Containers: css_put() dilemma
pj at sgi.com
Tue Jul 17 10:23:27 PDT 2007
> Because as soon as you do the atomic_dec_and_test() on css->refcnt and
> the refcnt hits zero, then theoretically someone other thread (that
> already holds container_mutex) could check that the refcount is zero
> and free the container structure.
Not just theory ... I've debugged crashes from live customer systems
that were basically this race.
I won't rest till it's the best ...
Programmer, Linux Scalability
Paul Jackson <pj at sgi.com> 1.925.600.0401
More information about the Containers