Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Mon, 6 Jun 2022 12:14:44 -0700
From:   Yury Norov <yury.norov@gmail.com>
To:     Sander Vanheule <sander@svanheule.net>
Cc:     Peter Zijlstra <peterz@infradead.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        Valentin Schneider <vschneid@redhat.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Marco Elver <elver@google.com>, linux-kernel@vger.kernel.org,
        Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Subject: Re: [PATCH v3 1/4] cpumask: Fix invalid uniprocessor mask assumption
Message-ID: <Yp5SJAkq8whC8EBI@yury-laptop>
References: <cover.1654410109.git.sander@svanheule.net>
 <cc1f7d3334348cccbf9fde091015a802f379a9be.1654410109.git.sander@svanheule.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <cc1f7d3334348cccbf9fde091015a802f379a9be.1654410109.git.sander@svanheule.net>
Precedence: bulk

On Sun, Jun 05, 2022 at 08:22:38AM +0200, Sander Vanheule wrote:
> On uniprocessor builds, any CPU mask is assumed to contain exactly one
> CPU (cpu0). This assumption ignores the existence of empty masks,
> resulting in incorrect behaviour.
> cpumask_first_zero(), cpumask_next_zero(), and for_each_cpu_not() don't
> provide behaviour matching the assumption that a UP mask is always "1",
> and instead provide behaviour matching the empty mask.
> 
> Drop the incorrectly optimised code and use the generic implementations
> in all cases.
> 
> Signed-off-by: Sander Vanheule <sander@svanheule.net>
> ---
> Changes since v1:
> - Drop UP implementations instead of trying to fix them
> ---
>  include/linux/cpumask.h | 80 -----------------------------------------
>  lib/Makefile            |  3 +-
>  2 files changed, 1 insertion(+), 82 deletions(-)
> 
> diff --git a/include/linux/cpumask.h b/include/linux/cpumask.h
> index fe29ac7cc469..d6add0e29ef4 100644
> --- a/include/linux/cpumask.h
> +++ b/include/linux/cpumask.h
> @@ -116,85 +116,6 @@ static __always_inline unsigned int cpumask_check(unsigned int cpu)
>  	return cpu;
>  }
>  
> -#if NR_CPUS == 1
> -/* Uniprocessor.  Assume all masks are "1". */
> -static inline unsigned int cpumask_first(const struct cpumask *srcp)
> -{
> -	return 0;
> -}
> -
> -static inline unsigned int cpumask_first_zero(const struct cpumask *srcp)
> -{
> -	return 0;
> -}
> -
> -static inline unsigned int cpumask_first_and(const struct cpumask *srcp1,
> -					     const struct cpumask *srcp2)
> -{
> -	return 0;
> -}
> -
> -static inline unsigned int cpumask_last(const struct cpumask *srcp)
> -{
> -	return 0;
> -}
> -
> -/* Valid inputs for n are -1 and 0. */
> -static inline unsigned int cpumask_next(int n, const struct cpumask *srcp)
> -{
> -	return n+1;
> -}
> -
> -static inline unsigned int cpumask_next_zero(int n, const struct cpumask *srcp)
> -{
> -	return n+1;
> -}
> -
> -static inline unsigned int cpumask_next_and(int n,
> -					    const struct cpumask *srcp,
> -					    const struct cpumask *andp)
> -{
> -	return n+1;
> -}
> -
> -static inline unsigned int cpumask_next_wrap(int n, const struct cpumask *mask,
> -					     int start, bool wrap)
> -{
> -	/* cpu0 unless stop condition, wrap and at cpu0, then nr_cpumask_bits */
> -	return (wrap && n == 0);
> -}
> -
> -/* cpu must be a valid cpu, ie 0, so there's no other choice. */
> -static inline unsigned int cpumask_any_but(const struct cpumask *mask,
> -					   unsigned int cpu)
> -{
> -	return 1;
> -}
> -
> -static inline unsigned int cpumask_local_spread(unsigned int i, int node)
> -{
> -	return 0;
> -}
> -
> -static inline int cpumask_any_and_distribute(const struct cpumask *src1p,
> -					     const struct cpumask *src2p) {
> -	return cpumask_first_and(src1p, src2p);
> -}
> -
> -static inline int cpumask_any_distribute(const struct cpumask *srcp)
> -{
> -	return cpumask_first(srcp);
> -}

It looks like cpumask_local_spread, cpumask_any_and_distribute and
cpumask_any_distribute were correct and better optimized in UP case.
cpumask_local_spread - for sure. I think it's worth keeping them
optimized.

Thanks,
Yury

> -#define for_each_cpu(cpu, mask)			\
> -	for ((cpu) = 0; (cpu) < 1; (cpu)++, (void)mask)
> -#define for_each_cpu_not(cpu, mask)		\
> -	for ((cpu) = 0; (cpu) < 1; (cpu)++, (void)mask)
> -#define for_each_cpu_wrap(cpu, mask, start)	\
> -	for ((cpu) = 0; (cpu) < 1; (cpu)++, (void)mask, (void)(start))
> -#define for_each_cpu_and(cpu, mask1, mask2)	\
> -	for ((cpu) = 0; (cpu) < 1; (cpu)++, (void)mask1, (void)mask2)
> -#else
>  /**
>   * cpumask_first - get the first cpu in a cpumask
>   * @srcp: the cpumask pointer
> @@ -324,7 +245,6 @@ extern int cpumask_next_wrap(int n, const struct cpumask *mask, int start, bool
>  	for ((cpu) = -1;						\
>  		(cpu) = cpumask_next_and((cpu), (mask1), (mask2)),	\
>  		(cpu) < nr_cpu_ids;)
> -#endif /* SMP */
>  
>  #define CPU_BITS_NONE						\
>  {								\
> diff --git a/lib/Makefile b/lib/Makefile
> index 89fcae891361..6f26a429115b 100644
> --- a/lib/Makefile
> +++ b/lib/Makefile
> @@ -34,10 +34,9 @@ lib-y := ctype.o string.o vsprintf.o cmdline.o \
>  	 is_single_threaded.o plist.o decompress.o kobject_uevent.o \
>  	 earlycpio.o seq_buf.o siphash.o dec_and_lock.o \
>  	 nmi_backtrace.o nodemask.o win_minmax.o memcat_p.o \
> -	 buildid.o
> +	 buildid.o cpumask.o
>  
>  lib-$(CONFIG_PRINTK) += dump_stack.o
> -lib-$(CONFIG_SMP) += cpumask.o
>  
>  lib-y	+= kobject.o klist.o
>  obj-y	+= lockref.o
> -- 
> 2.36.1