net: improve devres helpers
Subsystems (10)
v2 cover net: improve devres helpers
From: Bartosz Golaszewski <[email protected]>
So it seems like there's no support for relaxing certain networking devres helpers to not require previously allocated structures to also be managed. However the way mdio devres variants are implemented is still wrong and I modified my series to address it while keeping the functions strict.
First two patches modify the ixgbe driver to get rid of the last user of devm_mdiobus_free().
Patches 3, 4, 5 and 6 are mostly cosmetic.
Patch 7 fixes the way devm_mdiobus_register() is implemented.
Patches 8 & 9 provide a managed variant of of_mdiobus_register() and last patch uses it in mtk-star-emac driver.
v1 -> v2:
- drop the patch relaxing devm_register_netdev()
- require struct mii_bus to be managed in devm_mdiobus_register() and devm_of_mdiobus_register() but don't store that information in the structure itself: use devres_find() instead
Bartosz Golaszewski (10):
net: ethernet: ixgbe: check the return value of ixgbe_mii_bus_init()
net: ethernet: ixgbe: don't call devm_mdiobus_free()
net: devres: rename the release callback of devm_register_netdev()
Documentation: devres: add missing mdio helper
phy: un-inline devm_mdiobus_register()
phy: mdio: add kerneldoc for __devm_mdiobus_register()
net: phy: don't abuse devres in devm_mdiobus_register()
of: mdio: remove the 'extern' keyword from function declarations
of: mdio: provide devm_of_mdiobus_register()
net: ethernet: mtk-star-emac: use devm_of_mdiobus_register()
.../driver-api/driver-model/devres.rst | 3 +- drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 6 +- drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c | 14 +- drivers/net/ethernet/mediatek/mtk_star_emac.c | 13 +- drivers/net/ethernet/realtek/r8169_main.c | 2 +- drivers/net/phy/Makefile | 2 + drivers/net/phy/mdio_bus.c | 73 ---------- drivers/net/phy/mdio_devres.c | 133 ++++++++++++++++++ include/linux/of_mdio.h | 40 +++--- include/linux/phy.h | 21 +-- net/devres.c | 4 +- 11 files changed, 174 insertions(+), 137 deletions(-) create mode 100644 drivers/net/phy/mdio_devres.c
Signature · 7 lines
2.26.1 _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
PATCH v2 1/10 net: ethernet: ixgbe: check the return value of ixgbe_mii_bus_init()
From: Bartosz Golaszewski <[email protected]>
This function may fail. Check its return value and propagate the error code.
--- drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-)
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c +5 -1
@@ -11175,10 +11175,14 @@ static int ixgbe_probe(struct pci_dev *pdev, const struct pci_device_id *ent)IXGBE_LINK_SPEED_10GB_FULL |IXGBE_LINK_SPEED_1GB_FULL , true); -ixgbe_mii_bus_init (hw); + err =ixgbe_mii_bus_init (hw); +if (err) + goto err_netdev;return 0 ; +err_netdev: +unregister_netdev (netdev); err_register:ixgbe_release_hw_control (adapter);ixgbe_clear_interrupt_scheme (adapter);
Signature · 7 lines
2.26.1 _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
Re: [PATCH v2 01/10] net: ethernet: ixgbe: check the return value of ixgbe_mii_bus_init()
Quoted 6 lines · Bartosz Golaszewski
From: Bartosz Golaszewski <[email protected]>
This function may fail. Check its return value and propagate the error code.
