Move CONFIG_PHY_ADDR to Kconfig
Subsystems (3)
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
CONFIG_PHY_ADDR is used for old-style configuration. This makes impossible changing the PHY address, if multiple boards share a same config header file (for example include/configs/sunxi-common.h).
Moving this to Kconfig helps overcoming this issue. It's defined as entry inside PHYLIB section.
After the implemention, moveconfig was run. The issues are:
- edb9315a - CONFIG_PHYLIB is not enabled. Entry is
deleted.
- ds414 - CONFIG_PHYLIB is in incompatible format:
{ 0x1, 0x0 }. This entry is also deleted.
- devkit3250 - The PHY_ADDR is in hex format (0x1F).
Manually CONFIG_PHY_ADDR=31 is added in the defconfig.
After the changes the suspicious defconfigs passes building.
---
Changes for v2:
- Replaced CONFIG_SUNXI_PHY_ADDR with a common one CONFIG_PHY_ADDR, using moveconfig.
README | 4 ---- configs/devkit3250_defconfig | 1 + configs/khadas-vim_defconfig | 1 + configs/libretech-cc_defconfig | 1 + configs/p212_defconfig | 1 + drivers/net/phy/Kconfig | 7 +++++++ include/configs/am335x_shc.h | 1 - include/configs/baltos.h | 1 - include/configs/devkit3250.h | 1 - include/configs/ds414.h | 1 - include/configs/edb93xx.h | 1 - include/configs/khadas-vim.h | 2 -- include/configs/libretech-cc.h | 2 -- include/configs/p212.h | 2 -- include/configs/pepper.h | 1 - include/configs/sunxi-common.h | 2 -- include/configs/work_92105.h | 1 - include/configs/x600.h | 1 - scripts/config_whitelist.txt | 1 - 19 files changed, 11 insertions(+), 21 deletions(-)
README +0 -4
@@ -1436,10 +1436,6 @@ The following options need to be configured: be at least 4MB. - MII/PHY support: - CONFIG_PHY_ADDR - - The address of PHY on MII bus. - CONFIG_PHY_CLOCK_FREQ (ppc4xx) The clock frequency of the MII bus
configs/devkit3250_defconfig +1 -0
@@ -36,6 +36,7 @@ CONFIG_MTD_NOR_FLASH=y CONFIG_NAND=y CONFIG_SPL_NAND_SIMPLE=y CONFIG_PHYLIB=y +CONFIG_PHY_ADDR=31 CONFIG_DM_SERIAL=y CONFIG_SYS_NS16550=y CONFIG_USB=y
configs/khadas-vim_defconfig +1 -0
@@ -20,6 +20,7 @@ CONFIG_NET_RANDOM_ETHADDR=y CONFIG_DM_GPIO=y CONFIG_DM_MMC=y CONFIG_MMC_MESON_GX=y +CONFIG_PHY_ADDR=8 CONFIG_PHY_MESON_GXL=y CONFIG_DM_ETH=y CONFIG_ETH_DESIGNWARE=y
configs/libretech-cc_defconfig +1 -0
@@ -20,6 +20,7 @@ CONFIG_NET_RANDOM_ETHADDR=y CONFIG_DM_GPIO=y CONFIG_DM_MMC=y CONFIG_MMC_MESON_GX=y +CONFIG_PHY_ADDR=8 CONFIG_PHY_MESON_GXL=y CONFIG_DM_ETH=y CONFIG_ETH_DESIGNWARE=y
configs/p212_defconfig +1 -0
@@ -20,6 +20,7 @@ CONFIG_NET_RANDOM_ETHADDR=y CONFIG_DM_GPIO=y CONFIG_DM_MMC=y CONFIG_MMC_MESON_GX=y +CONFIG_PHY_ADDR=8 CONFIG_PHY_MESON_GXL=y CONFIG_DM_ETH=y CONFIG_ETH_DESIGNWARE=y
drivers/net/phy/Kconfig +7 -0
@@ -12,6 +12,13 @@ menuconfig PHYLIB if PHYLIB +config PHY_ADDR + int "PHY address" + default 1 if ARCH_SUNXI + default 0 + help + The address of PHY on MII bus. Usually in range of 0 to 31. + config B53_SWITCH bool "Broadcom BCM53xx (RoboSwitch) Ethernet switch PHY support." help
include/configs/am335x_shc.h +0 -1
@@ -264,7 +264,6 @@#define CONFIG_BOOTP_GATEWAY #define CONFIG_BOOTP_SUBNETMASK #define CONFIG_NET_RETRY_COUNT 10 -#define CONFIG_PHY_ADDR 0#define CONFIG_PHY_SMSC /* I2C configuration */
include/configs/baltos.h +0 -1
@@ -285,7 +285,6 @@#endif /* Network. */ -#define CONFIG_PHY_ADDR 0#define CONFIG_PHY_SMSC #define CONFIG_MII #define CONFIG_PHY_ATHEROS
include/configs/devkit3250.h +0 -1
@@ -73,7 +73,6 @@#define CONFIG_RMII #define CONFIG_PHY_SMSC #define CONFIG_LPC32XX_ETH -#define CONFIG_PHY_ADDR 0x1F#define CONFIG_SYS_FAULT_ECHO_LINK_DOWN /*
include/configs/ds414.h +0 -1
