Hi Geert-san,
> From: Geert Uytterhoeven, Sent: Friday, October 13, 2023 4:29 PM
>
> Hi Shimoda-san,
>
> On Fri, Oct 13, 2023 at 4:10 AM Yoshihiro Shimoda
> <yoshihiro.shimoda.uh@xxxxxxxxxxx> wrote:
> > > From: Geert Uytterhoeven, Sent: Thursday, October 12, 2023 9:35 PM
> > > On Thu, Oct 12, 2023 at 2:16 PM Yoshihiro Shimoda
> > > <yoshihiro.shimoda.uh@xxxxxxxxxxx> wrote:
> > > > Add PM ops for Suspend to Idle. When the system suspended,
> > > > the Ethernet Serdes's clock will be stopped. So, this driver needs
> > > > to re-initialize the Ethernet Serdes by phy_init() in
> > > > renesas_eth_sw_resume(). Otherwise, timeout happened in phy_power_on().
> > > >
> > > > Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@xxxxxxxxxxx>
> > >
> > > Thanks for your patch!
> >
> > Thank you for your review!
> >
> > > > --- a/drivers/net/ethernet/renesas/rswitch.c
> > > > +++ b/drivers/net/ethernet/renesas/rswitch.c
> > > > @@ -17,6 +17,7 @@
> > > > #include <linux/of_net.h>
> > > > #include <linux/phy/phy.h>
> > > > #include <linux/platform_device.h>
> > > > +#include <linux/pm.h>
> > > > #include <linux/pm_runtime.h>
> > > > #include <linux/rtnetlink.h>
> > > > #include <linux/slab.h>
> > > > @@ -1315,6 +1316,7 @@ static int rswitch_phy_device_init(struct rswitch_device *rdev)
> > > > if (!phydev)
> > > > goto out;
> > > > __set_bit(rdev->etha->phy_interface, phydev->host_interfaces);
> > > > + phydev->mac_managed_pm = true;
> > > >
> > > > phydev = of_phy_connect(rdev->ndev, phy, rswitch_adjust_link, 0,
> > > > rdev->etha->phy_interface);
> > > > @@ -1991,11 +1993,52 @@ static void renesas_eth_sw_remove(struct platform_device *pdev)
> > > > platform_set_drvdata(pdev, NULL);
> > > > }
> > > >
> > > > +static int __maybe_unused renesas_eth_sw_suspend(struct device *dev)
> > > > +{
> > > > + struct rswitch_private *priv = dev_get_drvdata(dev);
> > > > + struct net_device *ndev;
> > > > + int i;
> > >
> > > unsigned int (also below)
> >
> > I don't know why unsigned int is needed. Other functions use
>
> There's this old mantra "prefer unsigned over signed in C",
> and a valid port array index here is always unsigned.
I understood it.
> > rswitch_for_each_enabled_port{_continue_reverse}() with int.
> > Especially, rswitch_for_each_enabled_port_continue_reverse()
> > has the following code, unsigned int will not work correctly:
>
> Oh, there is also a reverse variant, which indeed needs a signed
> iterator, currently...
>
> > ---
> > #define rswitch_for_each_enabled_port_continue_reverse(priv, i) \
> > for (i--; i >= 0; i--) \
>
> I think this can be made to work with an unsigned iterator using
>
> for (; i-- > 0; )
I think that this loop cannot access index 0 :)
> > if (priv->rdev[i]->disabled) \
> > continue; \
> > else
> > ---
> >
> > So, I would like to keep this for consistency with other functions'
> > implementation. But, what do you think?
>
> Consistency is good...
> Surprising readers (why is this signed?) is bad...
> It's hard to keep a good balance...
>
> BTW, perhaps it would make sense to use the reverse order in suspend?
> Although it probably doesn't matter, as rswitch_deinit() uses the
> non-reverse order, too.
Thank you for your suggestion. I also think that we can use the
non-reverse order. So,
1) Change "_reverse()" macro to others somehow.
2) Change the type of i for "for_each" to unsigned int.
3) Add PM ops.
Perhaps, can/should we merge the 1) and 2) to one patch?
Best regards,
Yoshihiro Shimoda
> Gr{oetje,eeting}s,
>
> Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
>
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
> -- Linus Torvalds
