fix endianness

classfile.c

@@ -1,4 +1,5 @@
 #include "classfile.h"
+#include <assert.h>
 #include <memory.h>
 #include <stddef.h>
 #include <stdio.h>
@@ -14,12 +15,12 @@ void read_u2(void *ptr, size_t n, FILE *file) {
     buffer[i] = (buffer[i] >> 8) | (buffer[i] << 8);
   }
 #endif
-  memcpy(ptr, &buffer, n*sizeof(u2));
+  memcpy(ptr, &buffer, n * sizeof(u2));
 }
 
 void read_u4(void *ptr, size_t n, FILE *file) {
   u4 buffer[n];
-  fread(&buffer, sizeof(u2), n, file);
+  fread(&buffer, sizeof(u4), n, file);
 #ifdef SWAP_ENDIANNESS
   for (size_t i = 0; i < n; i++) {
     buffer[i] = ((buffer[i] >> 24) & 0xff) |    // move byte 3 to byte 0
@@ -28,7 +29,7 @@ void read_u4(void *ptr, size_t n, FILE *file) {
                 ((buffer[i] << 24) & 0xff000000);
   }
 #endif
-  memcpy(ptr, &buffer, n*sizeof(u4));
+  memcpy(ptr, &buffer, n * sizeof(u4));
 }
 
 ClassFile *ClassFile_load(const char *path) {
@@ -40,96 +41,125 @@ ClassFile *ClassFile_load(const char *path) {
 
   // magic 4bytes, minor_version 2bytes, major_version 2bytes,
   // constant_pool_count 2bytes
-  fread(&(cf->magic), 4 + 2 + 2 + 2, 1, file);
+  read_u4(&(cf->magic), 1, file);
+  read_u2(&(cf->minor_version), 3, file);
 
   // constant_pool
   cf->constant_pool =
       (cp_info *)malloc(sizeof(cp_info) * (cf->constant_pool_count - 1));
   for (u2 i = 0; i < (cf->constant_pool_count - 1); i++) {
     // cp_info.tag 2bytes
-    fread(&(cf->constant_pool[i].tag), 2, 1, file);
+    fread(&(cf->constant_pool[i].tag), 1, 1, file);
     size_t size;
     switch (cf->constant_pool[i].tag) {
     case CONSTANT_Class:
-      size = sizeof(CONSTANT_Class_info);
+      read_u2(&(cf->constant_pool[i].info.class_info.name_index), 1, file);
       break;
     case CONSTANT_Fieldref:
-      size = sizeof(CONSTANT_Fieldref_info);
+      read_u2(&(cf->constant_pool[i].info.fieldref_info.class_index), 1, file);
+      read_u2(&(cf->constant_pool[i].info.fieldref_info.name_and_type_index), 1,
+              file);
       break;
     case CONSTANT_Methodref:
-      size = sizeof(CONSTANT_Methodref_info);
+      read_u2(&(cf->constant_pool[i].info.methodref_info.class_index), 1, file);
+      read_u2(&(cf->constant_pool[i].info.methodref_info.name_and_type_index),
+              1, file);
       break;
     case CONSTANT_InterfaceMethodref:
-      size = sizeof(CONSTANT_InterfaceMethodref_info);
+      read_u2(&(cf->constant_pool[i].info.interface_methodred_info.class_index),
+              1, file);
+      read_u2(&(cf->constant_pool[i]
+                    .info.interface_methodred_info.name_and_type_index),
+              1, file);
       break;
     case CONSTANT_String:
-      size = sizeof(CONSTANT_String_info);
+      read_u2(&(cf->constant_pool[i].info.string_info.string_index), 1, file);
       break;
     case CONSTANT_Integer:
-      size = sizeof(CONSTANT_Integer_info);
+      read_u4(&(cf->constant_pool[i].info.integer_info.bytes), 1, file);
       break;
     case CONSTANT_Float:
-      size = sizeof(CONSTANT_Float_info);
+      read_u4(&(cf->constant_pool[i].info.float_info.bytes), 1, file);
       break;
     case CONSTANT_Long:
-      size = sizeof(CONSTANT_Long_info);
+      read_u4(&(cf->constant_pool[i].info.long_info.high_bytes), 1, file);
+      read_u4(&(cf->constant_pool[i].info.long_info.low_bytes), 1, file);
       break;
     case CONSTANT_Double:
-      size = sizeof(CONSTANT_Double_info);
+      read_u4(&(cf->constant_pool[i].info.double_info.high_bytes), 1, file);
+      read_u4(&(cf->constant_pool[i].info.double_info.low_bytes), 1, file);
       break;
     case CONSTANT_NameAndType:
-      size = sizeof(CONSTANT_NameAndType_info);
+      read_u2(&(cf->constant_pool[i].info.name_and_type_info.name_index), 1,
+              file);
+      read_u2(&(cf->constant_pool[i].info.name_and_type_info.descriptor_index),
+              1, file);
       break;
     case CONSTANT_Utf8:
-      size = sizeof(CONSTANT_Utf8_info);
+      read_u2(&(cf->constant_pool[i].info.utf8_info.length), 1, file);
+      fread(&(cf->constant_pool[i].info.utf8_info.bytes),
+            cf->constant_pool[i].info.utf8_info.length, 1, file);
       break;
     case CONSTANT_MethodHandle:
-      size = sizeof(CONSTANT_MethodHandle_info);
+      fread(&(cf->constant_pool[i].info.method_handle_info.reference_kind), 1,
+            1, file);
+      read_u2(&(cf->constant_pool[i].info.method_handle_info.reference_index),
+              1, file);
       break;
     case CONSTANT_MethodType:
-      size = sizeof(CONSTANT_MethodType_info);
+      read_u2(&(cf->constant_pool[i].info.method_type_info.descriptor_index), 1,
+              file);
       break;
     case CONSTANT_Dynamic:
-      size = sizeof(CONSTANT_Dynamic_info);
+      read_u2(
+          &(cf->constant_pool[i].info.dynamic_info.bootstrap_method_attr_index),
+          1, file);
+      read_u2(&(cf->constant_pool[i].info.dynamic_info.name_and_type_index), 1,
+              file);
       break;
     case CONSTANT_InvokeDynamic:
-      size = sizeof(CONSTANT_InvokeDynamic_info);
+      read_u2(&(cf->constant_pool[i]
+                    .info.invoke_dynamic_info.bootstrap_method_attr_index),
+              1, file);
+      read_u2(
+          &(cf->constant_pool[i].info.invoke_dynamic_info.name_and_type_index),
+          1, file);
       break;
     case CONSTANT_Module:
-      size = sizeof(CONSTANT_Module_info);
+      read_u2(&(cf->constant_pool[i].info.module_info.name_index), 1, file);
       break;
     case CONSTANT_Package:
-      size = sizeof(CONSTANT_Package_info);
+      read_u2(&(cf->constant_pool[i].info.package_info.name_index), 1, file);
       break;
+    default:
+      printf("unknown constant tag: %hhu\n", cf->constant_pool[i].tag);
+      // assert(0);
     }
-
-    cf->constant_pool[i].info = (void *)malloc(size);
-    fread(cf->constant_pool[i].info, size, 1, file);
   }
 
