2020-04-17 18:11:47

by Alim Akhtar

[permalink] [raw]
Subject: [PATCH v6 0/10] exynos-ufs: Add support for UFS HCI

This patch-set introduces UFS (Universal Flash Storage) host controller support
for Samsung family SoC. Mostly, it consists of UFS PHY and host specific driver.

- Changes since v5:
* re-introduce various quicks which was removed because of no driver
* consumer of those quirks, initial 4 patches does the same.
* Added Reviewed-by tags
* rebased on top of v5.7-rc1
* included Kiwoong's patch in this series, which this driver needs

- Changes since v4:
* Addressed review comments from Avir and Rob
* Minor improvment on the ufs phy and ufshc drivers
* Added Tested-by from Pawel
* Change UFS binding to DT schema format


- Changes since v3:
* Addressed Kishon's and Avir's review comments
* fixed make dt_binding_check error as pointed by Rob

- Changes since v2:
* fixed build warning by kbuild test robot
* Added Reported-by tags

- Changes since v1:
* fixed make dt_binding_check error as pointed by Rob
* Addressed Krzysztof's review comments
* Added Reviewed-by tags

Note: This series is based on Linux-5.7-rc1 (commit: 8f3d9f354286)

Alim Akhtar (9):
scsi: ufs: add quirk to fix mishandling utrlclr/utmrlclr
scsi: ufs: add quirk to disallow reset of interrupt aggregation
scsi: ufs: add quirk to enable host controller without hce
scsi: ufs: introduce UFSHCD_QUIRK_PRDT_BYTE_GRAN quirk
dt-bindings: phy: Document Samsung UFS PHY bindings
phy: samsung-ufs: add UFS PHY driver for samsung SoC
dt-bindings: ufs: Add DT binding documentation for ufs
scsi: ufs-exynos: add UFS host support for Exynos SoCs
arm64: dts: Add node for ufs exynos7

Kiwoong Kim (1):
scsi: ufs: add quirk to fix abnormal ocs fatal error

.../bindings/phy/samsung,ufs-phy.yaml | 74 +
.../bindings/ufs/samsung,exynos-ufs.yaml | 93 ++
.../boot/dts/exynos/exynos7-espresso.dts | 4 +
arch/arm64/boot/dts/exynos/exynos7.dtsi | 44 +-
drivers/phy/samsung/Kconfig | 9 +
drivers/phy/samsung/Makefile | 1 +
drivers/phy/samsung/phy-exynos7-ufs.h | 85 ++
drivers/phy/samsung/phy-samsung-ufs.c | 369 +++++
drivers/phy/samsung/phy-samsung-ufs.h | 142 ++
drivers/scsi/ufs/Kconfig | 12 +
drivers/scsi/ufs/Makefile | 1 +
drivers/scsi/ufs/ufs-exynos.c | 1289 +++++++++++++++++
drivers/scsi/ufs/ufs-exynos.h | 284 ++++
drivers/scsi/ufs/ufshcd.c | 126 +-
drivers/scsi/ufs/ufshcd.h | 29 +
drivers/scsi/ufs/unipro.h | 36 +
16 files changed, 2584 insertions(+), 14 deletions(-)
create mode 100644 Documentation/devicetree/bindings/phy/samsung,ufs-phy.yaml
create mode 100644 Documentation/devicetree/bindings/ufs/samsung,exynos-ufs.yaml
create mode 100644 drivers/phy/samsung/phy-exynos7-ufs.h
create mode 100644 drivers/phy/samsung/phy-samsung-ufs.c
create mode 100644 drivers/phy/samsung/phy-samsung-ufs.h
create mode 100644 drivers/scsi/ufs/ufs-exynos.c
create mode 100644 drivers/scsi/ufs/ufs-exynos.h


base-commit: 8f3d9f354286745c751374f5f1fcafee6b3f3136
--
2.17.1


2020-04-17 18:11:52

by Alim Akhtar

[permalink] [raw]
Subject: [PATCH v6 01/10] scsi: ufs: add quirk to fix mishandling utrlclr/utmrlclr

In the right behavior, setting the bit to '0' indicates clear and '1'
indicates no change. If host controller handles this the other way,
UFSHCI_QUIRK_BROKEN_REQ_LIST_CLR can be used.

Signed-off-by: Seungwon Jeon <[email protected]>
Signed-off-by: Alim Akhtar <[email protected]>
---
drivers/scsi/ufs/ufshcd.c | 11 +++++++++--
drivers/scsi/ufs/ufshcd.h | 5 +++++
2 files changed, 14 insertions(+), 2 deletions(-)

diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
index 698e8d20b4ba..3655b88fc862 100644
--- a/drivers/scsi/ufs/ufshcd.c
+++ b/drivers/scsi/ufs/ufshcd.c
@@ -645,7 +645,11 @@ static inline int ufshcd_get_tr_ocs(struct ufshcd_lrb *lrbp)
*/
static inline void ufshcd_utrl_clear(struct ufs_hba *hba, u32 pos)
{
- ufshcd_writel(hba, ~(1 << pos), REG_UTP_TRANSFER_REQ_LIST_CLEAR);
+ if (hba->quirks & UFSHCI_QUIRK_BROKEN_REQ_LIST_CLR)
+ ufshcd_writel(hba, (1 << pos), REG_UTP_TRANSFER_REQ_LIST_CLEAR);
+ else
+ ufshcd_writel(hba, ~(1 << pos),
+ REG_UTP_TRANSFER_REQ_LIST_CLEAR);
}

/**
@@ -655,7 +659,10 @@ static inline void ufshcd_utrl_clear(struct ufs_hba *hba, u32 pos)
*/
static inline void ufshcd_utmrl_clear(struct ufs_hba *hba, u32 pos)
{
- ufshcd_writel(hba, ~(1 << pos), REG_UTP_TASK_REQ_LIST_CLEAR);
+ if (hba->quirks & UFSHCI_QUIRK_BROKEN_REQ_LIST_CLR)
+ ufshcd_writel(hba, (1 << pos), REG_UTP_TASK_REQ_LIST_CLEAR);
+ else
+ ufshcd_writel(hba, ~(1 << pos), REG_UTP_TASK_REQ_LIST_CLEAR);
}

/**
diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
index 6ffc08ad85f6..071f0edf3f64 100644
--- a/drivers/scsi/ufs/ufshcd.h
+++ b/drivers/scsi/ufs/ufshcd.h
@@ -518,6 +518,11 @@ enum ufshcd_quirks {
* ops (get_ufs_hci_version) to get the correct version.
*/
UFSHCD_QUIRK_BROKEN_UFS_HCI_VERSION = 1 << 5,
+
+ /*
+ * Clear handling for transfer/task request list is just opposite.
+ */
+ UFSHCI_QUIRK_BROKEN_REQ_LIST_CLR = 1 << 6,
};

enum ufshcd_caps {

base-commit: 8f3d9f354286745c751374f5f1fcafee6b3f3136
--
2.17.1

2020-04-17 18:12:06

by Alim Akhtar

[permalink] [raw]
Subject: [PATCH v6 04/10] scsi: ufs: introduce UFSHCD_QUIRK_PRDT_BYTE_GRAN quirk

Some UFS host controllers may think granularities of PRDT length and
offset as bytes, not double words.

Signed-off-by: Kiwoong Kim <[email protected]>
Signed-off-by: Alim Akhtar <[email protected]>
---
drivers/scsi/ufs/ufshcd.c | 30 +++++++++++++++++++++++-------
drivers/scsi/ufs/ufshcd.h | 6 ++++++
2 files changed, 29 insertions(+), 7 deletions(-)

diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
index ee30ed6cc805..b32fcedcdcb9 100644
--- a/drivers/scsi/ufs/ufshcd.c
+++ b/drivers/scsi/ufs/ufshcd.c
@@ -2151,8 +2151,14 @@ static int ufshcd_map_sg(struct ufs_hba *hba, struct ufshcd_lrb *lrbp)
return sg_segments;

if (sg_segments) {
- lrbp->utr_descriptor_ptr->prd_table_length =
- cpu_to_le16((u16)sg_segments);
+
+ if (hba->quirks & UFSHCD_QUIRK_PRDT_BYTE_GRAN)
+ lrbp->utr_descriptor_ptr->prd_table_length =
+ cpu_to_le16((u16)(sg_segments *
+ sizeof(struct ufshcd_sg_entry)));
+ else
+ lrbp->utr_descriptor_ptr->prd_table_length =
+ cpu_to_le16((u16) (sg_segments));

prd_table = (struct ufshcd_sg_entry *)lrbp->ucd_prdt_ptr;

@@ -3500,11 +3506,21 @@ static void ufshcd_host_memory_configure(struct ufs_hba *hba)
cpu_to_le32(upper_32_bits(cmd_desc_element_addr));

/* Response upiu and prdt offset should be in double words */
- utrdlp[i].response_upiu_offset =
- cpu_to_le16(response_offset >> 2);
- utrdlp[i].prd_table_offset = cpu_to_le16(prdt_offset >> 2);
- utrdlp[i].response_upiu_length =
- cpu_to_le16(ALIGNED_UPIU_SIZE >> 2);
+ if (hba->quirks & UFSHCD_QUIRK_PRDT_BYTE_GRAN) {
+ utrdlp[i].response_upiu_offset =
+ cpu_to_le16(response_offset);
+ utrdlp[i].prd_table_offset =
+ cpu_to_le16(prdt_offset);
+ utrdlp[i].response_upiu_length =
+ cpu_to_le16(ALIGNED_UPIU_SIZE);
+ } else {
+ utrdlp[i].response_upiu_offset =
+ cpu_to_le16(response_offset >> 2);
+ utrdlp[i].prd_table_offset =
+ cpu_to_le16(prdt_offset >> 2);
+ utrdlp[i].response_upiu_length =
+ cpu_to_le16(ALIGNED_UPIU_SIZE >> 2);
+ }

ufshcd_init_lrb(hba, &hba->lrb[i], i);
}
diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
index f8d08cb9caf7..a9b9ace9fc72 100644
--- a/drivers/scsi/ufs/ufshcd.h
+++ b/drivers/scsi/ufs/ufshcd.h
@@ -535,6 +535,12 @@ enum ufshcd_quirks {
* enabled via HCE register.
*/
UFSHCI_QUIRK_BROKEN_HCE = 1 << 8,
+
+ /*
+ * This quirk needs to be enabled if the host controller regards
+ * resolution of the values of PRDTO and PRDTL in UTRD as byte.
+ */
+ UFSHCD_QUIRK_PRDT_BYTE_GRAN = 1 << 9,
};

enum ufshcd_caps {
--
2.17.1

2020-04-17 18:12:11

by Alim Akhtar

[permalink] [raw]
Subject: [PATCH v6 07/10] phy: samsung-ufs: add UFS PHY driver for samsung SoC

This patch introduces Samsung UFS PHY driver. This driver
supports to deal with phy calibration and power control
according to UFS host driver's behavior.

Reviewed-by: Kiwoong Kim <[email protected]>
Signed-off-by: Seungwon Jeon <[email protected]>
Signed-off-by: Alim Akhtar <[email protected]>
Cc: Kishon Vijay Abraham I <[email protected]>
Tested-by: Paweł Chmiel <[email protected]>
---
drivers/phy/samsung/Kconfig | 9 +
drivers/phy/samsung/Makefile | 1 +
drivers/phy/samsung/phy-exynos7-ufs.h | 85 ++++++
drivers/phy/samsung/phy-samsung-ufs.c | 369 ++++++++++++++++++++++++++
drivers/phy/samsung/phy-samsung-ufs.h | 142 ++++++++++
5 files changed, 606 insertions(+)
create mode 100644 drivers/phy/samsung/phy-exynos7-ufs.h
create mode 100644 drivers/phy/samsung/phy-samsung-ufs.c
create mode 100644 drivers/phy/samsung/phy-samsung-ufs.h

diff --git a/drivers/phy/samsung/Kconfig b/drivers/phy/samsung/Kconfig
index 9e483d1fdaf2..fc1e3c17f842 100644
--- a/drivers/phy/samsung/Kconfig
+++ b/drivers/phy/samsung/Kconfig
@@ -29,6 +29,15 @@ config PHY_EXYNOS_PCIE
Enable PCIe PHY support for Exynos SoC series.
This driver provides PHY interface for Exynos PCIe controller.

