Re: [PATCH rcu 10/11] srcu: Add FORCE_NEED_SRCU_NMI_SAFE Kconfig for testing

[Paul E. McKenney]

On Tue, Mar 25, 2025 at 09:04:31AM +0100, Geert Uytterhoeven wrote:
> Hi Boqun, Paul,
> 
> On Wed, 19 Feb 2025 at 16:44, Boqun Feng <boqun.feng@xxxxxxxxx> wrote:
> > From: "Paul E. McKenney" <paulmck@xxxxxxxxxx>
> >
> > The srcu_read_lock_nmisafe() and srcu_read_unlock_nmisafe() functions
> > map to __srcu_read_lock() and __srcu_read_unlock() on systems like x86
> > that have NMI-safe this_cpu_inc() operations.  This makes the underlying
> > __srcu_read_lock_nmisafe() and __srcu_read_unlock_nmisafe() functions
> > difficult to test on (for example) x86 systems, allowing bugs to creep in.
> >
> > This commit therefore creates a FORCE_NEED_SRCU_NMI_SAFE Kconfig that
> > forces those underlying functions to be used even on systems where they
> > are not needed, thus providing better testing coverage.
> >
> > Signed-off-by: Paul E. McKenney <paulmck@xxxxxxxxxx>
> > Signed-off-by: Boqun Feng <boqun.feng@xxxxxxxxx>
> 
> Thanks for your patch, which is now commit 536e8b9b80bc7a0a ("srcu:
> Add FORCE_NEED_SRCU_NMI_SAFE Kconfig for testing") in linus/master
> 
> > --- a/kernel/rcu/Kconfig
> > +++ b/kernel/rcu/Kconfig
> > @@ -65,6 +65,17 @@ config TREE_SRCU
> >         help
> >           This option selects the full-fledged version of SRCU.
> >
> > +config FORCE_NEED_SRCU_NMI_SAFE
> > +       bool "Force selection of NEED_SRCU_NMI_SAFE"
> 
> What am I supposed to answer here? "n" I guess.
> What about distro and allmodconfig kernels?

Yes, you should select "n" unless ...

> > +       depends on !TINY_SRCU
> > +       select NEED_SRCU_NMI_SAFE
> > +       default n
> > +       help
> > +         This option forces selection of the NEED_SRCU_NMI_SAFE
> > +         Kconfig option, allowing testing of srcu_read_lock_nmisafe()
> > +         and srcu_read_unlock_nmisafe() on architectures (like x86)
> > +         that select the ARCH_HAS_NMI_SAFE_THIS_CPU_OPS Kconfig option.
> 
> Perhaps this should depend on ARCH_HAS_NMI_SAFE_THIS_CPU_OPS?

... you are on a system selecting ARCH_HAS_NMI_SAFE_THIS_CPU_OPS and
you would like to test the SRCU setup that needed only by systems that
do not select ARCH_HAS_NMI_SAFE_THIS_CPU_OPS.

Ah.  I forgot to add "depends on RCU_EXPERT".

Apologies, I will fix this.  Does the patch show below do the trick?

							Thanx, Paul

------------------------------------------------------------------------

commit b5c8c6f89c6d7ac778e961ad4b883eada0c1f42a
Author: Paul E. McKenney <paulmck@xxxxxxxxxx>
Date:   Tue Mar 25 07:31:45 2025 -0700

    srcu: Make FORCE_NEED_SRCU_NMI_SAFE depend on RCU_EXPERT
    
    The FORCE_NEED_SRCU_NMI_SAFE is useful only for those wishing to test
    the SRCU code paths that accommodate architectures that do not have
    NMI-safe per-CPU operations, that is, those architectures that do not
    select the ARCH_HAS_NMI_SAFE_THIS_CPU_OPS Kconfig option.  As such, this
    is a specialized Kconfig option that is not intended for casual users.
    
    This commit therefore hides it behind the RCU_EXPERT Kconfig option.
    
    Reported-by: Geert Uytterhoeven <geert@xxxxxxxxxxxxxx>
    Closes: https://lore.kernel.org/all/CAMuHMdX6dy9_tmpLkpcnGzxyRbe6qSWYukcPp=H1GzZdyd3qBQ@xxxxxxxxxxxxxx/
    Signed-off-by: Paul E. McKenney <paulmck@xxxxxxxxxx>

diff --git a/kernel/rcu/Kconfig b/kernel/rcu/Kconfig
index b3f985d41717a..cc4ce79f58aa6 100644
--- a/kernel/rcu/Kconfig
+++ b/kernel/rcu/Kconfig
@@ -68,6 +68,7 @@ config TREE_SRCU
 config FORCE_NEED_SRCU_NMI_SAFE
 	bool "Force selection of NEED_SRCU_NMI_SAFE"
 	depends on !TINY_SRCU
+	depends on RCU_EXPERT
 	select NEED_SRCU_NMI_SAFE
 	default n
 	help