In get_token(), we could read one byte past the end of the
region. This is only a big problem in the case where the region
itself is (a) potentially hostile, and (b) not explicitly
nul-terminated.
This patch fixes the underlying bug, and also makes sure that the
one remaining case of not-NUL-terminated potentially hostile data
gets NUL-terminated.
Fix for bug 21018, TROVE-2016-12-002, and CVE-2016-1254
--- /dev/null
+ o Major bugfixes (parsing, security):
+
+ - Fix a bug in parsing that could cause clients to read a single
+ byte past the end of an allocated region. This bug could be
+ used to cause hardened clients (built with
+ --enable-expensive-hardening) to crash if they tried to visit
+ a hostile hidden service. Non-hardened clients are only
+ affected depending on the details of their platform's memory
+ allocator. Fixes bug 21018; bugfix on 0.2.0.8-alpha. Found by
+ using libFuzzer. Also tracked as TROVE-2016-12-002 and as
+ CVE-2016-1254.
if (tok->tp == ERR_) {
/* No keyword matched; call it an "K_opt" or "A_unrecognized" */
- if (**s == '@')
+ if (*s < eol && **s == '@')
tok->tp = A_UNKNOWN_;
else
tok->tp = K_OPT;
crypto_cipher_free(cipher);
len = ipos_encrypted_size - 2 - client_entries_len - CIPHER_IV_LEN;
- dec = tor_malloc(len);
+ dec = tor_malloc_zero(len + 1);
declen = crypto_cipher_decrypt_with_iv(session_key, dec, len,
ipos_encrypted + 2 + client_entries_len,
ipos_encrypted_size - 2 - client_entries_len);
"small.");
return -1;
}
- dec = tor_malloc_zero(ipos_encrypted_size - CIPHER_IV_LEN - 1);
+ dec = tor_malloc_zero(ipos_encrypted_size - CIPHER_IV_LEN - 1 + 1);
declen = crypto_cipher_decrypt_with_iv(descriptor_cookie, dec,
ipos_encrypted_size -