linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH v1 0/4] MHI event ring priority updates
@ 2021-06-17 21:30 Bhaumik Bhatt
  2021-06-17 21:30 ` [PATCH v1 1/4] bus: mhi: core: Add support for processing priority of event ring Bhaumik Bhatt
                   ` (3 more replies)
  0 siblings, 4 replies; 11+ messages in thread
From: Bhaumik Bhatt @ 2021-06-17 21:30 UTC (permalink / raw)
  To: manivannan.sadhasivam
  Cc: linux-arm-msm, hemantk, jhugo, linux-kernel, carl.yin,
	naveen.kumar, loic.poulain, Bhaumik Bhatt

Event ring priorities have remained out of use and are set to 1 by default.
This series brings about an enum entry expanding use of the priority field by
adding high priority tasklet scheduling to the menu.

In some cases, tasklet scheduling delays can be a nuisance. An example would be
ath11k adding a610f3c65d3b ("bus: mhi: Wait for M2 state during system resume")
in response to it. This issue can be potentially avoided if the controller uses
a dedicated event ring for control packets with high priority tasklet scheduling
in place.

There are no functional changes added to the controller (pci_generic or ath11k)
drivers with this series and the only changes there are use of the enum in place
of a harcoded value of "1". In the future, controllers can opt in to have any
event ring scheduled in high priority.

Tested on: X86_64 architecture with SDX65 on Ubuntu 18.04 distribution.

Bhaumik Bhatt (3):
  bus: mhi: pci_generic: Use enum entry for event ring priority
  ath11k: Use enum entry for event ring priority
  bus: mhi: core: Enable support for event ring priorities

Hemant Kumar (1):
  bus: mhi: core: Add support for processing priority of event ring

 drivers/bus/mhi/core/init.c           |  3 +-
 drivers/bus/mhi/core/internal.h       |  2 +-
 drivers/bus/mhi/core/main.c           | 19 ++++++++--
 drivers/bus/mhi/pci_generic.c         | 66 +++++++++++++++++------------------
 drivers/net/wireless/ath/ath11k/mhi.c |  4 +--
 include/linux/mhi.h                   | 14 ++++++--
 6 files changed, 65 insertions(+), 43 deletions(-)

-- 
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project


^ permalink raw reply	[flat|nested] 11+ messages in thread

* [PATCH v1 1/4] bus: mhi: core: Add support for processing priority of event ring
  2021-06-17 21:30 [PATCH v1 0/4] MHI event ring priority updates Bhaumik Bhatt
@ 2021-06-17 21:30 ` Bhaumik Bhatt
  2021-06-18  7:03   ` Manivannan Sadhasivam
  2021-06-17 21:30 ` [PATCH v1 2/4] bus: mhi: pci_generic: Use enum entry for event ring priority Bhaumik Bhatt
                   ` (2 subsequent siblings)
  3 siblings, 1 reply; 11+ messages in thread
From: Bhaumik Bhatt @ 2021-06-17 21:30 UTC (permalink / raw)
  To: manivannan.sadhasivam
  Cc: linux-arm-msm, hemantk, jhugo, linux-kernel, carl.yin,
	naveen.kumar, loic.poulain, Bhaumik Bhatt

From: Hemant Kumar <hemantk@codeaurora.org>

Event ring priorities are currently set to 1 and are unused.
Default processing priority for event rings is set to regular
tasklet. Controllers can choose to use high priority tasklet
scheduling for certain event rings critical for processing such
as ones transporting control information if they wish to avoid
with system scheduling delays for those packets. In order to
support these use cases, allow controllers to set event ring
priority to high. This patch only adds support and does not
enable usage of these priorities.

Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
---
 drivers/bus/mhi/core/internal.h |  2 +-
 drivers/bus/mhi/core/main.c     | 19 ++++++++++++++++---
 include/linux/mhi.h             | 14 ++++++++++++--
 3 files changed, 29 insertions(+), 6 deletions(-)

diff --git a/drivers/bus/mhi/core/internal.h b/drivers/bus/mhi/core/internal.h
index 672052f..666e102 100644
--- a/drivers/bus/mhi/core/internal.h
+++ b/drivers/bus/mhi/core/internal.h
@@ -535,7 +535,7 @@ struct mhi_event {
 	u32 intmod;
 	u32 irq;
 	int chan; /* this event ring is dedicated to a channel (optional) */
-	u32 priority;
+	enum mhi_er_priority priority;
 	enum mhi_er_data_type data_type;
 	struct mhi_ring ring;
 	struct db_cfg db_cfg;
diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c
index 8ac73f9..bfc9776 100644
--- a/drivers/bus/mhi/core/main.c
+++ b/drivers/bus/mhi/core/main.c
@@ -425,10 +425,11 @@ void mhi_create_devices(struct mhi_controller *mhi_cntrl)
 	}
 }
 
