Message ID | 20200107050846.16838-1-f.fainelli@gmail.com |
---|---|
State | Rejected |
Delegated to: | David Miller |
Headers | show |
Series | [net] Revert "net: stmmac: platform: Fix MDIO init for platforms without PHY" | expand |
> From: Florian Fainelli <f.fainelli@gmail.com> > Subject: [PATCH net] Revert "net: stmmac: platform: Fix MDIO init for platforms > without PHY" > > This reverts commit d3e014ec7d5ebe9644b5486bc530b91e62bbf624 ("net: > stmmac: platform: Fix MDIO init for platforms without PHY") because it breaks > existing systems with stmmac which do not have a MDIO bus sub-node nor a > 'phy-handle' property declared in their Device Tree. On those systems, the > stmmac MDIO bus is expected to be created and then scanned by > of_mdiobus_register() to create PHY devices. > > While these systems should arguably make use of a more accurate Device Tree > reprensentation with the use of the MDIO bus sub-node an appropriate 'phy- > handle', we cannot break them, therefore restore the behavior prior to the said > commit. > > Fixes: d3e014ec7d5e ("net: stmmac: platform: Fix MDIO init for platforms > without PHY") > Reported-by: Heiko Stuebner <heiko@sntech.de> > Reported-by: kernelci.org bot <bot@kernelci.org> > Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> Nacked-by: Sriram Dash <Sriram.dash@samsung.com> > --- > Heiko, > > I did not add the Tested-by because the patch is a little bit different from what > you tested, even if you most likely were not hitting the other part that I was > changing. Thanks! > > drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > index cc8d7e7bf9ac..bedaff0c13bd 100644 > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > @@ -320,7 +320,7 @@ static int stmmac_mtl_setup(struct platform_device > *pdev, static int stmmac_dt_phy(struct plat_stmmacenet_data *plat, > struct device_node *np, struct device *dev) { > - bool mdio = false; > + bool mdio = true; This is breaking for the platforms with fixed-link. stih418-b2199.dts and 169445.dts to name a few. For the newer platforms, they should provide the mdio/ snps,dwmac-mdio property in the device tree as we are checking the mdio/ snps,dwmac-mdio property in the stmmac_platform driver for the mdio bus memory allocation. For existing platforms, I agree we should not break them, but we should make the code correct. And make the existing platforms adapt to the proper code. There is a proposed solution. https://lkml.org/lkml/2020/1/7/14 What do you think? > static const struct of_device_id need_mdio_ids[] = { > { .compatible = "snps,dwc-qos-ethernet-4.10" }, > {}, > -- > 2.19.1
On 07/01/2020 06:46, Sriram Dash wrote: >> From: Florian Fainelli <f.fainelli@gmail.com> >> Subject: [PATCH net] Revert "net: stmmac: platform: Fix MDIO init for > platforms >> without PHY" >> >> This reverts commit d3e014ec7d5ebe9644b5486bc530b91e62bbf624 ("net: >> stmmac: platform: Fix MDIO init for platforms without PHY") because it > breaks >> existing systems with stmmac which do not have a MDIO bus sub-node nor a >> 'phy-handle' property declared in their Device Tree. On those systems, the >> stmmac MDIO bus is expected to be created and then scanned by >> of_mdiobus_register() to create PHY devices. >> >> While these systems should arguably make use of a more accurate Device > Tree >> reprensentation with the use of the MDIO bus sub-node an appropriate 'phy- >> handle', we cannot break them, therefore restore the behavior prior to the > said >> commit. >> >> Fixes: d3e014ec7d5e ("net: stmmac: platform: Fix MDIO init for platforms >> without PHY") >> Reported-by: Heiko Stuebner <heiko@sntech.de> >> Reported-by: kernelci.org bot <bot@kernelci.org> >> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> > Nacked-by: Sriram Dash <Sriram.dash@samsung.com> > >> --- >> Heiko, >> >> I did not add the Tested-by because the patch is a little bit different > from what >> you tested, even if you most likely were not hitting the other part that I > was >> changing. Thanks! >> >> drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c >> b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c >> index cc8d7e7bf9ac..bedaff0c13bd 100644 >> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c >> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c >> @@ -320,7 +320,7 @@ static int stmmac_mtl_setup(struct platform_device >> *pdev, static int stmmac_dt_phy(struct plat_stmmacenet_data *plat, >> struct device_node *np, struct device *dev) { >> - bool mdio = false; >> + bool mdio = true; > > > This is breaking for the platforms with fixed-link. > stih418-b2199.dts and 169445.dts to name a few. > > For the newer platforms, they should provide the mdio/ snps,dwmac-mdio > property in the device tree as we are checking the mdio/ snps,dwmac-mdio > property in the stmmac_platform driver for the mdio bus memory allocation. > For existing platforms, I agree we should not break them, but we should make > the code correct. And make the existing platforms adapt to the proper code. > There is a proposed solution. > https://lkml.org/lkml/2020/1/7/14 Can you post it as a patch then and add me in CC since it also breaks Oxnas as reported as Kci ? Neil > > What do you think? > >> static const struct of_device_id need_mdio_ids[] = { >> { .compatible = "snps,dwc-qos-ethernet-4.10" }, >> {}, >> -- >> 2.19.1 > > > > _______________________________________________ > linux-arm-kernel mailing list > linux-arm-kernel@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-arm-kernel >
On 07/01/2020 5:46 am, Sriram Dash wrote: >> From: Florian Fainelli <f.fainelli@gmail.com> >> Subject: [PATCH net] Revert "net: stmmac: platform: Fix MDIO init for > platforms >> without PHY" >> >> This reverts commit d3e014ec7d5ebe9644b5486bc530b91e62bbf624 ("net: >> stmmac: platform: Fix MDIO init for platforms without PHY") because it > breaks >> existing systems with stmmac which do not have a MDIO bus sub-node nor a >> 'phy-handle' property declared in their Device Tree. On those systems, the >> stmmac MDIO bus is expected to be created and then scanned by >> of_mdiobus_register() to create PHY devices. >> >> While these systems should arguably make use of a more accurate Device > Tree >> reprensentation with the use of the MDIO bus sub-node an appropriate 'phy- >> handle', we cannot break them, therefore restore the behavior prior to the > said >> commit. >> >> Fixes: d3e014ec7d5e ("net: stmmac: platform: Fix MDIO init for platforms >> without PHY") >> Reported-by: Heiko Stuebner <heiko@sntech.de> >> Reported-by: kernelci.org bot <bot@kernelci.org> >> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> > Nacked-by: Sriram Dash <Sriram.dash@samsung.com> > >> --- >> Heiko, >> >> I did not add the Tested-by because the patch is a little bit different > from what >> you tested, even if you most likely were not hitting the other part that I > was >> changing. Thanks! >> >> drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c >> b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c >> index cc8d7e7bf9ac..bedaff0c13bd 100644 >> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c >> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c >> @@ -320,7 +320,7 @@ static int stmmac_mtl_setup(struct platform_device >> *pdev, static int stmmac_dt_phy(struct plat_stmmacenet_data *plat, >> struct device_node *np, struct device *dev) { >> - bool mdio = false; >> + bool mdio = true; > > > This is breaking for the platforms with fixed-link. > stih418-b2199.dts and 169445.dts to name a few. > > For the newer platforms, they should provide the mdio/ snps,dwmac-mdio > property in the device tree as we are checking the mdio/ snps,dwmac-mdio > property in the stmmac_platform driver for the mdio bus memory allocation. > For existing platforms, I agree we should not break them, but we should make > the code correct. And make the existing platforms adapt to the proper code. > There is a proposed solution. > https://lkml.org/lkml/2020/1/7/14 > > What do you think? The binding says that the phy handle and mdio child node are optional, so "update all of the DTBs!" is not a viable solution. I'm far from an expert here, but AFAICS the fault of the current code is that it assumes the lack of a phy handle implies a fixed link, so the obvious answer is to actually check whether the "fixed-link" property is present. Robin. > >> static const struct of_device_id need_mdio_ids[] = { >> { .compatible = "snps,dwc-qos-ethernet-4.10" }, >> {}, >> -- >> 2.19.1 > > > > _______________________________________________ > linux-arm-kernel mailing list > linux-arm-kernel@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-arm-kernel >
