Re: [PATCH v5] debugobjects: Don't call fill_pool() in early boot hardirq context

[Waiman Long]

On 6/7/26 1:15 AM, Ingo Molnar wrote:
> * Waiman Long <longman@xxxxxxxxxx> wrote:
>
> > When booting a debug PREEMPT_RT kernel on an arm64 system with grace
> > processor, a "inconsistent {HARDIRQ-ON-W} -> {IN-HARDIRQ-W} usage"
> > lockdep warning message was reported to the console.
> >
> > During early boot, interrupts are getting enabled before the scheduler
> > is enabled. In this window (before SYSTEM_SCHEDULING is set) interrupts
> > can fire and attempt to fill the pool from within the hardirq. This can
> > lead to a deadlock the interrupt occurred while in the memory allocator.
> >
> > Add a new can_fill_pool() helper and reorder the exception rule and
> > forbid this scenario by excluding allocations from hardirq.
> >
> > Fixes: 06e0ae988f6e ("debugobjects: Allow to refill the pool before SYSTEM_SCHEDULING")
> > Co-developed-by: Waiman Long <longman@xxxxxxxxxx>
> > Co-developed-by: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
> > Co-developed-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> > Reviewed-by: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
> > Signed-off-by: Waiman Long <longman@xxxxxxxxxx>
> > ---
> >   lib/debugobjects.c | 46 +++++++++++++++++++++++++++++++++++++---------
> >   1 file changed, 37 insertions(+), 9 deletions(-)
> >
> > diff --git a/lib/debugobjects.c b/lib/debugobjects.c
> > index b18a682fe3da..6fb00e08a4e2 100644
> > --- a/lib/debugobjects.c
> > +++ b/lib/debugobjects.c
> > @@ -720,6 +720,41 @@ static inline bool debug_objects_is_pi_blocked_on(void)
> >   #endif
> >   }
> >   
> > +static inline bool can_fill_pool(void)
> > +{
> > +	/*
> > +	 * On !RT enabled kernels there are no restrictions and spinlock_t and
> > +	 * raw_spinlock_t are the same types.
> > +	 */
> > +	if (!IS_ENABLED(CONFIG_PREEMPT_RT))
> > +		return true;
> > -	if (!IS_ENABLED(CONFIG_PREEMPT_RT) || system_state < SYSTEM_SCHEDULING ||
> > -	    (preemptible() && !debug_objects_is_pi_blocked_on())) {
> > +	if (can_fill_pool()) {
> Would have been nice to have this as two patches: first one
> factors out the check into can_fill_pool() without changing the
> logic, while the second one applies the fix to can_fill_pool().

OK, I will send a v6 patch as suggested. However, Thomas had merged the 
v5 patch into tip. I am fine whether it will be one patch or 2 as long 
as the problem is going to be fixed.

Cheers,
Longman