-irqreturn_t mhi_irq_handler(int irq_number, void *dev)
+irqreturn_t mhi_irq_handler(int irq_number, void *priv)
 {
-	struct mhi_event *mhi_event = dev;
+	struct mhi_event *mhi_event = priv;
 	struct mhi_controller *mhi_cntrl = mhi_event->mhi_cntrl;
+	struct device *dev = &mhi_cntrl->mhi_dev->dev;
 	struct mhi_event_ctxt *er_ctxt =
 		&mhi_cntrl->mhi_ctxt->er_ctxt[mhi_event->er_index];
 	struct mhi_ring *ev_ring = &mhi_event->ring;
@@ -454,8 +455,20 @@ irqreturn_t mhi_irq_handler(int irq_number, void *dev)
 
 		if (mhi_dev)
 			mhi_notify(mhi_dev, MHI_CB_PENDING_DATA);
-	} else {
+
+		return IRQ_HANDLED;
+	}
+
+	switch (mhi_event->priority) {
+	case MHI_ER_PRIORITY_HI:
+		tasklet_hi_schedule(&mhi_event->task);
+		break;
+	case MHI_ER_PRIORITY_DEFAULT:
 		tasklet_schedule(&mhi_event->task);
+		break;
+	default:
+		dev_err(dev, "Skip event of unknown priority\n");
+		break;
 	}
 
 	return IRQ_HANDLED;
diff --git a/include/linux/mhi.h b/include/linux/mhi.h
index 86cea52..25ee312 100644
--- a/include/linux/mhi.h
+++ b/include/linux/mhi.h
@@ -198,6 +198,16 @@ enum mhi_er_data_type {
 };
 
 /**
+ * enum mhi_er_priority - Event ring processing priority
+ * @MHI_ER_PRIORITY_HI: processed by hi-priority tasklet
+ * @MHI_ER_PRIORITY_DEFAULT: processed by regular tasklet
+ */
+enum mhi_er_priority {
+	MHI_ER_PRIORITY_HI,
+	MHI_ER_PRIORITY_DEFAULT,
+};
+
+/**
  * enum mhi_db_brst_mode - Doorbell mode
  * @MHI_DB_BRST_DISABLE: Burst mode disable
  * @MHI_DB_BRST_ENABLE: Burst mode enable
@@ -250,7 +260,7 @@ struct mhi_channel_config {
  * @irq_moderation_ms: Delay irq for additional events to be aggregated
  * @irq: IRQ associated with this ring
  * @channel: Dedicated channel number. U32_MAX indicates a non-dedicated ring
- * @priority: Priority of this ring. Use 1 for now
+ * @priority: Processing priority of this ring.
  * @mode: Doorbell mode
  * @data_type: Type of data this ring will process
  * @hardware_event: This ring is associated with hardware channels
@@ -262,7 +272,7 @@ struct mhi_event_config {
 	u32 irq_moderation_ms;
 	u32 irq;
 	u32 channel;
-	u32 priority;
+	enum mhi_er_priority priority;
 	enum mhi_db_brst_mode mode;
 	enum mhi_er_data_type data_type;
 	bool hardware_event;
-- 
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project


^ permalink raw reply related	[flat|nested] 11+ messages in thread

* [PATCH v1 2/4] bus: mhi: pci_generic: Use enum entry for event ring priority
  2021-06-17 21:30 [PATCH v1 0/4] MHI event ring priority updates Bhaumik Bhatt
  2021-06-17 21:30 ` [PATCH v1 1/4] bus: mhi: core: Add support for processing priority of event ring Bhaumik Bhatt
@ 2021-06-17 21:30 ` Bhaumik Bhatt
  2021-06-17 22:21   ` Hemant Kumar
  2021-06-17 21:30 ` [PATCH v1 3/4] ath11k: " Bhaumik Bhatt
  2021-06-17 21:30 ` [PATCH v1 4/4] bus: mhi: core: Enable support for event ring priorities Bhaumik Bhatt
  3 siblings, 1 reply; 11+ messages in thread
From: Bhaumik Bhatt @ 2021-06-17 21:30 UTC (permalink / raw)
  To: manivannan.sadhasivam
  Cc: linux-arm-msm, hemantk, jhugo, linux-kernel, carl.yin,
	naveen.kumar, loic.poulain, Bhaumik Bhatt

Instead of using a default event ring priority of 1, use the enum
provided and set it to default.

Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
---
 drivers/bus/mhi/pci_generic.c | 66 +++++++++++++++++++++----------------------
 1 file changed, 33 insertions(+), 33 deletions(-)

diff --git a/drivers/bus/mhi/pci_generic.c b/drivers/bus/mhi/pci_generic.c
index 31360a2..5886547 100644
--- a/drivers/bus/mhi/pci_generic.c
+++ b/drivers/bus/mhi/pci_generic.c
@@ -74,17 +74,17 @@ struct mhi_pci_dev_info {
 		.doorbell_mode_switch = false,		\
 	}
 
-#define MHI_EVENT_CONFIG_CTRL(ev_ring, el_count) \
-	{					\
-		.num_elements = el_count,	\
-		.irq_moderation_ms = 0,		\
-		.irq = (ev_ring) + 1,		\
-		.priority = 1,			\
-		.mode = MHI_DB_BRST_DISABLE,	\
-		.data_type = MHI_ER_CTRL,	\
-		.hardware_event = false,	\
-		.client_managed = false,	\
-		.offload_channel = false,	\
+#define MHI_EVENT_CONFIG_CTRL(ev_ring, el_count)	\
+	{						\
+		.num_elements = el_count,		\
+		.irq_moderation_ms = 0,			\
+		.irq = (ev_ring) + 1,			\
+		.priority = MHI_ER_PRIORITY_DEFAULT,	\
+		.mode = MHI_DB_BRST_DISABLE,		\
+		.data_type = MHI_ER_CTRL,		\
+		.hardware_event = false,		\
+		.client_managed = false,		\
+		.offload_channel = false,		\
 	}
 
