[Bug 137145] [mbuf] Reference count computing isn't correct when more than one threads call function m_copypacket

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=137145

--- Comment #12 from Andriy Gapon <avg@FreeBSD.org> ---
(In reply to Konstantin Belousov from comment #11)
Perhaps Tx that is keeping its reference.
I don't know what kind of code people can write to manipulate mbuf, this is
just a theoretical example.

Larger point is why have the special case, does it provide a lot of win for
performance?
I know that atomic operations are expensive. A branch is not free, however,
too.
Always doing the atomic update is straightforward and safe(r).

-- 
You are receiving this mail because:
You are the assignee for the bug.