Re: [Ocf-linux-users] [PATCH - EP80579 driver update patch]

[David M. <dav...@mc...>]

Jivin Kennedy, Brendan lays it down ...
> Hi Dave, All,
> 
> Attached is an update to the EP80579 driver. It allows it to compile
> under FreeBSD Opencrypto framework and fixes a few bugs.
> 
> Would it be possible to include the updated code in your later releases
> of OCF?

Done,  will be in the next release.

> There??s also an update to cryptodev.c to allow a distinction between
> a process fail and algorithm fail. This helps DSA verify functionality
> when, for example, certain key sizes are not supported.
> 
> There is a corresponding OpenSSL eng_cryptodev patch (third attachment)
> if you want to check it out. This fixes some of the PKE ?? OCF calls.

Ok,  some questions for you on this one.

For the change:

-   if (cryptodev_asym(&kop, BN_num_bytes(m), r, 0, NULL) == -1) {
+   if (cryptodev_asym(&kop, BN_num_bytes(m), r, 0, NULL)) {
+       printf("OCF asym process failed, Running in software\n");
+       const RSA_METHOD *meth = RSA_PKCS1_SSLeay();
+       ret = meth->bn_mod_exp(r, a, p, m, ctx, in_mont);
+
+   } else if (ECANCELED == kop.crk_status) {
+       printf("OCF hardware operation cancelled. Running in Software\n");
        const RSA_METHOD *meth = RSA_PKCS1_SSLeay();
        ret = meth->bn_mod_exp(r, a, p, m, ctx, in_mont);
    }
+   /* else cryptodev operation worked ok ==> ret = 1*/
+

This changes the behaviour for other drivers (safe/ixp) that will fall back
to software on an ERANGE/E2BIG or any other error.

Did you just want to be able to identify the HW condition that causes this
separate to an error ?  If so,  then something like:

  if (cryptodev_asym(&kop, BN_num_bytes(m), r, 0, NULL)) {
      const RSA_METHOD *meth = RSA_PKCS1_SSLeay();
      printf("OCF asym process failed, Running in software\n");
      ret = meth->bn_mod_exp(r, a, p, m, ctx, in_mont);
  } else if (kop.crk_status) {
      const RSA_METHOD *meth = RSA_PKCS1_SSLeay();
      printf("OCF hardware operation failed(%d). Running in Software\n",
              kop.crk_status);
      ret = meth->bn_mod_exp(r, a, p, m, ctx, in_mont);
  }
  /* else cryptodev operation worked ok ==> ret = 1*/

Should do.  I don't think ECANCELED needs to be any different than any other
driver error,  but perhaps I am missing something.

If this sounds ok I will fix up teh patch on the way in,

Cheers,
Davidm

-- 
David McCullough,      dav...@mc...,  Ph:+61 734352815
McAfee - SnapGear      http://www.mcafee.com         http://www.uCdot.org