 #define MHI_CHANNEL_CONFIG_HW_UL(ch_num, ch_name, el_count, ev_ring) \
@@ -177,31 +177,31 @@ struct mhi_pci_dev_info {
 		.doorbell_mode_switch = false,		\
 	}
 
-#define MHI_EVENT_CONFIG_DATA(ev_ring, el_count) \
-	{					\
-		.num_elements = el_count,	\
-		.irq_moderation_ms = 5,		\
-		.irq = (ev_ring) + 1,		\
-		.priority = 1,			\
-		.mode = MHI_DB_BRST_DISABLE,	\
-		.data_type = MHI_ER_DATA,	\
-		.hardware_event = false,	\
-		.client_managed = false,	\
-		.offload_channel = false,	\
+#define MHI_EVENT_CONFIG_DATA(ev_ring, el_count)	\
+	{						\
+		.num_elements = el_count,		\
+		.irq_moderation_ms = 5,			\
+		.irq = (ev_ring) + 1,			\
+		.priority = MHI_ER_PRIORITY_DEFAULT,	\
+		.mode = MHI_DB_BRST_DISABLE,		\
+		.data_type = MHI_ER_DATA,		\
+		.hardware_event = false,		\
+		.client_managed = false,		\
+		.offload_channel = false,		\
 	}
 
 #define MHI_EVENT_CONFIG_HW_DATA(ev_ring, el_count, ch_num) \
-	{					\
-		.num_elements = el_count,	\
-		.irq_moderation_ms = 1,		\
-		.irq = (ev_ring) + 1,		\
-		.priority = 1,			\
-		.mode = MHI_DB_BRST_DISABLE,	\
-		.data_type = MHI_ER_DATA,	\
-		.hardware_event = true,		\
-		.client_managed = false,	\
-		.offload_channel = false,	\
-		.channel = ch_num,		\
+	{						\
+		.num_elements = el_count,		\
+		.irq_moderation_ms = 1,			\
+		.irq = (ev_ring) + 1,			\
+		.priority = MHI_ER_PRIORITY_DEFAULT,	\
+		.mode = MHI_DB_BRST_DISABLE,		\
+		.data_type = MHI_ER_DATA,		\
+		.hardware_event = true,			\
+		.client_managed = false,		\
+		.offload_channel = false,		\
+		.channel = ch_num,			\
 	}
 
 static const struct mhi_channel_config modem_qcom_v1_mhi_channels[] = {
-- 
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project


^ permalink raw reply related	[flat|nested] 11+ messages in thread

* [PATCH v1 3/4] ath11k: Use enum entry for event ring priority
  2021-06-17 21:30 [PATCH v1 0/4] MHI event ring priority updates Bhaumik Bhatt
  2021-06-17 21:30 ` [PATCH v1 1/4] bus: mhi: core: Add support for processing priority of event ring Bhaumik Bhatt
  2021-06-17 21:30 ` [PATCH v1 2/4] bus: mhi: pci_generic: Use enum entry for event ring priority Bhaumik Bhatt
@ 2021-06-17 21:30 ` Bhaumik Bhatt
  2021-06-17 21:30 ` [PATCH v1 4/4] bus: mhi: core: Enable support for event ring priorities Bhaumik Bhatt
  3 siblings, 0 replies; 11+ messages in thread
From: Bhaumik Bhatt @ 2021-06-17 21:30 UTC (permalink / raw)
  To: manivannan.sadhasivam
  Cc: linux-arm-msm, hemantk, jhugo, linux-kernel, carl.yin,
	naveen.kumar, loic.poulain, Bhaumik Bhatt

Instead of using a default event ring priority of 1, use the enum
provided and set it to default.

Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
---
 drivers/net/wireless/ath/ath11k/mhi.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/wireless/ath/ath11k/mhi.c b/drivers/net/wireless/ath/ath11k/mhi.c
index 27b394d..b7864fc 100644
--- a/drivers/net/wireless/ath/ath11k/mhi.c
+++ b/drivers/net/wireless/ath/ath11k/mhi.c
@@ -86,7 +86,7 @@ static struct mhi_event_config ath11k_mhi_events_qca6390[] = {
 		.irq_moderation_ms = 1,
 		.irq = 2,
 		.mode = MHI_DB_BRST_DISABLE,
-		.priority = 1,
+		.priority = MHI_ER_PRIORITY_DEFAULT,
 		.hardware_event = false,
 		.client_managed = false,
 		.offload_channel = false,
@@ -179,7 +179,7 @@ static struct mhi_event_config ath11k_mhi_events_qcn9074[] = {
 		.irq_moderation_ms = 1,
 		.irq = 2,
 		.mode = MHI_DB_BRST_DISABLE,
-		.priority = 1,
+		.priority = MHI_ER_PRIORITY_DEFAULT,
 		.hardware_event = false,
 		.client_managed = false,
 		.offload_channel = false,
-- 
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project


^ permalink raw reply related	[flat|nested] 11+ messages in thread

* [PATCH v1 4/4] bus: mhi: core: Enable support for event ring priorities
  2021-06-17 21:30 [PATCH v1 0/4] MHI event ring priority updates Bhaumik Bhatt
                   ` (2 preceding siblings ...)
  2021-06-17 21:30 ` [PATCH v1 3/4] ath11k: " Bhaumik Bhatt
@ 2021-06-17 21:30 ` Bhaumik Bhatt
  2021-06-17 22:20   ` Hemant Kumar
  3 siblings, 1 reply; 11+ messages in thread
From: Bhaumik Bhatt @ 2021-06-17 21:30 UTC (permalink / raw)
  To: manivannan.sadhasivam
  Cc: linux-arm-msm, hemantk, jhugo, linux-kernel, carl.yin,
	naveen.kumar, loic.poulain, Bhaumik Bhatt

Priority set in the event ring configuration by the controllers
has been unused by the MHI host driver until now. Enable usage of
those priorities set with the introduction of the enums for
default or high priority tasklet scheduling options.

Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
---
 drivers/bus/mhi/core/init.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/bus/mhi/core/init.c b/drivers/bus/mhi/core/init.c
index c81b377..4446760 100644
--- a/drivers/bus/mhi/core/init.c
+++ b/drivers/bus/mhi/core/init.c
@@ -673,8 +673,7 @@ static int parse_ev_cfg(struct mhi_controller *mhi_cntrl,
 				&mhi_cntrl->mhi_chan[mhi_event->chan];
 		}
 
-		/* Priority is fixed to 1 for now */
-		mhi_event->priority = 1;
+		mhi_event->priority = event_cfg->priority;
 
 		mhi_event->db_cfg.brstmode = event_cfg->mode;
 		if (MHI_INVALID_BRSTMODE(mhi_event->db_cfg.brstmode))
-- 
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project


^ permalink raw reply related	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 4/4] bus: mhi: core: Enable support for event ring priorities
  2021-06-17 21:30 ` [PATCH v1 4/4] bus: mhi: core: Enable support for event ring priorities Bhaumik Bhatt
@ 2021-06-17 22:20   ` Hemant Kumar
  0 siblings, 0 replies; 11+ messages in thread
From: Hemant Kumar @ 2021-06-17 22:20 UTC (permalink / raw)
  To: Bhaumik Bhatt, manivannan.sadhasivam
  Cc: linux-arm-msm, jhugo, linux-kernel, carl.yin, naveen.kumar, loic.poulain

On Thu, 2021-06-17 at 14:30 -0700, Bhaumik Bhatt wrote:
> Priority set in the event ring configuration by the controllers
> has been unused by the MHI host driver until now. Enable usage of
> those priorities set with the introduction of the enums for
> default or high priority tasklet scheduling options.
> 
> Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
> 
Reviewed-by: Hemant Kumar <hemantk@codeaurora.org>
-- 
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project


^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 2/4] bus: mhi: pci_generic: Use enum entry for event ring priority
  2021-06-17 21:30 ` [PATCH v1 2/4] bus: mhi: pci_generic: Use enum entry for event ring priority Bhaumik Bhatt
@ 2021-06-17 22:21   ` Hemant Kumar
  0 siblings, 0 replies; 11+ messages in thread
From: Hemant Kumar @ 2021-06-17 22:21 UTC (permalink / raw)
  To: Bhaumik Bhatt, manivannan.sadhasivam
  Cc: linux-arm-msm, jhugo, linux-kernel, carl.yin, naveen.kumar, loic.poulain

On Thu, 2021-06-17 at 14:30 -0700, Bhaumik Bhatt wrote:
> Instead of using a default event ring priority of 1, use the enum
> provided and set it to default.
> 
> Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
> ---
Reviewed-by: Hemant Kumar <hemantk@codeaurora.org>
-- 
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project


^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 1/4] bus: mhi: core: Add support for processing priority of event ring
  2021-06-17 21:30 ` [PATCH v1 1/4] bus: mhi: core: Add support for processing priority of event ring Bhaumik Bhatt
@ 2021-06-18  7:03   ` Manivannan Sadhasivam
  2021-06-18 17:17     ` Bhaumik Bhatt
  0 siblings, 1 reply; 11+ messages in thread
From: Manivannan Sadhasivam @ 2021-06-18  7:03 UTC (permalink / raw)
  To: Bhaumik Bhatt
  Cc: linux-arm-msm, hemantk, jhugo, linux-kernel, carl.yin,
	naveen.kumar, loic.poulain

On Thu, Jun 17, 2021 at 02:30:32PM -0700, Bhaumik Bhatt wrote:
> From: Hemant Kumar <hemantk@codeaurora.org>
> 
> Event ring priorities are currently set to 1 and are unused.
> Default processing priority for event rings is set to regular
> tasklet. Controllers can choose to use high priority tasklet
> scheduling for certain event rings critical for processing such
> as ones transporting control information if they wish to avoid
> with system scheduling delays for those packets. In order to
> support these use cases, allow controllers to set event ring
> priority to high. This patch only adds support and does not
> enable usage of these priorities.
> 
> Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
> Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
> ---
>  drivers/bus/mhi/core/internal.h |  2 +-
>  drivers/bus/mhi/core/main.c     | 19 ++++++++++++++++---
>  include/linux/mhi.h             | 14 ++++++++++++--
>  3 files changed, 29 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/bus/mhi/core/internal.h b/drivers/bus/mhi/core/internal.h
> index 672052f..666e102 100644
> --- a/drivers/bus/mhi/core/internal.h
> +++ b/drivers/bus/mhi/core/internal.h
> @@ -535,7 +535,7 @@ struct mhi_event {
>  	u32 intmod;
>  	u32 irq;
>  	int chan; /* this event ring is dedicated to a channel (optional) */
> -	u32 priority;
> +	enum mhi_er_priority priority;

Instead of using enum for priorities, can we just make use of the
existing "priority" field? Since the existing controllers set it to "1",
can we use "0" as the high priority?

