new time.c code: a caveat

C. Scott Ananian (cananian@lesser-magoo.lcs.mit.edu)
Sun, 12 Jul 1998 09:46:13 -0400 (EDT)

Note that *checking* the new time.c code for bugs is actually extremely
difficult: not that many things use sub-second time accuracy, and those
that do are often difficult to monitor. In many cases monotony is
sufficient to keep any problems from showing up. The old time.c code
would warp ahead to the next jiffy on machines with slowed TSCs, yielding
*no* *sub-jiffy* *accuracy* *at* *all* and to my knowledge no one ever
noticed this until things got so bad that a kernel divide-by-zero error
finally resulted (and even then, the kernel trapped and logged the
exception and continued merrily on).

So, before/as code comes in to revamp time.c, the big question ought to be
testability. It is not trivial to make some sort of primitive test suite,
but it ought to be done anyway. How else do we know the proposed codes
work *at all*?
--Scott
[as before, please make sure 'time.c' is somewhere in the subject of your
reponse if you'd like me to be able to see it]
@ @
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-oOO-(_)-OOo-=-=-=-=-=
C. Scott Ananian: cananian@lcs.mit.edu / Declare the Truth boldly and
Laboratory for Computer Science/Crypto / without hindrance.
Massachusetts Institute of Technology /META-PARRESIAS AKOLUTOS:Acts 28:31
-.-. .-.. .. ..-. ..-. --- .-. -.. ... -.-. --- - - .- -. .- -. .. .- -.
PGP key available via finger and from http://www.pdos.lcs.mit.edu/~cananian

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html