Signed-off-by Bartosz Golaszewski <[email protected]>
Signature · 6 lines
Florian _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
RE: [PATCH v2 01/10] net: ethernet: ixgbe: check the return value of ixgbe_mii_bus_init()
In reply to Bartosz GolaszewskiQuoted 18 lines
-----Original Message-----
From: Bartosz Golaszewski <[email protected]>
Sent: Monday, June 29, 2020 05:04
To: Kirsher, Jeffrey T <[email protected]>; David S . Miller <[email protected]>; Jakub Kicinski <[email protected]>; John Crispin <[email protected]>; Sean Wang <[email protected]>; Mark Lee <[email protected]>; Matthias Brugger <[email protected]>; Heiner Kallweit <[email protected]>; Andrew Lunn <[email protected]>; Florian Fainelli <[email protected]>; Russell King <[email protected]>; Rob Herring <[email protected]>; Frank Rowand <[email protected]>
Cc: [email protected]; [email protected]; linux-arm- [email protected]; [email protected]; [email protected]; Bartosz Golaszewski <[email protected]>
Subject: [PATCH v2 01/10] net: ethernet: ixgbe: check the return value ofixgbe_mii_bus_init()
Quoted 5 lines
From: Bartosz Golaszewski <[email protected]>
This function may fail. Check its return value and propagate the error code.
Signed-off-by Bartosz Golaszewski <[email protected]>
Quoted 3 lines
--- drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-)
_______________________________________________
Linux-mediatek mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-mediatek
PATCH v2 2/10 net: ethernet: ixgbe: don't call devm_mdiobus_free()
In reply to Bartosz GolaszewskiFrom: Bartosz Golaszewski <[email protected]>
The idea behind devres is that the release callbacks are called if probe fails. As we now check the return value of ixgbe_mii_bus_init(), we can drop the call devm_mdiobus_free() in error path as the release callback will be called automatically.
--- drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c | 14 +++----------- 1 file changed, 3 insertions(+), 11 deletions(-)
drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c +3 -11
@@ -905,7 +905,6 @@ s32 ixgbe_mii_bus_init(struct ixgbe_hw *hw)struct pci_dev *pdev = adapter->pdev;struct device *dev = &adapter->netdev->dev;struct mii_bus *bus; -int err = -ENODEV ; bus =devm_mdiobus_alloc (dev);if (!bus)
@@ -923,7 +922,7 @@ s32 ixgbe_mii_bus_init(struct ixgbe_hw *hw)case IXGBE_DEV_ID_X550EM_A_1G_T :case IXGBE_DEV_ID_X550EM_A_1G_T_L :if (!ixgbe_x550em_a_has_mii (hw)) - goto ixgbe_no_mii_bus; +return -ENODEV ; bus->read = &ixgbe_x550em_a_mii_bus_read; bus->write = &ixgbe_x550em_a_mii_bus_write;break ;
@@ -948,15 +947,8 @@ s32 ixgbe_mii_bus_init(struct ixgbe_hw *hw)*/ hw->phy.mdio.mode_support =MDIO_SUPPORTS_C45 |MDIO_SUPPORTS_C22 ; - err =mdiobus_register (bus); -if (!err) { - adapter->mii_bus = bus; -return 0 ; - } - -ixgbe_no_mii_bus: -devm_mdiobus_free (dev, bus); -return err; + adapter->mii_bus = bus; +return mdiobus_register (bus); }/**
Signature · 7 lines
2.26.1 _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
Re: [PATCH v2 02/10] net: ethernet: ixgbe: don't call devm_mdiobus_free()
Quoted 8 lines · Bartosz Golaszewski
From: Bartosz Golaszewski <[email protected]>
The idea behind devres is that the release callbacks are called if probe fails. As we now check the return value of
ixgbe_mii_bus_init(), we can drop the calldevm_mdiobus_free()in error path as the release callback will be called automatically.Signed-off-by Bartosz Golaszewski <[email protected]>
Signature · 6 lines
Florian _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
RE: [PATCH v2 02/10] net: ethernet: ixgbe: don't call devm_mdiobus_free()
In reply to Bartosz GolaszewskiQuoted 18 lines
-----Original Message-----
From: Bartosz Golaszewski <[email protected]>
Sent: Monday, June 29, 2020 05:04
