Make room for new ZEND_VM_ flags

Split OP flags and EXT flags in two separate sets, in zend_vm_gen.php to
maintain 32bit support. In Zend/zend_vm_opcodes.c we can unify them again.

Author: arnaud-lb

Zend/Optimizer/zend_dump.c

@@ -460,7 +460,7 @@ static void zend_dump_range_constraint(const zend_op_array *op_array, const zend
 ZEND_API void zend_dump_op(const zend_op_array *op_array, const zend_basic_block *b, const zend_op *opline, uint32_t dump_flags, const zend_ssa *ssa, const zend_ssa_op *ssa_op)
 {
 	const char *name = zend_get_opcode_name(opline->opcode);
-	uint32_t flags = zend_get_opcode_flags(opline->opcode);
+	uint64_t flags = zend_get_opcode_flags(opline->opcode);
 	uint32_t n = 0;
 
 	if (!ssa_op || ssa_op->result_use < 0) {

Zend/zend_vm_gen.php

@@ -56,21 +56,36 @@
 const ZEND_VM_KIND_TAILCALL = 5;
 
 $vm_op_flags = array(
-    "ZEND_VM_OP_SPEC"         => 1<<0,
+    "ZEND_VM_OP_SHIFT"        => 16,
+
+    "ZEND_VM_OP_SPEC_MASK"    => 0x1ff,
     "ZEND_VM_OP_CONST"        => 1<<1,
-    "ZEND_VM_OP_TMPVAR"       => 1<<2,
-    "ZEND_VM_OP_TMPVARCV"     => 1<<3,
-    "ZEND_VM_OP_MASK"         => 0xf0,
-    "ZEND_VM_OP_NUM"          => 0x10,
-    "ZEND_VM_OP_JMP_ADDR"     => 0x20,
-    "ZEND_VM_OP_TRY_CATCH"    => 0x30,
-    "ZEND_VM_OP_LOOP_END"     => 0x40,
-    "ZEND_VM_OP_THIS"         => 0x50,
-    "ZEND_VM_OP_NEXT"         => 0x60,
-    "ZEND_VM_OP_CLASS_FETCH"  => 0x70,
-    "ZEND_VM_OP_CONSTRUCTOR"  => 0x80,
-    "ZEND_VM_OP_CONST_FETCH"  => 0x90,
-    "ZEND_VM_OP_CACHE_SLOT"   => 0xa0,
+    "ZEND_VM_OP_TMP"          => 1<<2,
+    "ZEND_VM_OP_VAR"          => 1<<3,
+    "ZEND_VM_OP_UNUSED"       => 1<<4,
+    "ZEND_VM_OP_CV"           => 1<<5,
+    "ZEND_VM_OP_TMPVAR"       => 1<<6,
+    "ZEND_VM_OP_TMPVARCV"     => 1<<7,
+    "ZEND_VM_OP_REG"          => 1<<8,
+
+    // unused bits: 9-11
+
+    "ZEND_VM_OP_MASK"         => 0xf000,
+    "ZEND_VM_OP_NUM"          => 1<<12,
+    "ZEND_VM_OP_JMP_ADDR"     => 2<<12,
+    "ZEND_VM_OP_TRY_CATCH"    => 3<<12,
+    "ZEND_VM_OP_LOOP_END"     => 4<<12,
+    "ZEND_VM_OP_THIS"         => 5<<12,
+    "ZEND_VM_OP_NEXT"         => 6<<12,
+    "ZEND_VM_OP_CLASS_FETCH"  => 7<<12,
+    "ZEND_VM_OP_CONSTRUCTOR"  => 8<<12,
+    "ZEND_VM_OP_CONST_FETCH"  => 9<<12,
+    "ZEND_VM_OP_CACHE_SLOT"   => 10<<12,
+    // unused: (11<<12)-(15<<12)
+);
+
+$vm_ext_flags = array(
+    // unused: bits 1-15
 
     "ZEND_VM_EXT_VAR_FETCH"   => 1<<16,
     "ZEND_VM_EXT_ISSET"       => 1<<17,
@@ -100,15 +115,20 @@
     define($name, $val);
 }
 