This way we don't need to change the controller drivers.

>  	enum mhi_er_data_type data_type;
>  	struct mhi_ring ring;
>  	struct db_cfg db_cfg;
> diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c
> index 8ac73f9..bfc9776 100644
> --- a/drivers/bus/mhi/core/main.c
> +++ b/drivers/bus/mhi/core/main.c
> @@ -425,10 +425,11 @@ void mhi_create_devices(struct mhi_controller *mhi_cntrl)
>  	}
>  }
>  
> -irqreturn_t mhi_irq_handler(int irq_number, void *dev)
> +irqreturn_t mhi_irq_handler(int irq_number, void *priv)
>  {
> -	struct mhi_event *mhi_event = dev;
> +	struct mhi_event *mhi_event = priv;
>  	struct mhi_controller *mhi_cntrl = mhi_event->mhi_cntrl;
> +	struct device *dev = &mhi_cntrl->mhi_dev->dev;
>  	struct mhi_event_ctxt *er_ctxt =
>  		&mhi_cntrl->mhi_ctxt->er_ctxt[mhi_event->er_index];
>  	struct mhi_ring *ev_ring = &mhi_event->ring;
> @@ -454,8 +455,20 @@ irqreturn_t mhi_irq_handler(int irq_number, void *dev)
>  
>  		if (mhi_dev)
>  			mhi_notify(mhi_dev, MHI_CB_PENDING_DATA);
> -	} else {
> +
> +		return IRQ_HANDLED;
> +	}
> +
> +	switch (mhi_event->priority) {
> +	case MHI_ER_PRIORITY_HI:

This could be,

	/* Use high priority tasklet for high priority event ring */
	if (!mhi_event->priority)
		tasklet_hi_schedule(&mhi_event->task);
	else
		tasklet_schedule(&mhi_event->task);

Thanks,
Mani

> +		tasklet_hi_schedule(&mhi_event->task);
> +		break;
> +	case MHI_ER_PRIORITY_DEFAULT:
>  		tasklet_schedule(&mhi_event->task);
> +		break;
> +	default:
> +		dev_err(dev, "Skip event of unknown priority\n");
> +		break;
>  	}
>  
>  	return IRQ_HANDLED;
> diff --git a/include/linux/mhi.h b/include/linux/mhi.h
> index 86cea52..25ee312 100644
> --- a/include/linux/mhi.h
> +++ b/include/linux/mhi.h
> @@ -198,6 +198,16 @@ enum mhi_er_data_type {
>  };
>  
>  /**
> + * enum mhi_er_priority - Event ring processing priority
> + * @MHI_ER_PRIORITY_HI: processed by hi-priority tasklet
> + * @MHI_ER_PRIORITY_DEFAULT: processed by regular tasklet
> + */
> +enum mhi_er_priority {
> +	MHI_ER_PRIORITY_HI,
> +	MHI_ER_PRIORITY_DEFAULT,
> +};
> +
> +/**
>   * enum mhi_db_brst_mode - Doorbell mode
>   * @MHI_DB_BRST_DISABLE: Burst mode disable
>   * @MHI_DB_BRST_ENABLE: Burst mode enable
> @@ -250,7 +260,7 @@ struct mhi_channel_config {
>   * @irq_moderation_ms: Delay irq for additional events to be aggregated
>   * @irq: IRQ associated with this ring
>   * @channel: Dedicated channel number. U32_MAX indicates a non-dedicated ring
> - * @priority: Priority of this ring. Use 1 for now
> + * @priority: Processing priority of this ring.
>   * @mode: Doorbell mode
>   * @data_type: Type of data this ring will process
>   * @hardware_event: This ring is associated with hardware channels
> @@ -262,7 +272,7 @@ struct mhi_event_config {
>  	u32 irq_moderation_ms;
>  	u32 irq;
>  	u32 channel;
> -	u32 priority;
> +	enum mhi_er_priority priority;
>  	enum mhi_db_brst_mode mode;
>  	enum mhi_er_data_type data_type;
>  	bool hardware_event;
> -- 
> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
> a Linux Foundation Collaborative Project
> 

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 1/4] bus: mhi: core: Add support for processing priority of event ring
  2021-06-18  7:03   ` Manivannan Sadhasivam
@ 2021-06-18 17:17     ` Bhaumik Bhatt
  2021-06-18 17:31       ` Manivannan Sadhasivam
  0 siblings, 1 reply; 11+ messages in thread
From: Bhaumik Bhatt @ 2021-06-18 17:17 UTC (permalink / raw)
  To: Manivannan Sadhasivam
  Cc: linux-arm-msm, hemantk, jhugo, linux-kernel, carl.yin,
	naveen.kumar, loic.poulain

