[U-Boot] [PATCH] net: Multiple updates/enhancements to designware.c
This patch adds the following changes to designware ethernet driver found on the ST SPEAr SoC:
- Don't init MAC & PHY upon startup. This causes a delay, waiting for the auto negotiation to complete. And we don't want this delay to always happen. Especially not on platforms where ethernet is not used at all (e.g. booting via flash). Instead postpone the MAC / PHY configuration to the stage, where ethernet is first used. - Add possibility for board specific PHY init code. This is needed for example on the X600 board, where the Vitesse PHY needs to be configured for GMII mode. This board specific PHY init is done via the function designware_board_phy_init(). And this driver now adds a weak default which can be overridden by board code. - Use common functions miiphy_speed() & miiphy_duplex() to read link status from PHY. - Print status and progress of auto negotiation. - Print link status (speed, dupex) upon first usage.
Signed-off-by: Stefan Roese sr@denx.de Cc: Amit Virdi amit.virdi@st.com Cc: Vipin Kumar vipin.kumar@st.com Cc: Joe Hershberger joe.hershberger@gmail.com --- drivers/net/designware.c | 114 +++++++++++++++++----------------------- drivers/net/designware.h | 1 + include/configs/spear-common.h | 1 + 3 files changed, 49 insertions(+), 67 deletions(-)
diff --git a/drivers/net/designware.c b/drivers/net/designware.c index e263022..945e57b 100644 --- a/drivers/net/designware.c +++ b/drivers/net/designware.c @@ -28,6 +28,7 @@ #include <common.h> #include <miiphy.h> #include <malloc.h> +#include <linux/compiler.h> #include <linux/err.h> #include <asm/io.h> #include "designware.h" @@ -153,6 +154,13 @@ static int dw_eth_init(struct eth_device *dev, bd_t *bis) if (priv->phy_configured != 1) configure_phy(dev);
+ /* Print link status only once */ + if (!priv->link_printed) { + printf("ENET Speed is %d Mbps - %s duplex connection\n", + priv->speed, (priv->duplex == HALF) ? "HALF" : "FULL"); + priv->link_printed = 1; + } + /* Reset ethernet hardware */ if (mac_reset(dev) < 0) return -1; @@ -168,17 +176,17 @@ static int dw_eth_init(struct eth_device *dev, bd_t *bis)
conf = FRAMEBURSTENABLE | DISABLERXOWN;
- if (priv->speed != SPEED_1000M) + if (priv->speed != 1000) conf |= MII_PORTSELECT;
if ((priv->interface != PHY_INTERFACE_MODE_MII) && (priv->interface != PHY_INTERFACE_MODE_GMII)) {
- if (priv->speed == SPEED_100M) + if (priv->speed == 100) conf |= FES_100; }
- if (priv->duplex == FULL_DUPLEX) + if (priv->duplex == FULL) conf |= FULLDPLXMODE;
writel(conf, &mac_p->conf); @@ -397,6 +405,16 @@ static int dw_reset_phy(struct eth_device *dev) return 0; }
+/* + * Add weak default function for board specific PHY configuration + */ +int __weak designware_board_phy_init(struct eth_device *dev, int phy_addr, + int (*mii_write)(struct eth_device *, u8, u8, u16), + int dw_reset_phy(struct eth_device *)) +{ + return 0; +} + static int configure_phy(struct eth_device *dev) { struct dw_eth_dev *priv = dev->priv; @@ -406,9 +424,6 @@ static int configure_phy(struct eth_device *dev) u16 bmsr; u32 timeout; ulong start; - u16 anlpar, btsr; -#else - u16 ctrl; #endif
#if defined(CONFIG_DW_SEARCH_PHY) @@ -420,6 +435,16 @@ static int configure_phy(struct eth_device *dev) #else phy_addr = priv->address; #endif + + /* + * Some boards need board specific PHY initialization. This is + * after the main driver init code but before the auto negotiation + * is run. + */ + if (designware_board_phy_init(dev, phy_addr, + eth_mdio_write, dw_reset_phy) < 0) + return -1; + if (dw_reset_phy(dev) < 0) return -1;
