linux-kernel - Re: [PATCH 2/2] compiler: Prevent evaluation of WRITE

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: Windows password security audit tool. GUI, reports in PDF.

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

Date:   Fri, 24 May 2019 13:24:57 +0200
From:   Andrea Parri <andrea.parri@...rulasolutions.com>
To:     Mark Rutland <mark.rutland@....com>
Cc:     linux-kernel@...r.kernel.org, Arnd Bergmann <arnd@...db.de>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Jorgen Hansen <jhansen@...are.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Will Deacon <will.deacon@....com>,
        "Paul E. McKenney" <paulmck@...ux.ibm.com>
Subject: Re: [PATCH 2/2] compiler: Prevent evaluation of WRITE_ONCE()

On Fri, May 24, 2019 at 11:53:40AM +0100, Mark Rutland wrote:
> 
> This would be better titled as:
> 
>   compiler: don't return a value from WRITE_ONCE()

No strong opinion here: I'll adopt your suggestion in v2 if there are
no objections.  And similarly for the rcu_assign_pointer() patch.


> 
> ... since we do want the WRITE_ONCE() itself to be evaluated.
> 
> On Fri, May 24, 2019 at 12:35:36PM +0200, Andrea Parri wrote:
> > Now that there's no single use of the value of WRITE_ONCE(), change
> > the implementation to eliminate it.
> 
> I hope that's the case, but it's possible that some macros might be
> relying on this, so it's probably worth waiting to see if the kbuild
> test robot screams.

Absolutely!  Does kbuild process your tree?  I might be worth to apply
the patch to just see what kbuild 'think' about it...


> 
> Otherwise, I agree that WRITE_ONCE() returning a value is surprising,
> and unnecessary. IIRC you said that trying to suport that in other
> implementations was painful, so aligning on a non-returning version
> sounds reasonable to me.

And I should probably also modify the few #define-s under tools/ (that
I missed in this iteration...)

Thanks,
  Andrea