On 1/8/2020 5:42 PM, Andrey Smirnov wrote:
> @@ -275,12 +276,25 @@ static int instantiate_rng(struct device *ctrldev, int state_handle_mask,
> return -ENOMEM;
>
> for (sh_idx = 0; sh_idx < RNG4_MAX_HANDLES; sh_idx++) {
> + const u32 rdsta_if = RDSTA_IF0 << sh_idx;
> + const u32 rdsta_pr = RDSTA_PR0 << sh_idx;
> + const u32 rdsta_mask = rdsta_if | rdsta_pr;
> /*
> * If the corresponding bit is set, this state handle
> * was initialized by somebody else, so it's left alone.
> */
> - if ((1 << sh_idx) & state_handle_mask)
> - continue;
> + if (rdsta_if & state_handle_mask) {
> + if (rdsta_pr & state_handle_mask)
> + continue;
> +
> + dev_info(ctrldev,
> + "RNG4 SH%d was previously instantiated without prediction resistance. Tearing it down\n",
> + sh_idx);
> +
> + ret = deinstantiate_rng(ctrldev, rdsta_if);
> + if (ret)
> + break;
In case state handle 0 is deinstantiated, its reinstantiation with PR support
will have the side effect of re-generating JDKEK, TDKEK, TDSK.
This needs to be avoided, since other SW components (like OP-TEE f/w)
could have black keys in use. Overwriting the KEK registers would no longer
allow CAAM to decrypt them.
Horia