@@ -445,72 +470,32 @@ static int configure_phy(struct eth_device *dev) #if defined(CONFIG_DW_AUTONEG) timeout = CONFIG_AUTONEG_TIMEOUT; start = get_timer(0); - + puts("Waiting for PHY auto negotiation to complete"); while (get_timer(start) < timeout) { eth_mdio_read(dev, phy_addr, MII_BMSR, &bmsr); - if (bmsr & BMSR_ANEGCOMPLETE) + if (bmsr & BMSR_ANEGCOMPLETE) { + priv->phy_configured = 1; break; + } + + /* Print dot all 1s to show progress */ + if ((get_timer(start) % 100000) == 0) + putc('.');
/* Try again after 10usec */ udelay(10); };
- eth_mdio_read(dev, phy_addr, MII_LPA, &anlpar); - eth_mdio_read(dev, phy_addr, MII_STAT1000, &btsr); - - if (bmsr & BMSR_ANEGCOMPLETE) { - if (btsr & PHY_1000BTSR_1000FD) { - priv->speed = SPEED_1000M; - bmcr |= BMCR_SPEED1000; - priv->duplex = FULL_DUPLEX; - bmcr |= BMCR_FULLDPLX; - } else if (btsr & PHY_1000BTSR_1000HD) { - priv->speed = SPEED_1000M; - bmcr |= BMCR_SPEED1000; - priv->duplex = HALF_DUPLEX; - bmcr &= ~BMCR_FULLDPLX; - } else if (anlpar & LPA_100FULL) { - priv->speed = SPEED_100M; - bmcr |= BMCR_SPEED100; - priv->duplex = FULL_DUPLEX; - bmcr |= BMCR_FULLDPLX; - } else if (anlpar & LPA_100HALF) { - priv->speed = SPEED_100M; - bmcr |= BMCR_SPEED100; - priv->duplex = HALF_DUPLEX; - bmcr &= ~BMCR_FULLDPLX; - } else if (anlpar & LPA_10FULL) { - priv->speed = SPEED_10M; - bmcr &= ~BMCR_SPEED100; - priv->duplex = FULL_DUPLEX; - bmcr |= BMCR_FULLDPLX; - } else { - priv->speed = SPEED_10M; - bmcr &= ~BMCR_SPEED100; - priv->duplex = HALF_DUPLEX; - bmcr &= ~BMCR_FULLDPLX; - } - if (eth_mdio_write(dev, phy_addr, MII_BMCR, bmcr) < 0) - return -1; - } else - return -1; -#else - if (eth_mdio_read(dev, phy_addr, MII_BMCR, &ctrl) < 0) - return -1; - - if (ctrl & BMCR_FULLDPLX) - priv->duplex = FULL_DUPLEX; - else - priv->duplex = HALF_DUPLEX; - - if (ctrl & BMCR_SPEED1000) - priv->speed = SPEED_1000M; - else if (ctrl & BMCR_SPEED100) - priv->speed = SPEED_100M; + if (!(bmsr & BMSR_ANEGCOMPLETE)) + puts(" TIMEOUT!\n"); else - priv->speed = SPEED_10M; -#endif + puts(" done\n"); +#else priv->phy_configured = 1; +#endif + + priv->speed = miiphy_speed(dev->name, phy_addr); + priv->duplex = miiphy_duplex(dev->name, phy_addr);
return 0; } @@ -575,11 +560,6 @@ int designware_initialize(u32 id, ulong base_addr, u32 phy_addr, u32 interface) priv->phy_configured = 0; priv->interface = interface;
- if (mac_reset(dev) < 0) - return -1; - - configure_phy(dev); - dev->init = dw_eth_init; dev->send = dw_eth_send; dev->recv = dw_eth_recv; diff --git a/drivers/net/designware.h b/drivers/net/designware.h index 40020bf..d668f8f 100644 --- a/drivers/net/designware.h +++ b/drivers/net/designware.h @@ -240,6 +240,7 @@ struct dw_eth_dev { u32 tx_currdescnum; u32 rx_currdescnum; u32 phy_configured; + int link_printed; u32 padding;
struct dmamacdescr tx_mac_descrtable[CONFIG_TX_DESCR_NUM]; diff --git a/include/configs/spear-common.h b/include/configs/spear-common.h index 3366895..78cc33f 100644 --- a/include/configs/spear-common.h +++ b/include/configs/spear-common.h @@ -38,6 +38,7 @@ #define CONFIG_NET_MULTI #define CONFIG_PHY_RESET_DELAY 10000 /* in usec */ #define CONFIG_DW_AUTONEG +#define CONFIG_PHY_GIGE /* Include GbE speed/duplex detection */
/* USBD driver configuration */ #if defined(CONFIG_SPEAR_USBTTY)
Stefan,
On 5/8/2012 12:59 PM, Stefan Roese wrote:
This patch adds the following changes to designware ethernet driver found on the ST SPEAr SoC:
- Don't init MAC& PHY upon startup. This causes a delay, waiting for the auto negotiation to complete. And we don't want this delay to always happen. Especially not on platforms where ethernet is not used at all (e.g. booting via flash). Instead postpone the MAC / PHY configuration to the stage, where ethernet is first used.
