Re: [PATCH v3 1/6] crypto: SHA1 multibuffer crypto hash infrastructure

[Tim Chen <tim.c.chen@linux.intel.com>]

On Sat, 2014-06-21 at 01:57 +0800, Herbert Xu wrote:
> On Wed, Jun 11, 2014 at 02:03:45PM -0700, Tim Chen wrote:
> >
> > diff --git a/crypto/shash.c b/crypto/shash.c
> > index 929058a..234ece0 100644
> > --- a/crypto/shash.c
> > +++ b/crypto/shash.c
> > @@ -271,6 +271,18 @@ static int shash_async_finup(struct ahash_request *req)
> >  	return shash_ahash_finup(req, desc);
> >  }
> >  
> > +int shash_ahash_mcryptd_digest(struct ahash_request *req,
> > +			       struct shash_desc *desc)
> > +{
> > +	int err;
> > +
> > +	err = crypto_shash_init(desc) ?:
> > +	      shash_ahash_mcryptd_finup(req, desc);
> > +
> > +	return err;
> > +}
> > +EXPORT_SYMBOL_GPL(shash_ahash_mcryptd_digest);
> > +
> 
> This bit causes your patch to not build for me.
> 
> Please fix and resubmit.

Herbert,

Upon further testing, I realize that I have to change my logic
for doing work in the idle_notifier path to clear out partial tasks.
The notifier is actually in an atomic context and I may page fault
when working on the data buffer but I cannot sleep here.

I'll need to wake up the kernel crypto thread instead.  So it
will take me a bit longer to refine this logic.

Thanks.

Tim