On Sat, Jul 29, 2017 at 10:43:55PM +0200, Ingo Schwarze wrote: > > All the same, i'd like to point out that there is a minor documentation > issue. Some time ago, we decided that having duplicate descriptions > of each and every sysctl variable in sysctl(3) and sysctl(8) is bad > for maintainability, deleted them all from sysctl(8), and made sure > users of sysctl(8) can understand the full listing in sysctl(3) by > adding the full string-names used by sysctl(8) to the sysctl(3) > manual page. > > So after your patch, we would have to document in sysctl(3) that > CTL_MACHDEP.CPU_LIDACTION = machdep.lidaction takes int on the C > interface level and strings on the command line, and that > 0=none, 1=suspend, 2=hibernate. >
that would be doable, but... > While i don't know whether documenting *all* machdep variables in > sysctl(3) would make sense, machdep.lidaction certainly seems > important enough to document it. Same for machdep.kbdreset. > ... at present sysctl(3) does not describe any machdep syscts at all. it never has (or we removed them long ago and i forget). the theory was that the machdep ctls were a moving target and, by definition, only relevant to specific platforms. so sysctl(3) didn;t even attempt to describe them, sysctl(8) listed some of them, and sysctl.conf described some of them. maybe there's a bigger discussion possible, but for now it would be simple, and not inconsistent, to just keep that in the example /etc file. no immediate doc change would be needed. jmc > Yours, > Ingo >
