Re: [PATCH 02/11] preempt: Track NMI nesting to separate per-CPU counter

[Boqun Feng <boqun@kernel.org>]

On Tue, May 12, 2026 at 03:22:39PM -0400, Joel Fernandes wrote:
> 
> 
> On 5/12/2026 12:30 PM, Steven Rostedt wrote:
> > On Thu,  7 May 2026 21:21:02 -0700
> > Boqun Feng <boqun@kernel.org> wrote:
> > 
> >> From: Joel Fernandes <joelagnelf@nvidia.com>
> >>
> >> Move NMI nesting tracking from the preempt_count bits to a separate per-CPU
> >> counter (nmi_nesting). This is to free up the NMI bits in the preempt_count,
> >> allowing those bits to be repurposed for other uses.  This also has the benefit
> >> of tracking more than 16-levels deep if there is ever a need.
> >>
> >> Reduce multiple bits in preempt_count for NMI tracking. Reduce NMI_BITS
> >> from 3 to 1, using it only to detect if we're in an NMI.
> >>
> >> Suggested-by: Boqun Feng <boqun.feng@gmail.com>
> >> Signed-off-by: Joel Fernandes <joelaf@google.com>
> >> Signed-off-by: Lyude Paul <lyude@redhat.com>
> >> Signed-off-by: Boqun Feng <boqun@kernel.org>
> >> Link: https://patch.msgid.link/20260121223933.1568682-3-lyude@redhat.com
> >> ---
> >>  include/linux/hardirq.h | 16 ++++++++++++----
> >>  include/linux/preempt.h | 13 +++++++++----
> >>  kernel/softirq.c        |  2 ++
> >>  3 files changed, 23 insertions(+), 8 deletions(-)
> >>
> >> diff --git a/include/linux/hardirq.h b/include/linux/hardirq.h
> >> index d57cab4d4c06..cc06bda52c3e 100644
> >> --- a/include/linux/hardirq.h
> >> +++ b/include/linux/hardirq.h
> >> @@ -10,6 +10,8 @@
> >>  #include <linux/vtime.h>
> >>  #include <asm/hardirq.h>
> >>  
> >> +DECLARE_PER_CPU(unsigned int, nmi_nesting);
> >> +
> >>  extern void synchronize_irq(unsigned int irq);
> >>  extern bool synchronize_hardirq(unsigned int irq);
> >>  
> >> @@ -102,14 +104,16 @@ void irq_exit_rcu(void);
> >>   */
> >>  
> >>  /*
> >> - * nmi_enter() can nest up to 15 times; see NMI_BITS.
> >> + * nmi_enter() can nest - nesting is tracked in a per-CPU counter.
> >>   */
> >>  #define __nmi_enter()						\
> >>  	do {							\
> >>  		lockdep_off();					\
> >>  		arch_nmi_enter();				\
> >> -		BUG_ON(in_nmi() == NMI_MASK);			\
> >> -		__preempt_count_add(NMI_OFFSET + HARDIRQ_OFFSET);	\
> >> +		BUG_ON(__this_cpu_read(nmi_nesting) == UINT_MAX);	\
> > 
> > I think we should keep the max nesting fixed to 15. If this doesn't trigger
> > until UINT_MAX, it may take a long time to see that, and there's no reason
> > NMIs should nest more than 15 anyway.



> > 
> > Just because the counter allows it, doesn't me the system should allow it.
> 
> That's fine with me. Boqun, do you want to make the one-line change to the patch?
> 

Something like this on top of your patch?

diff --git a/include/linux/hardirq.h b/include/linux/hardirq.h
index cc06bda52c3e..a59a33e0f5ca 100644
--- a/include/linux/hardirq.h
+++ b/include/linux/hardirq.h
@@ -110,7 +110,8 @@ void irq_exit_rcu(void);
 	do {							\
 		lockdep_off();					\
 		arch_nmi_enter();				\
-		BUG_ON(__this_cpu_read(nmi_nesting) == UINT_MAX);	\
+		/* Maximum NMI nesting is 15 */			\
+		BUG_ON(__this_cpu_read(nmi_nesting) == 15);	\
 		__this_cpu_inc(nmi_nesting);			\
 		__preempt_count_add(HARDIRQ_OFFSET);		\
 		preempt_count_set(preempt_count() | NMI_MASK);	\

I will need to adjust this in patch #10 as well, but shouldn't be hard.

Regards,
Boqun

> Thanks.
>