Re: [U-Boot] [PATCH 07/10] ufs: Add glue layer driver for TI J721E devices

On 09/09/19 1:49 PM, Faiz Abbas wrote:

Add glue layer driver for the controller present on TI's J721E devices.

Signed-off-by: Faiz Abbas faiz_abbas@ti.com

drivers/ufs/Kconfig | 6 +++ drivers/ufs/Makefile | 1 + drivers/ufs/ti-j721e-ufs.c | 75 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 82 insertions(+) create mode 100644 drivers/ufs/ti-j721e-ufs.c

diff --git a/drivers/ufs/Kconfig b/drivers/ufs/Kconfig index a320b4561f..4875e9448e 100644 --- a/drivers/ufs/Kconfig +++ b/drivers/ufs/Kconfig @@ -14,4 +14,10 @@ config CADENCE_UFS This selects the platform driver for the Cadence UFS host controller present on present TI's J721e devices.

+config TI_J721E_UFS

• bool "Glue Layer driver for UFS on TI J721E devices"
• help

•  This selects the glue layer driver for Cadence controller
  

•  present on TI's J721E devices.
  

endmenu diff --git a/drivers/ufs/Makefile b/drivers/ufs/Makefile index 9262bd6cd0..62ed016608 100644 --- a/drivers/ufs/Makefile +++ b/drivers/ufs/Makefile @@ -5,3 +5,4 @@

obj-$(CONFIG_UFS) += ufs.o ufs-uclass.o obj-$(CONFIG_CADENCE_UFS) += cdns-platform.o +obj-$(CONFIG_TI_J721E_UFS) += ti-j721e-ufs.o diff --git a/drivers/ufs/ti-j721e-ufs.c b/drivers/ufs/ti-j721e-ufs.c new file mode 100644 index 0000000000..249d3796c0 --- /dev/null +++ b/drivers/ufs/ti-j721e-ufs.c @@ -0,0 +1,75 @@ +// SPDX-License-Identifier: GPL-2.0+ +/*

• • Copyright (C) 2019 Texas Instruments Incorporated - http://www.ti.com/
• */

+#include <clk.h> +#include <common.h> +#include <dm/device.h> +#include <dm/read.h>

#include <dm.h>

should be sufficient instead of two includes.

+#include <power-domain.h> +#include <regmap.h>

+#define UFS_SS_CTRL 0x4 +#define UFS_SS_RST_N_PCS BIT(0) +#define UFS_SS_CLK_26MHZ BIT(4)

+static int ti_j721e_ufs_probe(struct udevice *dev) +{

• struct power_domain ufs_pwrdmn;
• struct regmap *map;
• unsigned int clock;
• struct clk clk;
• u32 reg = 0;
• int ret;
• ret = power_domain_get(dev, &ufs_pwrdmn);
• if (ret) {

• dev_err(dev, "failed to get power domain %d\n", ret);
  

• return ret;
  

• }
• ret = power_domain_on(&ufs_pwrdmn);
• if (ret) {

• dev_err(dev, "Power domain on failed\n");
  

• return ret;
  

• }

Core takes care of enabling power domain before calling probe. So this can be dropped

• ret = regmap_init_mem(dev_ofnode(dev), &map);
• if (ret)

• return ret;
  

I think regmap is an overkill here.. We are just accessing a single register and within dedicated wrapper space. Just use regular IO accessors.

• ret = clk_get_by_index(dev, 0, &clk);
• if (ret) {

• dev_err(dev, "failed to get M-PHY clock\n");
  

• return ret;
  

• }
• clock = clk_get_rate(&clk);
• if (IS_ERR_VALUE(clock)) {

• dev_err(dev, "failed to get rate\n");
  

• return ret;
  

• }
• if (clock == 26000000)

• reg |= UFS_SS_CLK_26MHZ;
  

• /* Take UFS slave device out of reset */
• reg |= UFS_SS_RST_N_PCS;
• regmap_write(map, UFS_SS_CTRL, reg);

To be safe, put the slave back to reset before jumping to kernel so that card does not see any glitches when kernel is doing any re-initialization of clocks etc. This can be done in driver remove callback (and needs DM_FLAG_OS_PREPARE flag to be set in driver declaration)

• return dm_scan_fdt_dev(dev);

+}

+static const struct udevice_id ti_j721e_ufs_ids[] = {

• {

• .compatible = "ti,j721e-ufs",
  

• },
• {},

+};

+U_BOOT_DRIVER(ti_j721e_ufs) = {

• .name = "ti-j721e-ufs",
• .id = UCLASS_UFS,
• .of_match = ti_j721e_ufs_ids,
• .probe = ti_j721e_ufs_probe,

+};