@@ -41,7 +41,6 @@#define CONFIG_ENV_SECT_SIZE (64 << 10)/* 64KiB sectors */ #define CONFIG_PHY_MARVELL /* there is a marvell phy */ -#define CONFIG_PHY_ADDR { 0x1, 0x0 }#define CONFIG_SYS_NETA_INTERFACE_TYPE PHY_INTERFACE_MODE_RGMII#define CONFIG_SYS_ALT_MEMTEST
include/configs/edb93xx.h +0 -1
@@ -97,7 +97,6 @@#define CONFIG_DRIVER_EP93XX_MAC #define CONFIG_MII_SUPPRESS_PREAMBLE #define CONFIG_MII -#define CONFIG_PHY_ADDR 1#undef CONFIG_NETCONSOLE/* SDRAM configuration */
include/configs/khadas-vim.h +0 -2
@@ -12,8 +12,6 @@#define CONFIG_MISC_INIT_R -#define CONFIG_PHY_ADDR 8 -#define MESON_FDTFILE_SETTING "fdtfile=amlogic/meson-gxl-s905x-khadas-vim.dtb\0"#include <configs/meson-gxbb-common.h>
include/configs/libretech-cc.h +0 -2
@@ -12,8 +12,6 @@#define CONFIG_MISC_INIT_R -#define CONFIG_PHY_ADDR 8 -#define MESON_FDTFILE_SETTING "fdtfile=amlogic/meson-gxl-s905x-libretech-cc.dtb\0"#include <configs/meson-gxbb-common.h>
include/configs/p212.h +0 -2
@@ -12,8 +12,6 @@#define CONFIG_MISC_INIT_R -#define CONFIG_PHY_ADDR 8 -/* Serial setup */ #define CONFIG_CONS_INDEX 0
include/configs/pepper.h +0 -1
@@ -77,7 +77,6 @@#define CONFIG_SYS_NS16550_COM1 0x44e09000/* Ethernet support */ -#define CONFIG_PHY_ADDR 0#define CONFIG_PHY_RESET_DELAY 1000/* SPL */
include/configs/sunxi-common.h +0 -2
@@ -296,12 +296,10 @@ extern int soft_i2c_gpio_scl;/* Ethernet support */ #ifdef CONFIG_SUN4I_EMAC -#define CONFIG_PHY_ADDR 1#define CONFIG_MII /* MII PHY management */ #endif #ifdef CONFIG_SUN7I_GMAC -#define CONFIG_PHY_ADDR 1#define CONFIG_MII /* MII PHY management */ #define CONFIG_PHY_REALTEK #endif
include/configs/work_92105.h +0 -1
@@ -57,7 +57,6 @@#define CONFIG_PHY_SMSC #define CONFIG_LPC32XX_ETH -#define CONFIG_PHY_ADDR 0#define CONFIG_SYS_FAULT_ECHO_LINK_DOWN /* FIXME: remove "Waiting for PHY auto negotiation to complete..." message */
include/configs/x600.h +0 -1
@@ -71,7 +71,6 @@/* Ethernet config options */ #define CONFIG_MII #define CONFIG_PHY_RESET_DELAY 10000/* in usec */ -#define CONFIG_PHY_ADDR 0/* PHY address */ #define CONFIG_SPEAR_GPIO
scripts/config_whitelist.txt +0 -1
@@ -1593,7 +1593,6 @@ CONFIG_PERIF2_FREQ CONFIG_PERIF3_FREQ CONFIG_PERIF4_FREQ CONFIG_PHYSMEM -CONFIG_PHY_ADDR CONFIG_PHY_BASE_ADR CONFIG_PHY_BCM5421S CONFIG_PHY_ET1011C_TX_CLK_FIX
Signature · 1 line
2.7.4
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
Hi,
Quoted 3 lines
From: U-Boot [mailto:u-boot-bounces at lists.denx.de] On Behalf Of Stefan Mavrodiev
Sent: Friday, February 2, 2018 7:24 PM
<snip>
Quoted 5 lines
drivers/net/phy/Kconfig index renamed from drivers/net/phy/Kconfig +0
-095b7534..c934aed 100644
drivers/net/phy/Kconfig menuconfig PHYLIB · line 20
Quoted 9 lines
drivers/net/phy/Kconfig +6
-0@@ -12,6 +12,13 @@ menuconfig PHYLIB if PHYLIB +config PHY_ADDR + int "PHY address" + default 1 if ARCH_SUNXI + default 0 + help + The address of PHY on MII bus. Usually in range of 0 to 31.
Isn't board Kconfig, the right place to keep this?
Regards
Calvin
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
Quoted 14 lines
Hi,
Quoted 3 lines
From: U-Boot [mailto:u-boot-bounces at lists.denx.de] On Behalf Of Stefan Mavrodiev
Sent: Friday, February 2, 2018 7:24 PM<snip>
Quoted 6 lines
drivers/net/phy/Kconfig index renamed from drivers/net/phy/Kconfig +0
-095b7534..c934aed 100644
drivers/net/phy/Kconfig +0
-0@@ -12,6 +12,13 @@ menuconfig PHYLIB
Quoted 5 lines
drivers/net/phy/Kconfig menuconfig PHYLIB · line 17
Quoted 5 lines
drivers/net/phy/Kconfig +3
-0if PHYLIB +config PHY_ADDR + int "PHY address" + default 1 if ARCH_SUNXI
Sorry, I meant the default value can be defined in board/sunxi/Kconfig. Remaining definition of PHY_ADDR config can be here.