To: Kirsher, Jeffrey T <[email protected]>; David S . Miller <[email protected]>; Jakub Kicinski <[email protected]>; John Crispin <[email protected]>; Sean Wang <[email protected]>; Mark Lee <[email protected]>; Matthias Brugger <[email protected]>; Heiner Kallweit <[email protected]>; Andrew Lunn <[email protected]>; Florian Fainelli <[email protected]>; Russell King <[email protected]>; Rob Herring <[email protected]>; Frank Rowand <[email protected]>
Cc: [email protected]; [email protected]; linux-arm- [email protected]; [email protected]; [email protected]; Bartosz Golaszewski <[email protected]>
Subject: [PATCH v2 02/10] net: ethernet: ixgbe: don't calldevm_mdiobus_free()
Quoted 8 lines
From: Bartosz Golaszewski <[email protected]>
The idea behind devres is that the release callbacks are called if probe fails. As we now check the return value of
ixgbe_mii_bus_init(), we can drop the calldevm_mdiobus_free()in error path as the release callback will be called automatically.Signed-off-by Bartosz Golaszewski <[email protected]>
Quoted 3 lines
--- drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c | 14 +++----------- 1 file changed, 3 insertions(+), 11 deletions(-)
_______________________________________________
Linux-mediatek mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-mediatek
PATCH v2 3/10 net: devres: rename the release callback of devm_register_netdev()
In reply to Bartosz GolaszewskiFrom: Bartosz Golaszewski <[email protected]>
Make it an explicit counterpart to devm_register_netdev() just like we do with devm_free_netdev() for better clarity.
--- net/devres.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
net/devres.c +2 -2
@@ -39,7 +39,7 @@ struct net_device *devm_alloc_etherdev_mqs(struct device *dev, int sizeof_priv, }EXPORT_SYMBOL (devm_alloc_etherdev_mqs); -static void devm_netdev_release (struct device *dev,void *this) +static void devm_unregister_netdev (struct device *dev,void *this) {struct net_device_devres *res = this;
@@ -77,7 +77,7 @@ int devm_register_netdev(struct device *dev, struct net_device *ndev) netdev_devres_match, ndev)))return -EINVAL ; - dr =devres_alloc (devm_netdev_release,sizeof (*dr),GFP_KERNEL ); + dr =devres_alloc (devm_unregister_netdev,sizeof (*dr),GFP_KERNEL );if (!dr)return -ENOMEM ;
Signature · 7 lines
2.26.1 _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
Re: [PATCH v2 03/10] net: devres: rename the release callback of devm_register_netdev()
Quoted 6 lines · Bartosz Golaszewski
From: Bartosz Golaszewski <[email protected]>
Make it an explicit counterpart to
devm_register_netdev()just like we do withdevm_free_netdev()for better clarity.Signed-off-by Bartosz Golaszewski <[email protected]>
Signature · 6 lines
Florian _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
PATCH v2 4/10 Documentation: devres: add missing mdio helper
In reply to Bartosz GolaszewskiFrom: Bartosz Golaszewski <[email protected]>
We have a devres variant of mdiobus_register() but it's not listed in devres.rst. Add it under other mdio devm functions.
--- Documentation/driver-api/driver-model/devres.rst | 1 + 1 file changed, 1 insertion(+)
Documentation/driver-api/driver-model/devres.rst +1 -0
@@ -343,6 +343,7 @@ MDIO devm_mdiobus_alloc() devm_mdiobus_alloc_size() devm_mdiobus_free() + devm_mdiobus_register() MEM devm_free_pages()
Signature · 7 lines
2.26.1 _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
Re: [PATCH v2 04/10] Documentation: devres: add missing mdio helper
Quoted 6 lines · Bartosz Golaszewski
From: Bartosz Golaszewski <[email protected]>
We have a devres variant of
mdiobus_register()but it's not listed indevres.rst. Add it under other mdio devm functions.Signed-off-by Bartosz Golaszewski <[email protected]>
Signature · 6 lines
Florian _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
PATCH v2 5/10 phy: un-inline devm_mdiobus_register()
In reply to Bartosz GolaszewskiFrom: Bartosz Golaszewski <[email protected]>
Functions should only be static inline if they're very short. This devres helper is already over 10 lines and it will grow soon as we'll be improving upon its approach. Pull it into mdio_devres.c.
