On Wed, 11 Nov 2015 10:25:01 -0700 Stephen Warren <swarren@xxxxxxxxxxxxx> wrote: > On 11/09/2015 10:19 AM, Alban Bedel wrote: > > Add the support code needed to sign the RCM with RSA-PSS as needed > > to communicate with secured production devices. This mode is enabled > > by passing the --key command line argument. If such a key is set the > > RCM messages will be signed with it. > > IIRC, (at least some) Tegra chips support both SBK (which I believe uses > the CMAC hash) and (RSA) PKC. "--key" is a bit of a generic term. It > seems best to rename this cmdline option --pkc to make it clear which of > the two options it represents, and to allow possible future addition of > --sbk support without backwards compatibility issues or > inconsistency/confusion in cmdline option naming. You are right, I'll change this. > > diff --git a/src/main.c b/src/main.c > > > @@ -123,6 +124,10 @@ static void usage(char *progname) > > > + fprintf(stderr, "\t--key=<key.ber>\n"); > > + fprintf(stderr, "\t\tSpecify the key file for secured devices. The key should be\n"); > > + fprintf(stderr, "\t\tin DER format\n"); > > Is that the same format cbootimage uses for its keys? I want to make > sure we're not requiring users to convert keys to different formats in > order to use different tools. cbootimage only take binary blobs as input, so at the moment it doesn't really matter. The example signing script use openssl, which use PEM format per default. I don't know if openssl can autodetect the key format but the script could definitely be made to also support keys in DER format. However I used DER because it is the only format directly supported by Crypto++. I would prefer PEM as that's the more common format, but then we would need to add a PEM parser. > > diff --git a/src/rsa-pss.cpp b/src/rsa-pss.cpp > > Please add a copyright header to the new files. I'll do. Alban
Attachment:
signature.asc
Description: PGP signature
