mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Sander Vanheule <sander@svanheule.net>
To: Yury Norov <yury.norov@gmail.com>
Cc: x86@kernel.org, linux-kernel@vger.kernel.org,
	Andrew Morton <akpm@linux-foundation.org>,
	Andy Shevchenko <andriy.shevchenko@linux.intel.com>,
	elver@google.com, gregkh@linuxfoundation.org,
	Peter Zijlstra <peterz@infradead.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	vschneid@redhat.com, Ingo Molnar <mingo@redhat.com>,
	Borislav Petkov <bp@alien8.de>,
	Dave Hansen <dave.hansen@linux.intel.com>,
	"H . Peter Anvin" <hpa@zytor.com>
Subject: Re: [PATCH v4 2/5] cpumask: Fix invalid uniprocessor mask assumption
Date: Sun, 03 Jul 2022 16:17:30 +0200	[thread overview]
Message-ID: <9a412be021706d048be6a868d2f7eecf8a239cbe.camel@svanheule.net> (raw)
In-Reply-To: <YsC7qlQba4+VJ3W1@yury-laptop>

On Sat, 2022-07-02 at 14:42 -0700, Yury Norov wrote:
> On Sat, Jul 02, 2022 at 06:08:25PM +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>
> 
> Suggested-by: Yury Norov <yury.norov@gmail.com>
> 

To be honest, I was somewhat taken aback by this Suggested-by. I certainly appreciate the feedback
you've provided, and I think the patch has gotten better for it, so I do want to acknowledge that.
During reviews of other patches however, I haven't had a reviewer request this for suggestions that
couldn't be spun off into a separate patch. If you are OK with this patch, a Reviewed-by would
definitely be warranted and seems more appropriate IMHO.


Best,
Sander

> > ---
> > 
> > Notes:
> >     Changes since v3:
> >     - Add back UP-optimised cpumask_local_spread, cpumask_any_distribute,
> >       cpumask_any_and_distribute
> >     
> >     Changes since v1:
> >     - Drop UP implementations instead of trying to fix them
> > 
> >  include/linux/cpumask.h | 99 ++++++++---------------------------------
> >  lib/Makefile            |  3 +-
> >  lib/cpumask.c           |  2 +
> >  3 files changed, 22 insertions(+), 82 deletions(-)
> > 
> > diff --git a/include/linux/cpumask.h b/include/linux/cpumask.h
> > index fe29ac7cc469..7fbef41b3093 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);
> > -}
> > -
> > -#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
> > @@ -260,10 +181,29 @@ static inline unsigned int cpumask_next_zero(int n, const struct cpumask
> > *srcp)
> >  
> >  int __pure cpumask_next_and(int n, const struct cpumask *, const struct cpumask *);
> >  int __pure cpumask_any_but(const struct cpumask *mask, unsigned int cpu);
> > +
> > +#if NR_CPUS == 1
> > +/* Uniprocessor: there is only one valid CPU */
> > +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);
> > +}
> > +#else
> >  unsigned int cpumask_local_spread(unsigned int i, int node);
> >  int cpumask_any_and_distribute(const struct cpumask *src1p,
> >                                const struct cpumask *src2p);
> >  int cpumask_any_distribute(const struct cpumask *srcp);
> > +#endif /* NR_CPUS */
> >  
> >  /**
> >   * for_each_cpu - iterate over every cpu in a mask
> > @@ -324,7 +264,6 @@ 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 f99bf61f8bbc..bcc7e8ea0cde 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
> > diff --git a/lib/cpumask.c b/lib/cpumask.c
> > index a971a82d2f43..b9728513a4d4 100644
> > --- a/lib/cpumask.c
> > +++ b/lib/cpumask.c
> > @@ -192,6 +192,7 @@ void __init free_bootmem_cpumask_var(cpumask_var_t mask)
> >  }
> >  #endif
> >  
> > +#if NR_CPUS > 1
> >  /**
> >   * cpumask_local_spread - select the i'th cpu with local numa cpu's first
> >   * @i: index number
> > @@ -279,3 +280,4 @@ int cpumask_any_distribute(const struct cpumask *srcp)
> >         return next;
> >  }
> >  EXPORT_SYMBOL(cpumask_any_distribute);
> > +#endif /* NR_CPUS */
> > -- 
> > 2.36.1


  reply	other threads:[~2022-07-03 14:17 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-07-02 16:08 [PATCH v4 0/5] cpumask: Fix invalid uniprocessor assumptions Sander Vanheule
2022-07-02 16:08 ` [PATCH v4 1/5] x86/cacheinfo: move shared cache map definitions Sander Vanheule
2022-07-02 16:08 ` [PATCH v4 2/5] cpumask: Fix invalid uniprocessor mask assumption Sander Vanheule
2022-07-02 21:42   ` Yury Norov
2022-07-03 14:17     ` Sander Vanheule [this message]
2022-07-02 16:08 ` [PATCH v4 3/5] lib/test: Introduce cpumask KUnit test suite Sander Vanheule
2022-07-02 21:45   ` Yury Norov
2022-07-03  7:26     ` Sander Vanheule
2022-07-19 21:31   ` Maíra Canal
2022-07-20  5:24     ` David Gow
2022-07-20 12:43       ` Sander Vanheule
2022-07-20 12:47     ` Sander Vanheule
2022-07-02 16:08 ` [PATCH v4 4/5] cpumask: Add UP optimised for_each_*_cpu versions Sander Vanheule
2022-07-02 21:47   ` Yury Norov
2022-07-02 21:50     ` Yury Norov
2022-07-03  7:37       ` Sander Vanheule
2022-07-02 16:08 ` [PATCH v4 5/5] cpumask: Update cpumask_next_wrap() signature Sander Vanheule
2022-07-02 20:38 ` [PATCH v4 0/5] cpumask: Fix invalid uniprocessor assumptions Andrew Morton
2022-07-03  7:50   ` Sander Vanheule
2022-07-03 20:39     ` Andrew Morton
2022-07-10  6:51       ` Sander Vanheule
2022-07-11 20:22         ` Andrew Morton

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=9a412be021706d048be6a868d2f7eecf8a239cbe.camel@svanheule.net \
    --to=sander@svanheule.net \
    --cc=akpm@linux-foundation.org \
    --cc=andriy.shevchenko@linux.intel.com \
    --cc=bp@alien8.de \
    --cc=dave.hansen@linux.intel.com \
    --cc=elver@google.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=hpa@zytor.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=peterz@infradead.org \
    --cc=tglx@linutronix.de \
    --cc=vschneid@redhat.com \
    --cc=x86@kernel.org \
    --cc=yury.norov@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®