Quoted 5 lines
drivers/net/phy/Kconfig menuconfig PHYLIB · line 20
Quoted 3 lines
drivers/net/phy/Kconfig +3
-0+ default 0 + help + The address of PHY on MII bus. Usually in range of 0 to 31.Isn't board
Kconfig, the right place to keep this?
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
Quoted 14 lines · Calvin Johnson
Quoted 14 lines
Hi,
Quoted 3 lines
From: U-Boot [mailto:u-boot-bounces at lists.denx.de] On Behalf Of Stefan Mavrodiev
Sent: Friday, February 2, 2018 7:24 PM<snip>
Quoted 6 lines
drivers/net/phy/Kconfig index renamed from drivers/net/phy/Kconfig +0
-095b7534..c934aed 100644
drivers/net/phy/Kconfig +0
-0@@ -12,6 +12,13 @@ menuconfig PHYLIB
Quoted 8 lines
Quoted 5 lines
drivers/net/phy/Kconfig menuconfig PHYLIB · line 17
Quoted 5 lines
drivers/net/phy/Kconfig +3
-0if PHYLIB +config PHY_ADDR + int "PHY address" + default 1 if ARCH_SUNXISorry, I meant the default value can be defined in board/sunxi/Kconfig. Remaining definition of
PHY_ADDRconfig can be here.
Tom has asked a few times not to do this but to put the defaults where the Kconfig option is defined. I'm not sure I remember the details though.
Maxime
Signature · 10 lines
Maxime Ripard, Bootlin (formerly Free Electrons) Embedded Linux and Kernel engineering http://bootlin.com -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 833 bytes Desc: not available URL: <http://lists.denx.de/pipermail/u-boot/attachments/20180205/eafd1a57/attachment.sig>
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
Quoted 24 lines · Maxime Ripard
Quoted 14 lines · Calvin Johnson
Quoted 14 lines
Hi,
Quoted 3 lines
From: U-Boot [mailto:u-boot-bounces at lists.denx.de] On Behalf Of Stefan Mavrodiev
Sent: Friday, February 2, 2018 7:24 PM<snip>
Quoted 6 lines
drivers/net/phy/Kconfig index renamed from drivers/net/phy/Kconfig +0
-095b7534..c934aed 100644
drivers/net/phy/Kconfig +0
-0@@ -12,6 +12,13 @@ menuconfig PHYLIBQuoted 8 lines
Quoted 5 lines
drivers/net/phy/Kconfig menuconfig PHYLIB · line 17
Quoted 5 lines
drivers/net/phy/Kconfig +3
-0if PHYLIB +config PHY_ADDR + int "PHY address" + default 1 if ARCH_SUNXISorry, I meant the default value can be defined in board/sunxi/Kconfig. Remaining definition of
PHY_ADDRconfig can be here.
Quoted 3 lines
Tom has asked a few times not to do this but to put the defaults where the
Kconfigoption is defined. I'm not sure I remember the details though.
For now, yes, the preferred way of dealing with this is 'default FOO if BAR' in the main Kconfig entry line. What I wish for long term is for 'imply' to get extended to support 'imply FOO BAR' but I've not had more to add there other than "Gee, this would make life easier".
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 819 bytes Desc: not available URL: <http://lists.denx.de/pipermail/u-boot/attachments/20180205/a25979bb/attachment.sig>
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
Hi,
Quoted 28 lines
-----Original Message-----
From: Tom Rini [mailto:trini at konsulko.com]Quoted 24 lines · Maxime Ripard
Quoted 22 lines · Calvin Johnson
Quoted 14 lines
Hi,
Quoted 3 lines
From: U-Boot [mailto:u-boot-bounces at lists.denx.de] On Behalf Of Stefan Mavrodiev
Sent: Friday, February 2, 2018 7:24 PM<snip>
Quoted 6 lines
drivers/net/phy/Kconfig +0
-0@@ -12,6 +12,13 @@ menuconfig PHYLIBQuoted 5 lines
drivers/net/phy/Kconfig menuconfig PHYLIB · line 17
Quoted 5 lines
drivers/net/phy/Kconfig +3
-0if PHYLIB +config PHY_ADDR + int "PHY address" + default 1 if ARCH_SUNXISorry, I meant the default value can be defined in board/sunxi/Kconfig. Remaining definition of
PHY_ADDRconfig can be here.
Quoted 8 lines
Quoted 3 lines
Tom has asked a few times not to do this but to put the defaults where the
Kconfigoption is defined. I'm not sure I remember the details though.For now, yes, the preferred way of dealing with this is 'default FOO if BAR' in the main
Kconfigentry line. What I wish for long term is for 'imply' to get extended to support 'imply FOO BAR' but I've not had more to add there other than "Gee, this would make life easier".
Got it. Thanks!