Hi Mani,

On 2021-06-18 12:03 AM, Manivannan Sadhasivam wrote:
> On Thu, Jun 17, 2021 at 02:30:32PM -0700, Bhaumik Bhatt wrote:
>> From: Hemant Kumar <hemantk@codeaurora.org>
>> 
>> Event ring priorities are currently set to 1 and are unused.
>> Default processing priority for event rings is set to regular
>> tasklet. Controllers can choose to use high priority tasklet
>> scheduling for certain event rings critical for processing such
>> as ones transporting control information if they wish to avoid
>> with system scheduling delays for those packets. In order to
>> support these use cases, allow controllers to set event ring
>> priority to high. This patch only adds support and does not
>> enable usage of these priorities.
>> 
>> Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
>> Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
>> ---
>>  drivers/bus/mhi/core/internal.h |  2 +-
>>  drivers/bus/mhi/core/main.c     | 19 ++++++++++++++++---
>>  include/linux/mhi.h             | 14 ++++++++++++--
>>  3 files changed, 29 insertions(+), 6 deletions(-)
>> 
>> diff --git a/drivers/bus/mhi/core/internal.h 
>> b/drivers/bus/mhi/core/internal.h
>> index 672052f..666e102 100644
>> --- a/drivers/bus/mhi/core/internal.h
>> +++ b/drivers/bus/mhi/core/internal.h
>> @@ -535,7 +535,7 @@ struct mhi_event {
>>  	u32 intmod;
>>  	u32 irq;
>>  	int chan; /* this event ring is dedicated to a channel (optional) */
>> -	u32 priority;
>> +	enum mhi_er_priority priority;
> 
> Instead of using enum for priorities, can we just make use of the
> existing "priority" field? Since the existing controllers set it to 
> "1",
> can we use "0" as the high priority?
> 
> This way we don't need to change the controller drivers.
> 
I agree but the reasons to do the enum approach was to allow for future
expansion of the handling if it becomes necessary and provide clarity 
for
the field.

I can always do it this way for now and we can have the enum for another
time but would prefer updating what we have now.
>>  	enum mhi_er_data_type data_type;
>>  	struct mhi_ring ring;
>>  	struct db_cfg db_cfg;
>> diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c
>> index 8ac73f9..bfc9776 100644
>> --- a/drivers/bus/mhi/core/main.c
>> +++ b/drivers/bus/mhi/core/main.c
>> @@ -425,10 +425,11 @@ void mhi_create_devices(struct mhi_controller 
>> *mhi_cntrl)
>>  	}
>>  }
>> 
>> -irqreturn_t mhi_irq_handler(int irq_number, void *dev)
>> +irqreturn_t mhi_irq_handler(int irq_number, void *priv)
>>  {
>> -	struct mhi_event *mhi_event = dev;
>> +	struct mhi_event *mhi_event = priv;
>>  	struct mhi_controller *mhi_cntrl = mhi_event->mhi_cntrl;
>> +	struct device *dev = &mhi_cntrl->mhi_dev->dev;
>>  	struct mhi_event_ctxt *er_ctxt =
>>  		&mhi_cntrl->mhi_ctxt->er_ctxt[mhi_event->er_index];
>>  	struct mhi_ring *ev_ring = &mhi_event->ring;
>> @@ -454,8 +455,20 @@ irqreturn_t mhi_irq_handler(int irq_number, void 
>> *dev)
>> 
>>  		if (mhi_dev)
>>  			mhi_notify(mhi_dev, MHI_CB_PENDING_DATA);
>> -	} else {
>> +
>> +		return IRQ_HANDLED;
>> +	}
>> +
>> +	switch (mhi_event->priority) {
>> +	case MHI_ER_PRIORITY_HI:
> 
> This could be,
> 
> 	/* Use high priority tasklet for high priority event ring */
> 	if (!mhi_event->priority)
> 		tasklet_hi_schedule(&mhi_event->task);
> 	else
> 		tasklet_schedule(&mhi_event->task);
> 
> Thanks,
> Mani
> 
Yes, this works too if we keep the current non-enum approach.
>> +		tasklet_hi_schedule(&mhi_event->task);
>> +		break;
>> +	case MHI_ER_PRIORITY_DEFAULT:
>>  		tasklet_schedule(&mhi_event->task);
>> +		break;
>> +	default:
>> +		dev_err(dev, "Skip event of unknown priority\n");
>> +		break;
>>  	}
>> 
>>  	return IRQ_HANDLED;
>> diff --git a/include/linux/mhi.h b/include/linux/mhi.h
>> index 86cea52..25ee312 100644
>> --- a/include/linux/mhi.h
>> +++ b/include/linux/mhi.h
>> @@ -198,6 +198,16 @@ enum mhi_er_data_type {
>>  };
>> 
>>  /**
>> + * enum mhi_er_priority - Event ring processing priority
>> + * @MHI_ER_PRIORITY_HI: processed by hi-priority tasklet
>> + * @MHI_ER_PRIORITY_DEFAULT: processed by regular tasklet
>> + */
>> +enum mhi_er_priority {
>> +	MHI_ER_PRIORITY_HI,
>> +	MHI_ER_PRIORITY_DEFAULT,
>> +};
>> +
>> +/**
>>   * enum mhi_db_brst_mode - Doorbell mode
>>   * @MHI_DB_BRST_DISABLE: Burst mode disable
>>   * @MHI_DB_BRST_ENABLE: Burst mode enable
>> @@ -250,7 +260,7 @@ struct mhi_channel_config {
>>   * @irq_moderation_ms: Delay irq for additional events to be 
>> aggregated
>>   * @irq: IRQ associated with this ring
>>   * @channel: Dedicated channel number. U32_MAX indicates a 
>> non-dedicated ring
>> - * @priority: Priority of this ring. Use 1 for now
>> + * @priority: Processing priority of this ring.
>>   * @mode: Doorbell mode
>>   * @data_type: Type of data this ring will process
>>   * @hardware_event: This ring is associated with hardware channels
>> @@ -262,7 +272,7 @@ struct mhi_event_config {
>>  	u32 irq_moderation_ms;
>>  	u32 irq;
>>  	u32 channel;
>> -	u32 priority;
>> +	enum mhi_er_priority priority;
>>  	enum mhi_db_brst_mode mode;
>>  	enum mhi_er_data_type data_type;
>>  	bool hardware_event;
>> --
>> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora 
>> Forum,
>> a Linux Foundation Collaborative Project
>> 
Existing controllers would be fine.