   // access_flags 2bytes, this_class 2bytes, super_class 2bytes, interface_count
-  // 2bytes
-  fread(&(cf->access_flags), 2 + 2 + 2 + 2, 1, file);
+  // 2byte
+  read_u2(&(cf->access_flags), 4, file);
 
   // interfaces
   cf->interfaces = (u2 *)malloc(2 * cf->interfaces_count);
-  fread(cf->interfaces, 2 * cf->interfaces_count, 1, file);
+  read_u2(cf->interfaces, cf->interfaces_count, file);
 
   // fields_count 2bytes
-  fread(&(cf->fields_count), 2, 1, file);
+  read_u2(&(cf->fields_count), 1, file);
 
   // fields
   cf->fields = (field_info *)malloc(sizeof(field_info) * cf->fields_count);
   for (u2 i = 0; i < cf->fields_count; i++) {
     // access_flags 2bytes, name_index 2bytes, descriptor_index 2bytes,
     // attributes_count 2bytes
-    fread(&(cf->fields[i].access_flags), 2 + 2 + 2 + 2, 1, file);
+    read_u2(&(cf->fields[i].access_flags), 4, file);
     cf->fields[i].attributes = (attribute_info *)malloc(
         sizeof(attribute_info) * cf->fields[i].attributes_count);
     for (u2 j = 0; j < cf->fields[i].attributes_count; j++) {
       // attribute_name_index 2bytes, attribute_length 4bytes
-      fread(&(cf->fields[i].attributes[j].attribute_name_index), 2 + 4, 1,
-            file);
+      read_u2(&(cf->fields[i].attributes[j].attribute_name_index), 1, file);
+      read_u4(&(cf->fields[i].attributes[j].attribute_length), 1, file);
       cf->fields[i].attributes[j].info =
           (u1 *)malloc(cf->fields[i].attributes[j].attribute_length);
       fread(cf->fields[i].attributes[j].info,
@@ -138,20 +168,20 @@ ClassFile *ClassFile_load(const char *path) {
   }
 
   // methods_count 2bytes
-  fread(&(cf->methods_count), 2, 1, file);
+  read_u2(&(cf->methods_count), 1, file);
 
   // methods
   cf->methods = (method_info *)malloc(sizeof(method_info) * cf->methods_count);
   for (u2 i = 0; i < cf->methods_count; i++) {
     // access_flags 2bytes, name_index 2bytes, descriptor_index 2bytes,
     // attributes_count 2bytes
-    fread(&(cf->methods[i].access_flags), 2 + 2 + 2 + 2, 1, file);
+    read_u2(&(cf->methods[i].access_flags), 4, file);
     cf->methods[i].attributes = (attribute_info *)malloc(
         sizeof(attribute_info) * cf->methods[i].attributes_count);
     for (u2 j = 0; j < cf->methods[i].attributes_count; j++) {
       // attribute_name_index 2bytes, attribute_length 4bytes
-      fread(&(cf->methods[i].attributes[j].attribute_name_index), 2 + 4, 1,
-            file);
+      read_u2(&(cf->methods[i].attributes[j].attribute_name_index), 1, file);
+      read_u4(&(cf->methods[i].attributes[j].attribute_length), 1, file);
       cf->methods[i].attributes[j].info =
           (u1 *)malloc(cf->methods[i].attributes[j].attribute_length);
       fread(cf->methods[i].attributes[j].info,
@@ -160,14 +190,15 @@ ClassFile *ClassFile_load(const char *path) {
   }
 
