Current ms7724se LAN doesn't work

Current ms7724se LAN doesn't work

[Kuninori Morimoto]

Dear all

I got latest kernel from Paul's git,
and current ms7724se LAN doesn't work.

When If I remove 
archdata = {
         .hwblk_id = HWBLK_ETHER,
}
from ${LINUX}/arch/sh/boards/mach-se/7724/setup.c
then, it start works for me.

Are there any CONFIG to use SH-ETH or HWBLK ?

Best regards
--
Kuninori Morimoto
 

Re: Current ms7724se LAN doesn't work

[Paul Mundt]

On Mon, Aug 24, 2009 at 05:19:47PM +0900, Kuninori Morimoto wrote:
> 
> Dear all
> 
> I got latest kernel from Paul's git,
> and current ms7724se LAN doesn't work.
> 
> When If I remove 
> archdata = {
>          .hwblk_id = HWBLK_ETHER,
> }
> from ${LINUX}/arch/sh/boards/mach-se/7724/setup.c
> then, it start works for me.
> 
> Are there any CONFIG to use SH-ETH or HWBLK ?
> 
This would be a runtime PM issue, which SH-Mobile unconditionally
selects. The problem is that the module is most likely disabled via its
MSTP bit under the new scheme, so sh-eth will need to tie in to the
runtime PM framework to get that resolved.

Can you give this a try?

---

 drivers/net/sh_eth.c |   28 ++++++++++++++++++++++++++--
 1 file changed, 26 insertions(+), 2 deletions(-)

diff --git a/drivers/net/sh_eth.c b/drivers/net/sh_eth.c
index a2d82dd..c195baa 100644
--- a/drivers/net/sh_eth.c
+++ b/drivers/net/sh_eth.c
@@ -30,7 +30,7 @@
 #include <linux/phy.h>
 #include <linux/cache.h>
 #include <linux/io.h>
-
+#include <linux/pm_runtime.h>
 #include "sh_eth.h"
 
 /* There is CPU dependent code */
@@ -1006,6 +1006,8 @@ static int sh_eth_open(struct net_device *ndev)
 	int ret = 0;
 	struct sh_eth_private *mdp = netdev_priv(ndev);
 
+	pm_runtime_get_sync(&ndev->dev);
+
 	ret = request_irq(ndev->irq, &sh_eth_interrupt,
 #if defined(CONFIG_CPU_SUBTYPE_SH7763) || defined(CONFIG_CPU_SUBTYPE_SH7764)
 				IRQF_SHARED,
@@ -1173,6 +1175,8 @@ static int sh_eth_close(struct net_device *ndev)
 	ringsize = sizeof(struct sh_eth_txdesc) * TX_RING_SIZE;
 	dma_free_coherent(NULL, ringsize, mdp->tx_ring, mdp->tx_desc_dma);
 
+	pm_runtime_put_sync(&ndev->dev);
+
 	return 0;
 }
 
@@ -1404,6 +1408,7 @@ static int sh_eth_drv_probe(struct platform_device *pdev)
 
 	mdp = netdev_priv(ndev);
 	spin_lock_init(&mdp->lock);
+	pm_runtime_enable(&pdev->dev);
 
 	pd = (struct sh_eth_plat_data *)(pdev->dev.platform_data);
 	/* get PHY ID */
@@ -1477,18 +1482,37 @@ static int sh_eth_drv_remove(struct platform_device *pdev)
 	sh_mdio_release(ndev);
 	unregister_netdev(ndev);
 	flush_scheduled_work();
-
+	pm_runtime_disable(&pdev->dev);
 	free_netdev(ndev);
 	platform_set_drvdata(pdev, NULL);
 
 	return 0;
 }
 
+static int sh_eth_runtime_nop(struct device *dev)
+{
+	/*
+	 * Runtime PM callback shared between ->runtime_suspend()
+	 * and ->runtime_resume(). Simply returns success.
+	 *
+	 * This driver re-initializes all registers after
+	 * pm_runtime_get_sync() anyway so there is no need
+	 * to save and restore registers here.
+	 */
+	return 0;
+}
+
+static struct dev_pm_ops sh_eth_dev_pm_ops = {
+	.runtime_suspend = sh_eth_runtime_nop,
+	.runtime_resume = sh_eth_runtime_nop,
+};
+
 static struct platform_driver sh_eth_driver = {
 	.probe = sh_eth_drv_probe,
 	.remove = sh_eth_drv_remove,
 	.driver = {
 		   .name = CARDNAME,
+		   .pm = &sh_eth_dev_pm_ops,
 	},
 };
 

