[PATCH v3 3/3] Add test for %clock trace class

[Dmitry V. Levin]

On Sat, Apr 18, 2020 at 06:32:37PM +0300, Dmitry V. Levin wrote:
> On Sat, Apr 18, 2020 at 05:12:18PM +0200, Ákos Uzonyi wrote:
> > On Sat, 18 Apr 2020 at 00:26, Dmitry V. Levin wrote:
> > > On Fri, Apr 17, 2020 at 11:32:50PM +0200, Ákos Uzonyi wrote:
> > > > +trace_clock  test_trace_expr 'clock_nanosleep|times' -e%clock
> > >
> > > Apparently, this test fails on some platforms.
> > > I see it failing on ppc and mips, but other platforms also could be
> > > affected.  On these platforms, mkstemp implementation invokes
> > > gettimeofday, this results to fcntl and fcntl64 executables being filtered
> > > by -e%clock.
> > 
> > Now I have executed a Travis CI build, it fails on x86_64 also for the
> > same reason (i think it passed on my system because gettimeofday is
> > invoked through vDSO).
> > 
> > Adding "fcntl(64)?" to exclusion regexp makes the travis build pass.
> > Should I send a v4 patch?
> 
> I've found a way to tweak fcntl tests a bit so they won't use mkstemp,
> so thank you but no change in the test is needed so far.

And the whole series is now merged, thanks!


-- 
ldv