[PATCH v5 01/16] ufs: core: Optimize duplicate code to read extended feature

[PATCH v5 01/16] ufs: core: Optimize duplicate code to read extended feature

[Asutosh Das]

The code to parse the extended feature is duplicated twice
in the ufs core. Replace the duplicated code with a
function.

Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/ufs/core/ufshcd.c | 21 +++++++++++++--------
 1 file changed, 13 insertions(+), 8 deletions(-)

diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 768cb49..c9d7b78 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -215,6 +215,17 @@ ufs_get_desired_pm_lvl_for_dev_link_state(enum ufs_dev_pwr_mode dev_state,
 	return UFS_PM_LVL_0;
 }
 
+static unsigned int ufs_get_ext_ufs_feature(struct ufs_hba *hba,
+					    const u8 *desc_buf)
+{
+	if (hba->desc_size[QUERY_DESC_IDN_DEVICE] <
+	    DEVICE_DESC_PARAM_EXT_UFS_FEATURE_SUP + 4)
+		return 0;
+
+	return get_unaligned_be32(desc_buf +
+				  DEVICE_DESC_PARAM_EXT_UFS_FEATURE_SUP);
+}
+
 static const struct ufs_dev_quirk ufs_fixups[] = {
 	/* UFS cards deviations table */
 	{ .wmanufacturerid = UFS_VENDOR_MICRON,
@@ -7584,13 +7595,7 @@ static void ufshcd_wb_probe(struct ufs_hba *hba, const u8 *desc_buf)
 	     (hba->dev_quirks & UFS_DEVICE_QUIRK_SUPPORT_EXTENDED_FEATURES)))
 		goto wb_disabled;
 
-	if (hba->desc_size[QUERY_DESC_IDN_DEVICE] <
-	    DEVICE_DESC_PARAM_EXT_UFS_FEATURE_SUP + 4)
-		goto wb_disabled;
-
-	ext_ufs_feature = get_unaligned_be32(desc_buf +
-					DEVICE_DESC_PARAM_EXT_UFS_FEATURE_SUP);
-
+	ext_ufs_feature = ufs_get_ext_ufs_feature(hba, desc_buf);
 	if (!(ext_ufs_feature & UFS_DEV_WRITE_BOOSTER_SUP))
 		goto wb_disabled;
 
@@ -7644,7 +7649,7 @@ static void ufshcd_temp_notif_probe(struct ufs_hba *hba, const u8 *desc_buf)
 	if (!(hba->caps & UFSHCD_CAP_TEMP_NOTIF) || dev_info->wspecversion < 0x300)
 		return;
 
-	ext_ufs_feature = get_unaligned_be32(desc_buf + DEVICE_DESC_PARAM_EXT_UFS_FEATURE_SUP);
+	ext_ufs_feature = ufs_get_ext_ufs_feature(hba, desc_buf);
 
 	if (ext_ufs_feature & UFS_DEV_LOW_TEMP_NOTIF)
 		mask |= MASK_EE_TOO_LOW_TEMP;
-- 
2.7.4

Re: [PATCH v5 01/16] ufs: core: Optimize duplicate code to read extended feature

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:14PM -0800, Asutosh Das wrote:
> The code to parse the extended feature is duplicated twice
> in the ufs core. Replace the duplicated code with a
> function.
> 
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>

Reviewed-by: Manivannan Sadhasivam <mani@kernel.org>

Thanks,
Mani

> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
> ---
>  drivers/ufs/core/ufshcd.c | 21 +++++++++++++--------
>  1 file changed, 13 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 768cb49..c9d7b78 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -215,6 +215,17 @@ ufs_get_desired_pm_lvl_for_dev_link_state(enum ufs_dev_pwr_mode dev_state,
>  	return UFS_PM_LVL_0;
>  }
>  
> +static unsigned int ufs_get_ext_ufs_feature(struct ufs_hba *hba,
> +					    const u8 *desc_buf)
> +{
> +	if (hba->desc_size[QUERY_DESC_IDN_DEVICE] <
> +	    DEVICE_DESC_PARAM_EXT_UFS_FEATURE_SUP + 4)
> +		return 0;
> +
> +	return get_unaligned_be32(desc_buf +
> +				  DEVICE_DESC_PARAM_EXT_UFS_FEATURE_SUP);
> +}
> +
>  static const struct ufs_dev_quirk ufs_fixups[] = {
>  	/* UFS cards deviations table */
>  	{ .wmanufacturerid = UFS_VENDOR_MICRON,
> @@ -7584,13 +7595,7 @@ static void ufshcd_wb_probe(struct ufs_hba *hba, const u8 *desc_buf)
>  	     (hba->dev_quirks & UFS_DEVICE_QUIRK_SUPPORT_EXTENDED_FEATURES)))
>  		goto wb_disabled;
>  
> -	if (hba->desc_size[QUERY_DESC_IDN_DEVICE] <
> -	    DEVICE_DESC_PARAM_EXT_UFS_FEATURE_SUP + 4)
> -		goto wb_disabled;
> -
> -	ext_ufs_feature = get_unaligned_be32(desc_buf +
> -					DEVICE_DESC_PARAM_EXT_UFS_FEATURE_SUP);
> -
> +	ext_ufs_feature = ufs_get_ext_ufs_feature(hba, desc_buf);
>  	if (!(ext_ufs_feature & UFS_DEV_WRITE_BOOSTER_SUP))
>  		goto wb_disabled;
>  
> @@ -7644,7 +7649,7 @@ static void ufshcd_temp_notif_probe(struct ufs_hba *hba, const u8 *desc_buf)
>  	if (!(hba->caps & UFSHCD_CAP_TEMP_NOTIF) || dev_info->wspecversion < 0x300)
>  		return;
>  
> -	ext_ufs_feature = get_unaligned_be32(desc_buf + DEVICE_DESC_PARAM_EXT_UFS_FEATURE_SUP);
> +	ext_ufs_feature = ufs_get_ext_ufs_feature(hba, desc_buf);
>  
>  	if (ext_ufs_feature & UFS_DEV_LOW_TEMP_NOTIF)
>  		mask |= MASK_EE_TOO_LOW_TEMP;
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

[PATCH v5 02/16] ufs: core: Probe for ext_iid support

[Asutosh Das]

Task Tag is limited to 8 bits and this restricts the number
of active IOs to 255.
In Multi-circular queue mode, this may not be enough.
The specification provides EXT_IID which can be used to increase
the number of IOs if the UFS device and UFSHC support it.
This patch adds support to probe for ext_iid support in
ufs device and UFSHC.

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
---
 drivers/ufs/core/ufshcd.c | 31 +++++++++++++++++++++++++++++++
 include/ufs/ufs.h         |  4 ++++
 include/ufs/ufshcd.h      |  4 ++++
 include/ufs/ufshci.h      |  7 +++++++
 4 files changed, 46 insertions(+)

diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index c9d7b78..66b797f 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -2258,6 +2258,10 @@ static inline int ufshcd_hba_capabilities(struct ufs_hba *hba)
 	if (err)
 		dev_err(hba->dev, "crypto setup failed\n");
 
+	hba->mcq_capabilities = ufshcd_readl(hba, REG_MCQCAP);
+	hba->ext_iid_sup = FIELD_GET(MASK_EXT_IID_SUPPORT,
+				     hba->mcq_capabilities);
+
 	return err;
 }
 
@@ -7663,6 +7667,30 @@ static void ufshcd_temp_notif_probe(struct ufs_hba *hba, const u8 *desc_buf)
 	}
 }
 
+static void ufshcd_ext_iid_probe(struct ufs_hba *hba, u8 *desc_buf)
+{
+	struct ufs_dev_info *dev_info = &hba->dev_info;
+	u32 ext_ufs_feature;
+	u32 ext_iid_en = 0;
+	int err;
+
+	/* Only UFS-4.0 and above may support EXT_IID */
+	if (dev_info->wspecversion < 0x400)
+		goto out;
+
+	ext_ufs_feature = ufs_get_ext_ufs_feature(hba, desc_buf);
+	if (!(ext_ufs_feature & UFS_DEV_EXT_IID_SUP))
+		goto out;
+
+	err = ufshcd_query_attr_retry(hba, UPIU_QUERY_OPCODE_READ_ATTR,
+				      QUERY_ATTR_IDN_EXT_IID_EN, 0, 0, &ext_iid_en);
+	if (err)
+		dev_err(hba->dev, "failed reading bEXTIIDEn. err = %d\n", err);
+
+out:
+	dev_info->b_ext_iid_en = ext_iid_en;
+}
+
 void ufshcd_fixup_dev_quirks(struct ufs_hba *hba,
 			     const struct ufs_dev_quirk *fixups)
 {
@@ -7761,6 +7789,9 @@ static int ufs_get_device_desc(struct ufs_hba *hba)
 
 	ufshcd_temp_notif_probe(hba, desc_buf);
 
+	if (hba->ext_iid_sup)
+		ufshcd_ext_iid_probe(hba, desc_buf);
+
 	/*
 	 * ufshcd_read_string_desc returns size of the string
 	 * reset the error value
diff --git a/include/ufs/ufs.h b/include/ufs/ufs.h
index 1bba3fe..ba2a1d8 100644
--- a/include/ufs/ufs.h
+++ b/include/ufs/ufs.h
@@ -165,6 +165,7 @@ enum attr_idn {
 	QUERY_ATTR_IDN_AVAIL_WB_BUFF_SIZE       = 0x1D,
 	QUERY_ATTR_IDN_WB_BUFF_LIFE_TIME_EST    = 0x1E,
 	QUERY_ATTR_IDN_CURR_WB_BUFF_SIZE        = 0x1F,
+	QUERY_ATTR_IDN_EXT_IID_EN		= 0x2A,
 };
 
 /* Descriptor idn for Query requests */
@@ -352,6 +353,7 @@ enum {
 	UFS_DEV_EXT_TEMP_NOTIF		= BIT(6),
 	UFS_DEV_HPB_SUPPORT		= BIT(7),
 	UFS_DEV_WRITE_BOOSTER_SUP	= BIT(8),
+	UFS_DEV_EXT_IID_SUP		= BIT(16),
 };
 #define UFS_DEV_HPB_SUPPORT_VERSION		0x310
 
@@ -601,6 +603,8 @@ struct ufs_dev_info {
 
 	bool	b_rpm_dev_flush_capable;
 	u8	b_presrv_uspc_en;
+	/* UFS EXT_IID Enable */
+	bool	b_ext_iid_en;
 };
 
 /*
diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
index 5cf81df..aec37cb9 100644
--- a/include/ufs/ufshcd.h
+++ b/include/ufs/ufshcd.h
@@ -747,6 +747,7 @@ struct ufs_hba_monitor {
  * @outstanding_lock: Protects @outstanding_reqs.
  * @outstanding_reqs: Bits representing outstanding transfer requests
  * @capabilities: UFS Controller Capabilities
+ * @mcq_capabilities: UFS Multi Circular Queue capabilities
  * @nutrs: Transfer Request Queue depth supported by controller
  * @nutmrs: Task Management Queue depth supported by controller
  * @reserved_slot: Used to submit device commands. Protected by @dev_cmd.lock.
@@ -830,6 +831,7 @@ struct ufs_hba_monitor {
  *	device
  * @complete_put: whether or not to call ufshcd_rpm_put() from inside
  *	ufshcd_resume_complete()
+ * @ext_iid_sup: is EXT_IID is supported by UFSHC
  */
 struct ufs_hba {
 	void __iomem *mmio_base;
@@ -871,6 +873,7 @@ struct ufs_hba {
 
 	u32 capabilities;
 	int nutrs;
+	u32 mcq_capabilities;
 	int nutmrs;
 	u32 reserved_slot;
 	u32 ufs_version;
@@ -978,6 +981,7 @@ struct ufs_hba {
 #endif
 	u32 luns_avail;
 	bool complete_put;
+	bool ext_iid_sup;
 };
 
 /* Returns true if clocks can be gated. Otherwise false */
diff --git a/include/ufs/ufshci.h b/include/ufs/ufshci.h
index f525566..4d4da06 100644
--- a/include/ufs/ufshci.h
+++ b/include/ufs/ufshci.h
@@ -22,6 +22,7 @@ enum {
 /* UFSHCI Registers */
 enum {
 	REG_CONTROLLER_CAPABILITIES		= 0x00,
+	REG_MCQCAP				= 0x04,
 	REG_UFS_VERSION				= 0x08,
 	REG_CONTROLLER_DEV_ID			= 0x10,
 	REG_CONTROLLER_PROD_ID			= 0x14,
@@ -68,6 +69,12 @@ enum {
 	MASK_OUT_OF_ORDER_DATA_DELIVERY_SUPPORT	= 0x02000000,
 	MASK_UIC_DME_TEST_MODE_SUPPORT		= 0x04000000,
 	MASK_CRYPTO_SUPPORT			= 0x10000000,
+	MASK_MCQ_SUPPORT			= 0x40000000,
+};
+
+/* MCQ capability mask */
+enum {
+	MASK_EXT_IID_SUPPORT = 0x00000400,
 };
 
 #define UFS_MASK(mask, offset)		((mask) << (offset))
-- 
2.7.4

Re: [PATCH v5 02/16] ufs: core: Probe for ext_iid support

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:15PM -0800, Asutosh Das wrote:
> Task Tag is limited to 8 bits and this restricts the number
> of active IOs to 255.
> In Multi-circular queue mode, this may not be enough.
> The specification provides EXT_IID which can be used to increase
> the number of IOs if the UFS device and UFSHC support it.
> This patch adds support to probe for ext_iid support in
> ufs device and UFSHC.
> 
> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>

One nitpick below. But even without addressing that, this patch looks good to
me.

Reviewed-by: Manivannan Sadhasivam <mani@kernel.org>

> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
> Reviewed-by: Avri Altman <avri.altman@wdc.com>
> ---
>  drivers/ufs/core/ufshcd.c | 31 +++++++++++++++++++++++++++++++
>  include/ufs/ufs.h         |  4 ++++
>  include/ufs/ufshcd.h      |  4 ++++
>  include/ufs/ufshci.h      |  7 +++++++
>  4 files changed, 46 insertions(+)
> 
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index c9d7b78..66b797f 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -2258,6 +2258,10 @@ static inline int ufshcd_hba_capabilities(struct ufs_hba *hba)
>  	if (err)
>  		dev_err(hba->dev, "crypto setup failed\n");
>  
> +	hba->mcq_capabilities = ufshcd_readl(hba, REG_MCQCAP);
> +	hba->ext_iid_sup = FIELD_GET(MASK_EXT_IID_SUPPORT,
> +				     hba->mcq_capabilities);
> +
>  	return err;
>  }
>  
> @@ -7663,6 +7667,30 @@ static void ufshcd_temp_notif_probe(struct ufs_hba *hba, const u8 *desc_buf)
>  	}
>  }
>  
> +static void ufshcd_ext_iid_probe(struct ufs_hba *hba, u8 *desc_buf)
> +{
> +	struct ufs_dev_info *dev_info = &hba->dev_info;
> +	u32 ext_ufs_feature;
> +	u32 ext_iid_en = 0;

Given that scsi_host_alloc() uses kzalloc(), there is no need to explicitly set
the b_ext_iid_en to 0 in case of failure.

Thanks,
Mani

> +	int err;
> +
> +	/* Only UFS-4.0 and above may support EXT_IID */
> +	if (dev_info->wspecversion < 0x400)
> +		goto out;
> +
> +	ext_ufs_feature = ufs_get_ext_ufs_feature(hba, desc_buf);
> +	if (!(ext_ufs_feature & UFS_DEV_EXT_IID_SUP))
> +		goto out;
> +
> +	err = ufshcd_query_attr_retry(hba, UPIU_QUERY_OPCODE_READ_ATTR,
> +				      QUERY_ATTR_IDN_EXT_IID_EN, 0, 0, &ext_iid_en);
> +	if (err)
> +		dev_err(hba->dev, "failed reading bEXTIIDEn. err = %d\n", err);
> +
> +out:
> +	dev_info->b_ext_iid_en = ext_iid_en;
> +}
> +
>  void ufshcd_fixup_dev_quirks(struct ufs_hba *hba,
>  			     const struct ufs_dev_quirk *fixups)
>  {
> @@ -7761,6 +7789,9 @@ static int ufs_get_device_desc(struct ufs_hba *hba)
>  
>  	ufshcd_temp_notif_probe(hba, desc_buf);
>  
> +	if (hba->ext_iid_sup)
> +		ufshcd_ext_iid_probe(hba, desc_buf);
> +
>  	/*
>  	 * ufshcd_read_string_desc returns size of the string
>  	 * reset the error value
> diff --git a/include/ufs/ufs.h b/include/ufs/ufs.h
> index 1bba3fe..ba2a1d8 100644
> --- a/include/ufs/ufs.h
> +++ b/include/ufs/ufs.h
> @@ -165,6 +165,7 @@ enum attr_idn {
>  	QUERY_ATTR_IDN_AVAIL_WB_BUFF_SIZE       = 0x1D,
>  	QUERY_ATTR_IDN_WB_BUFF_LIFE_TIME_EST    = 0x1E,
>  	QUERY_ATTR_IDN_CURR_WB_BUFF_SIZE        = 0x1F,
> +	QUERY_ATTR_IDN_EXT_IID_EN		= 0x2A,
>  };
>  
>  /* Descriptor idn for Query requests */
> @@ -352,6 +353,7 @@ enum {
>  	UFS_DEV_EXT_TEMP_NOTIF		= BIT(6),
>  	UFS_DEV_HPB_SUPPORT		= BIT(7),
>  	UFS_DEV_WRITE_BOOSTER_SUP	= BIT(8),
> +	UFS_DEV_EXT_IID_SUP		= BIT(16),
>  };
>  #define UFS_DEV_HPB_SUPPORT_VERSION		0x310
>  
> @@ -601,6 +603,8 @@ struct ufs_dev_info {
>  
>  	bool	b_rpm_dev_flush_capable;
>  	u8	b_presrv_uspc_en;
> +	/* UFS EXT_IID Enable */
> +	bool	b_ext_iid_en;
>  };
>  
>  /*
> diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
> index 5cf81df..aec37cb9 100644
> --- a/include/ufs/ufshcd.h
> +++ b/include/ufs/ufshcd.h
> @@ -747,6 +747,7 @@ struct ufs_hba_monitor {
>   * @outstanding_lock: Protects @outstanding_reqs.
>   * @outstanding_reqs: Bits representing outstanding transfer requests
>   * @capabilities: UFS Controller Capabilities
> + * @mcq_capabilities: UFS Multi Circular Queue capabilities
>   * @nutrs: Transfer Request Queue depth supported by controller
>   * @nutmrs: Task Management Queue depth supported by controller
>   * @reserved_slot: Used to submit device commands. Protected by @dev_cmd.lock.
> @@ -830,6 +831,7 @@ struct ufs_hba_monitor {
>   *	device
>   * @complete_put: whether or not to call ufshcd_rpm_put() from inside
>   *	ufshcd_resume_complete()
> + * @ext_iid_sup: is EXT_IID is supported by UFSHC
>   */
>  struct ufs_hba {
>  	void __iomem *mmio_base;
> @@ -871,6 +873,7 @@ struct ufs_hba {
>  
>  	u32 capabilities;
>  	int nutrs;
> +	u32 mcq_capabilities;
>  	int nutmrs;
>  	u32 reserved_slot;
>  	u32 ufs_version;
> @@ -978,6 +981,7 @@ struct ufs_hba {
>  #endif
>  	u32 luns_avail;
>  	bool complete_put;
> +	bool ext_iid_sup;
>  };
>  
>  /* Returns true if clocks can be gated. Otherwise false */
> diff --git a/include/ufs/ufshci.h b/include/ufs/ufshci.h
> index f525566..4d4da06 100644
> --- a/include/ufs/ufshci.h
> +++ b/include/ufs/ufshci.h
> @@ -22,6 +22,7 @@ enum {
>  /* UFSHCI Registers */
>  enum {
>  	REG_CONTROLLER_CAPABILITIES		= 0x00,
> +	REG_MCQCAP				= 0x04,
>  	REG_UFS_VERSION				= 0x08,
>  	REG_CONTROLLER_DEV_ID			= 0x10,
>  	REG_CONTROLLER_PROD_ID			= 0x14,
> @@ -68,6 +69,12 @@ enum {
>  	MASK_OUT_OF_ORDER_DATA_DELIVERY_SUPPORT	= 0x02000000,
>  	MASK_UIC_DME_TEST_MODE_SUPPORT		= 0x04000000,
>  	MASK_CRYPTO_SUPPORT			= 0x10000000,
> +	MASK_MCQ_SUPPORT			= 0x40000000,
> +};
> +
> +/* MCQ capability mask */
> +enum {
> +	MASK_EXT_IID_SUPPORT = 0x00000400,
>  };
>  
>  #define UFS_MASK(mask, offset)		((mask) << (offset))
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

[PATCH v5 03/16] ufs: core: Introduce Multi-circular queue capability

[Asutosh Das]

Add support to check for MCQ capability in the UFSHC.
Add a module parameter to disable MCQ if needed.

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
---
 drivers/ufs/core/ufshcd.c | 31 +++++++++++++++++++++++++++++++
 include/ufs/ufshcd.h      |  2 ++
 2 files changed, 33 insertions(+)

diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 66b797f..08be8ad 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -89,6 +89,33 @@
 /* Polling time to wait for fDeviceInit */
 #define FDEVICEINIT_COMPL_TIMEOUT 1500 /* millisecs */
 
+/* UFSHC 4.0 compliant HC support this mode, refer param_set_mcq_mode() */
+static bool use_mcq_mode = true;
+
+static inline bool is_mcq_supported(struct ufs_hba *hba)
+{
+	return hba->mcq_sup && use_mcq_mode;
+}
+
+static int param_set_mcq_mode(const char *val, const struct kernel_param *kp)
+{
+	int ret;
+
+	ret = param_set_bool(val, kp);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static const struct kernel_param_ops mcq_mode_ops = {
+	.set = param_set_mcq_mode,
+	.get = param_get_bool,
+};
+
+module_param_cb(use_mcq_mode, &mcq_mode_ops, &use_mcq_mode, 0644);
+MODULE_PARM_DESC(mcq_mode, "Control MCQ mode for UFSHCI 4.0 controllers");
+
 #define ufshcd_toggle_vreg(_dev, _vreg, _on)				\
 	({                                                              \
 		int _ret;                                               \
@@ -2258,6 +2285,10 @@ static inline int ufshcd_hba_capabilities(struct ufs_hba *hba)
 	if (err)
 		dev_err(hba->dev, "crypto setup failed\n");
 
+	hba->mcq_sup = FIELD_GET(MASK_MCQ_SUPPORT, hba->capabilities);
+	if (!hba->mcq_sup)
+		return err;
+
 	hba->mcq_capabilities = ufshcd_readl(hba, REG_MCQCAP);
 	hba->ext_iid_sup = FIELD_GET(MASK_EXT_IID_SUPPORT,
 				     hba->mcq_capabilities);
diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
index aec37cb9..70c0f9f 100644
--- a/include/ufs/ufshcd.h
+++ b/include/ufs/ufshcd.h
@@ -832,6 +832,7 @@ struct ufs_hba_monitor {
  * @complete_put: whether or not to call ufshcd_rpm_put() from inside
  *	ufshcd_resume_complete()
  * @ext_iid_sup: is EXT_IID is supported by UFSHC
+ * @mcq_sup: is mcq supported by UFSHC
  */
 struct ufs_hba {
 	void __iomem *mmio_base;
@@ -982,6 +983,7 @@ struct ufs_hba {
 	u32 luns_avail;
 	bool complete_put;
 	bool ext_iid_sup;
+	bool mcq_sup;
 };
 
 /* Returns true if clocks can be gated. Otherwise false */
-- 
2.7.4

Re: [PATCH v5 03/16] ufs: core: Introduce Multi-circular queue capability

[Bart Van Assche]

On 11/22/22 20:10, Asutosh Das wrote:
> +module_param_cb(use_mcq_mode, &mcq_mode_ops, &use_mcq_mode, 0644);
> +MODULE_PARM_DESC(mcq_mode, "Control MCQ mode for UFSHCI 4.0 controllers");

Please make this description more detailed. The following information 
should be added:
* 0 disables MCQ.
* 1 enables MCQ.
* MCQ is enabled by default.

Once that information has been added, feel free to add:

Reviewed-by: Bart Van Assche <bvanassche@acm.org>

Re: [PATCH v5 03/16] ufs: core: Introduce Multi-circular queue capability

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:16PM -0800, Asutosh Das wrote:
> Add support to check for MCQ capability in the UFSHC.
> Add a module parameter to disable MCQ if needed.
> 
> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>

Couple of nitpicks below, with those addressed:

Reviewed-by: Manivannan Sadhasivam <mani@kernel.org>

> ---
>  drivers/ufs/core/ufshcd.c | 31 +++++++++++++++++++++++++++++++
>  include/ufs/ufshcd.h      |  2 ++
>  2 files changed, 33 insertions(+)
> 
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 66b797f..08be8ad 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -89,6 +89,33 @@
>  /* Polling time to wait for fDeviceInit */
>  #define FDEVICEINIT_COMPL_TIMEOUT 1500 /* millisecs */
>  
> +/* UFSHC 4.0 compliant HC support this mode, refer param_set_mcq_mode() */
> +static bool use_mcq_mode = true;
> +
> +static inline bool is_mcq_supported(struct ufs_hba *hba)

Please get rid of inline keyword and let the compiler handle it.

> +{
> +	return hba->mcq_sup && use_mcq_mode;
> +}
> +
> +static int param_set_mcq_mode(const char *val, const struct kernel_param *kp)
> +{
> +	int ret;
> +
> +	ret = param_set_bool(val, kp);
> +	if (ret)
> +		return ret;
> +
> +	return 0;
> +}
> +
> +static const struct kernel_param_ops mcq_mode_ops = {
> +	.set = param_set_mcq_mode,
> +	.get = param_get_bool,
> +};
> +
> +module_param_cb(use_mcq_mode, &mcq_mode_ops, &use_mcq_mode, 0644);
> +MODULE_PARM_DESC(mcq_mode, "Control MCQ mode for UFSHCI 4.0 controllers");

Is it ok to mention only 4.0? What about future revisions?

Thanks,
Mani
> +
>  #define ufshcd_toggle_vreg(_dev, _vreg, _on)				\
>  	({                                                              \
>  		int _ret;                                               \
> @@ -2258,6 +2285,10 @@ static inline int ufshcd_hba_capabilities(struct ufs_hba *hba)
>  	if (err)
>  		dev_err(hba->dev, "crypto setup failed\n");
>  
> +	hba->mcq_sup = FIELD_GET(MASK_MCQ_SUPPORT, hba->capabilities);
> +	if (!hba->mcq_sup)
> +		return err;
> +
>  	hba->mcq_capabilities = ufshcd_readl(hba, REG_MCQCAP);
>  	hba->ext_iid_sup = FIELD_GET(MASK_EXT_IID_SUPPORT,
>  				     hba->mcq_capabilities);
> diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
> index aec37cb9..70c0f9f 100644
> --- a/include/ufs/ufshcd.h
> +++ b/include/ufs/ufshcd.h
> @@ -832,6 +832,7 @@ struct ufs_hba_monitor {
>   * @complete_put: whether or not to call ufshcd_rpm_put() from inside
>   *	ufshcd_resume_complete()
>   * @ext_iid_sup: is EXT_IID is supported by UFSHC
> + * @mcq_sup: is mcq supported by UFSHC
>   */
>  struct ufs_hba {
>  	void __iomem *mmio_base;
> @@ -982,6 +983,7 @@ struct ufs_hba {
>  	u32 luns_avail;
>  	bool complete_put;
>  	bool ext_iid_sup;
> +	bool mcq_sup;
>  };
>  
>  /* Returns true if clocks can be gated. Otherwise false */
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

[PATCH v5 04/16] ufs: core: Defer adding host to scsi if mcq is supported

[Asutosh Das]

If MCQ support is present, enabling it after MCQ support
has been configured would require reallocating tags and memory.
It would also free up the already allocated memory in
Single Doorbell Mode. So defer invoking scsi_add_host() until
MCQ is configured.