- Add possibility for board specific PHY init code. This is needed for example on the X600 board, where the Vitesse PHY needs to be configured for GMII mode. This board specific PHY init is done via the function designware_board_phy_init(). And this driver now adds a weak default which can be overridden by board code.
- Use common functions miiphy_speed()& miiphy_duplex() to read link status from PHY.
- Print status and progress of auto negotiation.
- Print link status (speed, dupex) upon first usage.
Signed-off-by: Stefan Roesesr@denx.de Cc: Amit Virdiamit.virdi@st.com Cc: Vipin Kumarvipin.kumar@st.com Cc: Joe Hershbergerjoe.hershberger@gmail.com
drivers/net/designware.c | 114 +++++++++++++++++----------------------- drivers/net/designware.h | 1 + include/configs/spear-common.h | 1 + 3 files changed, 49 insertions(+), 67 deletions(-)
diff --git a/drivers/net/designware.c b/drivers/net/designware.c index e263022..945e57b 100644 --- a/drivers/net/designware.c +++ b/drivers/net/designware.c @@ -28,6 +28,7 @@ #include<common.h> #include<miiphy.h> #include<malloc.h> +#include<linux/compiler.h> #include<linux/err.h> #include<asm/io.h> #include "designware.h" @@ -153,6 +154,13 @@ static int dw_eth_init(struct eth_device *dev, bd_t *bis) if (priv->phy_configured != 1) configure_phy(dev);
- /* Print link status only once */
- if (!priv->link_printed) {
printf("ENET Speed is %d Mbps - %s duplex connection\n",priv->speed, (priv->duplex == HALF) ? "HALF" : "FULL");priv->link_printed = 1;- }
- /* Reset ethernet hardware */ if (mac_reset(dev)< 0) return -1;
@@ -168,17 +176,17 @@ static int dw_eth_init(struct eth_device *dev, bd_t *bis)
conf = FRAMEBURSTENABLE | DISABLERXOWN;
- if (priv->speed != SPEED_1000M)
if (priv->speed != 1000) conf |= MII_PORTSELECT;
if ((priv->interface != PHY_INTERFACE_MODE_MII)&& (priv->interface != PHY_INTERFACE_MODE_GMII)) {
if (priv->speed == SPEED_100M)
if (priv->speed == 100) conf |= FES_100;
- if (priv->duplex == FULL_DUPLEX)
if (priv->duplex == FULL) conf |= FULLDPLXMODE;
writel(conf,&mac_p->conf);
@@ -397,6 +405,16 @@ static int dw_reset_phy(struct eth_device *dev) return 0; }
+/*
- Add weak default function for board specific PHY configuration
- */
+int __weak designware_board_phy_init(struct eth_device *dev, int phy_addr,
int (*mii_write)(struct eth_device *, u8, u8, u16),int dw_reset_phy(struct eth_device *))+{
- return 0;
+}
- static int configure_phy(struct eth_device *dev) { struct dw_eth_dev *priv = dev->priv;
@@ -406,9 +424,6 @@ static int configure_phy(struct eth_device *dev) u16 bmsr; u32 timeout; ulong start;
- u16 anlpar, btsr;
-#else
u16 ctrl; #endif
#if defined(CONFIG_DW_SEARCH_PHY)
@@ -420,6 +435,16 @@ static int configure_phy(struct eth_device *dev) #else phy_addr = priv->address; #endif
- /*
* Some boards need board specific PHY initialization. This is* after the main driver init code but before the auto negotiation* is run.*/- if (designware_board_phy_init(dev, phy_addr,
eth_mdio_write, dw_reset_phy)< 0)return -1;- if (dw_reset_phy(dev)< 0) return -1;
