More updates to bcm63xx support (resend)

More updates to bcm63xx support (resend)

[Maxime Bizon]

[Resend, I forgot to generate an actual patch series and missed one
patch doing so, sorry for the noise]

Hi Ralf,

The following patch series (on top of Florian's latest post) are
mostly trivial fixes & cleanup requested by subsystem maintainers.

Patch 2 and 4 are actual bugfixes.

Florian has some more patches to send to you, I will wait for them to
be folded in the bcm63xx tree and then repost the whole serie.

I only got feedback from USB maintainer, and addressed most of them.
Davem has accidentally dropped the patch from its patchwork, so I
think just resubmitting will do. Nothing from serial & pcmcia.

The ethernet PHY support patch should be dropped, the only feature we
get is IRQ support which after field testing seems to be buggy. Maybe
later.

Is it ok for you if I ask for the patches to go through your tree once
they get acked ?

Thanks !

[PATCH 01/10] bcm63xx: remove duplicate init fields.

[Maxime Bizon]

This patch removes duplicate init fields in resource.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
---
 arch/mips/bcm63xx/dev-enet.c |    2 --
 1 files changed, 0 insertions(+), 2 deletions(-)

diff --git a/arch/mips/bcm63xx/dev-enet.c b/arch/mips/bcm63xx/dev-enet.c
index 51c2e5a..188fa66 100644
--- a/arch/mips/bcm63xx/dev-enet.c
+++ b/arch/mips/bcm63xx/dev-enet.c
@@ -42,12 +42,10 @@ static struct resource enet0_res[] = {
 	},
 	{
 		.start		= -1, /* filled at runtime */
-		.start		= IRQ_ENET0_RXDMA,
 		.flags		= IORESOURCE_IRQ,
 	},
 	{
 		.start		= -1, /* filled at runtime */
-		.start		= IRQ_ENET0_TXDMA,
 		.flags		= IORESOURCE_IRQ,
 	},
 };
-- 
1.6.0.4

[PATCH 02/10] bcm63xx: use napi_complete instead of __napi_complete

[Maxime Bizon]

We're not disabling IRQ, so we must call the irq safe flavour of
napi_complete.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
---
 drivers/net/bcm63xx_enet.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/net/bcm63xx_enet.c b/drivers/net/bcm63xx_enet.c
index a91f909..20e08ef 100644
--- a/drivers/net/bcm63xx_enet.c
+++ b/drivers/net/bcm63xx_enet.c
@@ -450,7 +450,7 @@ static int bcm_enet_poll(struct napi_struct *napi, int budget)
 
 	/* no more packet in rx/tx queue, remove device from poll
 	 * queue */
-	__napi_complete(napi);
+	napi_complete(napi);
 
 	/* restore rx/tx interrupt */
 	enet_dma_writel(priv, ENETDMA_IR_PKTDONE_MASK,
-- 
1.6.0.4

[PATCH 03/10] bcm63xx: convert bcm63xx_enet to netdev ops.

[Maxime Bizon]

This patch makes bcm63xx_enet driver use netdevice ops.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
---
 drivers/net/bcm63xx_enet.c |   24 ++++++++++++++----------
 1 files changed, 14 insertions(+), 10 deletions(-)

diff --git a/drivers/net/bcm63xx_enet.c b/drivers/net/bcm63xx_enet.c
index 20e08ef..36324b3 100644
--- a/drivers/net/bcm63xx_enet.c
+++ b/drivers/net/bcm63xx_enet.c
@@ -1551,6 +1551,19 @@ static void bcm_enet_hw_preinit(struct bcm_enet_priv *priv)
 	enet_writel(priv, val, ENET_MIBCTL_REG);
 }
 
+static const struct net_device_ops bcm_enet_ops = {
+	.ndo_open		= bcm_enet_open,
+	.ndo_stop		= bcm_enet_stop,
+	.ndo_start_xmit		= bcm_enet_start_xmit,
+	.ndo_get_stats		= bcm_enet_get_stats,
+	.ndo_set_mac_address	= bcm_enet_set_mac_address,
+	.ndo_set_multicast_list = bcm_enet_set_multicast_list,
+	.ndo_do_ioctl		= bcm_enet_ioctl,
+#ifdef CONFIG_NET_POLL_CONTROLLER
+	.ndo_poll_controller = bcm_enet_netpoll,
+#endif
+};
+
 /*
  * allocate netdevice, request register memory and register device.
  */
@@ -1716,17 +1729,8 @@ static int __devinit bcm_enet_probe(struct platform_device *pdev)
 		enet_writel(priv, 0, ENET_MIB_REG(i));
 
