https://bugs.dragonflybsd.org/https://bugs.dragonflybsd.org/favicon.ico?16293952082008-01-31T02:54:00ZDragonFlyBSD bugtrackerDragonFlyBSD - Bug #935: Test on IBAAhttps://bugs.dragonflybsd.org/issues/935?journal_id=42522008-01-31T02:54:00Zdillon
<ul></ul><p>:Hi Matt,<br />:...<br />:<br />:What I would recommend is that after initialisation<br />:(after the generator has been seeded from the<br />:"entropy" file - the seeding method that you added),<br />:the generator is warmed up as described in:<br />:<br />:http://www.burtleburtle.net/bob/rand/isaac.html<br />:<br />:I would recommend that the generator is warmed up by<br />:making 10 (ten) calls to IBAA_Call () (as described in<br />:the above URL). This has the effect of discarding the<br />:output of the generator for the first ten<br />:rounds/calls, and thereby eliminating the<br />:internal-state leakage produced by the first<br />:round/rounds.<br />:<br />:Sincerely,<br />:R Carey.</p>
<pre><code>Ok, so looking at our kernel implementation:</code></pre>
<pre><code>/usr/src/sys/kern/kern_nrandom.c</code></pre>
<pre><code>You are suggesting that I make 10 calls to IBAA_Call() at the end<br /> of the rand_initialize() function?</code></pre>
<pre><code>Should I also make 10 calls to IBAA_Call() at the end of the<br /> add_buffer_randomness() function? That function is called when<br /> userland (root) writes to /dev/random during early boot.</code></pre>
<pre><code>-Matt<br /> Matthew Dillon <br /> &lt;<a class="email" href="mailto:dillon@backplane.com">dillon@backplane.com</a>&gt;</code></pre> DragonFlyBSD - Bug #935: Test on IBAAhttps://bugs.dragonflybsd.org/issues/935?journal_id=42542008-01-31T03:34:00Zdillon
<ul></ul><p>I have committed both adjustments to head. The two warmups<br /> and the adjustment to L15's 'y' index.</p>
<pre><code>-Matt</code></pre>