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
---
 drivers/ufs/core/ufshcd.c | 19 +++++++++++++++----
 1 file changed, 15 insertions(+), 4 deletions(-)

diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 08be8ad..42c49ce 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -8208,6 +8208,7 @@ static int ufshcd_probe_hba(struct ufs_hba *hba, bool init_dev_params)
 	int ret;
 	unsigned long flags;
 	ktime_t start = ktime_get();
+	struct Scsi_Host *host = hba->host;
 
 	hba->ufshcd_state = UFSHCD_STATE_RESET;
 
@@ -8242,6 +8243,14 @@ static int ufshcd_probe_hba(struct ufs_hba *hba, bool init_dev_params)
 		ret = ufshcd_device_params_init(hba);
 		if (ret)
 			goto out;
+
+		if (is_mcq_supported(hba)) {
+			ret = scsi_add_host(host, hba->dev);
+			if (ret) {
+				dev_err(hba->dev, "scsi_add_host failed\n");
+				goto out;
+			}
+		}
 	}
 
 	ufshcd_tune_unipro_params(hba);
@@ -9838,10 +9847,12 @@ int ufshcd_init(struct ufs_hba *hba, void __iomem *mmio_base, unsigned int irq)
 		hba->is_irq_enabled = true;
 	}
 
-	err = scsi_add_host(host, hba->dev);
-	if (err) {
-		dev_err(hba->dev, "scsi_add_host failed\n");
-		goto out_disable;
+	if (!is_mcq_supported(hba)) {
+		err = scsi_add_host(host, hba->dev);
+		if (err) {
+			dev_err(hba->dev, "scsi_add_host failed\n");
+			goto out_disable;
+		}
 	}
 
 	hba->tmf_tag_set = (struct blk_mq_tag_set) {
-- 
2.7.4

Re: [PATCH v5 04/16] ufs: core: Defer adding host to scsi if mcq is supported

[Bart Van Assche]

On 11/22/22 20:10, Asutosh Das wrote:
> If MCQ support is present, enabling it after MCQ support
> has been configured would require reallocating tags and memory.
> It would also free up the already allocated memory in
> Single Doorbell Mode. So defer invoking scsi_add_host() until
> MCQ is configured.

Reviewed-by: Bart Van Assche <bvanassche@acm.org>

Re: [PATCH v5 04/16] ufs: core: Defer adding host to scsi if mcq is supported

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:17PM -0800, Asutosh Das wrote:
> If MCQ support is present, enabling it after MCQ support
> has been configured would require reallocating tags and memory.
> It would also free up the already allocated memory in
> Single Doorbell Mode. So defer invoking scsi_add_host() until
> MCQ is configured.

Why cannot we do it for non MCQ case as well?

Thanks,
Mani

> 
> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
> ---
>  drivers/ufs/core/ufshcd.c | 19 +++++++++++++++----
>  1 file changed, 15 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 08be8ad..42c49ce 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -8208,6 +8208,7 @@ static int ufshcd_probe_hba(struct ufs_hba *hba, bool init_dev_params)
>  	int ret;
>  	unsigned long flags;
>  	ktime_t start = ktime_get();
> +	struct Scsi_Host *host = hba->host;
>  
>  	hba->ufshcd_state = UFSHCD_STATE_RESET;
>  
> @@ -8242,6 +8243,14 @@ static int ufshcd_probe_hba(struct ufs_hba *hba, bool init_dev_params)
>  		ret = ufshcd_device_params_init(hba);
>  		if (ret)
>  			goto out;
> +
> +		if (is_mcq_supported(hba)) {
> +			ret = scsi_add_host(host, hba->dev);
> +			if (ret) {
> +				dev_err(hba->dev, "scsi_add_host failed\n");
> +				goto out;
> +			}
> +		}
>  	}
>  
>  	ufshcd_tune_unipro_params(hba);
> @@ -9838,10 +9847,12 @@ int ufshcd_init(struct ufs_hba *hba, void __iomem *mmio_base, unsigned int irq)
>  		hba->is_irq_enabled = true;
>  	}
>  
> -	err = scsi_add_host(host, hba->dev);
> -	if (err) {
> -		dev_err(hba->dev, "scsi_add_host failed\n");
> -		goto out_disable;
> +	if (!is_mcq_supported(hba)) {
> +		err = scsi_add_host(host, hba->dev);
> +		if (err) {
> +			dev_err(hba->dev, "scsi_add_host failed\n");
> +			goto out_disable;
> +		}
>  	}
>  
>  	hba->tmf_tag_set = (struct blk_mq_tag_set) {
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

Re: [PATCH v5 04/16] ufs: core: Defer adding host to scsi if mcq is supported

[Asutosh Das]

On Mon, Nov 28 2022 at 06:42 -0800, Manivannan Sadhasivam wrote:
>On Tue, Nov 22, 2022 at 08:10:17PM -0800, Asutosh Das wrote:
>> If MCQ support is present, enabling it after MCQ support
>> has been configured would require reallocating tags and memory.
>> It would also free up the already allocated memory in
>> Single Doorbell Mode. So defer invoking scsi_add_host() until
>> MCQ is configured.
>
>Why cannot we do it for non MCQ case as well?
>
Hello Mani,
Thanks for taking a look.

I don't think there was any specific reason to defer it for SDB mode.

-asd

>-- 
>மணிவண்ணன் சதாசிவம்

[PATCH v5 05/16] ufs: core: mcq: Add Multi Circular Queue support

[Asutosh Das]

Add support for multi-circular queue (MCQ) which has been added
in UFSHC v4.0 standard in addition to the Single Doorbell mode.
The MCQ mode supports multiple submission and completion queues.
Add support to configure the number of queues.

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
---
 drivers/ufs/core/Makefile      |   2 +-
 drivers/ufs/core/ufs-mcq.c     | 125 +++++++++++++++++++++++++++++++++++++++++
 drivers/ufs/core/ufshcd-priv.h |   1 +
 drivers/ufs/core/ufshcd.c      |  12 ++++
 include/ufs/ufshcd.h           |   4 ++
 5 files changed, 143 insertions(+), 1 deletion(-)
 create mode 100644 drivers/ufs/core/ufs-mcq.c

diff --git a/drivers/ufs/core/Makefile b/drivers/ufs/core/Makefile
index 62f38c5..4d02e0f 100644
--- a/drivers/ufs/core/Makefile
+++ b/drivers/ufs/core/Makefile
@@ -1,7 +1,7 @@
 # SPDX-License-Identifier: GPL-2.0
 
 obj-$(CONFIG_SCSI_UFSHCD)		+= ufshcd-core.o
-ufshcd-core-y				+= ufshcd.o ufs-sysfs.o
+ufshcd-core-y				+= ufshcd.o ufs-sysfs.o ufs-mcq.o
 ufshcd-core-$(CONFIG_DEBUG_FS)		+= ufs-debugfs.o
 ufshcd-core-$(CONFIG_SCSI_UFS_BSG)	+= ufs_bsg.o
 ufshcd-core-$(CONFIG_SCSI_UFS_CRYPTO)	+= ufshcd-crypto.o
diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
new file mode 100644
index 0000000..3818f45
--- /dev/null
+++ b/drivers/ufs/core/ufs-mcq.c
@@ -0,0 +1,125 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2022 Qualcomm Innovation Center. All rights reserved.
+ *
+ * Authors:
+ *	Asutosh Das <quic_asutoshd@quicinc.com>
+ *	Can Guo <quic_cang@quicinc.com>
+ */
+
+#include <asm/unaligned.h>
+#include <linux/dma-mapping.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
+#include "ufshcd-priv.h"
+
+#define UFS_MCQ_MIN_RW_QUEUES 2
+#define UFS_MCQ_MIN_READ_QUEUES 0
+#define UFS_MCQ_NUM_DEV_CMD_QUEUES 1
+#define UFS_MCQ_MIN_POLL_QUEUES 0
+
+
+static int rw_queue_count_set(const char *val, const struct kernel_param *kp)
+{
+	return param_set_uint_minmax(val, kp, UFS_MCQ_MIN_RW_QUEUES,
+				     num_possible_cpus());
+}
+
+static const struct kernel_param_ops rw_queue_count_ops = {
+	.set = rw_queue_count_set,
+	.get = param_get_uint,
+};
+
+static unsigned int rw_queues;
+module_param_cb(rw_queues, &rw_queue_count_ops, &rw_queues, 0644);
+MODULE_PARM_DESC(rw_queues,
+		 "Number of interrupt driven I/O queues used for rw. Default value is nr_cpus");
+
+static int read_queue_count_set(const char *val, const struct kernel_param *kp)
+{
+	return param_set_uint_minmax(val, kp, UFS_MCQ_MIN_READ_QUEUES,
+				     num_possible_cpus());
+}
+
+static const struct kernel_param_ops read_queue_count_ops = {
+	.set = read_queue_count_set,
+	.get = param_get_uint,
+};
+
+static unsigned int read_queues;
+module_param_cb(read_queues, &read_queue_count_ops, &read_queues, 0644);
+MODULE_PARM_DESC(read_queues,
+		 "Number of interrupt driven read queues used for read. Default value is 0");
+
+static int poll_queue_count_set(const char *val, const struct kernel_param *kp)
+{
+	return param_set_uint_minmax(val, kp, UFS_MCQ_MIN_POLL_QUEUES,
+				     num_possible_cpus());
+}
+
+static const struct kernel_param_ops poll_queue_count_ops = {
+	.set = poll_queue_count_set,
+	.get = param_get_uint,
+};
+
+static unsigned int poll_queues = 1;
+module_param_cb(poll_queues, &poll_queue_count_ops, &poll_queues, 0644);
+MODULE_PARM_DESC(poll_queues,
+		 "Number of poll queues used for r/w. Default value is 1");
+
+static int ufshcd_mcq_config_nr_queues(struct ufs_hba *hba)
+{
+	int i;
+	u32 hba_maxq, rem, tot_queues;
+	struct Scsi_Host *host = hba->host;
+
+	hba_maxq = FIELD_GET(GENMASK(7, 0), hba->mcq_capabilities);
+
+	tot_queues = UFS_MCQ_NUM_DEV_CMD_QUEUES + read_queues + poll_queues +
+			rw_queues;
+
+	if (hba_maxq < tot_queues) {
+		dev_err(hba->dev, "Total queues (%d) exceeds HC capacity (%d)\n",
+			tot_queues, hba_maxq);
+		return -EOPNOTSUPP;
+	}
+
+	rem = hba_maxq - UFS_MCQ_NUM_DEV_CMD_QUEUES;
+
+	if (rw_queues) {
+		hba->nr_queues[HCTX_TYPE_DEFAULT] = rw_queues;
+		rem -= hba->nr_queues[HCTX_TYPE_DEFAULT];
+	} else {
+		rw_queues = num_possible_cpus();
+	}
+
+	if (poll_queues) {
+		hba->nr_queues[HCTX_TYPE_POLL] = poll_queues;
+		rem -= hba->nr_queues[HCTX_TYPE_POLL];
+	}
+
+	if (read_queues) {
+		hba->nr_queues[HCTX_TYPE_READ] = read_queues;
+		rem -= hba->nr_queues[HCTX_TYPE_READ];
+	}
+
+	if (!hba->nr_queues[HCTX_TYPE_DEFAULT])
+		hba->nr_queues[HCTX_TYPE_DEFAULT] = min3(rem, rw_queues,
+							 num_possible_cpus());
+
+	for (i = 0; i < HCTX_MAX_TYPES; i++)
+		host->nr_hw_queues += hba->nr_queues[i];
+
+	hba->nr_hw_queues = host->nr_hw_queues + UFS_MCQ_NUM_DEV_CMD_QUEUES;
+	return 0;
+}
+
+int ufshcd_mcq_init(struct ufs_hba *hba)
+{
+	int ret;
+
+	ret = ufshcd_mcq_config_nr_queues(hba);
+
+	return ret;
+}
+
diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
index a9e8e1f..9368ba2 100644
--- a/drivers/ufs/core/ufshcd-priv.h
+++ b/drivers/ufs/core/ufshcd-priv.h
@@ -61,6 +61,7 @@ int ufshcd_query_attr(struct ufs_hba *hba, enum query_opcode opcode,
 int ufshcd_query_flag(struct ufs_hba *hba, enum query_opcode opcode,
 	enum flag_idn idn, u8 index, bool *flag_res);
 void ufshcd_auto_hibern8_update(struct ufs_hba *hba, u32 ahit);
+int ufshcd_mcq_init(struct ufs_hba *hba);
 
 #define SD_ASCII_STD true
 #define SD_RAW false
diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 42c49ce..0c4cd8f 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -8196,6 +8196,11 @@ static int ufshcd_add_lus(struct ufs_hba *hba)
 	return ret;
 }
 
+static int ufshcd_alloc_mcq(struct ufs_hba *hba)
+{
+	return ufshcd_mcq_init(hba);
+}
+
 /**
  * ufshcd_probe_hba - probe hba to detect device and initialize it
  * @hba: per-adapter instance
@@ -8245,6 +8250,13 @@ static int ufshcd_probe_hba(struct ufs_hba *hba, bool init_dev_params)
 			goto out;
 
 		if (is_mcq_supported(hba)) {
+			ret = ufshcd_alloc_mcq(hba);
+			if (ret) {
+				/* Continue with SDB mode */
+				use_mcq_mode = false;
+				dev_err(hba->dev, "MCQ mode is disabled, err=%d\n",
+					 ret);
+			}
 			ret = scsi_add_host(host, hba->dev);
 			if (ret) {
 				dev_err(hba->dev, "scsi_add_host failed\n");
diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
index 70c0f9f..dee0b37 100644
--- a/include/ufs/ufshcd.h
+++ b/include/ufs/ufshcd.h
@@ -833,6 +833,8 @@ struct ufs_hba_monitor {
  *	ufshcd_resume_complete()
  * @ext_iid_sup: is EXT_IID is supported by UFSHC
  * @mcq_sup: is mcq supported by UFSHC
+ * @nr_hw_queues: number of hardware queues configured
+ * @nr_queues: number of Queues of different queue types
  */
 struct ufs_hba {
 	void __iomem *mmio_base;
@@ -984,6 +986,8 @@ struct ufs_hba {
 	bool complete_put;
 	bool ext_iid_sup;
 	bool mcq_sup;
+	unsigned int nr_hw_queues;
+	unsigned int nr_queues[HCTX_MAX_TYPES];
 };
 
 /* Returns true if clocks can be gated. Otherwise false */
-- 
2.7.4

Re: [PATCH v5 05/16] ufs: core: mcq: Add Multi Circular Queue support

[Bart Van Assche]

On 11/22/22 20:10, Asutosh Das wrote:
> Add support for multi-circular queue (MCQ) which has been added
> in UFSHC v4.0 standard in addition to the Single Doorbell mode.
> The MCQ mode supports multiple submission and completion queues.
> Add support to configure the number of queues.

Reviewed-by: Bart Van Assche <bvanassche@acm.org>

Re: [PATCH v5 05/16] ufs: core: mcq: Add Multi Circular Queue support

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:18PM -0800, Asutosh Das wrote:
> Add support for multi-circular queue (MCQ) which has been added
> in UFSHC v4.0 standard in addition to the Single Doorbell mode.
> The MCQ mode supports multiple submission and completion queues.
> Add support to configure the number of queues.
> 

The patch subject is pretty opaque. Please use something like "Add initial
Multi Circular Queue support" or something similar to specify that this patch
only adds support for configuring the queues and not the full MCQ support.

Also, this patch adds the module params for queues, so that should be mentioned
in the description.

> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
> ---
>  drivers/ufs/core/Makefile      |   2 +-
>  drivers/ufs/core/ufs-mcq.c     | 125 +++++++++++++++++++++++++++++++++++++++++
>  drivers/ufs/core/ufshcd-priv.h |   1 +
>  drivers/ufs/core/ufshcd.c      |  12 ++++
>  include/ufs/ufshcd.h           |   4 ++
>  5 files changed, 143 insertions(+), 1 deletion(-)
>  create mode 100644 drivers/ufs/core/ufs-mcq.c
> 
> diff --git a/drivers/ufs/core/Makefile b/drivers/ufs/core/Makefile
> index 62f38c5..4d02e0f 100644
> --- a/drivers/ufs/core/Makefile
> +++ b/drivers/ufs/core/Makefile
> @@ -1,7 +1,7 @@
>  # SPDX-License-Identifier: GPL-2.0
>  
>  obj-$(CONFIG_SCSI_UFSHCD)		+= ufshcd-core.o
> -ufshcd-core-y				+= ufshcd.o ufs-sysfs.o
> +ufshcd-core-y				+= ufshcd.o ufs-sysfs.o ufs-mcq.o
>  ufshcd-core-$(CONFIG_DEBUG_FS)		+= ufs-debugfs.o
>  ufshcd-core-$(CONFIG_SCSI_UFS_BSG)	+= ufs_bsg.o
>  ufshcd-core-$(CONFIG_SCSI_UFS_CRYPTO)	+= ufshcd-crypto.o
> diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
> new file mode 100644
> index 0000000..3818f45
> --- /dev/null
> +++ b/drivers/ufs/core/ufs-mcq.c
> @@ -0,0 +1,125 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +/*
> + * Copyright (c) 2022 Qualcomm Innovation Center. All rights reserved.
> + *
> + * Authors:
> + *	Asutosh Das <quic_asutoshd@quicinc.com>
> + *	Can Guo <quic_cang@quicinc.com>
> + */
> +
> +#include <asm/unaligned.h>
> +#include <linux/dma-mapping.h>
> +#include <linux/module.h>
> +#include <linux/platform_device.h>
> +#include "ufshcd-priv.h"
> +
> +#define UFS_MCQ_MIN_RW_QUEUES 2
> +#define UFS_MCQ_MIN_READ_QUEUES 0
> +#define UFS_MCQ_NUM_DEV_CMD_QUEUES 1
> +#define UFS_MCQ_MIN_POLL_QUEUES 0
> +

Remove extra new line

> +
> +static int rw_queue_count_set(const char *val, const struct kernel_param *kp)
> +{
> +	return param_set_uint_minmax(val, kp, UFS_MCQ_MIN_RW_QUEUES,
> +				     num_possible_cpus());
> +}
> +
> +static const struct kernel_param_ops rw_queue_count_ops = {
> +	.set = rw_queue_count_set,
> +	.get = param_get_uint,
> +};
> +
> +static unsigned int rw_queues;
> +module_param_cb(rw_queues, &rw_queue_count_ops, &rw_queues, 0644);
> +MODULE_PARM_DESC(rw_queues,
> +		 "Number of interrupt driven I/O queues used for rw. Default value is nr_cpus");
> +
> +static int read_queue_count_set(const char *val, const struct kernel_param *kp)
> +{
> +	return param_set_uint_minmax(val, kp, UFS_MCQ_MIN_READ_QUEUES,
> +				     num_possible_cpus());
> +}
> +
> +static const struct kernel_param_ops read_queue_count_ops = {
> +	.set = read_queue_count_set,
> +	.get = param_get_uint,
> +};
> +
> +static unsigned int read_queues;
> +module_param_cb(read_queues, &read_queue_count_ops, &read_queues, 0644);
> +MODULE_PARM_DESC(read_queues,
> +		 "Number of interrupt driven read queues used for read. Default value is 0");
> +
> +static int poll_queue_count_set(const char *val, const struct kernel_param *kp)
> +{
> +	return param_set_uint_minmax(val, kp, UFS_MCQ_MIN_POLL_QUEUES,
> +				     num_possible_cpus());
> +}
> +
> +static const struct kernel_param_ops poll_queue_count_ops = {
> +	.set = poll_queue_count_set,
> +	.get = param_get_uint,
> +};
> +
> +static unsigned int poll_queues = 1;
> +module_param_cb(poll_queues, &poll_queue_count_ops, &poll_queues, 0644);
> +MODULE_PARM_DESC(poll_queues,
> +		 "Number of poll queues used for r/w. Default value is 1");
> +
> +static int ufshcd_mcq_config_nr_queues(struct ufs_hba *hba)
> +{
> +	int i;
> +	u32 hba_maxq, rem, tot_queues;
> +	struct Scsi_Host *host = hba->host;
> +
> +	hba_maxq = FIELD_GET(GENMASK(7, 0), hba->mcq_capabilities);

It'd be good to add a definition for GENMASK(7, 0).

> +
> +	tot_queues = UFS_MCQ_NUM_DEV_CMD_QUEUES + read_queues + poll_queues +
> +			rw_queues;
> +
> +	if (hba_maxq < tot_queues) {
> +		dev_err(hba->dev, "Total queues (%d) exceeds HC capacity (%d)\n",
> +			tot_queues, hba_maxq);
> +		return -EOPNOTSUPP;
> +	}
> +
> +	rem = hba_maxq - UFS_MCQ_NUM_DEV_CMD_QUEUES;
> +
> +	if (rw_queues) {
> +		hba->nr_queues[HCTX_TYPE_DEFAULT] = rw_queues;
> +		rem -= hba->nr_queues[HCTX_TYPE_DEFAULT];
> +	} else {
> +		rw_queues = num_possible_cpus();
> +	}
> +
> +	if (poll_queues) {
> +		hba->nr_queues[HCTX_TYPE_POLL] = poll_queues;
> +		rem -= hba->nr_queues[HCTX_TYPE_POLL];
> +	}
> +
> +	if (read_queues) {
> +		hba->nr_queues[HCTX_TYPE_READ] = read_queues;
> +		rem -= hba->nr_queues[HCTX_TYPE_READ];
> +	}
> +
> +	if (!hba->nr_queues[HCTX_TYPE_DEFAULT])
> +		hba->nr_queues[HCTX_TYPE_DEFAULT] = min3(rem, rw_queues,
> +							 num_possible_cpus());
> +
> +	for (i = 0; i < HCTX_MAX_TYPES; i++)
> +		host->nr_hw_queues += hba->nr_queues[i];
> +
> +	hba->nr_hw_queues = host->nr_hw_queues + UFS_MCQ_NUM_DEV_CMD_QUEUES;
> +	return 0;
> +}
> +
> +int ufshcd_mcq_init(struct ufs_hba *hba)
> +{
> +	int ret;
> +
> +	ret = ufshcd_mcq_config_nr_queues(hba);
> +
> +	return ret;
> +}
> +
> diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
> index a9e8e1f..9368ba2 100644
> --- a/drivers/ufs/core/ufshcd-priv.h
> +++ b/drivers/ufs/core/ufshcd-priv.h
> @@ -61,6 +61,7 @@ int ufshcd_query_attr(struct ufs_hba *hba, enum query_opcode opcode,
>  int ufshcd_query_flag(struct ufs_hba *hba, enum query_opcode opcode,
>  	enum flag_idn idn, u8 index, bool *flag_res);
>  void ufshcd_auto_hibern8_update(struct ufs_hba *hba, u32 ahit);
> +int ufshcd_mcq_init(struct ufs_hba *hba);
>  
>  #define SD_ASCII_STD true
>  #define SD_RAW false
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 42c49ce..0c4cd8f 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -8196,6 +8196,11 @@ static int ufshcd_add_lus(struct ufs_hba *hba)
>  	return ret;
>  }
>  
> +static int ufshcd_alloc_mcq(struct ufs_hba *hba)
> +{
> +	return ufshcd_mcq_init(hba);
> +}
> +
>  /**
>   * ufshcd_probe_hba - probe hba to detect device and initialize it
>   * @hba: per-adapter instance
> @@ -8245,6 +8250,13 @@ static int ufshcd_probe_hba(struct ufs_hba *hba, bool init_dev_params)
>  			goto out;
>  
>  		if (is_mcq_supported(hba)) {
> +			ret = ufshcd_alloc_mcq(hba);
> +			if (ret) {
> +				/* Continue with SDB mode */
> +				use_mcq_mode = false;
> +				dev_err(hba->dev, "MCQ mode is disabled, err=%d\n",
> +					 ret);
> +			}
>  			ret = scsi_add_host(host, hba->dev);
>  			if (ret) {
>  				dev_err(hba->dev, "scsi_add_host failed\n");
> diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
> index 70c0f9f..dee0b37 100644
> --- a/include/ufs/ufshcd.h
> +++ b/include/ufs/ufshcd.h
> @@ -833,6 +833,8 @@ struct ufs_hba_monitor {
>   *	ufshcd_resume_complete()
>   * @ext_iid_sup: is EXT_IID is supported by UFSHC
>   * @mcq_sup: is mcq supported by UFSHC
> + * @nr_hw_queues: number of hardware queues configured
> + * @nr_queues: number of Queues of different queue types
>   */
>  struct ufs_hba {
>  	void __iomem *mmio_base;
> @@ -984,6 +986,8 @@ struct ufs_hba {
>  	bool complete_put;
>  	bool ext_iid_sup;
>  	bool mcq_sup;
> +	unsigned int nr_hw_queues;
> +	unsigned int nr_queues[HCTX_MAX_TYPES];

Can these two members added before bool types to avoid any holes?

Thanks,
Mani

>  };
>  
>  /* Returns true if clocks can be gated. Otherwise false */
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

[PATCH v5 06/16] ufs: core: mcq: Configure resource regions

[Asutosh Das]

Define the mcq resources and add support to ioremap
the resource regions.

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
---
 drivers/ufs/core/ufs-mcq.c | 104 +++++++++++++++++++++++++++++++++++++++++++++
 include/ufs/ufshcd.h       |  28 ++++++++++++
 2 files changed, 132 insertions(+)

diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
index 3818f45..4aaa6aa 100644
--- a/drivers/ufs/core/ufs-mcq.c
+++ b/drivers/ufs/core/ufs-mcq.c
@@ -18,6 +18,11 @@
 #define UFS_MCQ_NUM_DEV_CMD_QUEUES 1
 #define UFS_MCQ_MIN_POLL_QUEUES 0
 
+#define MCQ_QCFGPTR_MASK	GENMASK(7, 0)
+#define MCQ_QCFGPTR_UNIT	0x200
+#define MCQ_SQATTR_OFFSET(c) \
+	((((c) >> 16) & MCQ_QCFGPTR_MASK) * MCQ_QCFGPTR_UNIT)
+#define MCQ_QCFG_SIZE	0x40
 
 static int rw_queue_count_set(const char *val, const struct kernel_param *kp)
 {
@@ -67,6 +72,102 @@ module_param_cb(poll_queues, &poll_queue_count_ops, &poll_queues, 0644);
 MODULE_PARM_DESC(poll_queues,
 		 "Number of poll queues used for r/w. Default value is 1");
 
+/* Resources */
+static const struct ufshcd_res_info ufs_res_info[RES_MAX] = {
+	{.name = "ufs_mem",},
+	{.name = "mcq",},
+	/* Submission Queue DAO */
+	{.name = "mcq_sqd",},
+	/* Submission Queue Interrupt Status */
+	{.name = "mcq_sqis",},
+	/* Completion Queue DAO */
+	{.name = "mcq_cqd",},
+	/* Completion Queue Interrupt Status */
+	{.name = "mcq_cqis",},
+	/* MCQ vendor specific */
+	{.name = "mcq_vs",},
+};
+
+static int ufshcd_mcq_config_resource(struct ufs_hba *hba)
+{
+	struct platform_device *pdev = to_platform_device(hba->dev);
+	struct ufshcd_res_info *res;
+	struct resource *res_mem, *res_mcq;
+	int i, ret = 0;
+
+	memcpy(hba->res, ufs_res_info, sizeof(ufs_res_info));
+
+	for (i = 0; i < RES_MAX; i++) {
+		res = &hba->res[i];
+		res->resource = platform_get_resource_byname(pdev,
+							     IORESOURCE_MEM,
+							     res->name);
+		if (!res->resource) {
+			dev_info(hba->dev, "Resource %s not provided\n", res->name);
+			if (i == RES_UFS)
+				return -ENOMEM;
+			continue;
+		} else if (i == RES_UFS) {
+			res_mem = res->resource;
+			res->base = hba->mmio_base;
+			continue;
+		}
+
+		res->base = devm_ioremap_resource(hba->dev, res->resource);
+		if (IS_ERR(res->base)) {
+			dev_err(hba->dev, "Failed to map res %s, err=%d\n",
+					 res->name, (int)PTR_ERR(res->base));
+			res->base = NULL;
+			ret = PTR_ERR(res->base);
+			return ret;
+		}
+	}
+
+	/* MCQ resource provided in DT */
+	res = &hba->res[RES_MCQ];
+	/* Bail if MCQ resource is provided */
+	if (res->base)
+		goto out;
+
+	/* Explicitly allocate MCQ resource from ufs_mem */
+	res_mcq = devm_kzalloc(hba->dev, sizeof(*res_mcq), GFP_KERNEL);
+	if (!res_mcq) {
+		dev_err(hba->dev, "Failed to allocate MCQ resource\n");
+		return ret;
+	}
+
+	res_mcq->start = res_mem->start +
+			 MCQ_SQATTR_OFFSET(hba->mcq_capabilities);
+	res_mcq->end = res_mcq->start + hba->nr_hw_queues * MCQ_QCFG_SIZE - 1;
+	res_mcq->flags = res_mem->flags;
+	res_mcq->name = "mcq";
+
+	ret = insert_resource(&iomem_resource, res_mcq);
+	if (ret) {
+		dev_err(hba->dev, "Failed to insert MCQ resource, err=%d\n",
+			ret);
+		goto insert_res_err;
+	}
+
+	res->base = devm_ioremap_resource(hba->dev, res_mcq);
+	if (IS_ERR(res->base)) {
+		dev_err(hba->dev, "MCQ registers mapping failed, err=%d\n",
+			(int)PTR_ERR(res->base));
+		ret = PTR_ERR(res->base);
+		goto ioremap_err;
+	}
+
+out:
+	hba->mcq_base = res->base;
+	return 0;
+ioremap_err:
+	res->base = NULL;
+	remove_resource(res_mcq);
+insert_res_err:
+	devm_kfree(hba->dev, res_mcq);
+	return ret;
+}
+
 static int ufshcd_mcq_config_nr_queues(struct ufs_hba *hba)
 {
 	int i;
@@ -119,7 +220,10 @@ int ufshcd_mcq_init(struct ufs_hba *hba)
 	int ret;
 
 	ret = ufshcd_mcq_config_nr_queues(hba);
+	if (ret)
+		return ret;
 
+	ret = ufshcd_mcq_config_resource(hba);
 	return ret;
 }
 
diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
index dee0b37..7bf7599 100644
--- a/include/ufs/ufshcd.h
+++ b/include/ufs/ufshcd.h
@@ -724,6 +724,30 @@ struct ufs_hba_monitor {
 };
 
 /**
+ * struct ufshcd_res_info_t - MCQ related resource regions
+ *
+ * @name: resource name
+ * @resource: pointer to resource region
+ * @base: register base address
+ */
+struct ufshcd_res_info {
+	const char *name;
+	struct resource *resource;
+	void __iomem *base;
+};
+
+enum ufshcd_res {
+	RES_UFS,
+	RES_MCQ,
+	RES_MCQ_SQD,
+	RES_MCQ_SQIS,
+	RES_MCQ_CQD,
+	RES_MCQ_CQIS,
+	RES_MCQ_VS,
+	RES_MAX,
+};
+
+/**
  * struct ufs_hba - per adapter private structure
  * @mmio_base: UFSHCI base register address
  * @ucdl_base_addr: UFS Command Descriptor base address
@@ -835,6 +859,8 @@ struct ufs_hba_monitor {
  * @mcq_sup: is mcq supported by UFSHC
  * @nr_hw_queues: number of hardware queues configured
  * @nr_queues: number of Queues of different queue types
+ * @res: array of resource info of MCQ registers
+ * @mcq_base: Multi circular queue registers base address
  */
 struct ufs_hba {
 	void __iomem *mmio_base;
@@ -988,6 +1014,8 @@ struct ufs_hba {
 	bool mcq_sup;
 	unsigned int nr_hw_queues;
 	unsigned int nr_queues[HCTX_MAX_TYPES];
+	struct ufshcd_res_info res[RES_MAX];
+	void __iomem *mcq_base;
 };
 
 /* Returns true if clocks can be gated. Otherwise false */
-- 
2.7.4

Re: [PATCH v5 06/16] ufs: core: mcq: Configure resource regions

[Bart Van Assche]

On 11/22/22 20:10, Asutosh Das wrote:
> +/* Resources */
> +static const struct ufshcd_res_info ufs_res_info[RES_MAX] = {
> +	{.name = "ufs_mem",},
> +	{.name = "mcq",},
> +	/* Submission Queue DAO */
> +	{.name = "mcq_sqd",},
> +	/* Submission Queue Interrupt Status */
> +	{.name = "mcq_sqis",},
> +	/* Completion Queue DAO */
> +	{.name = "mcq_cqd",},
> +	/* Completion Queue Interrupt Status */
> +	{.name = "mcq_cqis",},
> +	/* MCQ vendor specific */
> +	{.name = "mcq_vs",},
> +};

Which names to associate with nodes in the device tree for UFS MCQ 
resources has not been standardized. Additionally, the UFS driver is 
also used on platforms that do not support the device tree (e.g. Intel 
x86 platforms). So I don't think that the platform_get_resource_byname() 
calls should occur in the UFS driver core. How about moving the code 
that queries device tree nodes into the Qualcomm driver until a second 
user of this code is added? If a second user of this code appears this 
code could e.g. become a kernel module shared by both UFS host 
controller drivers.