+config PHY_SAMSUNG_UFS
+ tristate "SAMSUNG SoC series UFS PHY driver"
+ depends on OF && (ARCH_EXYNOS || COMPILE_TEST)
+ select GENERIC_PHY
+ help
+ Enable this to support the Samsung UFS PHY driver for
+ Samsung SoCs. This driver provides the interface for UFS
+ host controller to do PHY related programming.
+
config PHY_SAMSUNG_USB2
tristate "Samsung USB 2.0 PHY driver"
depends on HAS_IOMEM
diff --git a/drivers/phy/samsung/Makefile b/drivers/phy/samsung/Makefile
index db9b1aa0de6e..3959100fe8a2 100644
--- a/drivers/phy/samsung/Makefile
+++ b/drivers/phy/samsung/Makefile
@@ -2,6 +2,7 @@
obj-$(CONFIG_PHY_EXYNOS_DP_VIDEO) += phy-exynos-dp-video.o
obj-$(CONFIG_PHY_EXYNOS_MIPI_VIDEO) += phy-exynos-mipi-video.o
obj-$(CONFIG_PHY_EXYNOS_PCIE) += phy-exynos-pcie.o
+obj-$(CONFIG_PHY_SAMSUNG_UFS) += phy-samsung-ufs.o
obj-$(CONFIG_PHY_SAMSUNG_USB2) += phy-exynos-usb2.o
phy-exynos-usb2-y += phy-samsung-usb2.o
phy-exynos-usb2-$(CONFIG_PHY_EXYNOS4210_USB2) += phy-exynos4210-usb2.o
diff --git a/drivers/phy/samsung/phy-exynos7-ufs.h b/drivers/phy/samsung/phy-exynos7-ufs.h
new file mode 100644
index 000000000000..da981c1ac040
--- /dev/null
+++ b/drivers/phy/samsung/phy-exynos7-ufs.h
@@ -0,0 +1,85 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * UFS PHY driver data for Samsung EXYNOS7 SoC
+ *
+ * Copyright (C) 2015 Samsung Electronics Co., Ltd.
+ */
+#ifndef _PHY_EXYNOS7_UFS_H_
+#define _PHY_EXYNOS7_UFS_H_
+
+#include "phy-samsung-ufs.h"
+
+#define EXYNOS7_EMBEDDED_COMBO_PHY_CTRL 0x720
+#define EXYNOS7_EMBEDDED_COMBO_PHY_CTRL_MASK 0x1
+#define EXYNOS7_EMBEDDED_COMBO_PHY_CTRL_EN BIT(0)
+
+/* Calibration for phy initialization */
+static const struct samsung_ufs_phy_cfg exynos7_pre_init_cfg[] = {
+ PHY_COMN_REG_CFG(0x00f, 0xfa, PWR_MODE_ANY),
+ PHY_COMN_REG_CFG(0x010, 0x82, PWR_MODE_ANY),
+ PHY_COMN_REG_CFG(0x011, 0x1e, PWR_MODE_ANY),
+ PHY_COMN_REG_CFG(0x017, 0x84, PWR_MODE_ANY),
+ PHY_TRSV_REG_CFG(0x035, 0x58, PWR_MODE_ANY),
+ PHY_TRSV_REG_CFG(0x036, 0x32, PWR_MODE_ANY),
+ PHY_TRSV_REG_CFG(0x037, 0x40, PWR_MODE_ANY),
+ PHY_TRSV_REG_CFG(0x03b, 0x83, PWR_MODE_ANY),
+ PHY_TRSV_REG_CFG(0x042, 0x88, PWR_MODE_ANY),
+ PHY_TRSV_REG_CFG(0x043, 0xa6, PWR_MODE_ANY),
+ PHY_TRSV_REG_CFG(0x048, 0x74, PWR_MODE_ANY),
+ PHY_TRSV_REG_CFG(0x04c, 0x5b, PWR_MODE_ANY),
+ PHY_TRSV_REG_CFG(0x04d, 0x83, PWR_MODE_ANY),
+ PHY_TRSV_REG_CFG(0x05c, 0x14, PWR_MODE_ANY),
+ END_UFS_PHY_CFG
+};
+
+static const struct samsung_ufs_phy_cfg exynos7_post_init_cfg[] = {
+ END_UFS_PHY_CFG
+};
+
+/* Calibration for HS mode series A/B */
+static const struct samsung_ufs_phy_cfg exynos7_pre_pwr_hs_cfg[] = {
+ PHY_COMN_REG_CFG(0x00f, 0xfa, PWR_MODE_HS_ANY),
+ PHY_COMN_REG_CFG(0x010, 0x82, PWR_MODE_HS_ANY),
+ PHY_COMN_REG_CFG(0x011, 0x1e, PWR_MODE_HS_ANY),
+ /* Setting order: 1st(0x16, 2nd(0x15) */
+ PHY_COMN_REG_CFG(0x016, 0xff, PWR_MODE_HS_ANY),
+ PHY_COMN_REG_CFG(0x015, 0x80, PWR_MODE_HS_ANY),
+ PHY_COMN_REG_CFG(0x017, 0x94, PWR_MODE_HS_ANY),
+ PHY_TRSV_REG_CFG(0x036, 0x32, PWR_MODE_HS_ANY),
+ PHY_TRSV_REG_CFG(0x037, 0x43, PWR_MODE_HS_ANY),
+ PHY_TRSV_REG_CFG(0x038, 0x3f, PWR_MODE_HS_ANY),
+ PHY_TRSV_REG_CFG(0x042, 0x88, PWR_MODE_HS_G2_SER_A),
+ PHY_TRSV_REG_CFG(0x042, 0xbb, PWR_MODE_HS_G2_SER_B),
+ PHY_TRSV_REG_CFG(0x043, 0xa6, PWR_MODE_HS_ANY),
+ PHY_TRSV_REG_CFG(0x048, 0x74, PWR_MODE_HS_ANY),
+ PHY_TRSV_REG_CFG(0x034, 0x35, PWR_MODE_HS_G2_SER_A),
+ PHY_TRSV_REG_CFG(0x034, 0x36, PWR_MODE_HS_G2_SER_B),
+ PHY_TRSV_REG_CFG(0x035, 0x5b, PWR_MODE_HS_G2_SER_A),
+ PHY_TRSV_REG_CFG(0x035, 0x5c, PWR_MODE_HS_G2_SER_B),
+ END_UFS_PHY_CFG
+};
+
+/* Calibration for HS mode series A/B atfer PMC */
+static const struct samsung_ufs_phy_cfg exynos7_post_pwr_hs_cfg[] = {
+ PHY_COMN_REG_CFG(0x015, 0x00, PWR_MODE_HS_ANY),
+ PHY_TRSV_REG_CFG(0x04d, 0x83, PWR_MODE_HS_ANY),
+ END_UFS_PHY_CFG
+};
+
+static const struct samsung_ufs_phy_cfg *exynos7_ufs_phy_cfgs[CFG_TAG_MAX] = {
+ [CFG_PRE_INIT] = exynos7_pre_init_cfg,
+ [CFG_POST_INIT] = exynos7_post_init_cfg,
+ [CFG_PRE_PWR_HS] = exynos7_pre_pwr_hs_cfg,
+ [CFG_POST_PWR_HS] = exynos7_post_pwr_hs_cfg,
+};
+
+static struct samsung_ufs_phy_drvdata exynos7_ufs_phy = {
+ .cfg = exynos7_ufs_phy_cfgs,
+ .isol = {
+ .offset = EXYNOS7_EMBEDDED_COMBO_PHY_CTRL,
+ .mask = EXYNOS7_EMBEDDED_COMBO_PHY_CTRL_MASK,
+ .en = EXYNOS7_EMBEDDED_COMBO_PHY_CTRL_EN,
+ },
+};
+
+#endif /* _PHY_EXYNOS7_UFS_H_ */
diff --git a/drivers/phy/samsung/phy-samsung-ufs.c b/drivers/phy/samsung/phy-samsung-ufs.c
new file mode 100644
index 000000000000..4c8334bba3e9
--- /dev/null
+++ b/drivers/phy/samsung/phy-samsung-ufs.c
@@ -0,0 +1,369 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * UFS PHY driver for Samsung SoC
+ *
+ * Copyright (C) 2015 Samsung Electronics Co., Ltd.
+ * Author: Seungwon Jeon <[email protected]>
+ * Author: Alim Akhtar <[email protected]>
+ *
+ */
+#include <linux/clk.h>
+#include <linux/delay.h>
+#include <linux/err.h>
+#include <linux/of.h>
+#include <linux/io.h>
+#include <linux/iopoll.h>
+#include <linux/mfd/syscon.h>
+#include <linux/module.h>
+#include <linux/phy/phy.h>
+#include <linux/platform_device.h>
+#include <linux/regmap.h>
+
+#include "phy-samsung-ufs.h"
+
+#define for_each_phy_lane(phy, i) \
+ for (i = 0; i < (phy)->lane_cnt; i++)
+#define for_each_phy_cfg(cfg) \
+ for (; (cfg)->id; (cfg)++)
+
+#define PHY_DEF_LANE_CNT 1
+
+static void samsung_ufs_phy_config(struct samsung_ufs_phy *phy,
+ const struct samsung_ufs_phy_cfg *cfg, u8 lane)
+{
+ enum {LANE_0, LANE_1}; /* lane index */
+
+ switch (lane) {
+ case LANE_0:
+ writel(cfg->val, (phy)->reg_pma + cfg->off_0);
+ break;
+ case LANE_1:
+ if (cfg->id == PHY_TRSV_BLK)
+ writel(cfg->val, (phy)->reg_pma + cfg->off_1);
+ break;
+ }
+}
+
+int samsung_ufs_phy_wait_for_lock_acq(struct phy *phy)
+{
+ struct samsung_ufs_phy *ufs_phy = get_samsung_ufs_phy(phy);
+ const unsigned int timeout_us = 100000;
+ const unsigned int sleep_us = 10;
+ u32 val;
+ int err;
+
+ err = readl_poll_timeout(
+ ufs_phy->reg_pma + PHY_APB_ADDR(PHY_PLL_LOCK_STATUS),
+ val, (val & PHY_PLL_LOCK_BIT), sleep_us, timeout_us);
+ if (err) {
+ dev_err(ufs_phy->dev,
+ "failed to get phy pll lock acquisition %d\n", err);
+ goto out;
+ }
+
+ err = readl_poll_timeout(
+ ufs_phy->reg_pma + PHY_APB_ADDR(PHY_CDR_LOCK_STATUS),
+ val, (val & PHY_CDR_LOCK_BIT), sleep_us, timeout_us);
+ if (err) {
+ dev_err(ufs_phy->dev,
+ "failed to get phy cdr lock acquisition %d\n", err);
+ goto out;
+ }
+
+out:
+ return err;
+}
+
+int samsung_ufs_phy_calibrate(struct phy *phy)
+{
+ struct samsung_ufs_phy *ufs_phy = get_samsung_ufs_phy(phy);
+ struct samsung_ufs_phy_cfg **cfgs = ufs_phy->cfg;
+ const struct samsung_ufs_phy_cfg *cfg;
+ int i;
+ int err = 0;
+
+ if (unlikely(ufs_phy->ufs_phy_state < CFG_PRE_INIT ||
+ ufs_phy->ufs_phy_state >= CFG_TAG_MAX)) {
+ dev_err(ufs_phy->dev, "invalid phy config index %d\n",
+ ufs_phy->ufs_phy_state);
+ return -EINVAL;
+ }
+
+ if (ufs_phy->is_pre_init)
+ ufs_phy->is_pre_init = false;
+ if (ufs_phy->is_post_init) {
+ ufs_phy->is_post_init = false;
+ ufs_phy->ufs_phy_state = CFG_POST_INIT;
+ }
+ if (ufs_phy->is_pre_pmc) {
+ ufs_phy->is_pre_pmc = false;
+ ufs_phy->ufs_phy_state = CFG_PRE_PWR_HS;
+ }
+ if (ufs_phy->is_post_pmc) {
+ ufs_phy->is_post_pmc = false;
+ ufs_phy->ufs_phy_state = CFG_POST_PWR_HS;
+ }
+
+ switch (ufs_phy->ufs_phy_state) {
+ case CFG_PRE_INIT:
+ ufs_phy->is_post_init = true;
+ break;
+ case CFG_POST_INIT:
+ ufs_phy->is_pre_pmc = true;
+ break;
+ case CFG_PRE_PWR_HS:
+ ufs_phy->is_post_pmc = true;
+ break;
+ case CFG_POST_PWR_HS:
+ break;
+ default:
+ dev_err(ufs_phy->dev, "wrong state for phy calibration\n");
+ }
+
+ cfg = cfgs[ufs_phy->ufs_phy_state];
+ if (!cfg)
+ goto out;
+
+ for_each_phy_cfg(cfg) {
+ for_each_phy_lane(ufs_phy, i) {
+ samsung_ufs_phy_config(ufs_phy, cfg, i);
+ }
+ }
+
+ if (ufs_phy->ufs_phy_state == CFG_POST_PWR_HS)
+ err = samsung_ufs_phy_wait_for_lock_acq(phy);
+out:
+ return err;
+}
+
+static int samsung_ufs_phy_symbol_clk_init(struct samsung_ufs_phy *phy)
+{
+ struct clk *clk;
+ int ret = 0;
+
+ clk = devm_clk_get(phy->dev, "tx0_symbol_clk");
+ if (IS_ERR(clk)) {
+ dev_err(phy->dev, "failed to get tx0_symbol_clk clock\n");
+ goto out;
+ } else {
+ phy->tx0_symbol_clk = clk;
+ }
+
+ clk = devm_clk_get(phy->dev, "rx0_symbol_clk");
+ if (IS_ERR(clk)) {
+ dev_err(phy->dev, "failed to get rx0_symbol_clk clock\n");
+ goto out;
+ } else {
+ phy->rx0_symbol_clk = clk;
+ }
+
+ clk = devm_clk_get(phy->dev, "rx1_symbol_clk");
+ if (IS_ERR(clk)) {
+ dev_err(phy->dev, "failed to get rx1_symbol_clk clock\n");
+ goto out;
+ } else {
+ phy->rx1_symbol_clk = clk;
+ }
+
+ ret = clk_prepare_enable(phy->tx0_symbol_clk);
+ if (ret) {
+ dev_err(phy->dev, "%s: tx0_symbol_clk enable failed %d\n",
+ __func__, ret);
+ goto out;
+ }
+ ret = clk_prepare_enable(phy->rx0_symbol_clk);
+ if (ret) {
+ dev_err(phy->dev, "%s: rx0_symbol_clk enable failed %d\n",
+ __func__, ret);
+ goto out;
+ }
+ ret = clk_prepare_enable(phy->rx1_symbol_clk);
+ if (ret) {
+ dev_err(phy->dev, "%s: rx1_symbol_clk enable failed %d\n",
+ __func__, ret);
+ goto out;
+ }
+out:
+ return ret;
+}
+
+static int samsung_ufs_phy_clks_init(struct samsung_ufs_phy *phy)
+{
+ struct clk *phy_ref_clk;
+ int ret;
+
+ phy_ref_clk = devm_clk_get(phy->dev, "ref_clk");
+ if (IS_ERR(phy_ref_clk))
+ dev_err(phy->dev, "failed to get ref_clk clock\n");
+ else
+ phy->ref_clk = phy_ref_clk;
+
+ ret = clk_prepare_enable(phy->ref_clk);
+ if (ret) {
+ dev_err(phy->dev, "%s: ref_clk enable failed %d\n",
+ __func__, ret);
+ return ret;
+ }
+
+ dev_info(phy->dev, "UFS MPHY ref_clk_rate = %ld\n", clk_get_rate(phy_ref_clk));
+
+ return 0;
+}
+
+static int samsung_ufs_phy_init(struct phy *phy)
+{
+ struct samsung_ufs_phy *_phy = get_samsung_ufs_phy(phy);
+ int ret;
+
+ _phy->lane_cnt = phy->attrs.bus_width;
+ _phy->ufs_phy_state = CFG_PRE_INIT;
+
+ _phy->is_pre_init = true;
+ _phy->is_post_init = false;
+ _phy->is_pre_pmc = false;
+ _phy->is_post_pmc = false;
+
+
+ if (of_device_is_compatible(_phy->dev->of_node,
+ "samsung,exynos7-ufs-phy")) {
+ ret = samsung_ufs_phy_symbol_clk_init(_phy);
+ if (ret)
+ dev_err(_phy->dev,
+ "failed to set ufs phy symbol clocks\n");
+ }
+
+ ret = samsung_ufs_phy_clks_init(_phy);
+ if (ret)
+ dev_err(_phy->dev, "failed to set ufs phy clocks\n");
+
+ samsung_ufs_phy_calibrate(phy);
+
+ return 0;
+}
+
+static int samsung_ufs_phy_power_on(struct phy *phy)
+{
+ struct samsung_ufs_phy *_phy = get_samsung_ufs_phy(phy);
+
+ samsung_ufs_phy_ctrl_isol(_phy, false);
+ return 0;
+}
+
+static int samsung_ufs_phy_power_off(struct phy *phy)
+{
+ struct samsung_ufs_phy *_phy = get_samsung_ufs_phy(phy);
+
+ samsung_ufs_phy_ctrl_isol(_phy, true);
+ clk_disable_unprepare(_phy->ref_clk);
+ return 0;
+}
+
+static int samsung_ufs_phy_set_mode(struct phy *generic_phy,
+ enum phy_mode mode, int submode)
+{
+ struct samsung_ufs_phy *_phy = get_samsung_ufs_phy(generic_phy);
+
+ _phy->mode = PHY_MODE_INVALID;
+
+ if (mode > 0)
+ _phy->mode = mode;
+
+ return 0;
+}
+
+static struct phy_ops samsung_ufs_phy_ops = {
+ .init = samsung_ufs_phy_init,
+ .power_on = samsung_ufs_phy_power_on,
+ .power_off = samsung_ufs_phy_power_off,
+ .calibrate = samsung_ufs_phy_calibrate,
+ .set_mode = samsung_ufs_phy_set_mode,
+}
+;
+static const struct of_device_id samsung_ufs_phy_match[];
+
+static int samsung_ufs_phy_probe(struct platform_device *pdev)
+{
+ struct device *dev = &pdev->dev;
+ struct resource *res;
+ const struct of_device_id *match;
+ struct samsung_ufs_phy *phy;
+ struct phy *gen_phy;
+ struct phy_provider *phy_provider;
+ const struct samsung_ufs_phy_drvdata *drvdata;
+ int err = 0;
+
+ match = of_match_node(samsung_ufs_phy_match, dev->of_node);
+ if (!match) {
+ err = -EINVAL;
+ dev_err(dev, "failed to get match_node\n");
+ goto out;
+ }
+
+ phy = devm_kzalloc(dev, sizeof(*phy), GFP_KERNEL);
+ if (!phy) {
+ err = -ENOMEM;
+ goto out;
+ }
+
+ res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "phy-pma");
+ phy->reg_pma = devm_ioremap_resource(dev, res);
+ if (IS_ERR(phy->reg_pma)) {
+ err = PTR_ERR(phy->reg_pma);
+ goto out;
+ }
+
+ phy->reg_pmu = syscon_regmap_lookup_by_phandle(
+ dev->of_node, "samsung,pmu-syscon");
+ if (IS_ERR(phy->reg_pmu)) {
+ err = PTR_ERR(phy->reg_pmu);
+ dev_err(dev, "failed syscon remap for pmu\n");
+ goto out;
+ }
+
+ gen_phy = devm_phy_create(dev, NULL, &samsung_ufs_phy_ops);
+ if (IS_ERR(gen_phy)) {
+ err = PTR_ERR(gen_phy);
+ dev_err(dev, "failed to create PHY for ufs-phy\n");
+ goto out;
+ }
+
+ drvdata = match->data;
+ phy->dev = dev;
+ phy->drvdata = drvdata;
+ phy->cfg = (struct samsung_ufs_phy_cfg **)drvdata->cfg;
+ phy->isol = &drvdata->isol;
+ phy->lane_cnt = PHY_DEF_LANE_CNT;
+
+ phy_set_drvdata(gen_phy, phy);
+
+ phy_provider = devm_of_phy_provider_register(dev, of_phy_simple_xlate);
+ if (IS_ERR(phy_provider)) {
+ err = PTR_ERR(phy_provider);
+ dev_err(dev, "failed to register phy-provider\n");
+ goto out;
+ }
+out:
+ return err;
+}
+
+static const struct of_device_id samsung_ufs_phy_match[] = {
+ {
+ .compatible = "samsung,exynos7-ufs-phy",
+ .data = &exynos7_ufs_phy,
+ },
+ {},
+};
+MODULE_DEVICE_TABLE(of, samsung_ufs_phy_match);
+
+static struct platform_driver samsung_ufs_phy_driver = {
+ .probe = samsung_ufs_phy_probe,
+ .driver = {
+ .name = "samsung-ufs-phy",
+ .of_match_table = samsung_ufs_phy_match,
+ },
+};
+module_platform_driver(samsung_ufs_phy_driver);
+MODULE_DESCRIPTION("Samsung SoC UFS PHY Driver");
+MODULE_AUTHOR("Seungwon Jeon <[email protected]>");
+MODULE_AUTHOR("Alim Akhtar <[email protected]>");
+MODULE_LICENSE("GPL v2");
diff --git a/drivers/phy/samsung/phy-samsung-ufs.h b/drivers/phy/samsung/phy-samsung-ufs.h
new file mode 100644
index 000000000000..27dc1b573469
--- /dev/null
+++ b/drivers/phy/samsung/phy-samsung-ufs.h
@@ -0,0 +1,142 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * UFS PHY driver for Samsung EXYNOS SoC
+ *
+ * Copyright (C) 2015 Samsung Electronics Co., Ltd.
+ * Author: Seungwon Jeon <[email protected]>
+ * Author: Alim Akhtar <[email protected]>
+ *
+ */
+#ifndef _PHY_SAMSUNG_UFS_
+#define _PHY_SAMSUNG_UFS_
+
+#define PHY_COMN_BLK 1
+#define PHY_TRSV_BLK 2
+#define END_UFS_PHY_CFG { 0 }
+#define PHY_TRSV_CH_OFFSET 0x30
+#define PHY_APB_ADDR(off) ((off) << 2)
+
+#define PHY_COMN_REG_CFG(o, v, d) { \
+ .off_0 = PHY_APB_ADDR((o)), \
+ .off_1 = 0, \
+ .val = (v), \
+ .desc = (d), \
+ .id = PHY_COMN_BLK, \
+}
+
+#define PHY_TRSV_REG_CFG(o, v, d) { \
+ .off_0 = PHY_APB_ADDR((o)), \
+ .off_1 = PHY_APB_ADDR((o) + PHY_TRSV_CH_OFFSET), \
+ .val = (v), \
+ .desc = (d), \
+ .id = PHY_TRSV_BLK, \
+}
+
+/* UFS PHY registers */
+#define PHY_PLL_LOCK_STATUS 0x1e
+#define PHY_CDR_LOCK_STATUS 0x5e
+
+#define PHY_PLL_LOCK_BIT BIT(5)
+#define PHY_CDR_LOCK_BIT BIT(4)
+
+/* description for PHY calibration */
+enum {
+ /* applicable to any */
+ PWR_DESC_ANY = 0,
+ /* mode */
+ PWR_DESC_PWM = 1,
+ PWR_DESC_HS = 2,
+ /* series */
+ PWR_DESC_SER_A = 1,
+ PWR_DESC_SER_B = 2,
+ /* gear */
+ PWR_DESC_G1 = 1,
+ PWR_DESC_G2 = 2,
+ PWR_DESC_G3 = 3,
+ /* field mask */
+ MD_MASK = 0x3,
+ SR_MASK = 0x3,
+ GR_MASK = 0x7,
+};
+
+#define PWR_MODE_HS_G1_ANY PWR_MODE_HS(PWR_DESC_G1, PWR_DESC_ANY)
+#define PWR_MODE_HS_G1_SER_A PWR_MODE_HS(PWR_DESC_G1, PWR_DESC_SER_A)
+#define PWR_MODE_HS_G1_SER_B PWR_MODE_HS(PWR_DESC_G1, PWR_DESC_SER_B)
+#define PWR_MODE_HS_G2_ANY PWR_MODE_HS(PWR_DESC_G2, PWR_DESC_ANY)
+#define PWR_MODE_HS_G2_SER_A PWR_MODE_HS(PWR_DESC_G2, PWR_DESC_SER_A)
+#define PWR_MODE_HS_G2_SER_B PWR_MODE_HS(PWR_DESC_G2, PWR_DESC_SER_B)
+#define PWR_MODE_HS_G3_ANY PWR_MODE_HS(PWR_DESC_G3, PWR_DESC_ANY)
+#define PWR_MODE_HS_G3_SER_A PWR_MODE_HS(PWR_DESC_G3, PWR_DESC_SER_A)
+#define PWR_MODE_HS_G3_SER_B PWR_MODE_HS(PWR_DESC_G3, PWR_DESC_SER_B)
+#define PWR_MODE(g, s, m) ((((g) & GR_MASK) << 4) |\
+ (((s) & SR_MASK) << 2) | ((m) & MD_MASK))
+#define PWR_MODE_PWM_ANY PWR_MODE(PWR_DESC_ANY,\
+ PWR_DESC_ANY, PWR_DESC_PWM)
+#define PWR_MODE_HS(g, s) ((((g) & GR_MASK) << 4) |\
+ (((s) & SR_MASK) << 2) | PWR_DESC_HS)
+#define PWR_MODE_HS_ANY PWR_MODE(PWR_DESC_ANY,\
+ PWR_DESC_ANY, PWR_DESC_HS)
+#define PWR_MODE_ANY PWR_MODE(PWR_DESC_ANY,\
+ PWR_DESC_ANY, PWR_DESC_ANY)
+/* PHY calibration point/state */
+enum {
+ CFG_PRE_INIT,
+ CFG_POST_INIT,
+ CFG_PRE_PWR_HS,
+ CFG_POST_PWR_HS,
+ CFG_TAG_MAX,
+};
+
+struct samsung_ufs_phy_cfg {
+ u32 off_0;
+ u32 off_1;
+ u32 val;
+ u8 desc;
+ u8 id;
+};
+
+struct samsung_ufs_phy_drvdata {
+ const struct samsung_ufs_phy_cfg **cfg;
+ struct pmu_isol {
+ u32 offset;
+ u32 mask;
+ u32 en;
+ } isol;
+};
+
+struct samsung_ufs_phy {
+ struct device *dev;
+ void __iomem *reg_pma;
+ struct regmap *reg_pmu;
+ struct clk *ref_clk;
+ struct clk *ref_clk_parent;
+ struct clk *tx0_symbol_clk;
+ struct clk *rx0_symbol_clk;
+ struct clk *rx1_symbol_clk;
+ const struct samsung_ufs_phy_drvdata *drvdata;
+ struct samsung_ufs_phy_cfg **cfg;
+ const struct pmu_isol *isol;
+ u8 lane_cnt;
+ int ufs_phy_state;
+ enum phy_mode mode;
+ bool is_pre_init;
+ bool is_post_init;
+ bool is_pre_pmc;
+ bool is_post_pmc;
+};
+
+static inline struct samsung_ufs_phy *get_samsung_ufs_phy(struct phy *phy)
+{
+ return (struct samsung_ufs_phy *)phy_get_drvdata(phy);
+}
+
+static inline void samsung_ufs_phy_ctrl_isol(
+ struct samsung_ufs_phy *phy, u32 isol)
+{
+ regmap_update_bits(phy->reg_pmu, phy->isol->offset,
+ phy->isol->mask, isol ? 0 : phy->isol->en);
+}
+
+#include "phy-exynos7-ufs.h"
+
+#endif /* _PHY_SAMSUNG_UFS_ */
--
2.17.1

2020-04-17 18:12:20

by Alim Akhtar

[permalink] [raw]
Subject: [PATCH v6 06/10] dt-bindings: phy: Document Samsung UFS PHY bindings

This patch documents Samsung UFS PHY device tree bindings

Signed-off-by: Alim Akhtar <[email protected]>
Tested-by: Paweł Chmiel <[email protected]>
---
.../bindings/phy/samsung,ufs-phy.yaml | 74 +++++++++++++++++++
1 file changed, 74 insertions(+)
create mode 100644 Documentation/devicetree/bindings/phy/samsung,ufs-phy.yaml

diff --git a/Documentation/devicetree/bindings/phy/samsung,ufs-phy.yaml b/Documentation/devicetree/bindings/phy/samsung,ufs-phy.yaml
new file mode 100644
index 000000000000..352d5dda320d
--- /dev/null
+++ b/Documentation/devicetree/bindings/phy/samsung,ufs-phy.yaml
@@ -0,0 +1,74 @@
+# SPDX-License-Identifier: GPL-2.0-only OR BSD-2-Clause
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/phy/samsung,ufs-phy.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Samsung SoC series UFS PHY Device Tree Bindings
+
+maintainers:
+ - Alim Akhtar <[email protected]>
+
+properties:
+ "#phy-cells":
+ const: 0
+
+ compatible:
+ enum:
+ - samsung,exynos7-ufs-phy
+
+ reg:
+ maxItems: 1
+ description: PHY base register address
+
+ reg-names:
+ items:
+ - const: phy-pma
+
+ clocks:
+ items:
+ - description: PLL reference clock
+ - description: symbol clock for input symbol ( rx0-ch0 symbol clock)
+ - description: symbol clock for input symbol ( rx1-ch1 symbol clock)
+ - description: symbol clock for output symbol ( tx0 symbol clock)
+
+ clock-names:
+ items:
+ - const: ref_clk
+ - const: rx1_symbol_clk
+ - const: rx0_symbol_clk
+ - const: tx0_symbol_clk
+
+ samsung,pmu-syscon:
+ $ref: '/schemas/types.yaml#/definitions/phandle'
+ description: phandle for PMU system controller interface, used to
+ control pmu registers bits for ufs m-phy
+
+required:
+ - "#phy-cells"
+ - compatible
+ - reg
+ - reg-names
+ - clocks
+ - clock-names
+ - samsung,pmu-syscon
+
+examples:
+ - |
+ #include <dt-bindings/clock/exynos7-clk.h>
+
+ ufs_phy: ufs-phy@15571800 {
+ compatible = "samsung,exynos7-ufs-phy";
+ reg = <0x15571800 0x240>;
+ reg-names = "phy-pma";
+ samsung,pmu-syscon = <&pmu_system_controller>;
+ #phy-cells = <0>;
+ clocks = <&clock_fsys1 SCLK_COMBO_PHY_EMBEDDED_26M>,
+ <&clock_fsys1 PHYCLK_UFS20_RX1_SYMBOL_USER>,
+ <&clock_fsys1 PHYCLK_UFS20_RX0_SYMBOL_USER>,
+ <&clock_fsys1 PHYCLK_UFS20_TX0_SYMBOL_USER>;
+ clock-names = "ref_clk", "rx1_symbol_clk",
+ "rx0_symbol_clk", "tx0_symbol_clk";
+
+ };
+...
--
2.17.1

2020-04-17 18:12:22

by Alim Akhtar

[permalink] [raw]
Subject: [PATCH v6 10/10] arm64: dts: Add node for ufs exynos7

Adding dt node foe UFS and UFS-PHY for exynos7 SoC.

Signed-off-by: Alim Akhtar <[email protected]>
Tested-by: Paweł Chmiel <[email protected]>
---
.../boot/dts/exynos/exynos7-espresso.dts | 4 ++
arch/arm64/boot/dts/exynos/exynos7.dtsi | 44 ++++++++++++++++++-
2 files changed, 46 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/boot/dts/exynos/exynos7-espresso.dts b/arch/arm64/boot/dts/exynos/exynos7-espresso.dts
index 7af288fa9475..790f12ca8981 100644
--- a/arch/arm64/boot/dts/exynos/exynos7-espresso.dts
+++ b/arch/arm64/boot/dts/exynos/exynos7-espresso.dts
@@ -406,6 +406,10 @@
};
};

+&ufs {
+ status = "okay";
+};
+
&usbdrd_phy {
vbus-supply = <&usb30_vbus_reg>;
vbus-boost-supply = <&usb3drd_boost_5v>;
diff --git a/arch/arm64/boot/dts/exynos/exynos7.dtsi b/arch/arm64/boot/dts/exynos/exynos7.dtsi
index 5558045637ac..0c1ebd3ea294 100644
--- a/arch/arm64/boot/dts/exynos/exynos7.dtsi
+++ b/arch/arm64/boot/dts/exynos/exynos7.dtsi
@@ -220,9 +220,14 @@
#clock-cells = <1>;
clocks = <&fin_pll>, <&clock_top1 DOUT_ACLK_FSYS1_200>,
<&clock_top1 DOUT_SCLK_MMC0>,
- <&clock_top1 DOUT_SCLK_MMC1>;
+ <&clock_top1 DOUT_SCLK_MMC1>,
+ <&clock_top1 DOUT_SCLK_UFSUNIPRO20>,
+ <&clock_top1 DOUT_SCLK_PHY_FSYS1>,
+ <&clock_top1 DOUT_SCLK_PHY_FSYS1_26M>;
clock-names = "fin_pll", "dout_aclk_fsys1_200",
- "dout_sclk_mmc0", "dout_sclk_mmc1";
+ "dout_sclk_mmc0", "dout_sclk_mmc1",
+ "dout_sclk_ufsunipro20", "dout_sclk_phy_fsys1",
+ "dout_sclk_phy_fsys1_26m";
};

serial_0: serial@13630000 {
@@ -601,6 +606,41 @@
};
};

+ ufs: ufs@15570000 {
+ compatible = "samsung,exynos7-ufs";
+ reg = <0x15570000 0x100>, /* 0: HCI standard */
+ <0x15570100 0x100>, /* 1: Vendor specificed */
+ <0x15571000 0x200>, /* 2: UNIPRO */
+ <0x15572000 0x300>; /* 3: UFS protector */
+ reg-names = "hci", "vs_hci", "unipro", "ufsp";
+ interrupts = <GIC_SPI 200 IRQ_TYPE_LEVEL_HIGH>;
+ clocks = <&clock_fsys1 ACLK_UFS20_LINK>,
+ <&clock_fsys1 SCLK_UFSUNIPRO20_USER>;
+ clock-names = "core_clk", "sclk_unipro_main";
+ freq-table-hz = <0 0>, <0 0>;
+ pinctrl-names = "default";
+ pinctrl-0 = <&ufs_rst_n &ufs_refclk_out>;
+ pclk-freq-avail-range = <70000000 133000000>;
+ phys = <&ufs_phy>;
+ phy-names = "ufs-phy";
+ status = "disabled";
+ };
+
+ ufs_phy: ufs-phy@0x15571800 {
+ compatible = "samsung,exynos7-ufs-phy";
+ reg = <0x15571800 0x240>;
+ reg-names = "phy-pma";
+ samsung,pmu-syscon = <&pmu_system_controller>;
+ #phy-cells = <0>;
+ clocks = <&clock_fsys1 SCLK_COMBO_PHY_EMBEDDED_26M>,
+ <&clock_fsys1 PHYCLK_UFS20_RX1_SYMBOL_USER>,
+ <&clock_fsys1 PHYCLK_UFS20_RX0_SYMBOL_USER>,
+ <&clock_fsys1 PHYCLK_UFS20_TX0_SYMBOL_USER>;
+ clock-names = "ref_clk", "rx1_symbol_clk",
+ "rx0_symbol_clk",
+ "tx0_symbol_clk";
+ };
+
usbdrd_phy: phy@15500000 {
compatible = "samsung,exynos7-usbdrd-phy";
reg = <0x15500000 0x100>;
--
2.17.1

2020-04-17 18:12:30

by Alim Akhtar

[permalink] [raw]
Subject: [PATCH v6 09/10] scsi: ufs-exynos: add UFS host support for Exynos SoCs

This patch introduces Exynos UFS host controller driver,
which mainly handles vendor-specific operations including
link startup, power mode change and hibernation/unhibernation.