Calvin
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
In reply to Stefan MavrodievQuoted 18 lines · Stefan Mavrodiev
CONFIG_PHY_ADDRis used for old-style configuration. This makes impossible changing the PHY address, if multiple boards share a same config header file (for exampleinclude/configs/sunxi-common.h).Moving this to
Kconfighelps overcoming this issue. It's defined as entry inside PHYLIB section.After the implemention, moveconfig was run. The issues are:
-edb9315a-CONFIG_PHYLIBis not enabled. Entry is
deleted.- ds414 -
CONFIG_PHYLIBis in incompatible format:
{0x1,0x0}. This entry is also deleted.- devkit3250 - The
PHY_ADDRis in hex format (0x1F).
ManuallyCONFIG_PHY_ADDR=31is added in the defconfig.
Quoted 5 lines
After the changes the suspicious defconfigs passes building.
Signed-off-by Stefan Mavrodiev <[email protected]>Acked-by Maxime Ripard <[email protected]>---
rebased on master and
Applied to u-boot-sunxi/master
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
In reply to Stefan MavrodievHi,
Quoted 48 lines · Stefan Mavrodiev
CONFIG_PHY_ADDRis used for old-style configuration. This makes impossible changing the PHY address, if multiple boards share a same config header file (for exampleinclude/configs/sunxi-common.h).Moving this to
Kconfighelps overcoming this issue. It's defined as entry inside PHYLIB section.After the implemention, moveconfig was run. The issues are:
-edb9315a-CONFIG_PHYLIBis not enabled. Entry is
deleted.- ds414 -
CONFIG_PHYLIBis in incompatible format:
{0x1,0x0}. This entry is also deleted.- devkit3250 - The
PHY_ADDRis in hex format (0x1F).
ManuallyCONFIG_PHY_ADDR=31is added in the defconfig.After the changes the suspicious defconfigs passes building.
Signed-off-by Stefan Mavrodiev <[email protected]>Acked-by Maxime Ripard <[email protected]>---
Changes for v2:
- ReplacedCONFIG_SUNXI_PHY_ADDRwith a common oneCONFIG_PHY_ADDR, using moveconfig.README | 4 ---- configs/devkit3250_defconfig | 1 + configs/khadas-vim_defconfig | 1 + configs/libretech-cc_defconfig | 1 + configs/p212_defconfig | 1 + drivers/net/phy/Kconfig | 7 +++++++ include/configs/am335x_shc.h | 1 - include/configs/baltos.h | 1 - include/configs/devkit3250.h | 1 - include/configs/ds414.h | 1 - include/configs/edb93xx.h | 1 - include/configs/khadas-vim.h | 2 -- include/configs/libretech-cc.h | 2 -- include/configs/p212.h | 2 -- include/configs/pepper.h | 1 - include/configs/sunxi-common.h | 2 -- include/configs/work_92105.h | 1 - include/configs/x600.h | 1 - scripts/config_whitelist.txt | 1 - 19 files changed, 11 insertions(+), 21 deletions(-)
[snip]
Quoted 5 lines
drivers/net/phy/Kconfig +0
-0
drivers/net/phy/Kconfig menuconfig PHYLIB · line 21
Quoted 10 lines
drivers/net/phy/Kconfig +7
-0@@ -12,6 +12,13 @@ menuconfig PHYLIB if PHYLIB +config PHY_ADDR + int "PHY address" + default 1 if ARCH_SUNXI + default 0 + help + The address of PHY on MII bus. Usually in range of 0 to 31. +
Sorry for jumping out so late, but this commit breaks Intel Galileo ethernet. Previously the board boots with the following log:
Net: eth0: eth_designware#0, eth1: eth_designware#1
With this commit it becomes:
Net: No ethernet found.
The reason is that the board has two designware ethernet controllers, and PHY_ADDR has been set to zero for both. A simple fix is to:
drivers/net/designware.c +0 -4
@@ -471,10 +471,6 @@ static int dw_phy_init(struct dw_eth_dev *priv, void *dev)struct phy_device *phydev;int mask =0xffffffff , ret; -#ifdef CONFIG_PHY_ADDR - mask =1 <<CONFIG_PHY_ADDR ; -#endif - phydev =phy_find_by_mask (priv->bus, mask, priv->interface);if (!phydev)return -ENODEV ;
But the real question is that: why do we introduce this PHY_ADDR Kconfig? It for sure won't work for multiple ethernet controllers.This should be eliminated IMHO. Comments?
[snip]
Regards,
Bin
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
Quoted 98 lines · Bin Meng
Hi,
Quoted 48 lines · Stefan Mavrodiev
CONFIG_PHY_ADDRis used for old-style configuration. This makes impossible changing the PHY address, if multiple boards share a same config header file (for exampleinclude/configs/sunxi-common.h).Moving this to
Kconfighelps overcoming this issue. It's defined as entry inside PHYLIB section.After the implemention, moveconfig was run. The issues are:
-edb9315a-CONFIG_PHYLIBis not enabled. Entry is
deleted.- ds414 -
CONFIG_PHYLIBis in incompatible format:
{0x1,0x0}. This entry is also deleted.- devkit3250 - The
PHY_ADDRis in hex format (0x1F).
