[PATCH 1/3] can: c_can: Add d_can raminit support
AnilKumar, Chimata
anilkumar at ti.com
Wed Nov 21 00:05:47 EST 2012
On Tue, Nov 20, 2012 at 16:20:41, Marc Kleine-Budde wrote:
> On 11/14/2012 07:08 PM, AnilKumar Ch wrote:
> > Add D_CAN raminit support to C_CAN driver to enable D_CAN RAM,
> > which holds all the message objects during transmission or
> > receiving of data. This initialization/de-initialization should
> > be done in synchronous with D_CAN clock.
> >
> > In case of AM335X-EVM (active user of D_CAN driver) message RAM is
> > controlled through control module register for both instances. So
> > control module register details is required to initialization or
> > de-initialization of message RAM according to instance number.
> >
> > Control module memory resource is obtained from D_CAN dt node and
> > instance number obtained from device tree aliases node.
> >
> > Signed-off-by: AnilKumar Ch <anilkumar at ti.com>
>
> Some nitpicks inline.
Thanks for the review.
>
> > ---
> > drivers/net/can/c_can/c_can.c | 12 +++++++++++
> > drivers/net/can/c_can/c_can.h | 3 +++
> > drivers/net/can/c_can/c_can_platform.c | 34 +++++++++++++++++++++++++++++++-
> > 3 files changed, 48 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/net/can/c_can/c_can.c b/drivers/net/can/c_can/c_can.c
> > index e5180df..c15830c 100644
> > --- a/drivers/net/can/c_can/c_can.c
> > +++ b/drivers/net/can/c_can/c_can.c
> > @@ -233,6 +233,12 @@ static inline void c_can_pm_runtime_put_sync(const struct c_can_priv *priv)
> > pm_runtime_put_sync(priv->device);
> > }
> >
> > +static inline void c_can_reset_ram(const struct c_can_priv *priv, bool enable)
> > +{
> > + if (priv->ram_init)
> > + priv->ram_init(priv, enable);
> > +}
> > +
> > static inline int get_tx_next_msg_obj(const struct c_can_priv *priv)
> > {
> > return (priv->tx_next & C_CAN_NEXT_MSG_OBJ_MASK) +
> > @@ -1090,6 +1096,7 @@ static int c_can_open(struct net_device *dev)
> > struct c_can_priv *priv = netdev_priv(dev);
> >
> > c_can_pm_runtime_get_sync(priv);
> > + c_can_reset_ram(priv, true);
> >
> > /* open the can device */
> > err = open_candev(dev);
> > @@ -1118,6 +1125,7 @@ static int c_can_open(struct net_device *dev)
> > exit_irq_fail:
> > close_candev(dev);
> > exit_open_fail:
> > + c_can_reset_ram(priv, false);
> > c_can_pm_runtime_put_sync(priv);
> > return err;
> > }
> > @@ -1131,6 +1139,8 @@ static int c_can_close(struct net_device *dev)
> > c_can_stop(dev);
> > free_irq(dev->irq, dev);
> > close_candev(dev);
> > +
> > + c_can_reset_ram(priv, false);
> > c_can_pm_runtime_put_sync(priv);
> >
> > return 0;
> > @@ -1188,6 +1198,7 @@ int c_can_power_down(struct net_device *dev)
> >
> > c_can_stop(dev);
> >
> > + c_can_reset_ram(priv, false);
> > c_can_pm_runtime_put_sync(priv);
> >
> > return 0;
> > @@ -1206,6 +1217,7 @@ int c_can_power_up(struct net_device *dev)
> > WARN_ON(priv->type != BOSCH_D_CAN);
> >
> > c_can_pm_runtime_get_sync(priv);
> > + c_can_reset_ram(priv, true);
> >
> > /* Clear PDR and INIT bits */
> > val = priv->read_reg(priv, C_CAN_CTRL_EX_REG);
> > diff --git a/drivers/net/can/c_can/c_can.h b/drivers/net/can/c_can/c_can.h
> > index e5ed41d..419de5c 100644
> > --- a/drivers/net/can/c_can/c_can.h
> > +++ b/drivers/net/can/c_can/c_can.h
> > @@ -169,6 +169,9 @@ struct c_can_priv {
> > void *priv; /* for board-specific data */
> > u16 irqstatus;
> > enum c_can_dev_id type;
> > + u32 __iomem *raminit_ctrlreg;
