Skip to contentSkip to conversation

mfd: iqs62x: Minor cosmetic and functional improvements

PATCH Applied Jeff LaBundy 2 participants 13 messages Jan 18, 2021 – Jan 20, 2021 Multifunction devices (MFD)

lore
Share & tools
lore
#0 lore raw

v2 cover mfd: iqs62x: Minor cosmetic and functional improvements

This series includes a variety of minor cosmetic and functional improvements to the core support for the Azoteq IQS620A, IQS621, IQS622, IQS624 and IQS625 multi-function sensors.

The first three patches are purely cosmetic, while the remaining three incorporate learnings during recent work with other Azoteq devices that have similar functions.

Jeff LaBundy (6):
mfd: iqs62x: Remove superfluous whitespace above fallthroughs
mfd: iqs62x: Remove unused bit mask
mfd: iqs62x: Rename regmap_config struct
mfd: iqs62x: Increase interrupt handler return delay
mfd: iqs62x: Do not poll during ATI
mfd: iqs62x: Do not change clock frequency during ATI

 drivers/mfd/iqs62x.c       | 144 +++++++++++++++++++++++++--------------------
 include/linux/mfd/iqs62x.h |  11 ++--
 2 files changed, 85 insertions(+), 70 deletions(-)

--
2.7.4

#1 lore raw

PATCH v2 1/6 mfd: iqs62x: Remove superfluous whitespace above fallthroughs

Previously, all instances of the /* fall through */ comment were preceded by a newline to improve readability.

Now that /* fall through */ comments have been replaced with the fallthrough pseudo-keyword, the leftover whitespace looks out of place and can simply be removed.

Fixes df561f6688fe ("treewide: Use fallthrough pseudo-keyword")

---
Changes in v2:
- Added Acked-for-MFD-by trailer

 drivers/mfd/iqs62x.c | 3 ---
 1 file changed, 3 deletions(-)
drivers/mfd/iqs62x.c +0 -3
index 761b4ef..ec4c790 100644
@@ -135,7 +135,6 @@ static int iqs62x_dev_init(struct iqs62x_core *iqs62x)
 
 		if (val & IQS620_PROX_SETTINGS_4_SAR_EN)
 			iqs62x->ui_sel = IQS62X_UI_SAR1;
-
 		fallthrough;
 
 	case IQS621_PROD_NUM:
@@ -469,7 +468,6 @@ static irqreturn_t iqs62x_irq(int irq, void *context)
 		switch (event_reg) {
 		case IQS62X_EVENT_UI_LO:
 			event_data.ui_data = get_unaligned_le16(&event_map[i]);
-
 			fallthrough;
 
 		case IQS62X_EVENT_UI_HI:
@@ -490,7 +488,6 @@ static irqreturn_t iqs62x_irq(int irq, void *context)
 
 		case IQS62X_EVENT_HYST:
 			event_map[i] <<= iqs62x->dev_desc->hyst_shift;
-
 			fallthrough;
 
 		case IQS62X_EVENT_WHEEL:

--
2.7.4

Applied
#2 lore raw

Re: [PATCH v2 1/6] mfd: iqs62x: Remove superfluous whitespace above fallthroughs

Quoted 10 lines · Jeff LaBundy

Previously, all instances of the /* fall through */ comment were preceded by a newline to improve readability.

Now that /* fall through */ comments have been replaced with the fallthrough pseudo-keyword, the leftover whitespace looks out of place and can simply be removed.

Fixes df561f6688fe ("treewide: Use fallthrough pseudo-keyword")
Quoted 6 lines

---
Changes in v2:
- Added Acked-for-MFD-by trailer

 drivers/mfd/iqs62x.c | 3 ---
 1 file changed, 3 deletions(-)

Applied, thanks.

Signature · 4 lines
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog
#3 lore raw

PATCH v2 2/6 mfd: iqs62x: Remove unused bit mask

In reply to Jeff LaBundy

The register write that performed a mandatory soft reset during probe was removed during development of the driver, however the IQS62X_SYS_SETTINGS_SOFT_RESET bit mask was left behind. Remove it to keep stray macros out of the driver.

