netdev - Re: [net-next: PATCH v3 2/6] net: mdiobus: Introduce fwnode_mdbiobus

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: <CAPv3WKdSNZzvDS5-FNK4eTMfRt6xuk6xYJ6eQf0N01Q4OhRS9Q@mail.gmail.com>
Date:   Thu, 24 Jun 2021 13:10:04 +0200
From:   Marcin Wojtas <mw@...ihalf.com>
To:     Andrew Lunn <andrew@...n.ch>
Cc:     Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        netdev <netdev@...r.kernel.org>,
        "David S. Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>,
        Russell King - ARM Linux <linux@...linux.org.uk>,
        Grzegorz Jaszczyk <jaz@...ihalf.com>,
        Grzegorz Bernacki <gjb@...ihalf.com>, upstream@...ihalf.com,
        Samer El-Haj-Mahmoud <Samer.El-Haj-Mahmoud@....com>,
        Jon Nettleton <jon@...id-run.com>,
        Tomasz Nowicki <tn@...ihalf.com>, rjw@...ysocki.net,
        lenb@...nel.org, Stephen Rothwell <sfr@...b.auug.org.au>
Subject: Re: [net-next: PATCH v3 2/6] net: mdiobus: Introduce fwnode_mdbiobus_register()

Hi,

czw., 24 cze 2021 o 00:10 Marcin Wojtas <mw@...ihalf.com> napisał(a):
>
> Hi,
>
> śr., 23 cze 2021 o 22:22 Andrew Lunn <andrew@...n.ch> napisał(a):
> >
> > On Mon, Jun 21, 2021 at 07:30:24PM +0200, Marcin Wojtas wrote:
> > > This patch introduces a new helper function that
> > > wraps acpi_/of_ mdiobus_register() and allows its
> > > usage via common fwnode_ interface.
> > >
> > > Fall back to raw mdiobus_register() in case CONFIG_FWNODE_MDIO
> > > is not enabled, in order to satisfy compatibility
> > > in all future user drivers.
> > >
> > > Signed-off-by: Marcin Wojtas <mw@...ihalf.com>
> > > ---
> > >  include/linux/fwnode_mdio.h    | 12 +++++++++++
> > >  drivers/net/mdio/fwnode_mdio.c | 22 ++++++++++++++++++++
> > >  2 files changed, 34 insertions(+)
> > >
> > > diff --git a/include/linux/fwnode_mdio.h b/include/linux/fwnode_mdio.h
> > > index faf603c48c86..13d4ae8fee0a 100644
> > > --- a/include/linux/fwnode_mdio.h
> > > +++ b/include/linux/fwnode_mdio.h
> > > @@ -16,6 +16,7 @@ int fwnode_mdiobus_phy_device_register(struct mii_bus *mdio,
> > >  int fwnode_mdiobus_register_phy(struct mii_bus *bus,
> > >                               struct fwnode_handle *child, u32 addr);
> > >
> > > +int fwnode_mdiobus_register(struct mii_bus *bus, struct fwnode_handle *fwnode);
> > >  #else /* CONFIG_FWNODE_MDIO */
> > >  int fwnode_mdiobus_phy_device_register(struct mii_bus *mdio,
> > >                                      struct phy_device *phy,
> > > @@ -30,6 +31,17 @@ static inline int fwnode_mdiobus_register_phy(struct mii_bus *bus,
> > >  {
> > >       return -EINVAL;
> > >  }
> > > +
> > > +static inline int fwnode_mdiobus_register(struct mii_bus *bus,
> > > +                                       struct fwnode_handle *fwnode)
> > > +{
> > > +     /*
> > > +      * Fall back to mdiobus_register() function to register a bus.
> > > +      * This way, we don't have to keep compat bits around in drivers.
> > > +      */
> > > +
> > > +     return mdiobus_register(mdio);
> > > +}
> > >  #endif
> >
> > I looked at this some more, and in the end i decided it was O.K.
> >
> > > +/**
> > > + * fwnode_mdiobus_register - bring up all the PHYs on a given MDIO bus and
> > > + *   attach them to it.
> > > + * @bus: Target MDIO bus.
> > > + * @fwnode: Pointer to fwnode of the MDIO controller.
> > > + *
> > > + * Return values are determined accordingly to acpi_/of_ mdiobus_register()
> > > + * operation.
> > > + */
> > > +int fwnode_mdiobus_register(struct mii_bus *bus, struct fwnode_handle *fwnode)
> > > +{
> > > +     if (is_acpi_node(fwnode))
> > > +             return acpi_mdiobus_register(bus, fwnode);
> > > +     else if (is_of_node(fwnode))
> > > +             return of_mdiobus_register(bus, to_of_node(fwnode));
> > > +     else
> > > +             return -EINVAL;
> >
> > I wounder if here you should call mdiobus_register(mdio), rather than
> > -EINVAL?
> >
> > I don't have a strong opinion.
>
> Currently (and in foreseeable future) we support only DT/ACPI as a
> firmware description, reaching the last "else" means something really
> wrong. The case of lack of DT/ACPI and the fallback is handled on the
> include/linux/fwnode_mdio.h level.
>
> >
> > Reviewed-by: Andrew Lunn <andrew@...n.ch>
> >

Unfortunately I have to withdraw this patch, as well as xgmac_mdio
one. In case the fwnode_/of_/acpi_mdio are built as modules, we get a
cycle dependency during depmod phase of modules_install, eg.:

depmod: ERROR: Cycle detected: fwnode_mdio -> of_mdio -> fwnode_mdio
depmod: ERROR: Found 2 modules in dependency cycles!

OR:

depmod: ERROR: Cycle detected: acpi_mdio -> fwnode_mdio -> acpi_mdio
depmod: ERROR: Found 2 modules in dependency cycles!

The proper solution would be to merge contents of
acpi_mdiobus_register and of_mdiobus_register inside the common
fwnode_mdiobus_register (so that the former would only call the
latter). However this change seems feasible, but I'd expect it to be a
patchset bigger than this one alone and deserves its own thorough
review and testing, as it would affect huge amount of current
of_mdiobus_register users.

Given above, for now I will resubmit this patchset in the shape as
proposed in v1, i.e. using the 'if' condition explicitly in mvmdio
driver.

Best regards,
Marcin