Unknown licensing terms for binary blobs [Was: Re: [PATCH 1/2] firmware: Add cavium CPT hardware crypto firmware]

[Tim Harvey]

On Sat, Feb 19, 2022 at 11:22 PM Daniel Danzberger <daniel at dd-wrt.com> wrote:
>
> On Sat, 2022-02-19 at 20:21 +0100, Petr Štetiar wrote:
> > Hauke Mehrtens <hauke at hauke-m.de> [2021-05-22 20:31:27]:
> >
> > Hi Daniel, Tim,
> >
> > > On 4/18/21 9:46 AM, Daniel Danzberger wrote:
> > > > The firmware consists of 2 images:
> > > >   - cpt8x-mc-ae.out
> > > >   - cpt8x-mc-se.out
> > > >
> > > > Both are required and requests by the kernel module 'cptpf'
> > > > (drivers/crypto/cavium/cpt) to provide hardware crpyto support
> > > > on cavium platforms like the octeontx
> > > >
> > > > Signed-off-by: Daniel Danzberger <daniel at dd-wrt.com>
> > > > ---
> > > >   package/firmware/cavium-cpt/Makefile | 52
> > > > ++++++++++++++++++++++++++++
> > > >   1 file changed, 52 insertions(+)
> > > >   create mode 100644 package/firmware/cavium-cpt/Makefile
> > > Hi,
> > >
> > > Under which license are these two files distributed?
> >
> > do you've have an idea? Thanks!
> Hi,
>
> I couldn't find any information about the license, but I played around
> with this hardware crypto on some devices a while ago and found that
> they are performing terrible :)
> Much worse than the CPU. Gateworks even notices this on their website:
> http://trac.gateworks.com/wiki/linux/encryption#PerformaceComparisons
> http://trac.gateworks.com/wiki/newport/encryption
>
> With that in mind I think my 2 original patches can be dropped.

Correct, the performance is not good and the only benefit of using it
is if you need to offload crypto because your CPU is constrained (and
don't care about the crypto performance).

Tim