Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 23 May 2019 13:20:13 +0200
From:   Andrea Parri <andrea.parri@amarulasolutions.com>
To:     Mark Rutland <mark.rutland@arm.com>
Cc:     linux-kernel@vger.kernel.org, peterz@infradead.org,
        will.deacon@arm.com, aou@eecs.berkeley.edu, arnd@arndb.de,
        bp@alien8.de, catalin.marinas@arm.com, davem@davemloft.net,
        fenghua.yu@intel.com, heiko.carstens@de.ibm.com,
        herbert@gondor.apana.org.au, ink@jurassic.park.msu.ru,
        jhogan@kernel.org, linux@armlinux.org.uk, mattst88@gmail.com,
        mingo@kernel.org, mpe@ellerman.id.au, palmer@sifive.com,
        paul.burton@mips.com, paulus@samba.org, ralf@linux-mips.org,
        rth@twiddle.net, stable@vger.kernel.org, tglx@linutronix.de,
        tony.luck@intel.com, vgupta@synopsys.com
Subject: Re: [PATCH 00/18] locking/atomic: atomic64 type cleanup
Message-ID: <20190523112013.GA14035@andrea>
References: <20190522132250.26499-1-mark.rutland@arm.com>
 <20190523083013.GA4616@andrea>
 <20190523101926.GA3370@lakrids.cambridge.arm.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20190523101926.GA3370@lakrids.cambridge.arm.com>
User-Agent: Mutt/1.5.24 (2015-08-30)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

> > While reading the series, I realized that the following expression:
> > 
> > 	atomic64_t v;
> >         ...
> > 	typeof(v.counter) my_val = atomic64_set(&v, VAL);
> > 
> > is a valid expression on some architectures (in part., on architectures
> > which #define atomic64_set() to WRITE_ONCE()) but is invalid on others.
> > (This is due to the fact that WRITE_ONCE() can be used as an rvalue in
> > the above assignment; TBH, I ignore the reasons for having such rvalue?)
> > 
> > IIUC, similar considerations hold for atomic_set().
> > 
> > The question is whether this is a known/"expected" inconsistency in the
> > implementation of atomic64_set() or if this would also need to be fixed
> > /addressed (say in a different patchset)?
> 
> In either case, I don't think the intent is that they should be used that way,
> and from a quick scan, I can only fine a single relevant instance today:
> 
> [mark@lakrids:~/src/linux]% git grep '\(return\|=\)\s\+atomic\(64\)\?_set'
> include/linux/vmw_vmci_defs.h:  return atomic_set((atomic_t *)var, (u32)new_val);
> include/linux/vmw_vmci_defs.h:  return atomic64_set(var, new_val);
> 
> 
> [mark@lakrids:~/src/linux]% git grep '=\s+atomic_set' | wc -l
> 0
> [mark@lakrids:~/src/linux]% git grep '=\s+atomic64_set' | wc -l
> 0
> 
> Any architectures implementing arch_atomic_* will have both of these functions
> returning void. Currently that's x86 and arm64, but (time permitting) I intend
> to migrate other architectures, so I guess we'll have to fix the above up as
> required.
> 
> I think it's best to avoid the construct above.

Thank you for the clarification, Mark.  I agree with you that it'd be
better to avoid such constructs.  (FWIW, it is not currently possible
to use them in litmus tests for the LKMM...)

Thanks,
  Andrea