Re: Current ms7724se LAN doesn't work

[Kuninori Morimoto]

Dear Paul

Thank you for your patch

> This would be a runtime PM issue, which SH-Mobile unconditionally
> selects. The problem is that the module is most likely disabled via its
> MSTP bit under the new scheme, so sh-eth will need to tie in to the
> runtime PM framework to get that resolved.
> 
> Can you give this a try?

hmmm..
I tried this patch, but it still doesn't work.
log is 

----------
(snip)
usbcore: registered new interface driver usbhid                                 
usbhid: v2.6:USB HID core driver                                                
heartbeat: version 0.1.1 loaded                                                 
TCP cubic registered                                                            
NET: Registered protocol family 17                                              
RPC: Registered udp transport module.                                           
RPC: Registered tcp transport module.                                           
/opt/usr/src/WORK/morimoto/gitlinux/sh-2.6/drivers/rtc/hctosys.c: unable to ope)
IP-Config: Failed to open eth0                                                  
eth1: link down                                                                 
Sending DHCP requests ......
----------

Best regards
--
Kuninori Morimoto
 

Re: Current ms7724se LAN doesn't work

[Paul Mundt]

On Mon, Aug 24, 2009 at 06:45:45PM +0900, Kuninori Morimoto wrote:
> 
> Dear Paul
> 
> Thank you for your patch
> 
> > This would be a runtime PM issue, which SH-Mobile unconditionally
> > selects. The problem is that the module is most likely disabled via its
> > MSTP bit under the new scheme, so sh-eth will need to tie in to the
> > runtime PM framework to get that resolved.
> > 
> > Can you give this a try?
> 
> hmmm..
> I tried this patch, but it still doesn't work.
> log is 
> 
Ok, we'll have to let Magnus take a look at it then, as it's certainly
runtime PM related.

Re: Current ms7724se LAN doesn't work

[Kuninori Morimoto]

Dear Paul

> > I tried this patch, but it still doesn't work.
> > log is 
> > 
> Ok, we'll have to let Magnus take a look at it then, as it's certainly
> runtime PM related.

OK thanks

Best regards
--
Kuninori Morimoto
 

Re: Current ms7724se LAN doesn't work

[Magnus Damm]

On Mon, Aug 24, 2009 at 6:55 PM, Paul Mundt<lethal@linux-sh.org> wrote:
> On Mon, Aug 24, 2009 at 06:45:45PM +0900, Kuninori Morimoto wrote:
>>
>> Dear Paul
>>
>> Thank you for your patch
>>
>> > This would be a runtime PM issue, which SH-Mobile unconditionally
>> > selects. The problem is that the module is most likely disabled via its
>> > MSTP bit under the new scheme, so sh-eth will need to tie in to the
>> > runtime PM framework to get that resolved.
>> >
>> > Can you give this a try?
>>
>> hmmm..
>> I tried this patch, but it still doesn't work.
>> log is
>>
> Ok, we'll have to let Magnus take a look at it then, as it's certainly
> runtime PM related.

It looks like the driver is accessing hardware from probe, so you need
to force a resume after enable. The LCDC driver does just that:

       pm_runtime_enable(priv->dev);
       pm_runtime_resume(priv->dev);

Hopefully that should do the trick. Can you try this fix on top of
Paul's patch Morimoto-san?

Cheers,

/ magnus

Re: Current ms7724se LAN doesn't work

[Kuninori Morimoto]

Hi Magnus

Thank you for your hard work

> It looks like the driver is accessing hardware from probe, so you need
> to force a resume after enable. The LCDC driver does just that:
> 
>        pm_runtime_enable(priv->dev);
>        pm_runtime_resume(priv->dev);
> 
> Hopefully that should do the trick. Can you try this fix on top of
> Paul's patch Morimoto-san?

hmm...
I tried Paul's patch + pm_runtime_resume
But it still doesn't work for me...