---
Changes in v2:
- Added Acked-for-MFD-by trailer

 drivers/mfd/iqs62x.c | 1 -
 1 file changed, 1 deletion(-)
drivers/mfd/iqs62x.c +0 -1
index ec4c790..ff968dc 100644
@@ -57,7 +57,6 @@
 #define IQS620_TEMP_CAL_OFFS			0xC4
 
 #define IQS62X_SYS_SETTINGS			0xD0
-#define IQS62X_SYS_SETTINGS_SOFT_RESET		BIT(7)
 #define IQS62X_SYS_SETTINGS_ACK_RESET		BIT(6)
 #define IQS62X_SYS_SETTINGS_EVENT_MODE		BIT(5)
 #define IQS62X_SYS_SETTINGS_CLK_DIV		BIT(4)

--
2.7.4

Applied
#4 lore raw

Re: [PATCH v2 2/6] mfd: iqs62x: Remove unused bit mask

Quoted 5 lines · Jeff LaBundy

The register write that performed a mandatory soft reset during probe was removed during development of the driver, however the IQS62X_SYS_SETTINGS_SOFT_RESET bit mask was left behind. Remove it to keep stray macros out of the driver.

Quoted 8 lines

---
Changes in v2:
- Added Acked-for-MFD-by trailer

 drivers/mfd/iqs62x.c | 1 -
 1 file changed, 1 deletion(-)

Applied, thanks.

Signature · 4 lines
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog
#5 lore raw

PATCH v2 3/6 mfd: iqs62x: Rename regmap_config struct

In reply to Jeff LaBundy

The regmap member of the driver's private data is called 'regmap', but the regmap_config struct is called 'iqs62x_map_config'. Rename the latter to 'iqs62x_regmap_config' for consistency.

---
Changes in v2:
- Added Acked-for-MFD-by trailer

 drivers/mfd/iqs62x.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/mfd/iqs62x.c +2 -2
index ff968dc..7a1ff7c 100644
@@ -866,7 +866,7 @@ static const struct iqs62x_dev_desc iqs62x_devs[] = {
 	},
 };
 
-static const struct regmap_config iqs62x_map_config = {
+static const struct regmap_config iqs62x_regmap_config = {
 	.reg_bits = 8,
 	.val_bits = 8,
 	.max_register = IQS62X_MAX_REG,
@@ -892,7 +892,7 @@ static int iqs62x_probe(struct i2c_client *client)
 	INIT_LIST_HEAD(&iqs62x->fw_blk_head);
 	init_completion(&iqs62x->fw_done);
 
-	iqs62x->regmap = devm_regmap_init_i2c(client, &iqs62x_map_config);
+	iqs62x->regmap = devm_regmap_init_i2c(client, &iqs62x_regmap_config);
 	if (IS_ERR(iqs62x->regmap)) {
 		ret = PTR_ERR(iqs62x->regmap);
 		dev_err(&client->dev, "Failed to initialize register map: %d\n",

--
2.7.4

Applied
#6 lore raw

Re: [PATCH v2 3/6] mfd: iqs62x: Rename regmap_config struct

Quoted 4 lines · Jeff LaBundy

The regmap member of the driver's private data is called 'regmap', but the regmap_config struct is called 'iqs62x_map_config'. Rename the latter to 'iqs62x_regmap_config' for consistency.

Quoted 8 lines

---
Changes in v2:
- Added Acked-for-MFD-by trailer

 drivers/mfd/iqs62x.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Applied, thanks.

Signature · 4 lines
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog
#7 lore raw

PATCH v2 4/6 mfd: iqs62x: Increase interrupt handler return delay

In reply to Jeff LaBundy

The time the device takes to deassert its RDY output following an I2C stop condition scales with the core clock frequency.

To prevent level-triggered interrupts from being reasserted after the interrupt handler returns, increase the time before returning to account for the worst-case delay (~90 us) plus margin.

---
Changes in v2:
- Added Acked-for-MFD-by trailer

 drivers/mfd/iqs62x.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mfd/iqs62x.c +1 -1
index 7a1ff7c..07c9725 100644
@@ -533,7 +533,7 @@ static irqreturn_t iqs62x_irq(int irq, void *context)
 	 * ensure the device's RDY output has been deasserted by the time the
 	 * interrupt handler returns.
 	 */
-	usleep_range(50, 100);
+	usleep_range(150, 200);
 
 	return IRQ_HANDLED;
 }

--
2.7.4

Applied
#8 lore raw

Re: [PATCH v2 4/6] mfd: iqs62x: Increase interrupt handler return delay

Quoted 7 lines · Jeff LaBundy

The time the device takes to deassert its RDY output following an I2C stop condition scales with the core clock frequency.

