Alan/Andreas:
I'm guessing that the value assigned to ssh_key_struct.type
(ssh_keytypes_e) comes directly from the client. Is this true?
If so, the ssh_keytypes_e enum is incorrectly defined.
The definition in libssh is:
enum ssh_keytypes_e{
SSH_KEYTYPE_UNKNOWN=0,
SSH_KEYTYPE_DSS=1,
SSH_KEYTYPE_RSA,
SSH_KEYTYPE_RSA1,
SSH_KEYTYPE_ECDSA
};
However, this disagrees from the key-type enum in the OpenSSH client
(key.h):
enum types {
KEY_RSA1,
KEY_RSA,
KEY_DSA,
KEY_ECDSA,
KEY_ED25519,
KEY_RSA_CERT,
KEY_DSA_CERT,
KEY_ECDSA_CERT,
KEY_ED25519_CERT,
KEY_RSA_CERT_V00,
KEY_DSA_CERT_V00,
KEY_UNSPEC
};
It's a bug, right?
Dustin
On Mon, Feb 17, 2014 at 11:37 AM, Alan Dunn <[email protected]> wrote:
> These issues are just the way Andreas wanted me to report the problems
> I found. The patches contained in those issues are just the original
> patch in this thread broken into two smaller patches. So it's
> perfectly fine to just apply the initial patch in this thread for
> testing (though you'll need the other patch I proposed to be able to
> activate ECDSA signatures and some code to actually use an ECDSA key
> with an ssh_bind), though I imagine the patches in the issue links are
> what will be ultimately applied in libssh master.
>
> On Mon, Feb 17, 2014 at 10:16 AM, Dustin Oprea <[email protected]>
> wrote:
> >
> > On Feb 17, 2014 11:10 AM, "Alan Dunn" <[email protected]> wrote:
> >>
> >> Done:
> >>
> >> https://red.libssh.org/issues/147
> >> https://red.libssh.org/issues/148
> >>
> >> Thanks,
> >> - Alan
> >>
> >
> > Do the issues have to be fixed prior to trying your patch out, or does
> your
> > patch account for them?
> >
> > Dustin
>
>