Thanks,

Bart.

[PATCH v5 07/16] ufs: core: mcq: Calculate queue depth

[Asutosh Das]

The ufs device defines the supported queuedepth by
bqueuedepth which has a max value of 256.
The HC defines MAC (Max Active Commands) that define
the max number of commands that in flight to the ufs
device.
Calculate and configure the nutrs based on both these
values.

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
---
 drivers/ufs/core/ufs-mcq.c     | 32 ++++++++++++++++++++++++++++++++
 drivers/ufs/core/ufshcd-priv.h |  9 +++++++++
 drivers/ufs/core/ufshcd.c      | 17 ++++++++++++++++-
 drivers/ufs/host/ufs-qcom.c    |  8 ++++++++
 include/ufs/ufs.h              |  2 ++
 include/ufs/ufshcd.h           |  2 ++
 include/ufs/ufshci.h           |  1 +
 7 files changed, 70 insertions(+), 1 deletion(-)

diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
index 4aaa6aa..e95f748 100644
--- a/drivers/ufs/core/ufs-mcq.c
+++ b/drivers/ufs/core/ufs-mcq.c
@@ -18,6 +18,8 @@
 #define UFS_MCQ_NUM_DEV_CMD_QUEUES 1
 #define UFS_MCQ_MIN_POLL_QUEUES 0
 
+#define MAX_DEV_CMD_ENTRIES	2
+#define MCQ_CFG_MAC_MASK	GENMASK(16, 8)
 #define MCQ_QCFGPTR_MASK	GENMASK(7, 0)
 #define MCQ_QCFGPTR_UNIT	0x200
 #define MCQ_SQATTR_OFFSET(c) \
@@ -88,6 +90,36 @@ static const struct ufshcd_res_info ufs_res_info[RES_MAX] = {
 	{.name = "mcq_vs",},
 };
 
+/**
+ * ufshcd_mcq_decide_queue_depth - decide the queue depth
+ * @hba - per adapter instance
+ *
+ * MAC - Max. Active Command of the Host Controller (HC)
+ * HC wouldn't send more than this commands to the device.
+ * It is mandatory to implement get_hba_mac() to enable MCQ mode.
+ * Calculates and adjusts the queue depth based on the depth
+ * supported by the HC and ufs device.
+ */
+int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba)
+{
+	int mac;
+
+	/* Mandatory to implement get_hba_mac() */
+	mac = ufshcd_mcq_vops_get_hba_mac(hba);
+	if (mac < 0) {
+		dev_err(hba->dev, "Failed to get mac, err=%d\n", mac);
+		return mac;
+	}
+
+	WARN_ON(!hba->dev_info.bqueuedepth);
+	/*
+	 * max. value of bqueuedepth = 256, mac is host dependent.
+	 * It is mandatory for UFS device to define bQueueDepth if
+	 * shared queuing architecture is enabled.
+	 */
+	return min_t(int, mac, hba->dev_info.bqueuedepth);
+}
+
 static int ufshcd_mcq_config_resource(struct ufs_hba *hba)
 {
 	struct platform_device *pdev = to_platform_device(hba->dev);
diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
index 9368ba2..9f40fa5 100644
--- a/drivers/ufs/core/ufshcd-priv.h
+++ b/drivers/ufs/core/ufshcd-priv.h
@@ -62,6 +62,7 @@ int ufshcd_query_flag(struct ufs_hba *hba, enum query_opcode opcode,
 	enum flag_idn idn, u8 index, bool *flag_res);
 void ufshcd_auto_hibern8_update(struct ufs_hba *hba, u32 ahit);
 int ufshcd_mcq_init(struct ufs_hba *hba);
+int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba);
 
 #define SD_ASCII_STD true
 #define SD_RAW false
@@ -227,6 +228,14 @@ static inline void ufshcd_vops_config_scaling_param(struct ufs_hba *hba,
 		hba->vops->config_scaling_param(hba, p, data);
 }
 
+static inline int ufshcd_mcq_vops_get_hba_mac(struct ufs_hba *hba)
+{
+	if (hba->vops && hba->vops->get_hba_mac)
+		return hba->vops->get_hba_mac(hba);
+
+	return -EOPNOTSUPP;
+}
+
 extern const struct ufs_pm_lvl_states ufs_pm_lvl_states[];
 
 /**
diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 0c4cd8f..ae065da 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -7783,6 +7783,7 @@ static int ufs_get_device_desc(struct ufs_hba *hba)
 	/* getting Specification Version in big endian format */
 	dev_info->wspecversion = desc_buf[DEVICE_DESC_PARAM_SPEC_VER] << 8 |
 				      desc_buf[DEVICE_DESC_PARAM_SPEC_VER + 1];
+	dev_info->bqueuedepth = desc_buf[DEVICE_DESC_PARAM_Q_DPTH];
 	b_ufs_feature_sup = desc_buf[DEVICE_DESC_PARAM_UFS_FEAT];
 
 	model_index = desc_buf[DEVICE_DESC_PARAM_PRDCT_NAME];
@@ -8198,7 +8199,21 @@ static int ufshcd_add_lus(struct ufs_hba *hba)
 
 static int ufshcd_alloc_mcq(struct ufs_hba *hba)
 {
-	return ufshcd_mcq_init(hba);
+	int ret;
+	int old_nutrs = hba->nutrs;
+
+	ret = ufshcd_mcq_decide_queue_depth(hba);
+	if (ret < 0)
+		return ret;
+
+	hba->nutrs = ret;
+	ret = ufshcd_mcq_init(hba);
+	if (ret) {
+		hba->nutrs = old_nutrs;
+		return ret;
+	}
+
+	return 0;
 }
 
 /**
diff --git a/drivers/ufs/host/ufs-qcom.c b/drivers/ufs/host/ufs-qcom.c
index 8ad1415..7bd3c37 100644
--- a/drivers/ufs/host/ufs-qcom.c
+++ b/drivers/ufs/host/ufs-qcom.c
@@ -25,6 +25,7 @@
 #define UFS_QCOM_DEFAULT_DBG_PRINT_EN	\
 	(UFS_QCOM_DBG_PRINT_REGS_EN | UFS_QCOM_DBG_PRINT_TEST_BUS_EN)
 
+#define MAX_SUPP_MAC 64
 enum {
 	TSTBUS_UAWM,
 	TSTBUS_UARM,
@@ -1424,6 +1425,12 @@ static void ufs_qcom_config_scaling_param(struct ufs_hba *hba,
 }
 #endif
 
+static int ufs_qcom_get_hba_mac(struct ufs_hba *hba)
+{
+	/* Qualcomm HC supports up to 64 */
+	return MAX_SUPP_MAC;
+}
+
 /*
  * struct ufs_hba_qcom_vops - UFS QCOM specific variant operations
  *
@@ -1447,6 +1454,7 @@ static const struct ufs_hba_variant_ops ufs_hba_qcom_vops = {
 	.device_reset		= ufs_qcom_device_reset,
 	.config_scaling_param = ufs_qcom_config_scaling_param,
 	.program_key		= ufs_qcom_ice_program_key,
+	.get_hba_mac		= ufs_qcom_get_hba_mac,
 };
 
 /**
diff --git a/include/ufs/ufs.h b/include/ufs/ufs.h
index ba2a1d8..5112418 100644
--- a/include/ufs/ufs.h
+++ b/include/ufs/ufs.h
@@ -591,6 +591,8 @@ struct ufs_dev_info {
 	u8	*model;
 	u16	wspecversion;
 	u32	clk_gating_wait_us;
+	/* Stores the depth of queue in UFS device */
+	u8	bqueuedepth;
 
 	/* UFS HPB related flag */
 	bool	hpb_enabled;
diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
index 7bf7599..e03b310 100644
--- a/include/ufs/ufshcd.h
+++ b/include/ufs/ufshcd.h
@@ -297,6 +297,7 @@ struct ufs_pwr_mode_info {
  * @config_scaling_param: called to configure clock scaling parameters
  * @program_key: program or evict an inline encryption key
  * @event_notify: called to notify important events
+ * @get_hba_mac: called to get vendor specific mac value, mandatory for mcq mode
  */
 struct ufs_hba_variant_ops {
 	const char *name;
@@ -335,6 +336,7 @@ struct ufs_hba_variant_ops {
 			       const union ufs_crypto_cfg_entry *cfg, int slot);
 	void	(*event_notify)(struct ufs_hba *hba,
 				enum ufs_event_type evt, void *data);
+	int	(*get_hba_mac)(struct ufs_hba *hba);
 };
 
 /* clock gating state  */
diff --git a/include/ufs/ufshci.h b/include/ufs/ufshci.h
index 4d4da06..67fcebd 100644
--- a/include/ufs/ufshci.h
+++ b/include/ufs/ufshci.h
@@ -57,6 +57,7 @@ enum {
 	REG_UFS_CCAP				= 0x100,
 	REG_UFS_CRYPTOCAP			= 0x104,
 
+	REG_UFS_MCQ_CFG				= 0x380,
 	UFSHCI_CRYPTO_REG_SPACE_SIZE		= 0x400,
 };
 
-- 
2.7.4

Re: [PATCH v5 07/16] ufs: core: mcq: Calculate queue depth

[Bart Van Assche]

On 11/22/22 20:10, Asutosh Das wrote:
> +int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba)
> +{
> +	int mac;
> +
> +	/* Mandatory to implement get_hba_mac() */
> +	mac = ufshcd_mcq_vops_get_hba_mac(hba);
> +	if (mac < 0) {
> +		dev_err(hba->dev, "Failed to get mac, err=%d\n", mac);
> +		return mac;
> +	}
> +
> +	WARN_ON(!hba->dev_info.bqueuedepth);
> +	/*
> +	 * max. value of bqueuedepth = 256, mac is host dependent.
> +	 * It is mandatory for UFS device to define bQueueDepth if
> +	 * shared queuing architecture is enabled.
> +	 */
> +	return min_t(int, mac, hba->dev_info.bqueuedepth);
> +}

Isn't WARN_ON_ONCE() recommended over WARN_ON()?

Otherwise this patch looks good to me.

Thanks,

Bart.

Re: [PATCH v5 07/16] ufs: core: mcq: Calculate queue depth

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:20PM -0800, Asutosh Das wrote:
> The ufs device defines the supported queuedepth by
> bqueuedepth which has a max value of 256.
> The HC defines MAC (Max Active Commands) that define
> the max number of commands that in flight to the ufs
> device.
> Calculate and configure the nutrs based on both these
> values.
> 
> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
> ---
>  drivers/ufs/core/ufs-mcq.c     | 32 ++++++++++++++++++++++++++++++++
>  drivers/ufs/core/ufshcd-priv.h |  9 +++++++++
>  drivers/ufs/core/ufshcd.c      | 17 ++++++++++++++++-
>  drivers/ufs/host/ufs-qcom.c    |  8 ++++++++
>  include/ufs/ufs.h              |  2 ++
>  include/ufs/ufshcd.h           |  2 ++
>  include/ufs/ufshci.h           |  1 +
>  7 files changed, 70 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
> index 4aaa6aa..e95f748 100644
> --- a/drivers/ufs/core/ufs-mcq.c
> +++ b/drivers/ufs/core/ufs-mcq.c
> @@ -18,6 +18,8 @@
>  #define UFS_MCQ_NUM_DEV_CMD_QUEUES 1
>  #define UFS_MCQ_MIN_POLL_QUEUES 0
>  
> +#define MAX_DEV_CMD_ENTRIES	2
> +#define MCQ_CFG_MAC_MASK	GENMASK(16, 8)
>  #define MCQ_QCFGPTR_MASK	GENMASK(7, 0)
>  #define MCQ_QCFGPTR_UNIT	0x200
>  #define MCQ_SQATTR_OFFSET(c) \
> @@ -88,6 +90,36 @@ static const struct ufshcd_res_info ufs_res_info[RES_MAX] = {
>  	{.name = "mcq_vs",},
>  };
>  
> +/**
> + * ufshcd_mcq_decide_queue_depth - decide the queue depth
> + * @hba - per adapter instance
> + *

Kernel doc should define the return value also.

> + * MAC - Max. Active Command of the Host Controller (HC)
> + * HC wouldn't send more than this commands to the device.
> + * It is mandatory to implement get_hba_mac() to enable MCQ mode.
> + * Calculates and adjusts the queue depth based on the depth
> + * supported by the HC and ufs device.
> + */
> +int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba)
> +{
> +	int mac;
> +
> +	/* Mandatory to implement get_hba_mac() */
> +	mac = ufshcd_mcq_vops_get_hba_mac(hba);
> +	if (mac < 0) {
> +		dev_err(hba->dev, "Failed to get mac, err=%d\n", mac);
> +		return mac;
> +	}
> +
> +	WARN_ON(!hba->dev_info.bqueuedepth);

Instead of panic, you could just print and return an error.

> +	/*
> +	 * max. value of bqueuedepth = 256, mac is host dependent.
> +	 * It is mandatory for UFS device to define bQueueDepth if
> +	 * shared queuing architecture is enabled.
> +	 */
> +	return min_t(int, mac, hba->dev_info.bqueuedepth);
> +}
> +
>  static int ufshcd_mcq_config_resource(struct ufs_hba *hba)
>  {
>  	struct platform_device *pdev = to_platform_device(hba->dev);
> diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
> index 9368ba2..9f40fa5 100644
> --- a/drivers/ufs/core/ufshcd-priv.h
> +++ b/drivers/ufs/core/ufshcd-priv.h
> @@ -62,6 +62,7 @@ int ufshcd_query_flag(struct ufs_hba *hba, enum query_opcode opcode,
>  	enum flag_idn idn, u8 index, bool *flag_res);
>  void ufshcd_auto_hibern8_update(struct ufs_hba *hba, u32 ahit);
>  int ufshcd_mcq_init(struct ufs_hba *hba);
> +int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba);
>  
>  #define SD_ASCII_STD true
>  #define SD_RAW false
> @@ -227,6 +228,14 @@ static inline void ufshcd_vops_config_scaling_param(struct ufs_hba *hba,
>  		hba->vops->config_scaling_param(hba, p, data);
>  }
>  
> +static inline int ufshcd_mcq_vops_get_hba_mac(struct ufs_hba *hba)

Again, no inline please.

> +{
> +	if (hba->vops && hba->vops->get_hba_mac)
> +		return hba->vops->get_hba_mac(hba);
> +
> +	return -EOPNOTSUPP;
> +}
> +
>  extern const struct ufs_pm_lvl_states ufs_pm_lvl_states[];
>  
>  /**
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 0c4cd8f..ae065da 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -7783,6 +7783,7 @@ static int ufs_get_device_desc(struct ufs_hba *hba)
>  	/* getting Specification Version in big endian format */
>  	dev_info->wspecversion = desc_buf[DEVICE_DESC_PARAM_SPEC_VER] << 8 |
>  				      desc_buf[DEVICE_DESC_PARAM_SPEC_VER + 1];
> +	dev_info->bqueuedepth = desc_buf[DEVICE_DESC_PARAM_Q_DPTH];
>  	b_ufs_feature_sup = desc_buf[DEVICE_DESC_PARAM_UFS_FEAT];
>  
>  	model_index = desc_buf[DEVICE_DESC_PARAM_PRDCT_NAME];
> @@ -8198,7 +8199,21 @@ static int ufshcd_add_lus(struct ufs_hba *hba)
>  
>  static int ufshcd_alloc_mcq(struct ufs_hba *hba)
>  {
> -	return ufshcd_mcq_init(hba);
> +	int ret;
> +	int old_nutrs = hba->nutrs;
> +
> +	ret = ufshcd_mcq_decide_queue_depth(hba);
> +	if (ret < 0)
> +		return ret;
> +
> +	hba->nutrs = ret;
> +	ret = ufshcd_mcq_init(hba);
> +	if (ret) {
> +		hba->nutrs = old_nutrs;
> +		return ret;
> +	}
> +
> +	return 0;
>  }
>  
>  /**
> diff --git a/drivers/ufs/host/ufs-qcom.c b/drivers/ufs/host/ufs-qcom.c
> index 8ad1415..7bd3c37 100644
> --- a/drivers/ufs/host/ufs-qcom.c
> +++ b/drivers/ufs/host/ufs-qcom.c
> @@ -25,6 +25,7 @@
>  #define UFS_QCOM_DEFAULT_DBG_PRINT_EN	\
>  	(UFS_QCOM_DBG_PRINT_REGS_EN | UFS_QCOM_DBG_PRINT_TEST_BUS_EN)
>  
> +#define MAX_SUPP_MAC 64

Similar definitions are part of ufs-qcom.h.

Thanks,
Mani

>  enum {
>  	TSTBUS_UAWM,
>  	TSTBUS_UARM,
> @@ -1424,6 +1425,12 @@ static void ufs_qcom_config_scaling_param(struct ufs_hba *hba,
>  }
>  #endif
>  
> +static int ufs_qcom_get_hba_mac(struct ufs_hba *hba)
> +{
> +	/* Qualcomm HC supports up to 64 */
> +	return MAX_SUPP_MAC;
> +}
> +
>  /*
>   * struct ufs_hba_qcom_vops - UFS QCOM specific variant operations
>   *
> @@ -1447,6 +1454,7 @@ static const struct ufs_hba_variant_ops ufs_hba_qcom_vops = {
>  	.device_reset		= ufs_qcom_device_reset,
>  	.config_scaling_param = ufs_qcom_config_scaling_param,
>  	.program_key		= ufs_qcom_ice_program_key,
> +	.get_hba_mac		= ufs_qcom_get_hba_mac,
>  };
>  
>  /**
> diff --git a/include/ufs/ufs.h b/include/ufs/ufs.h
> index ba2a1d8..5112418 100644
> --- a/include/ufs/ufs.h
> +++ b/include/ufs/ufs.h
> @@ -591,6 +591,8 @@ struct ufs_dev_info {
>  	u8	*model;
>  	u16	wspecversion;
>  	u32	clk_gating_wait_us;
> +	/* Stores the depth of queue in UFS device */
> +	u8	bqueuedepth;
>  
>  	/* UFS HPB related flag */
>  	bool	hpb_enabled;
> diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
> index 7bf7599..e03b310 100644
> --- a/include/ufs/ufshcd.h
> +++ b/include/ufs/ufshcd.h
> @@ -297,6 +297,7 @@ struct ufs_pwr_mode_info {
>   * @config_scaling_param: called to configure clock scaling parameters
>   * @program_key: program or evict an inline encryption key
>   * @event_notify: called to notify important events
> + * @get_hba_mac: called to get vendor specific mac value, mandatory for mcq mode
>   */
>  struct ufs_hba_variant_ops {
>  	const char *name;
> @@ -335,6 +336,7 @@ struct ufs_hba_variant_ops {
>  			       const union ufs_crypto_cfg_entry *cfg, int slot);
>  	void	(*event_notify)(struct ufs_hba *hba,
>  				enum ufs_event_type evt, void *data);
> +	int	(*get_hba_mac)(struct ufs_hba *hba);
>  };
>  
>  /* clock gating state  */
> diff --git a/include/ufs/ufshci.h b/include/ufs/ufshci.h
> index 4d4da06..67fcebd 100644
> --- a/include/ufs/ufshci.h
> +++ b/include/ufs/ufshci.h
> @@ -57,6 +57,7 @@ enum {
>  	REG_UFS_CCAP				= 0x100,
>  	REG_UFS_CRYPTOCAP			= 0x104,
>  
> +	REG_UFS_MCQ_CFG				= 0x380,
>  	UFSHCI_CRYPTO_REG_SPACE_SIZE		= 0x400,
>  };
>  
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

Re: [PATCH v5 07/16] ufs: core: mcq: Calculate queue depth

[Asutosh Das]

On Mon, Nov 28 2022 at 07:15 -0800, Manivannan Sadhasivam wrote:
>On Tue, Nov 22, 2022 at 08:10:20PM -0800, Asutosh Das wrote:
>> The ufs device defines the supported queuedepth by
>> bqueuedepth which has a max value of 256.
>> The HC defines MAC (Max Active Commands) that define
>> the max number of commands that in flight to the ufs
>> device.
>> Calculate and configure the nutrs based on both these
>> values.
>>
>> Co-developed-by: Can Guo <quic_cang@quicinc.com>
>> Signed-off-by: Can Guo <quic_cang@quicinc.com>
>> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
>> ---
>>  drivers/ufs/core/ufs-mcq.c     | 32 ++++++++++++++++++++++++++++++++
>>  drivers/ufs/core/ufshcd-priv.h |  9 +++++++++
>>  drivers/ufs/core/ufshcd.c      | 17 ++++++++++++++++-
>>  drivers/ufs/host/ufs-qcom.c    |  8 ++++++++
>>  include/ufs/ufs.h              |  2 ++
>>  include/ufs/ufshcd.h           |  2 ++
>>  include/ufs/ufshci.h           |  1 +
>>  7 files changed, 70 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
>> index 4aaa6aa..e95f748 100644
>> --- a/drivers/ufs/core/ufs-mcq.c
>> +++ b/drivers/ufs/core/ufs-mcq.c
>> @@ -18,6 +18,8 @@
>>  #define UFS_MCQ_NUM_DEV_CMD_QUEUES 1
>>  #define UFS_MCQ_MIN_POLL_QUEUES 0
>>
>> +#define MAX_DEV_CMD_ENTRIES	2
>> +#define MCQ_CFG_MAC_MASK	GENMASK(16, 8)
>>  #define MCQ_QCFGPTR_MASK	GENMASK(7, 0)
>>  #define MCQ_QCFGPTR_UNIT	0x200
>>  #define MCQ_SQATTR_OFFSET(c) \
>> @@ -88,6 +90,36 @@ static const struct ufshcd_res_info ufs_res_info[RES_MAX] = {
>>  	{.name = "mcq_vs",},
>>  };
[...]

Hello Mani,
Thanks for taking a look.

>> +	WARN_ON(!hba->dev_info.bqueuedepth);
>
>Instead of panic, you could just print and return an error.
>
I'd make it WARN_ON_ONCE();

>> +	/*
>> +	 * max. value of bqueuedepth = 256, mac is host dependent.
>> +	 * It is mandatory for UFS device to define bQueueDepth if
>> +	 * shared queuing architecture is enabled.
>> +	 */
>> +	return min_t(int, mac, hba->dev_info.bqueuedepth);
>> +}
>> +
>>  static int ufshcd_mcq_config_resource(struct ufs_hba *hba)
>>  {
>>  	struct platform_device *pdev = to_platform_device(hba->dev);
>> diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
>> index 9368ba2..9f40fa5 100644
>> --- a/drivers/ufs/core/ufshcd-priv.h
>> +++ b/drivers/ufs/core/ufshcd-priv.h
>> @@ -62,6 +62,7 @@ int ufshcd_query_flag(struct ufs_hba *hba, enum query_opcode opcode,
>>  	enum flag_idn idn, u8 index, bool *flag_res);
>>  void ufshcd_auto_hibern8_update(struct ufs_hba *hba, u32 ahit);
>>  int ufshcd_mcq_init(struct ufs_hba *hba);
>> +int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba);
>>
>>  #define SD_ASCII_STD true
>>  #define SD_RAW false
>> @@ -227,6 +228,14 @@ static inline void ufshcd_vops_config_scaling_param(struct ufs_hba *hba,
>>  		hba->vops->config_scaling_param(hba, p, data);
>>  }
>>
>> +static inline int ufshcd_mcq_vops_get_hba_mac(struct ufs_hba *hba)
>
>Again, no inline please.
>
It spits out the following warning for all files that include this header, when
inline is removed:
warning: 'ufshcd_mcq_vops_get_hba_mac' defined but not used [-Wunused-function]

[...]
>> +#define MAX_SUPP_MAC 64
>
>Similar definitions are part of ufs-qcom.h.
>
>Thanks,
>Mani
>
>
>-- 
>மணிவண்ணன் சதாசிவம்

Re: [PATCH v5 07/16] ufs: core: mcq: Calculate queue depth

[Bart Van Assche]

On 11/28/22 11:54, Asutosh Das wrote:
> On Mon, Nov 28 2022 at 07:15 -0800, Manivannan Sadhasivam wrote:
>> On Tue, Nov 22, 2022 at 08:10:20PM -0800, Asutosh Das wrote:
>>> +static inline int ufshcd_mcq_vops_get_hba_mac(struct ufs_hba *hba)
>>
>> Again, no inline please.
>>
> It spits out the following warning for all files that include this 
> header, when
> inline is removed:
> warning: 'ufshcd_mcq_vops_get_hba_mac' defined but not used 
> [-Wunused-function]

My understanding is that the "no inline" rule applies to .c files only 
and also that functions defined in header files should be declared 
"static inline".

Thanks,

Bart.

[PATCH v5 08/16] ufs: core: mcq: Allocate memory for mcq mode

[Asutosh Das]

To read the bqueuedepth, the device descriptor is fetched
in Single Doorbell Mode. This allocated memory may not be
enough for MCQ mode because the number of tags supported
in MCQ mode may be larger than in SDB mode.
Hence, release the memory allocated in SDB mode and allocate
memory for MCQ mode operation.
Define the ufs hardware queue and Completion Queue Entry.

