Android dex在内存中的存放位置

在Dalvik虚拟机对dex的加载过程的基础上，我们来分析下面代码中的第三步。

/dalvik/vm/native/dalvik_system_DexFile.cpp

static void Dalvik_dalvik_system_DexFile_openDexFileNative(const u4* args,
    JValue* pResult)
{
    StringObject* sourceNameObj = (StringObject*) args[0];
    StringObject* outputNameObj = (StringObject*) args[1];
    DexOrJar* pDexOrJar = NULL;
    JarFile* pJarFile;
    RawDexFile* pRawDexFile;
    char* sourceName;
    char* outputName;

    sourceName = dvmCreateCstrFromString(sourceNameObj);
    if (outputNameObj != NULL)
        outputName = dvmCreateCstrFromString(outputNameObj);
    else
        outputName = NULL;
     
    // 1、尝试把它当做一个后缀为.dex的DEX文件进行打开，得到RawDexFile结构数据
    // 2、如果打开失败，则把它当做一个包含有classes.dex文件的Zip文件进行打开，得到JarFile结构数据
    if (hasDexExtension(sourceName)
            && dvmRawDexFileOpen(sourceName, outputName, &pRawDexFile, false) == 0) {
        pDexOrJar = (DexOrJar*) malloc(sizeof(DexOrJar));
        pDexOrJar->isDex = true;
        pDexOrJar->pRawDexFile = pRawDexFile;
        pDexOrJar->pDexMemory = NULL;
    } else if (dvmJarFileOpen(sourceName, outputName, &pJarFile, false) == 0) {
        ALOGV("Opening DEX file '%s' (Jar)", sourceName);

        pDexOrJar = (DexOrJar*) malloc(sizeof(DexOrJar));
        pDexOrJar->isDex = false;
        pDexOrJar->pJarFile = pJarFile;
        pDexOrJar->pDexMemory = NULL;
    }

    if (pDexOrJar != NULL) {
        pDexOrJar->fileName = sourceName;
        // 3、把pDexOrJar这个结构体中的内容加到gDvm中的userDexFile结构的hash表中，以便dalvik以后的查找
        addToDexFileTable(pDexOrJar);
    } else {
        free(sourceName);
    }

    free(outputName);
    RETURN_PTR(pDexOrJar);
}

可以看到DexOrJar文件最终添加到gDvm中的userDexFile结构的hash表中，下面我们来具体查看addToDexFileTable方法。

static void addToDexFileTable(DexOrJar* pDexOrJar) {
    /*
     * Later on, we will receive this pointer as an argument and need
     * to find it in the hash table without knowing if it's valid or
     * not, which means we can't compute a hash value from anything
     * inside DexOrJar. We don't share DexOrJar structs when the same
     * file is opened multiple times, so we can just use the low 32
     * bits of the pointer as the hash.
     */
    u4 hash = (u4) pDexOrJar;
    void* result;

    dvmHashTableLock(gDvm.userDexFiles);
    result = dvmHashTableLookup(gDvm.userDexFiles, hash, pDexOrJar,
            hashcmpDexOrJar, true);
    dvmHashTableUnlock(gDvm.userDexFiles);

    if (result != pDexOrJar) {
        ALOGE("Pointer has already been added?");
        dvmAbort();
    }

    pDexOrJar->okayToFree = true;
}

需要说明的是，gDvm是一个DvmGlobals结构的全局变量。在DvmGlobals结构中有一个成员HashTable* userDexFiles;

DvmGlobals在/dalvik/vm/Globals.h中

也就是说，userDexFiles是一个HashTabled的引用，HashTabled的结构为：

/dalvik/vm/Hash.h

struct HashTable {
    int         tableSize;          /* must be power of 2 */
    int         numEntries;         /* current #of "live" entries */
    int         numDeadEntries;     /* current #of tombstone entries */
    HashEntry*  pEntries;           /* array on heap */
    HashFreeFunc freeFunc;
    pthread_mutex_t lock;
};

HashEntry* pEntries指向一个HashEntry的数组,HashEntry结构体中的data变量存放的就是pDexOrJar。

/dalvik/vm/Hash.h

struct HashEntry {
    u4 hashValue;
    void* data; //DexOrJar *pDexOrJar
};

上面代码的关键就是dvmHashTableLookup方法，所以我们继续查看该方法。

/dalvik/vm/Hash.cpp

void* dvmHashTableLookup(HashTable* pHashTable, u4 itemHash, void* item,
    HashCompareFunc cmpFunc, bool doAdd)
{
    HashEntry* pEntry;
    HashEntry* pEnd;
    void* result = NULL;

    assert(pHashTable->tableSize > 0);
    assert(item != HASH_TOMBSTONE);
    assert(item != NULL);

    /* jump to the first entry and probe for a match */
    pEntry = &pHashTable->pEntries[itemHash & (pHashTable->tableSize-1)];
    pEnd = &pHashTable->pEntries[pHashTable->tableSize];
    while (pEntry->data != NULL) {
        if (pEntry->data != HASH_TOMBSTONE &&
            pEntry->hashValue == itemHash &&
            (*cmpFunc)(pEntry->data, item) == 0)
        {
            /* match */
            //ALOGD("+++ match on entry %d", pEntry - pHashTable->pEntries);
            break;
        }

        pEntry++;
        if (pEntry == pEnd) {     /* wrap around to start */
            if (pHashTable->tableSize == 1)
                break;      /* edge case - single-entry table */
            pEntry = pHashTable->pEntries;
        }

        //ALOGI("+++ look probing %d...", pEntry - pHashTable->pEntries);
    }

    if (pEntry->data == NULL) {
        if (doAdd) {
            pEntry->hashValue = itemHash;
            pEntry->data = item;
            pHashTable->numEntries++;

            /*
             * We've added an entry.  See if this brings us too close to full.
             */
            if ((pHashTable->numEntries+pHashTable->numDeadEntries) * LOAD_DENOM
                > pHashTable->tableSize * LOAD_NUMER)
            {
                if (!resizeHash(pHashTable, pHashTable->tableSize * 2)) {
                    /* don't really have a way to indicate failure */
                    ALOGE("Dalvik hash resize failure");
                    dvmAbort();
                }
                /* note "pEntry" is now invalid */
            } else {
                //ALOGW("okay %d/%d/%d",
                //    pHashTable->numEntries, pHashTable->tableSize,
                //    (pHashTable->tableSize * LOAD_NUMER) / LOAD_DENOM);
            }

            /* full table is bad -- search for nonexistent never halts */
            assert(pHashTable->numEntries < pHashTable->tableSize);
            result = item;
        } else {
            assert(result == NULL);
        }
    } else {
        result = pEntry->data;
    }

    return result;
}

上面的操作就是将pDexOrJar以及它对应的hashValue封装到一个HashEntry，然后将这个HashEntry存放到HashEntry* pEntries指向的HashEntry数组中去。

他们之间的整个引用过程如下图所示：

dexlocation.png

也就是说我们可以通过gDvm.userDexFiles找到具体DexOrJar的位置，然后通过DexOrJar的pRawDexFile定位到RawDexFile的位置，进而通过RawDexFile的pDvmDex找到DvmDex的位置，最终就可以通过DvmDex的pDexFile来定位到DexFile的位置，它就是内存中的.odex文件内容，它里面同时也包含了.dex的内容。

参考文章：
http://bbs.pediy.com/thread-207190.htm