On Tue, 18 Apr 2017 17:25:25 +0100 Zefram <zef...@fysh.org> wrote: > Roger Shimizu wrote: > >You tagged as minor, so I think it's just a display issue of raw time, > >and it doesn't affect adjtimex's time calculation, right? > > I didn't look at the calculating parts of adjtimex, which I never use. > I use adjtimex mainly for display, and just occasionally to poke > status flags. I tagged this bug as minor because I was only reporting > a display problem.
Understand. > I've now audited adjtimex fully for nanosecond issues. In addition to > the display problem, STA_NANO affects the interpretation of offsets input > with the -o and -s options; this is a documentation issue rather than a > functionality issue. The parts of adjtimex that perform computations are > not affected, because they don't use the adjtimex syscall to retrieve > the time: they use gettimeofday, which yields microseconds regardless > of STA_NANO. Thanks for your detailed explanation! Now I realized that you're using jessie (stable), and the part of the issue is already resolved in stretch (testing). See: https://github.com/rogers0/adjtimex/blob/master/debian/patches/07-update-manpage-adjtimex.8.patch And yes, only manpage is upated for the above patch. The adjtimex command result is still wrong about ns/us. > >And it'd be appreciated if you can kindly provide a patch. > > Attached. Thanks for your patch! I guess "usec or nsec" is still confusing to end-user. Better to add "refer to the status code" or something like that. > Would you also be interested in patches to make the interface a bit nicer? > I'm thinking of things like showing units on the values displayed with -p > (the unit for "offset" depends on STA_NANO), using symbolic names for > the status flags, and permitting input option values to use units (like > "-o +200us", whose meaning would be unaffected by STA_NANO). Yes, patch is always welcome! But please add new lines for new stuff. I wonder whether other program may read the result of adjtimex and do parsing stuff. Adding new lines can make those changes at minimum. Cheers, -- Roger Shimizu, GMT +9 Tokyo PGP/GPG: 4096R/6C6ACD6417B3ACB1
pgpNCEqTUoYzJ.pgp
Description: PGP signature