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
---
 drivers/ufs/core/ufs-mcq.c     | 58 ++++++++++++++++++++++++++++++++++++++++--
 drivers/ufs/core/ufshcd-priv.h |  1 +
 drivers/ufs/core/ufshcd.c      | 42 +++++++++++++++++++++++++++---
 include/ufs/ufshcd.h           | 19 ++++++++++++++
 include/ufs/ufshci.h           | 22 ++++++++++++++++
 5 files changed, 137 insertions(+), 5 deletions(-)

diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
index e95f748..51f0e40 100644
--- a/drivers/ufs/core/ufs-mcq.c
+++ b/drivers/ufs/core/ufs-mcq.c
@@ -247,15 +247,69 @@ static int ufshcd_mcq_config_nr_queues(struct ufs_hba *hba)
 	return 0;
 }
 
+int ufshcd_mcq_memory_alloc(struct ufs_hba *hba)
+{
+	struct ufs_hw_queue *hwq;
+	size_t utrdl_size, cqe_size;
+	int i;
+
+	for (i = 0; i < hba->nr_hw_queues; i++) {
+		hwq = &hba->uhq[i];
+
+		utrdl_size = sizeof(struct utp_transfer_req_desc) *
+			     hwq->max_entries;
+		hwq->sqe_base_addr = dmam_alloc_coherent(hba->dev, utrdl_size,
+							 &hwq->sqe_dma_addr,
+							 GFP_KERNEL);
+		if (!hwq->sqe_dma_addr) {
+			dev_err(hba->dev, "SQE allocation failed\n");
+			return -ENOMEM;
+		}
+
+		cqe_size = sizeof(struct cq_entry) * hwq->max_entries;
+		hwq->cqe_base_addr = dmam_alloc_coherent(hba->dev, cqe_size,
+							 &hwq->cqe_dma_addr,
+							 GFP_KERNEL);
+		if (!hwq->cqe_dma_addr) {
+			dev_err(hba->dev, "CQE allocation failed\n");
+			return -ENOMEM;
+		}
+	}
+
+	return 0;
+}
+
 int ufshcd_mcq_init(struct ufs_hba *hba)
 {
-	int ret;
+	struct ufs_hw_queue *hwq;
+	int ret, i;
 
 	ret = ufshcd_mcq_config_nr_queues(hba);
 	if (ret)
 		return ret;
 
 	ret = ufshcd_mcq_config_resource(hba);
-	return ret;
+	if (ret)
+		return ret;
+
+	hba->uhq = devm_kzalloc(hba->dev,
+				hba->nr_hw_queues * sizeof(struct ufs_hw_queue),
+				GFP_KERNEL);
+	if (!hba->uhq) {
+		dev_err(hba->dev, "ufs hw queue memory allocation failed\n");
+		return -ENOMEM;
+	}
+
+	for (i = 0; i < hba->nr_hw_queues; i++) {
+		hwq = &hba->uhq[i];
+		hwq->max_entries = hba->nutrs;
+	}
+
+	/* The very first HW queue serves device commands */
+	hba->dev_cmd_queue = &hba->uhq[0];
+	/* Give dev_cmd_queue the minimal number of entries */
+	hba->dev_cmd_queue->max_entries = MAX_DEV_CMD_ENTRIES;
+
+	return 0;
 }
 
diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
index 9f40fa5..4d2bde2 100644
--- a/drivers/ufs/core/ufshcd-priv.h
+++ b/drivers/ufs/core/ufshcd-priv.h
@@ -63,6 +63,7 @@ int ufshcd_query_flag(struct ufs_hba *hba, enum query_opcode opcode,
 void ufshcd_auto_hibern8_update(struct ufs_hba *hba, u32 ahit);
 int ufshcd_mcq_init(struct ufs_hba *hba);
 int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba);
+int ufshcd_mcq_memory_alloc(struct ufs_hba *hba);
 
 #define SD_ASCII_STD true
 #define SD_RAW false
diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index ae065da..45686e8 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -3740,6 +3740,12 @@ static int ufshcd_memory_alloc(struct ufs_hba *hba)
 	}
 
 	/*
+	 * Not freed if MCQ is configured see ufshcd_release_sdb_queue() and
+	 * ufshcd_config_mcq()
+	 */
+	if (hba->utmrdl_base_addr)
+		goto skip_utmrdl;
+	/*
 	 * Allocate memory for UTP Task Management descriptors
 	 * UFSHCI requires 1024 byte alignment of UTMRD
 	 */
@@ -3755,6 +3761,7 @@ static int ufshcd_memory_alloc(struct ufs_hba *hba)
 		goto out;
 	}
 
+skip_utmrdl:
 	/* Allocate memory for local reference block */
 	hba->lrb = devm_kcalloc(hba->dev,
 				hba->nutrs, sizeof(struct ufshcd_lrb),
@@ -8197,6 +8204,22 @@ static int ufshcd_add_lus(struct ufs_hba *hba)
 	return ret;
 }
 
+/* SDB - Single Doorbell */
+static void ufshcd_release_sdb_queue(struct ufs_hba *hba, int nutrs)
+{
+	size_t ucdl_size, utrdl_size;
+
+	ucdl_size = sizeof(struct utp_transfer_cmd_desc) * nutrs;
+	dmam_free_coherent(hba->dev, ucdl_size, hba->ucdl_base_addr,
+			   hba->ucdl_dma_addr);
+
+	utrdl_size = sizeof(struct utp_transfer_req_desc) * nutrs;
+	dmam_free_coherent(hba->dev, utrdl_size, hba->utrdl_base_addr,
+			   hba->utrdl_dma_addr);
+
+	devm_kfree(hba->dev, hba->lrb);
+}
+
 static int ufshcd_alloc_mcq(struct ufs_hba *hba)
 {
 	int ret;
@@ -8208,12 +8231,25 @@ static int ufshcd_alloc_mcq(struct ufs_hba *hba)
 
 	hba->nutrs = ret;
 	ret = ufshcd_mcq_init(hba);
-	if (ret) {
-		hba->nutrs = old_nutrs;
-		return ret;
+	if (ret)
+		goto err;
+
+	if (hba->nutrs != old_nutrs) {
+		ufshcd_release_sdb_queue(hba, old_nutrs);
+		ret = ufshcd_memory_alloc(hba);
+		if (ret)
+			goto err;
+		ufshcd_host_memory_configure(hba);
 	}
 
+	ret = ufshcd_mcq_memory_alloc(hba);
+	if (ret)
+		goto err;
+
 	return 0;
+err:
+	hba->nutrs = old_nutrs;
+	return ret;
 }
 
 /**
diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
index e03b310..e478bab 100644
--- a/include/ufs/ufshcd.h
+++ b/include/ufs/ufshcd.h
@@ -863,6 +863,8 @@ enum ufshcd_res {
  * @nr_queues: number of Queues of different queue types
  * @res: array of resource info of MCQ registers
  * @mcq_base: Multi circular queue registers base address
+ * @uhq: array of supported hardware queues
+ * @dev_cmd_queue: Queue for issuing device management commands
  */
 struct ufs_hba {
 	void __iomem *mmio_base;
@@ -1018,6 +1020,23 @@ struct ufs_hba {
 	unsigned int nr_queues[HCTX_MAX_TYPES];
 	struct ufshcd_res_info res[RES_MAX];
 	void __iomem *mcq_base;
+	struct ufs_hw_queue *uhq;
+	struct ufs_hw_queue *dev_cmd_queue;
+};
+
+/**
+ * @sqe_base_addr: submission queue entry base address
+ * @sqe_dma_addr: submission queue dma address
+ * @cqe_base_addr: completion queue base address
+ * @cqe_dma_addr: completion queue dma address
+ * @max_entries: max number of slots in this hardware queue
+ */
+struct ufs_hw_queue {
+	void *sqe_base_addr;
+	dma_addr_t sqe_dma_addr;
+	struct cq_entry *cqe_base_addr;
+	dma_addr_t cqe_dma_addr;
+	u32 max_entries;
 };
 
 /* Returns true if clocks can be gated. Otherwise false */
diff --git a/include/ufs/ufshci.h b/include/ufs/ufshci.h
index 67fcebd..1aae5b2 100644
--- a/include/ufs/ufshci.h
+++ b/include/ufs/ufshci.h
@@ -486,6 +486,28 @@ struct utp_transfer_req_desc {
 	__le16  prd_table_offset;
 };
 
+/* MCQ Completion Queue Entry */
+struct cq_entry {
+	/* DW 0-1 */
+	__le64 command_desc_base_addr;
+
+	/* DW 2 */
+	__le16  response_upiu_length;
+	__le16  response_upiu_offset;
+
+	/* DW 3 */
+	__le16  prd_table_length;
+	__le16  prd_table_offset;
+
+	/* DW 4 */
+	__le32 status;
+
+	/* DW 5-7 */
+	u32 reserved[3];
+};
+
+static_assert(sizeof(struct cq_entry) == 32);
+
 /*
  * UTMRD structure.
  */
-- 
2.7.4

Re: [PATCH v5 08/16] ufs: core: mcq: Allocate memory for mcq mode

[Bart Van Assche]

On 11/22/22 20:10, Asutosh Das wrote:
> To read the bqueuedepth, the device descriptor is fetched
> in Single Doorbell Mode. This allocated memory may not be
> enough for MCQ mode because the number of tags supported
> in MCQ mode may be larger than in SDB mode.
> Hence, release the memory allocated in SDB mode and allocate
> memory for MCQ mode operation.
> Define the ufs hardware queue and Completion Queue Entry.

Reviewed-by: Bart Van Assche <bvanassche@acm.org>

Re: [PATCH v5 08/16] ufs: core: mcq: Allocate memory for mcq mode

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:21PM -0800, Asutosh Das wrote:
> To read the bqueuedepth, the device descriptor is fetched
> in Single Doorbell Mode. This allocated memory may not be
> enough for MCQ mode because the number of tags supported
> in MCQ mode may be larger than in SDB mode.
> Hence, release the memory allocated in SDB mode and allocate
> memory for MCQ mode operation.
> Define the ufs hardware queue and Completion Queue Entry.
> 
> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
> ---
>  drivers/ufs/core/ufs-mcq.c     | 58 ++++++++++++++++++++++++++++++++++++++++--
>  drivers/ufs/core/ufshcd-priv.h |  1 +
>  drivers/ufs/core/ufshcd.c      | 42 +++++++++++++++++++++++++++---
>  include/ufs/ufshcd.h           | 19 ++++++++++++++
>  include/ufs/ufshci.h           | 22 ++++++++++++++++
>  5 files changed, 137 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
> index e95f748..51f0e40 100644
> --- a/drivers/ufs/core/ufs-mcq.c
> +++ b/drivers/ufs/core/ufs-mcq.c
> @@ -247,15 +247,69 @@ static int ufshcd_mcq_config_nr_queues(struct ufs_hba *hba)
>  	return 0;
>  }
>  
> +int ufshcd_mcq_memory_alloc(struct ufs_hba *hba)
> +{
> +	struct ufs_hw_queue *hwq;
> +	size_t utrdl_size, cqe_size;
> +	int i;
> +
> +	for (i = 0; i < hba->nr_hw_queues; i++) {
> +		hwq = &hba->uhq[i];
> +
> +		utrdl_size = sizeof(struct utp_transfer_req_desc) *
> +			     hwq->max_entries;
> +		hwq->sqe_base_addr = dmam_alloc_coherent(hba->dev, utrdl_size,
> +							 &hwq->sqe_dma_addr,
> +							 GFP_KERNEL);
> +		if (!hwq->sqe_dma_addr) {
> +			dev_err(hba->dev, "SQE allocation failed\n");
> +			return -ENOMEM;
> +		}
> +
> +		cqe_size = sizeof(struct cq_entry) * hwq->max_entries;
> +		hwq->cqe_base_addr = dmam_alloc_coherent(hba->dev, cqe_size,
> +							 &hwq->cqe_dma_addr,
> +							 GFP_KERNEL);
> +		if (!hwq->cqe_dma_addr) {
> +			dev_err(hba->dev, "CQE allocation failed\n");
> +			return -ENOMEM;
> +		}
> +	}
> +
> +	return 0;
> +}
> +
>  int ufshcd_mcq_init(struct ufs_hba *hba)
>  {
> -	int ret;
> +	struct ufs_hw_queue *hwq;
> +	int ret, i;
>  
>  	ret = ufshcd_mcq_config_nr_queues(hba);
>  	if (ret)
>  		return ret;
>  
>  	ret = ufshcd_mcq_config_resource(hba);
> -	return ret;
> +	if (ret)
> +		return ret;
> +
> +	hba->uhq = devm_kzalloc(hba->dev,
> +				hba->nr_hw_queues * sizeof(struct ufs_hw_queue),
> +				GFP_KERNEL);
> +	if (!hba->uhq) {
> +		dev_err(hba->dev, "ufs hw queue memory allocation failed\n");
> +		return -ENOMEM;
> +	}
> +
> +	for (i = 0; i < hba->nr_hw_queues; i++) {
> +		hwq = &hba->uhq[i];
> +		hwq->max_entries = hba->nutrs;
> +	}
> +
> +	/* The very first HW queue serves device commands */
> +	hba->dev_cmd_queue = &hba->uhq[0];
> +	/* Give dev_cmd_queue the minimal number of entries */
> +	hba->dev_cmd_queue->max_entries = MAX_DEV_CMD_ENTRIES;
> +
> +	return 0;
>  }
>  
> diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
> index 9f40fa5..4d2bde2 100644
> --- a/drivers/ufs/core/ufshcd-priv.h
> +++ b/drivers/ufs/core/ufshcd-priv.h
> @@ -63,6 +63,7 @@ int ufshcd_query_flag(struct ufs_hba *hba, enum query_opcode opcode,
>  void ufshcd_auto_hibern8_update(struct ufs_hba *hba, u32 ahit);
>  int ufshcd_mcq_init(struct ufs_hba *hba);
>  int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba);
> +int ufshcd_mcq_memory_alloc(struct ufs_hba *hba);
>  
>  #define SD_ASCII_STD true
>  #define SD_RAW false
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index ae065da..45686e8 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -3740,6 +3740,12 @@ static int ufshcd_memory_alloc(struct ufs_hba *hba)
>  	}
>  
>  	/*
> +	 * Not freed if MCQ is configured see ufshcd_release_sdb_queue() and
> +	 * ufshcd_config_mcq()

The comment is vague. Use something like,

"Skip allocating memory for utmrdl if it has been allocated during the first
pass (i.e., prior to MCQ enablement)"

> +	 */
> +	if (hba->utmrdl_base_addr)
> +		goto skip_utmrdl;
> +	/*
>  	 * Allocate memory for UTP Task Management descriptors
>  	 * UFSHCI requires 1024 byte alignment of UTMRD
>  	 */
> @@ -3755,6 +3761,7 @@ static int ufshcd_memory_alloc(struct ufs_hba *hba)
>  		goto out;
>  	}
>  
> +skip_utmrdl:
>  	/* Allocate memory for local reference block */
>  	hba->lrb = devm_kcalloc(hba->dev,
>  				hba->nutrs, sizeof(struct ufshcd_lrb),
> @@ -8197,6 +8204,22 @@ static int ufshcd_add_lus(struct ufs_hba *hba)
>  	return ret;
>  }
>  
> +/* SDB - Single Doorbell */
> +static void ufshcd_release_sdb_queue(struct ufs_hba *hba, int nutrs)
> +{
> +	size_t ucdl_size, utrdl_size;
> +
> +	ucdl_size = sizeof(struct utp_transfer_cmd_desc) * nutrs;
> +	dmam_free_coherent(hba->dev, ucdl_size, hba->ucdl_base_addr,
> +			   hba->ucdl_dma_addr);
> +
> +	utrdl_size = sizeof(struct utp_transfer_req_desc) * nutrs;
> +	dmam_free_coherent(hba->dev, utrdl_size, hba->utrdl_base_addr,
> +			   hba->utrdl_dma_addr);
> +
> +	devm_kfree(hba->dev, hba->lrb);
> +}
> +
>  static int ufshcd_alloc_mcq(struct ufs_hba *hba)
>  {
>  	int ret;
> @@ -8208,12 +8231,25 @@ static int ufshcd_alloc_mcq(struct ufs_hba *hba)
>  
>  	hba->nutrs = ret;
>  	ret = ufshcd_mcq_init(hba);
> -	if (ret) {
> -		hba->nutrs = old_nutrs;
> -		return ret;
> +	if (ret)
> +		goto err;
> +

A comment should be added here on why the allocation happens again even though
it is part of the commit description. This helps while going through the code
later.

> +	if (hba->nutrs != old_nutrs) {
> +		ufshcd_release_sdb_queue(hba, old_nutrs);
> +		ret = ufshcd_memory_alloc(hba);
> +		if (ret)
> +			goto err;
> +		ufshcd_host_memory_configure(hba);
>  	}
>  
> +	ret = ufshcd_mcq_memory_alloc(hba);
> +	if (ret)
> +		goto err;
> +
>  	return 0;
> +err:
> +	hba->nutrs = old_nutrs;
> +	return ret;
>  }
>  
>  /**
> diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
> index e03b310..e478bab 100644
> --- a/include/ufs/ufshcd.h
> +++ b/include/ufs/ufshcd.h
> @@ -863,6 +863,8 @@ enum ufshcd_res {
>   * @nr_queues: number of Queues of different queue types
>   * @res: array of resource info of MCQ registers
>   * @mcq_base: Multi circular queue registers base address
> + * @uhq: array of supported hardware queues
> + * @dev_cmd_queue: Queue for issuing device management commands
>   */
>  struct ufs_hba {
>  	void __iomem *mmio_base;
> @@ -1018,6 +1020,23 @@ struct ufs_hba {
>  	unsigned int nr_queues[HCTX_MAX_TYPES];
>  	struct ufshcd_res_info res[RES_MAX];
>  	void __iomem *mcq_base;
> +	struct ufs_hw_queue *uhq;
> +	struct ufs_hw_queue *dev_cmd_queue;
> +};
> +
> +/**

Kernel doc requires the description of the struct itself.

> + * @sqe_base_addr: submission queue entry base address
> + * @sqe_dma_addr: submission queue dma address
> + * @cqe_base_addr: completion queue base address
> + * @cqe_dma_addr: completion queue dma address
> + * @max_entries: max number of slots in this hardware queue
> + */
> +struct ufs_hw_queue {
> +	void *sqe_base_addr;
> +	dma_addr_t sqe_dma_addr;
> +	struct cq_entry *cqe_base_addr;
> +	dma_addr_t cqe_dma_addr;
> +	u32 max_entries;
>  };
>  
>  /* Returns true if clocks can be gated. Otherwise false */
> diff --git a/include/ufs/ufshci.h b/include/ufs/ufshci.h
> index 67fcebd..1aae5b2 100644
> --- a/include/ufs/ufshci.h
> +++ b/include/ufs/ufshci.h
> @@ -486,6 +486,28 @@ struct utp_transfer_req_desc {
>  	__le16  prd_table_offset;
>  };
>  
> +/* MCQ Completion Queue Entry */
> +struct cq_entry {
> +	/* DW 0-1 */
> +	__le64 command_desc_base_addr;
> +
> +	/* DW 2 */
> +	__le16  response_upiu_length;
> +	__le16  response_upiu_offset;
> +
> +	/* DW 3 */
> +	__le16  prd_table_length;
> +	__le16  prd_table_offset;
> +
> +	/* DW 4 */
> +	__le32 status;
> +
> +	/* DW 5-7 */
> +	u32 reserved[3];

It'd be good to use __le32 for the sake of uniformity even though these 3
DWORDS are reserved.

Thanks,
Mani

> +};
> +
> +static_assert(sizeof(struct cq_entry) == 32);
> +
>  /*
>   * UTMRD structure.
>   */
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

[PATCH v5 09/16] ufs: core: mcq: Configure operation and runtime interface

[Asutosh Das]

Runtime and operation registers are defined per Submission
and Completion queue.
The location of these registers is not defined in the spec;
meaning the offsets and stride may vary for different
HC vendors. Establish the stride, base address and doorbell
address offsets from vendor host driver and program it.

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/ufs/core/ufs-mcq.c     | 101 +++++++++++++++++++++++++++++++++++++++++
 drivers/ufs/core/ufshcd-priv.h |  11 +++++
 drivers/ufs/core/ufshcd.c      |  28 ++++++++++++
 drivers/ufs/host/ufs-qcom.c    |  24 ++++++++++
 include/ufs/ufshcd.h           |  52 +++++++++++++++++++++
 include/ufs/ufshci.h           |  31 +++++++++++++
 6 files changed, 247 insertions(+)

diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
index 51f0e40..ebecc47 100644
--- a/drivers/ufs/core/ufs-mcq.c
+++ b/drivers/ufs/core/ufs-mcq.c
@@ -17,6 +17,8 @@
 #define UFS_MCQ_MIN_READ_QUEUES 0
 #define UFS_MCQ_NUM_DEV_CMD_QUEUES 1
 #define UFS_MCQ_MIN_POLL_QUEUES 0
+#define QUEUE_EN_OFFSET 31
+#define QUEUE_ID_OFFSET 16
 
 #define MAX_DEV_CMD_ENTRIES	2
 #define MCQ_CFG_MAC_MASK	GENMASK(16, 8)
@@ -25,6 +27,7 @@
 #define MCQ_SQATTR_OFFSET(c) \
 	((((c) >> 16) & MCQ_QCFGPTR_MASK) * MCQ_QCFGPTR_UNIT)
 #define MCQ_QCFG_SIZE	0x40
+#define MCQ_ENTRY_SIZE_IN_DWORD	8
 
 static int rw_queue_count_set(const char *val, const struct kernel_param *kp)
 {
@@ -120,6 +123,24 @@ int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba)
 	return min_t(int, mac, hba->dev_info.bqueuedepth);
 }
 
+/**
+ * ufshcd_mcq_config_mac - Set the #Max Activ Cmds.
+ * @hba - per adapter instance
+ * @max_active_cmds - maximum # of active commands to the device at any time.
+ *
+ * The controller won't send more than the max_active_cmds to the device at
+ * any time.
+ */
+void ufshcd_mcq_config_mac(struct ufs_hba *hba, u32 max_active_cmds)
+{
+	u32 val;
+
+	val = ufshcd_readl(hba, REG_UFS_MCQ_CFG);
+	val &= ~MCQ_CFG_MAC_MASK;
+	val |= FIELD_PREP(MCQ_CFG_MAC_MASK, max_active_cmds);
+	ufshcd_writel(hba, val, REG_UFS_MCQ_CFG);
+}
+
 static int ufshcd_mcq_config_resource(struct ufs_hba *hba)
 {
 	struct platform_device *pdev = to_platform_device(hba->dev);
@@ -279,6 +300,80 @@ int ufshcd_mcq_memory_alloc(struct ufs_hba *hba)
 	return 0;
 }
 
+/* Operation and runtime registers configuration */
+#define MCQ_CFG_n(r, i)	((r) + MCQ_QCFG_SIZE * (i))
+#define MCQ_OPR_OFFSET_n(p, i) \
+	(hba->mcq_opr[(p)].offset + hba->mcq_opr[(p)].stride * (i))
+
+static void __iomem *mcq_opr_base(struct ufs_hba *hba,
+					 enum ufshcd_mcq_opr n, int i)
+{
+	struct ufshcd_mcq_opr_info_t *opr = &hba->mcq_opr[n];
+
+	return opr->base + opr->stride * i;
+}
+
+void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba)
+{
+	struct ufs_hw_queue *hwq;
+	u16 qsize;
+	int i;
+
+	for (i = 0; i < hba->nr_hw_queues; i++) {
+		hwq = &hba->uhq[i];
+		hwq->id = i;
+		qsize = hwq->max_entries * MCQ_ENTRY_SIZE_IN_DWORD - 1;
+
+		/* Submission Queue Lower Base Address */
+		ufsmcq_writelx(hba, lower_32_bits(hwq->sqe_dma_addr),
+			      MCQ_CFG_n(REG_SQLBA, i));
+		/* Submission Queue Upper Base Address */
+		ufsmcq_writelx(hba, upper_32_bits(hwq->sqe_dma_addr),
+			      MCQ_CFG_n(REG_SQUBA, i));
+		/* Submission Queue Doorbell Address Offset */
+		ufsmcq_writelx(hba, MCQ_OPR_OFFSET_n(OPR_SQD, i),
+			      MCQ_CFG_n(REG_SQDAO, i));
+		/* Submission Queue Interrupt Status Address Offset */
+		ufsmcq_writelx(hba, MCQ_OPR_OFFSET_n(OPR_SQIS, i),
+			      MCQ_CFG_n(REG_SQISAO, i));
+
+		/* Completion Queue Lower Base Address */
+		ufsmcq_writelx(hba, lower_32_bits(hwq->cqe_dma_addr),
+			      MCQ_CFG_n(REG_CQLBA, i));
+		/* Completion Queue Upper Base Address */
+		ufsmcq_writelx(hba, upper_32_bits(hwq->cqe_dma_addr),
+			      MCQ_CFG_n(REG_CQUBA, i));
+		/* Completion Queue Doorbell Address Offset */
+		ufsmcq_writelx(hba, MCQ_OPR_OFFSET_n(OPR_CQD, i),
+			      MCQ_CFG_n(REG_CQDAO, i));
+		/* Completion Queue Interrupt Status Address Offset */
+		ufsmcq_writelx(hba, MCQ_OPR_OFFSET_n(OPR_CQIS, i),
+			      MCQ_CFG_n(REG_CQISAO, i));
+
+		/* Save the base addresses for quicker access */
+		hwq->mcq_sq_head = mcq_opr_base(hba, OPR_SQD, i) + REG_SQHP;
+		hwq->mcq_sq_tail = mcq_opr_base(hba, OPR_SQD, i) + REG_SQTP;
+		hwq->mcq_cq_head = mcq_opr_base(hba, OPR_CQD, i) + REG_CQHP;
+		hwq->mcq_cq_tail = mcq_opr_base(hba, OPR_CQD, i) + REG_CQTP;
+
+		/* Enable Tail Entry Push Status interrupt only for non-poll queues */
+		if (i < hba->nr_hw_queues - hba->nr_queues[HCTX_TYPE_POLL])
+			writel(1, mcq_opr_base(hba, OPR_CQIS, i) + REG_CQIE);
+
+		/* Completion Queue Enable|Size to Completion Queue Attribute */
+		ufsmcq_writel(hba, (1 << QUEUE_EN_OFFSET) | qsize,
+			      MCQ_CFG_n(REG_CQATTR, i));
+
+		/*
+		 * Submission Qeueue Enable|Size|Completion Queue ID to
+		 * Submission Queue Attribute
+		 */
+		ufsmcq_writel(hba, (1 << QUEUE_EN_OFFSET) | qsize |
+			      (i << QUEUE_ID_OFFSET),
+			      MCQ_CFG_n(REG_SQATTR, i));
+	}
+}
+
 int ufshcd_mcq_init(struct ufs_hba *hba)
 {
 	struct ufs_hw_queue *hwq;
@@ -292,6 +387,12 @@ int ufshcd_mcq_init(struct ufs_hba *hba)
 	if (ret)
 		return ret;
 
+	ret = ufshcd_mcq_vops_op_runtime_config(hba);
+	if (ret) {
+		dev_err(hba->dev, "Operation runtime config failed, ret=%d\n",
+			ret);
+		return ret;
+	}
 	hba->uhq = devm_kzalloc(hba->dev,
 				hba->nr_hw_queues * sizeof(struct ufs_hw_queue),
 				GFP_KERNEL);
diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
index 4d2bde2..08963e1 100644
--- a/drivers/ufs/core/ufshcd-priv.h
+++ b/drivers/ufs/core/ufshcd-priv.h
@@ -64,6 +64,9 @@ void ufshcd_auto_hibern8_update(struct ufs_hba *hba, u32 ahit);
 int ufshcd_mcq_init(struct ufs_hba *hba);
 int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba);
 int ufshcd_mcq_memory_alloc(struct ufs_hba *hba);
+void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba);
+void ufshcd_mcq_config_mac(struct ufs_hba *hba, u32 max_active_cmds);
+void ufshcd_mcq_select_mcq_mode(struct ufs_hba *hba);
 
 #define SD_ASCII_STD true
 #define SD_RAW false
@@ -237,6 +240,14 @@ static inline int ufshcd_mcq_vops_get_hba_mac(struct ufs_hba *hba)
 	return -EOPNOTSUPP;
 }
 
+static inline int ufshcd_mcq_vops_op_runtime_config(struct ufs_hba *hba)
+{
+	if (hba->vops && hba->vops->op_runtime_config)
+		return hba->vops->op_runtime_config(hba);
+
+	return -EOPNOTSUPP;
+}
+
 extern const struct ufs_pm_lvl_states ufs_pm_lvl_states[];
 