 	/* register netdevice */
-	dev->open = bcm_enet_open;
-	dev->stop = bcm_enet_stop;
-	dev->hard_start_xmit = bcm_enet_start_xmit;
-	dev->get_stats = bcm_enet_get_stats;
-	dev->set_mac_address = bcm_enet_set_mac_address;
-	dev->set_multicast_list = bcm_enet_set_multicast_list;
+	dev->netdev_ops = &bcm_enet_ops;
 	netif_napi_add(dev, &priv->napi, bcm_enet_poll, 16);
-	dev->do_ioctl = bcm_enet_ioctl;
-#ifdef CONFIG_NET_POLL_CONTROLLER
-	dev->poll_controller = bcm_enet_netpoll;
-#endif
 
 	SET_ETHTOOL_OPS(dev, &bcm_enet_ethtool_ops);
 
-- 
1.6.0.4

[PATCH 04/10] bcm63xx: limit number of usb port to 1.

[Maxime Bizon]

This patch disables the use of more than one USB port. Hardware has
two ports, but one may be shared with USB slave port.

Until we have this information in the platform data, don't use the
second port.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
---
 drivers/usb/host/ohci-bcm63xx.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/drivers/usb/host/ohci-bcm63xx.c b/drivers/usb/host/ohci-bcm63xx.c
index 08807d9..74f432f 100644
--- a/drivers/usb/host/ohci-bcm63xx.c
+++ b/drivers/usb/host/ohci-bcm63xx.c
@@ -20,6 +20,8 @@ static int __devinit ohci_bcm63xx_start(struct usb_hcd *hcd)
 	struct ohci_hcd *ohci = hcd_to_ohci(hcd);
 	int ret;
 
+	ohci->num_ports = 1;
+
 	ret = ohci_init(ohci);
 	if (ret < 0)
 		return ret;
-- 
1.6.0.4

[PATCH 05/10] bcm63xx: use platform_get_irq in ohci-bcm63xx.c

[Maxime Bizon]

As requested by USB maintainer, use platform_get_irq instead of
platform_get_resource.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
---
 drivers/usb/host/ohci-bcm63xx.c |   10 +++++-----
 1 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/usb/host/ohci-bcm63xx.c b/drivers/usb/host/ohci-bcm63xx.c
index 74f432f..f0e4639 100644
--- a/drivers/usb/host/ohci-bcm63xx.c
+++ b/drivers/usb/host/ohci-bcm63xx.c
@@ -58,15 +58,15 @@ static const struct hc_driver ohci_bcm63xx_hc_driver = {
 
 static int __devinit ohci_hcd_bcm63xx_drv_probe(struct platform_device *pdev)
 {
-	struct resource *res_mem, *res_irq;
+	struct resource *res_mem;
 	struct usb_hcd *hcd;
 	struct ohci_hcd *ohci;
 	u32 reg;
-	int ret;
+	int ret, irq;
 
 	res_mem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
-	res_irq = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
-	if (!res_mem || !res_irq)
+	irq = platform_get_irq(pdev, 0);
+	if (!res_mem || irq < 0)
 		return -ENODEV;
 
 	if (BCMCPU_IS_6348()) {
@@ -114,7 +114,7 @@ static int __devinit ohci_hcd_bcm63xx_drv_probe(struct platform_device *pdev)
 		OHCI_QUIRK_FRAME_NO;
 	ohci_hcd_init(ohci);
 
-	ret = usb_add_hcd(hcd, res_irq->start, IRQF_DISABLED);
+	ret = usb_add_hcd(hcd, irq, IRQF_DISABLED);
 	if (ret)
 		goto out2;
 
-- 
1.6.0.4

[PATCH 06/10] bcm63xx: don't set bus type in ohci-bcm63xx.c

[Maxime Bizon]

The platform code already sets the bus type, so don't do it.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
---
 drivers/usb/host/ohci-bcm63xx.c |    1 -
 1 files changed, 0 insertions(+), 1 deletions(-)

diff --git a/drivers/usb/host/ohci-bcm63xx.c b/drivers/usb/host/ohci-bcm63xx.c
index f0e4639..d48c8ac 100644
--- a/drivers/usb/host/ohci-bcm63xx.c
+++ b/drivers/usb/host/ohci-bcm63xx.c
@@ -154,7 +154,6 @@ static struct platform_driver ohci_hcd_bcm63xx_driver = {
 	.driver		= {
 		.name	= "bcm63xx_ohci",
 		.owner	= THIS_MODULE,
-		.bus	= &platform_bus_type
 	},
 };
 
-- 
1.6.0.4

[PATCH 07/10] bcm63xx: clarify meaning of the magical value in ohci-bcm63xx.c

[Maxime Bizon]

