> Date: Mon, 8 Nov 2021 13:33:27 +
> From: nia
>
> On Sat, Nov 06, 2021 at 09:42:04AM +, Taylor R Campbell wrote:
> > That said, since we already argon2 logic as part of libcrypt, does it
> > make sense to have another copy in cgdconfig?
> >
> > I guess the main issue is with pthreads.
On Sat, Nov 06, 2021 at 09:42:04AM +, Taylor R Campbell wrote:
> Err... I don't think this will do what you want it to do. The effect
> will presumably be to add something like
>
> -Wl,-Bstatic -Wl,-Bdynamic ... -largon2
>
> to the linker command line eventually, because PROGDPLIBS ge
On 2021/11/06 18:42, Taylor R Campbell wrote:
--- /dev/null 1 Jan 1970 00:00:00 -
+++ sbin/cgdconfig/argon2_utils.c 6 Nov 2021 00:17:48 -
[...]
+ mem = usermem / 10;
What units are these in? Maybe add a comment explaining so the number
10 is a little more obvious?
> Date: Sat, 6 Nov 2021 00:20:58 +
> From: nia
>
> > Why dup what you just created? Why not kg->kg_key = bits_new(raw,
> > keylen)? This looks like a (minor) memory leak.
>
> This is what the existing code for pkcs5_pbkdf2/sha1 does. I believe
> that the returned pointer (raw) is immediate
On Fri, Nov 05, 2021 at 11:19:33PM +, Taylor R Campbell wrote:
> (For future patches: please use the `-p' option with cvs diff!)
Today I learned something new!
> It looks like we'll descend into external/apache2/argon2 twice this
> way. Am I mistaken? Is this intentional?
Uh, not intention
> Date: Fri, 5 Nov 2021 16:25:05 +
> From: nia
>
> This patch adds an "argon2id" password-based key generation method
> to cgdconfig(8).
Cool, thanks for working on this!
(For future patches: please use the `-p' option with cvs diff!)
> +++ lib/Makefile 5 Nov 2021 15:41:41 -
> @@
This patch adds an "argon2id" password-based key generation method
to cgdconfig(8).
To support Argon2, several new keygen parameters are supported
in cgd parameters files:
memory (integer, in kilobytes)
parallelism (integer, usually the number of CPU cores)
version (intege