> From: Robin Murphy <robin.murphy@arm.com> > Subject: Re: [PATCH net] Revert "net: stmmac: platform: Fix MDIO init for > platforms without PHY" > > On 07/01/2020 5:46 am, Sriram Dash wrote: > >> From: Florian Fainelli <f.fainelli@gmail.com> > >> Subject: [PATCH net] Revert "net: stmmac: platform: Fix MDIO init for > > platforms > >> without PHY" > >> > >> This reverts commit d3e014ec7d5ebe9644b5486bc530b91e62bbf624 ("net: > >> stmmac: platform: Fix MDIO init for platforms without PHY") because > >> it > > breaks > >> existing systems with stmmac which do not have a MDIO bus sub-node > >> nor a 'phy-handle' property declared in their Device Tree. On those > >> systems, the stmmac MDIO bus is expected to be created and then > >> scanned by > >> of_mdiobus_register() to create PHY devices. > >> > >> While these systems should arguably make use of a more accurate > >> Device > > Tree > >> reprensentation with the use of the MDIO bus sub-node an appropriate > >> 'phy- handle', we cannot break them, therefore restore the behavior > >> prior to the > > said > >> commit. > >> > >> Fixes: d3e014ec7d5e ("net: stmmac: platform: Fix MDIO init for > >> platforms without PHY") > >> Reported-by: Heiko Stuebner <heiko@sntech.de> > >> Reported-by: kernelci.org bot <bot@kernelci.org> > >> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> > > Nacked-by: Sriram Dash <Sriram.dash@samsung.com> > > > >> --- > >> Heiko, > >> > >> I did not add the Tested-by because the patch is a little bit > >> different > > from what > >> you tested, even if you most likely were not hitting the other part > >> that I > > was > >> changing. Thanks! > >> > >> drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c | 2 +- > >> 1 file changed, 1 insertion(+), 1 deletion(-) > >> > >> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > >> b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > >> index cc8d7e7bf9ac..bedaff0c13bd 100644 > >> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > >> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > >> @@ -320,7 +320,7 @@ static int stmmac_mtl_setup(struct > >> platform_device *pdev, static int stmmac_dt_phy(struct > plat_stmmacenet_data *plat, > >> struct device_node *np, struct device *dev) { > >> - bool mdio = false; > >> + bool mdio = true; > > > > > > This is breaking for the platforms with fixed-link. > > stih418-b2199.dts and 169445.dts to name a few. > > > > For the newer platforms, they should provide the mdio/ snps,dwmac-mdio > > property in the device tree as we are checking the mdio/ > > snps,dwmac-mdio property in the stmmac_platform driver for the mdio bus > memory allocation. > > For existing platforms, I agree we should not break them, but we > > should make the code correct. And make the existing platforms adapt to the > proper code. > > There is a proposed solution. > > https://protect2.fireeye.com/url?k=d075c0fc-8da69942-d0744bb3-0cc47a31 > > ba82-60be9f5a0fb38a27&u=https://lkml.org/lkml/2020/1/7/14 > > > > What do you think? > > The binding says that the phy handle and mdio child node are optional, so > "update all of the DTBs!" is not a viable solution. I'm far from an expert here, but I get your point Robin. Because the mdio child or snps,dwmac-mdio is not mandatory, there has been this problem. > AFAICS the fault of the current code is that it assumes the lack of a phy handle > implies a fixed link, so the obvious answer is to actually check whether the Kind of. > "fixed-link" property is present. I agree to have a fixed link check. Possibly this can be done: --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c @@ -320,7 +320,9 @@ static int stmmac_mtl_setup(struct platform_device *pdev, static