+foreach ($vm_ext_flags as $name => $val) {
+    define($name, $val);
+}
+
 $vm_op_decode = array(
     "ANY"                  => 0,
-    "CONST"                => ZEND_VM_OP_SPEC | ZEND_VM_OP_CONST,
-    "TMP"                  => ZEND_VM_OP_SPEC,
-    "VAR"                  => ZEND_VM_OP_SPEC,
-    "UNUSED"               => ZEND_VM_OP_SPEC,
-    "CV"                   => ZEND_VM_OP_SPEC,
-    "TMPVAR"               => ZEND_VM_OP_SPEC | ZEND_VM_OP_TMPVAR,
-    "TMPVARCV"             => ZEND_VM_OP_SPEC | ZEND_VM_OP_TMPVARCV,
+    "CONST"                => ZEND_VM_OP_CONST,
+    "TMP"                  => ZEND_VM_OP_TMP,
+    "VAR"                  => ZEND_VM_OP_VAR,
+    "UNUSED"               => ZEND_VM_OP_UNUSED,
+    "CV"                   => ZEND_VM_OP_CV,
+    "TMPVAR"               => ZEND_VM_OP_TMPVAR,
+    "TMPVARCV"             => ZEND_VM_OP_TMPVARCV,
+    "REG"                  => ZEND_VM_OP_REG,
     "NUM"                  => ZEND_VM_OP_NUM,
     "JMP_ADDR"             => ZEND_VM_OP_JMP_ADDR,
     "TRY_CATCH"            => ZEND_VM_OP_TRY_CATCH,
@@ -1885,7 +1905,7 @@ function gen_executor($f, $skl, $spec, $kind, $executor_name, $initializer_name)
                     out($f,"#define SPEC_RULE_ISSET        0x01000000\n");
                     out($f,"#define SPEC_RULE_OBSERVER     0x02000000\n");
                     out($f,"\n");
-                    out($f,"static const uint32_t *zend_spec_handlers;\n");
+                    out($f,"const uint32_t *zend_spec_handlers;\n");
                     out($f,"static zend_vm_opcode_handler_t const *zend_opcode_handlers;\n");
                     out($f,"static int zend_handlers_count;\n");
                     if ($kind == ZEND_VM_KIND_HYBRID) {
@@ -2389,7 +2409,7 @@ function parse_operand_spec($def, $lineno, $str, &$flags) {
             die("ERROR ($def:$lineno): Wrong operand type '$str'\n");
         }
     }
-    if (!($flags & ZEND_VM_OP_SPEC)) {
+    if (!($flags & ZEND_VM_OP_SPEC_MASK)) {
         if (count($a) != 1) {
             die("ERROR ($def:$lineno): Wrong operand type '$str'\n");
         }
@@ -2398,18 +2418,32 @@ function parse_operand_spec($def, $lineno, $str, &$flags) {
     return array_flip($a);
 }
 
-function parse_ext_spec($def, $lineno, $str) {
+function parse_ext_spec($def, $lineno, $str, $spec_str) {
     global $vm_ext_decode;
 
     $flags = 0;
-    $a = explode("|",$str);
-    foreach ($a as $val) {
-        if (isset($vm_ext_decode[$val])) {
-            $flags |= $vm_ext_decode[$val];
-        } else {
-            die("ERROR ($def:$lineno): Wrong extended_value type '$str'\n");
+
+    if ($str !== '') {
+        $a = explode("|",$str);
+        foreach ($a as $val) {
+            if (isset($vm_ext_decode[$val])) {
+                $flags |= $vm_ext_decode[$val];
+            } else {
+                die("ERROR ($def:$lineno): Wrong extended_value type '$str'\n");
+            }
+        }
+    }
+    if ($spec_str !== '') {
+        $a = explode(",",$spec_str);
+        foreach ($a as $val) {
+            if (isset($vm_ext_decode[$val])) {
+                $flags |= $vm_ext_decode[$val];
+            } else {
+                // Spec flags are validated separately.
+            }
         }
     }
+
     return $flags;
 }
 
