Message-ID: <907a5c11-e92f-a1dc-07fb-e2f6e5e9e59c@codeaurora.org>
Date:   Mon, 3 Jul 2017 11:54:12 +0530
From:   Sricharan R <sricharan@...eaurora.org>
To:     Abhishek Sahu <absahu@...eaurora.org>, dwmw2@...radead.org,
        computersforpeace@...il.com, boris.brezillon@...e-electrons.com,
        marek.vasut@...il.com, richard@....at, cyrille.pitchen@...ev4u.fr,
        robh+dt@...nel.org, mark.rutland@....com
Cc:     linux-mtd@...ts.infradead.org, devicetree@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-arm-msm@...r.kernel.org,
        andy.gross@...aro.org, architt@...eaurora.org
Subject: Re: [PATCH 02/14] qcom: mtd: nand: add and initialize QPIC DMA
 resources

Hi Abhishek,


On 6/29/2017 12:45 PM, Abhishek Sahu wrote:
> 1. The QPIC NAND uses 3 BAM channels: command, data tx and
>    data rx while EBI2 NAND uses only single ADM channel.
> 
> 2. The EBI2 NAND uses normal register read buffer since this
>    buffer will be remapped with dma_map_sg. The QPIC NAND will give
>    register read buffer in command descriptor and the command
>    descriptor will be mapped with dma_map_sg so the register buffer
>    should be DMA coherent.
> 
> Signed-off-by: Abhishek Sahu <absahu@...eaurora.org>
> ---
>  .../devicetree/bindings/mtd/qcom_nandc.txt         |  25 +++--
>  drivers/mtd/nand/qcom_nandc.c                      | 106 ++++++++++++++++-----
>  2 files changed, 99 insertions(+), 32 deletions(-)
> 
> diff --git a/Documentation/devicetree/bindings/mtd/qcom_nandc.txt b/Documentation/devicetree/bindings/mtd/qcom_nandc.txt
> index 5d0f7ae..87b9a56 100644
> --- a/Documentation/devicetree/bindings/mtd/qcom_nandc.txt
> +++ b/Documentation/devicetree/bindings/mtd/qcom_nandc.txt
> @@ -9,15 +9,17 @@ Required properties:
>  - clock-names:		must contain "core" for the core clock and "aon" for the
>  			always on clock
>  - dmas:			DMA specifier, consisting of a phandle to the ADM DMA
> -			controller node and the channel number to be used for
> -			NAND. Refer to dma.txt and qcom_adm.txt for more details
> -- dma-names:		must be "rxtx"
> -- qcom,cmd-crci:	must contain the ADM command type CRCI block instance
> -			number specified for the NAND controller on the given
> -			platform
> -- qcom,data-crci:	must contain the ADM data type CRCI block instance
> -			number specified for the NAND controller on the given
> -			platform
> +			or BAM DMA controller node and the channel number to
> +			be used for NAND. Refer to dma.txt, qcom_adm.txt(ADM)
> +			and qcom/bam_dma.txt(BAM) for more details
> +- dma-names:		"rxtx" - ADM
> +			"tx", "rx", "cmd" - BAM
> +- qcom,cmd-crci:	Only required for ADM DMA. must contain the ADM command
> +			type CRCI block instance number specified for the NAND
> +			controller on the given platform.
> +- qcom,data-crci:	Only required for ADM DMA. must contain the ADM data
> +			type CRCI block instance number specified for the NAND
> +			controller on the given platform.

  May be better to put the parameter list that are specific to ADMA and BAM under
  seperate groups and those common for both in one group, instead of mentioning
  at each of the property level.