ManuallyCONFIG_PHY_ADDR=31is added in the defconfig.After the changes the suspicious defconfigs passes building.
Signed-off-by Stefan Mavrodiev <[email protected]>Acked-by Maxime Ripard <[email protected]>---
Changes for v2:
- ReplacedCONFIG_SUNXI_PHY_ADDRwith a common oneCONFIG_PHY_ADDR, using moveconfig.README | 4 ---- configs/devkit3250_defconfig | 1 + configs/khadas-vim_defconfig | 1 + configs/libretech-cc_defconfig | 1 + configs/p212_defconfig | 1 + drivers/net/phy/Kconfig | 7 +++++++ include/configs/am335x_shc.h | 1 - include/configs/baltos.h | 1 - include/configs/devkit3250.h | 1 - include/configs/ds414.h | 1 - include/configs/edb93xx.h | 1 - include/configs/khadas-vim.h | 2 -- include/configs/libretech-cc.h | 2 -- include/configs/p212.h | 2 -- include/configs/pepper.h | 1 - include/configs/sunxi-common.h | 2 -- include/configs/work_92105.h | 1 - include/configs/x600.h | 1 - scripts/config_whitelist.txt | 1 - 19 files changed, 11 insertions(+), 21 deletions(-)[snip]
Quoted 5 lines
drivers/net/phy/Kconfig +0
-0drivers/net/phy/Kconfig menuconfig PHYLIB · line 21
Quoted 10 lines
drivers/net/phy/Kconfig +7
-0@@ -12,6 +12,13 @@ menuconfig PHYLIB if PHYLIB +config PHY_ADDR + int "PHY address" + default 1 if ARCH_SUNXI + default 0 + help + The address of PHY on MII bus. Usually in range of 0 to 31. +Sorry for jumping out so late, but this commit breaks Intel Galileo ethernet. Previously the board boots with the following log:
Net: eth0: eth_designware#0, eth1: eth_designware#1
With this commit it becomes:
Net: No ethernet found.
The reason is that the board has two designware ethernet controllers, and
PHY_ADDRhas been set to zero for both. A simple fix is to:drivers/net/designware.c +0
-4@@ -471,10 +471,6 @@ static int dw_phy_init(struct dw_eth_dev *priv, void *dev)struct phy_device *phydev;int mask =0xffffffff , ret; -#ifdef CONFIG_PHY_ADDR - mask =1 <<CONFIG_PHY_ADDR ; -#endif - phydev =phy_find_by_mask (priv->bus, mask, priv->interface);if (!phydev)return -ENODEV ;
Quoted 3 lines
But the real question is that: why do we introduce this
PHY_ADDRKconfig? It for sure won't work for multiple ethernet controllers.This should be eliminated IMHO. Comments?
This should be able to come from the device tree, ultimately. Can you undefine the phy addr for the Galileo board?
Quoted 8 lines
[snip]
Regards,
Bin
_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
Hi Joe,
Quoted 99 lines · Joe Hershberger
Quoted 98 lines · Bin Meng
Hi,
Quoted 48 lines · Stefan Mavrodiev
CONFIG_PHY_ADDRis used for old-style configuration. This makes impossible changing the PHY address, if multiple boards share a same config header file (for exampleinclude/configs/sunxi-common.h).Moving this to
Kconfighelps overcoming this issue. It's defined as entry inside PHYLIB section.After the implemention, moveconfig was run. The issues are:
-edb9315a-CONFIG_PHYLIBis not enabled. Entry is
deleted.- ds414 -
CONFIG_PHYLIBis in incompatible format:
{0x1,0x0}. This entry is also deleted.- devkit3250 - The
PHY_ADDRis in hex format (0x1F).
ManuallyCONFIG_PHY_ADDR=31is added in the defconfig.After the changes the suspicious defconfigs passes building.
Signed-off-by Stefan Mavrodiev <[email protected]>Acked-by Maxime Ripard <[email protected]>---
Changes for v2:
- ReplacedCONFIG_SUNXI_PHY_ADDRwith a common oneCONFIG_PHY_ADDR, using moveconfig.README | 4 ---- configs/devkit3250_defconfig | 1 + configs/khadas-vim_defconfig | 1 + configs/libretech-cc_defconfig | 1 + configs/p212_defconfig | 1 + drivers/net/phy/Kconfig | 7 +++++++ include/configs/am335x_shc.h | 1 - include/configs/baltos.h | 1 - include/configs/devkit3250.h | 1 - include/configs/ds414.h | 1 - include/configs/edb93xx.h | 1 - include/configs/khadas-vim.h | 2 -- include/configs/libretech-cc.h | 2 -- include/configs/p212.h | 2 -- include/configs/pepper.h | 1 - include/configs/sunxi-common.h | 2 -- include/configs/work_92105.h | 1 - include/configs/x600.h | 1 - scripts/config_whitelist.txt | 1 - 19 files changed, 11 insertions(+), 21 deletions(-)[snip]
Quoted 5 lines
drivers/net/phy/Kconfig +0
-0drivers/net/phy/Kconfig menuconfig PHYLIB · line 21
Quoted 10 lines
drivers/net/phy/Kconfig +7
-0@@ -12,6 +12,13 @@ menuconfig PHYLIB if PHYLIB +config PHY_ADDR + int "PHY address" + default 1 if ARCH_SUNXI + default 0 + help + The address of PHY on MII bus. Usually in range of 0 to 31. +Sorry for jumping out so late, but this commit breaks Intel Galileo ethernet. Previously the board boots with the following log:
Net: eth0: eth_designware#0, eth1: eth_designware#1
With this commit it becomes:
Net: No ethernet found.
