On Tue, Mar 31, 2020 at 09:42:46AM +0200, Vlastimil Babka wrote: > On 3/31/20 12:44 AM, Luis Chamberlain wrote: > > > > This is wonderful when we think about existing sysctls which have > > corresponding silly boot params that do the same thing. However, shoving > > a boot param capability down every possible built-in sysctl brings > > forward support considerations we should take serious, as this would > > add a new user interface and we'll have to support it. > > Hmm, if I boot with an initramfs with init process that does mount /proc and set > some sysctl there as the very first thing, then this will be almost the same > moment as my patch does it. There is no further kernel initialization in > between. So with your logic we already do support all non-modular sysctls to be > set so early. Yes, true. Then by all means: Reviewed-by: Luis Chamberlain <mcgrof@xxxxxxxxxx> Luis
