_t nameSize = Support::decodeInstName(nameData, InstDB::_instNameIndexTable[instId], InstDB::_instNameStringTable);
+
+  return output.append(nameData, nameSize);
 }
 
 InstId InstInternal::stringToInstId(Arch arch, const char* s, size_t len) noexcept {
@@ -64,30 +66,28 @@ InstId InstInternal::stringToInstId(Arch arch, const char* s, size_t len) noexce
   if (ASMJIT_UNLIKELY(prefix > 'z' - 'a'))
     return Inst::kIdNone;
 
-  uint32_t index = InstDB::instNameIndex[prefix].start;
-  if (ASMJIT_UNLIKELY(!index))
+  size_t base = InstDB::instNameIndex[prefix].start;
+  size_t end = InstDB::instNameIndex[prefix].end;
+
+  if (ASMJIT_UNLIKELY(!base))
     return Inst::kIdNone;
 
-  const char* nameData = InstDB::_nameData;
-  const InstDB::InstInfo* table = InstDB::_instInfoTable;
+  char nameData[32];
+  for (size_t lim = end - base; lim != 0; lim >>= 1) {
+    size_t instId = base + (lim >> 1);
+    size_t nameSize = Support::decodeInstName(nameData, InstDB::_instNameIndexTable[instId], InstDB::_instNameStringTable);
 
-  const InstDB::InstInfo* base = table + index;
-  const InstDB::InstInfo* end  = table + InstDB::instNameIndex[prefix].end;
+    int result = Support::compareStringViews(s, len, nameData, nameSize);
+    if (result < 0)
+      continue;
 
-  for (size_t lim = (size_t)(end - base); lim != 0; lim >>= 1) {
-    const InstDB::InstInfo* cur = base + (lim >> 1);
-    int result = Support::cmpInstName(nameData + cur[0]._nameDataIndex, s, len);
-
-    if (result < 0) {
-      base = cur + 1;
+    if (result > 0) {
+      base = instId + 1;
       lim--;
       continue;
     }
 
-    if (result > 0)
-      continue;
-
-    return InstId((size_t)(cur - table));
+    return InstId(instId);
   }
 
   return Inst::kIdNone;

src/asmjit/x86/x86instdb.h

@@ -442,8 +442,8 @@ ASMJIT_VARAPI const CommonInfo _commonInfoTable[];
 
 //! Instruction information.
 struct InstInfo {
-  //! Index to \ref _nameData.
-  uint32_t _nameDataIndex : 14;
+  //! Reserved for future use.
+  uint32_t _reserved : 14;
   //! Index to \ref _commonInfoTable.
   uint32_t _commonInfoIndex : 10;
   //! Index to \ref _additionalInfoTable.

src/asmjit/x86/x86instdb_p.h

@@ -295,7 +295,8 @@ extern const uint32_t _mainOpcodeTable[];
 extern const uint32_t _altOpcodeTable[];
 
 #ifndef ASMJIT_NO_TEXT
-extern const char _nameData[];
+extern const uint32_t _instNameIndexTable[];
+extern const char _instNameStringTable[];
 extern const InstNameIndex instNameIndex[26];
 #endif // !ASMJIT_NO_TEXT
 

tools/tablegen-arm.js

@@ -133,11 +133,8 @@ class ArmTableGen extends core.TableGen {
 
         // [06] OpcodeDataIndex.
         "([^\\)]+)" +
-        "\\s*,\\s*" +
-
-        // [07] NameDataIndex.
-        "([^\\)]+)" +
         "\\s*\\)"
+
       , "g");
 
     var m;
@@ -168,9 +165,7 @@ class ArmTableGen extends core.TableGen {
         opcodeData        : opcodeData,      // Opcode data.
         opcodeDataIndex   : -1,              // Opcode data index.
         rwInfo            : rwInfo,          // RW info.
-        flags             : instFlags,       // Instruction flags.
-
-        nameIndex         : -1               // Index to InstDB::_nameData.
+        flags             : instFlags        // Instruction flags.
       });
     }
 