Reported-by: kbuild test robot <[email protected]>
Reported-by: Julia Lawall <[email protected]>
[robot: drivers/scsi/ufs/ufs-exynos.c:931:8-10:
WARNING: possible condition with no effect (if == else)
]
Reviewed-by: Kiwoong Kim <[email protected]>
Reviewed-by: Avri Altman <[email protected]>
Signed-off-by: Seungwon Jeon <[email protected]>
Signed-off-by: Alim Akhtar <[email protected]>
Tested-by: Paweł Chmiel <[email protected]>
---
drivers/scsi/ufs/Kconfig | 12 +
drivers/scsi/ufs/Makefile | 1 +
drivers/scsi/ufs/ufs-exynos.c | 1289 +++++++++++++++++++++++++++++++++
drivers/scsi/ufs/ufs-exynos.h | 284 ++++++++
drivers/scsi/ufs/unipro.h | 33 +
5 files changed, 1619 insertions(+)
create mode 100644 drivers/scsi/ufs/ufs-exynos.c
create mode 100644 drivers/scsi/ufs/ufs-exynos.h

diff --git a/drivers/scsi/ufs/Kconfig b/drivers/scsi/ufs/Kconfig
index e2005aeddc2d..cc7e29c8c24f 100644
--- a/drivers/scsi/ufs/Kconfig
+++ b/drivers/scsi/ufs/Kconfig
@@ -160,3 +160,15 @@ config SCSI_UFS_BSG

