Kent, Bcache will put write_dirty_finish into system_wq, and also journal_write if the journal is full. The write lock of the root node is hold with journal_write, and the lock won't be released until journal_write_done, which means at this moment all work queue kthreads which run write_dirty_finish will block since they are trying to require the read lock of root node.
And as we know the max number of active kthreads belong to system_wq is min(256, 4*num_possible_cpus), so if all kthreads in the poll have been blocked in write_dirty_finish, the journal_write queued up has no chance to run, resulting to all the kthreads sleep forever. Is this ever possible? Have you seen such deadlocks in your systems? Thanks, Zhu Yanhai -- To unsubscribe from this list: send the line "unsubscribe linux-bcache" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
