On 09/22/2013 04:23 PM, Theodore Ts'o wrote: > On Sun, Sep 22, 2013 at 03:45:11PM -0700, H. Peter Anvin wrote: >> I understand the motivation, but I question basing it in a fixed amount of >> time. > > We already have a threshold based on the amount of the entropy in the > input pool. I could increase that threshold, but then instead of > having the entropy hover between say, 192 and 128, it would hover > between say, 576 and 512. What that means in practice is that there > will be a higher baseline, but we will still end up reseeding every 10 > seconds or so (that being approximately how much entropy I've seen > flowing into the input pool on my laptop system --- 64 bits every 10 > seconds or so). > > By basing it on a time-based threshold, it means that the input pool > can build up faster such that over time, such that entropy pool ends > up hovering around 3400 bits. >
So make it a threshold around 2048-3072 bits. A.k.a. "if the input pool is filling up fast enough, take advantage of it." > What is your concern is about having it being time-based --- or more > accurately, being partially time-based, since we also keep the entorpy > count based threshold? I'll note that the Fortuna Random Number > Generator, devised by Bruce Schneier and Niels Ferguson also uses as > part of its design a time-based reseed limit. Just that it is unnecessary in a scenario when entropy is plentiful. -hpa -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/