Select this if you need a bsg device node for your UFS controller.
If unsure, say N.
+
+config SCSI_UFS_EXYNOS
+ bool "EXYNOS specific hooks to UFS controller platform driver"
+ depends on SCSI_UFSHCD_PLATFORM && ARCH_EXYNOS || COMPILE_TEST
+ select PHY_SAMSUNG_UFS
+ help
+ This selects the EXYNOS specific additions to UFSHCD platform driver.
+ UFS host on EXYNOS includes HCI and UNIPRO layer, and associates with
+ UFS-PHY driver.
+
+ Select this if you have UFS host controller on EXYNOS chipset.
+ If unsure, say N.
diff --git a/drivers/scsi/ufs/Makefile b/drivers/scsi/ufs/Makefile
index 94c6c5d7334b..f0c5b95ec9cc 100644
--- a/drivers/scsi/ufs/Makefile
+++ b/drivers/scsi/ufs/Makefile
@@ -4,6 +4,7 @@ obj-$(CONFIG_SCSI_UFS_DWC_TC_PCI) += tc-dwc-g210-pci.o ufshcd-dwc.o tc-dwc-g210.
obj-$(CONFIG_SCSI_UFS_DWC_TC_PLATFORM) += tc-dwc-g210-pltfrm.o ufshcd-dwc.o tc-dwc-g210.o
obj-$(CONFIG_SCSI_UFS_CDNS_PLATFORM) += cdns-pltfrm.o
obj-$(CONFIG_SCSI_UFS_QCOM) += ufs-qcom.o
+obj-$(CONFIG_SCSI_UFS_EXYNOS) += ufs-exynos.o
obj-$(CONFIG_SCSI_UFSHCD) += ufshcd-core.o
ufshcd-core-y += ufshcd.o ufs-sysfs.o
ufshcd-core-$(CONFIG_SCSI_UFS_BSG) += ufs_bsg.o
diff --git a/drivers/scsi/ufs/ufs-exynos.c b/drivers/scsi/ufs/ufs-exynos.c
new file mode 100644
index 000000000000..068390f4b6ff
--- /dev/null
+++ b/drivers/scsi/ufs/ufs-exynos.c
@@ -0,0 +1,1289 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * UFS Host Controller driver for Exynos specific extensions
+ *
+ * Copyright (C) 2014-2015 Samsung Electronics Co., Ltd.
+ * Author: Seungwon Jeon <[email protected]>
+ * Author: Alim Akhtar <[email protected]>
+ *
+ */
+
+#include <linux/clk.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/phy/phy.h>
+#include <linux/platform_device.h>
+
+#include "ufshcd.h"
+#include "ufshcd-pltfrm.h"
+#include "ufshci.h"
+#include "unipro.h"
+
+#include "ufs-exynos.h"
+
+/*
+ * Exynos's Vendor specific registers for UFSHCI
+ */
+#define HCI_TXPRDT_ENTRY_SIZE 0x00
+#define PRDT_PREFECT_EN BIT(31)
+#define PRDT_SET_SIZE(x) ((x) & 0x1F)
+#define HCI_RXPRDT_ENTRY_SIZE 0x04
+#define HCI_1US_TO_CNT_VAL 0x0C
+#define CNT_VAL_1US_MASK 0x3FF
+#define HCI_UTRL_NEXUS_TYPE 0x40
+#define HCI_UTMRL_NEXUS_TYPE 0x44
+#define HCI_SW_RST 0x50
+#define UFS_LINK_SW_RST BIT(0)
+#define UFS_UNIPRO_SW_RST BIT(1)
+#define UFS_SW_RST_MASK (UFS_UNIPRO_SW_RST | UFS_LINK_SW_RST)
+#define HCI_DATA_REORDER 0x60
+#define HCI_UNIPRO_APB_CLK_CTRL 0x68
+#define UNIPRO_APB_CLK(v, x) (((v) & ~0xF) | ((x) & 0xF))
+#define HCI_AXIDMA_RWDATA_BURST_LEN 0x6C
+#define HCI_GPIO_OUT 0x70
+#define HCI_ERR_EN_PA_LAYER 0x78
+#define HCI_ERR_EN_DL_LAYER 0x7C
+#define HCI_ERR_EN_N_LAYER 0x80
+#define HCI_ERR_EN_T_LAYER 0x84
+#define HCI_ERR_EN_DME_LAYER 0x88
+#define HCI_CLKSTOP_CTRL 0xB0
+#define REFCLK_STOP BIT(2)
+#define UNIPRO_MCLK_STOP BIT(1)
+#define UNIPRO_PCLK_STOP BIT(0)
+#define CLK_STOP_MASK (REFCLK_STOP |\
+ UNIPRO_MCLK_STOP |\
+ UNIPRO_PCLK_STOP)
+#define HCI_MISC 0xB4
+#define REFCLK_CTRL_EN BIT(7)
+#define UNIPRO_PCLK_CTRL_EN BIT(6)
+#define UNIPRO_MCLK_CTRL_EN BIT(5)
+#define HCI_CORECLK_CTRL_EN BIT(4)
+#define CLK_CTRL_EN_MASK (REFCLK_CTRL_EN |\
+ UNIPRO_PCLK_CTRL_EN |\
+ UNIPRO_MCLK_CTRL_EN)
+/* Device fatal error */
+#define DFES_ERR_EN BIT(31)
+#define DFES_DEF_L2_ERRS (UIC_DATA_LINK_LAYER_ERROR_RX_BUF_OF |\
+ UIC_DATA_LINK_LAYER_ERROR_PA_INIT)
+#define DFES_DEF_L3_ERRS (UIC_NETWORK_UNSUPPORTED_HEADER_TYPE |\
+ UIC_NETWORK_BAD_DEVICEID_ENC |\
+ UIC_NETWORK_LHDR_TRAP_PACKET_DROPPING)
+#define DFES_DEF_L4_ERRS (UIC_TRANSPORT_UNSUPPORTED_HEADER_TYPE |\
+ UIC_TRANSPORT_UNKNOWN_CPORTID |\
+ UIC_TRANSPORT_NO_CONNECTION_RX |\
+ UIC_TRANSPORT_BAD_TC)
+
+enum {
+ UNIPRO_L1_5 = 0,/* PHY Adapter */
+ UNIPRO_L2, /* Data Link */
+ UNIPRO_L3, /* Network */
+ UNIPRO_L4, /* Transport */
+ UNIPRO_DME, /* DME */
+};
+
+/*
+ * UNIPRO registers
+ */
+#define UNIPRO_COMP_VERSION 0x000
+#define UNIPRO_DME_PWR_REQ 0x090
+#define UNIPRO_DME_PWR_REQ_POWERMODE 0x094
+#define UNIPRO_DME_PWR_REQ_LOCALL2TIMER0 0x098
+#define UNIPRO_DME_PWR_REQ_LOCALL2TIMER1 0x09C
+#define UNIPRO_DME_PWR_REQ_LOCALL2TIMER2 0x0A0
+#define UNIPRO_DME_PWR_REQ_REMOTEL2TIMER0 0x0A4
+#define UNIPRO_DME_PWR_REQ_REMOTEL2TIMER1 0x0A8
+#define UNIPRO_DME_PWR_REQ_REMOTEL2TIMER2 0x0AC
+
+/*
+ * UFS Protector registers
+ */
+#define UFSPRSECURITY 0x010
+#define NSSMU BIT(14)
+#define UFSPSBEGIN0 0x200
+#define UFSPSEND0 0x204
+#define UFSPSLUN0 0x208
+#define UFSPSCTRL0 0x20C
+
+#define CNTR_DIV_VAL 40
+
+static void exynos_ufs_auto_ctrl_hcc(struct exynos_ufs *ufs, bool en);
+static void exynos_ufs_ctrl_clkstop(struct exynos_ufs *ufs, bool en);
+
+static inline void exynos_ufs_enable_auto_ctrl_hcc(struct exynos_ufs *ufs)
+{
+ exynos_ufs_auto_ctrl_hcc(ufs, true);
+}
+
+static inline void exynos_ufs_disable_auto_ctrl_hcc(struct exynos_ufs *ufs)
+{
+ exynos_ufs_auto_ctrl_hcc(ufs, false);
+}
+
+static inline void exynos_ufs_disable_auto_ctrl_hcc_save(
+ struct exynos_ufs *ufs, u32 *val)
+{
+ *val = hci_readl(ufs, HCI_MISC);
+ exynos_ufs_auto_ctrl_hcc(ufs, false);
+}
+
+static inline void exynos_ufs_auto_ctrl_hcc_restore(
+ struct exynos_ufs *ufs, u32 *val)
+{
+ hci_writel(ufs, *val, HCI_MISC);
+}
+
+static inline void exynos_ufs_gate_clks(struct exynos_ufs *ufs)
+{
+ exynos_ufs_ctrl_clkstop(ufs, true);
+}
+
+static inline void exynos_ufs_ungate_clks(struct exynos_ufs *ufs)
+{
+ exynos_ufs_ctrl_clkstop(ufs, false);
+}
+
+static int exynos7_ufs_drv_init(struct device *dev, struct exynos_ufs *ufs)
+{
+ return 0;
+}
+
+static int exynos7_ufs_pre_link(struct exynos_ufs *ufs)
+{
+ struct ufs_hba *hba = ufs->hba;
+ u32 val = ufs->drv_data->uic_attr->pa_dbg_option_suite;
+ int i;
+
+ exynos_ufs_enable_ov_tm(hba);
+ for_each_ufs_tx_lane(ufs, i)
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(0x297, i), 0x17);
+ for_each_ufs_rx_lane(ufs, i) {
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(0x362, i), 0xff);
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(0x363, i), 0x00);
+ }
+ exynos_ufs_disable_ov_tm(hba);
+
+ for_each_ufs_tx_lane(ufs, i)
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(TX_HIBERN8_CONTROL, i), 0x0);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_TXPHY_CFGUPDT), 0x1);
+ udelay(1);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_OPTION_SUITE), val | (1 << 12));
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_SKIP_RESET_PHY), 0x1);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_SKIP_LINE_RESET), 0x1);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_LINE_RESET_REQ), 0x1);
+ udelay(1600);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_OPTION_SUITE), val);
+
+ return 0;
+}
+
+static int exynos7_ufs_post_link(struct exynos_ufs *ufs)
+{
+ struct ufs_hba *hba = ufs->hba;
+ int i;
+
+ exynos_ufs_enable_ov_tm(hba);
+ for_each_ufs_tx_lane(ufs, i) {
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(0x28b, i), 0x83);
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(0x29a, i), 0x07);
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(0x277, i),
+ TX_LINERESET_N(exynos_ufs_calc_time_cntr(ufs, 200000)));
+ }
+ exynos_ufs_disable_ov_tm(hba);
+
+ exynos_ufs_enable_dbg_mode(hba);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_SAVECONFIGTIME), 0xbb8);
+ exynos_ufs_disable_dbg_mode(hba);
+
+ return 0;
+}
+
+static int exynos7_ufs_pre_pwr_change(struct exynos_ufs *ufs,
+ struct ufs_pa_layer_attr *pwr)
+{
+ unipro_writel(ufs, 0x22, UNIPRO_DBG_FORCE_DME_CTRL_STATE);
+
+ return 0;
+}
+
+static int exynos7_ufs_post_pwr_change(struct exynos_ufs *ufs,
+ struct ufs_pa_layer_attr *pwr)
+{
+ struct ufs_hba *hba = ufs->hba;
+ int lanes = max_t(u32, pwr->lane_rx, pwr->lane_tx);
+
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_RXPHY_CFGUPDT), 0x1);
+
+ if (lanes == 1) {
+ exynos_ufs_enable_dbg_mode(hba);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_CONNECTEDTXDATALANES), 0x1);
+ exynos_ufs_disable_dbg_mode(hba);
+ }
+
+ return 0;
+}
+
+/**
+ * exynos_ufs_auto_ctrl_hcc - HCI core clock control by h/w
+ * Control should be disabled in the below cases
+ * - Before host controller S/W reset
+ * - Access to UFS protector's register
+ */
+static void exynos_ufs_auto_ctrl_hcc(struct exynos_ufs *ufs, bool en)
+{
+ u32 misc = hci_readl(ufs, HCI_MISC);
+
+ if (en)
+ hci_writel(ufs, misc | HCI_CORECLK_CTRL_EN, HCI_MISC);
+ else
+ hci_writel(ufs, misc & ~HCI_CORECLK_CTRL_EN, HCI_MISC);
+}
+
+static void exynos_ufs_ctrl_clkstop(struct exynos_ufs *ufs, bool en)
+{
+ u32 ctrl = hci_readl(ufs, HCI_CLKSTOP_CTRL);
+ u32 misc = hci_readl(ufs, HCI_MISC);
+
+ if (en) {
+ hci_writel(ufs, misc | CLK_CTRL_EN_MASK, HCI_MISC);
+ hci_writel(ufs, ctrl | CLK_STOP_MASK, HCI_CLKSTOP_CTRL);
+ } else {
+ hci_writel(ufs, ctrl & ~CLK_STOP_MASK, HCI_CLKSTOP_CTRL);
+ hci_writel(ufs, misc & ~CLK_CTRL_EN_MASK, HCI_MISC);
+ }
+}
+
+static int exynos_ufs_get_clk_info(struct exynos_ufs *ufs)
+{
+ struct ufs_hba *hba = ufs->hba;
+ struct list_head *head = &hba->clk_list_head;
+ struct ufs_clk_info *clki;
+ u32 pclk_rate;
+ u32 f_min, f_max;
+ u8 div = 0;
+ int ret = 0;
+
+ if (!head || list_empty(head))
+ goto out;
+
+ list_for_each_entry(clki, head, list) {
+ if (!IS_ERR(clki->clk)) {
+ if (!strcmp(clki->name, "core_clk"))
+ ufs->clk_hci_core = clki->clk;
+ else if (!strcmp(clki->name, "sclk_unipro_main"))
+ ufs->clk_unipro_main = clki->clk;
+ }
+ }
+
+ if (!ufs->clk_hci_core || !ufs->clk_unipro_main) {
+ dev_err(hba->dev, "failed to get clk info\n");
+ ret = -EINVAL;
+ goto out;
+ }
+
+ ufs->mclk_rate = clk_get_rate(ufs->clk_unipro_main);
+ pclk_rate = clk_get_rate(ufs->clk_hci_core);
+ f_min = ufs->pclk_avail_min;
+ f_max = ufs->pclk_avail_max;
+
+ if (ufs->opts & EXYNOS_UFS_OPT_HAS_APB_CLK_CTRL) {
+ do {
+ pclk_rate /= (div + 1);
+
+ if (pclk_rate <= f_max)
+ break;
+ div++;
+ } while (pclk_rate >= f_min);
+ }
+
+ if (unlikely(pclk_rate < f_min || pclk_rate > f_max)) {
+ dev_err(hba->dev, "not available pclk range %d\n", pclk_rate);
+ ret = -EINVAL;
+ goto out;
+ }
+
+ ufs->pclk_rate = pclk_rate;
+ ufs->pclk_div = div;
+
+out:
+ return ret;
+}
+
+static void exynos_ufs_set_unipro_pclk_div(struct exynos_ufs *ufs)
+{
+ if (ufs->opts & EXYNOS_UFS_OPT_HAS_APB_CLK_CTRL) {
+ u32 val;
+
+ val = hci_readl(ufs, HCI_UNIPRO_APB_CLK_CTRL);
+ hci_writel(ufs, UNIPRO_APB_CLK(val, ufs->pclk_div),
+ HCI_UNIPRO_APB_CLK_CTRL);
+ }
+}
+
+static void exynos_ufs_set_pwm_clk_div(struct exynos_ufs *ufs)
+{
+ struct ufs_hba *hba = ufs->hba;
+ struct exynos_ufs_uic_attr *attr = ufs->drv_data->uic_attr;
+
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB(CMN_PWM_CLK_CTRL), attr->cmn_pwm_clk_ctrl);
+}
+
+static void exynos_ufs_calc_pwm_clk_div(struct exynos_ufs *ufs)
+{
+ struct ufs_hba *hba = ufs->hba;
+ struct exynos_ufs_uic_attr *attr = ufs->drv_data->uic_attr;
+ const unsigned int div = 30, mult = 20;
+ const unsigned long pwm_min = 3 * 1000 * 1000;
+ const unsigned long pwm_max = 9 * 1000 * 1000;
+ const int divs[] = {32, 16, 8, 4};
+ unsigned long clk = 0, _clk, clk_period;
+ int i = 0, clk_idx = -1;
+
+ clk_period = UNIPRO_PCLK_PERIOD(ufs);
+ for (i = 0; i < ARRAY_SIZE(divs); i++) {
+ _clk = NSEC_PER_SEC * mult / (clk_period * divs[i] * div);
+ if (_clk >= pwm_min && _clk <= pwm_max) {
+ if (_clk > clk) {
+ clk_idx = i;
+ clk = _clk;
+ }
+ }
+ }
+
+ if (clk_idx == -1) {
+ ufshcd_dme_get(hba, UIC_ARG_MIB(CMN_PWM_CLK_CTRL), &clk_idx);
+ dev_err(hba->dev,
+ "failed to decide pwm clock divider, will not change\n");
+ }
+
+ attr->cmn_pwm_clk_ctrl = clk_idx & PWM_CLK_CTRL_MASK;
+}
+
+long exynos_ufs_calc_time_cntr(struct exynos_ufs *ufs, long period)
+{
+ const int precise = 10;
+ long pclk_rate = ufs->pclk_rate;
+ long clk_period, fraction;
+
+ clk_period = UNIPRO_PCLK_PERIOD(ufs);
+ fraction = ((NSEC_PER_SEC % pclk_rate) * precise) / pclk_rate;
+
+ return (period * precise) / ((clk_period * precise) + fraction);
+}
+
+static void exynos_ufs_specify_phy_time_attr(struct exynos_ufs *ufs)
+{
+ struct exynos_ufs_uic_attr *attr = ufs->drv_data->uic_attr;
+ struct ufs_phy_time_cfg *t_cfg = &ufs->t_cfg;
+
+ t_cfg->tx_linereset_p =
+ exynos_ufs_calc_time_cntr(ufs, attr->tx_dif_p_nsec);
+ t_cfg->tx_linereset_n =
+ exynos_ufs_calc_time_cntr(ufs, attr->tx_dif_n_nsec);
+ t_cfg->tx_high_z_cnt =
+ exynos_ufs_calc_time_cntr(ufs, attr->tx_high_z_cnt_nsec);
+ t_cfg->tx_base_n_val =
+ exynos_ufs_calc_time_cntr(ufs, attr->tx_base_unit_nsec);
+ t_cfg->tx_gran_n_val =
+ exynos_ufs_calc_time_cntr(ufs, attr->tx_gran_unit_nsec);
+ t_cfg->tx_sleep_cnt =
+ exynos_ufs_calc_time_cntr(ufs, attr->tx_sleep_cnt);
+
+ t_cfg->rx_linereset =
+ exynos_ufs_calc_time_cntr(ufs, attr->rx_dif_p_nsec);
+ t_cfg->rx_hibern8_wait =
+ exynos_ufs_calc_time_cntr(ufs, attr->rx_hibern8_wait_nsec);
+ t_cfg->rx_base_n_val =
+ exynos_ufs_calc_time_cntr(ufs, attr->rx_base_unit_nsec);
+ t_cfg->rx_gran_n_val =
+ exynos_ufs_calc_time_cntr(ufs, attr->rx_gran_unit_nsec);
+ t_cfg->rx_sleep_cnt =
+ exynos_ufs_calc_time_cntr(ufs, attr->rx_sleep_cnt);
+ t_cfg->rx_stall_cnt =
+ exynos_ufs_calc_time_cntr(ufs, attr->rx_stall_cnt);
+}
+
+static void exynos_ufs_config_phy_time_attr(struct exynos_ufs *ufs)
+{
+ struct ufs_hba *hba = ufs->hba;
+ struct ufs_phy_time_cfg *t_cfg = &ufs->t_cfg;
+ int i;
+
+ exynos_ufs_set_pwm_clk_div(ufs);
+
+ exynos_ufs_enable_ov_tm(hba);
+
+ for_each_ufs_rx_lane(ufs, i) {
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(RX_FILLER_ENABLE, i),
+ ufs->drv_data->uic_attr->rx_filler_enable);
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(RX_LINERESET_VAL, i),
+ RX_LINERESET(t_cfg->rx_linereset));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(RX_BASE_NVAL_07_00, i),
+ RX_BASE_NVAL_L(t_cfg->rx_base_n_val));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(RX_BASE_NVAL_15_08, i),
+ RX_BASE_NVAL_H(t_cfg->rx_base_n_val));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(RX_GRAN_NVAL_07_00, i),
+ RX_GRAN_NVAL_L(t_cfg->rx_gran_n_val));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(RX_GRAN_NVAL_10_08, i),
+ RX_GRAN_NVAL_H(t_cfg->rx_gran_n_val));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(RX_OV_SLEEP_CNT_TIMER, i),
+ RX_OV_SLEEP_CNT(t_cfg->rx_sleep_cnt));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(RX_OV_STALL_CNT_TIMER, i),
+ RX_OV_STALL_CNT(t_cfg->rx_stall_cnt));
+ }
+
+ for_each_ufs_tx_lane(ufs, i) {
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(TX_LINERESET_P_VAL, i),
+ TX_LINERESET_P(t_cfg->tx_linereset_p));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(TX_HIGH_Z_CNT_07_00, i),
+ TX_HIGH_Z_CNT_L(t_cfg->tx_high_z_cnt));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(TX_HIGH_Z_CNT_11_08, i),
+ TX_HIGH_Z_CNT_H(t_cfg->tx_high_z_cnt));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(TX_BASE_NVAL_07_00, i),
+ TX_BASE_NVAL_L(t_cfg->tx_base_n_val));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(TX_BASE_NVAL_15_08, i),
+ TX_BASE_NVAL_H(t_cfg->tx_base_n_val));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(TX_GRAN_NVAL_07_00, i),
+ TX_GRAN_NVAL_L(t_cfg->tx_gran_n_val));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(TX_GRAN_NVAL_10_08, i),
+ TX_GRAN_NVAL_H(t_cfg->tx_gran_n_val));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(TX_OV_SLEEP_CNT_TIMER, i),
+ TX_OV_H8_ENTER_EN |
+ TX_OV_SLEEP_CNT(t_cfg->tx_sleep_cnt));
+ ufshcd_dme_set(hba, UIC_ARG_MIB_SEL(TX_MIN_ACTIVATETIME, i),
+ ufs->drv_data->uic_attr->tx_min_activatetime);
+ }
+
+ exynos_ufs_disable_ov_tm(hba);
+}
+
+static void exynos_ufs_config_phy_cap_attr(struct exynos_ufs *ufs)
+{
+ struct ufs_hba *hba = ufs->hba;
+ struct exynos_ufs_uic_attr *attr = ufs->drv_data->uic_attr;
+ int i;
+
+ exynos_ufs_enable_ov_tm(hba);
+
+ for_each_ufs_rx_lane(ufs, i) {
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(RX_HS_G1_SYNC_LENGTH_CAP, i),
+ attr->rx_hs_g1_sync_len_cap);
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(RX_HS_G2_SYNC_LENGTH_CAP, i),
+ attr->rx_hs_g2_sync_len_cap);
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(RX_HS_G3_SYNC_LENGTH_CAP, i),
+ attr->rx_hs_g3_sync_len_cap);
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(RX_HS_G1_PREP_LENGTH_CAP, i),
+ attr->rx_hs_g1_prep_sync_len_cap);
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(RX_HS_G2_PREP_LENGTH_CAP, i),
+ attr->rx_hs_g2_prep_sync_len_cap);
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(RX_HS_G3_PREP_LENGTH_CAP, i),
+ attr->rx_hs_g3_prep_sync_len_cap);
+ }
+
+ if (attr->rx_adv_fine_gran_sup_en == 0) {
+ for_each_ufs_rx_lane(ufs, i) {
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(RX_ADV_GRANULARITY_CAP, i), 0);
+
+ if (attr->rx_min_actv_time_cap)
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(RX_MIN_ACTIVATETIME_CAP,
+ i), attr->rx_min_actv_time_cap);
+
+ if (attr->rx_hibern8_time_cap)
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(RX_HIBERN8TIME_CAP, i),
+ attr->rx_hibern8_time_cap);
+ }
+ } else if (attr->rx_adv_fine_gran_sup_en == 1) {
+ for_each_ufs_rx_lane(ufs, i) {
+ if (attr->rx_adv_fine_gran_step)
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(RX_ADV_GRANULARITY_CAP,
+ i), RX_ADV_FINE_GRAN_STEP(
+ attr->rx_adv_fine_gran_step));
+
+ if (attr->rx_adv_min_actv_time_cap)
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(
+ RX_ADV_MIN_ACTIVATETIME_CAP, i),
+ attr->rx_adv_min_actv_time_cap);
+
+ if (attr->rx_adv_hibern8_time_cap)
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(RX_ADV_HIBERN8TIME_CAP,
+ i),
+ attr->rx_adv_hibern8_time_cap);
+ }
+ }
+
+ exynos_ufs_disable_ov_tm(hba);
+}
+
+static void exynos_ufs_establish_connt(struct exynos_ufs *ufs)
+{
+ struct ufs_hba *hba = ufs->hba;
+ enum {
+ DEV_ID = 0x00,
+ PEER_DEV_ID = 0x01,
+ PEER_CPORT_ID = 0x00,
+ TRAFFIC_CLASS = 0x00,
+ };
+
+ /* allow cport attributes to be set */
+ ufshcd_dme_set(hba, UIC_ARG_MIB(T_CONNECTIONSTATE), CPORT_IDLE);
+
+ /* local unipro attributes */
+ ufshcd_dme_set(hba, UIC_ARG_MIB(N_DEVICEID), DEV_ID);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(N_DEVICEID_VALID), TRUE);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(T_PEERDEVICEID), PEER_DEV_ID);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(T_PEERCPORTID), PEER_CPORT_ID);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(T_CPORTFLAGS), CPORT_DEF_FLAGS);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(T_TRAFFICCLASS), TRAFFIC_CLASS);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(T_CONNECTIONSTATE), CPORT_CONNECTED);
+}
+
+static void exynos_ufs_config_smu(struct exynos_ufs *ufs)
+{
+ u32 reg, val;
+
+ exynos_ufs_disable_auto_ctrl_hcc_save(ufs, &val);
+
+ /* make encryption disabled by default */
+ reg = ufsp_readl(ufs, UFSPRSECURITY);
+ ufsp_writel(ufs, reg | NSSMU, UFSPRSECURITY);
+ ufsp_writel(ufs, 0x0, UFSPSBEGIN0);
+ ufsp_writel(ufs, 0xffffffff, UFSPSEND0);
+ ufsp_writel(ufs, 0xff, UFSPSLUN0);
+ ufsp_writel(ufs, 0xf1, UFSPSCTRL0);
+
+ exynos_ufs_auto_ctrl_hcc_restore(ufs, &val);
+}
+
+static void exynos_ufs_config_sync_pattern_mask(struct exynos_ufs *ufs,
+ struct ufs_pa_layer_attr *pwr)
+{
+ struct ufs_hba *hba = ufs->hba;
+ u8 g = max_t(u32, pwr->gear_rx, pwr->gear_tx);
+ u32 mask, sync_len;
+ enum {
+ SYNC_LEN_G1 = 80 * 1000, /* 80us */
+ SYNC_LEN_G2 = 40 * 1000, /* 44us */
+ SYNC_LEN_G3 = 20 * 1000, /* 20us */
+ };
+ int i;
+
+ if (g == 1)
+ sync_len = SYNC_LEN_G1;
+ else if (g == 2)
+ sync_len = SYNC_LEN_G2;
+ else if (g == 3)
+ sync_len = SYNC_LEN_G3;
+ else
+ return;
+
+ mask = exynos_ufs_calc_time_cntr(ufs, sync_len);
+ mask = (mask >> 8) & 0xff;
+
+ exynos_ufs_enable_ov_tm(hba);
+
+ for_each_ufs_rx_lane(ufs, i)
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB_SEL(RX_SYNC_MASK_LENGTH, i), mask);
+
+ exynos_ufs_disable_ov_tm(hba);
+}
+
+static int exynos_ufs_pre_pwr_mode(struct ufs_hba *hba,
+ struct ufs_pa_layer_attr *dev_max_params,
+ struct ufs_pa_layer_attr *dev_req_params)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+ struct phy *generic_phy = ufs->phy;
+ struct ufs_dev_params ufs_exynos_cap;
+ int ret;
+
+ if (!dev_req_params) {
+ pr_err("%s: incoming dev_req_params is NULL\n", __func__);
+ ret = -EINVAL;
+ goto out;
+ }
+
+
+ ufs_exynos_cap.tx_lanes = UFS_EXYNOS_LIMIT_NUM_LANES_TX;
+ ufs_exynos_cap.rx_lanes = UFS_EXYNOS_LIMIT_NUM_LANES_RX;
+ ufs_exynos_cap.hs_rx_gear = UFS_EXYNOS_LIMIT_HSGEAR_RX;
+ ufs_exynos_cap.hs_tx_gear = UFS_EXYNOS_LIMIT_HSGEAR_TX;
+ ufs_exynos_cap.pwm_rx_gear = UFS_EXYNOS_LIMIT_PWMGEAR_RX;
+ ufs_exynos_cap.pwm_tx_gear = UFS_EXYNOS_LIMIT_PWMGEAR_TX;
+ ufs_exynos_cap.rx_pwr_pwm = UFS_EXYNOS_LIMIT_RX_PWR_PWM;
+ ufs_exynos_cap.tx_pwr_pwm = UFS_EXYNOS_LIMIT_TX_PWR_PWM;
+ ufs_exynos_cap.rx_pwr_hs = UFS_EXYNOS_LIMIT_RX_PWR_HS;
+ ufs_exynos_cap.tx_pwr_hs = UFS_EXYNOS_LIMIT_TX_PWR_HS;
+ ufs_exynos_cap.hs_rate = UFS_EXYNOS_LIMIT_HS_RATE;
+ ufs_exynos_cap.desired_working_mode =
+ UFS_EXYNOS_LIMIT_DESIRED_MODE;
+
+ ret = ufshcd_get_pwr_dev_param(&ufs_exynos_cap,
+ dev_max_params, dev_req_params);
+ if (ret) {
+ pr_err("%s: failed to determine capabilities\n", __func__);
+ goto out;
+ }
+
+ if (ufs->drv_data->pre_pwr_change)
+ ufs->drv_data->pre_pwr_change(ufs, dev_req_params);
+
+ if (ufshcd_is_hs_mode(dev_req_params)) {
+ exynos_ufs_config_sync_pattern_mask(ufs, dev_req_params);
+
+ switch (dev_req_params->hs_rate) {
+ case PA_HS_MODE_A:
+ case PA_HS_MODE_B:
+ phy_calibrate(generic_phy);
+ break;
+ }
+ }
+
+ return 0;
+out:
+ return ret;
+}
+
+#define PWR_MODE_STR_LEN 64
+static int exynos_ufs_post_pwr_mode(struct ufs_hba *hba,
+ struct ufs_pa_layer_attr *pwr_max,
+ struct ufs_pa_layer_attr *pwr_req)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+ struct phy *generic_phy = ufs->phy;
+ int gear = max_t(u32, pwr_req->gear_rx, pwr_req->gear_tx);
+ int lanes = max_t(u32, pwr_req->lane_rx, pwr_req->lane_tx);
+ char pwr_str[PWR_MODE_STR_LEN] = "";
+
+ /* let default be PWM Gear 1, Lane 1 */
+ if (!gear)
+ gear = 1;
+
+ if (!lanes)
+ lanes = 1;
+
+ if (ufs->drv_data->post_pwr_change)
+ ufs->drv_data->post_pwr_change(ufs, pwr_req);
+
+ if ((ufshcd_is_hs_mode(pwr_req))) {
+ switch (pwr_req->hs_rate) {
+ case PA_HS_MODE_A:
+ case PA_HS_MODE_B:
+ phy_calibrate(generic_phy);
+ break;
+ }
+
+ snprintf(pwr_str, PWR_MODE_STR_LEN, "%s series_%s G_%d L_%d",
+ "FAST", pwr_req->hs_rate == PA_HS_MODE_A ? "A" : "B",
+ gear, lanes);
+ } else {
+ snprintf(pwr_str, PWR_MODE_STR_LEN, "%s G_%d L_%d",
+ "SLOW", gear, lanes);
+ }
+
+ dev_info(hba->dev, "Power mode changed to : %s\n", pwr_str);
+
+ return 0;
+}
+
+static void exynos_ufs_specify_nexus_t_xfer_req(struct ufs_hba *hba,
+ int tag, bool op)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+ u32 type;
+
+ type = hci_readl(ufs, HCI_UTRL_NEXUS_TYPE);
+
+ if (op)
+ hci_writel(ufs, type | (1 << tag), HCI_UTRL_NEXUS_TYPE);
+ else
+ hci_writel(ufs, type & ~(1 << tag), HCI_UTRL_NEXUS_TYPE);
+}
+
+static void exynos_ufs_specify_nexus_t_tm_req(struct ufs_hba *hba,
+ int tag, u8 func)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+ u32 type;
+
+ type = hci_readl(ufs, HCI_UTMRL_NEXUS_TYPE);
+
+ switch (func) {
+ case UFS_ABORT_TASK:
+ case UFS_QUERY_TASK:
+ hci_writel(ufs, type | (1 << tag), HCI_UTMRL_NEXUS_TYPE);
+ break;
+ case UFS_ABORT_TASK_SET:
+ case UFS_CLEAR_TASK_SET:
+ case UFS_LOGICAL_RESET:
+ case UFS_QUERY_TASK_SET:
+ hci_writel(ufs, type & ~(1 << tag), HCI_UTMRL_NEXUS_TYPE);
+ break;
+ }
+}
+
+static void exynos_ufs_phy_init(struct exynos_ufs *ufs)
+{
+ struct ufs_hba *hba = ufs->hba;
+ struct phy *generic_phy = ufs->phy;
+
+ if (ufs->avail_ln_rx == 0 || ufs->avail_ln_tx == 0) {
+ ufshcd_dme_get(hba, UIC_ARG_MIB(PA_AVAILRXDATALANES),
+ &ufs->avail_ln_rx);
+ ufshcd_dme_get(hba, UIC_ARG_MIB(PA_AVAILTXDATALANES),
+ &ufs->avail_ln_tx);
+ WARN(ufs->avail_ln_rx != ufs->avail_ln_tx,
+ "available data lane is not equal(rx:%d, tx:%d)\n",
+ ufs->avail_ln_rx, ufs->avail_ln_tx);
+ }
+
+ phy_set_bus_width(generic_phy, ufs->avail_ln_rx);
+ phy_init(generic_phy);
+}
+
+static void exynos_ufs_config_unipro(struct exynos_ufs *ufs)
+{
+ struct ufs_hba *hba = ufs->hba;
+
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_CLK_PERIOD),
+ DIV_ROUND_UP(NSEC_PER_SEC, ufs->mclk_rate));
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_TXTRAILINGCLOCKS),
+ ufs->drv_data->uic_attr->tx_trailingclks);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_OPTION_SUITE),
+ ufs->drv_data->uic_attr->pa_dbg_option_suite);
+}
+
+static void exynos_ufs_config_intr(struct exynos_ufs *ufs, u32 errs, u8 index)
+{
+ switch (index) {
+ case UNIPRO_L1_5:
+ hci_writel(ufs, DFES_ERR_EN | errs, HCI_ERR_EN_PA_LAYER);
+ break;
+ case UNIPRO_L2:
+ hci_writel(ufs, DFES_ERR_EN | errs, HCI_ERR_EN_DL_LAYER);
+ break;
+ case UNIPRO_L3:
+ hci_writel(ufs, DFES_ERR_EN | errs, HCI_ERR_EN_N_LAYER);
+ break;
+ case UNIPRO_L4:
+ hci_writel(ufs, DFES_ERR_EN | errs, HCI_ERR_EN_T_LAYER);
+ break;
+ case UNIPRO_DME:
+ hci_writel(ufs, DFES_ERR_EN | errs, HCI_ERR_EN_DME_LAYER);
+ break;
+ }
+}
+
+static int exynos_ufs_pre_link(struct ufs_hba *hba)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+
+ /* hci */
+ exynos_ufs_config_intr(ufs, DFES_DEF_L2_ERRS, UNIPRO_L2);
+ exynos_ufs_config_intr(ufs, DFES_DEF_L3_ERRS, UNIPRO_L3);
+ exynos_ufs_config_intr(ufs, DFES_DEF_L4_ERRS, UNIPRO_L4);
+ exynos_ufs_set_unipro_pclk_div(ufs);
+
+ /* unipro */
+ exynos_ufs_config_unipro(ufs);
+
+ /* m-phy */
+ exynos_ufs_phy_init(ufs);
+ exynos_ufs_config_phy_time_attr(ufs);
+ exynos_ufs_config_phy_cap_attr(ufs);
+
+ if (ufs->drv_data->pre_link)
+ ufs->drv_data->pre_link(ufs);
+
+ return 0;
+}
+
+static void exynos_ufs_fit_aggr_timeout(struct exynos_ufs *ufs)
+{
+ u32 val;
+
+ val = exynos_ufs_calc_time_cntr(ufs, IATOVAL_NSEC / CNTR_DIV_VAL);
+ hci_writel(ufs, val & CNT_VAL_1US_MASK, HCI_1US_TO_CNT_VAL);
+}
+
+static int exynos_ufs_post_link(struct ufs_hba *hba)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+ struct phy *generic_phy = ufs->phy;
+ struct exynos_ufs_uic_attr *attr = ufs->drv_data->uic_attr;
+
+ exynos_ufs_establish_connt(ufs);
+ exynos_ufs_fit_aggr_timeout(ufs);
+
+ hci_writel(ufs, 0xa, HCI_DATA_REORDER);
+ hci_writel(ufs, PRDT_SET_SIZE(12), HCI_TXPRDT_ENTRY_SIZE);
+ hci_writel(ufs, PRDT_SET_SIZE(12), HCI_RXPRDT_ENTRY_SIZE);
+ hci_writel(ufs, (1 << hba->nutrs) - 1, HCI_UTRL_NEXUS_TYPE);
+ hci_writel(ufs, (1 << hba->nutmrs) - 1, HCI_UTMRL_NEXUS_TYPE);
+ hci_writel(ufs, 0xf, HCI_AXIDMA_RWDATA_BURST_LEN);
+
+ if (ufs->opts & EXYNOS_UFS_OPT_SKIP_CONNECTION_ESTAB)
+ ufshcd_dme_set(hba,
+ UIC_ARG_MIB(T_DBG_SKIP_INIT_HIBERN8_EXIT), TRUE);
+
+ if (attr->pa_granularity) {
+ exynos_ufs_enable_dbg_mode(hba);
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_GRANULARITY),
+ attr->pa_granularity);
+ exynos_ufs_disable_dbg_mode(hba);
+
+ if (attr->pa_tactivate)
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_TACTIVATE),
+ attr->pa_tactivate);
+ if (attr->pa_hibern8time &&
+ !(ufs->opts & EXYNOS_UFS_OPT_USE_SW_HIBERN8_TIMER))
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_HIBERN8TIME),
+ attr->pa_hibern8time);
+ }
+
+ if (ufs->opts & EXYNOS_UFS_OPT_USE_SW_HIBERN8_TIMER) {
+ if (!attr->pa_granularity)
+ ufshcd_dme_get(hba, UIC_ARG_MIB(PA_GRANULARITY),
+ &attr->pa_granularity);
+ if (!attr->pa_hibern8time)
+ ufshcd_dme_get(hba, UIC_ARG_MIB(PA_HIBERN8TIME),
+ &attr->pa_hibern8time);
+ /*
+ * not wait for HIBERN8 time to exit hibernation
+ */
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_HIBERN8TIME), 0);
+
+ if (attr->pa_granularity < 1 || attr->pa_granularity > 6) {
+ /* Valid range for granularity: 1 ~ 6 */
+ dev_warn(hba->dev,
+ "%s: pa_granularty %d is invalid, assuming backwards compatibility\n",
+ __func__,
+ attr->pa_granularity);
+ attr->pa_granularity = 6;
+ }
+ }
+
+ phy_calibrate(generic_phy);
+
+ if (ufs->drv_data->post_link)
+ ufs->drv_data->post_link(ufs);
+
+ return 0;
+}
+
+static int exynos_ufs_parse_dt(struct device *dev, struct exynos_ufs *ufs)
+{
+ struct device_node *np = dev->of_node;
+ struct exynos_ufs_drv_data *drv_data = &exynos_ufs_drvs;
+ struct exynos_ufs_uic_attr *attr;
+ u32 freq[2];
+ int ret;
+
+ while (drv_data->compatible) {
+ if (of_device_is_compatible(np, drv_data->compatible)) {
+ ufs->drv_data = drv_data;
+ break;
+ }
+ drv_data++;
+ }
+
+ if (ufs->drv_data && ufs->drv_data->uic_attr) {
+ attr = ufs->drv_data->uic_attr;
+ } else {
+ dev_err(dev, "failed to get uic attributes\n");
+ ret = -EINVAL;
+ goto out;
+ }
+
+ ret = of_property_read_u32_array(np,
+ "pclk-freq-avail-range", freq, ARRAY_SIZE(freq));
+ if (!ret) {
+ ufs->pclk_avail_min = freq[0];
+ ufs->pclk_avail_max = freq[1];
+ } else {
+ dev_err(dev, "failed to get available pclk range\n");
+ goto out;
+ }
+
+ attr->rx_adv_fine_gran_sup_en = RX_ADV_FINE_GRAN_SUP_EN;
+ attr->rx_adv_fine_gran_step = RX_ADV_FINE_GRAN_STEP_VAL;
+ attr->rx_adv_min_actv_time_cap = RX_ADV_MIN_ACTV_TIME_CAP;
+ attr->pa_granularity = PA_GRANULARITY_VAL;
+ attr->pa_tactivate = PA_TACTIVATE_VAL;
+ attr->pa_hibern8time = PA_HIBERN8TIME_VAL;
+
+out:
+ return ret;
+}
+
+static int exynos_ufs_init(struct ufs_hba *hba)
+{
+ struct device *dev = hba->dev;
+ struct platform_device *pdev = to_platform_device(dev);
+ struct exynos_ufs *ufs;
+ struct resource *res;
+ int ret;
+
+ ufs = devm_kzalloc(dev, sizeof(*ufs), GFP_KERNEL);
+ if (!ufs)
+ return -ENOMEM;
+
+ /* exynos-specific hci */
+ res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "vs_hci");
+ ufs->reg_hci = devm_ioremap_resource(dev, res);
+ if (!ufs->reg_hci) {
+ dev_err(dev, "cannot ioremap for hci vendor register\n");
+ return -ENOMEM;
+ }
+
+ /* unipro */
+ res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "unipro");
+ ufs->reg_unipro = devm_ioremap_resource(dev, res);
+ if (!ufs->reg_unipro) {
+ dev_err(dev, "cannot ioremap for unipro register\n");
+ return -ENOMEM;
+ }
+
+ /* ufs protector */
+ res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "ufsp");
+ ufs->reg_ufsp = devm_ioremap_resource(dev, res);
+ if (!ufs->reg_ufsp) {
+ dev_err(dev, "cannot ioremap for ufs protector register\n");
+ return -ENOMEM;
+ }
+
+ ret = exynos_ufs_parse_dt(dev, ufs);
+ if (ret) {
+ dev_err(dev, "failed to get dt info.\n");
+ goto out;
+ }
+
+ ufs->phy = devm_phy_get(dev, "ufs-phy");
+ if (IS_ERR(ufs->phy)) {
+ ret = PTR_ERR(ufs->phy);
+ dev_err(dev, "failed to get ufs-phy\n");
+ goto out;
+ }
+
+ ret = phy_power_on(ufs->phy);
+ if (ret)
+ goto phy_exit;
+
+ ufs->hba = hba;
+ ufs->opts = ufs->drv_data->opts |
+ EXYNOS_UFS_OPT_SKIP_CONNECTION_ESTAB |
+ EXYNOS_UFS_OPT_USE_SW_HIBERN8_TIMER;
+ ufs->rx_sel_idx = PA_MAXDATALANES;
+ if (ufs->opts & EXYNOS_UFS_OPT_BROKEN_RX_SEL_IDX)
+ ufs->rx_sel_idx = 0;
+ hba->priv = (void *)ufs;
+ hba->quirks = ufs->drv_data->quirks;
+ if (ufs->drv_data->drv_init) {
+ ret = ufs->drv_data->drv_init(dev, ufs);
+ if (ret) {
+ dev_err(dev, "failed to init drv-data\n");
+ goto phy_off;
+ }
+ }
+
+ ret = exynos_ufs_get_clk_info(ufs);
+ if (ret)
+ goto phy_off;
+ exynos_ufs_specify_phy_time_attr(ufs);
+ exynos_ufs_config_smu(ufs);
+ return 0;
+
+phy_off:
+ phy_power_off(ufs->phy);
+phy_exit:
+ phy_exit(ufs->phy);
+ hba->priv = NULL;
+out:
+ return ret;
+}
+
+static int exynos_ufs_host_reset(struct ufs_hba *hba)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+ unsigned long timeout = jiffies + msecs_to_jiffies(1);
+ u32 val;
+ int ret = 0;
+
+ exynos_ufs_disable_auto_ctrl_hcc_save(ufs, &val);
+
+ hci_writel(ufs, UFS_SW_RST_MASK, HCI_SW_RST);
+
+ do {
+ if (!(hci_readl(ufs, HCI_SW_RST) & UFS_SW_RST_MASK))
+ goto out;
+ } while (time_before(jiffies, timeout));
+
+ dev_err(hba->dev, "timeout host sw-reset\n");
+ ret = -ETIMEDOUT;
+
+out:
+ exynos_ufs_auto_ctrl_hcc_restore(ufs, &val);
+ return ret;
+}
+
+static void exynos_ufs_dev_hw_reset(struct ufs_hba *hba)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+
+ hci_writel(ufs, 0 << 0, HCI_GPIO_OUT);
+ udelay(5);
+ hci_writel(ufs, 1 << 0, HCI_GPIO_OUT);
+}
+
+static void exynos_ufs_pre_hibern8(struct ufs_hba *hba, u8 enter)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+ struct exynos_ufs_uic_attr *attr = ufs->drv_data->uic_attr;
+
+ if (!enter) {
+ if (ufs->opts & EXYNOS_UFS_OPT_BROKEN_AUTO_CLK_CTRL)
+ exynos_ufs_disable_auto_ctrl_hcc(ufs);
+ exynos_ufs_ungate_clks(ufs);
+
+ if (ufs->opts & EXYNOS_UFS_OPT_USE_SW_HIBERN8_TIMER) {
+ const unsigned int granularity_tbl[] = {
+ 1, 4, 8, 16, 32, 100
+ };
+ int h8_time = attr->pa_hibern8time *
+ granularity_tbl[attr->pa_granularity - 1];
+ unsigned long us;
+ s64 delta;
+
+ do {
+ delta = h8_time - ktime_us_delta(ktime_get(),
+ ufs->entry_hibern8_t);
+ if (delta <= 0)
+ break;
+
+ us = min_t(s64, delta, USEC_PER_MSEC);
+ if (us >= 10)
+ usleep_range(us, us + 10);
+ } while (1);
+ }
+ }
+}
+
+static void exynos_ufs_post_hibern8(struct ufs_hba *hba, u8 enter)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+
+ if (!enter) {
+ u32 cur_mode = 0;
+ u32 pwrmode;
+
+ if (ufshcd_is_hs_mode(&ufs->dev_req_params))
+ pwrmode = FAST_MODE;
+ else
+ pwrmode = SLOW_MODE;
+
+ ufshcd_dme_get(hba, UIC_ARG_MIB(PA_PWRMODE), &cur_mode);
+ if (cur_mode != (pwrmode << 4 | pwrmode)) {
+ dev_warn(hba->dev, "%s: power mode change\n", __func__);
+ hba->pwr_info.pwr_rx = (cur_mode >> 4) & 0xf;
+ hba->pwr_info.pwr_tx = cur_mode & 0xf;
+ ufshcd_config_pwr_mode(hba, &hba->max_pwr_info.info);
+ }
+
+ if (!(ufs->opts & EXYNOS_UFS_OPT_SKIP_CONNECTION_ESTAB))
+ exynos_ufs_establish_connt(ufs);
+ } else {
+ ufs->entry_hibern8_t = ktime_get();
+ exynos_ufs_gate_clks(ufs);
+ if (ufs->opts & EXYNOS_UFS_OPT_BROKEN_AUTO_CLK_CTRL)
+ exynos_ufs_enable_auto_ctrl_hcc(ufs);
+ }
+}
+
+static int exynos_ufs_hce_enable_notify(struct ufs_hba *hba,
+ enum ufs_notify_change_status status)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+ int ret = 0;
+
+ switch (status) {
+ case PRE_CHANGE:
+ ret = exynos_ufs_host_reset(hba);
+ if (ret)
+ return ret;
+ exynos_ufs_dev_hw_reset(hba);
+ break;
+ case POST_CHANGE:
+ exynos_ufs_calc_pwm_clk_div(ufs);
+ if (!(ufs->opts & EXYNOS_UFS_OPT_BROKEN_AUTO_CLK_CTRL))
+ exynos_ufs_enable_auto_ctrl_hcc(ufs);
+ break;
+ }
+
+ return ret;
+}
+
+static int exynos_ufs_link_startup_notify(struct ufs_hba *hba,
+ enum ufs_notify_change_status status)
+{
+ int ret = 0;
+
+ switch (status) {
+ case PRE_CHANGE:
+ ret = exynos_ufs_pre_link(hba);
+ break;
+ case POST_CHANGE:
+ ret = exynos_ufs_post_link(hba);
+ break;
+ }
+
+ return ret;
+}
+
+static int exynos_ufs_pwr_change_notify(struct ufs_hba *hba,
+ enum ufs_notify_change_status status,
+ struct ufs_pa_layer_attr *dev_max_params,
+ struct ufs_pa_layer_attr *dev_req_params)
+{
+ int ret = 0;
+
+ switch (status) {
+ case PRE_CHANGE:
+ ret = exynos_ufs_pre_pwr_mode(hba, dev_max_params,
+ dev_req_params);
+ break;
+ case POST_CHANGE:
+ ret = exynos_ufs_post_pwr_mode(hba, NULL, dev_req_params);
+ break;
+ }
+
+ return ret;
+}
+
+static void exynos_ufs_hibern8_notify(struct ufs_hba *hba,
+ enum uic_cmd_dme enter,
+ enum ufs_notify_change_status notify)
+{
+ switch ((u8)notify) {
+ case PRE_CHANGE:
+ exynos_ufs_pre_hibern8(hba, enter);
+ break;
+ case POST_CHANGE:
+ exynos_ufs_post_hibern8(hba, enter);
+ break;
+ }
+}
+
+static int exynos_ufs_suspend(struct ufs_hba *hba, enum ufs_pm_op pm_op)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+
+ if (!ufshcd_is_link_active(hba))
+ phy_power_off(ufs->phy);
+
+ return 0;
+}
+
+static int exynos_ufs_resume(struct ufs_hba *hba, enum ufs_pm_op pm_op)
+{
+ struct exynos_ufs *ufs = ufshcd_get_variant(hba);
+
+ if (!ufshcd_is_link_active(hba))
+ phy_power_on(ufs->phy);
+
+ exynos_ufs_config_smu(ufs);
+
+ return 0;
+}
+
+static struct ufs_hba_variant_ops ufs_hba_exynos_ops = {
+ .name = "exynos_ufs",
+ .init = exynos_ufs_init,
+ .hce_enable_notify = exynos_ufs_hce_enable_notify,
+ .link_startup_notify = exynos_ufs_link_startup_notify,
+ .pwr_change_notify = exynos_ufs_pwr_change_notify,
+ .setup_xfer_req = exynos_ufs_specify_nexus_t_xfer_req,
+ .setup_task_mgmt = exynos_ufs_specify_nexus_t_tm_req,
+ .hibern8_notify = exynos_ufs_hibern8_notify,
+ .suspend = exynos_ufs_suspend,
+ .resume = exynos_ufs_resume,
+};
+
+static int exynos_ufs_probe(struct platform_device *pdev)
+{
+ int err;
+ struct device *dev = &pdev->dev;
+
+ err = ufshcd_pltfrm_init(pdev, &ufs_hba_exynos_ops);
+ if (err)
+ dev_err(dev, "ufshcd_pltfrm_init() failed %d\n", err);
+
+ return err;
+}
+
+static int exynos_ufs_remove(struct platform_device *pdev)
+{
+ struct ufs_hba *hba = platform_get_drvdata(pdev);
+
+ pm_runtime_get_sync(&(pdev)->dev);
+ ufshcd_remove(hba);
+ return 0;
+}
+
+struct exynos_ufs_drv_data exynos_ufs_drvs = {
+
+ .compatible = "samsung,exynos7-ufs",
+ .uic_attr = &exynos7_uic_attr,
+ .quirks = UFSHCD_QUIRK_PRDT_BYTE_GRAN |
+ UFSHCI_QUIRK_BROKEN_REQ_LIST_CLR |
+ UFSHCI_QUIRK_BROKEN_HCE |
+ UFSHCI_QUIRK_SKIP_RESET_INTR_AGGR |
+ UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR,
+ .opts = EXYNOS_UFS_OPT_HAS_APB_CLK_CTRL |
+ EXYNOS_UFS_OPT_BROKEN_AUTO_CLK_CTRL |
+ EXYNOS_UFS_OPT_BROKEN_RX_SEL_IDX,
+ .drv_init = exynos7_ufs_drv_init,
+ .pre_link = exynos7_ufs_pre_link,
+ .post_link = exynos7_ufs_post_link,
+ .pre_pwr_change = exynos7_ufs_pre_pwr_change,
+ .post_pwr_change = exynos7_ufs_post_pwr_change,
+};
+
+static const struct of_device_id exynos_ufs_of_match[] = {
+ { .compatible = "samsung,exynos7-ufs",
+ .data = &exynos_ufs_drvs },
+ {},
+};
+
+static const struct dev_pm_ops exynos_ufs_pm_ops = {
+ .suspend = ufshcd_pltfrm_suspend,
+ .resume = ufshcd_pltfrm_resume,
+ .runtime_suspend = ufshcd_pltfrm_runtime_suspend,
+ .runtime_resume = ufshcd_pltfrm_runtime_resume,
+ .runtime_idle = ufshcd_pltfrm_runtime_idle,
+};
+
+static struct platform_driver exynos_ufs_pltform = {
+ .probe = exynos_ufs_probe,
+ .remove = exynos_ufs_remove,
+ .shutdown = ufshcd_pltfrm_shutdown,
+ .driver = {
+ .name = "exynos-ufshc",
+ .pm = &exynos_ufs_pm_ops,
+ .of_match_table = of_match_ptr(exynos_ufs_of_match),
+ },
+};
+module_platform_driver(exynos_ufs_pltform);
diff --git a/drivers/scsi/ufs/ufs-exynos.h b/drivers/scsi/ufs/ufs-exynos.h
new file mode 100644
index 000000000000..813b286afd9d
--- /dev/null
+++ b/drivers/scsi/ufs/ufs-exynos.h
@@ -0,0 +1,284 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * UFS Host Controller driver for Exynos specific extensions
+ *
+ * Copyright (C) 2014-2015 Samsung Electronics Co., Ltd.
+ *
+ */
+
+#ifndef _UFS_EXYNOS_H_
+#define _UFS_EXYNOS_H_
+
+/*
+ * UNIPRO registers
+ */
+#define UNIPRO_DBG_FORCE_DME_CTRL_STATE 0x150
+
+/*
+ * MIBs for PA debug registers
+ */
+#define PA_DBG_CLK_PERIOD 0x9514
+#define PA_DBG_TXPHY_CFGUPDT 0x9518
+#define PA_DBG_RXPHY_CFGUPDT 0x9519
+#define PA_DBG_MODE 0x9529
+#define PA_DBG_SKIP_RESET_PHY 0x9539
+#define PA_DBG_OV_TM 0x9540
+#define PA_DBG_SKIP_LINE_RESET 0x9541
+#define PA_DBG_LINE_RESET_REQ 0x9543
+#define PA_DBG_OPTION_SUITE 0x9564
+#define PA_DBG_OPTION_SUITE_DYN 0x9565
+
+/*
+ * MIBs for Transport Layer debug registers
+ */
+#define T_DBG_SKIP_INIT_HIBERN8_EXIT 0xc001
+
+/*
+ * Exynos MPHY attributes
+ */
+#define TX_LINERESET_N_VAL 0x0277
+#define TX_LINERESET_N(v) (((v) >> 10) & 0xFF)
+#define TX_LINERESET_P_VAL 0x027D
+#define TX_LINERESET_P(v) (((v) >> 12) & 0xFF)
+#define TX_OV_SLEEP_CNT_TIMER 0x028E
+#define TX_OV_H8_ENTER_EN (1 << 7)
+#define TX_OV_SLEEP_CNT(v) (((v) >> 5) & 0x7F)
+#define TX_HIGH_Z_CNT_11_08 0x028C
+#define TX_HIGH_Z_CNT_H(v) (((v) >> 8) & 0xF)
+#define TX_HIGH_Z_CNT_07_00 0x028D
+#define TX_HIGH_Z_CNT_L(v) ((v) & 0xFF)
+#define TX_BASE_NVAL_07_00 0x0293
+#define TX_BASE_NVAL_L(v) ((v) & 0xFF)
+#define TX_BASE_NVAL_15_08 0x0294
+#define TX_BASE_NVAL_H(v) (((v) >> 8) & 0xFF)
+#define TX_GRAN_NVAL_07_00 0x0295
+#define TX_GRAN_NVAL_L(v) ((v) & 0xFF)
+#define TX_GRAN_NVAL_10_08 0x0296
+#define TX_GRAN_NVAL_H(v) (((v) >> 8) & 0x3)
+
+#define RX_FILLER_ENABLE 0x0316
+#define RX_FILLER_EN (1 << 1)
+#define RX_LINERESET_VAL 0x0317
+#define RX_LINERESET(v) (((v) >> 12) & 0xFF)
+#define RX_LCC_IGNORE 0x0318
+#define RX_SYNC_MASK_LENGTH 0x0321
+#define RX_HIBERN8_WAIT_VAL_BIT_20_16 0x0331
+#define RX_HIBERN8_WAIT_VAL_BIT_15_08 0x0332
+#define RX_HIBERN8_WAIT_VAL_BIT_07_00 0x0333
+#define RX_OV_SLEEP_CNT_TIMER 0x0340
+#define RX_OV_SLEEP_CNT(v) (((v) >> 6) & 0x1F)
+#define RX_OV_STALL_CNT_TIMER 0x0341
+#define RX_OV_STALL_CNT(v) (((v) >> 4) & 0xFF)
+#define RX_BASE_NVAL_07_00 0x0355
+#define RX_BASE_NVAL_L(v) ((v) & 0xFF)
+#define RX_BASE_NVAL_15_08 0x0354
+#define RX_BASE_NVAL_H(v) (((v) >> 8) & 0xFF)
+#define RX_GRAN_NVAL_07_00 0x0353
+#define RX_GRAN_NVAL_L(v) ((v) & 0xFF)
+#define RX_GRAN_NVAL_10_08 0x0352
+#define RX_GRAN_NVAL_H(v) (((v) >> 8) & 0x3)
+
+#define CMN_PWM_CLK_CTRL 0x0402
+#define PWM_CLK_CTRL_MASK 0x3
+
+#define IATOVAL_NSEC 20000 /* unit: ns */
+#define UNIPRO_PCLK_PERIOD(ufs) (NSEC_PER_SEC / ufs->pclk_rate)
+
+struct exynos_ufs;
+
+/* vendor specific pre-defined parameters */
+#define SLOW 1
+#define FAST 2
+
+#define UFS_EXYNOS_LIMIT_NUM_LANES_RX 2
+#define UFS_EXYNOS_LIMIT_NUM_LANES_TX 2
+#define UFS_EXYNOS_LIMIT_HSGEAR_RX UFS_HS_G3
+#define UFS_EXYNOS_LIMIT_HSGEAR_TX UFS_HS_G3
+#define UFS_EXYNOS_LIMIT_PWMGEAR_RX UFS_PWM_G4
+#define UFS_EXYNOS_LIMIT_PWMGEAR_TX UFS_PWM_G4
+#define UFS_EXYNOS_LIMIT_RX_PWR_PWM SLOW_MODE
+#define UFS_EXYNOS_LIMIT_TX_PWR_PWM SLOW_MODE
+#define UFS_EXYNOS_LIMIT_RX_PWR_HS FAST_MODE
+#define UFS_EXYNOS_LIMIT_TX_PWR_HS FAST_MODE
+#define UFS_EXYNOS_LIMIT_HS_RATE PA_HS_MODE_B
+#define UFS_EXYNOS_LIMIT_DESIRED_MODE FAST
+
+#define RX_ADV_FINE_GRAN_SUP_EN 0x1
+#define RX_ADV_FINE_GRAN_STEP_VAL 0x3
+#define RX_ADV_MIN_ACTV_TIME_CAP 0x9
+
+#define PA_GRANULARITY_VAL 0x6
+#define PA_TACTIVATE_VAL 0x3
+#define PA_HIBERN8TIME_VAL 0x20
+
+struct exynos_ufs_uic_attr {
+ /* TX Attributes */
+ unsigned int tx_trailingclks;
+ unsigned int tx_dif_p_nsec;
+ unsigned int tx_dif_n_nsec;
+ unsigned int tx_high_z_cnt_nsec;
+ unsigned int tx_base_unit_nsec;
+ unsigned int tx_gran_unit_nsec;
+ unsigned int tx_sleep_cnt;
+ unsigned int tx_min_activatetime;
+ /* RX Attributes */
+ unsigned int rx_filler_enable;
+ unsigned int rx_dif_p_nsec;
+ unsigned int rx_hibern8_wait_nsec;
+ unsigned int rx_base_unit_nsec;
+ unsigned int rx_gran_unit_nsec;
+ unsigned int rx_sleep_cnt;
+ unsigned int rx_stall_cnt;
+ unsigned int rx_hs_g1_sync_len_cap;
+ unsigned int rx_hs_g2_sync_len_cap;
+ unsigned int rx_hs_g3_sync_len_cap;
+ unsigned int rx_hs_g1_prep_sync_len_cap;
+ unsigned int rx_hs_g2_prep_sync_len_cap;
+ unsigned int rx_hs_g3_prep_sync_len_cap;
+ /* Common Attributes */
+ unsigned int cmn_pwm_clk_ctrl;
+ /* Internal Attributes */
+ unsigned int pa_dbg_option_suite;
+ /* Changeable Attributes */
+ unsigned int rx_adv_fine_gran_sup_en;
+ unsigned int rx_adv_fine_gran_step;
+ unsigned int rx_min_actv_time_cap;
+ unsigned int rx_hibern8_time_cap;
+ unsigned int rx_adv_min_actv_time_cap;
+ unsigned int rx_adv_hibern8_time_cap;
+ unsigned int pa_granularity;
+ unsigned int pa_tactivate;
+ unsigned int pa_hibern8time;
+};
+
+struct exynos_ufs_drv_data {
+ char *compatible;
+ struct exynos_ufs_uic_attr *uic_attr;
+ unsigned int quirks;
+ unsigned int opts;
+ /* SoC's specific operations */
+ int (*drv_init)(struct device *dev, struct exynos_ufs *ufs);
+ int (*pre_link)(struct exynos_ufs *ufs);
+ int (*post_link)(struct exynos_ufs *ufs);
+ int (*pre_pwr_change)(struct exynos_ufs *ufs,
+ struct ufs_pa_layer_attr *pwr);
+ int (*post_pwr_change)(struct exynos_ufs *ufs,
+ struct ufs_pa_layer_attr *pwr);
+};
+
+struct ufs_phy_time_cfg {
+ u32 tx_linereset_p;
+ u32 tx_linereset_n;
+ u32 tx_high_z_cnt;
+ u32 tx_base_n_val;
+ u32 tx_gran_n_val;
+ u32 tx_sleep_cnt;
+ u32 rx_linereset;
+ u32 rx_hibern8_wait;
+ u32 rx_base_n_val;
+ u32 rx_gran_n_val;
+ u32 rx_sleep_cnt;
+ u32 rx_stall_cnt;
+};
+
+struct exynos_ufs {
+ struct ufs_hba *hba;
+ struct phy *phy;
+ void __iomem *reg_hci;
+ void __iomem *reg_unipro;
+ void __iomem *reg_ufsp;
+ struct clk *clk_hci_core;
+ struct clk *clk_unipro_main;
+ struct clk *clk_apb;
+ u32 pclk_rate;
+ u32 pclk_div;
+ u32 pclk_avail_min;
+ u32 pclk_avail_max;
+ u32 mclk_rate;
+ int avail_ln_rx;
+ int avail_ln_tx;
+ int rx_sel_idx;
+ struct ufs_pa_layer_attr dev_req_params;
+ struct ufs_phy_time_cfg t_cfg;
+ ktime_t entry_hibern8_t;
+ struct exynos_ufs_drv_data *drv_data;
+
+ u32 opts;
+#define EXYNOS_UFS_OPT_HAS_APB_CLK_CTRL BIT(0)
+#define EXYNOS_UFS_OPT_SKIP_CONNECTION_ESTAB BIT(1)
+#define EXYNOS_UFS_OPT_BROKEN_AUTO_CLK_CTRL BIT(2)
+#define EXYNOS_UFS_OPT_BROKEN_RX_SEL_IDX BIT(3)
+#define EXYNOS_UFS_OPT_USE_SW_HIBERN8_TIMER BIT(4)
+};
+
+#define for_each_ufs_rx_lane(ufs, i) \
+ for (i = (ufs)->rx_sel_idx; \
+ i < (ufs)->rx_sel_idx + (ufs)->avail_ln_rx; i++)
+#define for_each_ufs_tx_lane(ufs, i) \
+ for (i = 0; i < (ufs)->avail_ln_tx; i++)
+
+#define EXYNOS_UFS_MMIO_FUNC(name) \
+static inline void name##_writel(struct exynos_ufs *ufs, u32 val, u32 reg)\
+{ \
+ writel(val, ufs->reg_##name + reg); \
+} \
+ \
+static inline u32 name##_readl(struct exynos_ufs *ufs, u32 reg) \
+{ \
+ return readl(ufs->reg_##name + reg); \
+}
+
+EXYNOS_UFS_MMIO_FUNC(hci);
+EXYNOS_UFS_MMIO_FUNC(unipro);
+EXYNOS_UFS_MMIO_FUNC(ufsp);
+#undef EXYNOS_UFS_MMIO_FUNC
+
+long exynos_ufs_calc_time_cntr(struct exynos_ufs *, long);
+
+static inline void exynos_ufs_enable_ov_tm(struct ufs_hba *hba)
+{
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_OV_TM), TRUE);
+}
+
+static inline void exynos_ufs_disable_ov_tm(struct ufs_hba *hba)
+{
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_OV_TM), FALSE);
+}
+
+static inline void exynos_ufs_enable_dbg_mode(struct ufs_hba *hba)
+{
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_MODE), TRUE);
+}
+
+static inline void exynos_ufs_disable_dbg_mode(struct ufs_hba *hba)
+{
+ ufshcd_dme_set(hba, UIC_ARG_MIB(PA_DBG_MODE), FALSE);
+}
+
+struct exynos_ufs_drv_data exynos_ufs_drvs;
+
+struct exynos_ufs_uic_attr exynos7_uic_attr = {
+ .tx_trailingclks = 0x10,
+ .tx_dif_p_nsec = 3000000, /* unit: ns */
+ .tx_dif_n_nsec = 1000000, /* unit: ns */
+ .tx_high_z_cnt_nsec = 20000, /* unit: ns */
+ .tx_base_unit_nsec = 100000, /* unit: ns */
+ .tx_gran_unit_nsec = 4000, /* unit: ns */
+ .tx_sleep_cnt = 1000, /* unit: ns */
+ .tx_min_activatetime = 0xa,
+ .rx_filler_enable = 0x2,
+ .rx_dif_p_nsec = 1000000, /* unit: ns */
+ .rx_hibern8_wait_nsec = 4000000, /* unit: ns */
+ .rx_base_unit_nsec = 100000, /* unit: ns */
+ .rx_gran_unit_nsec = 4000, /* unit: ns */
+ .rx_sleep_cnt = 1280, /* unit: ns */
+ .rx_stall_cnt = 320, /* unit: ns */
+ .rx_hs_g1_sync_len_cap = SYNC_LEN_COARSE(0xf),
+ .rx_hs_g2_sync_len_cap = SYNC_LEN_COARSE(0xf),
+ .rx_hs_g3_sync_len_cap = SYNC_LEN_COARSE(0xf),
+ .rx_hs_g1_prep_sync_len_cap = PREP_LEN(0xf),
+ .rx_hs_g2_prep_sync_len_cap = PREP_LEN(0xf),
+ .rx_hs_g3_prep_sync_len_cap = PREP_LEN(0xf),
+ .pa_dbg_option_suite = 0x30103,
+};
+#endif /* _UFS_EXYNOS_H_ */
diff --git a/drivers/scsi/ufs/unipro.h b/drivers/scsi/ufs/unipro.h
index 766d551df3fc..4ee64782fd48 100644
--- a/drivers/scsi/ufs/unipro.h
+++ b/drivers/scsi/ufs/unipro.h
@@ -64,8 +64,25 @@
#define CFGRXOVR4 0x00E9
#define RXSQCTRL 0x00B5
#define CFGRXOVR6 0x00BF
+#define RX_HS_G1_SYNC_LENGTH_CAP 0x008B
+#define RX_HS_G1_PREP_LENGTH_CAP 0x008C
+#define RX_HS_G2_SYNC_LENGTH_CAP 0x0094
+#define RX_HS_G3_SYNC_LENGTH_CAP 0x0095
+#define RX_HS_G2_PREP_LENGTH_CAP 0x0096
+#define RX_HS_G3_PREP_LENGTH_CAP 0x0097
+#define RX_ADV_GRANULARITY_CAP 0x0098
+#define RX_MIN_ACTIVATETIME_CAP 0x008F
+#define RX_HIBERN8TIME_CAP 0x0092
+#define RX_ADV_HIBERN8TIME_CAP 0x0099
+#define RX_ADV_MIN_ACTIVATETIME_CAP 0x009A
+