Best regards
--
Kuninori Morimoto

Re: Current ms7724se LAN doesn't work

[Magnus Damm]

[-- Attachment #1: Type: text/plain, Size: 588 bytes --]

Hi Morimoto-san,

[CC Iwamatsu-san]

On Wed, Aug 26, 2009 at 7:09 PM, Kuninori Morimoto
<morimoto.kuninori@renesas.com> wrote:
> I tried Paul's patch + pm_runtime_resume
> But it still doesn't work for me...

Can you give the attached patches a try?

Apply in this order:
linux-2.6.32-rc-net-sh_eth-platform-data-mac-20091007b.patch
linux-2.6.32-rc-net-sh_eth-runtime-pm-20091007.patch
linux-2.6.32-rc-sh-ecovec-sh_eth-pdata-mac-20091007.patch
linux-2.6.32-rc-sh-ecovec-sh_eth-hwblk_id-20091007.patch

They seem to work fine on ecovec for me. Do they solve your issue?

Cheers,

/ magnus

[-- Attachment #2: linux-2.6.32-rc-net-sh_eth-platform-data-mac-20091007b.patch --]
[-- Type: application/octet-stream, Size: 2167 bytes --]

From: Magnus Damm <damm@opensource.se>

Extend the sh_eth driver to allow passing the mac address
using the platform data structure. This to simplify board
setup code.

Signed-off-by: Magnus Damm <damm@opensource.se>
---

 arch/sh/include/asm/sh_eth.h |    1 +
 drivers/net/sh_eth.c         |   20 ++++++++++++--------
 2 files changed, 13 insertions(+), 8 deletions(-)

--- 0001/arch/sh/include/asm/sh_eth.h
+++ work/arch/sh/include/asm/sh_eth.h	2009-10-07 15:04:57.000000000 +0900
@@ -7,6 +7,7 @@ struct sh_eth_plat_data {
 	int phy;
 	int edmac_endian;
 
+	unsigned char mac_addr[6];
 	unsigned no_ether_link:1;
 	unsigned ether_link_active_low:1;
 };
--- 0001/drivers/net/sh_eth.c
+++ work/drivers/net/sh_eth.c	2009-10-07 16:54:29.000000000 +0900
@@ -298,16 +298,20 @@ static void update_mac_address(struct ne
  * When you want use this device, you must set MAC address in bootloader.
  *
  */
-static void read_mac_address(struct net_device *ndev)
+static void read_mac_address(struct net_device *ndev, unsigned char *mac)
 {
 	u32 ioaddr = ndev->base_addr;
 
-	ndev->dev_addr[0] = (ctrl_inl(ioaddr + MAHR) >> 24);
-	ndev->dev_addr[1] = (ctrl_inl(ioaddr + MAHR) >> 16) & 0xFF;
-	ndev->dev_addr[2] = (ctrl_inl(ioaddr + MAHR) >> 8) & 0xFF;
-	ndev->dev_addr[3] = (ctrl_inl(ioaddr + MAHR) & 0xFF);
-	ndev->dev_addr[4] = (ctrl_inl(ioaddr + MALR) >> 8) & 0xFF;
-	ndev->dev_addr[5] = (ctrl_inl(ioaddr + MALR) & 0xFF);
+	if (mac[0] || mac[1] || mac[2] || mac[3] || mac[4] || mac[5]) {
+		memcpy(ndev->dev_addr, mac, 6);
+	} else {
+		ndev->dev_addr[0] = (ctrl_inl(ioaddr + MAHR) >> 24);
+		ndev->dev_addr[1] = (ctrl_inl(ioaddr + MAHR) >> 16) & 0xFF;
+		ndev->dev_addr[2] = (ctrl_inl(ioaddr + MAHR) >> 8) & 0xFF;
+		ndev->dev_addr[3] = (ctrl_inl(ioaddr + MAHR) & 0xFF);
+		ndev->dev_addr[4] = (ctrl_inl(ioaddr + MALR) >> 8) & 0xFF;
+		ndev->dev_addr[5] = (ctrl_inl(ioaddr + MALR) & 0xFF);
+	}
 }
 
 struct bb_info {
@@ -1427,7 +1431,7 @@ static int sh_eth_drv_probe(struct platf
 	mdp->post_fw = POST_FW >> (devno << 1);
 
 	/* read and set MAC address */
-	read_mac_address(ndev);
+	read_mac_address(ndev, pd->mac_addr);
 
 	/* First device only init */
 	if (!devno) {

[-- Attachment #3: linux-2.6.32-rc-net-sh_eth-runtime-pm-20091007.patch --]
[-- Type: application/octet-stream, Size: 3498 bytes --]

From: Magnus Damm <damm@opensource.se>

Add Runtime PM support to the sh_eth driver.

