On Tue, Oct 04, 2011 at 07:59:53AM +0100, Herbert Xu wrote:
> On Mon, Oct 03, 2011 at 05:55:10PM -0700, Darrick J. Wong wrote:
> >
> > So what I think I'm hearing is...
> >
> > 1. Apply Bob's slice-by-8 algorithm patch to regular crc32.
> > 2. Adapt crc32's build code to generate crc32c as well.
>
On Mon, Oct 03, 2011 at 05:55:10PM -0700, Darrick J. Wong wrote:
>
> So what I think I'm hearing is...
>
> 1. Apply Bob's slice-by-8 algorithm patch to regular crc32.
> 2. Adapt crc32's build code to generate crc32c as well.
> 3. Remove crypto/crc32c.c's implementation and have it wrap the code ge
On Mon, Oct 03, 2011 at 09:35:13PM +0100, Herbert Xu wrote:
> On Mon, Oct 03, 2011 at 10:27:03PM +0200, Joakim Tjernlund wrote:
> >
> > > > Start from Bobs latest patches and add crc32c to lib/crc32.c
> > >
> > > If I did that, how should I handle patching in the hardware accelerated
> > > version
On Mon, Oct 03, 2011 at 10:27:03PM +0200, Joakim Tjernlund wrote:
>
> > > Start from Bobs latest patches and add crc32c to lib/crc32.c
> >
> > If I did that, how should I handle patching in the hardware accelerated
> > version
> > on Intel systems? That switcheroo ability seems to have been Herbe
"Darrick J. Wong" wrote on 2011/10/03 17:36:46:
>
> On Sat, Oct 01, 2011 at 04:02:10PM +0200, Joakim Tjernlund wrote:
> >
> > "Darrick J. Wong" wrote on 2011/09/30 21:29:56:
[SNIP]
> >
> > This is based on an old version of Bobs slice by 8 that has lots
> > duplication and
> > hard to maintain
On Sat, Oct 01, 2011 at 04:02:10PM +0200, Joakim Tjernlund wrote:
>
> "Darrick J. Wong" wrote on 2011/09/30 21:29:56:
> >
> > The existing CRC32c implementation uses Sarwate's algorithm to calculate the
> > code one byte at a time. Using a slicing-by-8 algorithm adapted from Bob
> > Pearson, we
"Darrick J. Wong" wrote on 2011/09/30 21:29:56:
>
> The existing CRC32c implementation uses Sarwate's algorithm to calculate the
> code one byte at a time. Using a slicing-by-8 algorithm adapted from Bob
> Pearson, we can process buffers 8 bytes at a time, for a substantial increase
> in perform