Hi:
This series of patches adds the backend support for asynchronous blkcipher
implementations. This includes the ablkcipher crypto type, generic crypto
request queueing, as well as a sample cryptd implementation that turns any
sync blkcipher algorithm into an async one by processing requests in a
dedicated kernel thread.
This allows drivers for hardware blkcipher devices to be written.
Here is a list of steps from this point onwards:
* Establish async users where appropriate (dm-crypt is done).
* Create async interface for hash/compress.
* Port existing crypto drivers across.
Any help is welcome, especially with respect to the last item since
I don't have any crypto devices here.
Cheers,
--
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
On Mon, Apr 16, 2007 at 08:51:05PM +1000, Herbert Xu ([email protected]) wrote:
> Hi:
>
> This series of patches adds the backend support for asynchronous blkcipher
> implementations. This includes the ablkcipher crypto type, generic crypto
> request queueing, as well as a sample cryptd implementation that turns any
> sync blkcipher algorithm into an async one by processing requests in a
> dedicated kernel thread.
>
> This allows drivers for hardware blkcipher devices to be written.
Hi Herbert.
Good to see async crypto processing is being (a bit slowly - more than a
year after we talked about it previously :) completed.
With this patchset included there will not be requirement for another async levels,
so I can port HIFN 795x driver from acrypto to this one.
Do you have a stable git tree to pull from, so I could prepare a testbed.
(althugh I can not start right now, but expect to do it soon).
And, hmm, your documentation really sucks, could you write a draft for
async api and put it into the tree?
Thanks for the work.
--
Evgeniy Polyakov
HI Evgeniy:
On Mon, Apr 16, 2007 at 03:28:42PM +0400, Evgeniy Polyakov wrote:
>
> Good to see async crypto processing is being (a bit slowly - more than a
> year after we talked about it previously :) completed.
Yeah I haven't really been able to put enough time into it unfortunately.
> With this patchset included there will not be requirement for another async levels,
> so I can port HIFN 795x driver from acrypto to this one.
Great!
> Do you have a stable git tree to pull from, so I could prepare a testbed.
> (althugh I can not start right now, but expect to do it soon).
These patches are now in cryptodev-2.6 and they will also show up in
mm soon.
> And, hmm, your documentation really sucks, could you write a draft for
> async api and put it into the tree?
Yes they suck so much that they don't exist :)
Once things settle down a bit I'll write something up. Of course
patches are welcome too.
For now the best place to start would be to look at the async crypto
users (tcrypt and dm-crypt) and driver (cryptd).
Cheers,
--
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