On Wed, Sep 16, 2015 at 12:00:10AM +0200, John Kacur wrote:
> On Mon, 31 Aug 2015, Josh Cartwright wrote:
>
> > tsnorm() is not used at all in signaltest. Remove it.
> >
> > Signed-off-by: Josh Cartwright <joshc@xxxxxx>
> > ---
> > src/signaltest/signaltest.c | 8 --------
> > 1 file changed, 8 deletions(-)
> >
> > diff --git a/src/signaltest/signaltest.c b/src/signaltest/signaltest.c
> > index 9454a26..c6d1cfd 100644
> > --- a/src/signaltest/signaltest.c
> > +++ b/src/signaltest/signaltest.c
> > @@ -69,14 +69,6 @@ static int shutdown;
> > static int tracelimit = 0;
> > static int oldtrace = 0;
> >
> > -static inline void tsnorm(struct timespec *ts)
> > -{
> > - while (ts->tv_nsec >= NSEC_PER_SEC) {
> > - ts->tv_nsec -= NSEC_PER_SEC;
> > - ts->tv_sec++;
> > - }
> > -}
> > -
> > static inline long calcdiff(struct timespec t1, struct timespec t2)
> > {
> > long diff;
> > --
> > 2.5.0
> >
> > --
>
> As you can see, many of the programs in this suite are modeled after
> cyclictest, but many of them don't unfortunately receive nearly the
> amount of testing that cyclictest does. Now the fact that this function
> was copied from cyclictest, but not used, sends off alarm bells in my
> head. We have various struct timespec in signaltest, could the fact that
> we are not calling tsnorm mean that there are some hidden potential
> defects? Rather than removing this function, I'd like to spend some time
> auditing the use of timespec here until I'm convinced.
Yeah, I did the cheap-and-easy thing here; but you raise a good point,
we should audit signaltest to see where any timespec math is done to
ensure proper normalization.
Josh
Attachment:
signature.asc
Description: PGP signature
