Re: [PATCH v5 11/15] spi: cadence-quadspi: Add compatible for AMD Pensando Elba SoC

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Brad,

On 13/06/22 12:56PM, Brad Larson wrote:
> From: Brad Larson <blarson@xxxxxxx>
> 
> The AMD Pensando Elba SoC has the Cadence QSPI controller integrated.
> 
> The quirk CQSPI_NEEDS_APB_AHB_HAZARD_WAR is added and if enabled
> a dummy readback from the controller is performed to ensure
> synchronization.
> 
> Signed-off-by: Brad Larson <blarson@xxxxxxx>
> ---
>  drivers/spi/spi-cadence-quadspi.c | 19 +++++++++++++++++++
>  1 file changed, 19 insertions(+)
> 
> diff --git a/drivers/spi/spi-cadence-quadspi.c b/drivers/spi/spi-cadence-quadspi.c
> index 72b1a5a2298c..ebb77ea8e6ba 100644
> --- a/drivers/spi/spi-cadence-quadspi.c
> +++ b/drivers/spi/spi-cadence-quadspi.c
> @@ -39,6 +39,7 @@
>  #define CQSPI_DISABLE_DAC_MODE		BIT(1)
>  #define CQSPI_SUPPORT_EXTERNAL_DMA	BIT(2)
>  #define CQSPI_NO_SUPPORT_WR_COMPLETION	BIT(3)
> +#define CQSPI_NEEDS_APB_AHB_HAZARD_WAR	BIT(4)
>  
>  /* Capabilities */
>  #define CQSPI_SUPPORTS_OCTAL		BIT(0)
> @@ -87,6 +88,7 @@ struct cqspi_st {
>  	bool			use_dma_read;
>  	u32			pd_dev_id;
>  	bool			wr_completion;
> +	bool			apb_ahb_hazard;
>  };
>  
>  struct cqspi_driver_platdata {
> @@ -952,6 +954,13 @@ static int cqspi_indirect_write_execute(struct cqspi_flash_pdata *f_pdata,
>  	if (cqspi->wr_delay)
>  		ndelay(cqspi->wr_delay);
>  
> +	/*
> +	 * If a hazard exists between the APB and AHB interfaces, perform a
> +	 * dummy readback from the controller to ensure synchronization.
> +	 */

This is needed for TI's SoCs as well. APB and AHB accesses are 
independent of each other on the interconnect and can be racy. I wrote a 
couple patches [0][1] to fix this on TI's fork. I never got around to 
sending them upstream. It would be great if you can pick those up. They 
fix the race in all paths, not just indirect write.

I would also prefer if we do this unconditionally. I don't think it has 
much downside even on platforms that do not strictly need this.

[0] https://git.ti.com/cgit/ti-linux-kernel/ti-linux-kernel/commit/drivers/spi/spi-cadence-quadspi.c?h=ti-linux-5.10.y&id=027f03a8512086e5ef05dc4e4ff53b2628848f95
[1] https://git.ti.com/cgit/ti-linux-kernel/ti-linux-kernel/commit/drivers/spi/spi-cadence-quadspi.c?h=ti-linux-5.10.y&id=4c367e58bab7d3f9c470c3778441f73546f20398

> +	if (cqspi->apb_ahb_hazard)
> +		(void)readl(reg_base + CQSPI_REG_INDIRECTWR);
> +
>  	while (remaining > 0) {
>  		size_t write_words, mod_bytes;
>  
> @@ -1667,6 +1676,8 @@ static int cqspi_probe(struct platform_device *pdev)
>  			cqspi->use_dma_read = true;
>  		if (ddata->quirks & CQSPI_NO_SUPPORT_WR_COMPLETION)
>  			cqspi->wr_completion = false;
> +		if (ddata->quirks & CQSPI_NEEDS_APB_AHB_HAZARD_WAR)
> +			cqspi->apb_ahb_hazard = true;
>  
>  		if (of_device_is_compatible(pdev->dev.of_node,
>  					    "xlnx,versal-ospi-1.0"))
> @@ -1789,6 +1800,10 @@ static const struct cqspi_driver_platdata versal_ospi = {
>  	.get_dma_status = cqspi_get_versal_dma_status,
>  };
>  
> +static const struct cqspi_driver_platdata pen_cdns_qspi = {
> +	.quirks = CQSPI_NEEDS_APB_AHB_HAZARD_WAR | CQSPI_DISABLE_DAC_MODE,
> +};
> +
>  static const struct of_device_id cqspi_dt_ids[] = {
>  	{
>  		.compatible = "cdns,qspi-nor",
> @@ -1814,6 +1829,10 @@ static const struct of_device_id cqspi_dt_ids[] = {
>  		.compatible = "intel,socfpga-qspi",
>  		.data = &socfpga_qspi,
>  	},
> +	{
> +		.compatible = "amd,pensando-elba-qspi",
> +		.data = &pen_cdns_qspi,
> +	},
>  	{ /* end of table */ }
>  };
>  
> -- 
> 2.17.1
> 

-- 
Regards,
Pratyush Yadav
Texas Instruments Inc.



[Index of Archives]     [Device Tree Compilter]     [Device Tree Spec]     [Linux Driver Backports]     [Video for Linux]     [Linux USB Devel]     [Linux PCI Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Yosemite Backpacking]


  Powered by Linux