koffi 0.9.27 → 0.9.28

package/README.md

@@ -265,7 +265,7 @@ Once this is done, you can execute each implementation, e.g. `build/atoi_cc` or
 
 ## atoi results
 
-Here are some results from 2022-04-24 on my Linux machine (AMD® Ryzen™ 7 5800H 16G):
+Here are some results from 2022-04-24 on Linux on my machine (AMD® Ryzen™ 7 5800H 16G):
 
 ```sh
 $ build/atoi_cc
@@ -278,7 +278,7 @@ Time: 1.10s
 
 $ ./atoi_koffi.js
 Iterations: 20000000
-Time: 2.34s
+Time: 1.91s
 
 # Note: the Node-FFI version does a few setTimeout calls to force the GC to run (around 20
 # for the example below), without which Node will consume all memory because the GC never appears
@@ -289,29 +289,29 @@ Iterations: 20000000
 Time: 640.49s
 ```
 
-And on my Windows machine (Intel® Core™ i5-4460 16G):
+And on Windows on the same machine (AMD® Ryzen™ 7 5800H 16G):
 
 ```sh
 $ build\atoi_cc.exe
 Iterations: 20000000
-Time: 0.66s
+Time: 0.25s
 
 $ node atoi_napi.js
 Iterations: 20000000
-Time: 3.23s
+Time: 1.94s
 
 $ node atoi_koffi.js
 Iterations: 20000000
-Time: 4.81s
+Time: 3.15s
 
 $ node atoi_node_ffi.js
 Iterations: 20000000
-Time: 491.99s
+Time: 267.20s
 ```
 
 ## Raylib results
 
-Here are some results from 2022-04-24 on my Linux machine (AMD® Ryzen™ 7 5800H 16G):
+Here are some results from 2022-04-24 on Linux on my machine (AMD® Ryzen™ 7 5800H 16G):
 
 ```sh
 $ build/raylib_cc
@@ -327,18 +327,18 @@ Iterations: 100
 Time: 27.13s
 ```
 
-And on my Windows machine (Intel® Core™ i5-4460 16G):
+And on Windows on the same machine (AMD® Ryzen™ 7 5800H 16G):
 
 ```sh
 $ build\raylib_cc.exe
 Iterations: 100
-Time: 10.53s
+Time: 8.39s
 
 $ node raylib_koffi.js
 Iterations: 100
-Time: 14.60s
+Time: 11.51s
 
 $ node raylib_node_ffi.js
 Iterations: 100
-Time: 44.97s
+Time: 32.47s
 ```

package/benchmark/CMakeLists.txt

@@ -16,7 +16,10 @@ project(koffi C CXX ASM)
 
 find_package(CNoke)
 
-set(CMAKE_CXX_STANDARD 17)
+set(CMAKE_CXX_STANDARD 20)
+
+set(THREADS_PREFER_PTHREAD_FLAG ON)
+find_package(Threads REQUIRED)
 
 if(NOT TARGET koffi)
     add_subdirectory(.. koffi)
@@ -27,6 +30,7 @@ add_subdirectory(../test test)
 
 add_executable(atoi_cc atoi_cc.cc ../vendor/libcc/libcc.cc)
 target_include_directories(atoi_cc PRIVATE ..)
+target_link_libraries(atoi_cc PRIVATE Threads::Threads)
 
 if(WIN32)
     target_compile_definitions(atoi_cc PRIVATE _CRT_SECURE_NO_WARNINGS _CRT_NONSTDC_NO_DEPRECATE)
@@ -35,19 +39,24 @@ endif()
 
 add_node_addon(NAME atoi_napi SOURCES atoi_napi.cc ../vendor/libcc/libcc.cc)
 target_include_directories(atoi_napi PRIVATE .. ../vendor/node-addon-api)
+target_link_libraries(atoi_napi PRIVATE Threads::Threads)
 
 if(WIN32)
     target_compile_definitions(atoi_napi PRIVATE _CRT_SECURE_NO_WARNINGS _CRT_NONSTDC_NO_DEPRECATE)
     target_link_libraries(atoi_napi PRIVATE ws2_32)
+else()
+    target_link_libraries(atoi_napi PRIVATE dl)
 endif()
 