USB maintainer asked for clarification  of the magic value used during
USB init. Be clear about the source of it.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
---
 drivers/usb/host/ohci-bcm63xx.c |    4 +++-
 1 files changed, 3 insertions(+), 1 deletions(-)

diff --git a/drivers/usb/host/ohci-bcm63xx.c b/drivers/usb/host/ohci-bcm63xx.c
index d48c8ac..bd66d5a 100644
--- a/drivers/usb/host/ohci-bcm63xx.c
+++ b/drivers/usb/host/ohci-bcm63xx.c
@@ -85,7 +85,9 @@ static int __devinit ohci_hcd_bcm63xx_drv_probe(struct platform_device *pdev)
 		reg &= ~USBH_PRIV_SWAP_OHCI_ENDN_MASK;
 		reg |= USBH_PRIV_SWAP_OHCI_DATA_MASK;
 		bcm_rset_writel(RSET_USBH_PRIV, reg, USBH_PRIV_SWAP_REG);
-		/* don't ask... */
+		/* the magic value comes for the original vendor BSP
+		 * and is needed for USB to work. Datasheet does not
+		 * help, so the magic value is used as-is. */
 		bcm_rset_writel(RSET_USBH_PRIV, 0x1c0020, USBH_PRIV_TEST_REG);
 	} else
 		return 0;
-- 
1.6.0.4

[PATCH 08/10] bcm63xx: use platform_get_irq in ehci-bcm63xx.c

[Maxime Bizon]

As requested by USB maintainer, use platform_get_irq instead of
platform_get_resource.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
---
 drivers/usb/host/ehci-bcm63xx.c |   10 +++++-----
 1 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/usb/host/ehci-bcm63xx.c b/drivers/usb/host/ehci-bcm63xx.c
index 2fef571..0aba0f9 100644
--- a/drivers/usb/host/ehci-bcm63xx.c
+++ b/drivers/usb/host/ehci-bcm63xx.c
@@ -62,15 +62,15 @@ static const struct hc_driver ehci_bcm63xx_hc_driver = {
 
 static int __devinit ehci_hcd_bcm63xx_drv_probe(struct platform_device *pdev)
 {
-	struct resource *res_mem, *res_irq;
+	struct resource *res_mem;
 	struct usb_hcd *hcd;
 	struct ehci_hcd *ehci;
 	u32 reg;
-	int ret;
+	int ret, irq;
 
 	res_mem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
-	res_irq = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
-	if (!res_mem || !res_irq)
+	irq = platform_get_irq(pdev, 0);;
+	if (!res_mem || irq < 0)
 		return -ENODEV;
 
 	reg = bcm_rset_readl(RSET_USBH_PRIV, USBH_PRIV_SWAP_REG);
@@ -109,7 +109,7 @@ static int __devinit ehci_hcd_bcm63xx_drv_probe(struct platform_device *pdev)
 	ehci->hcs_params = ehci_readl(ehci, &ehci->caps->hcs_params);
 	ehci->sbrn = 0x20;
 
-	ret = usb_add_hcd(hcd, res_irq->start, IRQF_DISABLED);
+	ret = usb_add_hcd(hcd, irq, IRQF_DISABLED);
 	if (ret)
 		goto out2;
 
-- 
1.6.0.4

[PATCH 09/10] bcm63xx: don't set bus type in ehci-bcm63xx.c

[Maxime Bizon]

The platform code already sets the bus type, so don't do it.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
---
 drivers/usb/host/ehci-bcm63xx.c |    1 -
 1 files changed, 0 insertions(+), 1 deletions(-)

diff --git a/drivers/usb/host/ehci-bcm63xx.c b/drivers/usb/host/ehci-bcm63xx.c
index 0aba0f9..8a62c0a 100644
--- a/drivers/usb/host/ehci-bcm63xx.c
+++ b/drivers/usb/host/ehci-bcm63xx.c
@@ -145,7 +145,6 @@ static struct platform_driver ehci_hcd_bcm63xx_driver = {
 	.driver		= {
 		.name	= "bcm63xx_ehci",
 		.owner	= THIS_MODULE,
-		.bus	= &platform_bus_type
 	},
 };
 
-- 
1.6.0.4

[PATCH 10/10] bcm63xx: clarify meaning of the magical value in ehci-bcm63xx.c

[Maxime Bizon]

USB maintainer asked for clarification of the magic value used during
USB init. Be clear about the source of it.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
---
 drivers/usb/host/ehci-bcm63xx.c |    4 +++-
 1 files changed, 3 insertions(+), 1 deletions(-)

diff --git a/drivers/usb/host/ehci-bcm63xx.c b/drivers/usb/host/ehci-bcm63xx.c
index 8a62c0a..5a03fdd 100644
--- a/drivers/usb/host/ehci-bcm63xx.c
+++ b/drivers/usb/host/ehci-bcm63xx.c
@@ -78,7 +78,9 @@ static int __devinit ehci_hcd_bcm63xx_drv_probe(struct platform_device *pdev)
 	reg |= USBH_PRIV_SWAP_EHCI_ENDN_MASK;
 	bcm_rset_writel(RSET_USBH_PRIV, reg, USBH_PRIV_SWAP_REG);
 