--- drivers/net/phy/Makefile | 2 +- drivers/net/phy/mdio_devres.c | 18 ++++++++++++++++++ include/linux/phy.h | 15 ++------------- 3 files changed, 21 insertions(+), 14 deletions(-) create mode 100644 drivers/net/phy/mdio_devres.c
drivers/net/phy/Makefile +1 -1
@@ -3,7 +3,7 @@ libphy-y := phy.o phy-c45.o phy-core.o phy_device.o \ linkmode.o -mdio-bus-y += mdio_bus.o mdio_device.o +mdio-bus-y += mdio_bus.o mdio_device.o mdio_devres.o ifdef CONFIG_MDIO_DEVICE obj-y += mdio-boardinfo.o
drivers/net/phy/mdio_devres.c new file +18 -0
@@ -0,0 +1,18 @@ +// SPDX-License-Identifier: GPL-2.0-or-later + +#include <linux/phy.h> + +int __devm_mdiobus_register (struct mii_bus *bus,struct module *owner) +{ +int ret; + +if (!bus->is_managed) +return -EPERM ; + + ret =__mdiobus_register (bus, owner); +if (!ret) + bus->is_managed_registered =1 ; + +return ret; +} +EXPORT_SYMBOL (__devm_mdiobus_register);
include/linux/phy.h +2 -13
@@ -313,20 +313,9 @@ static inline struct mii_bus *mdiobus_alloc(void) }int __mdiobus_register (struct mii_bus *bus,struct module *owner); +int __devm_mdiobus_register (struct mii_bus *bus,struct module *owner);#define mdiobus_register (bus) __mdiobus_register(bus, THIS_MODULE) -static inline int devm_mdiobus_register (struct mii_bus *bus) -{ -int ret; - -if (!bus->is_managed) -return -EPERM ; - - ret =mdiobus_register (bus); -if (!ret) - bus->is_managed_registered =1 ; - -return ret; -} +#define devm_mdiobus_register (bus) __devm_mdiobus_register(bus, THIS_MODULE)void mdiobus_unregister (struct mii_bus *bus);void mdiobus_free (struct mii_bus *bus);
Signature · 7 lines
2.26.1 _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
Re: [PATCH v2 05/10] phy: un-inline devm_mdiobus_register()
Quoted 7 lines · Bartosz Golaszewski
From: Bartosz Golaszewski <[email protected]>
Functions should only be static inline if they're very short. This devres helper is already over 10 lines and it will grow soon as we'll be improving upon its approach. Pull it into
mdio_devres.c.Signed-off-by Bartosz Golaszewski <[email protected]>
Signature · 6 lines
Florian _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
PATCH v2 6/10 phy: mdio: add kerneldoc for __devm_mdiobus_register()
In reply to Bartosz GolaszewskiFrom: Bartosz Golaszewski <[email protected]>
This function is not documented. Add a short kerneldoc description.
--- drivers/net/phy/mdio_devres.c | 7 +++++++ 1 file changed, 7 insertions(+)
drivers/net/phy/mdio_devres.c +7 -0
@@ -2,6 +2,13 @@#include <linux/phy.h> +/** +* __devm_mdiobus_register - Resource-managed variant of mdiobus_register() +* @bus: MII bus structure to register +* @owner: Owning module +* +* Returns 0 on success, negative error number on failure. +*/ int __devm_mdiobus_register (struct mii_bus *bus,struct module *owner) {int ret;
Signature · 7 lines
2.26.1 _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
Re: [PATCH v2 06/10] phy: mdio: add kerneldoc for __devm_mdiobus_register()
Quoted 5 lines · Bartosz Golaszewski
From: Bartosz Golaszewski <[email protected]>
This function is not documented. Add a short kerneldoc description.
