* Paolo Bonzini ([email protected]) wrote: > The first two patches are bugfixes meant for application. They allow > using threads that are readers and writers with implementations other > than urcu-qsbr. In addition, they allow early unregistering of > reader threads without risking deadlock with writers. > > The third patch is a sample implementation of lock-free queue following > discussion with Paul last week. > > They do not depend on the previous patch series. > > Paolo Bonzini (3): > add rcu_thread_{on,off}line to urcu-bp and urcu > add rcu_thread_offline before rcu_unregister_thread
It is illegal to call rcu_unregister_thread from within a RCU read-side critical section. So I don't see the point in adding support for explicit thread offlining. Maybe it's a documentation issue that lead you to assume that unregistering a thread from a read-side C.S. is valid ? Thanks, Mathieu > lock-free queue with RCU-based garbage collection > > README | 10 +- > tests/Makefile.am | 36 +++- > tests/test_qsbr_lfq.c | 587 > +++++++++++++++++++++++++++++++++++++++++++++++++ > tests/test_urcu_lfq.c | 585 ++++++++++++++++++++++++++++++++++++++++++++++++ > urcu-bp-static.h | 17 ++ > urcu-bp.c | 26 +++ > urcu-bp.h | 4 + > urcu-static.h | 15 ++ > urcu.c | 28 +++ > urcu.h | 4 + > 10 files changed, 1307 insertions(+), 5 deletions(-) > create mode 100644 tests/test_qsbr_lfq.c > create mode 100644 tests/test_urcu_lfq.c > > > _______________________________________________ > ltt-dev mailing list > [email protected] > http://lists.casi.polymtl.ca/cgi-bin/mailman/listinfo/ltt-dev > -- Mathieu Desnoyers Operating System Efficiency Consultant EfficiOS Inc. http://www.efficios.com _______________________________________________ ltt-dev mailing list [email protected] http://lists.casi.polymtl.ca/cgi-bin/mailman/listinfo/ltt-dev