@@ -445,72 +470,32 @@ static int configure_phy(struct eth_device *dev) #if defined(CONFIG_DW_AUTONEG) timeout = CONFIG_AUTONEG_TIMEOUT; start = get_timer(0);
- puts("Waiting for PHY auto negotiation to complete"); while (get_timer(start)< timeout) { eth_mdio_read(dev, phy_addr, MII_BMSR,&bmsr);
if (bmsr& BMSR_ANEGCOMPLETE)
if (bmsr& BMSR_ANEGCOMPLETE) {priv->phy_configured = 1; break;}/* Print dot all 1s to show progress */if ((get_timer(start) % 100000) == 0)putc('.');/* Try again after 10usec */ udelay(10); };
- eth_mdio_read(dev, phy_addr, MII_LPA,&anlpar);
- eth_mdio_read(dev, phy_addr, MII_STAT1000,&btsr);
- if (bmsr& BMSR_ANEGCOMPLETE) {
if (btsr& PHY_1000BTSR_1000FD) {priv->speed = SPEED_1000M;bmcr |= BMCR_SPEED1000;priv->duplex = FULL_DUPLEX;bmcr |= BMCR_FULLDPLX;} else if (btsr& PHY_1000BTSR_1000HD) {priv->speed = SPEED_1000M;bmcr |= BMCR_SPEED1000;priv->duplex = HALF_DUPLEX;bmcr&= ~BMCR_FULLDPLX;} else if (anlpar& LPA_100FULL) {priv->speed = SPEED_100M;bmcr |= BMCR_SPEED100;priv->duplex = FULL_DUPLEX;bmcr |= BMCR_FULLDPLX;} else if (anlpar& LPA_100HALF) {priv->speed = SPEED_100M;bmcr |= BMCR_SPEED100;priv->duplex = HALF_DUPLEX;bmcr&= ~BMCR_FULLDPLX;} else if (anlpar& LPA_10FULL) {priv->speed = SPEED_10M;bmcr&= ~BMCR_SPEED100;priv->duplex = FULL_DUPLEX;bmcr |= BMCR_FULLDPLX;} else {priv->speed = SPEED_10M;bmcr&= ~BMCR_SPEED100;priv->duplex = HALF_DUPLEX;bmcr&= ~BMCR_FULLDPLX;}if (eth_mdio_write(dev, phy_addr, MII_BMCR, bmcr)< 0)return -1;- } else
return -1;-#else
- if (eth_mdio_read(dev, phy_addr, MII_BMCR,&ctrl)< 0)
return -1;- if (ctrl& BMCR_FULLDPLX)
priv->duplex = FULL_DUPLEX;- else
priv->duplex = HALF_DUPLEX;- if (ctrl& BMCR_SPEED1000)
priv->speed = SPEED_1000M;- else if (ctrl& BMCR_SPEED100)
priv->speed = SPEED_100M;
- if (!(bmsr& BMSR_ANEGCOMPLETE))
puts(" TIMEOUT!\n");
priv->speed = SPEED_10M;-#endif
puts(" done\n");+#else priv->phy_configured = 1; +#endif
priv->speed = miiphy_speed(dev->name, phy_addr);
priv->duplex = miiphy_duplex(dev->name, phy_addr);
return 0; }
@@ -575,11 +560,6 @@ int designware_initialize(u32 id, ulong base_addr, u32 phy_addr, u32 interface) priv->phy_configured = 0; priv->interface = interface;
- if (mac_reset(dev)< 0)
return -1;- configure_phy(dev);
- dev->init = dw_eth_init; dev->send = dw_eth_send; dev->recv = dw_eth_recv;
diff --git a/drivers/net/designware.h b/drivers/net/designware.h index 40020bf..d668f8f 100644 --- a/drivers/net/designware.h +++ b/drivers/net/designware.h @@ -240,6 +240,7 @@ struct dw_eth_dev { u32 tx_currdescnum; u32 rx_currdescnum; u32 phy_configured;
int link_printed; u32 padding;
struct dmamacdescr tx_mac_descrtable[CONFIG_TX_DESCR_NUM];
diff --git a/include/configs/spear-common.h b/include/configs/spear-common.h index 3366895..78cc33f 100644 --- a/include/configs/spear-common.h +++ b/include/configs/spear-common.h @@ -38,6 +38,7 @@ #define CONFIG_NET_MULTI #define CONFIG_PHY_RESET_DELAY 10000 /* in usec */ #define CONFIG_DW_AUTONEG +#define CONFIG_PHY_GIGE /* Include GbE speed/duplex detection */
Shouldn't this come in a separate patch?