Signed-off-by Bartosz Golaszewski <[email protected]>
Signature · 6 lines
Florian _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
PATCH v2 7/10 net: phy: don't abuse devres in devm_mdiobus_register()
In reply to Bartosz GolaszewskiFrom: Bartosz Golaszewski <[email protected]>
We currently have two managed helpers for mdiobus - devm_mdiobus_alloc() and devm_mdiobus_register(). The idea behind devres is that the release callback releases whatever resource the devm function allocates. In the mdiobus case however there's no devres associated with the device by devm_mdiobus_register(). Instead the release callback for devm_mdiobus_alloc(): _devm_mdiobus_free() unregisters the device if it is marked as managed.
This all seems wrong. The managed structure shouldn't need to know or care about whether it's managed or not - and this is the case now for struct mii_bus. The devres wrapper should be opaque to the managed resource.
This changeset makes devm_mdiobus_alloc() and devm_mdiobus_register() conform to common devres standards: devm_mdiobus_alloc() allocates a devres structure and registers a callback that will call mdiobus_free(). __devm_mdiobus_register() allocated another devres and registers a callback that will unregister the bus.
--- .../driver-api/driver-model/devres.rst | 1 - drivers/net/ethernet/realtek/r8169_main.c | 2 +- drivers/net/phy/mdio_bus.c | 73 ---------------- drivers/net/phy/mdio_devres.c | 83 +++++++++++++++++-- include/linux/phy.h | 10 +-- 5 files changed, 82 insertions(+), 87 deletions(-)
Documentation/driver-api/driver-model/devres.rst +0 -1
@@ -342,7 +342,6 @@ LED MDIO devm_mdiobus_alloc() devm_mdiobus_alloc_size() - devm_mdiobus_free() devm_mdiobus_register() MEM
drivers/net/ethernet/realtek/r8169_main.c +1 -1
@@ -5103,7 +5103,7 @@ static int r8169_mdio_register(struct rtl8169_private *tp) new_bus->read = r8169_mdio_read_reg; new_bus->write = r8169_mdio_write_reg; - ret =devm_mdiobus_register (new_bus); + ret =devm_mdiobus_register (&pdev->dev, new_bus);if (ret)return ret;
drivers/net/phy/mdio_bus.c +0 -73
79 lines not on this page · raw message
drivers/net/phy/mdio_devres.c +77 -6
102 lines not on this page · raw message
include/linux/phy.h +4 -6
@@ -260,9 +260,6 @@ struct mii_bus {int (*reset)(struct mii_bus *bus);struct mdio_bus_stats stats[PHY_MAX_ADDR ]; - unsigned int is_managed:1; /* is device-managed */ - unsigned int is_managed_registered:1; -/* * A lock to ensure that only one thing can read/write * the MDIO bus at a time
@@ -313,9 +310,11 @@ static inline struct mii_bus *mdiobus_alloc(void) }int __mdiobus_register (struct mii_bus *bus,struct module *owner); -int __devm_mdiobus_register (struct mii_bus *bus,struct module *owner); +int __devm_mdiobus_register (struct device *dev,struct mii_bus *bus, +struct module *owner);#define mdiobus_register (bus) __mdiobus_register(bus, THIS_MODULE) -#define devm_mdiobus_register (bus) __devm_mdiobus_register(bus, THIS_MODULE) +#define devm_mdiobus_register (dev, bus) \ + __devm_mdiobus_register(dev, bus, THIS_MODULE)void mdiobus_unregister (struct mii_bus *bus);void mdiobus_free (struct mii_bus *bus);
@@ -326,7 +325,6 @@ static inline struct mii_bus *devm_mdiobus_alloc(struct device *dev) }struct mii_bus *mdio_find_bus (const char *mdio_name); -void devm_mdiobus_free (struct device *dev,struct mii_bus *bus);struct phy_device *mdiobus_scan (struct mii_bus *bus,int addr);#define PHY_INTERRUPT_DISABLED false
Signature · 7 lines
2.26.1 _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
PATCH v2 8/10 of: mdio: remove the 'extern' keyword from function declarations
In reply to Bartosz GolaszewskiFrom: Bartosz Golaszewski <[email protected]>
The 'extern' keyword in headers doesn't have any benefit. Remove them all from the of_mdio.h header.
