On Mon, Sep 25, 2017 at 05:54:59PM +0900, Byungchul Park wrote: > On Wed, Sep 20, 2017 at 07:00:20PM +0200, Peter Zijlstra wrote: > > With lockdep-crossrelease we get deadlock reports that span cpu-up and > > cpu-down chains. Such deadlocks cannot possibly happen because cpu-up > > and cpu-down are globally serialized. > > > > CPU0 CPU1 CPU2 > > cpuhp_up_callbacks: takedown_cpu: cpuhp_thread_fun: > > > > cpuhp_state > > irq_lock_sparse() > > irq_lock_sparse() > > wait_for_completion() > > cpuhp_state > > complete() > > > > Now that we have consistent AP state, we can trivially separate the > > AP-work class between up and down using st->bringup. > > Could you tell me what branch you worked the patches based on? > This is similar to the problem of workqueue so I want to fix it on > top of yours, as well.
I wrote the patches on top of tip/master. Thomas maintains these bits so hopefully he'll eventually merge them in the right tip tree.
