Re: [PATCH 1/5] firmware: xilinx: Adds new eemi call for reg access

[Sudeep Holla <sudeep.holla@arm.com>]

On Wed, Dec 04, 2019 at 03:29:15PM -0800, Jolly Shah wrote:
> From: Rajan Vaja <rajan.vaja@xilinx.com>
> 
> This patch adds new EEMI call which is used for CSU/PMU register
> access from linux.
> 
> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
> Signed-off-by: Jolly Shah <jolly.shah@xilinx.com>
> ---
>  drivers/firmware/xilinx/zynqmp.c     | 183 +++++++++++++++++++++++++++++++++++
>  include/linux/firmware/xlnx-zynqmp.h |   9 ++
>  2 files changed, 192 insertions(+)
> 
> diff --git a/drivers/firmware/xilinx/zynqmp.c b/drivers/firmware/xilinx/zynqmp.c
> index fd3d837..56431ad 100644
> --- a/drivers/firmware/xilinx/zynqmp.c
> +++ b/drivers/firmware/xilinx/zynqmp.c
> @@ -24,6 +24,8 @@
>  #include <linux/firmware/xlnx-zynqmp.h>
>  #include "zynqmp-debug.h"
>  
> +static unsigned long register_address;
> +
>  static const struct zynqmp_eemi_ops *eemi_ops_tbl;
>  
>  static const struct mfd_cell firmware_devs[] = {
> @@ -664,6 +666,26 @@ static int zynqmp_pm_set_requirement(const u32 node, const u32 capabilities,
>  				   qos, ack, NULL);
>  }
>  
> +/**
> + * zynqmp_pm_config_reg_access - PM Config API for Config register access
> + * @register_access_id:	ID of the requested REGISTER_ACCESS
> + * @address:		Address of the register to be accessed
> + * @mask:		Mask to be written to the register
> + * @value:		Value to be written to the register
> + * @out:		Returned output value
> + *
> + * This function calls REGISTER_ACCESS to configure CSU/PMU registers.
> + *
> + * Return:	Returns status, either success or error+reason
> + */
> +
> +static int zynqmp_pm_config_reg_access(u32 register_access_id, u32 address,
> +				       u32 mask, u32 value, u32 *out)
> +{
> +	return zynqmp_pm_invoke_fn(PM_REGISTER_ACCESS, register_access_id,
> +				   address, mask, value, out);
> +}
> +

If you have this API, can you remove all other APIs and implement them
using these ? This kills the EEMI abstraction.

NACK for this and any attempts to provide direct reas/write access to
the PM config space. EEMI should have better abstraction than this.

--
Regards,
Sudeep