   // attributes_count 2bytes
-  fread(&(cf->attributes_count), 2, 1, file);
+  read_u2(&(cf->attributes_count), 1, file);
 
   // attributes
   cf->attributes =
       (attribute_info *)malloc(sizeof(attribute_info) * cf->attributes_count);
   for (u2 i = 0; i < cf->attributes_count; i++) {
     // attribute_name_index 2bytes, attribute_length 4bytes
-    fread(&(cf->attributes[i].attribute_name_index), 2 + 4, 1, file);
+    read_u2(&(cf->attributes[i].attribute_name_index), 1, file);
+    read_u4(&(cf->attributes[i].attribute_length), 1, file);
     cf->attributes[i].info = (u1 *)malloc(cf->attributes[i].attribute_length);
     fread(cf->attributes[i].info, cf->attributes[i].attribute_length, 1, file);
   }
@@ -182,7 +213,7 @@ void ClassFile_info(const ClassFile *cf) {
          "minor_version=%hu\n"
          "major_version=%hu\n"
          "constant_pool_count=%hu\n"
-         "access_flags=%hu\n"
+         "access_flags=%hx\n"
          "this_class=%hu\n"
          "super_class=%hu\n"
          "interfaces_count=%hu\n"

classfile.h

@@ -52,17 +52,10 @@ typedef enum {
 } CONSTANT_TAGS;
 
 typedef struct {
-  u1 tag;
-  void *info;
-} cp_info;
-
-typedef struct {
-  u1 tag;
   u2 name_index;
 } CONSTANT_Class_info;
 
 typedef struct {
-  u1 tag;
   u2 class_index;
   u2 name_and_type_index;
 } CONSTANT_Fieldref_info;
@@ -70,63 +63,76 @@ typedef CONSTANT_Fieldref_info CONSTANT_Methodref_info;
 typedef CONSTANT_Fieldref_info CONSTANT_InterfaceMethodref_info;
 
 typedef struct {
-  u1 tag;
   u2 string_index;
 } CONSTANT_String_info;
 
 typedef struct {
-  u1 tag;
   u4 bytes;
 } CONSTANT_Integer_info;
 typedef CONSTANT_Integer_info CONSTANT_Float_info;
 
 typedef struct {
-  u1 tag;
   u4 high_bytes;
   u4 low_bytes;
 } CONSTANT_Long_info;
 typedef CONSTANT_Long_info CONSTANT_Double_info;
 
 typedef struct {
-  u1 tag;
   u2 name_index;
   u2 descriptor_index;
 } CONSTANT_NameAndType_info;
 
 typedef struct {
-  u1 tag;
   u2 length;
   u1 *bytes;
 } CONSTANT_Utf8_info;
 
 typedef struct {
-  u1 tag;
   u1 reference_kind;
   u2 reference_index;
 } CONSTANT_MethodHandle_info;
 
 typedef struct {
-  u1 tag;
   u2 descriptor_index;
 } CONSTANT_MethodType_info;
 
 typedef struct {
-  u1 tag;
   u2 bootstrap_method_attr_index;
   u2 name_and_type_index;
 } CONSTANT_Dynamic_info;
 typedef CONSTANT_Dynamic_info CONSTANT_InvokeDynamic_info;
 
 typedef struct {
-  u1 tag;
   u2 name_index;
 } CONSTANT_Module_info;
 
 typedef struct {
-  u1 tag;
   u2 name_index;
 } CONSTANT_Package_info;
 
+typedef struct {
+  u1 tag;
+  union {
+    CONSTANT_Class_info class_info;
+    CONSTANT_Fieldref_info fieldref_info;
+    CONSTANT_Methodref_info methodref_info;
+    CONSTANT_InterfaceMethodref_info interface_methodred_info;
+    CONSTANT_String_info string_info;
+    CONSTANT_Integer_info integer_info;
+    CONSTANT_Float_info float_info;
+    CONSTANT_Long_info long_info;
+    CONSTANT_Double_info double_info;
+    CONSTANT_NameAndType_info name_and_type_info;
+    CONSTANT_Utf8_info utf8_info;
+    CONSTANT_MethodHandle_info method_handle_info;
+    CONSTANT_MethodType_info method_type_info;
+    CONSTANT_Dynamic_info dynamic_info;
+    CONSTANT_InvokeDynamic_info invoke_dynamic_info;
+    CONSTANT_Module_info module_info;
+    CONSTANT_Package_info package_info;
+  } info;
+} cp_info;
+
 typedef struct {
   u2 attribute_name_index;
   u4 attribute_length;