-	/* don't ask... */
+	/* the magic value comes for the original vendor BSP and is
+	 * needed for USB to work. Datasheet does not help, so the
+	 * magic value is used as-is. */
 	bcm_rset_writel(RSET_USBH_PRIV, 0x1c0020, USBH_PRIV_TEST_REG);
 
 	hcd = usb_create_hcd(&ehci_bcm63xx_hc_driver, &pdev->dev, "bcm63xx");
-- 
1.6.0.4

Re: [PATCH 10/10] bcm63xx: clarify meaning of the magical value in ehci-bcm63xx.c

[Sergei Shtylyov]

Hello.

Maxime Bizon wrote:

> USB maintainer asked for clarification of the magic value used during
> USB init. Be clear about the source of it.

> Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
> ---
>  drivers/usb/host/ehci-bcm63xx.c |    4 +++-
>  1 files changed, 3 insertions(+), 1 deletions(-)

> diff --git a/drivers/usb/host/ehci-bcm63xx.c b/drivers/usb/host/ehci-bcm63xx.c
> index 8a62c0a..5a03fdd 100644
> --- a/drivers/usb/host/ehci-bcm63xx.c
> +++ b/drivers/usb/host/ehci-bcm63xx.c
> @@ -78,7 +78,9 @@ static int __devinit ehci_hcd_bcm63xx_drv_probe(struct platform_device *pdev)
>  	reg |= USBH_PRIV_SWAP_EHCI_ENDN_MASK;
>  	bcm_rset_writel(RSET_USBH_PRIV, reg, USBH_PRIV_SWAP_REG);
>  
> -	/* don't ask... */
> +	/* the magic value comes for the original vendor BSP and is
> +	 * needed for USB to work. Datasheet does not help, so the
> +	 * magic value is used as-is. */
>  	bcm_rset_writel(RSET_USBH_PRIV, 0x1c0020, USBH_PRIV_TEST_REG);
>  
>  	hcd = usb_create_hcd(&ehci_bcm63xx_hc_driver, &pdev->dev, "bcm63xx");

    Documentation/CodingStyle, chapter 8:

The preferred style for long (multi-line) comments is:
 

         /*
          * This is the preferred style for multi-line
          * comments in the Linux kernel source code.
          * Please use it consistently.
          *
          * Description:  A column of asterisks on the left side,
          * with beginning and ending almost-blank lines.
          */

WBR, Sergei

Re: [PATCH 01/10] bcm63xx: remove duplicate init fields.

[Florian Fainelli]

Le Monday 01 June 2009 19:21:49 Maxime Bizon, vous avez écrit :
> This patch removes duplicate init fields in resource.
>
> Signed-off-by: Maxime Bizon <mbizon@freebox.fr>

Acked-by: Florian Fainelli <florian@openwrt.org>

> ---
>  arch/mips/bcm63xx/dev-enet.c |    2 --
>  1 files changed, 0 insertions(+), 2 deletions(-)
>
> diff --git a/arch/mips/bcm63xx/dev-enet.c b/arch/mips/bcm63xx/dev-enet.c
> index 51c2e5a..188fa66 100644
> --- a/arch/mips/bcm63xx/dev-enet.c
> +++ b/arch/mips/bcm63xx/dev-enet.c
> @@ -42,12 +42,10 @@ static struct resource enet0_res[] = {
>  	},
>  	{
>  		.start		= -1, /* filled at runtime */
> -		.start		= IRQ_ENET0_RXDMA,
>  		.flags		= IORESOURCE_IRQ,
>  	},
>  	{
>  		.start		= -1, /* filled at runtime */
> -		.start		= IRQ_ENET0_TXDMA,
>  		.flags		= IORESOURCE_IRQ,
>  	},
>  };

Re: [PATCH 02/10] bcm63xx: use napi_complete instead of __napi_complete

[Florian Fainelli]

Le Monday 01 June 2009 19:21:50 Maxime Bizon, vous avez écrit :
> We're not disabling IRQ, so we must call the irq safe flavour of
> napi_complete.
>
> Signed-off-by: Maxime Bizon <mbizon@freebox.fr>

Tested-by: Florian Fainelli <florian@openwrt.org>
Acked-by: Florian Fainelli <florian@openwrt.org>

> ---
>  drivers/net/bcm63xx_enet.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/drivers/net/bcm63xx_enet.c b/drivers/net/bcm63xx_enet.c
> index a91f909..20e08ef 100644
> --- a/drivers/net/bcm63xx_enet.c
> +++ b/drivers/net/bcm63xx_enet.c
> @@ -450,7 +450,7 @@ static int bcm_enet_poll(struct napi_struct *napi, int
> budget)
>
>  	/* no more packet in rx/tx queue, remove device from poll
>  	 * queue */
> -	__napi_complete(napi);
> +	napi_complete(napi);
>
>  	/* restore rx/tx interrupt */
>  	enet_dma_writel(priv, ENETDMA_IR_PKTDONE_MASK,

Re: [PATCH 03/10] bcm63xx: convert bcm63xx_enet to netdev ops.