@@ -188,8 +183,7 @@ class ArmTableGen extends core.TableGen {
         String(inst.opcodeData      ).padEnd(86) + ", " +
         String(inst.rwInfo          ).padEnd(10) + ", " +
         String(inst.flags           ).padEnd(26) + ", " +
-        String(inst.opcodeDataIndex ).padEnd( 3) + ", " +
-        String(inst.nameIndex       ).padEnd( 4) + ")";
+        String(inst.opcodeDataIndex ).padEnd( 3) + ")" ;
     }) + "\n";
     return this.inject("InstInfo", s, this.insts.length * 4);
   }

tools/tablegen-x86.js

@@ -500,9 +500,8 @@ class X86TableGen extends core.TableGen {
         // --- autogenerated fields ---
         "([^\\)]+)"                        + "," +  // [05] MainOpcodeIndex.
         "([^\\)]+)"                        + "," +  // [06] AltOpcodeIndex.
-        "([^\\)]+)"                        + "," +  // [07] NameIndex.
-        "([^\\)]+)"                        + "," +  // [08] CommonDataIndex.
-        "([^\\)]+)"                        + "\\)", // [09] OperationDataIndex.
+        "([^\\)]+)"                        + "," +  // [07] CommonDataIndex.
+        "([^\\)]+)"                        + "\\)", // [08] OperationDataIndex.
       "g");
 
     var m;
@@ -562,7 +561,6 @@ class X86TableGen extends core.TableGen {
         String(inst.opcode1            ).padEnd(26) + ", " +
         String(inst.mainOpcodeIndex    ).padEnd( 3) + ", " +
         String(inst.altOpcodeIndex     ).padEnd( 3) + ", " +
-        String(inst.nameIndex          ).padEnd( 5) + ", " +
         String(inst.commonInfoIndex    ).padEnd( 3) + ", " +
         String(inst.additionalInfoIndex).padEnd( 3) + ")";
     }) + "\n";

tools/tablegen.js

@@ -580,6 +580,218 @@ class IndexedString {
 }
 exports.IndexedString = IndexedString;
 