 /**
diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 45686e8..042ecf04 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -43,6 +43,12 @@
 #define UFSHCD_ENABLE_INTRS	(UTP_TRANSFER_REQ_COMPL |\
 				 UTP_TASK_REQ_COMPL |\
 				 UFSHCD_ERROR_MASK)
+
+#define UFSHCD_ENABLE_MCQ_INTRS	(UTP_TASK_REQ_COMPL |\
+				 UFSHCD_ERROR_MASK |\
+				 MCQ_CQ_EVENT_STATUS)
+
+
 /* UIC command timeout, unit: ms */
 #define UIC_CMD_TIMEOUT	500
 
@@ -8252,6 +8258,20 @@ static int ufshcd_alloc_mcq(struct ufs_hba *hba)
 	return ret;
 }
 
+static void ufshcd_config_mcq(struct ufs_hba *hba)
+{
+	ufshcd_enable_intr(hba, UFSHCD_ENABLE_MCQ_INTRS);
+	ufshcd_mcq_make_queues_operational(hba);
+	ufshcd_mcq_config_mac(hba, hba->nutrs);
+
+	hba->host->can_queue = hba->nutrs - UFSHCD_NUM_RESERVED;
+	hba->reserved_slot = hba->nutrs - UFSHCD_NUM_RESERVED;
+	dev_info(hba->dev, "MCQ configured, nr_queues=%d, io_queues=%d, read_queue=%d, poll_queues=%d, queue_depth=%d\n",
+		 hba->nr_hw_queues, hba->nr_queues[HCTX_TYPE_DEFAULT],
+		 hba->nr_queues[HCTX_TYPE_READ], hba->nr_queues[HCTX_TYPE_POLL],
+		 hba->nutrs);
+}
+
 /**
  * ufshcd_probe_hba - probe hba to detect device and initialize it
  * @hba: per-adapter instance
@@ -8281,6 +8301,10 @@ static int ufshcd_probe_hba(struct ufs_hba *hba, bool init_dev_params)
 	/* UniPro link is active now */
 	ufshcd_set_link_active(hba);
 
+	/* Reconfigure MCQ upon reset */
+	if (is_mcq_enabled(hba) && !init_dev_params)
+		ufshcd_config_mcq(hba);
+
 	/* Verify device initialization by sending NOP OUT UPIU */
 	ret = ufshcd_verify_dev_init(hba);
 	if (ret)
@@ -8313,9 +8337,13 @@ static int ufshcd_probe_hba(struct ufs_hba *hba, bool init_dev_params)
 				dev_err(hba->dev, "scsi_add_host failed\n");
 				goto out;
 			}
+			/* MCQ may be disabled if ufshcd_alloc_mcq() fails */
+			if (use_mcq_mode)
+				ufshcd_config_mcq(hba);
 		}
 	}
 
+
 	ufshcd_tune_unipro_params(hba);
 
 	/* UFS device is also active now */
diff --git a/drivers/ufs/host/ufs-qcom.c b/drivers/ufs/host/ufs-qcom.c
index 7bd3c37..dd53e85 100644
--- a/drivers/ufs/host/ufs-qcom.c
+++ b/drivers/ufs/host/ufs-qcom.c
@@ -1425,6 +1425,29 @@ static void ufs_qcom_config_scaling_param(struct ufs_hba *hba,
 }
 #endif
 
+static int ufs_qcom_op_runtime_config(struct ufs_hba *hba)
+{
+	struct ufshcd_res_info *mem_res, *sqdao_res;
+	struct ufshcd_mcq_opr_info_t *opr;
+	int i;
+
+	mem_res = &hba->res[RES_UFS];
+	sqdao_res = &hba->res[RES_MCQ_SQD];
+
+	if (!mem_res->base || !sqdao_res->base)
+		return -EINVAL;
+
+	for (i = 0; i < OPR_MAX; i++) {
+		opr = &hba->mcq_opr[i];
+		opr->offset = sqdao_res->resource->start -
+			      mem_res->resource->start + 0x40 * i;
+		opr->stride = 0x100;
+		opr->base = sqdao_res->base + 0x40 * i;
+	}
+
+	return 0;
+}
+
 static int ufs_qcom_get_hba_mac(struct ufs_hba *hba)
 {
 	/* Qualcomm HC supports up to 64 */
@@ -1455,6 +1478,7 @@ static const struct ufs_hba_variant_ops ufs_hba_qcom_vops = {
 	.config_scaling_param = ufs_qcom_config_scaling_param,
 	.program_key		= ufs_qcom_ice_program_key,
 	.get_hba_mac		= ufs_qcom_get_hba_mac,
+	.op_runtime_config	= ufs_qcom_op_runtime_config,
 };
 
 /**
diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
index e478bab..76972fc 100644
--- a/include/ufs/ufshcd.h
+++ b/include/ufs/ufshcd.h
@@ -298,6 +298,7 @@ struct ufs_pwr_mode_info {
  * @program_key: program or evict an inline encryption key
  * @event_notify: called to notify important events
  * @get_hba_mac: called to get vendor specific mac value, mandatory for mcq mode
+ * @op_runtime_config: called to config Operation and runtime regs Pointers
  */
 struct ufs_hba_variant_ops {
 	const char *name;
@@ -337,6 +338,7 @@ struct ufs_hba_variant_ops {
 	void	(*event_notify)(struct ufs_hba *hba,
 				enum ufs_event_type evt, void *data);
 	int	(*get_hba_mac)(struct ufs_hba *hba);
+	int	(*op_runtime_config)(struct ufs_hba *hba);
 };
 
 /* clock gating state  */
@@ -750,6 +752,27 @@ enum ufshcd_res {
 };
 
 /**
+ * struct ufshcd_mcq_opr_info_t - Operation and Runtime registers
+ *
+ * @offset: Doorbell Address Offset
+ * @stride: Steps proportional to queue [0...31]
+ * @base: base address
+ */
+struct ufshcd_mcq_opr_info_t {
+	unsigned long offset;
+	unsigned long stride;
+	void __iomem *base;
+};
+
+enum ufshcd_mcq_opr {
+	OPR_SQD,
+	OPR_SQIS,
+	OPR_CQD,
+	OPR_CQIS,
+	OPR_MAX,
+};
+
+/**
  * struct ufs_hba - per adapter private structure
  * @mmio_base: UFSHCI base register address
  * @ucdl_base_addr: UFS Command Descriptor base address
@@ -859,6 +882,7 @@ enum ufshcd_res {
  *	ufshcd_resume_complete()
  * @ext_iid_sup: is EXT_IID is supported by UFSHC
  * @mcq_sup: is mcq supported by UFSHC
+ * @mcq_enabled: is mcq ready to accept requests
  * @nr_hw_queues: number of hardware queues configured
  * @nr_queues: number of Queues of different queue types
  * @res: array of resource info of MCQ registers
@@ -1016,29 +1040,47 @@ struct ufs_hba {
 	bool complete_put;
 	bool ext_iid_sup;
 	bool mcq_sup;
+	bool mcq_enabled;
 	unsigned int nr_hw_queues;
 	unsigned int nr_queues[HCTX_MAX_TYPES];
 	struct ufshcd_res_info res[RES_MAX];
 	void __iomem *mcq_base;
 	struct ufs_hw_queue *uhq;
 	struct ufs_hw_queue *dev_cmd_queue;
+	struct ufshcd_mcq_opr_info_t mcq_opr[OPR_MAX];
 };
 
 /**
+ * @mcq_sq_head: base address of submission queue head pointer
+ * @mcq_sq_tail: base address of submission queue tail pointer
+ * @mcq_cq_head: base address of completion queue head pointer
+ * @mcq_cq_tail: base address of completion queue tail pointer
  * @sqe_base_addr: submission queue entry base address
  * @sqe_dma_addr: submission queue dma address
  * @cqe_base_addr: completion queue base address
  * @cqe_dma_addr: completion queue dma address
  * @max_entries: max number of slots in this hardware queue
+ * @id: hardware queue ID
  */
 struct ufs_hw_queue {
+	void __iomem *mcq_sq_head;
+	void __iomem *mcq_sq_tail;
+	void __iomem *mcq_cq_head;
+	void __iomem *mcq_cq_tail;
+
 	void *sqe_base_addr;
 	dma_addr_t sqe_dma_addr;
 	struct cq_entry *cqe_base_addr;
 	dma_addr_t cqe_dma_addr;
 	u32 max_entries;
+	u32 id;
 };
 
+static inline bool is_mcq_enabled(struct ufs_hba *hba)
+{
+	return hba->mcq_enabled;
+}
+
 /* Returns true if clocks can be gated. Otherwise false */
 static inline bool ufshcd_is_clkgating_allowed(struct ufs_hba *hba)
 {
@@ -1094,6 +1136,16 @@ static inline bool ufshcd_enable_wb_if_scaling_up(struct ufs_hba *hba)
 	return hba->caps & UFSHCD_CAP_WB_WITH_CLK_SCALING;
 }
 
+#define ufsmcq_writel(hba, val, reg)	\
+	writel((val), (hba)->mcq_base + (reg))
+#define ufsmcq_readl(hba, reg)	\
+	readl((hba)->mcq_base + (reg))
+
+#define ufsmcq_writelx(hba, val, reg)	\
+	writel_relaxed((val), (hba)->mcq_base + (reg))
+#define ufsmcq_readlx(hba, reg)	\
+	readl_relaxed((hba)->mcq_base + (reg))
+
 #define ufshcd_writel(hba, val, reg)	\
 	writel((val), (hba)->mmio_base + (reg))
 #define ufshcd_readl(hba, reg)	\
diff --git a/include/ufs/ufshci.h b/include/ufs/ufshci.h
index 1aae5b2..c85bdf2 100644
--- a/include/ufs/ufshci.h
+++ b/include/ufs/ufshci.h
@@ -57,6 +57,7 @@ enum {
 	REG_UFS_CCAP				= 0x100,
 	REG_UFS_CRYPTOCAP			= 0x104,
 
+	REG_UFS_MEM_CFG				= 0x300,
 	REG_UFS_MCQ_CFG				= 0x380,
 	UFSHCI_CRYPTO_REG_SPACE_SIZE		= 0x400,
 };
@@ -78,6 +79,35 @@ enum {
 	MASK_EXT_IID_SUPPORT = 0x00000400,
 };
 
+enum {
+	REG_SQATTR		= 0x0,
+	REG_SQLBA		= 0x4,
+	REG_SQUBA		= 0x8,
+	REG_SQDAO		= 0xC,
+	REG_SQISAO		= 0x10,
+
+	REG_CQATTR		= 0x20,
+	REG_CQLBA		= 0x24,
+	REG_CQUBA		= 0x28,
+	REG_CQDAO		= 0x2C,
+	REG_CQISAO		= 0x30,
+};
+
+enum {
+	REG_SQHP		= 0x0,
+	REG_SQTP		= 0x4,
+};
+
+enum {
+	REG_CQHP		= 0x0,
+	REG_CQTP		= 0x4,
+};
+
+enum {
+	REG_CQIS		= 0x0,
+	REG_CQIE		= 0x4,
+};
+
 #define UFS_MASK(mask, offset)		((mask) << (offset))
 
 /* UFS Version 08h */
@@ -134,6 +164,7 @@ static inline u32 ufshci_version(u32 major, u32 minor)
 #define CONTROLLER_FATAL_ERROR			0x10000
 #define SYSTEM_BUS_FATAL_ERROR			0x20000
 #define CRYPTO_ENGINE_FATAL_ERROR		0x40000
+#define MCQ_CQ_EVENT_STATUS			0x100000
 
 #define UFSHCD_UIC_HIBERN8_MASK	(UIC_HIBERNATE_ENTER |\
 				UIC_HIBERNATE_EXIT)
-- 
2.7.4

Re: [PATCH v5 09/16] ufs: core: mcq: Configure operation and runtime interface

[Bart Van Assche]

On 11/22/22 20:10, Asutosh Das wrote:
> +static void ufshcd_config_mcq(struct ufs_hba *hba)
> +{
> +	ufshcd_enable_intr(hba, UFSHCD_ENABLE_MCQ_INTRS);
> +	ufshcd_mcq_make_queues_operational(hba);
> +	ufshcd_mcq_config_mac(hba, hba->nutrs);
> +
> +	hba->host->can_queue = hba->nutrs - UFSHCD_NUM_RESERVED;
> +	hba->reserved_slot = hba->nutrs - UFSHCD_NUM_RESERVED;
> +	dev_info(hba->dev, "MCQ configured, nr_queues=%d, io_queues=%d, read_queue=%d, poll_queues=%d, queue_depth=%d\n",
> +		 hba->nr_hw_queues, hba->nr_queues[HCTX_TYPE_DEFAULT],
> +		 hba->nr_queues[HCTX_TYPE_READ], hba->nr_queues[HCTX_TYPE_POLL],
> +		 hba->nutrs);
> +}

Please do not reserve a slot for device commands in MCQ mode but instead 
queue device management commands on the device management command queue 
in MCQ mode. Please also add WARN_ON_ONCE() statements in the code that 
uses hba->reserved_slot to make sure that that variable is not used in 
MCQ mode.

Thanks,

Bart.

Re: [PATCH v5 09/16] ufs: core: mcq: Configure operation and runtime interface

[Asutosh Das]

On Fri, Nov 25 2022 at 17:28 -0800, Bart Van Assche wrote:
>On 11/22/22 20:10, Asutosh Das wrote:
>>+static void ufshcd_config_mcq(struct ufs_hba *hba)
>>+{
>>+	ufshcd_enable_intr(hba, UFSHCD_ENABLE_MCQ_INTRS);
>>+	ufshcd_mcq_make_queues_operational(hba);
>>+	ufshcd_mcq_config_mac(hba, hba->nutrs);
>>+
>>+	hba->host->can_queue = hba->nutrs - UFSHCD_NUM_RESERVED;
>>+	hba->reserved_slot = hba->nutrs - UFSHCD_NUM_RESERVED;
>>+	dev_info(hba->dev, "MCQ configured, nr_queues=%d, io_queues=%d, read_queue=%d, poll_queues=%d, queue_depth=%d\n",
>>+		 hba->nr_hw_queues, hba->nr_queues[HCTX_TYPE_DEFAULT],
>>+		 hba->nr_queues[HCTX_TYPE_READ], hba->nr_queues[HCTX_TYPE_POLL],
>>+		 hba->nutrs);
>>+}
>
Hello Bart,

>Please do not reserve a slot for device commands in MCQ mode but 
>instead queue device management commands on the device management 
>command queue in MCQ mode. Please also add WARN_ON_ONCE() statements 
>in the code that uses hba->reserved_slot to make sure that that 
>variable is not used in MCQ mode.

Now it is queued to device management command queue in MCQ mode.
However, a task tag is needed for device management commands.
Do you mean that the task tag of device management command be default to 0
instead in MCQ mode?

>
>Thanks,
>
>Bart.

Re: [PATCH v5 09/16] ufs: core: mcq: Configure operation and runtime interface

[Bart Van Assche]

On 11/28/22 14:08, Asutosh Das wrote:
> Now it is queued to device management command queue in MCQ mode.
> However, a task tag is needed for device management commands.
> Do you mean that the task tag of device management command be default to 0
> instead in MCQ mode?

Right, we are using shared tags and that is why we need to reserve a tag 
for device management commands. Please ignore my previous comment.

Bart.

Re: [PATCH v5 09/16] ufs: core: mcq: Configure operation and runtime interface

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:22PM -0800, Asutosh Das wrote:
> Runtime and operation registers are defined per Submission
> and Completion queue.
> The location of these registers is not defined in the spec;
> meaning the offsets and stride may vary for different
> HC vendors. Establish the stride, base address and doorbell
> address offsets from vendor host driver and program it.
> 
> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>

Reviewed-by: Manivannan Sadhasivam <mani@kernel.org>

Thanks,
Mani

> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
> ---
>  drivers/ufs/core/ufs-mcq.c     | 101 +++++++++++++++++++++++++++++++++++++++++
>  drivers/ufs/core/ufshcd-priv.h |  11 +++++
>  drivers/ufs/core/ufshcd.c      |  28 ++++++++++++
>  drivers/ufs/host/ufs-qcom.c    |  24 ++++++++++
>  include/ufs/ufshcd.h           |  52 +++++++++++++++++++++
>  include/ufs/ufshci.h           |  31 +++++++++++++
>  6 files changed, 247 insertions(+)
> 
> diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
> index 51f0e40..ebecc47 100644
> --- a/drivers/ufs/core/ufs-mcq.c
> +++ b/drivers/ufs/core/ufs-mcq.c
> @@ -17,6 +17,8 @@
>  #define UFS_MCQ_MIN_READ_QUEUES 0
>  #define UFS_MCQ_NUM_DEV_CMD_QUEUES 1
>  #define UFS_MCQ_MIN_POLL_QUEUES 0
> +#define QUEUE_EN_OFFSET 31
> +#define QUEUE_ID_OFFSET 16
>  
>  #define MAX_DEV_CMD_ENTRIES	2
>  #define MCQ_CFG_MAC_MASK	GENMASK(16, 8)
> @@ -25,6 +27,7 @@
>  #define MCQ_SQATTR_OFFSET(c) \
>  	((((c) >> 16) & MCQ_QCFGPTR_MASK) * MCQ_QCFGPTR_UNIT)
>  #define MCQ_QCFG_SIZE	0x40
> +#define MCQ_ENTRY_SIZE_IN_DWORD	8
>  
>  static int rw_queue_count_set(const char *val, const struct kernel_param *kp)
>  {
> @@ -120,6 +123,24 @@ int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba)
>  	return min_t(int, mac, hba->dev_info.bqueuedepth);
>  }
>  
> +/**
> + * ufshcd_mcq_config_mac - Set the #Max Activ Cmds.
> + * @hba - per adapter instance
> + * @max_active_cmds - maximum # of active commands to the device at any time.
> + *
> + * The controller won't send more than the max_active_cmds to the device at
> + * any time.
> + */
> +void ufshcd_mcq_config_mac(struct ufs_hba *hba, u32 max_active_cmds)
> +{
> +	u32 val;
> +
> +	val = ufshcd_readl(hba, REG_UFS_MCQ_CFG);
> +	val &= ~MCQ_CFG_MAC_MASK;
> +	val |= FIELD_PREP(MCQ_CFG_MAC_MASK, max_active_cmds);
> +	ufshcd_writel(hba, val, REG_UFS_MCQ_CFG);
> +}
> +
>  static int ufshcd_mcq_config_resource(struct ufs_hba *hba)
>  {
>  	struct platform_device *pdev = to_platform_device(hba->dev);
> @@ -279,6 +300,80 @@ int ufshcd_mcq_memory_alloc(struct ufs_hba *hba)
>  	return 0;
>  }
>  
> +/* Operation and runtime registers configuration */
> +#define MCQ_CFG_n(r, i)	((r) + MCQ_QCFG_SIZE * (i))
> +#define MCQ_OPR_OFFSET_n(p, i) \
> +	(hba->mcq_opr[(p)].offset + hba->mcq_opr[(p)].stride * (i))
> +
> +static void __iomem *mcq_opr_base(struct ufs_hba *hba,
> +					 enum ufshcd_mcq_opr n, int i)
> +{
> +	struct ufshcd_mcq_opr_info_t *opr = &hba->mcq_opr[n];
> +
> +	return opr->base + opr->stride * i;
> +}
> +
> +void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba)
> +{
> +	struct ufs_hw_queue *hwq;
> +	u16 qsize;
> +	int i;
> +
> +	for (i = 0; i < hba->nr_hw_queues; i++) {
> +		hwq = &hba->uhq[i];
> +		hwq->id = i;
> +		qsize = hwq->max_entries * MCQ_ENTRY_SIZE_IN_DWORD - 1;
> +
> +		/* Submission Queue Lower Base Address */
> +		ufsmcq_writelx(hba, lower_32_bits(hwq->sqe_dma_addr),
> +			      MCQ_CFG_n(REG_SQLBA, i));
> +		/* Submission Queue Upper Base Address */
> +		ufsmcq_writelx(hba, upper_32_bits(hwq->sqe_dma_addr),
> +			      MCQ_CFG_n(REG_SQUBA, i));
> +		/* Submission Queue Doorbell Address Offset */
> +		ufsmcq_writelx(hba, MCQ_OPR_OFFSET_n(OPR_SQD, i),
> +			      MCQ_CFG_n(REG_SQDAO, i));
> +		/* Submission Queue Interrupt Status Address Offset */
> +		ufsmcq_writelx(hba, MCQ_OPR_OFFSET_n(OPR_SQIS, i),
> +			      MCQ_CFG_n(REG_SQISAO, i));
> +
> +		/* Completion Queue Lower Base Address */
> +		ufsmcq_writelx(hba, lower_32_bits(hwq->cqe_dma_addr),
> +			      MCQ_CFG_n(REG_CQLBA, i));
> +		/* Completion Queue Upper Base Address */
> +		ufsmcq_writelx(hba, upper_32_bits(hwq->cqe_dma_addr),
> +			      MCQ_CFG_n(REG_CQUBA, i));
> +		/* Completion Queue Doorbell Address Offset */
> +		ufsmcq_writelx(hba, MCQ_OPR_OFFSET_n(OPR_CQD, i),
> +			      MCQ_CFG_n(REG_CQDAO, i));
> +		/* Completion Queue Interrupt Status Address Offset */
> +		ufsmcq_writelx(hba, MCQ_OPR_OFFSET_n(OPR_CQIS, i),
> +			      MCQ_CFG_n(REG_CQISAO, i));
> +
> +		/* Save the base addresses for quicker access */
> +		hwq->mcq_sq_head = mcq_opr_base(hba, OPR_SQD, i) + REG_SQHP;
> +		hwq->mcq_sq_tail = mcq_opr_base(hba, OPR_SQD, i) + REG_SQTP;
> +		hwq->mcq_cq_head = mcq_opr_base(hba, OPR_CQD, i) + REG_CQHP;
> +		hwq->mcq_cq_tail = mcq_opr_base(hba, OPR_CQD, i) + REG_CQTP;
> +
> +		/* Enable Tail Entry Push Status interrupt only for non-poll queues */
> +		if (i < hba->nr_hw_queues - hba->nr_queues[HCTX_TYPE_POLL])
> +			writel(1, mcq_opr_base(hba, OPR_CQIS, i) + REG_CQIE);
> +
> +		/* Completion Queue Enable|Size to Completion Queue Attribute */
> +		ufsmcq_writel(hba, (1 << QUEUE_EN_OFFSET) | qsize,
> +			      MCQ_CFG_n(REG_CQATTR, i));
> +
> +		/*
> +		 * Submission Qeueue Enable|Size|Completion Queue ID to
> +		 * Submission Queue Attribute
> +		 */
> +		ufsmcq_writel(hba, (1 << QUEUE_EN_OFFSET) | qsize |
> +			      (i << QUEUE_ID_OFFSET),
> +			      MCQ_CFG_n(REG_SQATTR, i));
> +	}
> +}
> +
>  int ufshcd_mcq_init(struct ufs_hba *hba)
>  {
>  	struct ufs_hw_queue *hwq;
> @@ -292,6 +387,12 @@ int ufshcd_mcq_init(struct ufs_hba *hba)
>  	if (ret)
>  		return ret;
>  
> +	ret = ufshcd_mcq_vops_op_runtime_config(hba);
> +	if (ret) {
> +		dev_err(hba->dev, "Operation runtime config failed, ret=%d\n",
> +			ret);
> +		return ret;
> +	}
>  	hba->uhq = devm_kzalloc(hba->dev,
>  				hba->nr_hw_queues * sizeof(struct ufs_hw_queue),
>  				GFP_KERNEL);
> diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
> index 4d2bde2..08963e1 100644
> --- a/drivers/ufs/core/ufshcd-priv.h
> +++ b/drivers/ufs/core/ufshcd-priv.h
> @@ -64,6 +64,9 @@ void ufshcd_auto_hibern8_update(struct ufs_hba *hba, u32 ahit);
>  int ufshcd_mcq_init(struct ufs_hba *hba);
>  int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba);
>  int ufshcd_mcq_memory_alloc(struct ufs_hba *hba);
> +void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba);
> +void ufshcd_mcq_config_mac(struct ufs_hba *hba, u32 max_active_cmds);
> +void ufshcd_mcq_select_mcq_mode(struct ufs_hba *hba);
>  
>  #define SD_ASCII_STD true
>  #define SD_RAW false
> @@ -237,6 +240,14 @@ static inline int ufshcd_mcq_vops_get_hba_mac(struct ufs_hba *hba)
>  	return -EOPNOTSUPP;
>  }
>  
> +static inline int ufshcd_mcq_vops_op_runtime_config(struct ufs_hba *hba)
> +{
> +	if (hba->vops && hba->vops->op_runtime_config)
> +		return hba->vops->op_runtime_config(hba);
> +
> +	return -EOPNOTSUPP;
> +}
> +
>  extern const struct ufs_pm_lvl_states ufs_pm_lvl_states[];
>  
>  /**
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 45686e8..042ecf04 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -43,6 +43,12 @@
>  #define UFSHCD_ENABLE_INTRS	(UTP_TRANSFER_REQ_COMPL |\
>  				 UTP_TASK_REQ_COMPL |\
>  				 UFSHCD_ERROR_MASK)
> +
> +#define UFSHCD_ENABLE_MCQ_INTRS	(UTP_TASK_REQ_COMPL |\
> +				 UFSHCD_ERROR_MASK |\
> +				 MCQ_CQ_EVENT_STATUS)
> +
> +
>  /* UIC command timeout, unit: ms */
>  #define UIC_CMD_TIMEOUT	500
>  
> @@ -8252,6 +8258,20 @@ static int ufshcd_alloc_mcq(struct ufs_hba *hba)
>  	return ret;
>  }
>  
> +static void ufshcd_config_mcq(struct ufs_hba *hba)
> +{
> +	ufshcd_enable_intr(hba, UFSHCD_ENABLE_MCQ_INTRS);
> +	ufshcd_mcq_make_queues_operational(hba);
> +	ufshcd_mcq_config_mac(hba, hba->nutrs);
> +
> +	hba->host->can_queue = hba->nutrs - UFSHCD_NUM_RESERVED;
> +	hba->reserved_slot = hba->nutrs - UFSHCD_NUM_RESERVED;
> +	dev_info(hba->dev, "MCQ configured, nr_queues=%d, io_queues=%d, read_queue=%d, poll_queues=%d, queue_depth=%d\n",
> +		 hba->nr_hw_queues, hba->nr_queues[HCTX_TYPE_DEFAULT],
> +		 hba->nr_queues[HCTX_TYPE_READ], hba->nr_queues[HCTX_TYPE_POLL],
> +		 hba->nutrs);
> +}
> +
>  /**
>   * ufshcd_probe_hba - probe hba to detect device and initialize it
>   * @hba: per-adapter instance
> @@ -8281,6 +8301,10 @@ static int ufshcd_probe_hba(struct ufs_hba *hba, bool init_dev_params)
>  	/* UniPro link is active now */
>  	ufshcd_set_link_active(hba);
>  
> +	/* Reconfigure MCQ upon reset */
> +	if (is_mcq_enabled(hba) && !init_dev_params)
> +		ufshcd_config_mcq(hba);
> +
>  	/* Verify device initialization by sending NOP OUT UPIU */
>  	ret = ufshcd_verify_dev_init(hba);
>  	if (ret)
> @@ -8313,9 +8337,13 @@ static int ufshcd_probe_hba(struct ufs_hba *hba, bool init_dev_params)
>  				dev_err(hba->dev, "scsi_add_host failed\n");
>  				goto out;
>  			}
> +			/* MCQ may be disabled if ufshcd_alloc_mcq() fails */
> +			if (use_mcq_mode)
> +				ufshcd_config_mcq(hba);
>  		}
>  	}
>  
> +
>  	ufshcd_tune_unipro_params(hba);
>  
>  	/* UFS device is also active now */
> diff --git a/drivers/ufs/host/ufs-qcom.c b/drivers/ufs/host/ufs-qcom.c
> index 7bd3c37..dd53e85 100644
> --- a/drivers/ufs/host/ufs-qcom.c
> +++ b/drivers/ufs/host/ufs-qcom.c
> @@ -1425,6 +1425,29 @@ static void ufs_qcom_config_scaling_param(struct ufs_hba *hba,
>  }
>  #endif
>  
> +static int ufs_qcom_op_runtime_config(struct ufs_hba *hba)
> +{
> +	struct ufshcd_res_info *mem_res, *sqdao_res;
> +	struct ufshcd_mcq_opr_info_t *opr;
> +	int i;
> +
> +	mem_res = &hba->res[RES_UFS];
> +	sqdao_res = &hba->res[RES_MCQ_SQD];
> +
> +	if (!mem_res->base || !sqdao_res->base)
> +		return -EINVAL;
> +
> +	for (i = 0; i < OPR_MAX; i++) {
> +		opr = &hba->mcq_opr[i];
> +		opr->offset = sqdao_res->resource->start -
> +			      mem_res->resource->start + 0x40 * i;
> +		opr->stride = 0x100;
> +		opr->base = sqdao_res->base + 0x40 * i;
> +	}
> +
> +	return 0;
> +}
> +
>  static int ufs_qcom_get_hba_mac(struct ufs_hba *hba)
>  {
>  	/* Qualcomm HC supports up to 64 */
> @@ -1455,6 +1478,7 @@ static const struct ufs_hba_variant_ops ufs_hba_qcom_vops = {
>  	.config_scaling_param = ufs_qcom_config_scaling_param,
>  	.program_key		= ufs_qcom_ice_program_key,
>  	.get_hba_mac		= ufs_qcom_get_hba_mac,
> +	.op_runtime_config	= ufs_qcom_op_runtime_config,
>  };
>  
>  /**
> diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
> index e478bab..76972fc 100644
> --- a/include/ufs/ufshcd.h
> +++ b/include/ufs/ufshcd.h
> @@ -298,6 +298,7 @@ struct ufs_pwr_mode_info {
>   * @program_key: program or evict an inline encryption key
>   * @event_notify: called to notify important events
>   * @get_hba_mac: called to get vendor specific mac value, mandatory for mcq mode
> + * @op_runtime_config: called to config Operation and runtime regs Pointers
>   */
>  struct ufs_hba_variant_ops {
>  	const char *name;
> @@ -337,6 +338,7 @@ struct ufs_hba_variant_ops {
>  	void	(*event_notify)(struct ufs_hba *hba,
>  				enum ufs_event_type evt, void *data);
>  	int	(*get_hba_mac)(struct ufs_hba *hba);
> +	int	(*op_runtime_config)(struct ufs_hba *hba);
>  };
>  
>  /* clock gating state  */
> @@ -750,6 +752,27 @@ enum ufshcd_res {
>  };
>  
>  /**
> + * struct ufshcd_mcq_opr_info_t - Operation and Runtime registers
> + *
> + * @offset: Doorbell Address Offset
> + * @stride: Steps proportional to queue [0...31]
> + * @base: base address
> + */
> +struct ufshcd_mcq_opr_info_t {
> +	unsigned long offset;
> +	unsigned long stride;
> +	void __iomem *base;
> +};
> +
> +enum ufshcd_mcq_opr {
> +	OPR_SQD,
> +	OPR_SQIS,
> +	OPR_CQD,
> +	OPR_CQIS,
> +	OPR_MAX,
> +};
> +
> +/**
>   * struct ufs_hba - per adapter private structure
>   * @mmio_base: UFSHCI base register address
>   * @ucdl_base_addr: UFS Command Descriptor base address
> @@ -859,6 +882,7 @@ enum ufshcd_res {
>   *	ufshcd_resume_complete()
>   * @ext_iid_sup: is EXT_IID is supported by UFSHC
>   * @mcq_sup: is mcq supported by UFSHC
> + * @mcq_enabled: is mcq ready to accept requests
>   * @nr_hw_queues: number of hardware queues configured
>   * @nr_queues: number of Queues of different queue types
>   * @res: array of resource info of MCQ registers
> @@ -1016,29 +1040,47 @@ struct ufs_hba {
>  	bool complete_put;
>  	bool ext_iid_sup;
>  	bool mcq_sup;
> +	bool mcq_enabled;
>  	unsigned int nr_hw_queues;
>  	unsigned int nr_queues[HCTX_MAX_TYPES];
>  	struct ufshcd_res_info res[RES_MAX];
>  	void __iomem *mcq_base;
>  	struct ufs_hw_queue *uhq;
>  	struct ufs_hw_queue *dev_cmd_queue;
> +	struct ufshcd_mcq_opr_info_t mcq_opr[OPR_MAX];
>  };
>  
>  /**
> + * @mcq_sq_head: base address of submission queue head pointer
> + * @mcq_sq_tail: base address of submission queue tail pointer
> + * @mcq_cq_head: base address of completion queue head pointer
> + * @mcq_cq_tail: base address of completion queue tail pointer
>   * @sqe_base_addr: submission queue entry base address
>   * @sqe_dma_addr: submission queue dma address
>   * @cqe_base_addr: completion queue base address
>   * @cqe_dma_addr: completion queue dma address
>   * @max_entries: max number of slots in this hardware queue
> + * @id: hardware queue ID
>   */
>  struct ufs_hw_queue {
> +	void __iomem *mcq_sq_head;
> +	void __iomem *mcq_sq_tail;
> +	void __iomem *mcq_cq_head;
> +	void __iomem *mcq_cq_tail;
> +
>  	void *sqe_base_addr;
>  	dma_addr_t sqe_dma_addr;
>  	struct cq_entry *cqe_base_addr;
>  	dma_addr_t cqe_dma_addr;
>  	u32 max_entries;
> +	u32 id;
>  };
>  
> +static inline bool is_mcq_enabled(struct ufs_hba *hba)
> +{
> +	return hba->mcq_enabled;
> +}
> +
>  /* Returns true if clocks can be gated. Otherwise false */
>  static inline bool ufshcd_is_clkgating_allowed(struct ufs_hba *hba)
>  {
> @@ -1094,6 +1136,16 @@ static inline bool ufshcd_enable_wb_if_scaling_up(struct ufs_hba *hba)
>  	return hba->caps & UFSHCD_CAP_WB_WITH_CLK_SCALING;
>  }
>  
> +#define ufsmcq_writel(hba, val, reg)	\
> +	writel((val), (hba)->mcq_base + (reg))
> +#define ufsmcq_readl(hba, reg)	\
> +	readl((hba)->mcq_base + (reg))
> +
> +#define ufsmcq_writelx(hba, val, reg)	\
> +	writel_relaxed((val), (hba)->mcq_base + (reg))
> +#define ufsmcq_readlx(hba, reg)	\
> +	readl_relaxed((hba)->mcq_base + (reg))
> +
>  #define ufshcd_writel(hba, val, reg)	\
>  	writel((val), (hba)->mmio_base + (reg))
>  #define ufshcd_readl(hba, reg)	\
> diff --git a/include/ufs/ufshci.h b/include/ufs/ufshci.h
> index 1aae5b2..c85bdf2 100644
> --- a/include/ufs/ufshci.h
> +++ b/include/ufs/ufshci.h
> @@ -57,6 +57,7 @@ enum {
>  	REG_UFS_CCAP				= 0x100,
>  	REG_UFS_CRYPTOCAP			= 0x104,
>  
> +	REG_UFS_MEM_CFG				= 0x300,
>  	REG_UFS_MCQ_CFG				= 0x380,
>  	UFSHCI_CRYPTO_REG_SPACE_SIZE		= 0x400,
>  };
> @@ -78,6 +79,35 @@ enum {
>  	MASK_EXT_IID_SUPPORT = 0x00000400,
>  };
>  
> +enum {
> +	REG_SQATTR		= 0x0,
> +	REG_SQLBA		= 0x4,
> +	REG_SQUBA		= 0x8,
> +	REG_SQDAO		= 0xC,
> +	REG_SQISAO		= 0x10,
> +
> +	REG_CQATTR		= 0x20,
> +	REG_CQLBA		= 0x24,
> +	REG_CQUBA		= 0x28,
> +	REG_CQDAO		= 0x2C,
> +	REG_CQISAO		= 0x30,
> +};
> +
> +enum {
> +	REG_SQHP		= 0x0,
> +	REG_SQTP		= 0x4,
> +};
> +
> +enum {
> +	REG_CQHP		= 0x0,
> +	REG_CQTP		= 0x4,
> +};
> +
> +enum {
> +	REG_CQIS		= 0x0,
> +	REG_CQIE		= 0x4,
> +};
> +
>  #define UFS_MASK(mask, offset)		((mask) << (offset))
>  
>  /* UFS Version 08h */
> @@ -134,6 +164,7 @@ static inline u32 ufshci_version(u32 major, u32 minor)
>  #define CONTROLLER_FATAL_ERROR			0x10000
>  #define SYSTEM_BUS_FATAL_ERROR			0x20000
>  #define CRYPTO_ENGINE_FATAL_ERROR		0x40000
> +#define MCQ_CQ_EVENT_STATUS			0x100000
>  
>  #define UFSHCD_UIC_HIBERN8_MASK	(UIC_HIBERNATE_ENTER |\
>  				UIC_HIBERNATE_EXIT)
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

