A likely copy-pasta made its way through review which made the `rend_cell_auth`
never checked for all-zero.
This is not that problematic has the chances for SHA3 to generate a zeroed
buffer is close to impossible but defense in depth always.
Related to #41228
Signed-off-by: David Goulet <dgoulet@torproject.org>
crypto_mac_sha3_256(rend_cell_auth, sizeof(rend_cell_auth),
rend_auth_input, sizeof(rend_auth_input),
(const uint8_t *)T_HSMAC, strlen(T_HSMAC));
- bad |= safe_mem_is_zero(ntor_verify, DIGEST256_LEN);
+ bad |= safe_mem_is_zero(rend_cell_auth, DIGEST256_LEN);
{ /* Get the computed RENDEZVOUS1 material! */
memcpy(&hs_ntor_rend_cell_keys_out->rend_cell_auth_mac,