Re: [tip: locking/core] lockdep: Fix lockdep recursion

[Qian Cai]

On Fri, 2020-10-09 at 06:58 -0700, Paul E. McKenney wrote:
> On Fri, Oct 09, 2020 at 09:41:24AM -0400, Qian Cai wrote:
> > On Fri, 2020-10-09 at 07:58 +0000, tip-bot2 for Peter Zijlstra wrote:
> > > The following commit has been merged into the locking/core branch of tip:
> > > 
> > > Commit-ID:     4d004099a668c41522242aa146a38cc4eb59cb1e
> > > Gitweb:        
> > > https://git.kernel.org/tip/4d004099a668c41522242aa146a38cc4eb59cb1e
> > > Author:        Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> > > AuthorDate:    Fri, 02 Oct 2020 11:04:21 +02:00
> > > Committer:     Ingo Molnar <mingo@xxxxxxxxxx>
> > > CommitterDate: Fri, 09 Oct 2020 08:53:30 +02:00
> > > 
> > > lockdep: Fix lockdep recursion
> > > 
> > > Steve reported that lockdep_assert*irq*(), when nested inside lockdep
> > > itself, will trigger a false-positive.
> > > 
> > > One example is the stack-trace code, as called from inside lockdep,
> > > triggering tracing, which in turn calls RCU, which then uses
> > > lockdep_assert_irqs_disabled().
> > > 
> > > Fixes: a21ee6055c30 ("lockdep: Change hardirq{s_enabled,_context} to per-
> > > cpu
> > > variables")
> > > Reported-by: Steven Rostedt <rostedt@xxxxxxxxxxx>
> > > Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
> > > Signed-off-by: Ingo Molnar <mingo@xxxxxxxxxx>
> > 
> > Reverting this linux-next commit fixed booting RCU-list warnings everywhere.
> 
> Is it possible that the RCU-list warnings were being wrongly suppressed
> without a21ee6055c30?  As in are you certain that these RCU-list warnings
> are in fact false positives?

I guess you mean this commit a046a86082cc ("lockdep: Fix lockdep recursion")
instead of a21ee6055c30. It is unclear to me how that commit a046a86082cc would
suddenly start to generate those warnings, although I can see it starts to use
percpu variables even though the CPU is not yet set online.

DECLARE_PER_CPU(unsigned int, lockdep_recursion);

Anyway, the problem is that when we in the early boot:

start_secondary()
  smp_init_secondary()
    init_cpu_timer()
      clockevents_register_device()

We are taking a lock there but the CPU is not yet online, and the
__lock_acquire() would call things like hlist_for_each_entry_rcu() from
lookup_chain_cache() or register_lock_class(). Thus, triggering the RCU-list
from an offline CPU warnings.

I am not entirely sure how to fix those though.