[Florian Fainelli]

Le Monday 01 June 2009 19:21:51 Maxime Bizon, vous avez écrit :
> This patch makes bcm63xx_enet driver use netdevice ops.
>
> Signed-off-by: Maxime Bizon <mbizon@freebox.fr>

Acked-by: Florian Fainelli <florian@openwrt.org>

> ---
>  drivers/net/bcm63xx_enet.c |   24 ++++++++++++++----------
>  1 files changed, 14 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/net/bcm63xx_enet.c b/drivers/net/bcm63xx_enet.c
> index 20e08ef..36324b3 100644
> --- a/drivers/net/bcm63xx_enet.c
> +++ b/drivers/net/bcm63xx_enet.c
> @@ -1551,6 +1551,19 @@ static void bcm_enet_hw_preinit(struct bcm_enet_priv
> *priv) enet_writel(priv, val, ENET_MIBCTL_REG);
>  }
>
> +static const struct net_device_ops bcm_enet_ops = {
> +	.ndo_open		= bcm_enet_open,
> +	.ndo_stop		= bcm_enet_stop,
> +	.ndo_start_xmit		= bcm_enet_start_xmit,
> +	.ndo_get_stats		= bcm_enet_get_stats,
> +	.ndo_set_mac_address	= bcm_enet_set_mac_address,
> +	.ndo_set_multicast_list = bcm_enet_set_multicast_list,
> +	.ndo_do_ioctl		= bcm_enet_ioctl,
> +#ifdef CONFIG_NET_POLL_CONTROLLER
> +	.ndo_poll_controller = bcm_enet_netpoll,
> +#endif
> +};
> +
>  /*
>   * allocate netdevice, request register memory and register device.
>   */
> @@ -1716,17 +1729,8 @@ static int __devinit bcm_enet_probe(struct
> platform_device *pdev) enet_writel(priv, 0, ENET_MIB_REG(i));
>
>  	/* register netdevice */
> -	dev->open = bcm_enet_open;
> -	dev->stop = bcm_enet_stop;
> -	dev->hard_start_xmit = bcm_enet_start_xmit;
> -	dev->get_stats = bcm_enet_get_stats;
> -	dev->set_mac_address = bcm_enet_set_mac_address;
> -	dev->set_multicast_list = bcm_enet_set_multicast_list;
> +	dev->netdev_ops = &bcm_enet_ops;
>  	netif_napi_add(dev, &priv->napi, bcm_enet_poll, 16);
> -	dev->do_ioctl = bcm_enet_ioctl;
> -#ifdef CONFIG_NET_POLL_CONTROLLER
> -	dev->poll_controller = bcm_enet_netpoll;
> -#endif
>
>  	SET_ETHTOOL_OPS(dev, &bcm_enet_ethtool_ops);



-- 
Best regards, Florian Fainelli
Email : florian@openwrt.org
http://openwrt.org
-------------------------------

Re: [PATCH 04/10] bcm63xx: limit number of usb port to 1.

[Florian Fainelli]

Le Monday 01 June 2009 19:21:52 Maxime Bizon, vous avez écrit :
> This patch disables the use of more than one USB port. Hardware has
> two ports, but one may be shared with USB slave port.
>
> Until we have this information in the platform data, don't use the
> second port.
>
> Signed-off-by: Maxime Bizon <mbizon@freebox.fr>

Acked-by: Florian Fainelli <florian@openwrt.org>

> ---
>  drivers/usb/host/ohci-bcm63xx.c |    2 ++
>  1 files changed, 2 insertions(+), 0 deletions(-)
>
> diff --git a/drivers/usb/host/ohci-bcm63xx.c
> b/drivers/usb/host/ohci-bcm63xx.c index 08807d9..74f432f 100644
> --- a/drivers/usb/host/ohci-bcm63xx.c
> +++ b/drivers/usb/host/ohci-bcm63xx.c
> @@ -20,6 +20,8 @@ static int __devinit ohci_bcm63xx_start(struct usb_hcd
> *hcd) struct ohci_hcd *ohci = hcd_to_ohci(hcd);
>  	int ret;
>
> +	ohci->num_ports = 1;
> +
>  	ret = ohci_init(ohci);
>  	if (ret < 0)
>  		return ret;



