hifn_795x in a MacMini?

hifn_795x in a MacMini?

[Christian Kujau]

Hi Evgeniy,

while doing "make menuconfig" on a shiny new MacMini[0] I came across the 
option for the HIFN 795x crypto accelerator. I've set:

CONFIG_CRYPTO_DEV_HIFN_795X=m
CONFIG_CRYPTO_DEV_HIFN_795X_RNG=y

and after loading the module, dmesg shows:

   Driver for HIFN 795x crypto accelerator chip has been successfully registered.

Somehow this message sounds to me, as if this MacMini does have such a 
chip. This would be surprising, as I thought the chip is a highly 
specialized one, and expensive too. Also 
there's no ../devices/ symlink in /sys/bus/pci/drivers/hifn795x. Still, 
why would the driver load into the kernel when there's no such chip?

Thanks,
Christian.

[0] http://support.apple.com/specs/macmini/Mac_mini_Mid_2007.html
-- 
BOFH excuse #97:

Small animal kamikaze attack on power supplies

Re: hifn_795x in a MacMini?

[Kyle McMartin]

On Mon, Jul 14, 2008 at 09:18:09PM +0200, Christian Kujau wrote:
> and after loading the module, dmesg shows:
>
>   Driver for HIFN 795x crypto accelerator chip has been successfully registered.
>

The message is printing by the "module loading" function, which just
registers the pci driver, not the function that actually probes a
device. See hifn_init versus hifn_probe, in drivers/crypto/hifn_795x.c.

r, Kyle

Re: hifn_795x in a MacMini?

[Christian Kujau]

On Mon, 14 Jul 2008, Kyle McMartin wrote:
> The message is printing by the "module loading" function, which just
> registers the pci driver, not the function that actually probes a
> device. See hifn_init versus hifn_probe, in drivers/crypto/hifn_795x.c.

OK, understood now. Thanks for pointing that out!

Christian.
-- 
BOFH excuse #45:

virus attack, luser responsible

Re: hifn_795x in a MacMini?

[Evgeniy Polyakov]

Hi Christian.

On Mon, Jul 14, 2008 at 09:18:09PM +0200, Christian Kujau (lists@nerdbynature.de) wrote:
> Somehow this message sounds to me, as if this MacMini does have such a 
> chip. This would be surprising, as I thought the chip is a highly 
> specialized one, and expensive too. Also 
> there's no ../devices/ symlink in /sys/bus/pci/drivers/hifn795x. Still, 
> why would the driver load into the kernel when there's no such chip?

If you have appropriate hardware, then following string will be printed
in dmesg:

AES 128 ECB test has been successfully passed

-- 
	Evgeniy Polyakov