The clock to the ethernet hardware block will be
enabled as long as the network device is up.

Signed-off-by: Magnus Damm <damm@opensource.se>
---

 drivers/net/sh_eth.c |   35 +++++++++++++++++++++++++++++++++--
 drivers/net/sh_eth.h |    1 +
 2 files changed, 34 insertions(+), 2 deletions(-)

--- 0003/drivers/net/sh_eth.c
+++ work/drivers/net/sh_eth.c	2009-10-07 17:25:34.000000000 +0900
@@ -30,7 +30,7 @@
 #include <linux/phy.h>
 #include <linux/cache.h>
 #include <linux/io.h>
-
+#include <linux/pm_runtime.h>
 #include "sh_eth.h"
 
 /* There is CPU dependent code */
@@ -1012,6 +1012,8 @@ static int sh_eth_open(struct net_device
 	int ret = 0;
 	struct sh_eth_private *mdp = netdev_priv(ndev);
 
+	pm_runtime_get_sync(&mdp->pdev->dev);
+
 	ret = request_irq(ndev->irq, &sh_eth_interrupt,
 #if defined(CONFIG_CPU_SUBTYPE_SH7763) || defined(CONFIG_CPU_SUBTYPE_SH7764)
 				IRQF_SHARED,
@@ -1048,6 +1050,7 @@ static int sh_eth_open(struct net_device
 
 out_free_irq:
 	free_irq(ndev->irq, ndev);
+	pm_runtime_put_sync(&mdp->pdev->dev);
 	return ret;
 }
 
@@ -1179,6 +1182,8 @@ static int sh_eth_close(struct net_devic
 	ringsize = sizeof(struct sh_eth_txdesc) * TX_RING_SIZE;
 	dma_free_coherent(NULL, ringsize, mdp->tx_ring, mdp->tx_desc_dma);
 
+	pm_runtime_put_sync(&mdp->pdev->dev);
+
 	return 0;
 }
 
@@ -1187,6 +1192,8 @@ static struct net_device_stats *sh_eth_g
 	struct sh_eth_private *mdp = netdev_priv(ndev);
 	u32 ioaddr = ndev->base_addr;
 
+	pm_runtime_get_sync(&mdp->pdev->dev);
+
 	mdp->stats.tx_dropped += ctrl_inl(ioaddr + TROCR);
 	ctrl_outl(0, ioaddr + TROCR);	/* (write clear) */
 	mdp->stats.collisions += ctrl_inl(ioaddr + CDCR);
@@ -1202,6 +1209,8 @@ static struct net_device_stats *sh_eth_g
 	mdp->stats.tx_carrier_errors += ctrl_inl(ioaddr + CNDCR);
 	ctrl_outl(0, ioaddr + CNDCR);	/* (write clear) */
 #endif
+	pm_runtime_put_sync(&mdp->pdev->dev);
+
 	return &mdp->stats;
 }
 
@@ -1410,6 +1419,9 @@ static int sh_eth_drv_probe(struct platf
 
 	mdp = netdev_priv(ndev);
 	spin_lock_init(&mdp->lock);
+	mdp->pdev = pdev;
+	pm_runtime_enable(&pdev->dev);
+	pm_runtime_resume(&pdev->dev);
 
 	pd = (struct sh_eth_plat_data *)(pdev->dev.platform_data);
 	/* get PHY ID */
@@ -1485,18 +1497,37 @@ static int sh_eth_drv_remove(struct plat
 	sh_mdio_release(ndev);
 	unregister_netdev(ndev);
 	flush_scheduled_work();
-
+	pm_runtime_disable(&pdev->dev);
 	free_netdev(ndev);
 	platform_set_drvdata(pdev, NULL);
 