-- 
Best regards, Florian Fainelli
Email : florian@openwrt.org
http://openwrt.org
-------------------------------

Re: [PATCH 05/10] bcm63xx: use platform_get_irq in ohci-bcm63xx.c

[Florian Fainelli]

Le Monday 01 June 2009 19:21:53 Maxime Bizon, vous avez écrit :
> As requested by USB maintainer, use platform_get_irq instead of
> platform_get_resource.
>
> Signed-off-by: Maxime Bizon <mbizon@freebox.fr>

Acked-by: Florian Fainelli <florian@openwrt.org>
> ---
>  drivers/usb/host/ohci-bcm63xx.c |   10 +++++-----
>  1 files changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/usb/host/ohci-bcm63xx.c
> b/drivers/usb/host/ohci-bcm63xx.c index 74f432f..f0e4639 100644
> --- a/drivers/usb/host/ohci-bcm63xx.c
> +++ b/drivers/usb/host/ohci-bcm63xx.c
> @@ -58,15 +58,15 @@ static const struct hc_driver ohci_bcm63xx_hc_driver =
> {
>
>  static int __devinit ohci_hcd_bcm63xx_drv_probe(struct platform_device
> *pdev) {
> -	struct resource *res_mem, *res_irq;
> +	struct resource *res_mem;
>  	struct usb_hcd *hcd;
>  	struct ohci_hcd *ohci;
>  	u32 reg;
> -	int ret;
> +	int ret, irq;
>
>  	res_mem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> -	res_irq = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
> -	if (!res_mem || !res_irq)
> +	irq = platform_get_irq(pdev, 0);
> +	if (!res_mem || irq < 0)
>  		return -ENODEV;
>
>  	if (BCMCPU_IS_6348()) {
> @@ -114,7 +114,7 @@ static int __devinit ohci_hcd_bcm63xx_drv_probe(struct
> platform_device *pdev) OHCI_QUIRK_FRAME_NO;
>  	ohci_hcd_init(ohci);
>
> -	ret = usb_add_hcd(hcd, res_irq->start, IRQF_DISABLED);
> +	ret = usb_add_hcd(hcd, irq, IRQF_DISABLED);
>  	if (ret)
>  		goto out2;



-- 
Best regards, Florian Fainelli
Email : florian@openwrt.org
http://openwrt.org
-------------------------------

Re: [PATCH 06/10] bcm63xx: don't set bus type in ohci-bcm63xx.c

[Florian Fainelli]

Le Monday 01 June 2009 19:21:54 Maxime Bizon, vous avez écrit :
> The platform code already sets the bus type, so don't do it.
>
> Signed-off-by: Maxime Bizon <mbizon@freebox.fr>

Acked-by: Florian Fainelli <florian@openwrt.org>

> ---
>  drivers/usb/host/ohci-bcm63xx.c |    1 -
>  1 files changed, 0 insertions(+), 1 deletions(-)
>
> diff --git a/drivers/usb/host/ohci-bcm63xx.c
> b/drivers/usb/host/ohci-bcm63xx.c index f0e4639..d48c8ac 100644
> --- a/drivers/usb/host/ohci-bcm63xx.c
> +++ b/drivers/usb/host/ohci-bcm63xx.c
> @@ -154,7 +154,6 @@ static struct platform_driver ohci_hcd_bcm63xx_driver =
> { .driver		= {
>  		.name	= "bcm63xx_ohci",
>  		.owner	= THIS_MODULE,
> -		.bus	= &platform_bus_type
>  	},
>  };



-- 
Best regards, Florian Fainelli
Email : florian@openwrt.org
http://openwrt.org
-------------------------------

Re: [PATCH 07/10] bcm63xx: clarify meaning of the magical value in ohci-bcm63xx.c

[Florian Fainelli]

Le Monday 01 June 2009 19:21:55 Maxime Bizon, vous avez écrit :
> USB maintainer asked for clarification  of the magic value used during
> USB init. Be clear about the source of it.
>
> Signed-off-by: Maxime Bizon <mbizon@freebox.fr>

