Re: [PATCH 1/4] arm64: percpu: Always inline helpers used by noinstr code

[Jisheng Zhang <jszhang@kernel.org>]

On Sun, Nov 08, 2026 at 07:43:56AM +0000, sashiko-bot@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential
> issue to consider:
> - [Low] arm64: percpu: __PERCPU_OP_CASE() and sibling helpers not marked
>   __always_inline
> --
> 
> commit 3a2df0916b5200b227df21487897e171275b29f4
> Author: Jisheng Zhang <jszhang@kernel.org>
> 
> arm64: percpu: Always inline helpers used by noinstr code
> 
> This commit marks __hyp_my_cpu_offset(), __kern_my_cpu_offset(), and
> __percpu_read_##sz() as __always_inline so they are not emitted out of line
> when called from non-instrumentable code.
> 
> > diff --git a/arch/arm64/include/asm/percpu.h b/arch/arm64/include/asm/percpu.h
> > index 3cc316091d665..7c344de5052d5 100644
> > --- a/arch/arm64/include/asm/percpu.h
> > +++ b/arch/arm64/include/asm/percpu.h
> 
> [ ... ]
> 
> > @@ -129,7 +129,7 @@ static inline unsigned long __kern_my_cpu_offset(void)
> >  #endif
> >  
> >  #define PERCPU_RW_OPS(w, sfx, sz)						\
> > -static inline unsigned long __percpu_read_##sz(void __percpu *pcp)		\
> > +static __always_inline unsigned long __percpu_read_##sz(void __percpu *pcp)		\
> >  {										\
> >  	u16 *gprs = &current_thread_info()->pcpu_gprs;				\
> >  	unsigned long off;							\
> 
> [Severity: Low]
> Should __PERCPU_OP_CASE() and the remaining __percpu_* helpers in this header
> (__percpu_write_##sz(), __PERCPU_RET_OP_CASE(), PERCPU_XCHG_OP(),
> PERCPU_CMPXCHG_OP(), and __percpu_cmpxchg_128()) also be marked
> __always_inline?
> 
> The noinstr function lockdep_hardirqs_on() calls
> __debug_atomic_inc(redundant_hardirqs_on), which under CONFIG_DEBUG_LOCKDEP=y
> expands to this_cpu_inc(lockdep_stats.redundant_hardirqs_on) ->
> __percpu_add_case_64().  Because __PERCPU_OP_CASE() still uses plain static
> inline and is also called from instrumentable lockdep functions, builds with
> CONFIG_CC_OPTIMIZE_FOR_SIZE=y can emit __percpu_add_case_64() out of line in
> .text and call it from .noinstr.text before RCU and lockdep state are ready
> for instrumentation.

Good catch! I will address this in v2. Before that, I will wait
for a while to collect review comments.

Thanks
> 
> -- 
> Sashiko AI review · https://sashiko.dev/#/patchset/20261107071339.17467-1-jszhang@kernel.org?part=1