To prevent level-triggered interrupts from being reasserted after the interrupt handler returns, increase the time before returning to account for the worst-case delay (~90 us) plus margin.

Quoted 8 lines

---
Changes in v2:
- Added Acked-for-MFD-by trailer

 drivers/mfd/iqs62x.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Applied, thanks.

Signature · 4 lines
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog
#9 lore raw

PATCH v2 5/6 mfd: iqs62x: Do not poll during ATI

In reply to Jeff LaBundy

After loading firmware, the driver triggers ATI (calibration) with the newly loaded register configuration in place. Next, the driver polls a register field to ensure ATI completed in a timely fashion and that the device is ready to sense.

However, communicating with the device over I2C while ATI is under- way may induce noise in the device and cause ATI to fail. As such, the vendor recommends not to poll the device during ATI.

To solve this problem, let the device naturally signal to the host that ATI is complete by way of an interrupt. A completion prevents the sub-devices from being registered until this happens.

The former logic that scaled ATI timeout and filter settling delay is not carried forward with the new implementation, as it produces overly conservative delays at lower clock rates. Instead, a single pair of delays that covers all cases is used.

---
Changes in v2:
- Removed superfluous newlines throughout all iqs62x_dev_desc structs

 drivers/mfd/iqs62x.c       | 125 +++++++++++++++++++++++++--------------------
 include/linux/mfd/iqs62x.h |  11 ++--
 2 files changed, 73 insertions(+), 63 deletions(-)
drivers/mfd/iqs62x.c +69 -56
index 07c9725..9b5c389 100644
@@ -36,7 +36,6 @@
 #define IQS62X_PROD_NUM				0x00
 
 #define IQS62X_SYS_FLAGS			0x10
-#define IQS62X_SYS_FLAGS_IN_ATI			BIT(2)
 
 #define IQS620_HALL_FLAGS			0x16
 #define IQS621_HALL_FLAGS			0x19
@@ -60,6 +59,7 @@
 #define IQS62X_SYS_SETTINGS_ACK_RESET		BIT(6)
 #define IQS62X_SYS_SETTINGS_EVENT_MODE		BIT(5)
 #define IQS62X_SYS_SETTINGS_CLK_DIV		BIT(4)
+#define IQS62X_SYS_SETTINGS_COMM_ATI		BIT(3)
 #define IQS62X_SYS_SETTINGS_REDO_ATI		BIT(1)
 
 #define IQS62X_PWR_SETTINGS			0xD2
@@ -81,9 +81,7 @@
 #define IQS62X_FW_REC_TYPE_MASK			3
 #define IQS62X_FW_REC_TYPE_DATA			4
 
-#define IQS62X_ATI_POLL_SLEEP_US		10000
-#define IQS62X_ATI_POLL_TIMEOUT_US		500000
-#define IQS62X_ATI_STABLE_DELAY_MS		150
+#define IQS62X_FILT_SETTLE_MS			250
 
 struct iqs62x_fw_rec {
 	u8 type;
@@ -111,7 +109,6 @@ static int iqs62x_dev_init(struct iqs62x_core *iqs62x)
 	struct iqs62x_fw_blk *fw_blk;
 	unsigned int val;
 	int ret;
-	u8 clk_div = 1;
 
 	list_for_each_entry(fw_blk, &iqs62x->fw_blk_head, list) {
 		if (fw_blk->mask)
@@ -181,28 +178,32 @@ static int iqs62x_dev_init(struct iqs62x_core *iqs62x)
 			return ret;
 	}
 
