On Sat, Jan 14, 2017 at 03:05:30PM +0800, Baoyou Xie wrote:
> +static int zx2967_reset_assert(struct reset_controller_dev *rcdev,
> + unsigned long id)
> +{
> + struct zx2967_reset *reset = NULL;
> + int bank = id / 32;
> + int offset = id % 32;
> + unsigned int reg;
u32 is probably better for register value.
> + unsigned long flags;
> +
> + reset = container_of(rcdev, struct zx2967_reset, rcdev);
> +
> + spin_lock_irqsave(&reset->lock, flags);
> +
> + reg = readl(reset->reg_base + (bank * 4));
> + writel(reg & ~BIT(offset), reset->reg_base + (bank * 4));
> + reg = readl(reset->reg_base + (bank * 4));
Is this read on the register is necessary? If so, we should probably
have a comment for that.
> +
> + spin_unlock_irqrestore(&reset->lock, flags);
> +
> + return 0;
> +}
> +
> +static int zx2967_reset_deassert(struct reset_controller_dev *rcdev,
> + unsigned long id)
Please indent the line right after parentheses.
> +{
> + struct zx2967_reset *reset = NULL;
> + int bank = id / 32;
> + int offset = id % 32;
> + unsigned int reg;
> + unsigned long flags;
> +
> + reset = container_of(rcdev, struct zx2967_reset, rcdev);
> +
> + spin_lock_irqsave(&reset->lock, flags);
> +
> + reg = readl(reset->reg_base + (bank * 4));
> + writel(reg | BIT(offset), reset->reg_base + (bank * 4));
> + reg = readl(reset->reg_base + (bank * 4));
> +
> + spin_unlock_irqrestore(&reset->lock, flags);
> +
> + return 0;
> +}
Only difference between these two functions is only one line. Should we
consolidate them a bit?
Shawn
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
