Re: [RFC][PATCH] locking/refcount: Provide __refcount API to obtain the old value

From: Kees Cook <keescook@chromium.org>
To: peterz@infradead.org
Cc: David Howells <dhowells@redhat.com>,
	Xiaoming Ni <nixiaoming@huawei.com>,
	David Windsor <dwindsor@gmail.com>,
	Hans Liljestrand <ishkamiel@gmail.com>,
	Elena Reshetova <elena.reshetova@intel.com>,
	Paul Moore <paul@paul-moore.com>,
	edumazet@google.com, paulmck@kernel.org, shakeelb@google.com,
	James Morris <jamorris@linux.microsoft.com>,
	alex.huangjianhui@huawei.com, dylix.dailei@huawei.com,
	chenzefeng2@huawei.com, linux-kernel@vger.kernel.org,
	Will Deacon <will@kernel.org>
Subject: Re: [RFC][PATCH] locking/refcount: Provide __refcount API to obtain the old value
Date: Wed, 29 Jul 2020 12:29:14 -0700	[thread overview]
Message-ID: <202007291228.5DFF9315@keescook> (raw)
In-Reply-To: <20200729113731.GA2678@hirez.programming.kicks-ass.net>

On Wed, Jul 29, 2020 at 01:37:31PM +0200, peterz@infradead.org wrote:
> On Wed, Jul 29, 2020 at 01:11:20PM +0200, peterz@infradead.org wrote:
> 
> > Subject: locking/refcount: Provide __refcount API to obtain the old value
> > From: Peter Zijlstra <peterz@infradead.org>
> > Date: Wed Jul 29 13:00:57 CEST 2020
> > 
> > David requested means to obtain the old/previous value from the
> > refcount API for tracing purposes.
> > 
> > Duplicate (most of) the API as __refcount*() with an additional
> > 'int *' argument into which, if !NULL, the old value will be stored.
> > 
> > Requested-by: David Howells <dhowells@redhat.com>
> > Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> > ---
> >  include/linux/refcount.h |   65 +++++++++++++++++++++++++++++++++++++++++------
> >  1 file changed, 57 insertions(+), 8 deletions(-)
> > 
> > --- a/include/linux/refcount.h
> > +++ b/include/linux/refcount.h
> > @@ -165,7 +165,7 @@ static inline unsigned int refcount_read
> >   *
> >   * Return: false if the passed refcount is 0, true otherwise
> >   */
> > -static inline __must_check bool refcount_add_not_zero(int i, refcount_t *r)
> > +static inline __must_check bool __refcount_add_not_zero(int i, refcount_t *r, int *oldp)
> >  {
> >  	int old = refcount_read(r);
> >  
> > @@ -174,12 +174,20 @@ static inline __must_check bool refcount
> >  			break;
> >  	} while (!atomic_try_cmpxchg_relaxed(&r->refs, &old, old + i));
> >  
> > +	if (oldp)
> > +		*oldp = old;
> > +
> >  	if (unlikely(old < 0 || old + i < 0))
> >  		refcount_warn_saturate(r, REFCOUNT_ADD_NOT_ZERO_OVF);
> >  
> >  	return old;
> >  }
> >  
> > +static inline __must_check bool refcount_add_not_zero(int i, refcount_t *r)
> > +{
> > +	return __refcount_add_not_zero(i, r, NULL);
> > +}
> 
> so, I could also emulate C++'s
> 
> bool refcount_add_not_zero(int i, refcount_t *r, int *oldp = NULL)
> 
> style by going to town on this with a bunch of CPP magic, but I don't
> think that'll actually make things clearer.

Erg. No, I like the __-version better -- it looks more like other kernel
APIs.

-- 
Kees Cook