directory change notification patch

[Jeremy Allison]

On Sun, Mar 27, 2005 at 10:19:29PM -0500, Derrell.Lipman at UnwiredUniverse.com wrote:
> Jeremy Allison <jra at samba.org> writes:
> 
> > Sometime having your source code out there *sucks* :-) :-).  Especially when
> > it has subtle bugs (and I'm saying they're subtle 'cos this stuff has been
> > out there forever and no-one else noticed :-).
> 
> Really good engineers tend to be both highly opinionated and very defensive of
> their code, because in the vast majority of cases, their opinion ends up being
> the right way to go, and their code ends up being correct. :-)

Yeah, but not always. Especiall in this case We did work out whose original bug it was
but I'm not going to say who :-).

> Now the question is, do you agree with the priority that Mark gave in his
> patch?  Do interrupts deserve higher priority than data, as he indicated?  I
> believe they do.  Is his patch the best way to implement the fix?  I have not
> evaluated Mark's patch in detail to see specifically what he did; rather I
> verified that his explanation of the problem matched the existing code.  I
> take it you'll probably want to review (or already have reviewed) Mark's patch
> now...

I've reviewed it *very* crefully, and wrote sume test code to reproduce
the races. His fix (in sys_select) is correct, the fix in sys_select_intr
is a little overkill (IMHO) so I've trimmed down the paranoia a little.

Yes, he (and you) are correct in the EINTR must have a higher priority
than incoming data, otherwise an active client could starve asynchronous
processing.

BTW Derrell, thanks for explaining the problem in Very Simple Words of
one syllable so I got it :-). Sometimes I am an Engineer of Very Little
Brain :-).

Cheers,

	Jeremy.