On Tue, 19 Sept 2023 at 14:06, Roberto A. Foglietta <[email protected]> wrote: > > On Tue, 19 Sept 2023 at 13:30, Rob Landley <[email protected]> wrote:
> As the first init instruction or as the last kernel boot operation, is > THE general answer also when it is not the solution. Some systems need > a more dedicated effort but in no way THE answer puts them in a worse > condition than they were before. Confutations are welcome. :-) Yes, it hurts because on some specific systems the /dev/urandom is filled with a data stream which is constant for each boot or not good enough for seeding it in a secure way. This means that any later effort wil not have an immediate effect because of that initial seeding. Fine, then use > (write) to reset the entropy and >> (append) to add entropy. This will break the back-compatibility with the past and mess-up all the scripts or tools which are in the production nowadays. Fine, then use > (write) to add entropy as usual and >> (append) to reset entropy. Unfortunately this approach leaves behind some specific systems, when they will update the kernel. At that future time, they will choose to deselect the default option to seed the /dev/random OR use >> to overcome the issue. Hoping that those systems have a better way to seed the /dev/urandom engine which can not be the case at all because a low-frequency 1-single core NOMMU system, possibly with "none" as I/O and task scheduler can have a very small quantity of entropy and no means to amplify it. We may agree that those systems - because of their intrinsic limitations - are not sophisticated enough for being exposed in an untrustworthy environment like the Internet or they should integrate/connect a I2C/USB cheap thermal sensor that can provide true white noise, for example. With 1% of the effort, 99% of the work is done. The rest should be dealt with by experts which will select the more appropriate set of features and options case by case. Trying to deal with all the cases including that 1% lead to an impasse. This is wrong because a general solution should not necessarily be a mathematical truth but a the main way for almost all the cases which includes a reasonable exception handling that allows to deal properly also with the un-usual cases. This is a general solution. A mathematical truth is often useless or counterproductive in the real-world. Math says it is wrong in the most general case while practice says it works in almost all cases. Then add an exception handling and go with the practice. ;-) I hope this helps, R- _______________________________________________ busybox mailing list [email protected] http://lists.busybox.net/mailman/listinfo/busybox