--- include/linux/of_mdio.h | 37 ++++++++++++++++++------------------- 1 file changed, 18 insertions(+), 19 deletions(-)
include/linux/of_mdio.h +18 -19
@@ -12,27 +12,26 @@#include <linux/of.h> #if IS_ENABLED (CONFIG_OF_MDIO ) -extern bool of_mdiobus_child_is_phy (struct device_node *child); -extern int of_mdiobus_register (struct mii_bus *mdio,struct device_node *np); -extern struct phy_device *of_phy_find_device (struct device_node *phy_np); -extern struct phy_device *of_phy_connect (struct net_device *dev, -struct device_node *phy_np, -void (*hndlr)(struct net_device *), -u32 flags,phy_interface_t iface); -extern struct phy_device * +bool of_mdiobus_child_is_phy (struct device_node *child); +int of_mdiobus_register (struct mii_bus *mdio,struct device_node *np); +struct phy_device *of_phy_find_device (struct device_node *phy_np); +struct phy_device * +of_phy_connect (struct net_device *dev,struct device_node *phy_np, +void (*hndlr)(struct net_device *),u32 flags, +phy_interface_t iface); +struct phy_device *of_phy_get_and_connect (struct net_device *dev,struct device_node *np,void (*hndlr)(struct net_device *)); -struct phy_device *of_phy_attach (struct net_device *dev, -struct device_node *phy_np,u32 flags, -phy_interface_t iface); - -extern struct mii_bus *of_mdio_find_bus (struct device_node *mdio_np); -extern int of_phy_register_fixed_link (struct device_node *np); -extern void of_phy_deregister_fixed_link (struct device_node *np); -extern bool of_phy_is_fixed_link (struct device_node *np); -extern int of_mdiobus_phy_device_register (struct mii_bus *mdio, -struct phy_device *phy, -struct device_node *child,u32 addr); +struct phy_device * +of_phy_attach (struct net_device *dev,struct device_node *phy_np, +u32 flags,phy_interface_t iface); + +struct mii_bus *of_mdio_find_bus (struct device_node *mdio_np); +int of_phy_register_fixed_link (struct device_node *np); +void of_phy_deregister_fixed_link (struct device_node *np); +bool of_phy_is_fixed_link (struct device_node *np); +int of_mdiobus_phy_device_register (struct mii_bus *mdio,struct phy_device *phy, +struct device_node *child,u32 addr);static inline int of_mdio_parse_addr (struct device *dev,const struct device_node *np)
Signature · 7 lines
2.26.1 _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
Re: [PATCH v2 08/10] of: mdio: remove the 'extern' keyword from function declarations
Quoted 6 lines · Bartosz Golaszewski
From: Bartosz Golaszewski <[email protected]>
The 'extern' keyword in headers doesn't have any benefit. Remove them all from the
of_mdio.hheader.Signed-off-by Bartosz Golaszewski <[email protected]>
Signature · 6 lines
Florian _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
PATCH v2 9/10 of: mdio: provide devm_of_mdiobus_register()
In reply to Bartosz GolaszewskiFrom: Bartosz Golaszewski <[email protected]>
Implement a managed variant of of_mdiobus_register(). We need to make mdio_devres into its own module because otherwise we'd hit circular sumbol dependencies between phylib and of_mdio.
