linux-kernel - Re: [PATCH net-next] devlink: Make devlink

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 for Android: free password hash cracker in your pocket

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

Message-ID: <YUlBGk2Mq3iYhtku@unreal>
Date:   Tue, 21 Sep 2021 05:19:06 +0300
From:   Leon Romanovsky <leon@...nel.org>
To:     Jakub Kicinski <kuba@...nel.org>
Cc:     "David S . Miller" <davem@...emloft.net>,
        Alexandre Belloni <alexandre.belloni@...tlin.com>,
        Andrew Lunn <andrew@...n.ch>, Ariel Elior <aelior@...vell.com>,
        Bin Luo <luobin9@...wei.com>,
        Claudiu Manoil <claudiu.manoil@....com>,
        Coiby Xu <coiby.xu@...il.com>,
        Derek Chickles <dchickles@...vell.com>, drivers@...sando.io,
        Felix Manlunas <fmanlunas@...vell.com>,
        Florian Fainelli <f.fainelli@...il.com>,
        Geetha sowjanya <gakula@...vell.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        GR-everest-linux-l2@...vell.com, GR-Linux-NIC-Dev@...vell.com,
        hariprasad <hkelam@...vell.com>,
        Ido Schimmel <idosch@...dia.com>,
        intel-wired-lan@...ts.osuosl.org,
        Ioana Ciornei <ioana.ciornei@....com>,
        Jerin Jacob <jerinj@...vell.com>,
        Jesse Brandeburg <jesse.brandeburg@...el.com>,
        Jiri Pirko <jiri@...dia.com>,
        Jonathan Lemon <jonathan.lemon@...il.com>,
        Linu Cherian <lcherian@...vell.com>,
        linux-kernel@...r.kernel.org, linux-omap@...r.kernel.org,
        linux-rdma@...r.kernel.org, linux-staging@...ts.linux.dev,
        Manish Chopra <manishc@...vell.com>,
        Michael Chan <michael.chan@...adcom.com>,
        netdev@...r.kernel.org, oss-drivers@...igine.com,
        Richard Cochran <richardcochran@...il.com>,
        Saeed Mahameed <saeedm@...dia.com>,
        Salil Mehta <salil.mehta@...wei.com>,
        Satanand Burla <sburla@...vell.com>,
        Shannon Nelson <snelson@...sando.io>,
        Simon Horman <simon.horman@...igine.com>,
        Subbaraya Sundeep <sbhatta@...vell.com>,
        Sunil Goutham <sgoutham@...vell.com>,
        Taras Chornyi <tchornyi@...vell.com>,
        Tariq Toukan <tariqt@...dia.com>,
        Tony Nguyen <anthony.l.nguyen@...el.com>,
        UNGLinuxDriver@...rochip.com, Vadym Kochan <vkochan@...vell.com>,
        Vivien Didelot <vivien.didelot@...il.com>,
        Vladimir Oltean <vladimir.oltean@....com>,
        Yisen Zhuang <yisen.zhuang@...wei.com>
Subject: Re: [PATCH net-next] devlink: Make devlink_register to be void

On Mon, Sep 20, 2021 at 02:04:07PM -0700, Jakub Kicinski wrote:
> On Mon, 20 Sep 2021 13:39:15 -0700 Jakub Kicinski wrote:
> > On Mon, 20 Sep 2021 17:41:44 +0300 Leon Romanovsky wrote:
> > > From: Leon Romanovsky <leonro@...dia.com>
> > > 
> > > devlink_register() can't fail and always returns success, but all drivers
> > > are obligated to check returned status anyway. This adds a lot of boilerplate
> > > code to handle impossible flow.
> > > 
> > > Make devlink_register() void and simplify the drivers that use that
> > > API call.  
> > 
> > Unlike unused functions bringing back error handling may be
> > non-trivial. I'd rather you deferred such cleanups until you're 
> > ready to post your full rework and therefore give us some confidence 
> > the revert will not be needed.
> 
> If you disagree you gotta repost, new devlink_register call got added
> in the meantime.

This is exactly what I afraid, new devlink API users are added faster
than I can cleanup them.

For example, let's take a look on newly added ipc_devlink_init(), it is
called conditionally "if (stage == IPC_MEM_EXEC_STAGE_BOOT) {". How can
it be different stage if we are in driver .probe() routine?

They also introduced devlink_sio.devlink_read_pend and
devlink_sio.read_sem to protect from something that right position of
devlink_register() will fix. I also have serious doubts that their
current protection is correct, once they called to devlink_params_publish()
the user can crash the system, because he can access the parameters before
they initialized their protection.

So yes, I disagree. We will need to make sure that devlink_register()
can't fail and it will make life easier for everyone (no need to unwind)
while we put that command  being last in probe sequence.

If I repost, will you take it? I don't want to waste anyone time if it
is not.

Thanks