 	return 0;
 }
 
+static int sh_eth_runtime_nop(struct device *dev)
+{
+	/*
+	 * Runtime PM callback shared between ->runtime_suspend()
+	 * and ->runtime_resume(). Simply returns success.
+	 *
+	 * This driver re-initializes all registers after
+	 * pm_runtime_get_sync() anyway so there is no need
+	 * to save and restore registers here.
+	 */
+	return 0;
+}
+
+static struct dev_pm_ops sh_eth_dev_pm_ops = {
+	.runtime_suspend = sh_eth_runtime_nop,
+	.runtime_resume = sh_eth_runtime_nop,
+};
+
 static struct platform_driver sh_eth_driver = {
 	.probe = sh_eth_drv_probe,
 	.remove = sh_eth_drv_remove,
 	.driver = {
 		   .name = CARDNAME,
+		   .pm = &sh_eth_dev_pm_ops,
 	},
 };
 
--- 0001/drivers/net/sh_eth.h
+++ work/drivers/net/sh_eth.h	2009-10-07 17:22:24.000000000 +0900
@@ -703,6 +703,7 @@ struct sh_eth_cpu_data {
 };
 
 struct sh_eth_private {
+	struct platform_device *pdev;
 	struct sh_eth_cpu_data *cd;
 	dma_addr_t rx_desc_dma;
 	dma_addr_t tx_desc_dma;

[-- Attachment #4: linux-2.6.32-rc-sh-ecovec-sh_eth-pdata-mac-20091007.patch --]
[-- Type: application/octet-stream, Size: 1757 bytes --]

From: Magnus Damm <damm@opensource.se>

Convert the ecovec24 board code to pass the mac
address to the sh_eth driver using platform data.
Also, remove the static clock to allow Runtime PM.

Signed-off-by: Magnus Damm <damm@opensource.se>
---

 arch/sh/boards/mach-ecovec24/setup.c |   33 ++++-----------------------------
 1 file changed, 4 insertions(+), 29 deletions(-)

--- 0003/arch/sh/boards/mach-ecovec24/setup.c
+++ work/arch/sh/boards/mach-ecovec24/setup.c	2009-10-07 15:11:51.000000000 +0900
@@ -414,12 +414,9 @@ static u8 mac_read(struct i2c_adapter *a
 	return buf;
 }
 
-#define MAC_LEN 6
-static void __init sh_eth_init(void)
+static void __init sh_eth_init(struct sh_eth_plat_data *pd)
 {
 	struct i2c_adapter *a = i2c_get_adapter(1);
-	struct clk *eth_clk;
-	u8 mac[MAC_LEN];
 	int i;
 
 	if (!a) {
@@ -427,33 +424,11 @@ static void __init sh_eth_init(void)
 		return;
 	}
 
-	eth_clk = clk_get(NULL, "eth0");
-	if (!eth_clk) {
-		pr_err("can not get eth0 clk\n");
-		return;
-	}
-
 	/* read MAC address frome EEPROM */
-	for (i = 0; i < MAC_LEN; i++) {
-		mac[i] = mac_read(a, 0x10 + i);
+	for (i = 0; i < sizeof(pd->mac_addr); i++) {
+		pd->mac_addr[i] = mac_read(a, 0x10 + i);
 		msleep(10);
 	}
-
-	/* clock enable */
-	clk_enable(eth_clk);
-
-	/* reset sh-eth */
-	ctrl_outl(0x1, SH_ETH_ADDR + 0x0);
-
-	/* set MAC addr */
-	ctrl_outl((mac[0] << 24) |
-		  (mac[1] << 16) |
-		  (mac[2] <<  8) |
-		  (mac[3] <<  0), SH_ETH_MAHR);
-	ctrl_outl((mac[4] <<  8) |
-		  (mac[5] <<  0), SH_ETH_MALR);
-
-	clk_put(eth_clk);
 }
 
 #define PORT_HIZA 0xA4050158
@@ -665,7 +640,7 @@ arch_initcall(arch_setup);
 
 static int __init devices_setup(void)
 {
-	sh_eth_init();
+	sh_eth_init(&sh_eth_plat);
 	return 0;
 }
 device_initcall(devices_setup);

[-- Attachment #5: linux-2.6.32-rc-sh-ecovec-sh_eth-hwblk_id-20091007.patch --]
[-- Type: application/octet-stream, Size: 616 bytes --]

From: Magnus Damm <damm@opensource.se>

Add HWBLK_ETHER to the sh_eth platform device
to allow Runtime PM of the ethernet hardware.

