Re: lustre: why does cfs_get_random_bytes() exist?

["Dilger, Andreas" <andreas.dilger@intel.com>]

On 2013/10/03 5:45 PM, "Theodore Ts'o" <tytso@mit.edu> wrote:
>On Thu, Oct 03, 2013 at 11:06:58PM +0000, Dilger, Andreas wrote:
>> 
>> The Lustre cfs_get_random_bytes() incorporates (via cfs_rand()) a seed
>> which also hashes in the addresses from any network interfaces that are
>> configured.
>> Conversely, cfs_rand() also is seeded at startup from
>>get_random_bytes() in
>> case a hardware RNG is available.  This ensures even with identical
>>initial
>> conditions cfs_get_random_bytes() gets a different random stream on each
>> node.
>
>With modern kernels, the /dev/random driver has the
>add_device_randomness() interface which is used to mix in
>personalization information, which includes the network MAC address.
>So that particular concern should be covered without the hack of
>mixing in cfs_rand().

I think that depends on the network driver.  The Cray systems have some
very strange networking hardware that is beyond our control - definitely
not ethernet or Infiniband.

I'll have to ask the Cray folks if their network drivers do this today.

>> I'm not against cleaning this up, if there is some mechanism for the
>> startup code to add in the node interface addresses into the entropy
>> pool, and this is also used to perturb the prandom_u32() sequence
>> after that point.
>
>That's handled too, via the late initcall prandom_reseed().
>
>Cheers,
>
>					- Ted
>


Cheers, Andreas
-- 
Andreas Dilger

Lustre Software Architect
Intel High Performance Data Division