The reason is that the board has two designware ethernet controllers, and
PHY_ADDRhas been set to zero for both. A simple fix is to:drivers/net/designware.c +0
-4@@ -471,10 +471,6 @@ static int dw_phy_init(struct dw_eth_dev *priv, void *dev)struct phy_device *phydev;int mask =0xffffffff , ret; -#ifdef CONFIG_PHY_ADDR - mask =1 <<CONFIG_PHY_ADDR ; -#endif - phydev =phy_find_by_mask (priv->bus, mask, priv->interface);if (!phydev)return -ENODEV ;
Quoted 9 lines
Quoted 3 lines
But the real question is that: why do we introduce this
PHY_ADDRKconfig? It for sure won't work for multiple ethernet controllers.This should be eliminated IMHO. Comments?This should be able to come from the device tree, ultimately. Can you undefine the phy addr for the Galileo board?
Quoted 2 lines
[snip]
#undf the PHY_ADDR in Galileo board looks weird. This to me is a workaround.
Since the designware ethernet controller driver supports finding any PHY attached to its mdio bus, the changes suggested above can be a proper fix.
Regards,
Bin
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
Quoted 109 lines · Bin Meng
Hi Joe,
Quoted 106 lines · Joe Hershberger
Quoted 98 lines · Bin Meng
Hi,
Quoted 48 lines · Stefan Mavrodiev
CONFIG_PHY_ADDRis used for old-style configuration. This makes impossible changing the PHY address, if multiple boards share a same config header file (for exampleinclude/configs/sunxi-common.h).Moving this to
Kconfighelps overcoming this issue. It's defined as entry inside PHYLIB section.After the implemention, moveconfig was run. The issues are:
-edb9315a-CONFIG_PHYLIBis not enabled. Entry is
deleted.- ds414 -
CONFIG_PHYLIBis in incompatible format:
{0x1,0x0}. This entry is also deleted.- devkit3250 - The
PHY_ADDRis in hex format (0x1F).
ManuallyCONFIG_PHY_ADDR=31is added in the defconfig.After the changes the suspicious defconfigs passes building.
Signed-off-by Stefan Mavrodiev <[email protected]>Acked-by Maxime Ripard <[email protected]>---
Changes for v2:
- ReplacedCONFIG_SUNXI_PHY_ADDRwith a common oneCONFIG_PHY_ADDR, using moveconfig.README | 4 ---- configs/devkit3250_defconfig | 1 + configs/khadas-vim_defconfig | 1 + configs/libretech-cc_defconfig | 1 + configs/p212_defconfig | 1 + drivers/net/phy/Kconfig | 7 +++++++ include/configs/am335x_shc.h | 1 - include/configs/baltos.h | 1 - include/configs/devkit3250.h | 1 - include/configs/ds414.h | 1 - include/configs/edb93xx.h | 1 - include/configs/khadas-vim.h | 2 -- include/configs/libretech-cc.h | 2 -- include/configs/p212.h | 2 -- include/configs/pepper.h | 1 - include/configs/sunxi-common.h | 2 -- include/configs/work_92105.h | 1 - include/configs/x600.h | 1 - scripts/config_whitelist.txt | 1 - 19 files changed, 11 insertions(+), 21 deletions(-)[snip]
Quoted 5 lines
drivers/net/phy/Kconfig +0
-0drivers/net/phy/Kconfig menuconfig PHYLIB · line 21
Quoted 10 lines
drivers/net/phy/Kconfig +7
-0@@ -12,6 +12,13 @@ menuconfig PHYLIB if PHYLIB +config PHY_ADDR + int "PHY address" + default 1 if ARCH_SUNXI + default 0 + help + The address of PHY on MII bus. Usually in range of 0 to 31. +Sorry for jumping out so late, but this commit breaks Intel Galileo ethernet. Previously the board boots with the following log:
Net: eth0: eth_designware#0, eth1: eth_designware#1
With this commit it becomes:
Net: No ethernet found.
The reason is that the board has two designware ethernet controllers, and
PHY_ADDRhas been set to zero for both. A simple fix is to:drivers/net/designware.c +0
-4@@ -471,10 +471,6 @@ static int dw_phy_init(struct dw_eth_dev *priv, void *dev)struct phy_device *phydev;int mask =0xffffffff , ret; -#ifdef CONFIG_PHY_ADDR - mask =1 <<CONFIG_PHY_ADDR ; -#endif - phydev =phy_find_by_mask (priv->bus, mask, priv->interface);if (!phydev)return -ENODEV ;Quoted 3 lines
But the real question is that: why do we introduce this
PHY_ADDRKconfig? It for sure won't work for multiple ethernet controllers.This should be eliminated IMHO. Comments?This should be able to come from the device tree, ultimately. Can you undefine the phy addr for the Galileo board?