int stmmac_dt_phy(struct plat_stmmacenet_data *plat, struct device_node *np, struct device *dev) { - bool mdio = false; + bool mdio = true; + struct device_node *fixed_link; + static const struct of_device_id need_mdio_ids[] = { { .compatible = "snps,dwc-qos-ethernet-4.10" }, {}, @@ -340,9 +342,8 @@ static int stmmac_dt_phy(struct plat_stmmacenet_data *plat, } } - if (plat->mdio_node) { - dev_dbg(dev, "Found MDIO subnode\n"); - mdio = true; + if (of_get_child_by_name(np, "fixed-link")) { + mdio = false; } if (mdio) { Neil and Heiko, Can you guys please test this on your platforms? We can post a more cleaner version of the patch if all agree to it. > > Robin. > > > > >> static const struct of_device_id need_mdio_ids[] = { > >> { .compatible = "snps,dwc-qos-ethernet-4.10" }, > >> {}, > >> -- > >> 2.19.1 > > > > > > > > _______________________________________________ > > linux-arm-kernel mailing list > > linux-arm-kernel@lists.infradead.org > > https://protect2.fireeye.com/url?k=2cb95551-716a0cef-2cb8de1e-0cc47a31 > > ba82-d01ceb62a8a93fa2&u=http://lists.infradead.org/mailman/listinfo/li > > nux-arm-kernel > >
On 07/01/2020 12:14 pm, Sriram Dash wrote: >> From: Robin Murphy <robin.murphy@arm.com> >> Subject: Re: [PATCH net] Revert "net: stmmac: platform: Fix MDIO init for >> platforms without PHY" >> >> On 07/01/2020 5:46 am, Sriram Dash wrote: >>>> From: Florian Fainelli <f.fainelli@gmail.com> >>>> Subject: [PATCH net] Revert "net: stmmac: platform: Fix MDIO init for >>> platforms >>>> without PHY" >>>> >>>> This reverts commit d3e014ec7d5ebe9644b5486bc530b91e62bbf624 ("net: >>>> stmmac: platform: Fix MDIO init for platforms without PHY") because >>>> it >>> breaks >>>> existing systems with stmmac which do not have a MDIO bus sub-node >>>> nor a 'phy-handle' property declared in their Device Tree. On those >>>> systems, the stmmac MDIO bus is expected to be created and then >>>> scanned by >>>> of_mdiobus_register() to create PHY devices. >>>> >>>> While these systems should arguably make use of a more accurate >>>> Device >>> Tree >>>> reprensentation with the use of the MDIO bus sub-node an appropriate >>>> 'phy- handle', we cannot break them, therefore restore the behavior >>>> prior to the >>> said >>>> commit. >>>> >>>> Fixes: d3e014ec7d5e ("net: stmmac: platform: Fix MDIO init for >>>> platforms without PHY") >>>> Reported-by: Heiko Stuebner <heiko@sntech.de> >>>> Reported-by: kernelci.org bot <bot@kernelci.org> >>>> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> >>> Nacked-by: Sriram Dash <Sriram.dash@samsung.com> >>> >>>> --- >>>> Heiko, >>>> >>>> I did not add the Tested-by because the patch is a little bit >>>> different >>> from what >>>> you tested, even if you most likely were not hitting the other part >>>> that I >>> was >>>> changing. Thanks! >>>> >>>> drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c | 2 +- >>>> 1 file changed, 1 insertion(+), 1 deletion(-) >>>> >>>> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c >>>> b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c >>>> index cc8d7e7bf9ac..bedaff0c13bd 100644 >>>> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c >>>> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c >>>> @@ -320,7 +320,7 @@ static int stmmac_mtl_setup(struct >>>> platform_device *pdev, static int stmmac_dt_phy(struct >> plat_stmmacenet_data *plat, >>>> struct device_node *np, struct device *dev) { >>>> - bool mdio = false; >>>> + bool mdio = true; >>> >>> >>> This is breaking for the platforms with fixed-link. >>> stih418-b2199.dts and 169445.dts to name a few. >>> >>> For the newer platforms, they should provide the mdio/ snps,dwmac-mdio >>> property in the device tree as we are checking the mdio/ >>> snps,dwmac-mdio property in the stmmac_platform driver for the mdio bus >> memory allocation. >>> For existing platforms, I agree we should not break them, but we >>> should make the code correct. And make the existing platforms adapt to the >> proper code. >>> There is a proposed solution. >>> https://protect2.fireeye.com/url?k=d075c0fc-8da69942-d0744bb3-0cc47a31 >>> ba82-60be9f5a0fb38a27&u=https://lkml.org/lkml/2020/1/7/14 >>> >>> What do you think? >> >> The binding says that the phy handle and mdio child node are optional, so >> "update all of the DTBs!" is not a viable solution. I'm far from an expert here, but > > I get your point Robin. Because the mdio child or snps,dwmac-mdio is not mandatory, > there has been this problem. > >> AFAICS the fault of the current code is that it assumes the lack of a phy handle >> implies a fixed link, so the obvious answer is to actually check whether the > > Kind of. > >> "fixed-link" property is present. > > > I agree to have a fixed link check. Possibly this can be done: FWIW, given the 4th case in the kerneldoc table I think this check might need to come first. On the upside, there appears to be a handy helper already, so it may just be as simple as: bool mdio = !of_phy_is_fixed_link(np); Robin. > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > @@ -320,7 +320,9 @@ static int stmmac_mtl_setup(struct platform_device *pdev, > static int stmmac_dt_phy(struct plat_stmmacenet_data *plat, > struct device_node *np, struct device *dev) > { > - bool mdio = false; > + bool mdio = true; > + struct device_node *fixed_link; > + > static const struct of_device_id need_mdio_ids[] = { > { .compatible = "snps,dwc-qos-ethernet-4.10" }, > {}, > @@ -340,9 +342,8 @@ static int stmmac_dt_phy(struct plat_stmmacenet_data *plat, > } > } > > - if (plat->mdio_node) { > - dev_dbg(dev, "Found MDIO subnode\n"); > - mdio = true; > + if (of_get_child_by_name(np, "fixed-link")) { > + mdio = false; > } > > if (mdio) { > > Neil and Heiko, > > Can you guys please test this on your platforms? > We can post a more cleaner version of the patch if all agree to it. > >> >> Robin. >> >>> >>>> static const struct of_device_id need_mdio_ids[] = { >>>> { .compatible = "snps,dwc-qos-ethernet-4.10" }, >>>> {}, >>>> -- >>>> 2.19.1 >>> >>> >>> >>> _______________________________________________ >>> linux-arm-kernel mailing list >>> linux-arm-kernel@lists.infradead.org >>> https://protect2.fireeye.com/url?k=2cb95551-716a0cef-2cb8de1e-0cc47a31 >>> ba82-d01ceb62a8a93fa2&u=http://lists.infradead.org/mailman/listinfo/li >>> nux-arm-kernel >>> >
From: Sriram Dash <sriram.dash@samsung.com> Date: Jan/07/2020, 12:14:19 (UTC+00:00) > Can you guys please test this on your platforms? > We can post a more cleaner version of the patch if all agree to it. Can you also test this one [1] ? [1] https://patchwork.ozlabs.org/patch/1218762/ --- Thanks, Jose Miguel Abreu
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c index cc8d7e7bf9ac..bedaff0c13bd 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c @@ -320,7 +320,7 @@ static int stmmac_mtl_setup(struct platform_device *pdev, static int stmmac_dt_phy(struct plat_stmmacenet_data *plat, struct device_node *np, struct device *dev) { - bool mdio = false; + bool mdio = true; static const struct of_device_id need_mdio_ids[] = { { .compatible = "snps,dwc-qos-ethernet-4.10" }, {},
This reverts commit d3e014ec7d5ebe9644b5486bc530b91e62bbf624 ("net: stmmac: platform: Fix MDIO init for platforms without PHY") because it breaks existing systems with stmmac which do not have a MDIO bus sub-node nor a 'phy-handle' property declared in their Device Tree. On those systems, the stmmac MDIO bus is expected to be created and then scanned by of_mdiobus_register() to create PHY devices. While these systems should arguably make use of a more accurate Device Tree reprensentation with the use of the MDIO bus sub-node an appropriate 'phy-handle', we cannot break them, therefore restore the behavior prior to the said commit. Fixes: d3e014ec7d5e ("net: stmmac: platform: Fix MDIO init for platforms without PHY") Reported-by: Heiko Stuebner <heiko@sntech.de> Reported-by: kernelci.org bot <bot@kernelci.org> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> --- Heiko, I did not add the Tested-by because the patch is a little bit different from what you tested, even if you most likely were not hitting the other part that I was changing. Thanks! drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)