>  - #address-cells:	<1> - subnodes give the chip-select number
>  - #size-cells:		<0>
>  
> @@ -95,6 +97,11 @@ nand@...0000 {
>  		<&gcc GCC_QPIC_AHB_CLK>;
>  	clock-names = "core", "aon";
>  
> +	dmas = <&qpicbam 0>,
> +		<&qpicbam 1>,
> +		<&qpicbam 2>;
> +	dma-names = "tx", "rx", "cmd";
> +
>  	#address-cells = <1>;
>  	#size-cells = <0>;
>  
> diff --git a/drivers/mtd/nand/qcom_nandc.c b/drivers/mtd/nand/qcom_nandc.c
> index f55f728..520add9 100644
> --- a/drivers/mtd/nand/qcom_nandc.c
> +++ b/drivers/mtd/nand/qcom_nandc.c
> @@ -226,6 +226,7 @@ struct nandc_regs {
>   *				by upper layers directly
>   * @buf_size/count/start:	markers for chip->read_buf/write_buf functions
>   * @reg_read_buf:		local buffer for reading back registers via DMA
> + * @reg_read_buf_phys:		contains dma address for register read buffer
>   * @reg_read_pos:		marker for data read in reg_read_buf
>   *
>   * @regs:			a contiguous chunk of memory for DMA register
> @@ -249,9 +250,19 @@ struct qcom_nand_controller {
>  	struct clk *core_clk;
>  	struct clk *aon_clk;
>  
> -	struct dma_chan *chan;
> -	unsigned int cmd_crci;
> -	unsigned int data_crci;
> +	union {
> +		struct {
> +			struct dma_chan *tx_chan;
> +			struct dma_chan *rx_chan;
> +			struct dma_chan *cmd_chan;
> +		};
> +		struct {
> +			struct dma_chan *chan;
> +			unsigned int cmd_crci;
> +			unsigned int data_crci;
> +		};
> +	};
> +
>  	struct list_head desc_list;
>  
>  	u8		*data_buffer;
> @@ -261,6 +272,7 @@ struct qcom_nand_controller {
>  	int		buf_start;
>  
>  	__le32 *reg_read_buf;
> +	dma_addr_t reg_read_buf_phys;
>  	int reg_read_pos;
>  
>  	struct nandc_regs *regs;
> @@ -1956,16 +1968,48 @@ static int qcom_nandc_alloc(struct qcom_nand_controller *nandc)
>  	if (!nandc->regs)
>  		return -ENOMEM;
>  
> -	nandc->reg_read_buf = devm_kzalloc(nandc->dev,
> -				MAX_REG_RD * sizeof(*nandc->reg_read_buf),
> -				GFP_KERNEL);
> -	if (!nandc->reg_read_buf)
> -		return -ENOMEM;
>  
> -	nandc->chan = dma_request_slave_channel(nandc->dev, "rxtx");
> -	if (!nandc->chan) {
> -		dev_err(nandc->dev, "failed to request slave channel\n");
> -		return -ENODEV;
> +	if (!nandc->dma_bam_enabled) {

 Better to swap if, else part to avoid !

> +		nandc->reg_read_buf =
> +			devm_kzalloc(nandc->dev, MAX_REG_RD *
> +				     sizeof(*nandc->reg_read_buf), GFP_KERNEL);
> +
> +		if (!nandc->reg_read_buf)
> +			return -ENOMEM;
> +
> +		nandc->chan = dma_request_slave_channel(nandc->dev, "rxtx");
> +		if (!nandc->chan) {
> +			dev_err(nandc->dev,
> +				"failed to request slave channel\n");
> +			return -ENODEV;
> +		}
> +	} else {
> +		nandc->reg_read_buf =
> +			dmam_alloc_coherent(nandc->dev, MAX_REG_RD *
> +					    sizeof(*nandc->reg_read_buf),
> +					    &nandc->reg_read_buf_phys,
> +					    GFP_KERNEL);
> +
> +		if (!nandc->reg_read_buf)
> +			return -ENOMEM;
> +
> +		nandc->tx_chan = dma_request_slave_channel(nandc->dev, "tx");
> +		if (!nandc->tx_chan) {
> +			dev_err(nandc->dev, "failed to request tx channel\n");
> +			return -ENODEV;
> +		}
> +
> +		nandc->rx_chan = dma_request_slave_channel(nandc->dev, "rx");
> +		if (!nandc->rx_chan) {
> +			dev_err(nandc->dev, "failed to request rx channel\n");
> +			return -ENODEV;
> +		}
> +
> +		nandc->cmd_chan = dma_request_slave_channel(nandc->dev, "cmd");
> +		if (!nandc->cmd_chan) {
> +			dev_err(nandc->dev, "failed to request cmd channel\n");
> +			return -ENODEV;
> +		}
>  	}
>  
>  	INIT_LIST_HEAD(&nandc->desc_list);
> @@ -1978,7 +2022,19 @@ static int qcom_nandc_alloc(struct qcom_nand_controller *nandc)
>  
>  static void qcom_nandc_unalloc(struct qcom_nand_controller *nandc)
>  {
> -	dma_release_channel(nandc->chan);
> +	if (nandc->dma_bam_enabled) {
> +		if (nandc->tx_chan)
> +			dma_release_channel(nandc->tx_chan);
> +
> +		if (nandc->rx_chan)
> +			dma_release_channel(nandc->rx_chan);
> +
> +		if (nandc->cmd_chan)
> +			dma_release_channel(nandc->cmd_chan);
> +	} else {
> +		if (nandc->chan)
> +			dma_release_channel(nandc->chan);
> +	}
>  }
>  
>  /* one time setup of a few nand controller registers */
> @@ -2063,16 +2119,20 @@ static int qcom_nandc_parse_dt(struct platform_device *pdev)
>  	struct device_node *np = nandc->dev->of_node;
>  	int ret;
>  
> -	ret = of_property_read_u32(np, "qcom,cmd-crci", &nandc->cmd_crci);
> -	if (ret) {
> -		dev_err(nandc->dev, "command CRCI unspecified\n");
> -		return ret;
> -	}
> +	if (!nandc->dma_bam_enabled) {

 instead, can we simply do if (is_bam) return; here ?

Regards,
 Sricharan

-- 
"QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Powered by blists - more mailing lists