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines
[snip]
#undf the
PHY_ADDRin Galileo board looks weird. This to me is a workaround.
I didn't mean to add a #undef. I was just saying that if the "default 0" in the Kconfig were instead "default 0 if !X86" or something (or maybe if the board defconfig explicitly does unselects it).
Quoted 3 lines
Since the designware ethernet controller driver supports finding any PHY attached to its mdio bus, the changes suggested above can be a proper fix.
That is good for your board, but some board may have more than one phy and want to specify which to use in U-Boot.
Ultimately it should be possible to read it from the DT.
Cheers,
-Joe
Quoted 6 lines
Regards,
Bin
_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
Hi Joe,
Quoted 110 lines · Joe Hershberger
Quoted 109 lines · Bin Meng
Hi Joe,
Quoted 106 lines · Joe Hershberger
Quoted 98 lines · Bin Meng
Hi,
Quoted 48 lines · Stefan Mavrodiev
CONFIG_PHY_ADDRis used for old-style configuration. This makes impossible changing the PHY address, if multiple boards share a same config header file (for exampleinclude/configs/sunxi-common.h).Moving this to
Kconfighelps overcoming this issue. It's defined as entry inside PHYLIB section.After the implemention, moveconfig was run. The issues are:
-edb9315a-CONFIG_PHYLIBis not enabled. Entry is
deleted.- ds414 -
CONFIG_PHYLIBis in incompatible format:
{0x1,0x0}. This entry is also deleted.- devkit3250 - The
PHY_ADDRis in hex format (0x1F).
ManuallyCONFIG_PHY_ADDR=31is added in the defconfig.After the changes the suspicious defconfigs passes building.
Signed-off-by Stefan Mavrodiev <[email protected]>Acked-by Maxime Ripard <[email protected]>---
Changes for v2:
- ReplacedCONFIG_SUNXI_PHY_ADDRwith a common oneCONFIG_PHY_ADDR, using moveconfig.README | 4 ---- configs/devkit3250_defconfig | 1 + configs/khadas-vim_defconfig | 1 + configs/libretech-cc_defconfig | 1 + configs/p212_defconfig | 1 + drivers/net/phy/Kconfig | 7 +++++++ include/configs/am335x_shc.h | 1 - include/configs/baltos.h | 1 - include/configs/devkit3250.h | 1 - include/configs/ds414.h | 1 - include/configs/edb93xx.h | 1 - include/configs/khadas-vim.h | 2 -- include/configs/libretech-cc.h | 2 -- include/configs/p212.h | 2 -- include/configs/pepper.h | 1 - include/configs/sunxi-common.h | 2 -- include/configs/work_92105.h | 1 - include/configs/x600.h | 1 - scripts/config_whitelist.txt | 1 - 19 files changed, 11 insertions(+), 21 deletions(-)[snip]
Quoted 5 lines
drivers/net/phy/Kconfig +0
-0drivers/net/phy/Kconfig menuconfig PHYLIB · line 21
Quoted 10 lines
drivers/net/phy/Kconfig +7
-0@@ -12,6 +12,13 @@ menuconfig PHYLIB if PHYLIB +config PHY_ADDR + int "PHY address" + default 1 if ARCH_SUNXI + default 0 + help + The address of PHY on MII bus. Usually in range of 0 to 31. +Sorry for jumping out so late, but this commit breaks Intel Galileo ethernet. Previously the board boots with the following log:
Net: eth0: eth_designware#0, eth1: eth_designware#1
With this commit it becomes:
Net: No ethernet found.
The reason is that the board has two designware ethernet controllers, and
PHY_ADDRhas been set to zero for both. A simple fix is to:drivers/net/designware.c +0
-4@@ -471,10 +471,6 @@ static int dw_phy_init(struct dw_eth_dev *priv, void *dev)struct phy_device *phydev;int mask =0xffffffff , ret; -#ifdef CONFIG_PHY_ADDR - mask =1 <<CONFIG_PHY_ADDR ; -#endif - phydev =phy_find_by_mask (priv->bus, mask, priv->interface);if (!phydev)return -ENODEV ;Quoted 3 lines
But the real question is that: why do we introduce this
PHY_ADDRKconfig? It for sure won't work for multiple ethernet controllers.This should be eliminated IMHO. Comments?This should be able to come from the device tree, ultimately. Can you undefine the phy addr for the Galileo board?
Quoted 9 lines
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines
[snip]
#undf the
PHY_ADDRin Galileo board looks weird. This to me is a workaround.I didn't mean to add a #undef. I was just saying that if the "default 0" in the
Kconfigwere instead "default 0 if !X86" or something (or maybe if the board defconfig explicitly does unselects it).
This cannot be done as CONFIG_PHY_ADDR is an "int", not a "bool". We cannot do "# CONFIG_PHY_ADDR is not set" in the galileo_defconfig.
Quoted 9 lines
Quoted 3 lines
Since the designware ethernet controller driver supports finding any PHY attached to its mdio bus, the changes suggested above can be a proper fix.
That is good for your board, but some board may have more than one phy and want to specify which to use in U-Boot.
Ultimately it should be possible to read it from the DT.
When will this be fixed? At present the CONFIG_PHY_ADDR concept is just wrong. It can only handle one eth and one PHY.