Do you think we have a problem if a new controller blindly inputs a "0" 
in
the priority not knowing the impact of it?

If you give me a go ahead, I can make these changes in v2 and leave the 
enum
stuff out.

Thanks,
Bhaumik
---
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora 
Forum,
a Linux Foundation Collaborative Project

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 1/4] bus: mhi: core: Add support for processing priority of event ring
  2021-06-18 17:17     ` Bhaumik Bhatt
@ 2021-06-18 17:31       ` Manivannan Sadhasivam
  2021-06-18 17:43         ` Bhaumik Bhatt
  0 siblings, 1 reply; 11+ messages in thread
From: Manivannan Sadhasivam @ 2021-06-18 17:31 UTC (permalink / raw)
  To: Bhaumik Bhatt
  Cc: linux-arm-msm, hemantk, jhugo, linux-kernel, carl.yin,
	naveen.kumar, loic.poulain

On Fri, Jun 18, 2021 at 10:17:59AM -0700, Bhaumik Bhatt wrote:
> Hi Mani,
> 
> On 2021-06-18 12:03 AM, Manivannan Sadhasivam wrote:
> > On Thu, Jun 17, 2021 at 02:30:32PM -0700, Bhaumik Bhatt wrote:
> > > From: Hemant Kumar <hemantk@codeaurora.org>
> > > 
> > > Event ring priorities are currently set to 1 and are unused.
> > > Default processing priority for event rings is set to regular
> > > tasklet. Controllers can choose to use high priority tasklet
> > > scheduling for certain event rings critical for processing such
> > > as ones transporting control information if they wish to avoid
> > > with system scheduling delays for those packets. In order to
> > > support these use cases, allow controllers to set event ring
> > > priority to high. This patch only adds support and does not
> > > enable usage of these priorities.
> > > 
> > > Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
> > > Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
> > > ---
> > >  drivers/bus/mhi/core/internal.h |  2 +-
> > >  drivers/bus/mhi/core/main.c     | 19 ++++++++++++++++---
> > >  include/linux/mhi.h             | 14 ++++++++++++--
> > >  3 files changed, 29 insertions(+), 6 deletions(-)
> > > 
> > > diff --git a/drivers/bus/mhi/core/internal.h
> > > b/drivers/bus/mhi/core/internal.h
> > > index 672052f..666e102 100644
> > > --- a/drivers/bus/mhi/core/internal.h
> > > +++ b/drivers/bus/mhi/core/internal.h
> > > @@ -535,7 +535,7 @@ struct mhi_event {
> > >  	u32 intmod;
> > >  	u32 irq;
> > >  	int chan; /* this event ring is dedicated to a channel (optional) */
> > > -	u32 priority;
> > > +	enum mhi_er_priority priority;
> > 
> > Instead of using enum for priorities, can we just make use of the
> > existing "priority" field? Since the existing controllers set it to "1",
> > can we use "0" as the high priority?
> > 
> > This way we don't need to change the controller drivers.
> > 
> I agree but the reasons to do the enum approach was to allow for future
> expansion of the handling if it becomes necessary and provide clarity for
> the field.
> 
> I can always do it this way for now and we can have the enum for another
> time but would prefer updating what we have now.

Yeah, let's deal with it later once the necessity arises.

> > >  	enum mhi_er_data_type data_type;
> > >  	struct mhi_ring ring;
> > >  	struct db_cfg db_cfg;
> > > diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c
> > > index 8ac73f9..bfc9776 100644
> > > --- a/drivers/bus/mhi/core/main.c
> > > +++ b/drivers/bus/mhi/core/main.c
> > > @@ -425,10 +425,11 @@ void mhi_create_devices(struct mhi_controller
> > > *mhi_cntrl)
> > >  	}
> > >  }
> > > 

[...]