 # ---- Raylib ----
 
-add_executable(raylib_cc atoi_cc.cc ../vendor/libcc/libcc.cc)
+add_executable(raylib_cc raylib_cc.cc ../vendor/libcc/libcc.cc)
 target_include_directories(raylib_cc PRIVATE ..)
-add_dependencies(raylib_cc raylib)
+target_link_libraries(raylib_cc PRIVATE Threads::Threads raylib)
 
 if(WIN32)
     target_compile_definitions(raylib_cc PRIVATE _CRT_SECURE_NO_WARNINGS _CRT_NONSTDC_NO_DEPRECATE)
     target_link_libraries(raylib_cc PRIVATE ws2_32)
+else()
+    target_link_libraries(raylib_cc PRIVATE dl)
 endif()

package/benchmark/atoi_cc.cc

@@ -36,7 +36,7 @@ int Main(int argc, char **argv)
     int64_t start = GetMonotonicTime();
 
     for (int i = 0; i < iterations; i++) {
-        sum += (uint64_t)atoi(strings[i % RG_LEN(strings)]);
+        sum = sum + (uint64_t)atoi(strings[i % RG_LEN(strings)]);
     }
 
     // Help prevent optimisation of loop

package/benchmark/raylib_cc.cc

@@ -31,13 +31,13 @@ int Main(int argc, char **argv)
     SetWindowState(FLAG_WINDOW_HIDDEN);
     InitWindow(640, 480, "Raylib Test");
 
-    Image img = GenImageColor(800, 600, (Color){ .r = 0, .g = 0, .b = 0, .a = 255 });
+    Image img = GenImageColor(800, 600, Color { .r = 0, .g = 0, .b = 0, .a = 255 });
     Font font = GetFontDefault();
 
     int64_t start = GetMonotonicTime();
 
     for (int i = 0; i < iterations; i++) {
-        ImageClearBackground(&img, (Color){ .r = 0, .g = 0, .b = 0, .a = 255 });
+        ImageClearBackground(&img, Color { .r = 0, .g = 0, .b = 0, .a = 255 });
 
         for (int j = 0; j < 3600; j++) {
             const char *text = "Hello World!";

package/benchmark/raylib_koffi.js

@@ -82,7 +82,7 @@ function main() {
     }
     console.log('Iterations:', iterations);
 
-    let lib_filename = path.dirname(__filename) + '/test/build/raylib' + koffi.extension;
+    let lib_filename = path.dirname(__filename) + '/build/raylib' + koffi.extension;
     let lib = koffi.load(lib_filename);
 
     const InitWindow = lib.cdecl('InitWindow', 'void', ['int', 'int', 'string']);

package/benchmark/raylib_node_ffi.js

@@ -98,14 +98,7 @@ function main() {
     }
     console.log('Iterations:', iterations);
 
-    let iterations = parseInt(process.argv[2], 10);
-    if (Number.isNaN(iterations))
-        throw new Error('Not a valid number');
-    if (iterations < 1)
-        throw new Error('Value must be positive');
-    console.log('Iterations:', iterations);
-
-    let lib_filename = path.dirname(__filename) + '/test/build/raylib' + koffi.extension;
+    let lib_filename = path.dirname(__filename) + '/build/raylib' + koffi.extension;
 
     const r = ffi.Library(lib_filename, {
         InitWindow: ['void', ['int', 'int', 'string']],

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "koffi",
-  "version": "0.9.27",
+  "version": "0.9.28",
   "description": "Fast and simple FFI (foreign function interface) for Node.js",
   "keywords": [
     "foreign",

package/src/call_arm32.cc

@@ -233,10 +233,10 @@ static Napi::Value TranslateCall(const Napi::CallbackInfo &info)
     // Return through registers unless it's too big
     if (RG_UNLIKELY(!call.AllocStack(func->args_size, 16, &args_ptr)))
         return env.Null();
-    if (RG_UNLIKELY(!call.AllocStack(4 * 4, 8, &gpr_ptr)))
-        return env.Null();
     if (RG_UNLIKELY(!call.AllocStack(8 * 8, 8, &vec_ptr)))
         return env.Null();
+    if (RG_UNLIKELY(!call.AllocStack(4 * 4, 8, &gpr_ptr)))
+        return env.Null();
     if (func->ret.use_memory) {
         if (RG_UNLIKELY(!call.AllocHeap(func->ret.type->size, 16, &return_ptr)))
             return env.Null();

