2011-12-01 17:01:18

by Seth Jennings

[permalink] [raw]
Subject: Per-cpu tfm allocation/free functions


I am currently working on the zcache RAM compression driver (in the staging
tree) and have a question.

I am looking to use the crypto compression API instead of the hardcoded LZO
calls that are in there right now, but I need to allocate a tfm per cpu per
compression algorithm being used.

I've already written the code but it looks very similar to what is already in
ipcomp_alloc/free_tfms() in net/xfrm/xfrm_ipcomp.c.

Has there ever been any talk about implementing functions like that inside
the cryptographic API in order to prevent duplicate per-cpu code by the users?

Would there be any interest/objections regarding a patch that did implement them?