Thu 25 Apr 2013 06:27:22 PM UTC, comment #2:
I was able to reproduce with --set-asset-tag=12313 (not sure why it did not with your example). I'm surprised this bug lingered so long. I guess most vendors do not check the pad.
I did a different patch though, and found another location in the code that needed to be changed. Here's the patch I commited for 1.2.6. I've of course given credit to you & Broadcom in the ChangeLog for finding the bug and the fix location.
===================================================================
--- libfreeipmi/interface/ipmi-rmcpplus-interface.c (revision 9608)
+++ libfreeipmi/interface/ipmi-rmcpplus-interface.c (working copy)
@@ -739,7 +739,7 @@
uint8_t iv[IPMI_CRYPT_AES_CBC_128_IV_LENGTH];
int iv_len;
uint8_t payload_buf[IPMI_MAX_PAYLOAD_LENGTH];
- uint8_t pad_len;
+ uint8_t pad_len, pad_tmp;
int payload_len, cipher_keylen, cipher_blocklen, encrypt_len;
/* Note: Confidentiality Key for AES_CBS_128 is K2 */
@@ -808,7 +808,11 @@
/* Pad the data appropriately */
/* +1 is for the pad length field */
- pad_len = IPMI_CRYPT_AES_CBC_128_BLOCK_LENGTH - ((payload_len + 1) % IPMI_CRYPT_AES_CBC_128_BLOCK_LENGTH);
+ pad_tmp = ((payload_len + 1) % IPMI_CRYPT_AES_CBC_128_BLOCK_LENGTH);
+ if (pad_tmp)
+ pad_len = IPMI_CRYPT_AES_CBC_128_BLOCK_LENGTH - pad_tmp;
+ else
+ pad_len = 0;
if ((payload_len + pad_len + 1) > IPMI_MAX_PAYLOAD_LENGTH)
{
@@ -821,8 +825,8 @@
unsigned int i;
for (i = 0; i < pad_len; i++)
payload_buf[payload_len + i] = i + 1;
- payload_buf[payload_len + pad_len] = pad_len;
}
+ payload_buf[payload_len + pad_len] = pad_len;
/* +1 for pad length field */
if ((encrypt_len = crypt_cipher_encrypt (IPMI_CRYPT_CIPHER_AES,
|