Hi Herbert, > > On Fri, Apr 15, 2016 at 02:32:42PM +0000, Tudor-Dan Ambarus wrote: > > > > On Wed, Apr 06, 2016 at 04:37:04PM +0300, Tudor Ambarus wrote: > > > > > This approach has the advantage that users can select specific > > > > > parser actions by using a general parser with function pointers > > > > > to specific actions. > > > > > > > > I don't understand why we need different parsing functions in the > > > > first place. Can't they just return raw integers always? > > > > > > > > You can then trivially convert the raw integers to MPI, no? > > > > > > We need different parsing functions so that we don't allocate duplicate > > buffers for the same data. > > > > > > You need to allocate buffers when getting the raw integers and you need > > to allocate other (duplicate) buffers when converting the raw integers to > > MPI. > > > > > > Using the proposed API each user can select the format of data he > wants, > > eliminating the need of a double conversion (with its drawbacks: > duplicate > > buffers, unnecessary cycles). > > > > The double allocation only happens with software RSA, right? I > > don't really think that is going to matter considering how slow > > it is, no? > > The double allocation happens only with software RSA. I can't estimate the > performance hit for the double conversion. Naturally is to parse the data > only once. Plus, my suggestion is simple, faster and doesn't waste > resources. Do you want me to address some suggestions, or do you intend to apply this patch set? Thanks, ta -- To unsubscribe from this list: send the line "unsubscribe linux-crypto" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html