netdev - Re: Advice on MFD-style probing of DSA switch SoCs

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] [day] [month] [year] [list]

Message-ID: <20230211213643.talp76wceyc7mypp@skbuf>
Date:   Sat, 11 Feb 2023 23:36:43 +0200
From:   Vladimir Oltean <vladimir.oltean@....com>
To:     Andrew Lunn <andrew@...n.ch>
Cc:     Saravana Kannan <saravanak@...gle.com>,
        Krzysztof Kozlowski <krzysztof.kozlowski@...aro.org>,
        Rob Herring <robh+dt@...nel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
        Florian Fainelli <f.fainelli@...il.com>,
        Lee Jones <lee@...nel.org>,
        Colin Foster <colin.foster@...advantage.com>,
        Alexandre Belloni <alexandre.belloni@...tlin.com>,
        netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
        devicetree@...r.kernel.org
Subject: Re: Advice on MFD-style probing of DSA switch SoCs

On Sat, Feb 11, 2023 at 04:50:39PM +0100, Andrew Lunn wrote:
> I'm not sure that is true. The switches probe method should register
> with the driver core any resources a switch provides. So switch #1
> MDIO bus driver is registered during its probe, allowing the probe of
> switch #2 to happen. When switch #2 probes, it should register its
> clock with the common clock framework, etc.
> 
> However, the linking of resources together, the PTP clock in your
> example, should happen in the switches setup() call, which only
> happens once all the switches in the cluster have probed, so all the
> needed resources should be available.
> 
> Because we have these two phases, i think the above setup would work.

I was thinking it wouldn't work because the PTP timer sub-driver would
first have to request the clock input from the CCF, before registering
its own clock output with the CCF. So the driver writer would be forced
to request a clock from the CCF in the probe() phase, if it wanted to
also register as a clock provider in the probe() phase.

I.o.w., switch #1 would get an -EPROBE_DEFER waiting for the clock from
switch #2, before it would get to call dsa_register_switch().

TLDR: the rule "provide all you have in stage 1, request all you need
in stage 2" would only work if you don't need anything to register your
providers.