package/src/call_arm32_fwd.S

@@ -56,22 +56,22 @@
 
 // Prepare general purpose argument registers from array passed by caller.
 .macro forward_int
-    ldr r3, [r1, 76]
-    ldr r2, [r1, 72]
-    ldr r0, [r1, 64]
-    ldr r1, [r1, 68]
+    ldr r3, [r1, 12]
+    ldr r2, [r1, 8]
+    ldr r0, [r1, 0]
+    ldr r1, [r1, 4]
 .endm
 
 // Prepare vector argument registers from array passed by caller.
 .macro forward_vec
-    vldr d7, [r1, 56]
-    vldr d6, [r1, 48]
-    vldr d5, [r1, 40]
-    vldr d4, [r1, 32]
-    vldr d3, [r1, 24]
-    vldr d2, [r1, 16]
-    vldr d1, [r1, 8]
-    vldr d0, [r1, 0]
+    vldr d7, [r1, 72]
+    vldr d6, [r1, 64]
+    vldr d5, [r1, 56]
+    vldr d4, [r1, 48]
+    vldr d3, [r1, 40]
+    vldr d2, [r1, 32]
+    vldr d1, [r1, 24]
+    vldr d0, [r1, 16]
 .endm
 
 ForwardCallGG:

package/src/util.cc

@@ -81,23 +81,38 @@ const char *CallData::PushString(const Napi::Value &value)
 {
     RG_ASSERT(value.IsString());
 
+    const Size SmallSize = 32;
+
     Napi::Env env = value.Env();
     napi_status status;
 
+    Span<char> buf;
     size_t len = 0;
-    status = napi_get_value_string_utf8(env, value, nullptr, 0, &len);
-    RG_ASSERT(status == napi_ok);
 
-    Span<char> buf;
-    buf.len = (Size)len + 1;
+    // Optimize for small strings
+    buf.len = SmallSize;
     if (RG_UNLIKELY(!AllocHeap(buf.len, 1, &buf.ptr)))
         return nullptr;
-
     if (RG_UNLIKELY(napi_get_value_string_utf8(env, value, buf.ptr, (size_t)buf.len, &len) != napi_ok)) {
         ThrowError<Napi::Error>(env, "Failed to convert string to UTF-8");
         return nullptr;
     }
 
+    // Slow path for bigger strings
+    if (len >= SmallSize - 1) {
+        status = napi_get_value_string_utf8(env, value, nullptr, 0, &len);
+        RG_ASSERT(status == napi_ok);
+        RG_ASSERT(len >= SmallSize);
+
+        buf.len = (Size)len + 1;
+        if (RG_UNLIKELY(!AllocHeap(buf.len - SmallSize, 1)))
+            return nullptr;
+        if (RG_UNLIKELY(napi_get_value_string_utf8(env, value, buf.ptr, (size_t)buf.len, &len) != napi_ok)) {
+            ThrowError<Napi::Error>(env, "Failed to convert string to UTF-8");
+            return nullptr;
+        }
+    }
+
     return buf.ptr;
 }
 

package/src/util.hh

@@ -101,8 +101,8 @@ public:
 
     template <typename T = void>
     bool AllocStack(Size size, Size align, T **out_ptr = nullptr);
-    template <typename T>
-    bool AllocHeap(Size size, Size align, T **out_ptr);
+    template <typename T = void>
+    bool AllocHeap(Size size, Size align, T **out_ptr = nullptr);
 
     const char *PushString(const Napi::Value &value);
     bool PushObject(const Napi::Object &obj, const TypeInfo *type, uint8_t *dest);
@@ -151,7 +151,9 @@ bool CallData::AllocHeap(Size size, Size align, T **out_ptr)
     heap_mem->ptr += delta;
     heap_mem->len -= delta;
 
