On Wednesday, August 1, 2018 7:01:42 PM CEST you wrote: > On Wednesday, August 1, 2018 6:10:56 PM CEST you wrote: > > Looking at: [1](https://wiki.archlinux.org/index.php/Rng-tools), and then > > following to: [2](https://en.wikipedia.org/wiki/ > > Hardware_random_number_generator), one can think that rng-tools uses: [3] > > (https://en.wikipedia.org/wiki/RdRand). > > > > But it might not be so, since [2] says: "The RdRand opcode will return > > values from an onboard hardware random number generator. It is present in > > Intel Ivy Bridge processors and AMD64 processors since 2015", and > > IvyBridge > > started in 2012, and rng-tools code is very, very old (but there is some > > overlapping time, possibly). > > > > Therefore believing that rng-tools uses RdRand's CPU hardware random > > generator instruction, might be pure misconception. > > > > Consequently, haveged seeming much newer and maintained, it is in my > > opinion the best choice. > > It seems the overlapping time was sufficient to "Add RDRAND support" to > rng-tools: (debian's rng-tools5) > > [4](https://git.kernel.org/pub/scm/utils/kernel/rng-tools/rng-tools.git/log/ > ?qt=grep&q=rdrand) > > "Add RDRAND support" (2012-07-31) > [5](https://git.kernel.org/pub/scm/utils/kernel/rng-tools/rng-tools.git/comm > it/?id=2a01453686f065ac95916fe49a02407914e05dd4) > > Next, is the redhat version: > [6](https://github.com/nhorman/rng-tools/search?q=rdrand&unscoped_q=rdrand) > > debian's "rng-tools" (unofficial), seems to be to old, viz. anterior to > 2012's IvyBridge, to use RdRand instruction: > [7](https://tracker.debian.org/pkg/rng-tools) > (last news, as of today, is from 2011-11-10) >
My CPU is i5-3210m, which seems to be Ivy Bridge, and therefore it should implement the RdRand instruction... rng-tools started before the introduction of that instruction; So even though support for that instruction seems to have been introduced in 2012, we should not take for granted that the hardware generated random number from rng-tools, necessarily comes from that instruction. https://github.com/nhorman/rng-tools: "The rng-tools official repository (formerly part of the gkernel project on sourceforge)"... So what is this version: https://git.kernel.org/pub/scm/utils/kernel/rng-tools/rng-tools.git/ Which is the one used by debian, and pointed to by archlinux wiki, https:// wiki.archlinux.org/index.php/Rng-tools, and which seems to be several years older than the nhorman/rng-tools's version? Anyway, to be thorough, i removed haveged, and install rng-tools5 package, and rebooted: The reboot took the same time, to the second, as with haveged (counting in my head). (To be really thorough, I should remove also rng-tools5 and reboot, maybe the sddm start delay has just vanished into thin air during the while; But that I won't do now.) So, I'll stick to it, because, well, why not. Note: if it is for but one cpu's instruction, maybe that code could be integrated in the kernel? (I've not found if IBM z13 processor integrate RdRand instruction).