--- .../driver-api/driver-model/devres.rst | 1 + drivers/net/phy/Makefile | 4 +- drivers/net/phy/mdio_devres.c | 37 +++++++++++++++++++ include/linux/of_mdio.h | 3 ++ 4 files changed, 44 insertions(+), 1 deletion(-)
Documentation/driver-api/driver-model/devres.rst +1 -0
@@ -343,6 +343,7 @@ MDIO devm_mdiobus_alloc() devm_mdiobus_alloc_size() devm_mdiobus_register() + devm_of_mdiobus_register() MEM devm_free_pages()
drivers/net/phy/Makefile +3 -1
@@ -3,7 +3,8 @@ libphy-y := phy.o phy-c45.o phy-core.o phy_device.o \ linkmode.o -mdio-bus-y += mdio_bus.o mdio_device.o mdio_devres.o +mdio-bus-y += mdio_bus.o mdio_device.o +mdio-devres-y += mdio_devres.o ifdef CONFIG_MDIO_DEVICE obj-y += mdio-boardinfo.o
@@ -17,6 +18,7 @@ libphy-y += $(mdio-bus-y) else obj-$(CONFIG_MDIO_DEVICE) += mdio-bus.o endif +obj-$(CONFIG_MDIO_DEVICE) += mdio-devres.o libphy-$(CONFIG_SWPHY) += swphy.o libphy-$(CONFIG_LED_TRIGGER_PHY) += phy_led_triggers.o
drivers/net/phy/mdio_devres.c +37 -0
@@ -1,6 +1,7 @@// SPDX-License-Identifier: GPL-2.0-or-later #include <linux/device.h> +#include <linux/of_mdio.h> #include <linux/phy.h> #include <linux/stddef.h>
@@ -94,3 +95,39 @@ int __devm_mdiobus_register(struct device *dev, struct mii_bus *bus,return 0 ; }EXPORT_SYMBOL (__devm_mdiobus_register); + +#if IS_ENABLED (CONFIG_OF_MDIO ) +/** +* devm_of_mdiobus_register - Resource managed variant of of_mdiobus_register() +* @dev: Device to register mii_bus for +* @mdio: MII bus structure to register +* @np: Device node to parse +*/ +int devm_of_mdiobus_register (struct device *dev,struct mii_bus *mdio, +struct device_node *np) +{ +struct mdiobus_devres *dr; +int ret; + +if (WARN_ON (!devres_find (dev, devm_mdiobus_free, + mdiobus_devres_match, mdio))) +return -EINVAL ; + + dr =devres_alloc (devm_mdiobus_unregister,sizeof (*dr),GFP_KERNEL ); +if (!dr) +return -ENOMEM ; + + ret =of_mdiobus_register (mdio, np); +if (ret) { +devres_free (dr); +return ret; + } + + dr->mii = mdio; +devres_add (dev, dr); +return 0 ; +} +EXPORT_SYMBOL (devm_of_mdiobus_register); +#endif /* CONFIG_OF_MDIO */ + +MODULE_LICENSE ("GPL" );
include/linux/of_mdio.h +3 -0
@@ -8,12 +8,15 @@#ifndef __LINUX_OF_MDIO_H#define __LINUX_OF_MDIO_H +#include <linux/device.h> #include <linux/phy.h> #include <linux/of.h> #if IS_ENABLED (CONFIG_OF_MDIO )bool of_mdiobus_child_is_phy (struct device_node *child);int of_mdiobus_register (struct mii_bus *mdio,struct device_node *np); +int devm_of_mdiobus_register (struct device *dev,struct mii_bus *mdio, +struct device_node *np);struct phy_device *of_phy_find_device (struct device_node *phy_np);struct phy_device * of_phy_connect(struct net_device *dev, struct device_node *phy_np,
Signature · 7 lines
2.26.1 _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
PATCH v2 10/10 net: ethernet: mtk-star-emac: use devm_of_mdiobus_register()
In reply to Bartosz GolaszewskiFrom: Bartosz Golaszewski <[email protected]>
Shrink the code by using the managed variant of of_mdiobus_register().