+
+// ============================================================================
+// [InstructionNameData]
+// ============================================================================
+
+function decimalToHexString(number, pad) {
+  if (number < 0)
+    number = 0xFFFFFFFF + number + 1;
+
+  let s = number.toString(16).toUpperCase();
+  if (pad)
+    s = s.padStart(pad, "0")
+  return s;
+}
+
+function charTo5Bit(c) {
+  if (c >= 'a' && c <= 'z')
+    return 1 + (c.charCodeAt(0) - 'a'.charCodeAt(0));
+  else if (c >= '0' && c <= '4')
+    return 1 + 26 + (c.charCodeAt(0) - '0'.charCodeAt(0));
+  else
+    FAIL(`Character '${c}' cannot be encoded into a 5-bit string`);
+}
+
+class InstructionNameData {
+  constructor() {
+    this.names = [];
+    this.primaryTable = [];
+    this.stringTable = "";
+    this.size = 0;
+    this.indexComment = [];
+    this.maxNameLength = 0;
+  }
+
+  add(s) {
+    // First try to encode the string with 5-bit characters that fit into a 32-bit int.
+    if (/^[a-z0-4]{0,6}$/.test(s)) {
+      let index = 0;
+      for (let i = 0; i < s.length; i++)
+        index |= charTo5Bit(s[i]) << (i * 5);
+
+      this.names.push(s);
+      this.primaryTable.push(index | (1 << 31));
+      this.indexComment.push(`Small '${s}'.`);
+    }
+    else {
+      // Put the string into a string table.
+      this.names.push(s);
+      this.primaryTable.push(-1);
+      this.indexComment.push(``);
+    }
+
+    if (this.maxNameLength < s.length)
+      this.maxNameLength = s.length;
+  }
+
+  index() {
+    const kMaxPrefixSize = 15;
+    const kMaxSuffixSize = 7;
+    const names = [];
+
+    for (let idx = 0; idx < this.primaryTable.length; idx++) {
+      if (this.primaryTable[idx] === -1) {
+        names.push({ name: this.names[idx], index: idx });
+      }
+    }
+
+    names.sort(function(a, b) {
+      if (a.name.length > b.name.length)
+        return -1;
+      if (a.name.length < b.name.length)
+        return 1;
+      return (a > b) ? 1 : (a < b) ? -1 : 0;
+    });
+
+    for (let z = 0; z < names.length; z++) {
+      const idx = names[z].index;
+      const name = names[z].name;
+
+      let done = false;
+      let longestPrefix = 0;
+      let longestSuffix = 0;
+
+      let prefix = "";
+      let suffix = "";
+
+      for (let i = Math.min(name.length, kMaxPrefixSize); i > 0; i--) {
+        prefix = name.substring(0, i);
+        suffix = name.substring(i);
+
+        const prefixIndex = this.stringTable.indexOf(prefix);
+        const suffixIndex = this.stringTable.indexOf(suffix);
+
+        // Matched both parts?
+        if (prefixIndex !== -1 && suffix === "") {
+          done = true;
+          break;
+        }
+
+        if (prefixIndex !== -1 && suffixIndex !== -1) {
+          done = true;
+          break;
+        }
+
+        if (prefixIndex !== -1 && longestPrefix === 0)
+          longestPrefix = prefix.length;
+
+        if (suffixIndex !== -1 && suffix.length > longestSuffix)
+          longestSuffix = suffix.length;
+
+        if (suffix.length === kMaxSuffixSize)
+          break;
+      }
+
+      if (!done) {
+        let minPrefixSize = name.length >= 8 ? name.length / 2 + 1 : name.length - 2;
+
+        prefix = "";
+        suffix = "";
+
+        if (longestPrefix >= minPrefixSize) {
+          prefix = name.substring(0, longestPrefix);
+          suffix = name.substring(longestPrefix);
+        }
+        else if (longestSuffix) {
+          const splitAt = Math.min(name.length - longestSuffix, kMaxPrefixSize);;
+          prefix = name.substring(0, splitAt);
+          suffix = name.substring(splitAt);
+        }
+        else if (name.length > kMaxPrefixSize) {
+          prefix = name.substring(0, kMaxPrefixSize);
+          suffix = name.substring(kMaxPrefixSize);
+        }
+        else {
+          prefix = name;
+          suffix = "";
+        }
+      }
+
+      if (suffix) {
+        const prefixIndex = this.addOrReferenceString(prefix);
+        const suffixIndex = this.addOrReferenceString(suffix);
+
+        this.primaryTable[idx] = prefixIndex | (prefix.length << 12) | (suffixIndex << 16) | (suffix.length << 28);
+        this.indexComment[idx] = `Large '${prefix}|${suffix}'.`;
+      }
+      else {
+        const prefixIndex = this.addOrReferenceString(prefix);
+
+        this.primaryTable[idx] = prefixIndex | (prefix.length << 12);
+        this.indexComment[idx] = `Large '${prefix}'.`;
+      }
+    }
+  }
+
+  addOrReferenceString(s) {
+    let index = this.stringTable.indexOf(s);
+    if (index === -1) {
+      index = this.stringTable.length;
+      this.stringTable += s;
+    }
+    return index;
+  }
+
+  formatIndexTable(tableName) {
+    if (this.size === -1)
+      FAIL(`IndexedString.formatIndexTable(): Not indexed yet, call index()`);
+
+    let s = "";
+    for (let i = 0; i < this.primaryTable.length; i++) {
+      s += "0x" + decimalToHexString(this.primaryTable[i], 8);
+      s += i !== this.primaryTable.length - 1 ? "," : " ";
+      s += " // " + this.indexComment[i] + "\n";
+    }
+
+    return `const uint32_t ${tableName}[] = {\n${StringUtils.indent(s, "  ")}};\n`;
+  }
+
+  formatStringTable(tableName) {
+    if (this.size === -1)
+      FAIL(`IndexedString.formatStringTable(): Not indexed yet, call index()`);
+
+    let s = "";
+    for (let i = 0; i < this.stringTable.length; i += 80) {
+      if (s)
+        s += "\n"
+      s += '"' + this.stringTable.substring(i, i + 80) + '"';
+    }
+    s += ";\n";
+
+    return `const char ${tableName}[] =\n${StringUtils.indent(s, "  ")}\n`;
+  }
+
+  getSize() {
+    if (this.size === -1)
+      FAIL(`IndexedString.getSize(): Not indexed yet, call index()`);
+
+    return this.primaryTable.length * 4 + this.stringTable.length;
+  }
+
+  getIndex(k) {
+    if (this.size === -1)
+      FAIL(`IndexedString.getIndex(): Not indexed yet, call index()`);
+
+    if (!hasOwn.call(this.map, k))
+      FAIL(`IndexedString.getIndex(): Key '${k}' not found.`);
+
+    return this.map[k];
+  }
+}
+exports.InstructionNameData = InstructionNameData;
+
 // ============================================================================
 // [IndexedArray]
 // ============================================================================
@@ -891,41 +1103,37 @@ class NameTable extends Task {
   }
 