[PATCH v5 10/16] ufs: core: mcq: Use shared tags for MCQ mode

[Asutosh Das]

Enable shared tags for MCQ. For UFS, this should
not have a huge performance impact. It however
simplifies the MCQ implementation and reuses most of
the existing code in the issue and completion path.
Also add multiple queue mapping to map_queue().

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/ufs/core/ufs-mcq.c |  2 ++
 drivers/ufs/core/ufshcd.c  | 28 ++++++++++++++++------------
 2 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
index ebecc47..e4ddb90 100644
--- a/drivers/ufs/core/ufs-mcq.c
+++ b/drivers/ufs/core/ufs-mcq.c
@@ -376,6 +376,7 @@ void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba)
 
 int ufshcd_mcq_init(struct ufs_hba *hba)
 {
+	struct Scsi_Host *host = hba->host;
 	struct ufs_hw_queue *hwq;
 	int ret, i;
 
@@ -411,6 +412,7 @@ int ufshcd_mcq_init(struct ufs_hba *hba)
 	/* Give dev_cmd_queue the minimal number of entries */
 	hba->dev_cmd_queue->max_entries = MAX_DEV_CMD_ENTRIES;
 
+	host->host_tagset = 1;
 	return 0;
 }
 
diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 042ecf04..d61e99f 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -2763,24 +2763,28 @@ static inline bool is_device_wlun(struct scsi_device *sdev)
  */
 static void ufshcd_map_queues(struct Scsi_Host *shost)
 {
-	int i;
+	struct ufs_hba *hba = shost_priv(shost);
+	int i, queue_offset = 0;
+
+	if (!is_mcq_supported(hba)) {
+		hba->nr_queues[HCTX_TYPE_DEFAULT] = 1;
+		hba->nr_queues[HCTX_TYPE_READ] = 0;
+		hba->nr_queues[HCTX_TYPE_POLL] = 1;
+		hba->nr_hw_queues = 1;
+	}
 
 	for (i = 0; i < shost->nr_maps; i++) {
 		struct blk_mq_queue_map *map = &shost->tag_set.map[i];
 
-		switch (i) {
-		case HCTX_TYPE_DEFAULT:
-		case HCTX_TYPE_POLL:
-			map->nr_queues = 1;
-			break;
-		case HCTX_TYPE_READ:
-			map->nr_queues = 0;
+		map->nr_queues = hba->nr_queues[i];
+		if (!map->nr_queues)
 			continue;
-		default:
-			WARN_ON_ONCE(true);
-		}
-		map->queue_offset = 0;
+		map->queue_offset = queue_offset;
+		if (i == HCTX_TYPE_POLL && !is_mcq_supported(hba))
+			map->queue_offset = 0;
+
 		blk_mq_map_queues(map);
+		queue_offset += map->nr_queues;
 	}
 }
 
-- 
2.7.4

Re: [PATCH v5 10/16] ufs: core: mcq: Use shared tags for MCQ mode

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:23PM -0800, Asutosh Das wrote:
> Enable shared tags for MCQ. For UFS, this should
> not have a huge performance impact. It however
> simplifies the MCQ implementation and reuses most of
> the existing code in the issue and completion path.
> Also add multiple queue mapping to map_queue().
> 
> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>

Reviewed-by: Manivannan Sadhasivam <mani@kernel.org>

Thanks,
Mani

> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
> ---
>  drivers/ufs/core/ufs-mcq.c |  2 ++
>  drivers/ufs/core/ufshcd.c  | 28 ++++++++++++++++------------
>  2 files changed, 18 insertions(+), 12 deletions(-)
> 
> diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
> index ebecc47..e4ddb90 100644
> --- a/drivers/ufs/core/ufs-mcq.c
> +++ b/drivers/ufs/core/ufs-mcq.c
> @@ -376,6 +376,7 @@ void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba)
>  
>  int ufshcd_mcq_init(struct ufs_hba *hba)
>  {
> +	struct Scsi_Host *host = hba->host;
>  	struct ufs_hw_queue *hwq;
>  	int ret, i;
>  
> @@ -411,6 +412,7 @@ int ufshcd_mcq_init(struct ufs_hba *hba)
>  	/* Give dev_cmd_queue the minimal number of entries */
>  	hba->dev_cmd_queue->max_entries = MAX_DEV_CMD_ENTRIES;
>  
> +	host->host_tagset = 1;
>  	return 0;
>  }
>  
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 042ecf04..d61e99f 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -2763,24 +2763,28 @@ static inline bool is_device_wlun(struct scsi_device *sdev)
>   */
>  static void ufshcd_map_queues(struct Scsi_Host *shost)
>  {
> -	int i;
> +	struct ufs_hba *hba = shost_priv(shost);
> +	int i, queue_offset = 0;
> +
> +	if (!is_mcq_supported(hba)) {
> +		hba->nr_queues[HCTX_TYPE_DEFAULT] = 1;
> +		hba->nr_queues[HCTX_TYPE_READ] = 0;
> +		hba->nr_queues[HCTX_TYPE_POLL] = 1;
> +		hba->nr_hw_queues = 1;
> +	}
>  
>  	for (i = 0; i < shost->nr_maps; i++) {
>  		struct blk_mq_queue_map *map = &shost->tag_set.map[i];
>  
> -		switch (i) {
> -		case HCTX_TYPE_DEFAULT:
> -		case HCTX_TYPE_POLL:
> -			map->nr_queues = 1;
> -			break;
> -		case HCTX_TYPE_READ:
> -			map->nr_queues = 0;
> +		map->nr_queues = hba->nr_queues[i];
> +		if (!map->nr_queues)
>  			continue;
> -		default:
> -			WARN_ON_ONCE(true);
> -		}
> -		map->queue_offset = 0;
> +		map->queue_offset = queue_offset;
> +		if (i == HCTX_TYPE_POLL && !is_mcq_supported(hba))
> +			map->queue_offset = 0;
> +
>  		blk_mq_map_queues(map);
> +		queue_offset += map->nr_queues;
>  	}
>  }
>  
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

[PATCH v5 11/16] ufs: core: Prepare ufshcd_send_command for mcq

[Asutosh Das]

Add support to send commands using multiple submission
queues in MCQ mode.
Modify the functions that use ufshcd_send_command().

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/ufs/core/ufs-mcq.c     |  1 +
 drivers/ufs/core/ufshcd-priv.h | 10 ++++++++++
 drivers/ufs/core/ufshcd.c      | 36 ++++++++++++++++++++++++++----------
 include/ufs/ufshcd.h           |  5 +++++
 4 files changed, 42 insertions(+), 10 deletions(-)

diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
index e4ddb90..7179f86 100644
--- a/drivers/ufs/core/ufs-mcq.c
+++ b/drivers/ufs/core/ufs-mcq.c
@@ -405,6 +405,7 @@ int ufshcd_mcq_init(struct ufs_hba *hba)
 	for (i = 0; i < hba->nr_hw_queues; i++) {
 		hwq = &hba->uhq[i];
 		hwq->max_entries = hba->nutrs;
+		spin_lock_init(&hwq->sq_lock);
 	}
 
 	/* The very first HW queue serves device commands */
diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
index 08963e1..5616047 100644
--- a/drivers/ufs/core/ufshcd-priv.h
+++ b/drivers/ufs/core/ufshcd-priv.h
@@ -324,4 +324,14 @@ static inline bool ufs_is_valid_unit_desc_lun(struct ufs_dev_info *dev_info, u8
 	return lun == UFS_UPIU_RPMB_WLUN || (lun < dev_info->max_lu_supported);
 }
 
+static inline void ufshcd_inc_sq_tail(struct ufs_hw_queue *q)
+{
+	u32 mask = q->max_entries - 1;
+	u32 val;
+
+	q->sq_tail_slot = (q->sq_tail_slot + 1) & mask;
+	val = q->sq_tail_slot * sizeof(struct utp_transfer_req_desc);
+	writel(val, q->mcq_sq_tail);
+}
+
 #endif /* _UFSHCD_PRIV_H_ */
diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index d61e99f..93a9e38 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -2182,9 +2182,11 @@ static void ufshcd_update_monitor(struct ufs_hba *hba, const struct ufshcd_lrb *
  * ufshcd_send_command - Send SCSI or device management commands
  * @hba: per adapter instance
  * @task_tag: Task tag of the command
+ * @hwq: pointer to hardware queue instance
  */
 static inline
-void ufshcd_send_command(struct ufs_hba *hba, unsigned int task_tag)
+void ufshcd_send_command(struct ufs_hba *hba, unsigned int task_tag,
+			 struct ufs_hw_queue *hwq)
 {
 	struct ufshcd_lrb *lrbp = &hba->lrb[task_tag];
 	unsigned long flags;
@@ -2198,12 +2200,24 @@ void ufshcd_send_command(struct ufs_hba *hba, unsigned int task_tag)
 	if (unlikely(ufshcd_should_inform_monitor(hba, lrbp)))
 		ufshcd_start_monitor(hba, lrbp);
 
-	spin_lock_irqsave(&hba->outstanding_lock, flags);
-	if (hba->vops && hba->vops->setup_xfer_req)
-		hba->vops->setup_xfer_req(hba, task_tag, !!lrbp->cmd);
-	__set_bit(task_tag, &hba->outstanding_reqs);
-	ufshcd_writel(hba, 1 << task_tag, REG_UTP_TRANSFER_REQ_DOOR_BELL);
-	spin_unlock_irqrestore(&hba->outstanding_lock, flags);
+	if (is_mcq_enabled(hba)) {
+		int utrd_size = sizeof(struct utp_transfer_req_desc);
+
+		spin_lock(&hwq->sq_lock);
+		memcpy(hwq->sqe_base_addr + (hwq->sq_tail_slot * utrd_size),
+		       lrbp->utr_descriptor_ptr, utrd_size);
+		ufshcd_inc_sq_tail(hwq);
+		spin_unlock(&hwq->sq_lock);
+	} else {
+		spin_lock_irqsave(&hba->outstanding_lock, flags);
+		if (hba->vops && hba->vops->setup_xfer_req)
+			hba->vops->setup_xfer_req(hba, lrbp->task_tag,
+						  !!lrbp->cmd);
+		__set_bit(lrbp->task_tag, &hba->outstanding_reqs);
+		ufshcd_writel(hba, 1 << lrbp->task_tag,
+			      REG_UTP_TRANSFER_REQ_DOOR_BELL);
+		spin_unlock_irqrestore(&hba->outstanding_lock, flags);
+	}
 }
 
 /**
@@ -2822,6 +2836,7 @@ static int ufshcd_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *cmd)
 	int tag = scsi_cmd_to_rq(cmd)->tag;
 	struct ufshcd_lrb *lrbp;
 	int err = 0;
+	struct ufs_hw_queue *hwq = NULL;
 
 	WARN_ONCE(tag < 0 || tag >= hba->nutrs, "Invalid tag %d\n", tag);
 
@@ -2906,7 +2921,7 @@ static int ufshcd_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *cmd)
 		goto out;
 	}
 
-	ufshcd_send_command(hba, tag);
+	ufshcd_send_command(hba, tag, hwq);
 
 out:
 	rcu_read_unlock();
@@ -3101,10 +3116,11 @@ static int ufshcd_exec_dev_cmd(struct ufs_hba *hba,
 		goto out;
 
 	hba->dev_cmd.complete = &wait;
+	hba->dev_cmd.cqe = NULL;
 
 	ufshcd_add_query_upiu_trace(hba, UFS_QUERY_SEND, lrbp->ucd_req_ptr);
 
-	ufshcd_send_command(hba, tag);
+	ufshcd_send_command(hba, tag, hba->dev_cmd_queue);
 	err = ufshcd_wait_for_dev_cmd(hba, lrbp, timeout);
 	ufshcd_add_query_upiu_trace(hba, err ? UFS_QUERY_ERR : UFS_QUERY_COMP,
 				    (struct utp_upiu_req *)lrbp->ucd_rsp_ptr);
@@ -6926,7 +6942,7 @@ static int ufshcd_issue_devman_upiu_cmd(struct ufs_hba *hba,
 
 	ufshcd_add_query_upiu_trace(hba, UFS_QUERY_SEND, lrbp->ucd_req_ptr);
 
-	ufshcd_send_command(hba, tag);
+	ufshcd_send_command(hba, tag, hba->dev_cmd_queue);
 	/*
 	 * ignore the returning value here - ufshcd_check_query_response is
 	 * bound to fail since dev_cmd.query and dev_cmd.type were left empty.
diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
index 76972fc..853c525 100644
--- a/include/ufs/ufshcd.h
+++ b/include/ufs/ufshcd.h
@@ -222,6 +222,7 @@ struct ufs_dev_cmd {
 	struct mutex lock;
 	struct completion *complete;
 	struct ufs_query query;
+	struct cq_entry *cqe;
 };
 
 /**
@@ -1061,6 +1062,8 @@ struct ufs_hba {
  * @cqe_dma_addr: completion queue dma address
  * @max_entries: max number of slots in this hardware queue
  * @id: hardware queue ID
+ * @sq_tp_slot: current slot to which SQ tail pointer is pointing
+ * @sq_lock: serialize submission queue access
  */
 struct ufs_hw_queue {
 	void __iomem *mcq_sq_head;
@@ -1074,6 +1077,8 @@ struct ufs_hw_queue {
 	dma_addr_t cqe_dma_addr;
 	u32 max_entries;
 	u32 id;
+	u32 sq_tail_slot;
+	spinlock_t sq_lock;
 };
 
 static inline bool is_mcq_enabled(struct ufs_hba *hba)
-- 
2.7.4

Re: [PATCH v5 11/16] ufs: core: Prepare ufshcd_send_command for mcq

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:24PM -0800, Asutosh Das wrote:
> Add support to send commands using multiple submission
> queues in MCQ mode.
> Modify the functions that use ufshcd_send_command().
> 
> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>

Reviewed-by: Manivannan Sadhasivam <mani@kernel.org>

Thanks,
Mani

> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
> ---
>  drivers/ufs/core/ufs-mcq.c     |  1 +
>  drivers/ufs/core/ufshcd-priv.h | 10 ++++++++++
>  drivers/ufs/core/ufshcd.c      | 36 ++++++++++++++++++++++++++----------
>  include/ufs/ufshcd.h           |  5 +++++
>  4 files changed, 42 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
> index e4ddb90..7179f86 100644
> --- a/drivers/ufs/core/ufs-mcq.c
> +++ b/drivers/ufs/core/ufs-mcq.c
> @@ -405,6 +405,7 @@ int ufshcd_mcq_init(struct ufs_hba *hba)
>  	for (i = 0; i < hba->nr_hw_queues; i++) {
>  		hwq = &hba->uhq[i];
>  		hwq->max_entries = hba->nutrs;
> +		spin_lock_init(&hwq->sq_lock);
>  	}
>  
>  	/* The very first HW queue serves device commands */
> diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
> index 08963e1..5616047 100644
> --- a/drivers/ufs/core/ufshcd-priv.h
> +++ b/drivers/ufs/core/ufshcd-priv.h
> @@ -324,4 +324,14 @@ static inline bool ufs_is_valid_unit_desc_lun(struct ufs_dev_info *dev_info, u8
>  	return lun == UFS_UPIU_RPMB_WLUN || (lun < dev_info->max_lu_supported);
>  }
>  
> +static inline void ufshcd_inc_sq_tail(struct ufs_hw_queue *q)
> +{
> +	u32 mask = q->max_entries - 1;
> +	u32 val;
> +
> +	q->sq_tail_slot = (q->sq_tail_slot + 1) & mask;
> +	val = q->sq_tail_slot * sizeof(struct utp_transfer_req_desc);
> +	writel(val, q->mcq_sq_tail);
> +}
> +
>  #endif /* _UFSHCD_PRIV_H_ */
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index d61e99f..93a9e38 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -2182,9 +2182,11 @@ static void ufshcd_update_monitor(struct ufs_hba *hba, const struct ufshcd_lrb *
>   * ufshcd_send_command - Send SCSI or device management commands
>   * @hba: per adapter instance
>   * @task_tag: Task tag of the command
> + * @hwq: pointer to hardware queue instance
>   */
>  static inline
> -void ufshcd_send_command(struct ufs_hba *hba, unsigned int task_tag)
> +void ufshcd_send_command(struct ufs_hba *hba, unsigned int task_tag,
> +			 struct ufs_hw_queue *hwq)
>  {
>  	struct ufshcd_lrb *lrbp = &hba->lrb[task_tag];
>  	unsigned long flags;
> @@ -2198,12 +2200,24 @@ void ufshcd_send_command(struct ufs_hba *hba, unsigned int task_tag)
>  	if (unlikely(ufshcd_should_inform_monitor(hba, lrbp)))
>  		ufshcd_start_monitor(hba, lrbp);
>  
> -	spin_lock_irqsave(&hba->outstanding_lock, flags);
> -	if (hba->vops && hba->vops->setup_xfer_req)
> -		hba->vops->setup_xfer_req(hba, task_tag, !!lrbp->cmd);
> -	__set_bit(task_tag, &hba->outstanding_reqs);
> -	ufshcd_writel(hba, 1 << task_tag, REG_UTP_TRANSFER_REQ_DOOR_BELL);
> -	spin_unlock_irqrestore(&hba->outstanding_lock, flags);
> +	if (is_mcq_enabled(hba)) {
> +		int utrd_size = sizeof(struct utp_transfer_req_desc);
> +
> +		spin_lock(&hwq->sq_lock);
> +		memcpy(hwq->sqe_base_addr + (hwq->sq_tail_slot * utrd_size),
> +		       lrbp->utr_descriptor_ptr, utrd_size);
> +		ufshcd_inc_sq_tail(hwq);
> +		spin_unlock(&hwq->sq_lock);
> +	} else {
> +		spin_lock_irqsave(&hba->outstanding_lock, flags);
> +		if (hba->vops && hba->vops->setup_xfer_req)
> +			hba->vops->setup_xfer_req(hba, lrbp->task_tag,
> +						  !!lrbp->cmd);
> +		__set_bit(lrbp->task_tag, &hba->outstanding_reqs);
> +		ufshcd_writel(hba, 1 << lrbp->task_tag,
> +			      REG_UTP_TRANSFER_REQ_DOOR_BELL);
> +		spin_unlock_irqrestore(&hba->outstanding_lock, flags);
> +	}
>  }
>  
>  /**
> @@ -2822,6 +2836,7 @@ static int ufshcd_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *cmd)
>  	int tag = scsi_cmd_to_rq(cmd)->tag;
>  	struct ufshcd_lrb *lrbp;
>  	int err = 0;
> +	struct ufs_hw_queue *hwq = NULL;
>  
>  	WARN_ONCE(tag < 0 || tag >= hba->nutrs, "Invalid tag %d\n", tag);
>  
> @@ -2906,7 +2921,7 @@ static int ufshcd_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *cmd)
>  		goto out;
>  	}
>  
> -	ufshcd_send_command(hba, tag);
> +	ufshcd_send_command(hba, tag, hwq);
>  
>  out:
>  	rcu_read_unlock();
> @@ -3101,10 +3116,11 @@ static int ufshcd_exec_dev_cmd(struct ufs_hba *hba,
>  		goto out;
>  
>  	hba->dev_cmd.complete = &wait;
> +	hba->dev_cmd.cqe = NULL;
>  
>  	ufshcd_add_query_upiu_trace(hba, UFS_QUERY_SEND, lrbp->ucd_req_ptr);
>  
> -	ufshcd_send_command(hba, tag);
> +	ufshcd_send_command(hba, tag, hba->dev_cmd_queue);
>  	err = ufshcd_wait_for_dev_cmd(hba, lrbp, timeout);
>  	ufshcd_add_query_upiu_trace(hba, err ? UFS_QUERY_ERR : UFS_QUERY_COMP,
>  				    (struct utp_upiu_req *)lrbp->ucd_rsp_ptr);
> @@ -6926,7 +6942,7 @@ static int ufshcd_issue_devman_upiu_cmd(struct ufs_hba *hba,
>  
>  	ufshcd_add_query_upiu_trace(hba, UFS_QUERY_SEND, lrbp->ucd_req_ptr);
>  
> -	ufshcd_send_command(hba, tag);
> +	ufshcd_send_command(hba, tag, hba->dev_cmd_queue);
>  	/*
>  	 * ignore the returning value here - ufshcd_check_query_response is
>  	 * bound to fail since dev_cmd.query and dev_cmd.type were left empty.
> diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
> index 76972fc..853c525 100644
> --- a/include/ufs/ufshcd.h
> +++ b/include/ufs/ufshcd.h
> @@ -222,6 +222,7 @@ struct ufs_dev_cmd {
>  	struct mutex lock;
>  	struct completion *complete;
>  	struct ufs_query query;
> +	struct cq_entry *cqe;
>  };
>  
>  /**
> @@ -1061,6 +1062,8 @@ struct ufs_hba {
>   * @cqe_dma_addr: completion queue dma address
>   * @max_entries: max number of slots in this hardware queue
>   * @id: hardware queue ID
> + * @sq_tp_slot: current slot to which SQ tail pointer is pointing
> + * @sq_lock: serialize submission queue access
>   */
>  struct ufs_hw_queue {
>  	void __iomem *mcq_sq_head;
> @@ -1074,6 +1077,8 @@ struct ufs_hw_queue {
>  	dma_addr_t cqe_dma_addr;
>  	u32 max_entries;
>  	u32 id;
> +	u32 sq_tail_slot;
> +	spinlock_t sq_lock;
>  };
>  
>  static inline bool is_mcq_enabled(struct ufs_hba *hba)
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

[PATCH v5 12/16] ufs: core: mcq: Find hardware queue to queue request

[Asutosh Das]

Adds support to find the hardware queue on which the request
would be queued.
Since the very first queue is to serve device commands, an offset
of 1 is added to the index of the hardware queue.

