Re: [LTP] LTP failures

[David A. M. <dm...@re...>]

Dan, Craig:

Dan Kegel wrote:
> 
> Craig Thomas wrote:
> > I haven't quite figured out where to find the rationale for why certain
> > tests fail.  All I have been able to find are the lists of failures for a specific
> > distro/kernel/architecture/ltp version.  I would like to find out if these
> > failures are caused by the kernel or by a library (glibc version) or by
> > something else.
> 
> Let's take one failure you cite as an example.
> http://khack.osdl.org/stp/271579/results/fail_summary.txt
> says
> nanosleep02    1   FAIL : Remaining sleep time 3999392 usec doesn't match with the expected 3999949 usec time
> 
> Sounds like it's sleeping 0.5 milliseconds too little?
> (That's not just being picky -- it's ok to sleep too much,
> but not to sleep too little.)
> 

As this issue was explained to me, the kernel (RHL 2.4.18 in my case)
doesn't support better than 10000uS (10mS) precision (HZ is set to 100),
while this test expects 100uS precision:

	#define USEC_PRECISION 100 

Does this sound accurate for your environment as well?


d.marlin