linux-kernel - Re: [PATCH v3 09/11] Input: alps - remove unlikely() from IS

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:   Sat, 31 Aug 2019 14:03:56 -0700
From:   Dmitry Torokhov <dmitry.torokhov@...il.com>
To:     Joe Perches <joe@...ches.com>
Cc:     Pali Rohár <pali.rohar@...il.com>,
        Denis Efremov <efremov@...ux.com>,
        linux-kernel@...r.kernel.org,
        Andrew Morton <akpm@...ux-foundation.org>,
        linux-input@...r.kernel.org
Subject: Re: [PATCH v3 09/11] Input: alps - remove unlikely() from IS_ERR*()
 condition

On Sat, Aug 31, 2019 at 01:32:02PM -0700, Joe Perches wrote:
> On Sat, 2019-08-31 at 17:25 +0200, Pali Rohár wrote:
> > On Thursday 29 August 2019 10:50:39 Dmitry Torokhov wrote:
> > > On Thu, Aug 29, 2019 at 07:50:23PM +0300, Denis Efremov wrote:
> > > > "unlikely(IS_ERR_OR_NULL(x))" is excessive. IS_ERR_OR_NULL() already uses
> > > > unlikely() internally.
> > > 
> > > The keyword here is _internally_.
> > > 
> > > https://lore.kernel.org/lkml/20190821174857.GD76194@dtor-ws/
> > > 
> > > So please no.
> 
> I think it poor form not to simply restate your original
> objection from 4 message levels below this link

Thank you for the lesson in etiquette, but I posted reference to the
very message I wanted.

> 
> https://lists.gt.net/linux/kernel/2269724
> 
>    Hm... I do not like this change. If I read code 
>     
>     if (unlikely(IS_ERR_OR_NULL(priv->dev3))) 
>     
>    then I know that it is really unlikely that condition will be truth and 
>    so this is some case of error/exception or something that normally does 
>    not happen too much. 
>     
>    But if I read code 
>     
>     if (IS_ERR_OR_NULL(priv->dev3)) 
>     
>    I know nothing about chance that this condition will be truth. Explicit 
>    unlikely in previous example give me more information. 
>     
> I alslo think this argument is dubious as it also applies
> to any IS_ERR and all the unlikely uses have been removed
> from those.

No, if you read the reference I posted, the argument does not apply to
all IS_ERR() instances. Majority of them are in probe() paths where we
do not really care about likely/unlikely. Here we are dealing with
IS_ERR in a [fairly] hot path.

Thanks.

-- 
Dmitry