-	ret = regmap_read(iqs62x->regmap, IQS62X_SYS_SETTINGS, &val);
-	if (ret)
-		return ret;
-
-	if (val & IQS62X_SYS_SETTINGS_CLK_DIV)
-		clk_div = iqs62x->dev_desc->clk_div;
-
-	ret = regmap_write(iqs62x->regmap, IQS62X_SYS_SETTINGS, val |
-			   IQS62X_SYS_SETTINGS_ACK_RESET |
-			   IQS62X_SYS_SETTINGS_EVENT_MODE |
-			   IQS62X_SYS_SETTINGS_REDO_ATI);
-	if (ret)
-		return ret;
-
-	ret = regmap_read_poll_timeout(iqs62x->regmap, IQS62X_SYS_FLAGS, val,
-				       !(val & IQS62X_SYS_FLAGS_IN_ATI),
-				       IQS62X_ATI_POLL_SLEEP_US,
-				       IQS62X_ATI_POLL_TIMEOUT_US * clk_div);
+	/*
+	 * Place the device in streaming mode at first so as not to miss the
+	 * limited number of interrupts that would otherwise occur after ATI
+	 * completes. The device is subsequently placed in event mode by the
+	 * interrupt handler.
+	 *
+	 * In the meantime, mask interrupts during ATI to prevent the device
+	 * from soliciting I2C traffic until the noise-sensitive ATI process
+	 * is complete.
+	 */
+	ret = regmap_update_bits(iqs62x->regmap, IQS62X_SYS_SETTINGS,
+				 IQS62X_SYS_SETTINGS_ACK_RESET |
+				 IQS62X_SYS_SETTINGS_EVENT_MODE |
+				 IQS62X_SYS_SETTINGS_COMM_ATI |
+				 IQS62X_SYS_SETTINGS_REDO_ATI,
+				 IQS62X_SYS_SETTINGS_ACK_RESET |
+				 IQS62X_SYS_SETTINGS_REDO_ATI);
 	if (ret)
 		return ret;
 
-	msleep(IQS62X_ATI_STABLE_DELAY_MS * clk_div);
+	/*
+	 * The following delay gives the device time to deassert its RDY output
+	 * in case a communication window was open while the REDO_ATI field was
+	 * written. This prevents an interrupt from being serviced prematurely.
+	 */
+	usleep_range(5000, 5100);
 
 	return 0;
 }
@@ -433,6 +434,11 @@ const struct iqs62x_event_desc iqs62x_events[IQS62X_NUM_EVENTS] = {
 		.mask	= BIT(7),
 		.val	= BIT(7),
 	},
+	[IQS62X_EVENT_SYS_ATI] = {
+		.reg	= IQS62X_EVENT_SYS,
+		.mask	= BIT(2),
+		.val	= BIT(2),
+	},
 };
 EXPORT_SYMBOL_GPL(iqs62x_events);
 
@@ -521,12 +527,39 @@ static irqreturn_t iqs62x_irq(int irq, void *context)
 				"Failed to re-initialize device: %d\n", ret);
 			return IRQ_NONE;
 		}
+
+		iqs62x->event_cache |= BIT(IQS62X_EVENT_SYS_RESET);
+		reinit_completion(&iqs62x->ati_done);
+	} else if (event_flags & BIT(IQS62X_EVENT_SYS_ATI)) {
+		iqs62x->event_cache |= BIT(IQS62X_EVENT_SYS_ATI);
+		reinit_completion(&iqs62x->ati_done);
+	} else if (!completion_done(&iqs62x->ati_done)) {
+		ret = regmap_update_bits(iqs62x->regmap, IQS62X_SYS_SETTINGS,
+					 IQS62X_SYS_SETTINGS_EVENT_MODE, 0xFF);
+		if (ret) {
+			dev_err(&client->dev,
+				"Failed to enable event mode: %d\n", ret);
+			return IRQ_NONE;
+		}
+
+		msleep(IQS62X_FILT_SETTLE_MS);
+		complete_all(&iqs62x->ati_done);
 	}
 
-	ret = blocking_notifier_call_chain(&iqs62x->nh, event_flags,
-					   &event_data);
-	if (ret & NOTIFY_STOP_MASK)
-		return IRQ_NONE;
+	/*
+	 * Reset and ATI events are not broadcast to the sub-device drivers
+	 * until ATI has completed. Any other events that may have occurred
+	 * during ATI are ignored.
+	 */
+	if (completion_done(&iqs62x->ati_done)) {
+		event_flags |= iqs62x->event_cache;
+		ret = blocking_notifier_call_chain(&iqs62x->nh, event_flags,
+						   &event_data);
+		if (ret & NOTIFY_STOP_MASK)
+			return IRQ_NONE;
+
+		iqs62x->event_cache = 0;
+	}
 
 	/*
 	 * Once the communication window is closed, a small delay is added to
@@ -567,6 +600,12 @@ static void iqs62x_firmware_load(const struct firmware *fw, void *context)
 		goto err_out;
 	}
 