#define is_mphy_tx_attr(attr) (attr < RX_MODE)
+#define RX_ADV_FINE_GRAN_STEP(x) ((((x) & 0x3) << 1) | 0x1)
+#define SYNC_LEN_FINE(x) ((x) & 0x3F)
+#define SYNC_LEN_COARSE(x) ((1 << 6) | ((x) & 0x3F))
+#define PREP_LEN(x) ((x) & 0xF)
+
#define RX_MIN_ACTIVATETIME_UNIT_US 100
#define HIBERN8TIME_UNIT_US 100

@@ -124,6 +141,7 @@
#define PA_PACPREQEOBTIMEOUT 0x1591
#define PA_HIBERN8TIME 0x15A7
#define PA_LOCALVERINFO 0x15A9
+#define PA_GRANULARITY 0x15AA
#define PA_TACTIVATE 0x15A8
#define PA_PACPFRAMECOUNT 0x15C0
#define PA_PACPERRORCOUNT 0x15C1
@@ -291,4 +309,19 @@ enum {
TRUE,
};

+/* CPort setting */
+#define E2EFC_ON (1 << 0)
+#define E2EFC_OFF (0 << 0)
+#define CSD_N_ON (0 << 1)
+#define CSD_N_OFF (1 << 1)
+#define CSV_N_ON (0 << 2)
+#define CSV_N_OFF (1 << 2)
+#define CPORT_DEF_FLAGS (CSV_N_OFF | CSD_N_OFF | E2EFC_OFF)
+
+/* CPort connection state */
+enum {
+ CPORT_IDLE = 0,
+ CPORT_CONNECTED,
+};
+
#endif /* _UNIPRO_H_ */
--
2.17.1

2020-04-17 18:13:00

by Alim Akhtar

[permalink] [raw]
Subject: [PATCH v6 03/10] scsi: ufs: add quirk to enable host controller without hce

Some host controllers don't support host controller enable via HCE.