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/ufs/core/ufs-mcq.c     | 20 ++++++++++++++++++++
 drivers/ufs/core/ufshcd-priv.h |  3 +++
 drivers/ufs/core/ufshcd.c      |  3 +++
 3 files changed, 26 insertions(+)

diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
index 7179f86..10a0d0d7 100644
--- a/drivers/ufs/core/ufs-mcq.c
+++ b/drivers/ufs/core/ufs-mcq.c
@@ -94,6 +94,26 @@ static const struct ufshcd_res_info ufs_res_info[RES_MAX] = {
 };
 
 /**
+ * ufshcd_mcq_req_to_hwq - find the hardware queue on which the
+ * request would be issued.
+ * @hba - per adapter instance
+ * @req - pointer to the request to be issued
+ *
+ * Returns the hardware queue instance on which the request would
+ * be queued.
+ */
+struct ufs_hw_queue *ufshcd_mcq_req_to_hwq(struct ufs_hba *hba,
+					 struct request *req)
+{
+	u32 utag = blk_mq_unique_tag(req);
+	u32 hwq = blk_mq_unique_tag_to_hwq(utag);
+
+	/* uhq[0] is used to serve device commands */
+	return &hba->uhq[hwq + UFSHCD_MCQ_IO_QUEUE_OFFSET];
+}
+
+
+/**
  * ufshcd_mcq_decide_queue_depth - decide the queue depth
  * @hba - per adapter instance
  *
diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
index 5616047..14df7ce 100644
--- a/drivers/ufs/core/ufshcd-priv.h
+++ b/drivers/ufs/core/ufshcd-priv.h
@@ -67,7 +67,10 @@ int ufshcd_mcq_memory_alloc(struct ufs_hba *hba);
 void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba);
 void ufshcd_mcq_config_mac(struct ufs_hba *hba, u32 max_active_cmds);
 void ufshcd_mcq_select_mcq_mode(struct ufs_hba *hba);
+struct ufs_hw_queue *ufshcd_mcq_req_to_hwq(struct ufs_hba *hba,
+					   struct request *req);
 
+#define UFSHCD_MCQ_IO_QUEUE_OFFSET	1
 #define SD_ASCII_STD true
 #define SD_RAW false
 int ufshcd_read_string_desc(struct ufs_hba *hba, u8 desc_index,
diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 93a9e38..52c0386 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -2921,6 +2921,9 @@ static int ufshcd_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *cmd)
 		goto out;
 	}
 
+	if (is_mcq_enabled(hba))
+		hwq = ufshcd_mcq_req_to_hwq(hba, scsi_cmd_to_rq(cmd));
+
 	ufshcd_send_command(hba, tag, hwq);
 
 out:
-- 
2.7.4

Re: [PATCH v5 12/16] ufs: core: mcq: Find hardware queue to queue request

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:25PM -0800, Asutosh Das wrote:
> Adds support to find the hardware queue on which the request
> would be queued.
> Since the very first queue is to serve device commands, an offset
> of 1 is added to the index of the hardware queue.
> 
> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>

One small nitpick below...

Reviewed-by: Manivannan Sadhasivam <mani@kernel.org>

> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
> ---
>  drivers/ufs/core/ufs-mcq.c     | 20 ++++++++++++++++++++
>  drivers/ufs/core/ufshcd-priv.h |  3 +++
>  drivers/ufs/core/ufshcd.c      |  3 +++
>  3 files changed, 26 insertions(+)
> 
> diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
> index 7179f86..10a0d0d7 100644
> --- a/drivers/ufs/core/ufs-mcq.c
> +++ b/drivers/ufs/core/ufs-mcq.c
> @@ -94,6 +94,26 @@ static const struct ufshcd_res_info ufs_res_info[RES_MAX] = {
>  };
>  
>  /**
> + * ufshcd_mcq_req_to_hwq - find the hardware queue on which the
> + * request would be issued.
> + * @hba - per adapter instance
> + * @req - pointer to the request to be issued
> + *
> + * Returns the hardware queue instance on which the request would
> + * be queued.
> + */
> +struct ufs_hw_queue *ufshcd_mcq_req_to_hwq(struct ufs_hba *hba,
> +					 struct request *req)
> +{
> +	u32 utag = blk_mq_unique_tag(req);
> +	u32 hwq = blk_mq_unique_tag_to_hwq(utag);
> +
> +	/* uhq[0] is used to serve device commands */
> +	return &hba->uhq[hwq + UFSHCD_MCQ_IO_QUEUE_OFFSET];
> +}
> +

Remove extra newline.

Thanks,
Mani

> +
> +/**
>   * ufshcd_mcq_decide_queue_depth - decide the queue depth
>   * @hba - per adapter instance
>   *
> diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
> index 5616047..14df7ce 100644
> --- a/drivers/ufs/core/ufshcd-priv.h
> +++ b/drivers/ufs/core/ufshcd-priv.h
> @@ -67,7 +67,10 @@ int ufshcd_mcq_memory_alloc(struct ufs_hba *hba);
>  void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba);
>  void ufshcd_mcq_config_mac(struct ufs_hba *hba, u32 max_active_cmds);
>  void ufshcd_mcq_select_mcq_mode(struct ufs_hba *hba);
> +struct ufs_hw_queue *ufshcd_mcq_req_to_hwq(struct ufs_hba *hba,
> +					   struct request *req);
>  
> +#define UFSHCD_MCQ_IO_QUEUE_OFFSET	1
>  #define SD_ASCII_STD true
>  #define SD_RAW false
>  int ufshcd_read_string_desc(struct ufs_hba *hba, u8 desc_index,
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 93a9e38..52c0386 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -2921,6 +2921,9 @@ static int ufshcd_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *cmd)
>  		goto out;
>  	}
>  
> +	if (is_mcq_enabled(hba))
> +		hwq = ufshcd_mcq_req_to_hwq(hba, scsi_cmd_to_rq(cmd));
> +
>  	ufshcd_send_command(hba, tag, hwq);
>  
>  out:
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

[PATCH v5 13/16] ufs: core: Prepare for completion in mcq

[Asutosh Das]

Modify completion path APIs and add completion queue
entry.

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/ufs/core/ufshcd-priv.h |  2 ++
 drivers/ufs/core/ufshcd.c      | 80 ++++++++++++++++++++++++++----------------
 2 files changed, 51 insertions(+), 31 deletions(-)

diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
index 14df7ce..6453449 100644
--- a/drivers/ufs/core/ufshcd-priv.h
+++ b/drivers/ufs/core/ufshcd-priv.h
@@ -61,6 +61,8 @@ int ufshcd_query_attr(struct ufs_hba *hba, enum query_opcode opcode,
 int ufshcd_query_flag(struct ufs_hba *hba, enum query_opcode opcode,
 	enum flag_idn idn, u8 index, bool *flag_res);
 void ufshcd_auto_hibern8_update(struct ufs_hba *hba, u32 ahit);
+void ufshcd_compl_one_cqe(struct ufs_hba *hba, int task_tag,
+			  struct cq_entry *cqe);
 int ufshcd_mcq_init(struct ufs_hba *hba);
 int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba);
 int ufshcd_mcq_memory_alloc(struct ufs_hba *hba);
diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 52c0386..f16d02c 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -784,12 +784,17 @@ static inline bool ufshcd_is_device_present(struct ufs_hba *hba)
 /**
  * ufshcd_get_tr_ocs - Get the UTRD Overall Command Status
  * @lrbp: pointer to local command reference block
+ * @cqe: pointer to the completion queue entry
  *
  * This function is used to get the OCS field from UTRD
  * Returns the OCS field in the UTRD
  */
-static enum utp_ocs ufshcd_get_tr_ocs(struct ufshcd_lrb *lrbp)
+static enum utp_ocs ufshcd_get_tr_ocs(struct ufshcd_lrb *lrbp,
+				      struct cq_entry *cqe)
 {
+	if (cqe)
+		return le32_to_cpu(cqe->status) & MASK_OCS;
+
 	return le32_to_cpu(lrbp->utr_descriptor_ptr->header.dword_2) & MASK_OCS;
 }
 
@@ -3048,7 +3053,7 @@ static int ufshcd_wait_for_dev_cmd(struct ufs_hba *hba,
 		 * not trigger any race conditions.
 		 */
 		hba->dev_cmd.complete = NULL;
-		err = ufshcd_get_tr_ocs(lrbp);
+		err = ufshcd_get_tr_ocs(lrbp, hba->dev_cmd.cqe);
 		if (!err)
 			err = ufshcd_dev_cmd_completion(hba, lrbp);
 	} else {
@@ -5214,18 +5219,20 @@ ufshcd_scsi_cmd_status(struct ufshcd_lrb *lrbp, int scsi_status)
  * ufshcd_transfer_rsp_status - Get overall status of the response
  * @hba: per adapter instance
  * @lrbp: pointer to local reference block of completed command
+ * @cqe: pointer to the completion queue entry
  *
  * Returns result of the command to notify SCSI midlayer
  */
 static inline int
-ufshcd_transfer_rsp_status(struct ufs_hba *hba, struct ufshcd_lrb *lrbp)
+ufshcd_transfer_rsp_status(struct ufs_hba *hba, struct ufshcd_lrb *lrbp,
+			   struct cq_entry *cqe)
 {
 	int result = 0;
 	int scsi_status;
 	enum utp_ocs ocs;
 
 	/* overall command status of utrd */
-	ocs = ufshcd_get_tr_ocs(lrbp);
+	ocs = ufshcd_get_tr_ocs(lrbp, cqe);
 
 	if (hba->quirks & UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR) {
 		if (be32_to_cpu(lrbp->ucd_rsp_ptr->header.dword_1) &
@@ -5390,42 +5397,53 @@ static void ufshcd_release_scsi_cmd(struct ufs_hba *hba,
 }
 
 /**
- * __ufshcd_transfer_req_compl - handle SCSI and query command completion
+ * ufshcd_compl_one_cqe - handle a completion queue entry
  * @hba: per adapter instance
- * @completed_reqs: bitmask that indicates which requests to complete
+ * @task_tag: the task tag of the request to be completed
+ * @cqe: pointer to the completion queue entry
  */
-static void __ufshcd_transfer_req_compl(struct ufs_hba *hba,
-					unsigned long completed_reqs)
+void ufshcd_compl_one_cqe(struct ufs_hba *hba, int task_tag,
+			  struct cq_entry *cqe)
 {
 	struct ufshcd_lrb *lrbp;
 	struct scsi_cmnd *cmd;
-	int index;
-
-	for_each_set_bit(index, &completed_reqs, hba->nutrs) {
-		lrbp = &hba->lrb[index];
-		lrbp->compl_time_stamp = ktime_get();
-		lrbp->compl_time_stamp_local_clock = local_clock();
-		cmd = lrbp->cmd;
-		if (cmd) {
-			if (unlikely(ufshcd_should_inform_monitor(hba, lrbp)))
-				ufshcd_update_monitor(hba, lrbp);
-			ufshcd_add_command_trace(hba, index, UFS_CMD_COMP);
-			cmd->result = ufshcd_transfer_rsp_status(hba, lrbp);
-			ufshcd_release_scsi_cmd(hba, lrbp);
-			/* Do not touch lrbp after scsi done */
-			scsi_done(cmd);
-		} else if (lrbp->command_type == UTP_CMD_TYPE_DEV_MANAGE ||
-			lrbp->command_type == UTP_CMD_TYPE_UFS_STORAGE) {
-			if (hba->dev_cmd.complete) {
-				ufshcd_add_command_trace(hba, index,
-							 UFS_DEV_COMP);
-				complete(hba->dev_cmd.complete);
-				ufshcd_clk_scaling_update_busy(hba);
-			}
+
+	lrbp = &hba->lrb[task_tag];
+	lrbp->compl_time_stamp = ktime_get();
+	cmd = lrbp->cmd;
+	if (cmd) {
+		if (unlikely(ufshcd_should_inform_monitor(hba, lrbp)))
+			ufshcd_update_monitor(hba, lrbp);
+		ufshcd_add_command_trace(hba, task_tag, UFS_CMD_COMP);
+		cmd->result = ufshcd_transfer_rsp_status(hba, lrbp, cqe);
+		ufshcd_release_scsi_cmd(hba, lrbp);
+		/* Do not touch lrbp after scsi done */
+		scsi_done(cmd);
+	} else if (lrbp->command_type == UTP_CMD_TYPE_DEV_MANAGE ||
+		   lrbp->command_type == UTP_CMD_TYPE_UFS_STORAGE) {
+		if (hba->dev_cmd.complete) {
+			hba->dev_cmd.cqe = cqe;
+			ufshcd_add_command_trace(hba, task_tag, UFS_DEV_COMP);
+			complete(hba->dev_cmd.complete);
+			ufshcd_clk_scaling_update_busy(hba);
 		}
 	}
 }
 
+/**
+ * __ufshcd_transfer_req_compl - handle SCSI and query command completion
+ * @hba: per adapter instance
+ * @completed_reqs: bitmask that indicates which requests to complete
+ */
+static void __ufshcd_transfer_req_compl(struct ufs_hba *hba,
+					unsigned long completed_reqs)
+{
+	int tag;
+
+	for_each_set_bit(tag, &completed_reqs, hba->nutrs)
+		ufshcd_compl_one_cqe(hba, tag, NULL);
+}
+
 /*
  * Returns > 0 if one or more commands have been completed or 0 if no
  * requests have been completed.
-- 
2.7.4

Re: [PATCH v5 13/16] ufs: core: Prepare for completion in mcq

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:26PM -0800, Asutosh Das wrote:
> Modify completion path APIs and add completion queue
> entry.
> 
> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>

Reviewed-by: Manivannan Sadhasivam <mani@kernel.org>

Thanks,
Mani

> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
> ---
>  drivers/ufs/core/ufshcd-priv.h |  2 ++
>  drivers/ufs/core/ufshcd.c      | 80 ++++++++++++++++++++++++++----------------
>  2 files changed, 51 insertions(+), 31 deletions(-)
> 
> diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
> index 14df7ce..6453449 100644
> --- a/drivers/ufs/core/ufshcd-priv.h
> +++ b/drivers/ufs/core/ufshcd-priv.h
> @@ -61,6 +61,8 @@ int ufshcd_query_attr(struct ufs_hba *hba, enum query_opcode opcode,
>  int ufshcd_query_flag(struct ufs_hba *hba, enum query_opcode opcode,
>  	enum flag_idn idn, u8 index, bool *flag_res);
>  void ufshcd_auto_hibern8_update(struct ufs_hba *hba, u32 ahit);
> +void ufshcd_compl_one_cqe(struct ufs_hba *hba, int task_tag,
> +			  struct cq_entry *cqe);
>  int ufshcd_mcq_init(struct ufs_hba *hba);
>  int ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba);
>  int ufshcd_mcq_memory_alloc(struct ufs_hba *hba);
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 52c0386..f16d02c 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -784,12 +784,17 @@ static inline bool ufshcd_is_device_present(struct ufs_hba *hba)
>  /**
>   * ufshcd_get_tr_ocs - Get the UTRD Overall Command Status
>   * @lrbp: pointer to local command reference block
> + * @cqe: pointer to the completion queue entry
>   *
>   * This function is used to get the OCS field from UTRD
>   * Returns the OCS field in the UTRD
>   */
> -static enum utp_ocs ufshcd_get_tr_ocs(struct ufshcd_lrb *lrbp)
> +static enum utp_ocs ufshcd_get_tr_ocs(struct ufshcd_lrb *lrbp,
> +				      struct cq_entry *cqe)
>  {
> +	if (cqe)
> +		return le32_to_cpu(cqe->status) & MASK_OCS;
> +
>  	return le32_to_cpu(lrbp->utr_descriptor_ptr->header.dword_2) & MASK_OCS;
>  }
>  
> @@ -3048,7 +3053,7 @@ static int ufshcd_wait_for_dev_cmd(struct ufs_hba *hba,
>  		 * not trigger any race conditions.
>  		 */
>  		hba->dev_cmd.complete = NULL;
> -		err = ufshcd_get_tr_ocs(lrbp);
> +		err = ufshcd_get_tr_ocs(lrbp, hba->dev_cmd.cqe);
>  		if (!err)
>  			err = ufshcd_dev_cmd_completion(hba, lrbp);
>  	} else {
> @@ -5214,18 +5219,20 @@ ufshcd_scsi_cmd_status(struct ufshcd_lrb *lrbp, int scsi_status)
>   * ufshcd_transfer_rsp_status - Get overall status of the response
>   * @hba: per adapter instance
>   * @lrbp: pointer to local reference block of completed command
> + * @cqe: pointer to the completion queue entry
>   *
>   * Returns result of the command to notify SCSI midlayer
>   */
>  static inline int
> -ufshcd_transfer_rsp_status(struct ufs_hba *hba, struct ufshcd_lrb *lrbp)
> +ufshcd_transfer_rsp_status(struct ufs_hba *hba, struct ufshcd_lrb *lrbp,
> +			   struct cq_entry *cqe)
>  {
>  	int result = 0;
>  	int scsi_status;
>  	enum utp_ocs ocs;
>  
>  	/* overall command status of utrd */
> -	ocs = ufshcd_get_tr_ocs(lrbp);
> +	ocs = ufshcd_get_tr_ocs(lrbp, cqe);
>  
>  	if (hba->quirks & UFSHCD_QUIRK_BROKEN_OCS_FATAL_ERROR) {
>  		if (be32_to_cpu(lrbp->ucd_rsp_ptr->header.dword_1) &
> @@ -5390,42 +5397,53 @@ static void ufshcd_release_scsi_cmd(struct ufs_hba *hba,
>  }
>  
>  /**
> - * __ufshcd_transfer_req_compl - handle SCSI and query command completion
> + * ufshcd_compl_one_cqe - handle a completion queue entry
>   * @hba: per adapter instance
> - * @completed_reqs: bitmask that indicates which requests to complete
> + * @task_tag: the task tag of the request to be completed
> + * @cqe: pointer to the completion queue entry
>   */
> -static void __ufshcd_transfer_req_compl(struct ufs_hba *hba,
> -					unsigned long completed_reqs)
> +void ufshcd_compl_one_cqe(struct ufs_hba *hba, int task_tag,
> +			  struct cq_entry *cqe)
>  {
>  	struct ufshcd_lrb *lrbp;
>  	struct scsi_cmnd *cmd;
> -	int index;
> -
> -	for_each_set_bit(index, &completed_reqs, hba->nutrs) {
> -		lrbp = &hba->lrb[index];
> -		lrbp->compl_time_stamp = ktime_get();
> -		lrbp->compl_time_stamp_local_clock = local_clock();
> -		cmd = lrbp->cmd;
> -		if (cmd) {
> -			if (unlikely(ufshcd_should_inform_monitor(hba, lrbp)))
> -				ufshcd_update_monitor(hba, lrbp);
> -			ufshcd_add_command_trace(hba, index, UFS_CMD_COMP);
> -			cmd->result = ufshcd_transfer_rsp_status(hba, lrbp);
> -			ufshcd_release_scsi_cmd(hba, lrbp);
> -			/* Do not touch lrbp after scsi done */
> -			scsi_done(cmd);
> -		} else if (lrbp->command_type == UTP_CMD_TYPE_DEV_MANAGE ||
> -			lrbp->command_type == UTP_CMD_TYPE_UFS_STORAGE) {
> -			if (hba->dev_cmd.complete) {
> -				ufshcd_add_command_trace(hba, index,
> -							 UFS_DEV_COMP);
> -				complete(hba->dev_cmd.complete);
> -				ufshcd_clk_scaling_update_busy(hba);
> -			}
> +
> +	lrbp = &hba->lrb[task_tag];
> +	lrbp->compl_time_stamp = ktime_get();
> +	cmd = lrbp->cmd;
> +	if (cmd) {
> +		if (unlikely(ufshcd_should_inform_monitor(hba, lrbp)))
> +			ufshcd_update_monitor(hba, lrbp);
> +		ufshcd_add_command_trace(hba, task_tag, UFS_CMD_COMP);
> +		cmd->result = ufshcd_transfer_rsp_status(hba, lrbp, cqe);
> +		ufshcd_release_scsi_cmd(hba, lrbp);
> +		/* Do not touch lrbp after scsi done */
> +		scsi_done(cmd);
> +	} else if (lrbp->command_type == UTP_CMD_TYPE_DEV_MANAGE ||
> +		   lrbp->command_type == UTP_CMD_TYPE_UFS_STORAGE) {
> +		if (hba->dev_cmd.complete) {
> +			hba->dev_cmd.cqe = cqe;
> +			ufshcd_add_command_trace(hba, task_tag, UFS_DEV_COMP);
> +			complete(hba->dev_cmd.complete);
> +			ufshcd_clk_scaling_update_busy(hba);
>  		}
>  	}
>  }
>  
> +/**
> + * __ufshcd_transfer_req_compl - handle SCSI and query command completion
> + * @hba: per adapter instance
> + * @completed_reqs: bitmask that indicates which requests to complete
> + */
> +static void __ufshcd_transfer_req_compl(struct ufs_hba *hba,
> +					unsigned long completed_reqs)
> +{
> +	int tag;
> +
> +	for_each_set_bit(tag, &completed_reqs, hba->nutrs)
> +		ufshcd_compl_one_cqe(hba, tag, NULL);
> +}
> +
>  /*
>   * Returns > 0 if one or more commands have been completed or 0 if no
>   * requests have been completed.
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

[PATCH v5 14/16] ufs: mcq: Add completion support of a cqe

[Asutosh Das]

Add support for completing requests from Completion Queue.
Some host controllers support vendor specific registers
that provide a bitmap of all CQ's which have at least one
completed CQE. Add this support.
The MCQ specification doesn't provide the Task Tag or its
equivalent in the Completion Queue Entry.
So use an indirect method to find the Task Tag from the
Completion Queue Entry.

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
Reported-by: kernel test robot <lkp@intel.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/ufs/core/ufs-mcq.c     | 57 ++++++++++++++++++++++++++++++++++++++++++
 drivers/ufs/core/ufshcd-priv.h | 43 +++++++++++++++++++++++++++++++
 drivers/ufs/core/ufshcd.c      | 37 +++++++++++++++++++++++++++
 drivers/ufs/host/ufs-qcom.c    | 16 ++++++++++++
 drivers/ufs/host/ufs-qcom.h    |  4 +++
 include/ufs/ufshcd.h           |  7 ++++++
 include/ufs/ufshci.h           |  3 +++
 7 files changed, 167 insertions(+)

diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
index 10a0d0d7..365ad98 100644
--- a/drivers/ufs/core/ufs-mcq.c
+++ b/drivers/ufs/core/ufs-mcq.c
@@ -28,6 +28,7 @@
 	((((c) >> 16) & MCQ_QCFGPTR_MASK) * MCQ_QCFGPTR_UNIT)
 #define MCQ_QCFG_SIZE	0x40
 #define MCQ_ENTRY_SIZE_IN_DWORD	8
+#define CQE_UCD_BA GENMASK_ULL(63, 7)
 
 static int rw_queue_count_set(const char *val, const struct kernel_param *kp)
 {
@@ -333,6 +334,59 @@ static void __iomem *mcq_opr_base(struct ufs_hba *hba,
 	return opr->base + opr->stride * i;
 }
 
+u32 ufshcd_mcq_read_cqis(struct ufs_hba *hba, int i)
+{
+	return readl(mcq_opr_base(hba, OPR_CQIS, i) + REG_CQIS);
+}
+
+void ufshcd_mcq_write_cqis(struct ufs_hba *hba, u32 val, int i)
+{
+	writel(val, mcq_opr_base(hba, OPR_CQIS, i) + REG_CQIS);
+}
+
+static int ufshcd_mcq_get_tag(struct ufs_hba *hba,
+				     struct ufs_hw_queue *hwq,
+				     struct cq_entry *cqe)
+{
+	dma_addr_t dma_addr;
+
+	/* sizeof(struct utp_transfer_cmd_desc) must be a multiple of 128 */
+	BUILD_BUG_ON(sizeof(struct utp_transfer_cmd_desc) & GENMASK(6, 0));
+
+	/* Bits 63:7 UCD base address, 6:5 are reserved, 4:0 is SQ ID */
+	dma_addr = le64_to_cpu(cqe->command_desc_base_addr) & CQE_UCD_BA;
+
+	return (dma_addr - hba->ucdl_dma_addr) /
+		sizeof(struct utp_transfer_cmd_desc);
+}
+
+static void ufshcd_mcq_process_cqe(struct ufs_hba *hba,
+					    struct ufs_hw_queue *hwq)
+{
+	struct cq_entry *cqe = ufshcd_mcq_cur_cqe(hwq);
+	int tag = ufshcd_mcq_get_tag(hba, hwq, cqe);
+
+	ufshcd_compl_one_cqe(hba, tag, cqe);
+}
+
+unsigned long ufshcd_mcq_poll_cqe_nolock(struct ufs_hba *hba,
+					 struct ufs_hw_queue *hwq)
+{
+	unsigned long completed_reqs = 0;
+
+	ufshcd_mcq_update_cq_tail_slot(hwq);
+	while (!ufshcd_mcq_is_cq_empty(hwq)) {
+		ufshcd_mcq_process_cqe(hba, hwq);
+		ufshcd_mcq_inc_cq_head_slot(hwq);
+		completed_reqs++;
+	}
+
+	if (completed_reqs)
+		ufshcd_mcq_update_cq_head(hwq);
+
+	return completed_reqs;
+}
+
 void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba)
 {
 	struct ufs_hw_queue *hwq;
@@ -376,6 +430,9 @@ void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba)
 		hwq->mcq_cq_head = mcq_opr_base(hba, OPR_CQD, i) + REG_CQHP;
 		hwq->mcq_cq_tail = mcq_opr_base(hba, OPR_CQD, i) + REG_CQTP;
 
+		/* Reinitializing is needed upon HC reset */
+		hwq->sq_tail_slot = hwq->cq_tail_slot = hwq->cq_head_slot = 0;
+
 		/* Enable Tail Entry Push Status interrupt only for non-poll queues */
 		if (i < hba->nr_hw_queues - hba->nr_queues[HCTX_TYPE_POLL])
 			writel(1, mcq_opr_base(hba, OPR_CQIS, i) + REG_CQIE);
diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
index 6453449..c5b5bf3 100644
--- a/drivers/ufs/core/ufshcd-priv.h
+++ b/drivers/ufs/core/ufshcd-priv.h
@@ -69,6 +69,10 @@ int ufshcd_mcq_memory_alloc(struct ufs_hba *hba);
 void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba);
 void ufshcd_mcq_config_mac(struct ufs_hba *hba, u32 max_active_cmds);
 void ufshcd_mcq_select_mcq_mode(struct ufs_hba *hba);
+u32 ufshcd_mcq_read_cqis(struct ufs_hba *hba, int i);
+void ufshcd_mcq_write_cqis(struct ufs_hba *hba, u32 val, int i);
+unsigned long ufshcd_mcq_poll_cqe_nolock(struct ufs_hba *hba,
+					 struct ufs_hw_queue *hwq);
 struct ufs_hw_queue *ufshcd_mcq_req_to_hwq(struct ufs_hba *hba,
 					   struct request *req);
 
@@ -253,6 +257,15 @@ static inline int ufshcd_mcq_vops_op_runtime_config(struct ufs_hba *hba)
 	return -EOPNOTSUPP;
 }
 
+static inline int ufshcd_vops_get_outstanding_cqs(struct ufs_hba *hba,
+						  unsigned long *ocqs)
+{
+	if (hba->vops && hba->vops->get_outstanding_cqs)
+		return hba->vops->get_outstanding_cqs(hba, ocqs);
+
+	return -EOPNOTSUPP;
+}
+
 extern const struct ufs_pm_lvl_states ufs_pm_lvl_states[];
 
 /**
@@ -339,4 +352,34 @@ static inline void ufshcd_inc_sq_tail(struct ufs_hw_queue *q)
 	writel(val, q->mcq_sq_tail);
 }
 
