Difference between revisions of "KBAG"

From The iPhone Wiki
Jump to: navigation, search
m (How it Works: another link update)
m (v)
Line 5: Line 5:
 
===KBAG128===
 
===KBAG128===
 
typedef struct {
 
typedef struct {
uint32_t magic; // string with bits flipped = "KBAG" (magic = 0x4741424B)
+
uint32_t magic; // string with bytes flipped ("KBAG" in little endian)
 
uint32_t fullSize; // size of KBAG from beyond that point to the end of it
 
uint32_t fullSize; // size of KBAG from beyond that point to the end of it
 
uint32_t tagDataSize; // size of KBAG without this 0xC header
 
uint32_t tagDataSize; // size of KBAG without this 0xC header
Line 17: Line 17:
 
===KBAG192===
 
===KBAG192===
 
typedef struct Unparsed_KBAG_AES192 {
 
typedef struct Unparsed_KBAG_AES192 {
uint32_t magic; // string with bits flipped = "KBAG" (magic = 0x4741424B)
+
uint32_t magic; // string with bytes flipped ("KBAG" in little endian)
 
uint32_t fullSize; // size of KBAG from beyond that point to the end of it
 
uint32_t fullSize; // size of KBAG from beyond that point to the end of it
 
uint32_t tagDataSize; // size of KBAG without this 0xC header
 
uint32_t tagDataSize; // size of KBAG without this 0xC header
Line 29: Line 29:
 
===KBAG256===
 
===KBAG256===
 
typedef struct Unparsed_KBAG_256 {
 
typedef struct Unparsed_KBAG_256 {
uint32_t magic; // string with bits flipped = "KBAG" (magic = 0x4741424B)
+
uint32_t magic; // string with bytes flipped ("KBAG" in little endian)
 
uint32_t fullSize; // size of KBAG from beyond that point to the end of it
 
uint32_t fullSize; // size of KBAG from beyond that point to the end of it
 
uint32_t tagDataSize; // size of KBAG without this 0xC header
 
uint32_t tagDataSize; // size of KBAG without this 0xC header

Revision as of 21:01, 17 September 2014

Explanation

In Apple's new IMG3 security scheme, they have used something called a KBAG. At the bottom of a firmware file, you will see something that will, on the ASCII side of your hex editor, see "GABK", which, as ARM is little-endian based, is "KBAG" flipped. Look on the hex side and you will the KBAG according to this format:

KBAG Format

KBAG128

typedef struct {
     uint32_t magic;       // string with bytes flipped ("KBAG" in little endian)
     uint32_t fullSize;    // size of KBAG from beyond that point to the end of it
     uint32_t tagDataSize; // size of KBAG without this 0xC header
     uint32_t cryptState;  // 1 if the key and IV in the KBAG are encrypted with the GID Key
                           // 2 is used with a second KBAG for the S5L8920, use is unknown.
     uint32_t aesType;     // 0x80 = aes128 / 0xc0 = aes192 / 0x100 = aes256
     uint8_t encIV[16];    // IV for the firmware file, encrypted with the GID Key
     uint8_t encKey[16];   // Key for the firmware file, encrypted with the GID Key
} UnparsedKbagAes128_t;

KBAG192

typedef struct Unparsed_KBAG_AES192 {
     uint32_t magic;       // string with bytes flipped ("KBAG" in little endian)
     uint32_t fullSize;    // size of KBAG from beyond that point to the end of it
     uint32_t tagDataSize; // size of KBAG without this 0xC header
     uint32_t cryptState;  // 1 if the key and IV in the KBAG are encrypted with the GID Key
                           // 2 is used with a second KBAG for the S5L8920, use is unknown.
     uint32_t aesType;     // 0x80 = aes128 / 0xc0 = aes192 / 0x100 = aes256
     uint8_t encIV[16];    // IV for the firmware file, encrypted with the GID Key
     uint8_t encKey[24];   // Key for the firmware file, encrypted with the GID Key
} UparsedKbagAes192_t;

KBAG256

typedef struct Unparsed_KBAG_256 {
     uint32_t magic;       // string with bytes flipped ("KBAG" in little endian)
     uint32_t fullSize;    // size of KBAG from beyond that point to the end of it
     uint32_t tagDataSize; // size of KBAG without this 0xC header
     uint32_t cryptState;  // 1 if the key and IV in the KBAG are encrypted with the GID Key
                           // 2 is used with a second KBAG for the S5L8920, use is unknown.
     uint32_t aesType;     // 0x80 = aes128 / 0xc0 = aes192 / 0x100 = aes256
     uint8_t encIV[16];    // IV for the firmware file, encrypted with the GID Key
     uint8_t encKey[32];   // Key for the firmware file, encrypted with the GID Key
} UparsedKbagAes256_t;

How it Works

Basically, it just boils down to using the GID Key to decrypt encIV and encKey, then using that key and IV to decrypt the DATA section of the file (the code itself).

As an interesting side note, because of the circumstances with the IMG3 File Format, the kernel never needs to even touch the GID Key anymore, as it's job is to just flash the image to the NOR as is, with container and all.

in order to decrypt the KBAG you need to remove them from IMG3 using this command dd if=IMG3_FILE bs=1 skip=4741424B count=0x70