Signed-off-by: Magnus Damm <damm@opensource.se>
---

 arch/sh/boards/mach-ecovec24/setup.c |    3 +++
 1 file changed, 3 insertions(+)

--- 0004/arch/sh/boards/mach-ecovec24/setup.c
+++ work/arch/sh/boards/mach-ecovec24/setup.c	2009-10-07 16:58:33.000000000 +0900
@@ -132,6 +132,9 @@ static struct platform_device sh_eth_dev
 	},
 	.num_resources = ARRAY_SIZE(sh_eth_resources),
 	.resource = sh_eth_resources,
+	.archdata = {
+		.hwblk_id = HWBLK_ETHER,
+	},
 };
 
 /* USB0 host */

Re: Current ms7724se LAN doesn't work

[Kuninori Morimoto]

Hi Magnus

sorry for my late response

> Can you give the attached patches a try?
> 
> Apply in this order:
> linux-2.6.32-rc-net-sh_eth-platform-data-mac-20091007b.patch
> linux-2.6.32-rc-net-sh_eth-runtime-pm-20091007.patch
> linux-2.6.32-rc-sh-ecovec-sh_eth-pdata-mac-20091007.patch
> linux-2.6.32-rc-sh-ecovec-sh_eth-hwblk_id-20091007.patch
> 
> They seem to work fine on ecovec for me. Do they solve your issue?

How cool !!
It works well on ecovec / ms7724se !!
Thank you

Tested-by: Kuninori Morimoto <morimoto.kuninori@renesas.com>

Best regards
--
Kuninori Morimoto

Re: Current ms7724se LAN doesn't work

[Magnus Damm]

Hi Morimoto-san,

On Thu, Oct 8, 2009 at 1:45 PM, Kuninori Morimoto
<morimoto.kuninori@renesas.com> wrote:
>
> Hi Magnus
>
> sorry for my late response

No worries, you are not late.

>> Can you give the attached patches a try?
>>
>> Apply in this order:
>> linux-2.6.32-rc-net-sh_eth-platform-data-mac-20091007b.patch
>> linux-2.6.32-rc-net-sh_eth-runtime-pm-20091007.patch
>> linux-2.6.32-rc-sh-ecovec-sh_eth-pdata-mac-20091007.patch
>> linux-2.6.32-rc-sh-ecovec-sh_eth-hwblk_id-20091007.patch
>>
>> They seem to work fine on ecovec for me. Do they solve your issue?
>
> How cool !!
> It works well on ecovec / ms7724se !!
> Thank you
>
> Tested-by: Kuninori Morimoto <morimoto.kuninori@renesas.com>

Excellent, thanks for testing!

/ magnus

Re: Current ms7724se LAN doesn't work

[Paul Mundt]

On Thu, Oct 08, 2009 at 03:51:43PM +0900, Magnus Damm wrote:
> Hi Morimoto-san,
> 
> On Thu, Oct 8, 2009 at 1:45 PM, Kuninori Morimoto
> <morimoto.kuninori@renesas.com> wrote:
> >
> > Hi Magnus
> >
> > sorry for my late response
> 
> No worries, you are not late.
> 
> >> Can you give the attached patches a try?
> >>
> >> Apply in this order:
> >> linux-2.6.32-rc-net-sh_eth-platform-data-mac-20091007b.patch
> >> linux-2.6.32-rc-net-sh_eth-runtime-pm-20091007.patch
> >> linux-2.6.32-rc-sh-ecovec-sh_eth-pdata-mac-20091007.patch
> >> linux-2.6.32-rc-sh-ecovec-sh_eth-hwblk_id-20091007.patch
> >>
> >> They seem to work fine on ecovec for me. Do they solve your issue?
> >
> > How cool !!
> > It works well on ecovec / ms7724se !!
> > Thank you
> >
> > Tested-by: Kuninori Morimoto <morimoto.kuninori@renesas.com>
> 
> Excellent, thanks for testing!
> 
Please send the driver bits on to DaveM, and I'll merge the platform bits
independently.