@@ -2464,7 +2498,7 @@ function parse_spec_rules($def, $lineno, $str) {
 }
 
 function gen_vm_opcodes_header(
-    array $opcodes, int $max_opcode, int $max_opcode_len, array $vm_op_flags
+    array $opcodes, int $max_opcode, int $max_opcode_len, array $vm_op_flags, array $vm_ext_flags
 ): string {
     $str = HEADER_TEXT;
     $str .= "#ifndef ZEND_VM_OPCODES_H\n#define ZEND_VM_OPCODES_H\n\n";
@@ -2538,12 +2572,15 @@ function gen_vm_opcodes_header(
     foreach ($vm_op_flags as $name => $val) {
         $str .= sprintf("#define %-24s 0x%08x\n", $name, $val);
     }
-    $str .= "#define ZEND_VM_OP1_FLAGS(flags) (flags & 0xff)\n";
-    $str .= "#define ZEND_VM_OP2_FLAGS(flags) ((flags >> 8) & 0xff)\n";
+    foreach ($vm_ext_flags as $name => $val) {
+        $str .= sprintf("#define %-24s (UINT64_C(0x%08x) << 32)\n", $name, $val);
+    }
+    $str .= sprintf("#define ZEND_VM_OP1_FLAGS(flags) (flags & 0x%08x)\n", (1<<ZEND_VM_OP_SHIFT)-1);
+    $str .= sprintf("#define ZEND_VM_OP2_FLAGS(flags) ((flags >> %d) & 0x%08x)\n", ZEND_VM_OP_SHIFT, (1<<ZEND_VM_OP_SHIFT)-1);
     $str .= "\n";
     $str .= "BEGIN_EXTERN_C()\n\n";
     $str .= "ZEND_API const char* ZEND_FASTCALL zend_get_opcode_name(uint8_t opcode);\n";
-    $str .= "ZEND_API uint32_t ZEND_FASTCALL zend_get_opcode_flags(uint8_t opcode);\n";
+    $str .= "ZEND_API uint64_t ZEND_FASTCALL zend_get_opcode_flags(uint8_t opcode);\n";
     $str .= "ZEND_API uint8_t zend_get_opcode_id(const char *name, size_t length);\n\n";
     $str .= "END_EXTERN_C()\n\n";
 
@@ -2567,7 +2604,7 @@ function gen_vm_opcodes_header(
 function gen_vm($def, $skel) {
     global $definition_file, $skeleton_file, $executor_file,
         $op_types, $list, $opcodes, $helpers, $params, $opnames,
-        $vm_op_flags, $used_extra_spec;
+        $vm_op_flags, $vm_ext_flags, $used_extra_spec;
 
     // Load definition file
     $in = @file($def);
@@ -2632,10 +2669,8 @@ function gen_vm($def, $skel) {
             $len  = strlen($op);
             $op1  = parse_operand_spec($def, $lineno, $m[4], $flags1);
             $op2  = parse_operand_spec($def, $lineno, $m[5], $flags2);
-            $flags = $flags1 | ($flags2 << 8);
-            if (!empty($m[7])) {
-                $flags |= parse_ext_spec($def, $lineno, $m[7]);
-            }
+            $flags = $flags1 | ($flags2 << ZEND_VM_OP_SHIFT);
+            $ext_flags = parse_ext_spec($def, $lineno, $m[7] ?? '', $m[9] ?? '');
 
             if ($len > $max_opcode_len) {
                 $max_opcode_len = $len;
@@ -2649,14 +2684,14 @@ function gen_vm($def, $skel) {
             if (isset($opnames[$op])) {
                 die("ERROR ($def:$lineno): Opcode with name '$op' is already defined.\n");
             }
-            $opcodes[$code] = array("op"=>$op,"op1"=>$op1,"op2"=>$op2,"code"=>"","flags"=>$flags,"hot"=>$hot);
+            $opcodes[$code] = array("op"=>$op,"op1"=>$op1,"op2"=>$op2,"code"=>"","flags"=>$flags,"ext_flags"=>$ext_flags,"hot"=>$hot);
             if (isset($m[9])) {
                 $opcodes[$code]["spec"] = parse_spec_rules($def, $lineno, $m[9]);
                 if (isset($opcodes[$code]["spec"]["NO_CONST_CONST"])) {
-                    $opcodes[$code]["flags"] |= $vm_op_flags["ZEND_VM_NO_CONST_CONST"];
+                    $opcodes[$code]["ext_flags"] |= ZEND_VM_NO_CONST_CONST;
                 }
                 if (isset($opcodes[$code]["spec"]["COMMUTATIVE"])) {
-                    $opcodes[$code]["flags"] |= $vm_op_flags["ZEND_VM_COMMUTATIVE"];
+                    $opcodes[$code]["ext_flags"] |= ZEND_VM_COMMUTATIVE;
                 }
             }
             $opnames[$op] = $code;
@@ -2701,23 +2736,21 @@ function gen_vm($def, $skel) {
             $op = $m[4];
             $op1 = parse_operand_spec($def, $lineno, $m[5], $flags1);
             $op2 = parse_operand_spec($def, $lineno, $m[6], $flags2);
-            $flags = $flags1 | ($flags2 << 8);
-            if (!empty($m[8])) {
-                $flags |= parse_ext_spec($def, $lineno, $m[8]);
-            }
+            $flags = $flags1 | ($flags2 << ZEND_VM_OP_SHIFT);
+            $ext_flags = parse_ext_spec($def, $lineno, $m[8] ?? '', $m[10] ?? '');
 