Otherwise, Acked-by: Amit Virdi amit.virdi@st.com
Regards Amit Virdi
/* USBD driver configuration */ #if defined(CONFIG_SPEAR_USBTTY)
Amit,
On Wednesday 09 May 2012 11:29:04 Amit Virdi wrote:
+++ b/include/configs/spear-common.h @@ -38,6 +38,7 @@
#define CONFIG_NET_MULTI #define CONFIG_PHY_RESET_DELAY 10000 /*
in usec */
#define CONFIG_DW_AUTONEG
+#define CONFIG_PHY_GIGE /* Include GbE speed/duplex
detection */
Shouldn't this come in a separate patch?
Not sure. This change is needed for this change in the designware driver to really work with 1000mbps links:
- Use common functions miiphy_speed()& miiphy_duplex() to read link status from PHY.
Splitting it into a separate commit will result in a non working ethernet driver for such links. Even though it should still compile and be git bisectable. I would prefer to keep it in one patch.
Otherwise, Acked-by: Amit Virdi amit.virdi@st.com
Thanks, Stefan
-- DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-0 Fax: (+49)-8142-66989-80 Email: office@denx.de
Joe,
On Wednesday 09 May 2012 11:36:56 Stefan Roese wrote:
+++ b/include/configs/spear-common.h @@ -38,6 +38,7 @@
#define CONFIG_NET_MULTI #define CONFIG_PHY_RESET_DELAY 10000 /*
in usec */
#define CONFIG_DW_AUTONEG
+#define CONFIG_PHY_GIGE /* Include GbE speed/duplex
detection */
Shouldn't this come in a separate patch?
Not sure. This change is needed for this change in the designware driver to really work with 1000mbps links:
- Use common functions miiphy_speed()& miiphy_duplex() to read link status from PHY.
Splitting it into a separate commit will result in a non working ethernet driver for such links. Even though it should still compile and be git bisectable. I would prefer to keep it in one patch.
Are you okay with this patch? If yes, perhaps it would be best if I would push this patch with the other SPEAr platform patches upstream. What do you think? Can I have your Acked-by?
Thanks, Stefan
-- DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-0 Fax: (+49)-8142-66989-80 Email: office@denx.de
Hi Stefan,
On Fri, Jun 1, 2012 at 7:55 AM, Stefan Roese sr@denx.de wrote:
Joe,
On Wednesday 09 May 2012 11:36:56 Stefan Roese wrote:
+++ b/include/configs/spear-common.h @@ -38,6 +38,7 @@
#define CONFIG_NET_MULTI #define CONFIG_PHY_RESET_DELAY 10000 /*
in usec */
#define CONFIG_DW_AUTONEG
+#define CONFIG_PHY_GIGE /* Include GbE speed/duplex
detection */
Shouldn't this come in a separate patch?
Not sure. This change is needed for this change in the designware driver to really work with 1000mbps links:
- Use common functions miiphy_speed()& miiphy_duplex() to read
link status from PHY.
Splitting it into a separate commit will result in a non working ethernet driver for such links. Even though it should still compile and be git bisectable. I would prefer to keep it in one patch.
Are you okay with this patch? If yes, perhaps it would be best if I would push this patch with the other SPEAr platform patches upstream. What do you think? Can I have your Acked-by?
I'm not especially fond of adding prints, but it seems in line with the existing behavior.
Acked-by: Joe Hershberger joe.hershberger@gmail.com
participants (3)
-
Amit Virdi -
Joe Hershberger -
Stefan Roese