[U-Boot] [PATCH 1/4] Exynos: Add hardware accelerated SHA 256

[Kim Phillips]

On Fri, 1 Mar 2013 16:11:36 +0000
Akshay Saraswat <akshay.s at samsung.com> wrote:

> Samsung Enterprise Portal mySingle
> 
>  
> 
> Hi Kim,
> 
>  
> 
> >On Thu, 28 Feb 2013 11:08:21 +0000
> 
> >Akshay Saraswat <akshay.s at samsung.com> wrote:
> 
> >
> 
> >> >On Wed, 27 Feb 2013 10:24:39 -0500
> 
> >>
> 
> >> >Akshay Saraswat <akshay.s at samsung.com> wrote:
> 
> >
> 
> >can you fix your mailer to not double space lines?
> 
>  
> 
> Sorry for the disarranged text. I hope this time it is good.
> 
>  

As you can see, no.  It looks like you didn't change anything.  I
have to manually single-space things below.

> >> So, I have incremented this value
> >> to 500 which shall be good enough for the lowest of all frequencies. But I guess what we
> >> need here is some formula, to calculate timeout on the basis of frequency, which is nowhere
> >> defined.
> 
> >That's odd - I see a bunch of frequencies advertised in
> >arch/arm/cpu/armv7/exynos/clock.c.
> >Or how about using a cycle counter instead?
> >btw, where can I find documentation for the ACE?
> 
> I tried with different frequencies and data sizes and found that it takes 2 to 3 msecs in all the cases.
> Since, it is hardware accelerated encoding it is supposed to behave in this same manner.
> But the current version of u-boot behaves mysteriously and slow downs when pressed enter for long.
> In the above case mentioned I even saw time took to be more than 100 msecs for all the cases.
> I think it would not be a better idea to timeout encoding acceleration.
> But still if it is needed, cycle counter suggestion would be a great solution.

it depends on whether the h/w can fault during its operation.

> You can read Exynos5250 manual's "Security subsystem" section, which I referred for this.

you mean this?:

http://www.samsung.com/global/business/semiconductor/file/product/Exynos_5_Dual_User_Manaul_Public_REV100-0.pdf

There is no "Security subsystem" section.

Kim