> > + unsigned int instance;
> > + void (*ram_init) (const struct c_can_priv *priv, bool enable);
> > };
> >
> > struct net_device *alloc_c_can_dev(void);
> > diff --git a/drivers/net/can/c_can/c_can_platform.c b/drivers/net/can/c_can/c_can_platform.c
> > index ee141613..2e61d69 100644
> > --- a/drivers/net/can/c_can/c_can_platform.c
> > +++ b/drivers/net/can/c_can/c_can_platform.c
> > @@ -38,6 +38,8 @@
> >
> > #include "c_can.h"
> >
> > +#define CAN_RAMINIT_START_MASK(i) (1 << (i))
> > +
> > /*
> > * 16-bit c_can registers can be arranged differently in the memory
> > * architecture of different implementations. For example: 16-bit
> > @@ -68,6 +70,24 @@ static void c_can_plat_write_reg_aligned_to_32bit(struct c_can_priv *priv,
> > writew(val, priv->base + 2 * priv->regs[index]);
> > }
> >
> > +static void c_can_hw_raminit(const struct c_can_priv *priv, bool enable)
> > +{
> > + u32 val;
> > +
> > + if (!priv->raminit_ctrlreg || (priv->instance < 0))
> > + return;
>
> Can you move this sanity check to the probe function and only assign
> priv->ram_init if the above is true?
Sure, I will change
>
> > +
> > + val = readl(priv->raminit_ctrlreg);
> > + if (enable) {
> > + val &= ~CAN_RAMINIT_START_MASK(priv->instance);
> > + val |= CAN_RAMINIT_START_MASK(priv->instance);
> > + writel(val, priv->raminit_ctrlreg);
> > + } else {
> > + val &= ~CAN_RAMINIT_START_MASK(priv->instance);
> > + writel(val, priv->raminit_ctrlreg);
> > + }
> > +}
> > +
> > static struct platform_device_id c_can_id_table[] = {
> > [BOSCH_C_CAN_PLATFORM] = {
> > .name = KBUILD_MODNAME,
> > @@ -99,7 +119,7 @@ static int __devinit c_can_plat_probe(struct platform_device *pdev)
> > const struct of_device_id *match;
> > const struct platform_device_id *id;
> > struct pinctrl *pinctrl;
> > - struct resource *mem;
> > + struct resource *mem, *res;
> > int irq;
> > struct clk *clk;
> >
> > @@ -178,6 +198,18 @@ static int __devinit c_can_plat_probe(struct platform_device *pdev)
> > priv->can.ctrlmode_supported |= CAN_CTRLMODE_3_SAMPLES;
> > priv->read_reg = c_can_plat_read_reg_aligned_to_16bit;
> > priv->write_reg = c_can_plat_write_reg_aligned_to_16bit;
> > +
> > + res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
> > + priv->raminit_ctrlreg =
> > + devm_request_and_ioremap(&pdev->dev, res);
>
> What happens if there isn't a second IORESOURCE_MEM region? devm_request
> will print an error in this case and any other error, too [1]. Do we
> need streamlining here?
>
> [1] http://lxr.free-electrons.com/source/drivers/base/platform.c#L59
I did not get what the point is.
In most of the cases above two API's returns NULL. Even res is NULL
nothing to worry, first condition in devm_request_and_ioremap() is
NULL pointer check of res. If "res" is NULL then devm_xx will return
NULL which result into printing of below warning.
>
> > + if (!priv->raminit_ctrlreg)
> > + dev_warn(&pdev->dev, "failed to obtain control memory\n");
I will change this warning to info
if (!priv->raminit_ctrlreg)
dev_info(&pdev->dev, "control memory is not used for raminit\n");
> > +
> > + if (pdev->dev.of_node)
> > + pdev->id = of_alias_get_id(pdev->dev.of_node, "d_can");
> > +
> > + priv->instance = pdev->id;
>
> Please do not modify the pdev, better you use something like this:
>
> pdev->id < 0 ? of_alias_get_id(pdev->dev.of_node, "d_can") :
> pdev->id;
I will change accordingly.
Thanks
AnilKumar
More information about the devicetree-discuss
mailing list