+static inline void ufshcd_mcq_update_cq_tail_slot(struct ufs_hw_queue *q)
+{
+	u32 val = readl(q->mcq_cq_tail);
+
+	q->cq_tail_slot = val / sizeof(struct cq_entry);
+}
+
+static inline bool ufshcd_mcq_is_cq_empty(struct ufs_hw_queue *q)
+{
+	return q->cq_head_slot == q->cq_tail_slot;
+}
+
+static inline void ufshcd_mcq_inc_cq_head_slot(struct ufs_hw_queue *q)
+{
+	q->cq_head_slot++;
+	if (q->cq_head_slot == q->max_entries)
+		q->cq_head_slot = 0;
+}
+
+static inline void ufshcd_mcq_update_cq_head(struct ufs_hw_queue *q)
+{
+	writel(q->cq_head_slot * sizeof(struct cq_entry), q->mcq_cq_head);
+}
+
+static inline struct cq_entry *ufshcd_mcq_cur_cqe(struct ufs_hw_queue *q)
+{
+	struct cq_entry *cqe = q->cqe_base_addr;
+
+	return cqe + q->cq_head_slot;
+}
 #endif /* _UFSHCD_PRIV_H_ */
diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index f16d02c..7fb7c5f 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -6672,6 +6672,40 @@ static irqreturn_t ufshcd_tmc_handler(struct ufs_hba *hba)
 }
 
 /**
+ * ufshcd_handle_mcq_cq_events - handle MCQ completion queue events
+ * @hba: per adapter instance
+ *
+ * Returns IRQ_HANDLED if interrupt is handled
+ */
+static irqreturn_t ufshcd_handle_mcq_cq_events(struct ufs_hba *hba)
+{
+	struct ufs_hw_queue *hwq;
+	unsigned long outstanding_cqs;
+	unsigned int nr_queues;
+	int i, ret;
+	u32 events;
+
+	ret = ufshcd_vops_get_outstanding_cqs(hba, &outstanding_cqs);
+	if (ret)
+		outstanding_cqs = (1U << hba->nr_hw_queues) - 1;
+
+	/* Exclude the poll queues */
+	nr_queues = hba->nr_hw_queues - hba->nr_queues[HCTX_TYPE_POLL];
+	for_each_set_bit(i, &outstanding_cqs, nr_queues) {
+		hwq = &hba->uhq[i];
+
+		events = ufshcd_mcq_read_cqis(hba, i);
+		if (events)
+			ufshcd_mcq_write_cqis(hba, events, i);
+
+		if (events & UFSHCD_MCQ_CQIS_TAIL_ENT_PUSH_STS)
+			ufshcd_mcq_poll_cqe_nolock(hba, hwq);
+	}
+
+	return IRQ_HANDLED;
+}
+
+/**
  * ufshcd_sl_intr - Interrupt service routine
  * @hba: per adapter instance
  * @intr_status: contains interrupts generated by the controller
@@ -6696,6 +6730,9 @@ static irqreturn_t ufshcd_sl_intr(struct ufs_hba *hba, u32 intr_status)
 	if (intr_status & UTP_TRANSFER_REQ_COMPL)
 		retval |= ufshcd_transfer_req_compl(hba);
 
+	if (intr_status & MCQ_CQ_EVENT_STATUS)
+		retval |= ufshcd_handle_mcq_cq_events(hba);
+
 	return retval;
 }
 
diff --git a/drivers/ufs/host/ufs-qcom.c b/drivers/ufs/host/ufs-qcom.c
index dd53e85..89d29a1 100644
--- a/drivers/ufs/host/ufs-qcom.c
+++ b/drivers/ufs/host/ufs-qcom.c
@@ -1454,6 +1454,21 @@ static int ufs_qcom_get_hba_mac(struct ufs_hba *hba)
 	return MAX_SUPP_MAC;
 }
 
+static int ufs_qcom_get_outstanding_cqs(struct ufs_hba *hba,
+					unsigned long *ocqs)
+{
+	u32 cqis_vs;
+	struct ufshcd_res_info *mcq_vs_res = &hba->res[RES_MCQ_VS];
+
+	if (!mcq_vs_res->base)
+		return -EINVAL;
+
+	cqis_vs = readl(mcq_vs_res->base + UFS_MEM_CQIS_VS);
+	*ocqs = cqis_vs;
+
+	return 0;
+}
+
 /*
  * struct ufs_hba_qcom_vops - UFS QCOM specific variant operations
  *
@@ -1479,6 +1494,7 @@ static const struct ufs_hba_variant_ops ufs_hba_qcom_vops = {
 	.program_key		= ufs_qcom_ice_program_key,
 	.get_hba_mac		= ufs_qcom_get_hba_mac,
 	.op_runtime_config	= ufs_qcom_op_runtime_config,
+	.get_outstanding_cqs	= ufs_qcom_get_outstanding_cqs,
 };
 
 /**
diff --git a/drivers/ufs/host/ufs-qcom.h b/drivers/ufs/host/ufs-qcom.h
index 44466a3..7769f03 100644
--- a/drivers/ufs/host/ufs-qcom.h
+++ b/drivers/ufs/host/ufs-qcom.h
@@ -72,6 +72,10 @@ enum {
 	UFS_UFS_DBG_RD_EDTL_RAM			= 0x1900,
 };
 
+enum {
+	UFS_MEM_CQIS_VS		= 0x8,
+};
+
 #define UFS_CNTLR_2_x_x_VEN_REGS_OFFSET(x)	(0x000 + x)
 #define UFS_CNTLR_3_x_x_VEN_REGS_OFFSET(x)	(0x400 + x)
 
diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
index 853c525..d5fde64 100644
--- a/include/ufs/ufshcd.h
+++ b/include/ufs/ufshcd.h
@@ -300,6 +300,7 @@ struct ufs_pwr_mode_info {
  * @event_notify: called to notify important events
  * @get_hba_mac: called to get vendor specific mac value, mandatory for mcq mode
  * @op_runtime_config: called to config Operation and runtime regs Pointers
+ * @get_outstanding_cqs: called to get outstanding completion queues
  */
 struct ufs_hba_variant_ops {
 	const char *name;
@@ -340,6 +341,8 @@ struct ufs_hba_variant_ops {
 				enum ufs_event_type evt, void *data);
 	int	(*get_hba_mac)(struct ufs_hba *hba);
 	int	(*op_runtime_config)(struct ufs_hba *hba);
+	int	(*get_outstanding_cqs)(struct ufs_hba *hba,
+				       unsigned long *ocqs);
 };
 
 /* clock gating state  */
@@ -1064,6 +1067,8 @@ struct ufs_hba {
  * @id: hardware queue ID
  * @sq_tp_slot: current slot to which SQ tail pointer is pointing
  * @sq_lock: serialize submission queue access
+ * @cq_tail_slot: current slot to which CQ tail pointer is pointing
+ * @cq_head_slot: current slot to which CQ head pointer is pointing
  */
 struct ufs_hw_queue {
 	void __iomem *mcq_sq_head;
@@ -1079,6 +1084,8 @@ struct ufs_hw_queue {
 	u32 id;
 	u32 sq_tail_slot;
 	spinlock_t sq_lock;
+	u32 cq_tail_slot;
+	u32 cq_head_slot;
 };
 
 static inline bool is_mcq_enabled(struct ufs_hba *hba)
diff --git a/include/ufs/ufshci.h b/include/ufs/ufshci.h
index c85bdf2..1e557ba 100644
--- a/include/ufs/ufshci.h
+++ b/include/ufs/ufshci.h
@@ -262,6 +262,9 @@ enum {
 /* UTMRLRSR - UTP Task Management Request Run-Stop Register 80h */
 #define UTP_TASK_REQ_LIST_RUN_STOP_BIT		0x1
 
+/* CQISy - CQ y Interrupt Status Register  */
+#define UFSHCD_MCQ_CQIS_TAIL_ENT_PUSH_STS	0x1
+
 /* UICCMD - UIC Command */
 #define COMMAND_OPCODE_MASK		0xFF
 #define GEN_SELECTOR_INDEX_MASK		0xFFFF
-- 
2.7.4

Re: [PATCH v5 14/16] ufs: mcq: Add completion support of a cqe

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:27PM -0800, Asutosh Das wrote:
> Add support for completing requests from Completion Queue.
> Some host controllers support vendor specific registers
> that provide a bitmap of all CQ's which have at least one
> completed CQE. Add this support.
> The MCQ specification doesn't provide the Task Tag or its
> equivalent in the Completion Queue Entry.
> So use an indirect method to find the Task Tag from the
> Completion Queue Entry.
> 
> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
> Reported-by: kernel test robot <lkp@intel.com>

What is this reported by for?

> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
> ---
>  drivers/ufs/core/ufs-mcq.c     | 57 ++++++++++++++++++++++++++++++++++++++++++
>  drivers/ufs/core/ufshcd-priv.h | 43 +++++++++++++++++++++++++++++++
>  drivers/ufs/core/ufshcd.c      | 37 +++++++++++++++++++++++++++
>  drivers/ufs/host/ufs-qcom.c    | 16 ++++++++++++
>  drivers/ufs/host/ufs-qcom.h    |  4 +++
>  include/ufs/ufshcd.h           |  7 ++++++
>  include/ufs/ufshci.h           |  3 +++
>  7 files changed, 167 insertions(+)
> 
> diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
> index 10a0d0d7..365ad98 100644
> --- a/drivers/ufs/core/ufs-mcq.c
> +++ b/drivers/ufs/core/ufs-mcq.c
> @@ -28,6 +28,7 @@
>  	((((c) >> 16) & MCQ_QCFGPTR_MASK) * MCQ_QCFGPTR_UNIT)
>  #define MCQ_QCFG_SIZE	0x40
>  #define MCQ_ENTRY_SIZE_IN_DWORD	8
> +#define CQE_UCD_BA GENMASK_ULL(63, 7)
>  
>  static int rw_queue_count_set(const char *val, const struct kernel_param *kp)
>  {
> @@ -333,6 +334,59 @@ static void __iomem *mcq_opr_base(struct ufs_hba *hba,
>  	return opr->base + opr->stride * i;
>  }
>  
> +u32 ufshcd_mcq_read_cqis(struct ufs_hba *hba, int i)
> +{
> +	return readl(mcq_opr_base(hba, OPR_CQIS, i) + REG_CQIS);
> +}
> +
> +void ufshcd_mcq_write_cqis(struct ufs_hba *hba, u32 val, int i)
> +{
> +	writel(val, mcq_opr_base(hba, OPR_CQIS, i) + REG_CQIS);
> +}
> +

It'd be good to add a comment about tag extraction as you did in description.

> +static int ufshcd_mcq_get_tag(struct ufs_hba *hba,
> +				     struct ufs_hw_queue *hwq,
> +				     struct cq_entry *cqe)
> +{
> +	dma_addr_t dma_addr;
> +
> +	/* sizeof(struct utp_transfer_cmd_desc) must be a multiple of 128 */
> +	BUILD_BUG_ON(sizeof(struct utp_transfer_cmd_desc) & GENMASK(6, 0));
> +
> +	/* Bits 63:7 UCD base address, 6:5 are reserved, 4:0 is SQ ID */
> +	dma_addr = le64_to_cpu(cqe->command_desc_base_addr) & CQE_UCD_BA;
> +
> +	return (dma_addr - hba->ucdl_dma_addr) /
> +		sizeof(struct utp_transfer_cmd_desc);
> +}
> +

[...]

> diff --git a/drivers/ufs/host/ufs-qcom.c b/drivers/ufs/host/ufs-qcom.c
> index dd53e85..89d29a1 100644
> --- a/drivers/ufs/host/ufs-qcom.c
> +++ b/drivers/ufs/host/ufs-qcom.c
> @@ -1454,6 +1454,21 @@ static int ufs_qcom_get_hba_mac(struct ufs_hba *hba)
>  	return MAX_SUPP_MAC;
>  }
>  
> +static int ufs_qcom_get_outstanding_cqs(struct ufs_hba *hba,
> +					unsigned long *ocqs)
> +{
> +	u32 cqis_vs;
> +	struct ufshcd_res_info *mcq_vs_res = &hba->res[RES_MCQ_VS];
> +
> +	if (!mcq_vs_res->base)
> +		return -EINVAL;
> +
> +	cqis_vs = readl(mcq_vs_res->base + UFS_MEM_CQIS_VS);
> +	*ocqs = cqis_vs;

Why can't you assign readl to *ocqs?

Thanks,
Mani
> +
> +	return 0;
> +}
> +
>  /*
>   * struct ufs_hba_qcom_vops - UFS QCOM specific variant operations
>   *
> @@ -1479,6 +1494,7 @@ static const struct ufs_hba_variant_ops ufs_hba_qcom_vops = {
>  	.program_key		= ufs_qcom_ice_program_key,
>  	.get_hba_mac		= ufs_qcom_get_hba_mac,
>  	.op_runtime_config	= ufs_qcom_op_runtime_config,
> +	.get_outstanding_cqs	= ufs_qcom_get_outstanding_cqs,
>  };
>  
>  /**
> diff --git a/drivers/ufs/host/ufs-qcom.h b/drivers/ufs/host/ufs-qcom.h
> index 44466a3..7769f03 100644
> --- a/drivers/ufs/host/ufs-qcom.h
> +++ b/drivers/ufs/host/ufs-qcom.h
> @@ -72,6 +72,10 @@ enum {
>  	UFS_UFS_DBG_RD_EDTL_RAM			= 0x1900,
>  };
>  
> +enum {
> +	UFS_MEM_CQIS_VS		= 0x8,
> +};
> +
>  #define UFS_CNTLR_2_x_x_VEN_REGS_OFFSET(x)	(0x000 + x)
>  #define UFS_CNTLR_3_x_x_VEN_REGS_OFFSET(x)	(0x400 + x)
>  
> diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
> index 853c525..d5fde64 100644
> --- a/include/ufs/ufshcd.h
> +++ b/include/ufs/ufshcd.h
> @@ -300,6 +300,7 @@ struct ufs_pwr_mode_info {
>   * @event_notify: called to notify important events
>   * @get_hba_mac: called to get vendor specific mac value, mandatory for mcq mode
>   * @op_runtime_config: called to config Operation and runtime regs Pointers
> + * @get_outstanding_cqs: called to get outstanding completion queues
>   */
>  struct ufs_hba_variant_ops {
>  	const char *name;
> @@ -340,6 +341,8 @@ struct ufs_hba_variant_ops {
>  				enum ufs_event_type evt, void *data);
>  	int	(*get_hba_mac)(struct ufs_hba *hba);
>  	int	(*op_runtime_config)(struct ufs_hba *hba);
> +	int	(*get_outstanding_cqs)(struct ufs_hba *hba,
> +				       unsigned long *ocqs);
>  };
>  
>  /* clock gating state  */
> @@ -1064,6 +1067,8 @@ struct ufs_hba {
>   * @id: hardware queue ID
>   * @sq_tp_slot: current slot to which SQ tail pointer is pointing
>   * @sq_lock: serialize submission queue access
> + * @cq_tail_slot: current slot to which CQ tail pointer is pointing
> + * @cq_head_slot: current slot to which CQ head pointer is pointing
>   */
>  struct ufs_hw_queue {
>  	void __iomem *mcq_sq_head;
> @@ -1079,6 +1084,8 @@ struct ufs_hw_queue {
>  	u32 id;
>  	u32 sq_tail_slot;
>  	spinlock_t sq_lock;
> +	u32 cq_tail_slot;
> +	u32 cq_head_slot;
>  };
>  
>  static inline bool is_mcq_enabled(struct ufs_hba *hba)
> diff --git a/include/ufs/ufshci.h b/include/ufs/ufshci.h
> index c85bdf2..1e557ba 100644
> --- a/include/ufs/ufshci.h
> +++ b/include/ufs/ufshci.h
> @@ -262,6 +262,9 @@ enum {
>  /* UTMRLRSR - UTP Task Management Request Run-Stop Register 80h */
>  #define UTP_TASK_REQ_LIST_RUN_STOP_BIT		0x1
>  
> +/* CQISy - CQ y Interrupt Status Register  */
> +#define UFSHCD_MCQ_CQIS_TAIL_ENT_PUSH_STS	0x1
> +
>  /* UICCMD - UIC Command */
>  #define COMMAND_OPCODE_MASK		0xFF
>  #define GEN_SELECTOR_INDEX_MASK		0xFFFF
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

Re: [PATCH v5 14/16] ufs: mcq: Add completion support of a cqe

[Bart Van Assche]

On 11/28/22 09:00, Manivannan Sadhasivam wrote:
> On Tue, Nov 22, 2022 at 08:10:27PM -0800, Asutosh Das wrote:
>> Add support for completing requests from Completion Queue.
>> Some host controllers support vendor specific registers
>> that provide a bitmap of all CQ's which have at least one
>> completed CQE. Add this support.
>> The MCQ specification doesn't provide the Task Tag or its
>> equivalent in the Completion Queue Entry.
>> So use an indirect method to find the Task Tag from the
>> Completion Queue Entry.
>>
>> Co-developed-by: Can Guo <quic_cang@quicinc.com>
>> Signed-off-by: Can Guo <quic_cang@quicinc.com>
>> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
>> Reported-by: kernel test robot <lkp@intel.com>
> 
> What is this reported by for?

I think that tag should be left out. "Reported-by: kernel test robot" 
should only be used for patches that fix bugs reported by the kernel 
test robot. I assume that a fix has been folded in into this patch that 
was reported by the kernel test robot. If that is the case, the 
"Reported-by: kernel test robot" tag is inappropriate.

Bart.

[PATCH v5 15/16] ufs: core: mcq: Add completion support in poll

[Asutosh Das]

Complete cqe requests in poll. Assumption is that
several poll completion may happen in different CPUs
for the same completion queue. Hence a spin lock
protection is added.

Co-developed-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Can Guo <quic_cang@quicinc.com>
Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/ufs/core/ufs-mcq.c     | 13 +++++++++++++
 drivers/ufs/core/ufshcd-priv.h |  2 ++
 drivers/ufs/core/ufshcd.c      |  7 +++++++
 include/ufs/ufshcd.h           |  2 ++
 4 files changed, 24 insertions(+)

diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
index 365ad98..5311857 100644
--- a/drivers/ufs/core/ufs-mcq.c
+++ b/drivers/ufs/core/ufs-mcq.c
@@ -387,6 +387,18 @@ unsigned long ufshcd_mcq_poll_cqe_nolock(struct ufs_hba *hba,
 	return completed_reqs;
 }
 
+unsigned long ufshcd_mcq_poll_cqe_lock(struct ufs_hba *hba,
+				       struct ufs_hw_queue *hwq)
+{
+	unsigned long completed_reqs;
+
+	spin_lock(&hwq->cq_lock);
+	completed_reqs = ufshcd_mcq_poll_cqe_nolock(hba, hwq);
+	spin_unlock(&hwq->cq_lock);
+
+	return completed_reqs;
+}
+
 void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba)
 {
 	struct ufs_hw_queue *hwq;
@@ -483,6 +495,7 @@ int ufshcd_mcq_init(struct ufs_hba *hba)
 		hwq = &hba->uhq[i];
 		hwq->max_entries = hba->nutrs;
 		spin_lock_init(&hwq->sq_lock);
+		spin_lock_init(&hwq->cq_lock);
 	}
 
 	/* The very first HW queue serves device commands */
diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
index c5b5bf3..73ce8a2 100644
--- a/drivers/ufs/core/ufshcd-priv.h
+++ b/drivers/ufs/core/ufshcd-priv.h
@@ -75,6 +75,8 @@ unsigned long ufshcd_mcq_poll_cqe_nolock(struct ufs_hba *hba,
 					 struct ufs_hw_queue *hwq);
 struct ufs_hw_queue *ufshcd_mcq_req_to_hwq(struct ufs_hba *hba,
 					   struct request *req);
+unsigned long ufshcd_mcq_poll_cqe_lock(struct ufs_hba *hba,
+				       struct ufs_hw_queue *hwq);
 
 #define UFSHCD_MCQ_IO_QUEUE_OFFSET	1
 #define SD_ASCII_STD true
diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 7fb7c5f..8416d42 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -5453,6 +5453,13 @@ static int ufshcd_poll(struct Scsi_Host *shost, unsigned int queue_num)
 	struct ufs_hba *hba = shost_priv(shost);
 	unsigned long completed_reqs, flags;
 	u32 tr_doorbell;
+	struct ufs_hw_queue *hwq;
+
+	if (is_mcq_enabled(hba)) {
+		hwq = &hba->uhq[queue_num + UFSHCD_MCQ_IO_QUEUE_OFFSET];
+
+		return ufshcd_mcq_poll_cqe_lock(hba, hwq);
+	}
 
 	spin_lock_irqsave(&hba->outstanding_lock, flags);
 	tr_doorbell = ufshcd_readl(hba, REG_UTP_TRANSFER_REQ_DOOR_BELL);
diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
index d5fde64..a709391 100644
--- a/include/ufs/ufshcd.h
+++ b/include/ufs/ufshcd.h
@@ -1069,6 +1069,7 @@ struct ufs_hba {
  * @sq_lock: serialize submission queue access
  * @cq_tail_slot: current slot to which CQ tail pointer is pointing
  * @cq_head_slot: current slot to which CQ head pointer is pointing
+ * @cq_lock: Synchronize between multiple polling instances
  */
 struct ufs_hw_queue {
 	void __iomem *mcq_sq_head;
@@ -1086,6 +1087,7 @@ struct ufs_hw_queue {
 	spinlock_t sq_lock;
 	u32 cq_tail_slot;
 	u32 cq_head_slot;
+	spinlock_t cq_lock;
 };
 
 static inline bool is_mcq_enabled(struct ufs_hba *hba)
-- 
2.7.4

Re: [PATCH v5 15/16] ufs: core: mcq: Add completion support in poll

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:28PM -0800, Asutosh Das wrote:
> Complete cqe requests in poll. Assumption is that
> several poll completion may happen in different CPUs
> for the same completion queue. Hence a spin lock
> protection is added.
> 
> Co-developed-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Can Guo <quic_cang@quicinc.com>
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>

Reviewed-by: Manivannan Sadhasivam <mani@kernel.org>

Thanks,
Mani

> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
> ---
>  drivers/ufs/core/ufs-mcq.c     | 13 +++++++++++++
>  drivers/ufs/core/ufshcd-priv.h |  2 ++
>  drivers/ufs/core/ufshcd.c      |  7 +++++++
>  include/ufs/ufshcd.h           |  2 ++
>  4 files changed, 24 insertions(+)
> 
> diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
> index 365ad98..5311857 100644
> --- a/drivers/ufs/core/ufs-mcq.c
> +++ b/drivers/ufs/core/ufs-mcq.c
> @@ -387,6 +387,18 @@ unsigned long ufshcd_mcq_poll_cqe_nolock(struct ufs_hba *hba,
>  	return completed_reqs;
>  }
>  
> +unsigned long ufshcd_mcq_poll_cqe_lock(struct ufs_hba *hba,
> +				       struct ufs_hw_queue *hwq)
> +{
> +	unsigned long completed_reqs;
> +
> +	spin_lock(&hwq->cq_lock);
> +	completed_reqs = ufshcd_mcq_poll_cqe_nolock(hba, hwq);
> +	spin_unlock(&hwq->cq_lock);
> +
> +	return completed_reqs;
> +}
> +
>  void ufshcd_mcq_make_queues_operational(struct ufs_hba *hba)
>  {
>  	struct ufs_hw_queue *hwq;
> @@ -483,6 +495,7 @@ int ufshcd_mcq_init(struct ufs_hba *hba)
>  		hwq = &hba->uhq[i];
>  		hwq->max_entries = hba->nutrs;
>  		spin_lock_init(&hwq->sq_lock);
> +		spin_lock_init(&hwq->cq_lock);
>  	}
>  
>  	/* The very first HW queue serves device commands */
> diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
> index c5b5bf3..73ce8a2 100644
> --- a/drivers/ufs/core/ufshcd-priv.h
> +++ b/drivers/ufs/core/ufshcd-priv.h
> @@ -75,6 +75,8 @@ unsigned long ufshcd_mcq_poll_cqe_nolock(struct ufs_hba *hba,
>  					 struct ufs_hw_queue *hwq);
>  struct ufs_hw_queue *ufshcd_mcq_req_to_hwq(struct ufs_hba *hba,
>  					   struct request *req);
> +unsigned long ufshcd_mcq_poll_cqe_lock(struct ufs_hba *hba,
> +				       struct ufs_hw_queue *hwq);
>  
>  #define UFSHCD_MCQ_IO_QUEUE_OFFSET	1
>  #define SD_ASCII_STD true
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 7fb7c5f..8416d42 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -5453,6 +5453,13 @@ static int ufshcd_poll(struct Scsi_Host *shost, unsigned int queue_num)
>  	struct ufs_hba *hba = shost_priv(shost);
>  	unsigned long completed_reqs, flags;
>  	u32 tr_doorbell;
> +	struct ufs_hw_queue *hwq;
> +
> +	if (is_mcq_enabled(hba)) {
> +		hwq = &hba->uhq[queue_num + UFSHCD_MCQ_IO_QUEUE_OFFSET];
> +
> +		return ufshcd_mcq_poll_cqe_lock(hba, hwq);
> +	}
>  
>  	spin_lock_irqsave(&hba->outstanding_lock, flags);
>  	tr_doorbell = ufshcd_readl(hba, REG_UTP_TRANSFER_REQ_DOOR_BELL);
> diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
> index d5fde64..a709391 100644
> --- a/include/ufs/ufshcd.h
> +++ b/include/ufs/ufshcd.h
> @@ -1069,6 +1069,7 @@ struct ufs_hba {
>   * @sq_lock: serialize submission queue access
>   * @cq_tail_slot: current slot to which CQ tail pointer is pointing
>   * @cq_head_slot: current slot to which CQ head pointer is pointing
> + * @cq_lock: Synchronize between multiple polling instances
>   */
>  struct ufs_hw_queue {
>  	void __iomem *mcq_sq_head;
> @@ -1086,6 +1087,7 @@ struct ufs_hw_queue {
>  	spinlock_t sq_lock;
>  	u32 cq_tail_slot;
>  	u32 cq_head_slot;
> +	spinlock_t cq_lock;
>  };
>  
>  static inline bool is_mcq_enabled(struct ufs_hba *hba)
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்

[PATCH v5 16/16] ufs: core: mcq: Enable Multi Circular Queue

[Asutosh Das]

Enable MCQ in the Host Controller.

Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/ufs/core/ufshcd.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 8416d42..41005b8 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -8351,6 +8351,12 @@ static void ufshcd_config_mcq(struct ufs_hba *hba)
 
 	hba->host->can_queue = hba->nutrs - UFSHCD_NUM_RESERVED;
 	hba->reserved_slot = hba->nutrs - UFSHCD_NUM_RESERVED;
+
+	/* Select MCQ mode */
+	ufshcd_writel(hba, ufshcd_readl(hba, REG_UFS_MEM_CFG) | 0x1,
+		      REG_UFS_MEM_CFG);
+	hba->mcq_enabled = true;
+
 	dev_info(hba->dev, "MCQ configured, nr_queues=%d, io_queues=%d, read_queue=%d, poll_queues=%d, queue_depth=%d\n",
 		 hba->nr_hw_queues, hba->nr_queues[HCTX_TYPE_DEFAULT],
 		 hba->nr_queues[HCTX_TYPE_READ], hba->nr_queues[HCTX_TYPE_POLL],
-- 
2.7.4

Re: [PATCH v5 16/16] ufs: core: mcq: Enable Multi Circular Queue

[Manivannan Sadhasivam]

On Tue, Nov 22, 2022 at 08:10:29PM -0800, Asutosh Das wrote:
> Enable MCQ in the Host Controller.
> 
> Signed-off-by: Asutosh Das <quic_asutoshd@quicinc.com>

Reviewed-by: Manivannan Sadhasivam <mani@kernel.org>

Thanks,
Mani

> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
> ---
>  drivers/ufs/core/ufshcd.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 8416d42..41005b8 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -8351,6 +8351,12 @@ static void ufshcd_config_mcq(struct ufs_hba *hba)
>  
>  	hba->host->can_queue = hba->nutrs - UFSHCD_NUM_RESERVED;
>  	hba->reserved_slot = hba->nutrs - UFSHCD_NUM_RESERVED;
> +
> +	/* Select MCQ mode */
> +	ufshcd_writel(hba, ufshcd_readl(hba, REG_UFS_MEM_CFG) | 0x1,
> +		      REG_UFS_MEM_CFG);
> +	hba->mcq_enabled = true;
> +
>  	dev_info(hba->dev, "MCQ configured, nr_queues=%d, io_queues=%d, read_queue=%d, poll_queues=%d, queue_depth=%d\n",
>  		 hba->nr_hw_queues, hba->nr_queues[HCTX_TYPE_DEFAULT],
>  		 hba->nr_queues[HCTX_TYPE_READ], hba->nr_queues[HCTX_TYPE_POLL],
> -- 
> 2.7.4
> 

-- 
மணிவண்ணன் சதாசிவம்