Signed-off-by: Seungwon Jeon <[email protected]>
Signed-off-by: Alim Akhtar <[email protected]>
---
drivers/scsi/ufs/ufshcd.c | 76 +++++++++++++++++++++++++++++++++++++--
drivers/scsi/ufs/ufshcd.h | 6 ++++
2 files changed, 80 insertions(+), 2 deletions(-)

diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
index 0e9704da58bd..ee30ed6cc805 100644
--- a/drivers/scsi/ufs/ufshcd.c
+++ b/drivers/scsi/ufs/ufshcd.c
@@ -3534,6 +3534,52 @@ static int ufshcd_dme_link_startup(struct ufs_hba *hba)
"dme-link-startup: error code %d\n", ret);
return ret;
}
+/**
+ * ufshcd_dme_reset - UIC command for DME_RESET
+ * @hba: per adapter instance
+ *
+ * DME_RESET command is issued in order to reset UniPro stack.
+ * This function now deal with cold reset.
+ *
+ * Returns 0 on success, non-zero value on failure
+ */
+static int ufshcd_dme_reset(struct ufs_hba *hba)
+{
+ struct uic_command uic_cmd = {0};
+ int ret;
+
+ uic_cmd.command = UIC_CMD_DME_RESET;
+
+ ret = ufshcd_send_uic_cmd(hba, &uic_cmd);
+ if (ret)
+ dev_err(hba->dev,
+ "dme-reset: error code %d\n", ret);
+
+ return ret;
+}
+
+/**
+ * ufshcd_dme_enable - UIC command for DME_ENABLE
+ * @hba: per adapter instance
+ *
+ * DME_ENABLE command is issued in order to enable UniPro stack.
+ *
+ * Returns 0 on success, non-zero value on failure
+ */
+static int ufshcd_dme_enable(struct ufs_hba *hba)
+{
+ struct uic_command uic_cmd = {0};
+ int ret;
+
+ uic_cmd.command = UIC_CMD_DME_ENABLE;
+
+ ret = ufshcd_send_uic_cmd(hba, &uic_cmd);
+ if (ret)
+ dev_err(hba->dev,
+ "dme-reset: error code %d\n", ret);
+
+ return ret;
+}

static inline void ufshcd_add_delay_before_dme_cmd(struct ufs_hba *hba)
{
@@ -4251,7 +4297,7 @@ static inline void ufshcd_hba_stop(struct ufs_hba *hba, bool can_sleep)
}

/**
- * ufshcd_hba_enable - initialize the controller
+ * ufshcd_hba_execute_hce - initialize the controller
* @hba: per adapter instance
*
* The controller resets itself and controller firmware initialization
@@ -4260,7 +4306,7 @@ static inline void ufshcd_hba_stop(struct ufs_hba *hba, bool can_sleep)
*
* Returns 0 on success, non-zero value on failure
*/
-int ufshcd_hba_enable(struct ufs_hba *hba)
+static int ufshcd_hba_execute_hce(struct ufs_hba *hba)
{
int retry;

@@ -4308,6 +4354,32 @@ int ufshcd_hba_enable(struct ufs_hba *hba)

return 0;
}
+
+int ufshcd_hba_enable(struct ufs_hba *hba)
+{
+ int ret;
+
+ if (hba->quirks & UFSHCI_QUIRK_BROKEN_HCE) {
+ ufshcd_set_link_off(hba);
+ ufshcd_vops_hce_enable_notify(hba, PRE_CHANGE);
+
+ /* enable UIC related interrupts */
+ ufshcd_enable_intr(hba, UFSHCD_UIC_MASK);
+ ret = ufshcd_dme_reset(hba);
+ if (!ret) {
+ ret = ufshcd_dme_enable(hba);
+ if (!ret)
+ ufshcd_vops_hce_enable_notify(hba, POST_CHANGE);
+ if (ret)
+ dev_err(hba->dev,
+ "Host controller enable failed with non-hce\n");
+ }
+ } else {
+ ret = ufshcd_hba_execute_hce(hba);
+ }
+
+ return ret;
+}
EXPORT_SYMBOL_GPL(ufshcd_hba_enable);

static int ufshcd_disable_tx_lcc(struct ufs_hba *hba, bool peer)
diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
index 53096642f9a8..f8d08cb9caf7 100644
--- a/drivers/scsi/ufs/ufshcd.h
+++ b/drivers/scsi/ufs/ufshcd.h
@@ -529,6 +529,12 @@ enum ufshcd_quirks {
* that the interrupt aggregation timer and counter are reset by s/w.
*/
UFSHCI_QUIRK_SKIP_RESET_INTR_AGGR = 1 << 7,
+
+ /*
+ * This quirks needs to be enabled if host controller cannot be
+ * enabled via HCE register.
+ */
+ UFSHCI_QUIRK_BROKEN_HCE = 1 << 8,
};

enum ufshcd_caps {
--
2.17.1

2020-04-17 18:13:02

by Alim Akhtar

[permalink] [raw]
Subject: [PATCH v6 02/10] scsi: ufs: add quirk to disallow reset of interrupt aggregation

Some host controllers support interrupt aggregation but don't allow
resetting counter and timer in software.

Signed-off-by: Seungwon Jeon <[email protected]>
Signed-off-by: Alim Akhtar <[email protected]>
---
drivers/scsi/ufs/ufshcd.c | 3 ++-
drivers/scsi/ufs/ufshcd.h | 6 ++++++
2 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
index 3655b88fc862..0e9704da58bd 100644
--- a/drivers/scsi/ufs/ufshcd.c
+++ b/drivers/scsi/ufs/ufshcd.c
@@ -4884,7 +4884,8 @@ static irqreturn_t ufshcd_transfer_req_compl(struct ufs_hba *hba)
* false interrupt if device completes another request after resetting
* aggregation and before reading the DB.
*/
- if (ufshcd_is_intr_aggr_allowed(hba))
+ if (ufshcd_is_intr_aggr_allowed(hba) &&
+ !(hba->quirks & UFSHCI_QUIRK_SKIP_RESET_INTR_AGGR))
ufshcd_reset_intr_aggr(hba);

tr_doorbell = ufshcd_readl(hba, REG_UTP_TRANSFER_REQ_DOOR_BELL);
diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
index 071f0edf3f64..53096642f9a8 100644
--- a/drivers/scsi/ufs/ufshcd.h
+++ b/drivers/scsi/ufs/ufshcd.h
@@ -523,6 +523,12 @@ enum ufshcd_quirks {
* Clear handling for transfer/task request list is just opposite.
*/
UFSHCI_QUIRK_BROKEN_REQ_LIST_CLR = 1 << 6,
+
+ /*
+ * This quirk needs to be enabled if host controller doesn't allow
+ * that the interrupt aggregation timer and counter are reset by s/w.
+ */
+ UFSHCI_QUIRK_SKIP_RESET_INTR_AGGR = 1 << 7,
};

enum ufshcd_caps {
--
2.17.1

2020-04-17 18:14:46

by Alim Akhtar

[permalink] [raw]
Subject: [PATCH v6 08/10] dt-bindings: ufs: Add DT binding documentation for ufs

This patch adds DT binding for samsung ufs hci

Signed-off-by: Alim Akhtar <[email protected]>
---
.../bindings/ufs/samsung,exynos-ufs.yaml | 93 +++++++++++++++++++
1 file changed, 93 insertions(+)
create mode 100644 Documentation/devicetree/bindings/ufs/samsung,exynos-ufs.yaml

diff --git a/Documentation/devicetree/bindings/ufs/samsung,exynos-ufs.yaml b/Documentation/devicetree/bindings/ufs/samsung,exynos-ufs.yaml
new file mode 100644
index 000000000000..954338b7f37d
--- /dev/null
+++ b/Documentation/devicetree/bindings/ufs/samsung,exynos-ufs.yaml
@@ -0,0 +1,93 @@
+# SPDX-License-Identifier: GPL-2.0-only OR BSD-2-Clause
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/ufs/samsung,exynos-ufs.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Samsung SoC series UFS host controller Device Tree Bindings
+
+maintainers:
+ - Alim Akhtar <[email protected]>
+
+description: |
+ Each Samsung UFS host controller instance should have its own node.
+ This binding define Samsung specific binding other then what is used
+ in the common ufshcd bindings
+ [1] Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
+
+properties:
+
+ compatible:
+ enum:
+ - samsung,exynos7-ufs
+
+ reg:
+ items:
+ - description: HCI register
+ - description: vendor specific register
+ - description: unipro register
+ - description: UFS protector register
+
+ reg-names:
+ items:
+ - const: hci
+ - const: vs_hci
+ - const: unipro
+ - const: ufsp
+
+ clocks:
+ maxItems: 2
+ items:
+ - description: ufs link core clock
+ - description: unipro main clock
+
+ clock-names:
+ maxItems: 2
+ items:
+ - const: core_clk
+ - const: sclk_unipro_main
+
+ interrupts:
+ items:
+ - description: interrupt signal for various ufshc status
+
+ phys:
+ maxItems: 1
+ description:
+ phandle of the ufs phy node
+
+ phy-names:
+ const: ufs-phy
+
+required:
+ - compatible
+ - reg
+ - interrupts
+ - phys
+ - phy-names
+ - clocks
+ - clock-names
+
+examples:
+ - |
+ #include <dt-bindings/interrupt-controller/arm-gic.h>
+ #include <dt-bindings/clock/exynos7-clk.h>
+
+ ufs: ufs@15570000 {
+ compatible = "samsung,exynos7-ufs";
+ reg = <0x15570000 0x100>,
+ <0x15570100 0x100>,
+ <0x15571000 0x200>,
+ <0x15572000 0x300>;
+ reg-names = "hci", "vs_hci", "unipro", "ufsp";
+ interrupts = <GIC_SPI 200 IRQ_TYPE_LEVEL_HIGH>;
+ clocks = <&clock_fsys1 ACLK_UFS20_LINK>,
+ <&clock_fsys1 SCLK_UFSUNIPRO20_USER>;
+ clock-names = "core_clk", "sclk_unipro_main";
+ pinctrl-names = "default";
+ pinctrl-0 = <&ufs_rst_n &ufs_refclk_out>;
+ pclk-freq-avail-range = <70000000 133000000>;
+ phys = <&ufs_phy>;
+ phy-names = "ufs-phy";
+ };
+...
--
2.17.1

2020-04-17 18:15:25

by Alim Akhtar

[permalink] [raw]
Subject: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs fatal error

From: Kiwoong Kim <[email protected]>

Some architectures determines if fatal error for OCS
occurrs to check status in response upiu. This patch
is to prevent from reporting command results with that.

Signed-off-by: Kiwoong Kim <[email protected]>
Signed-off-by: Alim Akhtar <[email protected]>
---
drivers/scsi/ufs/ufshcd.c | 6 ++++++
drivers/scsi/ufs/ufshcd.h | 6 ++++++
2 files changed, 12 insertions(+)

diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
index b32fcedcdcb9..8c07caff0a5c 100644
--- a/drivers/scsi/ufs/ufshcd.c
+++ b/drivers/scsi/ufs/ufshcd.c
@@ -4794,6 +4794,12 @@ ufshcd_transfer_rsp_status(struct ufs_hba *hba, struct ufshcd_lrb *lrbp)
/* overall command status of utrd */
ocs = ufshcd_get_tr_ocs(lrbp);

+ if (hba->quirks & UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR) {
+ if (be32_to_cpu(lrbp->ucd_rsp_ptr->header.dword_1) &
+ MASK_RSP_UPIU_RESULT)
+ ocs = OCS_SUCCESS;
+ }
+
switch (ocs) {
case OCS_SUCCESS:
result = ufshcd_get_req_rsp(lrbp->ucd_rsp_ptr);
diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
index a9b9ace9fc72..e1d09c2c4302 100644
--- a/drivers/scsi/ufs/ufshcd.h
+++ b/drivers/scsi/ufs/ufshcd.h
@@ -541,6 +541,12 @@ enum ufshcd_quirks {
* resolution of the values of PRDTO and PRDTL in UTRD as byte.
*/
UFSHCD_QUIRK_PRDT_BYTE_GRAN = 1 << 9,
+
+ /*
+ * This quirk needs to be enabled if the host controller reports
+ * OCS FATAL ERROR with device error through sense data
+ */
+ UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR = 1 << 10,
};

enum ufshcd_caps {
--
2.17.1

2020-04-17 18:22:25

by Randy Dunlap

[permalink] [raw]
Subject: Re: [PATCH v6 09/10] scsi: ufs-exynos: add UFS host support for Exynos SoCs

On 4/17/20 10:59 AM, Alim Akhtar wrote:
> diff --git a/drivers/scsi/ufs/Kconfig b/drivers/scsi/ufs/Kconfig
> index e2005aeddc2d..cc7e29c8c24f 100644
> --- a/drivers/scsi/ufs/Kconfig
> +++ b/drivers/scsi/ufs/Kconfig
> @@ -160,3 +160,15 @@ config SCSI_UFS_BSG
>
> Select this if you need a bsg device node for your UFS controller.
> If unsure, say N.
> +
> +config SCSI_UFS_EXYNOS
> + bool "EXYNOS specific hooks to UFS controller platform driver"
> + depends on SCSI_UFSHCD_PLATFORM && ARCH_EXYNOS || COMPILE_TEST

"&&" has higher precedence than "||", so I'm thinking that line should be

+ depends on SCSI_UFSHCD_PLATFORM && (ARCH_EXYNOS || COMPILE_TEST)

> + select PHY_SAMSUNG_UFS
> + help
> + This selects the EXYNOS specific additions to UFSHCD platform driver.
> + UFS host on EXYNOS includes HCI and UNIPRO layer, and associates with
> + UFS-PHY driver.
> +
> + Select this if you have UFS host controller on EXYNOS chipset.
> + If unsure, say N.


--
~Randy

2020-04-18 12:41:47

by Avri Altman

[permalink] [raw]
Subject: RE: [PATCH v6 0/10] exynos-ufs: Add support for UFS HCI


>
> This patch-set introduces UFS (Universal Flash Storage) host controller
> support
> for Samsung family SoC. Mostly, it consists of UFS PHY and host specific
> driver.
>
> - Changes since v5:
> * re-introduce various quicks which was removed because of no driver
> * consumer of those quirks, initial 4 patches does the same.
You forgot to add those quirks to ufs_fixups.
Each patch that introduces a quirk needs to introduce its users as well -
This is the reason it was removed in the first place.

Thanks,
Avri

2020-04-18 13:10:31

by Alim Akhtar

[permalink] [raw]
Subject: RE: [PATCH v6 0/10] exynos-ufs: Add support for UFS HCI

Hi Avri,

> -----Original Message-----
> From: Avri Altman <[email protected]>
> Sent: 18 April 2020 18:09
> To: Alim Akhtar <[email protected]>; [email protected]
> Cc: [email protected]; [email protected]; [email protected];
> [email protected]; [email protected];
> [email protected]; [email protected]; linux-samsung-
> [email protected]; [email protected]; linux-
> [email protected]
> Subject: RE: [PATCH v6 0/10] exynos-ufs: Add support for UFS HCI
>
>
> >
> > This patch-set introduces UFS (Universal Flash Storage) host
> > controller support for Samsung family SoC. Mostly, it consists of UFS
> > PHY and host specific driver.
> >
> > - Changes since v5:
> > * re-introduce various quicks which was removed because of no driver
> > * consumer of those quirks, initial 4 patches does the same.
> You forgot to add those quirks to ufs_fixups.

ufs_fixups are for ufs __device__ related quirks, what I have posted are all host controller quirks.
Please have a look on the other quirks related to HCI like UFSHCD_QUIRK_BROKEN_UFS_HCI_VERSION
Which is used in other vendor HCI driver.
Let me know if I am missing anything here.

> Each patch that introduces a quirk needs to introduce its users as well - This is
> the reason it was removed in the first place.
>
> Thanks,
> Avri

2020-04-18 16:01:57

by Avri Altman

[permalink] [raw]
Subject: RE: [PATCH v6 0/10] exynos-ufs: Add support for UFS HCI

> > -----Original Message-----
> > From: Avri Altman <[email protected]>
> > Sent: 18 April 2020 18:09
> > To: Alim Akhtar <[email protected]>; [email protected]
> > Cc: [email protected]; [email protected];
> [email protected];
> > [email protected]; [email protected];
> > [email protected]; [email protected]; linux-samsung-
> > [email protected]; [email protected]; linux-
> > [email protected]
> > Subject: RE: [PATCH v6 0/10] exynos-ufs: Add support for UFS HCI
> >
> >
> > >
> > > This patch-set introduces UFS (Universal Flash Storage) host
> > > controller support for Samsung family SoC. Mostly, it consists of UFS
> > > PHY and host specific driver.
> > >
> > > - Changes since v5:
> > > * re-introduce various quicks which was removed because of no driver
> > > * consumer of those quirks, initial 4 patches does the same.
> > You forgot to add those quirks to ufs_fixups.
>
> ufs_fixups are for ufs __device__ related quirks, what I have posted are all
> host controller quirks.
Right.
So what I am saying is that I am missing the hba->quirks |= UFSHCI_QUIRK_<new-quirk>
In ufs-exynos.c for each one of the new quirks.

2020-04-18 16:06:18

by Avri Altman

[permalink] [raw]
Subject: RE: [PATCH v6 0/10] exynos-ufs: Add support for UFS HCI



> -----Original Message-----
> From: Avri Altman
> Sent: Saturday, April 18, 2020 7:00 PM
> To: Alim Akhtar <[email protected]>; [email protected]
> Cc: [email protected]; [email protected];
> [email protected]; [email protected]; [email protected];
> [email protected]; [email protected]; linux-samsung-
> [email protected]; [email protected]; linux-
> [email protected]
> Subject: RE: [PATCH v6 0/10] exynos-ufs: Add support for UFS HCI
>
> > > -----Original Message-----
> > > From: Avri Altman <[email protected]>
> > > Sent: 18 April 2020 18:09
> > > To: Alim Akhtar <[email protected]>; [email protected]
> > > Cc: [email protected]; [email protected];
> > [email protected];
> > > [email protected]; [email protected];
> > > [email protected]; [email protected]; linux-samsung-
> > > [email protected]; [email protected]; linux-
> > > [email protected]
> > > Subject: RE: [PATCH v6 0/10] exynos-ufs: Add support for UFS HCI
> > >
> > >
> > > >
> > > > This patch-set introduces UFS (Universal Flash Storage) host
> > > > controller support for Samsung family SoC. Mostly, it consists of UFS
> > > > PHY and host specific driver.
> > > >
> > > > - Changes since v5:
> > > > * re-introduce various quicks which was removed because of no driver
> > > > * consumer of those quirks, initial 4 patches does the same.
> > > You forgot to add those quirks to ufs_fixups.
> >
> > ufs_fixups are for ufs __device__ related quirks, what I have posted are all
> > host controller quirks.
> Right.
> So what I am saying is that I am missing the hba->quirks |=
> UFSHCI_QUIRK_<new-quirk>
> In ufs-exynos.c for each one of the new quirks.
Oh, but you add those in patch #9 -
Ok. Got it. Sorry about the confusion.

Thanks,
Avri

2020-04-20 06:56:36

by Avri Altman

[permalink] [raw]
Subject: RE: [PATCH v6 01/10] scsi: ufs: add quirk to fix mishandling utrlclr/utmrlclr

>
> In the right behavior, setting the bit to '0' indicates clear and '1'
> indicates no change. If host controller handles this the other way,
> UFSHCI_QUIRK_BROKEN_REQ_LIST_CLR can be used.
>
> Signed-off-by: Seungwon Jeon <[email protected]>
> Signed-off-by: Alim Akhtar <[email protected]>
Reviewed-by: Avri Altman <[email protected]>

2020-04-20 07:06:47

by Avri Altman

[permalink] [raw]
Subject: RE: [PATCH v6 02/10] scsi: ufs: add quirk to disallow reset of interrupt aggregation

>
> Some host controllers support interrupt aggregation but don't allow
> resetting counter and timer in software.
>
> Signed-off-by: Seungwon Jeon <[email protected]>
> Signed-off-by: Alim Akhtar <[email protected]>
Reviewed-by: Avri Altman <[email protected]>

2020-04-20 07:16:19

by Avri Altman

[permalink] [raw]
Subject: RE: [PATCH v6 03/10] scsi: ufs: add quirk to enable host controller without hce

>
> Some host controllers don't support host controller enable via HCE.
>
> Signed-off-by: Seungwon Jeon <[email protected]>
> Signed-off-by: Alim Akhtar <[email protected]>
Reviewed-by: Avri Altman <[email protected]>

2020-04-20 08:15:30

by Avri Altman

[permalink] [raw]
Subject: RE: [PATCH v6 04/10] scsi: ufs: introduce UFSHCD_QUIRK_PRDT_BYTE_GRAN quirk


>
> Some UFS host controllers may think granularities of PRDT length and
> offset as bytes, not double words.
>
> Signed-off-by: Kiwoong Kim <[email protected]>
> Signed-off-by: Alim Akhtar <[email protected]>
Reviewed-by: Avri Altman <[email protected]>

2020-04-20 08:58:29

by Avri Altman

[permalink] [raw]
Subject: RE: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs fatal error

>
> From: Kiwoong Kim <[email protected]>
>
> Some architectures determines if fatal error for OCS
> occurrs to check status in response upiu. This patch
Typo - occurs

> is to prevent from reporting command results with that.
>
> Signed-off-by: Kiwoong Kim <[email protected]>
> Signed-off-by: Alim Akhtar <[email protected]>
> ---
> drivers/scsi/ufs/ufshcd.c | 6 ++++++
> drivers/scsi/ufs/ufshcd.h | 6 ++++++
> 2 files changed, 12 insertions(+)
>
> diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> index b32fcedcdcb9..8c07caff0a5c 100644
> --- a/drivers/scsi/ufs/ufshcd.c
> +++ b/drivers/scsi/ufs/ufshcd.c
> @@ -4794,6 +4794,12 @@ ufshcd_transfer_rsp_status(struct ufs_hba *hba,
> struct ufshcd_lrb *lrbp)
> /* overall command status of utrd */
> ocs = ufshcd_get_tr_ocs(lrbp);
>
> + if (hba->quirks & UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR) {
> + if (be32_to_cpu(lrbp->ucd_rsp_ptr->header.dword_1) &
> + MASK_RSP_UPIU_RESULT)
> + ocs = OCS_SUCCESS;
> + }
> +
Not sure that I follow what this quirk is all about.
Your code overrides ocs by open coding ufshcd_get_rsp_upiu_result.

Normally OCS is in utp transfer req descriptor, dword 2, bits 0..7.
My understanding from your description, is that some fatal error might occur,
But the host controller does not report it, and it still needs to be checked in the response upiu.
Evidently you are not doing so.
Please elaborate your description.

P.S.
The ocs is being evaluated in device management commands as well,
Isn't this something you need to attend?

Thanks,
Avri

> switch (ocs) {
> case OCS_SUCCESS:
> result = ufshcd_get_req_rsp(lrbp->ucd_rsp_ptr);
> diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
> index a9b9ace9fc72..e1d09c2c4302 100644
> --- a/drivers/scsi/ufs/ufshcd.h
> +++ b/drivers/scsi/ufs/ufshcd.h
> @@ -541,6 +541,12 @@ enum ufshcd_quirks {
> * resolution of the values of PRDTO and PRDTL in UTRD as byte.
> */
> UFSHCD_QUIRK_PRDT_BYTE_GRAN = 1 << 9,
> +
> + /*
> + * This quirk needs to be enabled if the host controller reports
> + * OCS FATAL ERROR with device error through sense data
> + */
> + UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR = 1 << 10,
> };
>
> enum ufshcd_caps {
> --
> 2.17.1

2020-04-20 10:27:50

by Avri Altman

[permalink] [raw]
Subject: RE: [PATCH v6 09/10] scsi: ufs-exynos: add UFS host support for Exynos SoCs

> + struct exynos_ufs_drv_data *drv_data;
> +
> + u32 opts;
> +#define EXYNOS_UFS_OPT_HAS_APB_CLK_CTRL BIT(0)
> +#define EXYNOS_UFS_OPT_SKIP_CONNECTION_ESTAB BIT(1)
> +#define EXYNOS_UFS_OPT_BROKEN_AUTO_CLK_CTRL BIT(2)
> +#define EXYNOS_UFS_OPT_BROKEN_RX_SEL_IDX BIT(3)
> +#define EXYNOS_UFS_OPT_USE_SW_HIBERN8_TIMER BIT(4)
Could not find where the last 2 are being used.

Thanks,
Avri

2020-04-21 01:01:39

by Alim Akhtar

[permalink] [raw]
Subject: RE: [PATCH v6 09/10] scsi: ufs-exynos: add UFS host support for Exynos SoCs

Hi Avri,

> -----Original Message-----
> From: Avri Altman <[email protected]>
> Sent: 20 April 2020 15:56
> To: Alim Akhtar <[email protected]>; [email protected]
> Cc: [email protected]; [email protected]; [email protected];
> [email protected]; [email protected];
> [email protected]; [email protected]; linux-samsung-
> [email protected]; [email protected]; linux-
> [email protected]
> Subject: RE: [PATCH v6 09/10] scsi: ufs-exynos: add UFS host support for Exynos
> SoCs
>
> > + struct exynos_ufs_drv_data *drv_data;
> > +
> > + u32 opts;
> > +#define EXYNOS_UFS_OPT_HAS_APB_CLK_CTRL BIT(0)
> > +#define EXYNOS_UFS_OPT_SKIP_CONNECTION_ESTAB BIT(1)
> > +#define EXYNOS_UFS_OPT_BROKEN_AUTO_CLK_CTRL BIT(2)
> > +#define EXYNOS_UFS_OPT_BROKEN_RX_SEL_IDX BIT(3)
> > +#define EXYNOS_UFS_OPT_USE_SW_HIBERN8_TIMER BIT(4)
> Could not find where the last 2 are being used.
>
The assignment is done on line# 988 and 989 in exynos_ufs_init() and used in {pre,post}_hibern8,
let me see if I can add these in the list of .opts in drv_data.
Thanks for pointing this out.

> Thanks,
> Avri

2020-04-21 05:56:03

by Can Guo

[permalink] [raw]
Subject: Re: [PATCH v6 01/10] scsi: ufs: add quirk to fix mishandling utrlclr/utmrlclr

On 2020-04-18 01:59, Alim Akhtar wrote:
> In the right behavior, setting the bit to '0' indicates clear and '1'
> indicates no change. If host controller handles this the other way,
> UFSHCI_QUIRK_BROKEN_REQ_LIST_CLR can be used.
>
> Signed-off-by: Seungwon Jeon <[email protected]>
> Signed-off-by: Alim Akhtar <[email protected]>

Reviewed-by: Can Guo <[email protected]>

> ---
> drivers/scsi/ufs/ufshcd.c | 11 +++++++++--
> drivers/scsi/ufs/ufshcd.h | 5 +++++
> 2 files changed, 14 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> index 698e8d20b4ba..3655b88fc862 100644
> --- a/drivers/scsi/ufs/ufshcd.c
> +++ b/drivers/scsi/ufs/ufshcd.c
> @@ -645,7 +645,11 @@ static inline int ufshcd_get_tr_ocs(struct
> ufshcd_lrb *lrbp)
> */
> static inline void ufshcd_utrl_clear(struct ufs_hba *hba, u32 pos)
> {
> - ufshcd_writel(hba, ~(1 << pos), REG_UTP_TRANSFER_REQ_LIST_CLEAR);
> + if (hba->quirks & UFSHCI_QUIRK_BROKEN_REQ_LIST_CLR)
> + ufshcd_writel(hba, (1 << pos), REG_UTP_TRANSFER_REQ_LIST_CLEAR);
> + else
> + ufshcd_writel(hba, ~(1 << pos),
> + REG_UTP_TRANSFER_REQ_LIST_CLEAR);
> }
>
> /**
> @@ -655,7 +659,10 @@ static inline void ufshcd_utrl_clear(struct
> ufs_hba *hba, u32 pos)
> */
> static inline void ufshcd_utmrl_clear(struct ufs_hba *hba, u32 pos)
> {
> - ufshcd_writel(hba, ~(1 << pos), REG_UTP_TASK_REQ_LIST_CLEAR);
> + if (hba->quirks & UFSHCI_QUIRK_BROKEN_REQ_LIST_CLR)
> + ufshcd_writel(hba, (1 << pos), REG_UTP_TASK_REQ_LIST_CLEAR);
> + else
> + ufshcd_writel(hba, ~(1 << pos), REG_UTP_TASK_REQ_LIST_CLEAR);
> }
>
> /**
> diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
> index 6ffc08ad85f6..071f0edf3f64 100644
> --- a/drivers/scsi/ufs/ufshcd.h
> +++ b/drivers/scsi/ufs/ufshcd.h
> @@ -518,6 +518,11 @@ enum ufshcd_quirks {
> * ops (get_ufs_hci_version) to get the correct version.
> */
> UFSHCD_QUIRK_BROKEN_UFS_HCI_VERSION = 1 << 5,
> +
> + /*
> + * Clear handling for transfer/task request list is just opposite.
> + */
> + UFSHCI_QUIRK_BROKEN_REQ_LIST_CLR = 1 << 6,
> };
>
> enum ufshcd_caps {
>
> base-commit: 8f3d9f354286745c751374f5f1fcafee6b3f3136

2020-04-21 06:03:39

by Can Guo

[permalink] [raw]
Subject: Re: [PATCH v6 03/10] scsi: ufs: add quirk to enable host controller without hce

On 2020-04-21 13:59, Can Guo wrote:
> On 2020-04-18 01:59, Alim Akhtar wrote:
>> Some host controllers don't support host controller enable via HCE.
>>
>> Signed-off-by: Seungwon Jeon <[email protected]>
>> Signed-off-by: Alim Akhtar <[email protected]>
>
> They are back again finally...
>
> Reviewd-by: Can Guo <[email protected]>

Reviewed-by: Can Guo <[email protected]>

>
>> ---
>> drivers/scsi/ufs/ufshcd.c | 76
>> +++++++++++++++++++++++++++++++++++++--
>> drivers/scsi/ufs/ufshcd.h | 6 ++++
>> 2 files changed, 80 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
>> index 0e9704da58bd..ee30ed6cc805 100644
>> --- a/drivers/scsi/ufs/ufshcd.c
>> +++ b/drivers/scsi/ufs/ufshcd.c
>> @@ -3534,6 +3534,52 @@ static int ufshcd_dme_link_startup(struct
>> ufs_hba *hba)
>> "dme-link-startup: error code %d\n", ret);
>> return ret;
>> }
>> +/**
>> + * ufshcd_dme_reset - UIC command for DME_RESET
>> + * @hba: per adapter instance
>> + *
>> + * DME_RESET command is issued in order to reset UniPro stack.
>> + * This function now deal with cold reset.
>> + *
>> + * Returns 0 on success, non-zero value on failure
>> + */
>> +static int ufshcd_dme_reset(struct ufs_hba *hba)
>> +{
>> + struct uic_command uic_cmd = {0};
>> + int ret;
>> +
>> + uic_cmd.command = UIC_CMD_DME_RESET;
>> +
>> + ret = ufshcd_send_uic_cmd(hba, &uic_cmd);
>> + if (ret)
>> + dev_err(hba->dev,
>> + "dme-reset: error code %d\n", ret);
>> +
>> + return ret;
>> +}
>> +
>> +/**
>> + * ufshcd_dme_enable - UIC command for DME_ENABLE
>> + * @hba: per adapter instance
>> + *
>> + * DME_ENABLE command is issued in order to enable UniPro stack.
>> + *
>> + * Returns 0 on success, non-zero value on failure
>> + */
>> +static int ufshcd_dme_enable(struct ufs_hba *hba)
>> +{
>> + struct uic_command uic_cmd = {0};
>> + int ret;
>> +
>> + uic_cmd.command = UIC_CMD_DME_ENABLE;
>> +
>> + ret = ufshcd_send_uic_cmd(hba, &uic_cmd);
>> + if (ret)
>> + dev_err(hba->dev,
>> + "dme-reset: error code %d\n", ret);
>> +
>> + return ret;
>> +}
>>
>> static inline void ufshcd_add_delay_before_dme_cmd(struct ufs_hba
>> *hba)
>> {
>> @@ -4251,7 +4297,7 @@ static inline void ufshcd_hba_stop(struct
>> ufs_hba *hba, bool can_sleep)
>> }
>>
>> /**
>> - * ufshcd_hba_enable - initialize the controller
>> + * ufshcd_hba_execute_hce - initialize the controller
>> * @hba: per adapter instance
>> *
>> * The controller resets itself and controller firmware
>> initialization
>> @@ -4260,7 +4306,7 @@ static inline void ufshcd_hba_stop(struct
>> ufs_hba *hba, bool can_sleep)
>> *
>> * Returns 0 on success, non-zero value on failure
>> */
>> -int ufshcd_hba_enable(struct ufs_hba *hba)
>> +static int ufshcd_hba_execute_hce(struct ufs_hba *hba)
>> {
>> int retry;
>>
>> @@ -4308,6 +4354,32 @@ int ufshcd_hba_enable(struct ufs_hba *hba)
>>
>> return 0;
>> }
>> +
>> +int ufshcd_hba_enable(struct ufs_hba *hba)
>> +{
>> + int ret;
>> +
>> + if (hba->quirks & UFSHCI_QUIRK_BROKEN_HCE) {
>> + ufshcd_set_link_off(hba);
>> + ufshcd_vops_hce_enable_notify(hba, PRE_CHANGE);
>> +
>> + /* enable UIC related interrupts */
>> + ufshcd_enable_intr(hba, UFSHCD_UIC_MASK);
>> + ret = ufshcd_dme_reset(hba);
>> + if (!ret) {
>> + ret = ufshcd_dme_enable(hba);
>> + if (!ret)
>> + ufshcd_vops_hce_enable_notify(hba, POST_CHANGE);
>> + if (ret)
>> + dev_err(hba->dev,
>> + "Host controller enable failed with non-hce\n");
>> + }
>> + } else {
>> + ret = ufshcd_hba_execute_hce(hba);
>> + }
>> +
>> + return ret;
>> +}
>> EXPORT_SYMBOL_GPL(ufshcd_hba_enable);
>>
>> static int ufshcd_disable_tx_lcc(struct ufs_hba *hba, bool peer)
>> diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
>> index 53096642f9a8..f8d08cb9caf7 100644
>> --- a/drivers/scsi/ufs/ufshcd.h
>> +++ b/drivers/scsi/ufs/ufshcd.h
>> @@ -529,6 +529,12 @@ enum ufshcd_quirks {
>> * that the interrupt aggregation timer and counter are reset by
>> s/w.
>> */
>> UFSHCI_QUIRK_SKIP_RESET_INTR_AGGR = 1 << 7,
>> +
>> + /*
>> + * This quirks needs to be enabled if host controller cannot be
>> + * enabled via HCE register.
>> + */
>> + UFSHCI_QUIRK_BROKEN_HCE = 1 << 8,
>> };
>>
>> enum ufshcd_caps {

2020-04-21 06:03:43

by Can Guo

[permalink] [raw]
Subject: Re: [PATCH v6 03/10] scsi: ufs: add quirk to enable host controller without hce

On 2020-04-18 01:59, Alim Akhtar wrote:
> Some host controllers don't support host controller enable via HCE.
>
> Signed-off-by: Seungwon Jeon <[email protected]>
> Signed-off-by: Alim Akhtar <[email protected]>

They are back again finally...

Reviewd-by: Can Guo <[email protected]>

> ---
> drivers/scsi/ufs/ufshcd.c | 76 +++++++++++++++++++++++++++++++++++++--
> drivers/scsi/ufs/ufshcd.h | 6 ++++
> 2 files changed, 80 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> index 0e9704da58bd..ee30ed6cc805 100644
> --- a/drivers/scsi/ufs/ufshcd.c
> +++ b/drivers/scsi/ufs/ufshcd.c
> @@ -3534,6 +3534,52 @@ static int ufshcd_dme_link_startup(struct
> ufs_hba *hba)
> "dme-link-startup: error code %d\n", ret);
> return ret;
> }
> +/**
> + * ufshcd_dme_reset - UIC command for DME_RESET
> + * @hba: per adapter instance
> + *
> + * DME_RESET command is issued in order to reset UniPro stack.
> + * This function now deal with cold reset.
> + *
> + * Returns 0 on success, non-zero value on failure
> + */
> +static int ufshcd_dme_reset(struct ufs_hba *hba)
> +{
> + struct uic_command uic_cmd = {0};
> + int ret;
> +
> + uic_cmd.command = UIC_CMD_DME_RESET;
> +
> + ret = ufshcd_send_uic_cmd(hba, &uic_cmd);
> + if (ret)
> + dev_err(hba->dev,
> + "dme-reset: error code %d\n", ret);
> +
> + return ret;
> +}
> +
> +/**
> + * ufshcd_dme_enable - UIC command for DME_ENABLE
> + * @hba: per adapter instance
> + *
> + * DME_ENABLE command is issued in order to enable UniPro stack.
> + *
> + * Returns 0 on success, non-zero value on failure
> + */
> +static int ufshcd_dme_enable(struct ufs_hba *hba)
> +{
> + struct uic_command uic_cmd = {0};
> + int ret;
> +
> + uic_cmd.command = UIC_CMD_DME_ENABLE;
> +
> + ret = ufshcd_send_uic_cmd(hba, &uic_cmd);
> + if (ret)
> + dev_err(hba->dev,
> + "dme-reset: error code %d\n", ret);
> +
> + return ret;
> +}
>
> static inline void ufshcd_add_delay_before_dme_cmd(struct ufs_hba
> *hba)
> {
> @@ -4251,7 +4297,7 @@ static inline void ufshcd_hba_stop(struct
> ufs_hba *hba, bool can_sleep)
> }
>
> /**
> - * ufshcd_hba_enable - initialize the controller
> + * ufshcd_hba_execute_hce - initialize the controller
> * @hba: per adapter instance
> *
> * The controller resets itself and controller firmware initialization
> @@ -4260,7 +4306,7 @@ static inline void ufshcd_hba_stop(struct
> ufs_hba *hba, bool can_sleep)
> *
> * Returns 0 on success, non-zero value on failure
> */
> -int ufshcd_hba_enable(struct ufs_hba *hba)
> +static int ufshcd_hba_execute_hce(struct ufs_hba *hba)
> {
> int retry;
>
> @@ -4308,6 +4354,32 @@ int ufshcd_hba_enable(struct ufs_hba *hba)
>
> return 0;
> }
> +
> +int ufshcd_hba_enable(struct ufs_hba *hba)
> +{
> + int ret;
> +
> + if (hba->quirks & UFSHCI_QUIRK_BROKEN_HCE) {
> + ufshcd_set_link_off(hba);
> + ufshcd_vops_hce_enable_notify(hba, PRE_CHANGE);
> +
> + /* enable UIC related interrupts */
> + ufshcd_enable_intr(hba, UFSHCD_UIC_MASK);
> + ret = ufshcd_dme_reset(hba);
> + if (!ret) {
> + ret = ufshcd_dme_enable(hba);
> + if (!ret)
> + ufshcd_vops_hce_enable_notify(hba, POST_CHANGE);
> + if (ret)
> + dev_err(hba->dev,
> + "Host controller enable failed with non-hce\n");
> + }
> + } else {
> + ret = ufshcd_hba_execute_hce(hba);
> + }
> +
> + return ret;
> +}
> EXPORT_SYMBOL_GPL(ufshcd_hba_enable);
>
> static int ufshcd_disable_tx_lcc(struct ufs_hba *hba, bool peer)
> diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
> index 53096642f9a8..f8d08cb9caf7 100644
> --- a/drivers/scsi/ufs/ufshcd.h
> +++ b/drivers/scsi/ufs/ufshcd.h
> @@ -529,6 +529,12 @@ enum ufshcd_quirks {
> * that the interrupt aggregation timer and counter are reset by s/w.
> */
> UFSHCI_QUIRK_SKIP_RESET_INTR_AGGR = 1 << 7,
> +
> + /*
> + * This quirks needs to be enabled if host controller cannot be
> + * enabled via HCE register.
> + */
> + UFSHCI_QUIRK_BROKEN_HCE = 1 << 8,
> };
>
> enum ufshcd_caps {

2020-04-21 07:51:41

by Kiwoong Kim

[permalink] [raw]
Subject: RE: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs fatal error

> -----Original Message-----
> From: Avri Altman <[email protected]>
> Sent: Monday, April 20, 2020 5:56 PM
> To: Alim Akhtar <[email protected]>; [email protected]
> Cc: [email protected]; [email protected];
> [email protected]; [email protected]; [email protected];
> [email protected]; [email protected]; linux-samsung-
> [email protected]; [email protected]; linux-
> [email protected]
> Subject: RE: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs
> fatal error
>
> >
> > From: Kiwoong Kim <[email protected]>
> >
> > Some architectures determines if fatal error for OCS occurrs to check
> > status in response upiu. This patch
> Typo - occurs
>
> > is to prevent from reporting command results with that.
> >
> > Signed-off-by: Kiwoong Kim <[email protected]>
> > Signed-off-by: Alim Akhtar <[email protected]>
> > ---
> > drivers/scsi/ufs/ufshcd.c | 6 ++++++
> > drivers/scsi/ufs/ufshcd.h | 6 ++++++
> > 2 files changed, 12 insertions(+)
> >
> > diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> > index b32fcedcdcb9..8c07caff0a5c 100644
> > --- a/drivers/scsi/ufs/ufshcd.c
> > +++ b/drivers/scsi/ufs/ufshcd.c
> > @@ -4794,6 +4794,12 @@ ufshcd_transfer_rsp_status(struct ufs_hba *hba,
> > struct ufshcd_lrb *lrbp)
> > /* overall command status of utrd */
> > ocs = ufshcd_get_tr_ocs(lrbp);
> >
> > + if (hba->quirks & UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR) {
> > + if (be32_to_cpu(lrbp->ucd_rsp_ptr->header.dword_1) &
> > + MASK_RSP_UPIU_RESULT)
> > + ocs = OCS_SUCCESS;
> > + }
> > +
> Not sure that I follow what this quirk is all about.
> Your code overrides ocs by open coding ufshcd_get_rsp_upiu_result.
>
> Normally OCS is in utp transfer req descriptor, dword 2, bits 0..7.
> My understanding from your description, is that some fatal error might
> occur, But the host controller does not report it, and it still needs to
> be checked in the response upiu.
> Evidently you are not doing so.
> Please elaborate your description.
>
> P.S.
> The ocs is being evaluated in device management commands as well, Isn't
> this something you need to attend?
>
> Thanks,
> Avri
>
> > switch (ocs) {
> > case OCS_SUCCESS:
> > result = ufshcd_get_req_rsp(lrbp->ucd_rsp_ptr);
> > diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
> > index a9b9ace9fc72..e1d09c2c4302 100644
> > --- a/drivers/scsi/ufs/ufshcd.h
> > +++ b/drivers/scsi/ufs/ufshcd.h
> > @@ -541,6 +541,12 @@ enum ufshcd_quirks {
> > * resolution of the values of PRDTO and PRDTL in UTRD as byte.
> > */
> > UFSHCD_QUIRK_PRDT_BYTE_GRAN = 1 << 9,
> > +
> > + /*
> > + * This quirk needs to be enabled if the host controller reports
> > + * OCS FATAL ERROR with device error through sense data
> > + */
> > + UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR = 1 << 10,
> > };
> >
> > enum ufshcd_caps {
> > --
> > 2.17.1
Avri

As specified in the spec, OCS isn't supposed to refer to the contents of RESPONSE UPIU.
But, Exynos host behaves like that in some cases, e.g. a value of 'state' in is isn't GOOD(00h).

For QUERY RESPONSE, its offset, i.e. " dword_1" is reserved, so currently no impact, I think.
But if you feel another condition is necessary to identify if this request is QUERY REQEUST or not, we can add more.

Thanks


2020-04-21 07:54:47

by Kiwoong Kim

[permalink] [raw]
Subject: RE: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs fatal error



> -----Original Message-----
> From: Avri Altman <[email protected]>
> Sent: Monday, April 20, 2020 5:56 PM
> To: Alim Akhtar <[email protected]>; [email protected]
> Cc: [email protected]; [email protected];
> [email protected]; [email protected]; [email protected];
> [email protected]; [email protected]; linux-samsung-
> [email protected]; [email protected]; linux-
> [email protected]
> Subject: RE: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs
> fatal error
>
> >
> > From: Kiwoong Kim <[email protected]>
> >
> > Some architectures determines if fatal error for OCS occurrs to check
> > status in response upiu. This patch
> Typo - occurs
>
> > is to prevent from reporting command results with that.
> >
> > Signed-off-by: Kiwoong Kim <[email protected]>
> > Signed-off-by: Alim Akhtar <[email protected]>
> > ---
> > drivers/scsi/ufs/ufshcd.c | 6 ++++++
> > drivers/scsi/ufs/ufshcd.h | 6 ++++++
> > 2 files changed, 12 insertions(+)
> >
> > diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> > index b32fcedcdcb9..8c07caff0a5c 100644
> > --- a/drivers/scsi/ufs/ufshcd.c
> > +++ b/drivers/scsi/ufs/ufshcd.c
> > @@ -4794,6 +4794,12 @@ ufshcd_transfer_rsp_status(struct ufs_hba *hba,
> > struct ufshcd_lrb *lrbp)
> > /* overall command status of utrd */
> > ocs = ufshcd_get_tr_ocs(lrbp);
> >
> > + if (hba->quirks & UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR) {
> > + if (be32_to_cpu(lrbp->ucd_rsp_ptr->header.dword_1) &
> > + MASK_RSP_UPIU_RESULT)
> > + ocs = OCS_SUCCESS;
> > + }
> > +
> Not sure that I follow what this quirk is all about.
> Your code overrides ocs by open coding ufshcd_get_rsp_upiu_result.
>
> Normally OCS is in utp transfer req descriptor, dword 2, bits 0..7.
> My understanding from your description, is that some fatal error might
> occur, But the host controller does not report it, and it still needs to
> be checked in the response upiu.
> Evidently you are not doing so.
> Please elaborate your description.
>
> P.S.
> The ocs is being evaluated in device management commands as well, Isn't
> this something you need to attend?
>
> Thanks,
> Avri
>
> > switch (ocs) {
> > case OCS_SUCCESS:
> > result = ufshcd_get_req_rsp(lrbp->ucd_rsp_ptr);
> > diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
> > index a9b9ace9fc72..e1d09c2c4302 100644
> > --- a/drivers/scsi/ufs/ufshcd.h
> > +++ b/drivers/scsi/ufs/ufshcd.h
> > @@ -541,6 +541,12 @@ enum ufshcd_quirks {
> > * resolution of the values of PRDTO and PRDTL in UTRD as byte.
> > */
> > UFSHCD_QUIRK_PRDT_BYTE_GRAN = 1 << 9,
> > +
> > + /*
> > + * This quirk needs to be enabled if the host controller reports
> > + * OCS FATAL ERROR with device error through sense data
> > + */
> > + UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR = 1 << 10,
> > };
> >
> > enum ufshcd_caps {
> > --
> > 2.17.1
Avri

As specified in the spec, OCS isn't supposed to refer to the contents of RESPONSE UPIU.
But, Exynos host behaves like that in some cases, e.g. a value of 'state' in is isn't GOOD(00h).