Acked-by: Florian Fainelli <florian@openwrt.org>
> ---
>  drivers/usb/host/ohci-bcm63xx.c |    4 +++-
>  1 files changed, 3 insertions(+), 1 deletions(-)
>
> diff --git a/drivers/usb/host/ohci-bcm63xx.c
> b/drivers/usb/host/ohci-bcm63xx.c index d48c8ac..bd66d5a 100644
> --- a/drivers/usb/host/ohci-bcm63xx.c
> +++ b/drivers/usb/host/ohci-bcm63xx.c
> @@ -85,7 +85,9 @@ static int __devinit ohci_hcd_bcm63xx_drv_probe(struct
> platform_device *pdev) reg &= ~USBH_PRIV_SWAP_OHCI_ENDN_MASK;
>  		reg |= USBH_PRIV_SWAP_OHCI_DATA_MASK;
>  		bcm_rset_writel(RSET_USBH_PRIV, reg, USBH_PRIV_SWAP_REG);
> -		/* don't ask... */
> +		/* the magic value comes for the original vendor BSP
> +		 * and is needed for USB to work. Datasheet does not
> +		 * help, so the magic value is used as-is. */
>  		bcm_rset_writel(RSET_USBH_PRIV, 0x1c0020, USBH_PRIV_TEST_REG);
>  	} else
>  		return 0;



-- 
Best regards, Florian Fainelli
Email : florian@openwrt.org
http://openwrt.org
-------------------------------

Re: [PATCH 08/10] bcm63xx: use platform_get_irq in ehci-bcm63xx.c

[Florian Fainelli]

Le Monday 01 June 2009 19:21:56 Maxime Bizon, vous avez écrit :
> As requested by USB maintainer, use platform_get_irq instead of
> platform_get_resource.
>
> Signed-off-by: Maxime Bizon <mbizon@freebox.fr>

Acked-by: Florian Fainelli <florian@openwrt.org>

> ---
>  drivers/usb/host/ehci-bcm63xx.c |   10 +++++-----
>  1 files changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/usb/host/ehci-bcm63xx.c
> b/drivers/usb/host/ehci-bcm63xx.c index 2fef571..0aba0f9 100644
> --- a/drivers/usb/host/ehci-bcm63xx.c
> +++ b/drivers/usb/host/ehci-bcm63xx.c
> @@ -62,15 +62,15 @@ static const struct hc_driver ehci_bcm63xx_hc_driver =
> {
>
>  static int __devinit ehci_hcd_bcm63xx_drv_probe(struct platform_device
> *pdev) {
> -	struct resource *res_mem, *res_irq;
> +	struct resource *res_mem;
>  	struct usb_hcd *hcd;
>  	struct ehci_hcd *ehci;
>  	u32 reg;
> -	int ret;
> +	int ret, irq;
>
>  	res_mem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> -	res_irq = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
> -	if (!res_mem || !res_irq)
> +	irq = platform_get_irq(pdev, 0);;
> +	if (!res_mem || irq < 0)
>  		return -ENODEV;
>
>  	reg = bcm_rset_readl(RSET_USBH_PRIV, USBH_PRIV_SWAP_REG);
> @@ -109,7 +109,7 @@ static int __devinit ehci_hcd_bcm63xx_drv_probe(struct
> platform_device *pdev) ehci->hcs_params = ehci_readl(ehci,
> &ehci->caps->hcs_params);
>  	ehci->sbrn = 0x20;
>
> -	ret = usb_add_hcd(hcd, res_irq->start, IRQF_DISABLED);
> +	ret = usb_add_hcd(hcd, irq, IRQF_DISABLED);
>  	if (ret)
>  		goto out2;



-- 
Best regards, Florian Fainelli
Email : florian@openwrt.org
http://openwrt.org
-------------------------------

Re: [PATCH 09/10] bcm63xx: don't set bus type in ehci-bcm63xx.c

[Florian Fainelli]

Le Monday 01 June 2009 19:21:57 Maxime Bizon, vous avez écrit :
> The platform code already sets the bus type, so don't do it.
>
> Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
Acked-by: Florian Fainelli <florian@openwrt.org>


> ---
>  drivers/usb/host/ehci-bcm63xx.c |    1 -
>  1 files changed, 0 insertions(+), 1 deletions(-)
>
> diff --git a/drivers/usb/host/ehci-bcm63xx.c
> b/drivers/usb/host/ehci-bcm63xx.c index 0aba0f9..8a62c0a 100644
> --- a/drivers/usb/host/ehci-bcm63xx.c
> +++ b/drivers/usb/host/ehci-bcm63xx.c
> @@ -145,7 +145,6 @@ static struct platform_driver ehci_hcd_bcm63xx_driver =
> { .driver		= {
>  		.name	= "bcm63xx_ehci",
>  		.owner	= THIS_MODULE,
> -		.bus	= &platform_bus_type
>  	},
>  };



-- 
Best regards, Florian Fainelli
Email : florian@openwrt.org
http://openwrt.org
-------------------------------

Re: [PATCH 10/10] bcm63xx: clarify meaning of the magical value in ehci-bcm63xx.c

[Florian Fainelli]

Le Monday 01 June 2009 19:21:58 Maxime Bizon, vous avez écrit :
> USB maintainer asked for clarification of the magic value used during
> USB init. Be clear about the source of it.
>
> Signed-off-by: Maxime Bizon <mbizon@freebox.fr>