-    *out_ptr = (T *)ptr;
+    if (out_ptr) {
+        *out_ptr = (T *)ptr;
+    }
     return true;
 }
 

package/vendor/sqlite3/shell.c

@@ -7964,9 +7964,14 @@ static int zipfileFilter(
     zipfileCursorErr(pCsr, "zipfile() function requires an argument");
     return SQLITE_ERROR;
   }else if( sqlite3_value_type(argv[0])==SQLITE_BLOB ){
+    static const u8 aEmptyBlob = 0;
     const u8 *aBlob = (const u8*)sqlite3_value_blob(argv[0]);
     int nBlob = sqlite3_value_bytes(argv[0]);
     assert( pTab->pFirstEntry==0 );
+    if( aBlob==0 ){
+      aBlob = &aEmptyBlob;
+      nBlob = 0;
+    }
     rc = zipfileLoadDirectory(pTab, aBlob, nBlob);
     pCsr->pFreeEntry = pTab->pFirstEntry;
     pTab->pFirstEntry = pTab->pLastEntry = 0;
@@ -19906,7 +19911,7 @@ static int do_meta_command(char *zLine, ShellState *p){
 
   if( c=='i' && strncmp(azArg[0], "import", n)==0 ){
     char *zTable = 0;           /* Insert data into this table */
-    char *zSchema = "main";     /* within this schema */
+    char *zSchema = 0;          /* within this schema (may default to "main") */
     char *zFile = 0;            /* Name of file to extra content from */
     sqlite3_stmt *pStmt = NULL; /* A statement */
     int nCol;                   /* Number of columns in the table */
@@ -19915,11 +19920,13 @@ static int do_meta_command(char *zLine, ShellState *p){
     int needCommit;             /* True to COMMIT or ROLLBACK at end */
     int nSep;                   /* Number of bytes in p->colSeparator[] */
     char *zSql;                 /* An SQL statement */
+    char *zFullTabName;         /* Table name with schema if applicable */
     ImportCtx sCtx;             /* Reader context */
     char *(SQLITE_CDECL *xRead)(ImportCtx*); /* Func to read one value */
     int eVerbose = 0;           /* Larger for more console output */
     int nSkip = 0;              /* Initial lines to skip */
     int useOutputMode = 1;      /* Use output mode to determine separators */
+    char *zCreate = 0;          /* CREATE TABLE statement text */
 
