920899 - fipstest has bad memset in hmac_test

Status: RESOLVED FIXED

(NSS :: Test, defect, P2)

Description

[Mike Frysinger]

Attachment: fipstest.patch

when compiling with recent versions of gcc, this warning is issued:

fipstest.c: In function ‘hmac_test’:
fipstest.c:3680:35: warning: argument to ‘sizeof’ in ‘memset’ call is the same expression as the destination; did you mean to provide an explicit length? [-Wsizeof-pointer-memaccess]
             memset(msg, 0, sizeof msg);  
                                   ^

normally in this code base, buffers are declared on the stack like so:
    unsigned char HMAC[HASH_LENGTH_MAX];
that way when the code does the memset like it does for msg, it works.

but this particular case allocs msg dynamically:
    unsigned int msgLen = 128;    /* the length of the input  */
                                  /*  Message is always 128 Bytes */
    unsigned char *msg = NULL;    /* holds the message to digest.*/
...
    msg = PORT_ZAlloc(msgLen);
    memset(msg, 0, msgLen);

so this later memset in the code which does sizeof msg is wrong -- it only zereos out the first sizeof(void*) bytes instead of all 128.

attached patch fixes this.  it also fixes an earlier memset in this func which is run before the NULL pointer check.  it's a possible NULL pointer deref.

Comment 1

[Wan-Teh Chang]

Attachment: Patch v2, by Mike Frysinger

Mike: thank you for the patch. The first memset is not necessary
because PORT_ZAlloc (a simple wrapper around calloc) zeros the
returned memory.

In fact, the |msg| buffer can be allocated on the stack like the
|key| and |HMAC| buffers.

https://hg.mozilla.org/projects/nss/rev/574d0c70a2c7

Comment 2

[Kai Engert (:KaiE:)]

changing target milestone to 3.15.4