Crypto patches

Hi Herbert, Dave,

Sometime ago I sent a patch which makes twofish key setup code much smaller (2.5 times). Currently, it is so much unrolled that twofish module's code is ~35k on i386.

I am ok with people disagreeing with me and considering 35k not too big. However, IIRC that patch was not reviewed (I've got no reply) and I have no way of knowing whether it was deemed unacceptable or just was lost in the churn.

I want to rediff and resend it. I also want to send patches for other modules (e.g. camellia, aes).

Please let me know what is the best channels to send crypto patches through so that they are not lost. Thanks. --
vda

Sorry I can't find a copy in my mail box archive at all so it looks like it was lost on the way to my mail server.

Unfortunately emails are lossy so sometimes things to tend to get lost. The best thing to do if you get no reply at all is to resend your patch.

If I don't like something I will tell you in no uncertain terms :)

linux-crypto is the best list but please CC me too.

Thanks,

Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt

I'm going to unify the aes key setup on i386/x86_64/generic code on aes. It should work from what I thing but I am not sure :) Are planning/doing the same thing on aes?

Sebastian

I was thinking about measuring whether 64bit aes is really faster than i386 one. i386 code seems to be already maxed out (at the first glance), apart from using stack for temp storage. I wonder why 64bit one is completely different, and it worth testing - maybe i386 code (only with saves to REX registers instead of stack) will actually work faster in 64bit too.

But twofish and camellia are higher on my TODO list. --
vda