             if (isset($opcodes[$code])) {
                 die("ERROR ($def:$lineno): Opcode with name '$code' is already defined.\n");
             }
             $used_extra_spec["TYPE"] = 1;
-            $opcodes[$code] = array("op"=>$op,"op1"=>$op1,"op2"=>$op2,"code"=>"","flags"=>$flags,"hot"=>$hot,"is_type_spec"=>true);
+            $opcodes[$code] = array("op"=>$op,"op1"=>$op1,"op2"=>$op2,"code"=>"","flags"=>$flags,"ext_flags"=>$ext_flags,"hot"=>$hot,"is_type_spec"=>true);
             if (isset($m[10])) {
                 $opcodes[$code]["spec"] = parse_spec_rules($def, $lineno, $m[10]);
                 if (isset($opcodes[$code]["spec"]["NO_CONST_CONST"])) {
-                    $opcodes[$code]["flags"] |= $vm_op_flags["ZEND_VM_NO_CONST_CONST"];
+                    $opcodes[$code]["ext_flags"] |= ZEND_VM_NO_CONST_CONST;
                 }
                 if (isset($opcodes[$code]["spec"]["COMMUTATIVE"])) {
-                    $opcodes[$code]["flags"] |= $vm_op_flags["ZEND_VM_COMMUTATIVE"];
+                    $opcodes[$code]["ext_flags"] |= ZEND_VM_COMMUTATIVE;
                 }
             }
             $opnames[$op] = $code;
@@ -2836,7 +2869,7 @@ function gen_vm($def, $skel) {
     }
 
     // Generate opcode #defines (zend_vm_opcodes.h)
-    $str = gen_vm_opcodes_header($opcodes, $max_opcode, $max_opcode_len, $vm_op_flags);
+    $str = gen_vm_opcodes_header($opcodes, $max_opcode, $max_opcode_len, $vm_op_flags, $vm_ext_flags);
     write_file_if_changed(__DIR__ . "/zend_vm_opcodes.h", $str);
     echo "zend_vm_opcodes.h generated successfully.\n";
 
@@ -2855,9 +2888,9 @@ function gen_vm($def, $skel) {
     }
     out($f, "};\n\n");
 
-    out($f,"static uint32_t zend_vm_opcodes_flags[".($max_opcode + 1)."] = {\n");
+    out($f,"static uint64_t zend_vm_opcodes_flags[".($max_opcode + 1)."] = {\n");
     for ($i = 0; $i <= $max_opcode; $i++) {
-        out($f, sprintf("\t0x%08x,\n", isset($opcodes[$i]["flags"]) ? $opcodes[$i]["flags"] : 0));
+        out($f, sprintf("\tUINT64_C(0x%08x) | (UINT64_C(0x%08x) << 32), /* %s */\n", $opcodes[$i]["flags"] ?? 0, $opcodes[$i]["ext_flags"] ?? 0, $opcodes[$i]["op"] ?? ''));
     }
     out($f, "};\n\n");
 
@@ -2868,7 +2901,7 @@ function gen_vm($def, $skel) {
     out($f, "\treturn zend_vm_opcodes_names[opcode];\n");
     out($f, "}\n");
 
-    out($f, "ZEND_API uint32_t ZEND_FASTCALL zend_get_opcode_flags(uint8_t opcode) {\n");
+    out($f, "ZEND_API uint64_t ZEND_FASTCALL zend_get_opcode_flags(uint8_t opcode) {\n");
     out($f, "\tif (UNEXPECTED(opcode > ZEND_VM_LAST_OPCODE)) {\n");
     out($f, "\t\topcode = ZEND_NOP;\n");
     out($f, "\t}\n");