> Existing controllers would be fine.
> 
> Do you think we have a problem if a new controller blindly inputs a "0" in
> the priority not knowing the impact of it?
> 

We should document it in the kernel doc for the struct field and that
should be enough. We can't do much if people doesn't read the doc ;)

Thanks,
Mani

> If you give me a go ahead, I can make these changes in v2 and leave the enum
> stuff out.
> 
> Thanks,
> Bhaumik
> ---
> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
> a Linux Foundation Collaborative Project

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 1/4] bus: mhi: core: Add support for processing priority of event ring
  2021-06-18 17:31       ` Manivannan Sadhasivam
@ 2021-06-18 17:43         ` Bhaumik Bhatt
  0 siblings, 0 replies; 11+ messages in thread
From: Bhaumik Bhatt @ 2021-06-18 17:43 UTC (permalink / raw)
  To: Manivannan Sadhasivam
  Cc: linux-arm-msm, hemantk, jhugo, linux-kernel, carl.yin,
	naveen.kumar, loic.poulain

Hi Mani,
On 2021-06-18 10:31 AM, Manivannan Sadhasivam wrote:
> On Fri, Jun 18, 2021 at 10:17:59AM -0700, Bhaumik Bhatt wrote:
>> Hi Mani,
>> 
>> On 2021-06-18 12:03 AM, Manivannan Sadhasivam wrote:
>> > On Thu, Jun 17, 2021 at 02:30:32PM -0700, Bhaumik Bhatt wrote:
>> > > From: Hemant Kumar <hemantk@codeaurora.org>
>> > >
>> > > Event ring priorities are currently set to 1 and are unused.
>> > > Default processing priority for event rings is set to regular
>> > > tasklet. Controllers can choose to use high priority tasklet
>> > > scheduling for certain event rings critical for processing such
>> > > as ones transporting control information if they wish to avoid
>> > > with system scheduling delays for those packets. In order to
>> > > support these use cases, allow controllers to set event ring
>> > > priority to high. This patch only adds support and does not
>> > > enable usage of these priorities.
>> > >
>> > > Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
>> > > Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
>> > > ---
>> > >  drivers/bus/mhi/core/internal.h |  2 +-
>> > >  drivers/bus/mhi/core/main.c     | 19 ++++++++++++++++---
>> > >  include/linux/mhi.h             | 14 ++++++++++++--
>> > >  3 files changed, 29 insertions(+), 6 deletions(-)
>> > >
>> > > diff --git a/drivers/bus/mhi/core/internal.h
>> > > b/drivers/bus/mhi/core/internal.h
>> > > index 672052f..666e102 100644
>> > > --- a/drivers/bus/mhi/core/internal.h
>> > > +++ b/drivers/bus/mhi/core/internal.h
>> > > @@ -535,7 +535,7 @@ struct mhi_event {
>> > >  	u32 intmod;
>> > >  	u32 irq;
>> > >  	int chan; /* this event ring is dedicated to a channel (optional) */
>> > > -	u32 priority;
>> > > +	enum mhi_er_priority priority;
>> >
>> > Instead of using enum for priorities, can we just make use of the
>> > existing "priority" field? Since the existing controllers set it to "1",
>> > can we use "0" as the high priority?
>> >
>> > This way we don't need to change the controller drivers.
>> >
>> I agree but the reasons to do the enum approach was to allow for 
>> future
>> expansion of the handling if it becomes necessary and provide clarity 
>> for
>> the field.
>> 
>> I can always do it this way for now and we can have the enum for 
>> another
>> time but would prefer updating what we have now.
> 
> Yeah, let's deal with it later once the necessity arises.
> 
Sure. I will make the v2.

>> > >  	enum mhi_er_data_type data_type;
>> > >  	struct mhi_ring ring;
>> > >  	struct db_cfg db_cfg;
>> > > diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c
>> > > index 8ac73f9..bfc9776 100644
>> > > --- a/drivers/bus/mhi/core/main.c
>> > > +++ b/drivers/bus/mhi/core/main.c
>> > > @@ -425,10 +425,11 @@ void mhi_create_devices(struct mhi_controller
>> > > *mhi_cntrl)
>> > >  	}
>> > >  }
>> > >
> 
> [...]
> 
>> Existing controllers would be fine.
>> 
>> Do you think we have a problem if a new controller blindly inputs a 
>> "0" in
>> the priority not knowing the impact of it?
>> 
> 
> We should document it in the kernel doc for the struct field and that
> should be enough. We can't do much if people doesn't read the doc ;)
> 
> Thanks,
> Mani
> 
Totally agree :)

>> If you give me a go ahead, I can make these changes in v2 and leave 
>> the enum
>> stuff out.
>> 
>> Thanks,
>> Bhaumik
>> ---
>> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora 
>> Forum,
>> a Linux Foundation Collaborative Project

Thanks,
Bhaumik
---
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora 
Forum,
a Linux Foundation Collaborative Project

^ permalink raw reply	[flat|nested] 11+ messages in thread

end of thread, other threads:[~2021-06-18 17:43 UTC | newest]

Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-06-17 21:30 [PATCH v1 0/4] MHI event ring priority updates Bhaumik Bhatt
2021-06-17 21:30 ` [PATCH v1 1/4] bus: mhi: core: Add support for processing priority of event ring Bhaumik Bhatt
2021-06-18  7:03   ` Manivannan Sadhasivam
2021-06-18 17:17     ` Bhaumik Bhatt
2021-06-18 17:31       ` Manivannan Sadhasivam
2021-06-18 17:43         ` Bhaumik Bhatt
2021-06-17 21:30 ` [PATCH v1 2/4] bus: mhi: pci_generic: Use enum entry for event ring priority Bhaumik Bhatt
2021-06-17 22:21   ` Hemant Kumar
2021-06-17 21:30 ` [PATCH v1 3/4] ath11k: " Bhaumik Bhatt
2021-06-17 21:30 ` [PATCH v1 4/4] bus: mhi: core: Enable support for event ring priorities Bhaumik Bhatt
2021-06-17 22:20   ` Hemant Kumar

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).