+	if (!wait_for_completion_timeout(&iqs62x->ati_done,
+					 msecs_to_jiffies(2000))) {
+		dev_err(&client->dev, "Failed to complete ATI\n");
+		goto err_out;
+	}
+
 	ret = devm_mfd_add_devices(&client->dev, PLATFORM_DEVID_NONE,
 				   iqs62x->dev_desc->sub_devs,
 				   iqs62x->dev_desc->num_sub_devs,
@@ -748,22 +787,17 @@ static const struct iqs62x_dev_desc iqs62x_devs[] = {
 		.dev_name	= "iqs620at",
 		.sub_devs	= iqs620at_sub_devs,
 		.num_sub_devs	= ARRAY_SIZE(iqs620at_sub_devs),
-
 		.prod_num	= IQS620_PROD_NUM,
 		.sw_num		= 0x08,
 		.cal_regs	= iqs620at_cal_regs,
 		.num_cal_regs	= ARRAY_SIZE(iqs620at_cal_regs),
-
 		.prox_mask	= BIT(0),
 		.sar_mask	= BIT(1) | BIT(7),
 		.hall_mask	= BIT(2),
 		.hyst_mask	= BIT(3),
 		.temp_mask	= BIT(4),
-
 		.prox_settings	= IQS620_PROX_SETTINGS_4,
 		.hall_flags	= IQS620_HALL_FLAGS,
-
-		.clk_div	= 4,
 		.fw_name	= "iqs620a.bin",
 		.event_regs	= &iqs620a_event_regs[IQS62X_UI_PROX],
 	},
@@ -771,20 +805,15 @@ static const struct iqs62x_dev_desc iqs62x_devs[] = {
 		.dev_name	= "iqs620a",
 		.sub_devs	= iqs620a_sub_devs,
 		.num_sub_devs	= ARRAY_SIZE(iqs620a_sub_devs),
-
 		.prod_num	= IQS620_PROD_NUM,
 		.sw_num		= 0x08,
-
 		.prox_mask	= BIT(0),
 		.sar_mask	= BIT(1) | BIT(7),
 		.hall_mask	= BIT(2),
 		.hyst_mask	= BIT(3),
 		.temp_mask	= BIT(4),
-
 		.prox_settings	= IQS620_PROX_SETTINGS_4,
 		.hall_flags	= IQS620_HALL_FLAGS,
-
-		.clk_div	= 4,
 		.fw_name	= "iqs620a.bin",
 		.event_regs	= &iqs620a_event_regs[IQS62X_UI_PROX],
 	},
@@ -792,23 +821,18 @@ static const struct iqs62x_dev_desc iqs62x_devs[] = {
 		.dev_name	= "iqs621",
 		.sub_devs	= iqs621_sub_devs,
 		.num_sub_devs	= ARRAY_SIZE(iqs621_sub_devs),
-
 		.prod_num	= IQS621_PROD_NUM,
 		.sw_num		= 0x09,
 		.cal_regs	= iqs621_cal_regs,
 		.num_cal_regs	= ARRAY_SIZE(iqs621_cal_regs),
-
 		.prox_mask	= BIT(0),
 		.hall_mask	= BIT(1),
 		.als_mask	= BIT(2),
 		.hyst_mask	= BIT(3),
 		.temp_mask	= BIT(4),
-
 		.als_flags	= IQS621_ALS_FLAGS,
 		.hall_flags	= IQS621_HALL_FLAGS,
 		.hyst_shift	= 5,
-
-		.clk_div	= 2,
 		.fw_name	= "iqs621.bin",
 		.event_regs	= &iqs621_event_regs[IQS62X_UI_PROX],
 	},
@@ -816,21 +840,16 @@ static const struct iqs62x_dev_desc iqs62x_devs[] = {
 		.dev_name	= "iqs622",
 		.sub_devs	= iqs622_sub_devs,
 		.num_sub_devs	= ARRAY_SIZE(iqs622_sub_devs),
-
 		.prod_num	= IQS622_PROD_NUM,
 		.sw_num		= 0x06,
-
 		.prox_mask	= BIT(0),
 		.sar_mask	= BIT(1),
 		.hall_mask	= BIT(2),
 		.als_mask	= BIT(3),
 		.ir_mask	= BIT(4),
-
 		.prox_settings	= IQS622_PROX_SETTINGS_4,
 		.als_flags	= IQS622_ALS_FLAGS,
 		.hall_flags	= IQS622_HALL_FLAGS,
-
-		.clk_div	= 2,
 		.fw_name	= "iqs622.bin",
 		.event_regs	= &iqs622_event_regs[IQS62X_UI_PROX],
 	},
@@ -838,14 +857,10 @@ static const struct iqs62x_dev_desc iqs62x_devs[] = {
 		.dev_name	= "iqs624",
 		.sub_devs	= iqs624_sub_devs,
 		.num_sub_devs	= ARRAY_SIZE(iqs624_sub_devs),
-
 		.prod_num	= IQS624_PROD_NUM,
 		.sw_num		= 0x0B,
-
 		.interval	= IQS624_INTERVAL_NUM,
 		.interval_div	= 3,
-
-		.clk_div	= 2,
 		.fw_name	= "iqs624.bin",
 		.event_regs	= &iqs624_event_regs[IQS62X_UI_PROX],
 	},
@@ -853,14 +868,10 @@ static const struct iqs62x_dev_desc iqs62x_devs[] = {
 		.dev_name	= "iqs625",
 		.sub_devs	= iqs625_sub_devs,
 		.num_sub_devs	= ARRAY_SIZE(iqs625_sub_devs),
-
 		.prod_num	= IQS625_PROD_NUM,
 		.sw_num		= 0x0B,
-
 		.interval	= IQS625_INTERVAL_NUM,
 		.interval_div	= 10,
-
-		.clk_div	= 2,
 		.fw_name	= "iqs625.bin",
 		.event_regs	= &iqs625_event_regs[IQS62X_UI_PROX],
 	},
@@ -890,6 +901,8 @@ static int iqs62x_probe(struct i2c_client *client)
 
