Re: [Qemu-devel] [PATCH 1/3] qga: add support to get host time

[Lei Li]

On 01/08/2013 05:52 AM, Eric Blake wrote:
> On 01/06/2013 03:06 AM, Lei Li wrote:
> > Signed-off-by: Lei Li <address@hidden>
> > ---
> >   qga/commands-posix.c |   18 ++++++++++++++++++
> >   qga/qapi-schema.json |   17 +++++++++++++++++
> >   2 files changed, 35 insertions(+), 0 deletions(-)
> >
> > diff --git a/qga/commands-posix.c b/qga/commands-posix.c
> > index a657201..26b0fa0 100644
> > --- a/qga/commands-posix.c
> > +++ b/qga/commands-posix.c
> > @@ -91,6 +91,24 @@ exit_err:
> >       error_set(err, QERR_UNDEFINED_ERROR);
> >   }
> >   
> > +static HostTimeInfo *get_host_time(void)
> > +{
> > +    host_time = g_malloc0(sizeof(HostTimeInfo));
> > +    host_time->seconds = tq.tv_sec;
> > +    host_time->microseconds = tq.tv_usec;
> Why usec?  struct timespec with nanoseconds might be a nicer unit, even
> if for the initial implementation, you use
> qemu_gettimeofday().tv_usec*1000 rather than dragging in a realtime
> library for full ns resolution.  If nothing else, the lesson that ought
> to be learned from the proliferation of time types is that any time you
> don't report lots of precision, someone comes along later on having to
> add yet another interface adding more precision.

ok, I will have a try to change it to a ns resolution.

> > +++ b/qga/qapi-schema.json
> > @@ -83,6 +83,23 @@
> >   { 'command': 'guest-ping' }
> >   
> >   ##
> > +# @HostTimeInfo
> > +#
> > +# Information about host time.
> > +#
> > +# @seconds: "seconds" time from the host.
> Document that this is relative to the Epoch of 1970-01-01 (no matter
> what the host uses for its internal reference point).

Sure.

> > +#
> > +# @microseconds: "microseconds" time from the host.
> Again, nanoseconds (struct timespec) might be nicer.
>
> > +#
> > +# @utc-offset: information about utc offset.
> In what format?  Minutes away from UTC, a 4-digit decimal value, or
> something else (that is, is a one-hour offset represented as 60 or 100)?
>   Are negative values east or west of UTC?

For this version, it's a one-hour offset represented as:±[hh].
Negative values are west, andpositive values are east of UTC.

> > +#
> > +# Since: 1.4
> > +##
> > +{ 'type': 'HostTimeInfo',
> > +  'data': { 'seconds': 'int', 'microseconds': 'int',
> > +             'utc-offset': 'int' } }
> Indentation seems inconsistent.
>
> Ah, here you made them mandatory - only your cover letter implied that
> they were optional.

Sorry for the inconsistent indentation.
No, this is just the definition of the time structure. The optional arguments
mentioned in the cover letter are for command "guest-set-time". :)



--
Lei