Regards,
Bin
PATCH v2 1/3 Move CONFIG_PHY_ADDR to Kconfig
Quoted 118 lines · Bin Meng
Hi Joe,
Quoted 115 lines · Joe Hershberger
Quoted 113 lines · Bin Meng
Hi Joe,
Quoted 99 lines · Joe Hershberger
Quoted 98 lines · Bin Meng
Hi,
Quoted 48 lines · Stefan Mavrodiev
CONFIG_PHY_ADDRis used for old-style configuration. This makes impossible changing the PHY address, if multiple boards share a same config header file (for exampleinclude/configs/sunxi-common.h).Moving this to
Kconfighelps overcoming this issue. It's defined as entry inside PHYLIB section.After the implemention, moveconfig was run. The issues are:
-edb9315a-CONFIG_PHYLIBis not enabled. Entry is
deleted.- ds414 -
CONFIG_PHYLIBis in incompatible format:
{0x1,0x0}. This entry is also deleted.- devkit3250 - The
PHY_ADDRis in hex format (0x1F).
ManuallyCONFIG_PHY_ADDR=31is added in the defconfig.After the changes the suspicious defconfigs passes building.
Signed-off-by Stefan Mavrodiev <[email protected]>Acked-by Maxime Ripard <[email protected]>---
Changes for v2:
- ReplacedCONFIG_SUNXI_PHY_ADDRwith a common oneCONFIG_PHY_ADDR, using moveconfig.README | 4 ---- configs/devkit3250_defconfig | 1 + configs/khadas-vim_defconfig | 1 + configs/libretech-cc_defconfig | 1 + configs/p212_defconfig | 1 + drivers/net/phy/Kconfig | 7 +++++++ include/configs/am335x_shc.h | 1 - include/configs/baltos.h | 1 - include/configs/devkit3250.h | 1 - include/configs/ds414.h | 1 - include/configs/edb93xx.h | 1 - include/configs/khadas-vim.h | 2 -- include/configs/libretech-cc.h | 2 -- include/configs/p212.h | 2 -- include/configs/pepper.h | 1 - include/configs/sunxi-common.h | 2 -- include/configs/work_92105.h | 1 - include/configs/x600.h | 1 - scripts/config_whitelist.txt | 1 - 19 files changed, 11 insertions(+), 21 deletions(-)[snip]
Quoted 5 lines
drivers/net/phy/Kconfig +0
-0drivers/net/phy/Kconfig menuconfig PHYLIB · line 21
Quoted 10 lines
drivers/net/phy/Kconfig +7
-0@@ -12,6 +12,13 @@ menuconfig PHYLIB if PHYLIB +config PHY_ADDR + int "PHY address" + default 1 if ARCH_SUNXI + default 0 + help + The address of PHY on MII bus. Usually in range of 0 to 31. +Sorry for jumping out so late, but this commit breaks Intel Galileo ethernet. Previously the board boots with the following log:
Net: eth0: eth_designware#0, eth1: eth_designware#1
With this commit it becomes:
Net: No ethernet found.
The reason is that the board has two designware ethernet controllers, and
PHY_ADDRhas been set to zero for both. A simple fix is to:drivers/net/designware.c +0
-4@@ -471,10 +471,6 @@ static int dw_phy_init(struct dw_eth_dev *priv, void *dev)struct phy_device *phydev;int mask =0xffffffff , ret; -#ifdef CONFIG_PHY_ADDR - mask =1 <<CONFIG_PHY_ADDR ; -#endif - phydev =phy_find_by_mask (priv->bus, mask, priv->interface);if (!phydev)return -ENODEV ;Quoted 9 lines
Quoted 3 lines
But the real question is that: why do we introduce this
PHY_ADDRKconfig? It for sure won't work for multiple ethernet controllers.This should be eliminated IMHO. Comments?This should be able to come from the device tree, ultimately. Can you undefine the phy addr for the Galileo board?
Quoted 2 lines
[snip]
#undf the
PHY_ADDRin Galileo board looks weird. This to me is a workaround.
Quoted 7 lines
Quoted 4 lines
I didn't mean to add a #undef. I was just saying that if the "default 0" in the
Kconfigwere instead "default 0 if !X86" or something (or maybe if the board defconfig explicitly does unselects it).This cannot be done as
CONFIG_PHY_ADDRis an "int", not a "bool". We cannot do "#CONFIG_PHY_ADDRis not set" in thegalileo_defconfig.
I just sent an RFC.
Quoted 4 lines
Quoted 4 lines
Quoted 3 lines
Since the designware ethernet controller driver supports finding any PHY attached to its mdio bus, the changes suggested above can be a proper fix.
Quoted 8 lines
Quoted 5 lines
That is good for your board, but some board may have more than one phy and want to specify which to use in U-Boot.
Ultimately it should be possible to read it from the DT.
When will this be fixed? At present the
CONFIG_PHY_ADDRconcept is just wrong. It can only handle one eth and one PHY.
I'm not sure exactly when I'll have the DT sourced phy addr ready, but hopefully the RFC will fix things in the meantime.
Cheers,
-Joe
Quoted 6 lines
Regards,
Bin
_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot