Re: CVS commit: src/sys/kern

["Mindaugas R." <rmind%NetBSD.org@localhost>]

Elad Efrat <elad%bsd.org.il@localhost> wrote:

[... skip ...]
> the only way fileassoc_table_lookup() can fail in
> fileassoc_table_delete() is if there's a locking violation. since
> fileassoc lacks locking altogether, your introduced workaround isn't a
> real fix, but rather a way to silence a coverity "bug".
Hence, you did not prove me, that this change is somehow wrong. As you have
shown, fail is simply not possible because of such circumstances (which is
because of more abstract problem!) - but it DOES NOT MEAN, that the chain of
NULLs is wrong. It may simply be unnecessary.
Actually, leaving such parts may cause problems in future, when, for example,
that will be able to really fail in this case. Then someone should inspect and
handle such situations in code (and doesn't look that it would be you, even
this is yours code). I would probably tend to disagree that looking in
abstract, such behaviour is correct at all.

> and, still, you did not tell me who okay'd this commit. <...>
So as you are so experienced and enlightened, you may probably know, that it
is not possible to every time get an OK from others. Of course, we could
create tech-approves@, where you would be an officer with 24hours/7days per
week support. Would you agree with this? :)

-- 
Best regards,
Mindaugas
www.NetBSD.org