Acked-by: Florian Fainelli <florian@openwrt.org>

> ---
>  drivers/usb/host/ehci-bcm63xx.c |    4 +++-
>  1 files changed, 3 insertions(+), 1 deletions(-)
>
> diff --git a/drivers/usb/host/ehci-bcm63xx.c
> b/drivers/usb/host/ehci-bcm63xx.c index 8a62c0a..5a03fdd 100644
> --- a/drivers/usb/host/ehci-bcm63xx.c
> +++ b/drivers/usb/host/ehci-bcm63xx.c
> @@ -78,7 +78,9 @@ static int __devinit ehci_hcd_bcm63xx_drv_probe(struct
> platform_device *pdev) reg |= USBH_PRIV_SWAP_EHCI_ENDN_MASK;
>  	bcm_rset_writel(RSET_USBH_PRIV, reg, USBH_PRIV_SWAP_REG);
>
> -	/* don't ask... */
> +	/* the magic value comes for the original vendor BSP and is
> +	 * needed for USB to work. Datasheet does not help, so the
> +	 * magic value is used as-is. */
>  	bcm_rset_writel(RSET_USBH_PRIV, 0x1c0020, USBH_PRIV_TEST_REG);
>
>  	hcd = usb_create_hcd(&ehci_bcm63xx_hc_driver, &pdev->dev, "bcm63xx");



-- 
Best regards, Florian Fainelli
Email : florian@openwrt.org
http://openwrt.org
-------------------------------

[PATCH 10/10 (v2)] bcm63xx: clarify meaning of the magical value in ehci-bcm63xx.c

[Maxime Bizon]

USB maintainer asked for clarification of the magic value used during
USB init. Be clear about the source of it.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
---
 drivers/usb/host/ehci-bcm63xx.c |    6 +++++-
 1 files changed, 5 insertions(+), 1 deletions(-)

diff --git a/drivers/usb/host/ehci-bcm63xx.c b/drivers/usb/host/ehci-bcm63xx.c
index 8a62c0a..c2b3e7f 100644
--- a/drivers/usb/host/ehci-bcm63xx.c
+++ b/drivers/usb/host/ehci-bcm63xx.c
@@ -78,7 +78,11 @@ static int __devinit ehci_hcd_bcm63xx_drv_probe(struct platform_device *pdev)
 	reg |= USBH_PRIV_SWAP_EHCI_ENDN_MASK;
 	bcm_rset_writel(RSET_USBH_PRIV, reg, USBH_PRIV_SWAP_REG);
 
-	/* don't ask... */
+	/*
+	 * The magic value comes for the original vendor BSP and is
+	 * needed for USB to work. Datasheet does not help, so the
+	 * magic value is used as-is.
+	 */
 	bcm_rset_writel(RSET_USBH_PRIV, 0x1c0020, USBH_PRIV_TEST_REG);
 
 	hcd = usb_create_hcd(&ehci_bcm63xx_hc_driver, &pdev->dev, "bcm63xx");
-- 
1.6.0.4

[PATCH 07/10 (v2)] bcm63xx: clarify meaning of the magical value in ohci-bcm63xx.c

[Maxime Bizon]

USB maintainer asked for clarification  of the magic value used during
USB init. Be clear about the source of it.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
---
 drivers/usb/host/ohci-bcm63xx.c |    6 +++++-
 1 files changed, 5 insertions(+), 1 deletions(-)

diff --git a/drivers/usb/host/ohci-bcm63xx.c b/drivers/usb/host/ohci-bcm63xx.c
index d48c8ac..668808e 100644
--- a/drivers/usb/host/ohci-bcm63xx.c
+++ b/drivers/usb/host/ohci-bcm63xx.c
@@ -85,7 +85,11 @@ static int __devinit ohci_hcd_bcm63xx_drv_probe(struct platform_device *pdev)
 		reg &= ~USBH_PRIV_SWAP_OHCI_ENDN_MASK;
 		reg |= USBH_PRIV_SWAP_OHCI_DATA_MASK;
 		bcm_rset_writel(RSET_USBH_PRIV, reg, USBH_PRIV_SWAP_REG);
-		/* don't ask... */
+		/*
+		 * The magic value comes for the original vendor BSP
+		 * and is needed for USB to work. Datasheet does not
+		 * help, so the magic value is used as-is.
+		 */
 		bcm_rset_writel(RSET_USBH_PRIV, 0x1c0020, USBH_PRIV_TEST_REG);
 	} else
 		return 0;
-- 
1.6.0.4

Re: More updates to bcm63xx support (resend)

[Ralf Baechle]

On Mon, Jun 01, 2009 at 07:21:48PM +0200, Maxime Bizon wrote:

Whole series folded into the existing series.

Thanks guys,

  Ralf