 	BLOCKING_INIT_NOTIFIER_HEAD(&iqs62x->nh);
 	INIT_LIST_HEAD(&iqs62x->fw_blk_head);
+
+	init_completion(&iqs62x->ati_done);
 	init_completion(&iqs62x->fw_done);
 
 	iqs62x->regmap = devm_regmap_init_i2c(client, &iqs62x_regmap_config);
include/linux/mfd/iqs62x.h +4 -7
index 043d3b6..5ced55e 100644
@@ -28,7 +28,7 @@
 #define IQS620_GLBL_EVENT_MASK_PMU		BIT(6)
 
 #define IQS62X_NUM_KEYS				16
-#define IQS62X_NUM_EVENTS			(IQS62X_NUM_KEYS + 5)
+#define IQS62X_NUM_EVENTS			(IQS62X_NUM_KEYS + 6)
 
 #define IQS62X_EVENT_SIZE			10
 
@@ -78,6 +78,7 @@ enum iqs62x_event_flag {
 
 	/* everything else */
 	IQS62X_EVENT_SYS_RESET,
+	IQS62X_EVENT_SYS_ATI,
 };
 
 struct iqs62x_event_data {
@@ -97,12 +98,10 @@ struct iqs62x_dev_desc {
 	const char *dev_name;
 	const struct mfd_cell *sub_devs;
 	int num_sub_devs;
-
 	u8 prod_num;
 	u8 sw_num;
 	const u8 *cal_regs;
 	int num_cal_regs;
-
 	u8 prox_mask;
 	u8 sar_mask;
 	u8 hall_mask;
@@ -110,16 +109,12 @@ struct iqs62x_dev_desc {
 	u8 temp_mask;
 	u8 als_mask;
 	u8 ir_mask;
-
 	u8 prox_settings;
 	u8 als_flags;
 	u8 hall_flags;
 	u8 hyst_shift;
-
 	u8 interval;
 	u8 interval_div;
-
-	u8 clk_div;
 	const char *fw_name;
 	const enum iqs62x_event_reg (*event_regs)[IQS62X_EVENT_SIZE];
 };
@@ -130,8 +125,10 @@ struct iqs62x_core {
 	struct regmap *regmap;
 	struct blocking_notifier_head nh;
 	struct list_head fw_blk_head;
+	struct completion ati_done;
 	struct completion fw_done;
 	enum iqs62x_ui_sel ui_sel;
+	unsigned long event_cache;
 };
 
 extern const struct iqs62x_event_desc iqs62x_events[IQS62X_NUM_EVENTS];

--
2.7.4

Applied
#10 lore raw

Re: [PATCH v2 5/6] mfd: iqs62x: Do not poll during ATI

Quoted 18 lines · Jeff LaBundy

After loading firmware, the driver triggers ATI (calibration) with the newly loaded register configuration in place. Next, the driver polls a register field to ensure ATI completed in a timely fashion and that the device is ready to sense.

