Message-ID: <1481569373.7188.48.camel@linux.intel.com>
Subject: Re: [PATCH V3] i2c: designware: fix wrong tx/rx fifo for ACPI
From: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
To: Tin Huynh <tnhuynh@apm.com>,
        Jarkko Nikula <jarkko.nikula@linux.intel.com>,
        Mika Westerberg <mika.westerberg@linux.intel.com>,
        Wolfram Sang <wsa@the-dreams.de>
Cc: linux-i2c@vger.kernel.org, linux-kernel@vger.kernel.org,
        linux-acpi@vger.kernel.org, Loc Ho <lho@apm.com>,
        Thang Nguyen <tqnguyen@apm.com>, Phong Vo <pvo@apm.com>,
        patches@apm.com
Date: Mon, 12 Dec 2016 21:02:53 +0200
In-Reply-To: <1481531810-31695-1-git-send-email-tnhuynh@apm.com>
References: <1481531810-31695-1-git-send-email-tnhuynh@apm.com>
Organization: Intel Finland Oy
Content-Type: text/plain; charset="UTF-8"
Mime-Version: 1.0
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 3082
Lines: 104

Thanks for an update! My comments below.

On Mon, 2016-12-12 at 15:36 +0700, Tin Huynh wrote:
> ACPI always sets txfifo and rxfifo to 32. This configuration will
> cause problem if the IP core supports a fifo size of less than 32.
> The driver should read the fifo size from the IP and select the 
> smaller one of the two.

I would use FIFO in capital to be consistent with what you refer to
(apparently not a variable name), so

Tx FIFO, Rx FIFO, FIFO, and so on.


> 
> Signed-off-by: Tin Huynh <tnhuynh@apm.com>
> 
> ---
>  drivers/i2c/busses/i2c-designware-platdrv.c |   26
> ++++++++++++++++++++------
>  1 files changed, 20 insertions(+), 6 deletions(-)
> 
> Change from V2:
>  -Add a helper function to set fifo size.
> 
> Change from V1:
>  -Revert the default 32 for fifo, read parameter from IP core
>  and pick the smaller one of the two.
>  -Correct the title to describe new approach.
> 
> diff --git a/drivers/i2c/busses/i2c-designware-platdrv.c
> b/drivers/i2c/busses/i2c-designware-platdrv.c
> index 0b42a12..665f491 100644
> --- a/drivers/i2c/busses/i2c-designware-platdrv.c
> +++ b/drivers/i2c/busses/i2c-designware-platdrv.c
> @@ -150,6 +150,24 @@ static int i2c_dw_plat_prepare_clk(struct
> dw_i2c_dev *i_dev, bool prepare)
>  	return 0;
>  }
>  
> +static void dw_i2c_set_fifo_size(struct dw_i2c_dev *dev)
> +{
> +	u32 param1, tx_fifo_depth, rx_fifo_depth;
> +
> +	param1 = i2c_dw_read_comp_param(dev);

You name it as param1 because you read *_PARAM1? For me it's not clear
from the name of helper function.

u32 param would work otherwise.

> +	tx_fifo_depth = ((param1 >> 16) & 0xff) + 1;
> +	rx_fifo_depth = ((param1 >> 8)  & 0xff) + 1;
> +	if (!dev->tx_fifo_depth) {
> +		dev->tx_fifo_depth = tx_fifo_depth;
> +		dev->rx_fifo_depth = rx_fifo_depth;
> +	} else if (tx_fifo_depth) {
> +		dev->tx_fifo_depth = min_t(u32, dev->tx_fifo_depth,
> +				tx_fifo_depth);
> +		dev->rx_fifo_depth = min_t(u32, dev->rx_fifo_depth,
> +				rx_fifo_depth);
> +	}

So, let's clarify here:
Is it possible to have an IP without parameter block enabled? I mean to
read something arbitrary (or zeroes, or all-ones) from param1.

If not, just remove second condition at all.

> +}
> +
>  static int dw_i2c_plat_probe(struct platform_device *pdev)
>  {
>  	struct dw_i2c_platform_data *pdata = dev_get_platdata(&pdev-
> >dev);
> @@ -246,13 +264,9 @@ static int dw_i2c_plat_probe(struct
> platform_device *pdev)
>  				1000000);
>  	}
>  
> -	if (!dev->tx_fifo_depth) {
> -		u32 param1 = i2c_dw_read_comp_param(dev);
> -
> -		dev->tx_fifo_depth = ((param1 >> 16) & 0xff) + 1;
> -		dev->rx_fifo_depth = ((param1 >> 8)  & 0xff) + 1;
> 

> +	if (!dev->tx_fifo_depth)
>  		dev->adapter.nr = pdev->id;

Now you spread condition to two locations and it's hard to remember
ordering without looking closer to the code.

That's why I suggested to pass an ID parameter in the first place.

> -	}
> +	dw_i2c_set_fifo_size(dev);
>  
>  	adap = &dev->adapter;
>  	adap->owner = THIS_MODULE;

-- 
Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Intel Finland Oy