bugfix: if the len of iv is not 96-bit, y0 can be calculated incorrectly.
An initialization vector IV can have any number of bits between 1 and 2^64. So it should be filled to the lower 64-bit in the last step when computing ghash. Signed-off-by: openluopworld <luopengxq@gmail.com>
This commit is contained in:
parent
eb009232c0
commit
eab65acca4
1 changed files with 3 additions and 1 deletions
|
@ -254,6 +254,7 @@ int mbedtls_gcm_starts( mbedtls_gcm_context *ctx,
|
|||
size_t i;
|
||||
const unsigned char *p;
|
||||
size_t use_len, olen = 0;
|
||||
uint64_t iv_bits;
|
||||
|
||||
GCM_VALIDATE_RET( ctx != NULL );
|
||||
GCM_VALIDATE_RET( iv != NULL );
|
||||
|
@ -278,7 +279,8 @@ int mbedtls_gcm_starts( mbedtls_gcm_context *ctx,
|
|||
else
|
||||
{
|
||||
memset( work_buf, 0x00, 16 );
|
||||
MBEDTLS_PUT_UINT64_BE( iv_len * 8, work_buf, 8 );
|
||||
iv_bits = (uint64_t)iv_len * 8;
|
||||
MBEDTLS_PUT_UINT64_BE( iv_bits, work_buf, 8 );
|
||||
|
||||
p = iv;
|
||||
while( iv_len > 0 )
|
||||
|
|
Loading…
Reference in a new issue