#8089 NORM 8.2.0 (: /dev/hwrng isn't random.

[Zarro Boogs per Child]

#8089: /dev/hwrng isn't random.
-----------------------------------+----------------------------------------
   Reporter:  cscott               |       Owner:  cscott                 
       Type:  defect               |      Status:  new                    
   Priority:  normal               |   Milestone:  8.2.0 (was Update.2)   
  Component:  ofw - open firmware  |     Version:  not specified          
 Resolution:                       |    Keywords:  blocks?:8.2.0 cjbfor8.2
Next_action:  qa signoff           |    Verified:  0                      
  Blockedby:                       |    Blocking:  8024                   
-----------------------------------+----------------------------------------

Comment(by cscott):

 My original theory was disproven; it turned out that I was just testing a
 machine which happened to have a 'lucky 0'.  Other machines had 'lucky
 24s' and 'lucky 55s'.  I wrote some code to generate a histogram of the
 results, and was appalled.  For 100,000 numbers, each bin should have had
 roughly 390 hits.  Instead:
 {{{
 # ./test3.py | sort | uniq -c | sort -n | tail -10
 1687 99
 1803 4a
 1871 29
 2051 95
 2348 92
 2405 a4
 2552 aa
 3097 2a
 3412 55
 4116 24
 # ./test3.py | sort | uniq -c | sort -n | tail -10
 1598 99
 1676 4a
 1771 29
 2086 92
 2306 95
 2532 aa
 2761 a4
 2992 2a
 3174 55
 4424 24
 #
 }}}

 Testing on a different machine (also running E17, but a different kernel)
 revealed a different set of 'lucky numbers', but the lucky numbers stayed
 consistently lucky in run after run.

 Jordan figured it out, though: stay tuned for the exciting conclusion!

-- 
Ticket URL: <http://dev.laptop.org/ticket/8089#comment:21>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system