[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

RE: SI-ia64: LSB.os exec tests

> > The numbers are always the same - i.e., over two
> > runs I saw this same set of three "incorrect" values
> > 12 times with no variation.
> > 
> > What should I read into this?
> I think that the test is wrong, and will fail on any Linux machine
> with HZ != 100.  With a higher click we have more play in the
> numbers.  The itimer spec is a little vague on the exact semantics,
> though.

Thanks for the pointer. What I read in the sus under
setitimer/getitimer suggests that if the implementation
can't precisely represent, e.g. 10,000 microseconds, it
can round it up.  With a 100 Hz clock, it works out that
it can precisely represent it, but not with a 1024 hz
clock like the Itanium uses.  Thus I think I agree the
test questionable, it should be allowing for up to a 
one-tick overage.

In my error cases, I'm over by 736, 496 and 256 microseconds.
One tick at 1024/sec is 977 microseconds, so I guess I
think these tests pass :-)

But I await an opinion from someone who understands
this stuff better...


To UNSUBSCRIBE, email to lsb-test-request@lists.linuxbase.org
with subject of "unsubscribe". Trouble? Email listmaster@lists.linuxbase.org

Reply to: