linux-kernel - Re: [PATCH v17 02/16] preempt: Track NMI nesting to separate per-CPU counter

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20260203121520.GN1282955@noisy.programming.kicks-ass.net>
Date: Tue, 3 Feb 2026 13:15:20 +0100
From: Peter Zijlstra <peterz@...radead.org>
To: Lyude Paul <lyude@...hat.com>
Cc: rust-for-linux@...r.kernel.org, linux-kernel@...r.kernel.org,
	Thomas Gleixner <tglx@...utronix.de>,
	Boqun Feng <boqun.feng@...il.com>,
	Daniel Almeida <daniel.almeida@...labora.com>,
	Miguel Ojeda <ojeda@...nel.org>,
	Alex Gaynor <alex.gaynor@...il.com>, Gary Guo <gary@...yguo.net>,
	Björn Roy Baron <bjorn3_gh@...tonmail.com>,
	Benno Lossin <lossin@...nel.org>,
	Andreas Hindborg <a.hindborg@...nel.org>,
	Alice Ryhl <aliceryhl@...gle.com>, Trevor Gross <tmgross@...ch.edu>,
	Danilo Krummrich <dakr@...nel.org>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Ingo Molnar <mingo@...hat.com>, Will Deacon <will@...nel.org>,
	Waiman Long <longman@...hat.com>,
	Joel Fernandes <joelagnelf@...dia.com>
Subject: Re: [PATCH v17 02/16] preempt: Track NMI nesting to separate per-CPU
 counter

On Wed, Jan 21, 2026 at 05:39:05PM -0500, Lyude Paul wrote:

>  #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);	\
> +		__this_cpu_inc(nmi_nesting);			\
> +		__preempt_count_add(HARDIRQ_OFFSET);		\
> +		preempt_count_set(preempt_count() | NMI_MASK);	\
>  	} while (0)
>  
>  #define nmi_enter()						\
> @@ -124,8 +128,12 @@ void irq_exit_rcu(void);
>  
>  #define __nmi_exit()						\
>  	do {							\
> +		unsigned int nesting;				\
>  		BUG_ON(!in_nmi());				\
> -		__preempt_count_sub(NMI_OFFSET + HARDIRQ_OFFSET);	\
> +		__preempt_count_sub(HARDIRQ_OFFSET);		\
> +		nesting = __this_cpu_dec_return(nmi_nesting);	\
> +		if (!nesting)					\
> +			__preempt_count_sub(NMI_OFFSET);	\
>  		arch_nmi_exit();				\
>  		lockdep_on();					\
>  	} while (0)

While not wrong like last time; it is pretty awful. 

preempt_count_set() is a cmpxchg() loop.

Would not something like so be better?

#define __nmi_enter()                                           \
        do {                                                    \
+               unsigned int _o = NMI_MASK + HARDIRQ_OFFSET;    \
                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) == ~0U);    \
+               __this_cpu_inc(nmi_nesting);                    \
+               _o -= (preempt_count() & NMI_MASK);             \
+               __preempt_count_add(_o);                        \
        } while (0)


 #define __nmi_exit()                                           \
        do {                                                    \
+               unsigned int _o = HARDIRQ_OFFSET;               \
                BUG_ON(!in_nmi());                              \
-               __preempt_count_sub(NMI_OFFSET + HARDIRQ_OFFSET);       \
+               if (!__this_cpu_dec_return(nmi_nesting))        \
+                       _o += NMI_MASK;                         \
+               __preempt_count_sub(_o);                        \
                arch_nmi_exit();                                \
                lockdep_on();                                   \
        } while (0)


But I'm really somewhat sad that 64bit can't do better than this.