<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0//EN" "http://www.w3.org/TR/REC-html40/strict.dtd"> <html><head><meta name="qrichtext" content="1" /><style type="text/css"> p, li { white-space: pre-wrap; } </style></head><body style=" font-family:'monospace'; font-size:9pt; font-weight:400; font-style:normal;"> On Wednesday, January 27, 2016 01:57:10 PM Joachim Strömbergson wrote: > Aloha! > > Fredrik Thulin wrote: > > My point is that we will have to deal with warmup/bootup tests > > regardless of if we do 1 or N bootups of the entropy source for every > > 1 bootup of the HSM. > > No, N bootups will be much, much more complicated. > > Warmup once during cold boot of the board is one thing. A lot of things > will take time, initiialize etc. Doing warmup delay, testing multiple > times while the board is quite probably running in operations mode open > up for starvation in SW, possible timeouts etc. > > If the entropy source power control and thus warmup is under FPGA > control we can do this only during full reseed. But unless the reseed > gets a headsup, we either power up when it in fact wasn't needed, or we > risk adding the warmup delay right into the reseed time. > > One point of having the TRNG chain structured with separated mixer and > FIFOs is that we can do reseed while delivering random data to > applications. This will risk breaking that assumption. > > As I said. I can live with having the power down function on the board, > but possible to disable (i.e. it will be from the start.). But I think > it opens up a huge can of worms and will require quite a lot of thinking > and FPGA design to eliminate possible problems. This is turning into something we need to talk about next time we meet face to face =). I think we have had different levels of ambition in mind here. You talk about continuous delivery of random numbers and non-blocking reseeds. I was thinking that a reseed will likely always be a potentially blocking operation because you need to assess the entropy from the sources and, if it is not deemed good enough you need to wait for more or halt the reseed operation alltogether, or something. Anyway, as youv'e said - the jumper allows us to continue with the hardware design and leave the interesting discussion about how this is all going to work for later =). Please don't think I'm dismissing the subject, I'm really looking forward to talking more about it. /Fredrik </body></html>