   run() {
-    const arch = this.ctx.arch;
     const none = "Inst::kIdNone";
-
     const insts = this.ctx.insts;
-    const instNames = new IndexedString();
 
     const instFirst = new Array(26);
     const instLast  = new Array(26);
 
-    var maxLength = 0;
-    for (var i = 0; i < insts.length; i++) {
-      const inst = insts[i];
-      instNames.add(inst.displayName);
-      maxLength = Math.max(maxLength, inst.displayName.length);
-    }
-    instNames.index();
+    const instNameData = new InstructionNameData();
 
-    for (var i = 0; i < insts.length; i++) {
+    for (let i = 0; i < insts.length; i++)
+      instNameData.add(insts[i].displayName);
+    instNameData.index();
+
+    for (let i = 0; i < insts.length; i++) {
       const inst = insts[i];
       const name = inst.displayName;
-      const nameIndex = instNames.getIndex(name);
-
       const index = name.charCodeAt(0) - 'a'.charCodeAt(0);
+
       if (index < 0 || index >= 26)
         FAIL(`TableGen.generateNameData(): Invalid lookup character '${name[0]}' of '${name}'`);
 
-      inst.nameIndex = nameIndex;
       if (instFirst[index] === undefined)
         instFirst[index] = `Inst::kId${inst.enum}`;
       instLast[index] = `Inst::kId${inst.enum}`;
     }
 
     var s = "";
-    s += `const char InstDB::_nameData[] =\n${instNames.format(kIndent, kJustify)}\n`;
+    s += instNameData.formatIndexTable("InstDB::_instNameIndexTable");
     s += `\n`;
+    s += instNameData.formatStringTable("InstDB::_instNameStringTable");
+    s += `\n`;
+
     s += `const InstDB::InstNameIndex InstDB::instNameIndex[26] = {\n`;
     for (var i = 0; i < instFirst.length; i++) {
       const firstId = instFirst[i] || none;
@@ -939,9 +1147,9 @@ class NameTable extends Task {
     s += `};\n`;
 
     this.ctx.inject("NameLimits",
-      StringUtils.disclaimer(`enum : uint32_t { kMaxNameSize = ${maxLength} };\n`));
+      StringUtils.disclaimer(`enum : uint32_t { kMaxNameSize = ${instNameData.maxNameLength} };\n`));
 
-    return this.ctx.inject("NameData", StringUtils.disclaimer(s), instNames.getSize() + 26 * 4);
+    return this.ctx.inject("NameData", StringUtils.disclaimer(s), instNameData.getSize() + 26 * 4);
   }
 }
 exports.NameTable = NameTable;