     failIfSafeMode(p, "cannot run .import in safe mode");
     memset(&sCtx, 0, sizeof(sCtx));
@@ -20042,7 +20049,6 @@ static int do_meta_command(char *zLine, ShellState *p){
       import_cleanup(&sCtx);
       goto meta_command_exit;
     }
-    /* Below, resources must be freed before exit. */
     if( eVerbose>=2 || (eVerbose>=1 && useOutputMode) ){
       char zSep[2];
       zSep[1] = 0;
@@ -20054,11 +20060,17 @@ static int do_meta_command(char *zLine, ShellState *p){
       output_c_string(p->out, zSep);
       utf8_printf(p->out, "\n");
     }
+    /* Below, resources must be freed before exit. */
     while( (nSkip--)>0 ){
       while( xRead(&sCtx) && sCtx.cTerm==sCtx.cColSep ){}
     }
-    zSql = sqlite3_mprintf("SELECT * FROM \"%w\".\"%w\"", zSchema, zTable);
-    if( zSql==0 ){
+    if( zSchema!=0 ){
+      zFullTabName = sqlite3_mprintf("\"%w\".\"%w\"", zSchema, zTable);
+    }else{
+      zFullTabName = sqlite3_mprintf("\"%w\"", zTable);
+    }
+    zSql = sqlite3_mprintf("SELECT * FROM %s", zFullTabName);
+    if( zSql==0 || zFullTabName==0 ){
       import_cleanup(&sCtx);
       shell_out_of_memory();
     }
@@ -20066,11 +20078,10 @@ static int do_meta_command(char *zLine, ShellState *p){
     rc = sqlite3_prepare_v2(p->db, zSql, -1, &pStmt, 0);
     import_append_char(&sCtx, 0);    /* To ensure sCtx.z is allocated */
     if( rc && sqlite3_strglob("no such table: *", sqlite3_errmsg(p->db))==0 ){
-      char *zCreate = sqlite3_mprintf("CREATE TABLE \"%w\".\"%w\"",
-                                      zSchema, zTable);
       sqlite3 *dbCols = 0;
       char *zRenames = 0;
       char *zColDefs;
+      zCreate = sqlite3_mprintf("CREATE TABLE %s", zFullTabName);
       while( xRead(&sCtx) ){
         zAutoColumn(sCtx.z, &dbCols, 0);
         if( sCtx.cTerm!=sCtx.cColSep ) break;
@@ -20084,9 +20095,12 @@ static int do_meta_command(char *zLine, ShellState *p){
       }
       assert(dbCols==0);
       if( zColDefs==0 ){
+        utf8_printf(stderr,"%s: empty file\n", sCtx.zFile);
+      import_fail:
         sqlite3_free(zCreate);
+        sqlite3_free(zSql);
+        sqlite3_free(zFullTabName);
         import_cleanup(&sCtx);
-        utf8_printf(stderr,"%s: empty file\n", sCtx.zFile);
         rc = 1;
         goto meta_command_exit;
       }
@@ -20097,22 +20111,18 @@ static int do_meta_command(char *zLine, ShellState *p){
       rc = sqlite3_exec(p->db, zCreate, 0, 0, 0);
       if( rc ){
         utf8_printf(stderr, "%s failed:\n%s\n", zCreate, sqlite3_errmsg(p->db));
-        sqlite3_free(zCreate);
-        import_cleanup(&sCtx);
-        rc = 1;
-        goto meta_command_exit;
+        goto import_fail;
       }
       sqlite3_free(zCreate);
+      zCreate = 0;
       rc = sqlite3_prepare_v2(p->db, zSql, -1, &pStmt, 0);
     }
-    sqlite3_free(zSql);
     if( rc ){
       if (pStmt) sqlite3_finalize(pStmt);
       utf8_printf(stderr,"Error: %s\n", sqlite3_errmsg(p->db));
-      import_cleanup(&sCtx);
-      rc = 1;
-      goto meta_command_exit;
+      goto import_fail;
     }
+    sqlite3_free(zSql);
     nCol = sqlite3_column_count(pStmt);
     sqlite3_finalize(pStmt);
     pStmt = 0;
@@ -20122,8 +20132,7 @@ static int do_meta_command(char *zLine, ShellState *p){
       import_cleanup(&sCtx);
       shell_out_of_memory();
     }
-    sqlite3_snprintf(nByte+20, zSql, "INSERT INTO \"%w\".\"%w\" VALUES(?",
-                     zSchema, zTable);
+    sqlite3_snprintf(nByte+20, zSql, "INSERT INTO %s VALUES(?", zFullTabName);
     j = strlen30(zSql);
     for(i=1; i<nCol; i++){
       zSql[j++] = ',';
@@ -20135,14 +20144,13 @@ static int do_meta_command(char *zLine, ShellState *p){
       utf8_printf(p->out, "Insert using: %s\n", zSql);
     }
     rc = sqlite3_prepare_v2(p->db, zSql, -1, &pStmt, 0);
-    sqlite3_free(zSql);
     if( rc ){
       utf8_printf(stderr, "Error: %s\n", sqlite3_errmsg(p->db));
       if (pStmt) sqlite3_finalize(pStmt);
-      import_cleanup(&sCtx);
-      rc = 1;
-      goto meta_command_exit;
+      goto import_fail;
     }
+    sqlite3_free(zSql);
+    sqlite3_free(zFullTabName);
     needCommit = sqlite3_get_autocommit(p->db);
     if( needCommit ) sqlite3_exec(p->db, "BEGIN", 0, 0, 0);
     do{
@@ -22636,7 +22644,8 @@ static int process_input(ShellState *p){
       qss = QSS_Start;
     }
   }
-  if( nSql && QSS_PLAINDARK(qss) ){
+  if( nSql ){
+    /* This may be incomplete. Let the SQL parser deal with that. */
     errCnt += runOneSqlLine(p, zSql, p->in, startline);
   }
   free(zSql);