--- drivers/net/ethernet/mediatek/mtk_star_emac.c | 13 +------------ 1 file changed, 1 insertion(+), 12 deletions(-)
drivers/net/ethernet/mediatek/mtk_star_emac.c +1 -12
@@ -1389,7 +1389,7 @@ static int mtk_star_mdio_init(struct net_device *ndev) priv->mii->write = mtk_star_mdio_write; priv->mii->priv = priv; - ret =of_mdiobus_register (priv->mii, mdio_node); + ret =devm_of_mdiobus_register (dev, priv->mii, mdio_node); out_put_node:of_node_put (mdio_node);
@@ -1441,13 +1441,6 @@ static void mtk_star_clk_disable_unprepare(void *data)clk_bulk_disable_unprepare (MTK_STAR_NCLKS , priv->clks); } -static void mtk_star_mdiobus_unregister (void *data) -{ -struct mtk_star_priv *priv = data; - -mdiobus_unregister (priv->mii); -} -static int mtk_star_probe (struct platform_device *pdev) {struct device_node *of_node;
@@ -1549,10 +1542,6 @@ static int mtk_star_probe(struct platform_device *pdev)if (ret)return ret; - ret =devm_add_action_or_reset (dev, mtk_star_mdiobus_unregister, priv); -if (ret) -return ret; - ret =eth_platform_get_mac_address (dev, ndev->dev_addr);if (ret || !is_valid_ether_addr (ndev->dev_addr))eth_hw_addr_random (ndev);
Signature · 7 lines
2.26.1 _______________________________________________ Linux-mediatek mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-mediatek
Quoted 14 lines · On Mon, 29 Jun 2020 14:03:36 +0200 Bartosz Golaszewski wrote:
From: Bartosz Golaszewski <[email protected]>
So it seems like there's no support for relaxing certain networking devres helpers to not require previously allocated structures to also be managed. However the way mdio devres variants are implemented is still wrong and I modified my series to address it while keeping the functions strict.
First two patches modify the ixgbe driver to get rid of the last user of
devm_mdiobus_free().Patches 3, 4, 5 and 6 are mostly cosmetic.
Patch 7 fixes the way
devm_mdiobus_register()is implemented.
Quoted 8 lines
Patches 8 & 9 provide a managed variant of
of_mdiobus_register()and last patch uses it in mtk-star-emac driver.v1 -> v2:
- drop the patch relaxingdevm_register_netdev()
- require structmii_busto be managed indevm_mdiobus_register()anddevm_of_mdiobus_register()but don't store that information in the structure itself: usedevres_find()instead
David Miller
maintainerFrom: Bartosz Golaszewski <[email protected]>
Date: Mon, 29 Jun 2020 14:03:36 +0200
Quoted 14 lines
From: Bartosz Golaszewski <[email protected]>
So it seems like there's no support for relaxing certain networking devres helpers to not require previously allocated structures to also be managed. However the way mdio devres variants are implemented is still wrong and I modified my series to address it while keeping the functions strict.
First two patches modify the ixgbe driver to get rid of the last user of
devm_mdiobus_free().Patches 3, 4, 5 and 6 are mostly cosmetic.
Patch 7 fixes the way
devm_mdiobus_register()is implemented.
Quoted 8 lines
Patches 8 & 9 provide a managed variant of
of_mdiobus_register()and last patch uses it in mtk-star-emac driver.v1 -> v2:
- drop the patch relaxingdevm_register_netdev()
- require structmii_busto be managed indevm_mdiobus_register()anddevm_of_mdiobus_register()but don't store that information in the structure itself: usedevres_find()instead
Series applied, thank you.