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, 21 Nov 2019 09:11:58 -0800
From:   Kees Cook <keescook@chromium.org>
To:     dsterba@suse.cz, Will Deacon <will@kernel.org>,
        linux-kernel@vger.kernel.org, Ingo Molnar <mingo@kernel.org>,
        Elena Reshetova <elena.reshetova@intel.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Ard Biesheuvel <ard.biesheuvel@linaro.org>,
        Hanjun Guo <guohanjun@huawei.com>
Subject: Re: [RESEND PATCH v4 03/10] lib/refcount: Remove unused
 refcount_*_checked() variants
Message-ID: <201911210910.81231377F@keescook>
References: <20191121115902.2551-1-will@kernel.org>
 <20191121115902.2551-4-will@kernel.org>
 <20191121145533.GZ3001@twin.jikos.cz>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20191121145533.GZ3001@twin.jikos.cz>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Nov 21, 2019 at 03:55:33PM +0100, David Sterba wrote:
> On Thu, Nov 21, 2019 at 11:58:55AM +0000, Will Deacon wrote:
> > The full-fat refcount implementation is exposed via a set of functions
> > suffixed with "_checked()", the idea being that code can choose to use
> > the more expensive, yet more secure implementation on a case-by-case
> > basis.
> > 
> > In reality, this hasn't happened, so with a grand total of zero users,
> > let's remove the checked variants for now by simply dropping the suffix
> > and predicating the out-of-line functions on CONFIG_REFCOUNT_FULL=y.
> 
> I am still interested in the _checked versions and have a WIP patch that
> adds that to btrfs (that was my original plan) but haven't had enough
> time to finalize it. The patch itself is simple, the missing part is to
> understand and document what the saturated counters would do with the
> structures.

The good news is that this series removes the case of refcount_t _not_
being checked, so there's no need for _checked helpers.
CONFIG_REFCOUNT_FULL gets removed because all refcount_t ends up being
checked on all architectures. No extra work needed! :) (See patch 8)

-- 
Kees Cook