More detailed comments and documentation

native/jni/su/cert.cpp

@@ -2,9 +2,13 @@
 
 using namespace std;
 
+#define APK_SIGNING_BLOCK_MAGIC     "APK Sig Block 42"
+#define SIGNATURE_SCHEME_V2_MAGIC   0x7109871a
+#define EOCD_MAGIC                  0x6054b50
+
 // Top-level block container
 struct signing_block {
-    uint64_t len;
+    uint64_t block_sz;
 
     struct id_value_pair {
         uint64_t len;
@@ -14,7 +18,7 @@ struct signing_block {
         };
     } id_value_pair_sequence[0];
 
-    uint64_t block_len;   // *MUST* be same as len
+    uint64_t block_sz_;   // *MUST* be same as block_sz
     char magic[16];       // "APK Sig Block 42"
 };
 
@@ -59,78 +63,124 @@ struct v2_signature {
     } signer_sequence[0];
 };
 
-// The structures above are just for documentation purpose
+// End of central directory record
-// The real parsing logic is the following
+struct EOCD {
+    uint32_t magic;            // 0x6054b50
+    uint8_t pad[8];            // 8 bytes of irrelevant data
+    uint32_t central_dir_sz;   // size of central directory
+    uint32_t central_dir_off;  // offset of central directory
+    uint16_t comment_sz;       // size of comment
+    char comment[0];
+} __attribute__((packed));
 
+/*
+ * A v2/v3 signed APK has the format as following
+ *
+ * +---------------+
+ * | zip content   |
+ * +---------------+
+ * | signing block |
+ * +---------------+
+ * | central dir   |
+ * +---------------+
+ * | EOCD          |
+ * +---------------+
+ *
+ * Scan from end of file to find EOCD, and figure our way back to the
+ * offset of the signing block. Next, directly extract the certificate
+ * from the v2 signature block.
+ *
+ * All structures above are mostly just for documentation purpose.
+ *
+ * This method extracts the first certificate of the first signer
+ * within the APK v2 signature block.
+ */
 string read_certificate(int fd) {
-    string certificate;
     uint32_t size4;
-    uint64_t size8, size_of_block;
+    uint64_t size8;
 
+    // Find EOCD
     for (int i = 0;; i++) {
-        unsigned short n;
+        // i is the absolute offset to end of file
-        lseek(fd, -i - 2, SEEK_END);
+        uint16_t comment_sz = 0;
-        read(fd, &n, 2);
+        lseek(fd, -((off_t) sizeof(comment_sz)) - i, SEEK_END);
-        if (n == i) {
+        read(fd, &comment_sz, sizeof(comment_sz));
-            lseek(fd, -22, SEEK_CUR);
+        if (comment_sz == i) {
-            read(fd, &size4, 4);
+            // Double check if we actually found the structure
-            if (size4 == 0x6054b50u) { // central directory end magic
+            lseek(fd, -((off_t) sizeof(EOCD)), SEEK_CUR);
+            uint32_t magic = 0;
+            read(fd, &magic, sizeof(magic));
+            if (magic == EOCD_MAGIC) {
                 break;
             }
         }
         if (i == 0xffff) {
-            return certificate;
+            // Comments cannot be longer than 0xffff (overflow), abort
+            return {};
         }
     }
 
-    lseek(fd, 12, SEEK_CUR);
+    // We are now at EOCD + sizeof(magic)
+    // Seek and read central_dir_off to find start of central directory
+    uint32_t central_dir_off = 0;
+    {
+        constexpr off_t off = offsetof(EOCD, central_dir_off) - sizeof(EOCD::magic);
+        lseek(fd, off, SEEK_CUR);
+    }
+    read(fd, &central_dir_off, sizeof(central_dir_off));
 
-    read(fd, &size4, 0x4);
+    // Next, find the start of the APK signing block
-    lseek(fd, (off_t) (size4 - 0x18), SEEK_SET);
+    {
-
+        constexpr int off = sizeof(signing_block::block_sz_) + sizeof(signing_block::magic);
-    read(fd, &size8, 0x8);
+        lseek(fd, (off_t) (central_dir_off - off), SEEK_SET);
-    char magic[0x10] = {0};
+    }
+    read(fd, &size8, sizeof(size8));  // size8 = block_sz_
+    char magic[sizeof(signing_block::magic)] = {0};
     read(fd, magic, sizeof(magic));
-    if (memcmp(magic, "APK Sig Block 42", sizeof(magic)) != 0) {
+    if (memcmp(magic, APK_SIGNING_BLOCK_MAGIC, sizeof(magic)) != 0) {
-        return certificate;
+        // Invalid signing block magic, abort
+        return {};
+    }
+    uint64_t signing_blk_sz = 0;
+    lseek(fd, (off_t) (central_dir_off - size8 - sizeof(signing_blk_sz)), SEEK_SET);
+    read(fd, &signing_blk_sz, sizeof(signing_blk_sz));
+    if (signing_blk_sz != size8) {
+        // block_sz != block_sz_, invalid signing block format, abort
+        return {};
     }
 
-    lseek(fd, (off_t) (size4 - (size8 + 0x8)), SEEK_SET);
+    // Finally, we are now at the beginning of the id-value pair sequence
-    read(fd, &size_of_block, 0x8);
-    if (size_of_block != size8) {
-        return certificate;
-    }
 
     for (;;) {
-        uint32_t id;
+        read(fd, &size8, sizeof(size8)); // id-value pair length
-        uint32_t offset;
+        if (size8 == signing_blk_sz) {
-        read(fd, &size8, 0x8); // sequence length
+            // Outside of the id-value pair sequence; actually reading block_sz_
-        if (size8 == size_of_block) {
             break;
         }
-        read(fd, &id, 0x4); // id
-        offset = 4;
 
-        if (id == 0x7109871au) {
+        uint32_t id;
-            read(fd, &size4, 0x4); // signer-sequence length
+        read(fd, &id, sizeof(id));
-            read(fd, &size4, 0x4); // signer length
+        if (id == SIGNATURE_SCHEME_V2_MAGIC) {
-            read(fd, &size4, 0x4); // signed data length
+            read(fd, &size4, sizeof(size4)); // signer sequence length
-            offset += 0x4 * 3;
 
-            read(fd, &size4, 0x4); // digests-sequence length
+            read(fd, &size4, sizeof(size4)); // signer length
-            lseek(fd, (off_t) (size4), SEEK_CUR);// skip digests
+            read(fd, &size4, sizeof(size4)); // signed data length
-            offset += 0x4 + size4;
 
-            read(fd, &size4, 0x4); // certificates length
+            read(fd, &size4, sizeof(size4)); // digest sequence length
-            read(fd, &size4, 0x4); // certificate length
+            lseek(fd, (off_t) (size4), SEEK_CUR); // skip all digests
-            offset += 0x4 * 2;
 
-            certificate.resize(size4);
+            read(fd, &size4, sizeof(size4)); // cert sequence length
-            read(fd, certificate.data(), size4);
+            read(fd, &size4, sizeof(size4)); // cert length
 
-            offset += size4;
+            string cert;
+            cert.resize(size4);
+            read(fd, cert.data(), size4);
+
+            return cert;
+        } else {
+            // Skip this id-value pair
+            lseek(fd, (off_t) (size8 - sizeof(id)), SEEK_CUR);
         }
-        lseek(fd, (off_t) (size8 - offset), SEEK_CUR);
     }
-    return certificate;
+    return {};
 }