For QUERY RESPONSE, its offset, i.e. " dword_1" is reserved, so currently no impact, I think.
But if you feel another condition is necessary to identify if this request is QUERY REQEUST or not, we can add more.

Thanks

2020-04-21 12:08:55

by Avri Altman

[permalink] [raw]
Subject: RE: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs fatal error

>
> > -----Original Message-----
> > From: Avri Altman <[email protected]>
> > Sent: Monday, April 20, 2020 5:56 PM
> > To: Alim Akhtar <[email protected]>; [email protected]
> > Cc: [email protected]; [email protected];
> > [email protected]; [email protected];
> [email protected];
> > [email protected]; [email protected]; linux-samsung-
> > [email protected]; [email protected]; linux-
> > [email protected]
> > Subject: RE: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs
> > fatal error
> >
> > >
> > > From: Kiwoong Kim <[email protected]>
> > >
> > > Some architectures determines if fatal error for OCS occurrs to check
> > > status in response upiu. This patch
> > Typo - occurs
> >
> > > is to prevent from reporting command results with that.
> > >
> > > Signed-off-by: Kiwoong Kim <[email protected]>
> > > Signed-off-by: Alim Akhtar <[email protected]>
> > > ---
> > > drivers/scsi/ufs/ufshcd.c | 6 ++++++
> > > drivers/scsi/ufs/ufshcd.h | 6 ++++++
> > > 2 files changed, 12 insertions(+)
> > >
> > > diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> > > index b32fcedcdcb9..8c07caff0a5c 100644
> > > --- a/drivers/scsi/ufs/ufshcd.c
> > > +++ b/drivers/scsi/ufs/ufshcd.c
> > > @@ -4794,6 +4794,12 @@ ufshcd_transfer_rsp_status(struct ufs_hba
> *hba,
> > > struct ufshcd_lrb *lrbp)
> > > /* overall command status of utrd */
> > > ocs = ufshcd_get_tr_ocs(lrbp);
> > >
> > > + if (hba->quirks & UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR) {
> > > + if (be32_to_cpu(lrbp->ucd_rsp_ptr->header.dword_1) &
> > > + MASK_RSP_UPIU_RESULT)
> > > + ocs = OCS_SUCCESS;
> > > + }
> > > +
> > Not sure that I follow what this quirk is all about.
> > Your code overrides ocs by open coding ufshcd_get_rsp_upiu_result.
> >
> > Normally OCS is in utp transfer req descriptor, dword 2, bits 0..7.
> > My understanding from your description, is that some fatal error might
> > occur, But the host controller does not report it, and it still needs to
> > be checked in the response upiu.
> > Evidently you are not doing so.
> > Please elaborate your description.
> >
> > P.S.
> > The ocs is being evaluated in device management commands as well, Isn't
> > this something you need to attend?
> >
> > Thanks,
> > Avri
> >
> > > switch (ocs) {
> > > case OCS_SUCCESS:
> > > result = ufshcd_get_req_rsp(lrbp->ucd_rsp_ptr);
> > > diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
> > > index a9b9ace9fc72..e1d09c2c4302 100644
> > > --- a/drivers/scsi/ufs/ufshcd.h
> > > +++ b/drivers/scsi/ufs/ufshcd.h
> > > @@ -541,6 +541,12 @@ enum ufshcd_quirks {
> > > * resolution of the values of PRDTO and PRDTL in UTRD as byte.
> > > */
> > > UFSHCD_QUIRK_PRDT_BYTE_GRAN = 1 << 9,
> > > +
> > > + /*
> > > + * This quirk needs to be enabled if the host controller reports
> > > + * OCS FATAL ERROR with device error through sense data
> > > + */
> > > + UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR = 1 << 10,
> > > };
> > >
> > > enum ufshcd_caps {
> > > --
> > > 2.17.1
> Avri
>
> As specified in the spec, OCS isn't supposed to refer to the contents of
> RESPONSE UPIU.
> But, Exynos host behaves like that in some cases, e.g. a value of 'state' in is
> isn't GOOD(00h).
OK.
I still think that you might consider rewording your commit, explaining this quirk better.
Specifically you might not want to say "if fatal..." because fatal code (0x7) is just one error code out of many.
Also you might want to use ufshcd_get_rsp_upiu_result() in the quirk body instead of open coding it.

>
> For QUERY RESPONSE, its offset, i.e. " dword_1" is reserved, so currently no
> impact, I think.
> But if you feel another condition is necessary to identify if this request is
> QUERY REQEUST or not, we can add more.
No need, as long as you are ok with whatever ufshcd_get_tr_ocs() returns in ufshcd_wait_for_dev_cmd().

Thanks,
Avri

>
> Thanks

2020-04-21 16:59:21

by Alim Akhtar

[permalink] [raw]
Subject: Re: [PATCH v6 08/10] dt-bindings: ufs: Add DT binding documentation for ufs

Hi Rob
Request to comment on this dt-bindings documentation.
Thanks

On Fri, Apr 17, 2020 at 11:41 PM Alim Akhtar <[email protected]> wrote:
>
> This patch adds DT binding for samsung ufs hci
>
> Signed-off-by: Alim Akhtar <[email protected]>
> ---
> .../bindings/ufs/samsung,exynos-ufs.yaml | 93 +++++++++++++++++++
> 1 file changed, 93 insertions(+)
> create mode 100644 Documentation/devicetree/bindings/ufs/samsung,exynos-ufs.yaml
>
> diff --git a/Documentation/devicetree/bindings/ufs/samsung,exynos-ufs.yaml b/Documentation/devicetree/bindings/ufs/samsung,exynos-ufs.yaml
> new file mode 100644
> index 000000000000..954338b7f37d
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/ufs/samsung,exynos-ufs.yaml
> @@ -0,0 +1,93 @@
> +# SPDX-License-Identifier: GPL-2.0-only OR BSD-2-Clause
> +%YAML 1.2
> +---
> +$id: http://devicetree.org/schemas/ufs/samsung,exynos-ufs.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> +
> +title: Samsung SoC series UFS host controller Device Tree Bindings
> +
> +maintainers:
> + - Alim Akhtar <[email protected]>
> +
> +description: |
> + Each Samsung UFS host controller instance should have its own node.
> + This binding define Samsung specific binding other then what is used
> + in the common ufshcd bindings
> + [1] Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> +
> +properties:
> +
> + compatible:
> + enum:
> + - samsung,exynos7-ufs
> +
> + reg:
> + items:
> + - description: HCI register
> + - description: vendor specific register
> + - description: unipro register
> + - description: UFS protector register
> +
> + reg-names:
> + items:
> + - const: hci
> + - const: vs_hci
> + - const: unipro
> + - const: ufsp
> +
> + clocks:
> + maxItems: 2
> + items:
> + - description: ufs link core clock
> + - description: unipro main clock
> +
> + clock-names:
> + maxItems: 2
> + items:
> + - const: core_clk
> + - const: sclk_unipro_main
> +
> + interrupts:
> + items:
> + - description: interrupt signal for various ufshc status
> +
> + phys:
> + maxItems: 1
> + description:
> + phandle of the ufs phy node
> +
> + phy-names:
> + const: ufs-phy
> +
> +required:
> + - compatible
> + - reg
> + - interrupts
> + - phys
> + - phy-names
> + - clocks
> + - clock-names
> +
> +examples:
> + - |
> + #include <dt-bindings/interrupt-controller/arm-gic.h>
> + #include <dt-bindings/clock/exynos7-clk.h>
> +
> + ufs: ufs@15570000 {
> + compatible = "samsung,exynos7-ufs";
> + reg = <0x15570000 0x100>,
> + <0x15570100 0x100>,
> + <0x15571000 0x200>,
> + <0x15572000 0x300>;
> + reg-names = "hci", "vs_hci", "unipro", "ufsp";
> + interrupts = <GIC_SPI 200 IRQ_TYPE_LEVEL_HIGH>;
> + clocks = <&clock_fsys1 ACLK_UFS20_LINK>,
> + <&clock_fsys1 SCLK_UFSUNIPRO20_USER>;
> + clock-names = "core_clk", "sclk_unipro_main";
> + pinctrl-names = "default";
> + pinctrl-0 = <&ufs_rst_n &ufs_refclk_out>;
> + pclk-freq-avail-range = <70000000 133000000>;
> + phys = <&ufs_phy>;
> + phy-names = "ufs-phy";
> + };
> +...
> --
> 2.17.1
>


--
Regards,
Alim

2020-04-21 17:00:30

by Alim Akhtar

[permalink] [raw]
Subject: Re: [PATCH v6 06/10] dt-bindings: phy: Document Samsung UFS PHY bindings

Hi Rob
Request you to comment on this dt-bindings documentation.
Thanks

On Fri, Apr 17, 2020 at 11:43 PM Alim Akhtar <[email protected]> wrote:
>
> This patch documents Samsung UFS PHY device tree bindings
>
> Signed-off-by: Alim Akhtar <[email protected]>
> Tested-by: Paweł Chmiel <[email protected]>
> ---
> .../bindings/phy/samsung,ufs-phy.yaml | 74 +++++++++++++++++++
> 1 file changed, 74 insertions(+)
> create mode 100644 Documentation/devicetree/bindings/phy/samsung,ufs-phy.yaml
>
> diff --git a/Documentation/devicetree/bindings/phy/samsung,ufs-phy.yaml b/Documentation/devicetree/bindings/phy/samsung,ufs-phy.yaml
> new file mode 100644
> index 000000000000..352d5dda320d
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/phy/samsung,ufs-phy.yaml
> @@ -0,0 +1,74 @@
> +# SPDX-License-Identifier: GPL-2.0-only OR BSD-2-Clause
> +%YAML 1.2
> +---
> +$id: http://devicetree.org/schemas/phy/samsung,ufs-phy.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> +
> +title: Samsung SoC series UFS PHY Device Tree Bindings
> +
> +maintainers:
> + - Alim Akhtar <[email protected]>
> +
> +properties:
> + "#phy-cells":
> + const: 0
> +
> + compatible:
> + enum:
> + - samsung,exynos7-ufs-phy
> +
> + reg:
> + maxItems: 1
> + description: PHY base register address
> +
> + reg-names:
> + items:
> + - const: phy-pma
> +
> + clocks:
> + items:
> + - description: PLL reference clock
> + - description: symbol clock for input symbol ( rx0-ch0 symbol clock)
> + - description: symbol clock for input symbol ( rx1-ch1 symbol clock)
> + - description: symbol clock for output symbol ( tx0 symbol clock)
> +
> + clock-names:
> + items:
> + - const: ref_clk
> + - const: rx1_symbol_clk
> + - const: rx0_symbol_clk
> + - const: tx0_symbol_clk
> +
> + samsung,pmu-syscon:
> + $ref: '/schemas/types.yaml#/definitions/phandle'
> + description: phandle for PMU system controller interface, used to
> + control pmu registers bits for ufs m-phy
> +
> +required:
> + - "#phy-cells"
> + - compatible
> + - reg
> + - reg-names
> + - clocks
> + - clock-names
> + - samsung,pmu-syscon
> +
> +examples:
> + - |
> + #include <dt-bindings/clock/exynos7-clk.h>
> +
> + ufs_phy: ufs-phy@15571800 {
> + compatible = "samsung,exynos7-ufs-phy";
> + reg = <0x15571800 0x240>;
> + reg-names = "phy-pma";
> + samsung,pmu-syscon = <&pmu_system_controller>;
> + #phy-cells = <0>;
> + clocks = <&clock_fsys1 SCLK_COMBO_PHY_EMBEDDED_26M>,
> + <&clock_fsys1 PHYCLK_UFS20_RX1_SYMBOL_USER>,
> + <&clock_fsys1 PHYCLK_UFS20_RX0_SYMBOL_USER>,
> + <&clock_fsys1 PHYCLK_UFS20_TX0_SYMBOL_USER>;
> + clock-names = "ref_clk", "rx1_symbol_clk",
> + "rx0_symbol_clk", "tx0_symbol_clk";
> +
> + };
> +...
> --
> 2.17.1
>


--
Regards,
Alim

2020-04-22 06:55:44

by Christoph Hellwig

[permalink] [raw]
Subject: Re: [PATCH v6 04/10] scsi: ufs: introduce UFSHCD_QUIRK_PRDT_BYTE_GRAN quirk

On Fri, Apr 17, 2020 at 11:29:38PM +0530, Alim Akhtar wrote:
> Some UFS host controllers may think granularities of PRDT length and
> offset as bytes, not double words.
>
> Signed-off-by: Kiwoong Kim <[email protected]>
> Signed-off-by: Alim Akhtar <[email protected]>
> ---
> drivers/scsi/ufs/ufshcd.c | 30 +++++++++++++++++++++++-------
> drivers/scsi/ufs/ufshcd.h | 6 ++++++
> 2 files changed, 29 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> index ee30ed6cc805..b32fcedcdcb9 100644
> --- a/drivers/scsi/ufs/ufshcd.c
> +++ b/drivers/scsi/ufs/ufshcd.c
> @@ -2151,8 +2151,14 @@ static int ufshcd_map_sg(struct ufs_hba *hba, struct ufshcd_lrb *lrbp)
> return sg_segments;
>
> if (sg_segments) {
> - lrbp->utr_descriptor_ptr->prd_table_length =
> - cpu_to_le16((u16)sg_segments);
> +
> + if (hba->quirks & UFSHCD_QUIRK_PRDT_BYTE_GRAN)
> + lrbp->utr_descriptor_ptr->prd_table_length =
> + cpu_to_le16((u16)(sg_segments *
> + sizeof(struct ufshcd_sg_entry)));
> + else
> + lrbp->utr_descriptor_ptr->prd_table_length =
> + cpu_to_le16((u16) (sg_segments));

No double words here. "Normal" UFS uses the actual segment count,
while Samsumg uses bytes. Also no need fo the u16 count in
either the old or new version.

2020-04-22 06:57:20

by Christoph Hellwig

[permalink] [raw]
Subject: Re: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs fatal error

On Fri, Apr 17, 2020 at 11:29:39PM +0530, Alim Akhtar wrote:
> From: Kiwoong Kim <[email protected]>
>
> Some architectures determines if fatal error for OCS
> occurrs to check status in response upiu. This patch
> is to prevent from reporting command results with that.

What does "Some architectures" mean? All this seems to be about
error propagation to the SCSI midlyaer, so this sounds rather
strange.

2020-04-26 15:57:02

by Alim Akhtar

[permalink] [raw]
Subject: RE: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs fatal error

Hi Avri

> -----Original Message-----
> From: Avri Altman <[email protected]>
> Sent: 21 April 2020 17:37
> To: Kiwoong Kim <[email protected]>; 'Alim Akhtar'
> <[email protected]>; [email protected]; [email protected]
> Cc: [email protected]; [email protected]; [email protected];
> [email protected]; [email protected]; linux-samsung-
> [email protected]; [email protected]; linux-
> [email protected]
> Subject: RE: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs fatal error
>
> >
> > > -----Original Message-----
> > > From: Avri Altman <[email protected]>
> > > Sent: Monday, April 20, 2020 5:56 PM
> > > To: Alim Akhtar <[email protected]>; [email protected]
> > > Cc: [email protected]; [email protected];
> > > [email protected]; [email protected];
> > [email protected];
> > > [email protected]; [email protected]; linux-samsung-
> > > [email protected]; [email protected]; linux-
> > > [email protected]
> > > Subject: RE: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal
> > > ocs fatal error
> > >
> > > >
> > > > From: Kiwoong Kim <[email protected]>
> > > >
> > > > Some architectures determines if fatal error for OCS occurrs to
> > > > check status in response upiu. This patch
> > > Typo - occurs
> > >
> > > > is to prevent from reporting command results with that.
> > > >
> > > > Signed-off-by: Kiwoong Kim <[email protected]>
> > > > Signed-off-by: Alim Akhtar <[email protected]>
> > > > ---
> > > > drivers/scsi/ufs/ufshcd.c | 6 ++++++ drivers/scsi/ufs/ufshcd.h |
> > > > 6 ++++++
> > > > 2 files changed, 12 insertions(+)
> > > >
> > > > diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> > > > index b32fcedcdcb9..8c07caff0a5c 100644
> > > > --- a/drivers/scsi/ufs/ufshcd.c
> > > > +++ b/drivers/scsi/ufs/ufshcd.c
> > > > @@ -4794,6 +4794,12 @@ ufshcd_transfer_rsp_status(struct ufs_hba
> > *hba,
> > > > struct ufshcd_lrb *lrbp)
> > > > /* overall command status of utrd */
> > > > ocs = ufshcd_get_tr_ocs(lrbp);
> > > >
> > > > + if (hba->quirks & UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR) {
> > > > + if (be32_to_cpu(lrbp->ucd_rsp_ptr->header.dword_1) &
> > > > + MASK_RSP_UPIU_RESULT)
> > > > + ocs = OCS_SUCCESS;
> > > > + }
> > > > +
> > > Not sure that I follow what this quirk is all about.
> > > Your code overrides ocs by open coding ufshcd_get_rsp_upiu_result.
> > >
> > > Normally OCS is in utp transfer req descriptor, dword 2, bits 0..7.
> > > My understanding from your description, is that some fatal error
> > > might occur, But the host controller does not report it, and it
> > > still needs to be checked in the response upiu.
> > > Evidently you are not doing so.
> > > Please elaborate your description.
> > >
> > > P.S.
> > > The ocs is being evaluated in device management commands as well,
> > > Isn't this something you need to attend?
> > >
> > > Thanks,
> > > Avri
> > >
> > > > switch (ocs) {
> > > > case OCS_SUCCESS:
> > > > result = ufshcd_get_req_rsp(lrbp->ucd_rsp_ptr);
> > > > diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
> > > > index a9b9ace9fc72..e1d09c2c4302 100644
> > > > --- a/drivers/scsi/ufs/ufshcd.h
> > > > +++ b/drivers/scsi/ufs/ufshcd.h
> > > > @@ -541,6 +541,12 @@ enum ufshcd_quirks {
> > > > * resolution of the values of PRDTO and PRDTL in UTRD as byte.
> > > > */
> > > > UFSHCD_QUIRK_PRDT_BYTE_GRAN = 1 << 9,
> > > > +
> > > > + /*
> > > > + * This quirk needs to be enabled if the host controller reports
> > > > + * OCS FATAL ERROR with device error through sense data
> > > > + */
> > > > + UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR = 1 << 10,
> > > > };
> > > >
> > > > enum ufshcd_caps {
> > > > --
> > > > 2.17.1
> > Avri
> >
> > As specified in the spec, OCS isn't supposed to refer to the contents
> > of RESPONSE UPIU.
> > But, Exynos host behaves like that in some cases, e.g. a value of
> > 'state' in is isn't GOOD(00h).
> OK.
> I still think that you might consider rewording your commit, explaining this quirk
> better.
> Specifically you might not want to say "if fatal..." because fatal code (0x7) is just
> one error code out of many.
> Also you might want to use ufshcd_get_rsp_upiu_result() in the quirk body
> instead of open coding it.
>
> >
> > For QUERY RESPONSE, its offset, i.e. " dword_1" is reserved, so
> > currently no impact, I think.
> > But if you feel another condition is necessary to identify if this
> > request is QUERY REQEUST or not, we can add more.
> No need, as long as you are ok with whatever ufshcd_get_tr_ocs() returns in
> ufshcd_wait_for_dev_cmd().
>
I will update the commit message to make it clear in the next version of the patch set.

> Thanks,
> Avri
>
> >
> > Thanks


2020-04-26 15:58:47

by Alim Akhtar

[permalink] [raw]
Subject: RE: [PATCH v6 04/10] scsi: ufs: introduce UFSHCD_QUIRK_PRDT_BYTE_GRAN quirk

Hi Christoph,

> -----Original Message-----
> From: Christoph Hellwig <[email protected]>
> Sent: 22 April 2020 12:24
> To: Alim Akhtar <[email protected]>
> Cc: [email protected]; [email protected];
[email protected];
> [email protected]; [email protected]; [email protected];
> [email protected]; [email protected];
> [email protected]; [email protected]; linux-arm-
> [email protected]; [email protected]
> Subject: Re: [PATCH v6 04/10] scsi: ufs: introduce
> UFSHCD_QUIRK_PRDT_BYTE_GRAN quirk
>
> On Fri, Apr 17, 2020 at 11:29:38PM +0530, Alim Akhtar wrote:
> > Some UFS host controllers may think granularities of PRDT length and
> > offset as bytes, not double words.
> >
> > Signed-off-by: Kiwoong Kim <[email protected]>
> > Signed-off-by: Alim Akhtar <[email protected]>
> > ---
> > drivers/scsi/ufs/ufshcd.c | 30 +++++++++++++++++++++++-------
> > drivers/scsi/ufs/ufshcd.h | 6 ++++++
> > 2 files changed, 29 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> > index ee30ed6cc805..b32fcedcdcb9 100644
> > --- a/drivers/scsi/ufs/ufshcd.c
> > +++ b/drivers/scsi/ufs/ufshcd.c
> > @@ -2151,8 +2151,14 @@ static int ufshcd_map_sg(struct ufs_hba *hba,
> struct ufshcd_lrb *lrbp)
> > return sg_segments;
> >
> > if (sg_segments) {
> > - lrbp->utr_descriptor_ptr->prd_table_length =
> > - cpu_to_le16((u16)sg_segments);
> > +
> > + if (hba->quirks & UFSHCD_QUIRK_PRDT_BYTE_GRAN)
> > + lrbp->utr_descriptor_ptr->prd_table_length =
> > + cpu_to_le16((u16)(sg_segments *
> > + sizeof(struct ufshcd_sg_entry)));
> > + else
> > + lrbp->utr_descriptor_ptr->prd_table_length =
> > + cpu_to_le16((u16) (sg_segments));
>
> No double words here. "Normal" UFS uses the actual segment count, while
> Samsumg uses bytes. Also no need fo the u16 count in either the old or
new
> version.
Ok, will update the commit message and take your suggestion in the next
version.

2020-04-26 16:02:02

by Alim Akhtar

[permalink] [raw]
Subject: RE: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs fatal error



> -----Original Message-----
> From: Christoph Hellwig <[email protected]>
> Sent: 22 April 2020 12:26
> To: Alim Akhtar <[email protected]>
> Cc: [email protected]; [email protected];
[email protected];
> [email protected]; [email protected]; [email protected];
> [email protected]; [email protected];
> [email protected]; [email protected]; linux-arm-
> [email protected]; [email protected]
> Subject: Re: [PATCH v6 05/10] scsi: ufs: add quirk to fix abnormal ocs
fatal error
>
> On Fri, Apr 17, 2020 at 11:29:39PM +0530, Alim Akhtar wrote:
> > From: Kiwoong Kim <[email protected]>
> >
> > Some architectures determines if fatal error for OCS occurrs to check
> > status in response upiu. This patch is to prevent from reporting
> > command results with that.
>
> What does "Some architectures" mean? All this seems to be about error
> propagation to the SCSI midlyaer, so this sounds rather strange.
Ok will update the commit message with more details.