However, communicating with the device over I2C while ATI is under- way may induce noise in the device and cause ATI to fail. As such, the vendor recommends not to poll the device during ATI.

To solve this problem, let the device naturally signal to the host that ATI is complete by way of an interrupt. A completion prevents the sub-devices from being registered until this happens.

The former logic that scaled ATI timeout and filter settling delay is not carried forward with the new implementation, as it produces overly conservative delays at lower clock rates. Instead, a single pair of delays that covers all cases is used.

Quoted 8 lines

---
Changes in v2:
- Removed superfluous newlines throughout all iqs62x_dev_desc structs

 drivers/mfd/iqs62x.c       | 125 +++++++++++++++++++++++++--------------------
 include/linux/mfd/iqs62x.h |  11 ++--
 2 files changed, 73 insertions(+), 63 deletions(-)

Applied, thanks.

Signature · 4 lines
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog
#11 lore raw

PATCH v2 6/6 mfd: iqs62x: Do not change clock frequency during ATI

In reply to Jeff LaBundy

After a reset event, the device automatically triggers ATI at the default core clock frequency (16 MHz). Soon afterward, the driver loads firmware which may attempt to lower the frequency.

If this initial ATI cycle is still in progress when the frequency is changed, however, the device incorrectly reports channels 0, 1 and 2 to be in a state of touch once ATI finally completes.

To solve this problem, wait until ATI is complete before lowering the frequency. Because this particular ATI cycle occurs following a reset event, its duration is predictable and a simple delay can suffice.

---
Changes in v2:
- Added Acked-for-MFD-by trailer

 drivers/mfd/iqs62x.c | 9 +++++++++
 1 file changed, 9 insertions(+)
drivers/mfd/iqs62x.c +9 -0
index 9b5c389..d1fc38a 100644
@@ -81,6 +81,7 @@
 #define IQS62X_FW_REC_TYPE_MASK			3
 #define IQS62X_FW_REC_TYPE_DATA			4
 
+#define IQS62X_ATI_STARTUP_MS			350
 #define IQS62X_FILT_SETTLE_MS			250
 
 struct iqs62x_fw_rec {
@@ -111,6 +112,14 @@ static int iqs62x_dev_init(struct iqs62x_core *iqs62x)
 	int ret;
 
 	list_for_each_entry(fw_blk, &iqs62x->fw_blk_head, list) {
+		/*
+		 * In case ATI is in progress, wait for it to complete before
+		 * lowering the core clock frequency.
+		 */
+		if (fw_blk->addr == IQS62X_SYS_SETTINGS &&
+		    *fw_blk->data & IQS62X_SYS_SETTINGS_CLK_DIV)
+			msleep(IQS62X_ATI_STARTUP_MS);
+
 		if (fw_blk->mask)
 			ret = regmap_update_bits(iqs62x->regmap, fw_blk->addr,
 						 fw_blk->mask, *fw_blk->data);

--
2.7.4

Applied
#12 lore raw

Re: [PATCH v2 6/6] mfd: iqs62x: Do not change clock frequency during ATI

Quoted 13 lines · Jeff LaBundy

After a reset event, the device automatically triggers ATI at the default core clock frequency (16 MHz). Soon afterward, the driver loads firmware which may attempt to lower the frequency.

If this initial ATI cycle is still in progress when the frequency is changed, however, the device incorrectly reports channels 0, 1 and 2 to be in a state of touch once ATI finally completes.

To solve this problem, wait until ATI is complete before lowering the frequency. Because this particular ATI cycle occurs following a reset event, its duration is predictable and a simple delay can suffice.

Quoted 8 lines

---
Changes in v2:
- Added Acked-for-MFD-by trailer

 drivers/mfd/iqs62x.c | 9 +++++++++
 1 file changed, 9 insertions(+)

Applied, thanks.

Signature · 4 lines
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog