Skip to contentSkip to conversation

regmap: regmap-irq/gpio-max77620: add level-irq support

PATCH In review Matti Vaittinen 5 participants 12 messages Dec 18, 2018 – Jan 4, 2019

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 regmap: regmap-irq/gpio-max77620: add level-irq support

Add level active IRQ support to regmap-irq irqchip. Change breaks existing regmap-irq type setting. Convert the existing drivers which use regmap-irq with trigger type setting (gpio-max77620) to work with this new approach. So we do not magically support level-active IRQs on gpio-max77620 - but add support to the regmap-irq for chips which support them =)

We do not support distinguishing situation where HW supports rising and falling edge detection but not both. Separating this would require inventing yet another flags for IRQ types.

---

Version 3 of this patch is intended to be functionally identical to v2. This patch is rebased on top of a tree which contains changes:
"regmap: irq: handle HW using separate rising/falling edge interrupts" from Bartosz Golaszewski and the change
"regmap: regmap-irq: Remove default irq type setting from core" (proposed here):
https://lore.kernel.org/lkml/[email protected]/

There should not be direct dependency to "regmap: regmap-irq: Remove default irq type setting from core" though. Patch was also tested to apply cleany on regmap-tree.

Same statement regarding testing applies - gpio-max77620 are only tested to compile. All real testing would be _HIGHLY_ appreciated.

 drivers/base/regmap/regmap-irq.c | 35 ++++++++++-----
 drivers/gpio/gpio-max77620.c     | 96 ++++++++++++++++++++++++++--------------
 include/linux/regmap.h           | 27 ++++++++---
 3 files changed, 110 insertions(+), 48 deletions(-)
drivers/base/regmap/regmap-irq.c +25 -10
index 8b216b2e2c19..31d23c9a5ae7 100644
@@ -199,7 +199,7 @@ static void regmap_irq_enable(struct irq_data *data)
 	const struct regmap_irq *irq_data = irq_to_regmap_irq(d, data->hwirq);
 	unsigned int mask, type;
 
-	type = irq_data->type_falling_mask | irq_data->type_rising_mask;
+	type = irq_data->type.type_falling_val | irq_data->type.type_rising_val;
 
 	/*
 	 * The type_in_mask flag means that the underlying hardware uses
@@ -234,27 +234,42 @@ static int regmap_irq_set_type(struct irq_data *data, unsigned int type)
 	struct regmap_irq_chip_data *d = irq_data_get_irq_chip_data(data);
 	struct regmap *map = d->map;
 	const struct regmap_irq *irq_data = irq_to_regmap_irq(d, data->hwirq);
-	int reg = irq_data->type_reg_offset / map->reg_stride;
+	int reg;
+	const struct regmap_irq_type *t = &irq_data->type;
 
-	if (!(irq_data->type_rising_mask | irq_data->type_falling_mask))
-		return 0;
+	if ((t->types_supported & type) != type)
+		return -ENOTSUPP;
+
+	reg = t->type_reg_offset / map->reg_stride;
 
-	d->type_buf[reg] &= ~(irq_data->type_falling_mask |
-					irq_data->type_rising_mask);
+	if (t->type_reg_mask)
+		d->type_buf[reg] &= ~t->type_reg_mask;
+	else
+		d->type_buf[reg] &= ~(t->type_falling_val |
+				      t->type_rising_val |
+				      t->type_level_low_val |
+				      t->type_level_high_val);
 	switch (type) {
 	case IRQ_TYPE_EDGE_FALLING:
-		d->type_buf[reg] |= irq_data->type_falling_mask;
+		d->type_buf[reg] |= t->type_falling_val;
 		break;
 
 	case IRQ_TYPE_EDGE_RISING:
-		d->type_buf[reg] |= irq_data->type_rising_mask;
+		d->type_buf[reg] |= t->type_rising_val;
 		break;
 
 	case IRQ_TYPE_EDGE_BOTH:
-		d->type_buf[reg] |= (irq_data->type_falling_mask |
-					irq_data->type_rising_mask);
+		d->type_buf[reg] |= (t->type_falling_val |
+					t->type_rising_val);
 		break;
 
+	case IRQ_TYPE_LEVEL_HIGH:
+		d->type_buf[reg] |= t->type_level_high_val;
+		break;
+
+	case IRQ_TYPE_LEVEL_LOW:
+		d->type_buf[reg] |= t->type_level_low_val;
+		break;
 	default:
 		return -EINVAL;
 	}
drivers/gpio/gpio-max77620.c +64 -32

124 lines not on this page · raw message

include/linux/regmap.h +21 -6
index b7aa50cfb306..a904f87151e8 100644
@@ -1089,22 +1089,37 @@ int regmap_fields_read(struct regmap_field *field, unsigned int id,
 int regmap_fields_update_bits_base(struct regmap_field *field,  unsigned int id,
 				   unsigned int mask, unsigned int val,
 				   bool *change, bool async, bool force);
+/**
+ * struct regmap_irq_type - IRQ type definitions.
+ *
+ * @type_reg_offset: Offset register for the irq type setting.
+ * @type_rising_val: Register value to configure RISING type irq.
+ * @type_falling_val: Register value to configure FALLING type irq.
+ * @type_level_low_val: Register value to configure LEVEL_LOW type irq.
+ * @type_level_high_val: Register value to configure LEVEL_HIGH type irq.
+ * @types_supported: logical OR of IRQ_TYPE_* flags indicating supported types.
+ */
+struct regmap_irq_type {
+	unsigned int type_reg_offset;
+	unsigned int type_reg_mask;
+	unsigned int type_rising_val;
+	unsigned int type_falling_val;
+	unsigned int type_level_low_val;
+	unsigned int type_level_high_val;
+	unsigned int types_supported;
+};
 
 /**
  * struct regmap_irq - Description of an IRQ for the generic regmap irq_chip.
  *
  * @reg_offset: Offset of the status/mask register within the bank
  * @mask:       Mask used to flag/control the register.
- * @type_reg_offset: Offset register for the irq type setting.
- * @type_rising_mask: Mask bit to configure RISING type irq.
- * @type_falling_mask: Mask bit to configure FALLING type irq.
+ * @type:	IRQ trigger type setting details if supported.
  */
 struct regmap_irq {
 	unsigned int reg_offset;
 	unsigned int mask;
-	unsigned int type_reg_offset;
-	unsigned int type_rising_mask;
-	unsigned int type_falling_mask;
+	struct regmap_irq_type type;
 };
 
 #define REGMAP_IRQ_REG(_irq, _off, _mask)		\
Signature · 8 lines
2.14.3


-- 
Matti Vaittinen
ROHM Semiconductors

~~~ "I don't think so," said Rene Descartes.  Just then, he vanished ~~~
#1 lore raw
Thank you for the patch! Yet something to improve:

Hi Matti,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on regmap/for-next]
[also build test ERROR on next-20181218]
[cannot apply to v4.20-rc7]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]

url:    https://github.com/0day-ci/linux/commits/Matti-Vaittinen/regmap-regmap-irq-gpio-max77620-add-level-irq-support/20181218-225844
base:   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap.git for-next
config: x86_64-randconfig-x006-201850 (attached as .config)
compiler: gcc-7 (Debian 7.3.0-1) 7.3.0
reproduce:
        # save the attached .config to linux build tree
        make ARCH=x86_64 

All errors (new ones prefixed by >>):

drivers/base/regmap/regmap-irq.c: In function 'regmap_add_irq_chip':

>> drivers/base/regmap/regmap-irq.c:644:24: error: 'const struct regmap_irq' has no member named 'type_reg_offset'; did you mean 'reg_offset'?
       reg = chip->irqs[i].type_reg_offset / map->reg_stride;
                           ^~~~~~~~~~~~~~~
                           reg_offset
>> drivers/base/regmap/regmap-irq.c:645:41: error: 'const struct regmap_irq' has no member named 'type_rising_mask'
       d->type_buf_def[reg] |= chip->irqs[i].type_rising_mask |
                                            ^
>> drivers/base/regmap/regmap-irq.c:646:19: error: 'const struct regmap_irq' has no member named 'type_falling_mask'
         chip->irqs[i].type_falling_mask;
                      ^

vim +644 drivers/base/regmap/regmap-irq.c

4af8be67f Mark Brown          2012-05-13  446  
f8beab2bb Mark Brown          2011-10-28  447  /**
2cf8e2dfd Charles Keepax      2017-01-12  448   * regmap_add_irq_chip() - Use standard regmap IRQ controller handling
f8beab2bb Mark Brown          2011-10-28  449   *
2cf8e2dfd Charles Keepax      2017-01-12  450   * @map: The regmap for the device.
2cf8e2dfd Charles Keepax      2017-01-12  451   * @irq: The IRQ the device uses to signal interrupts.
2cf8e2dfd Charles Keepax      2017-01-12  452   * @irq_flags: The IRQF_ flags to use for the primary interrupt.
2cf8e2dfd Charles Keepax      2017-01-12  453   * @irq_base: Allocate at specific IRQ number if irq_base > 0.
2cf8e2dfd Charles Keepax      2017-01-12  454   * @chip: Configuration for the interrupt controller.
2cf8e2dfd Charles Keepax      2017-01-12  455   * @data: Runtime data structure for the controller, allocated on success.
f8beab2bb Mark Brown          2011-10-28  456   *
f8beab2bb Mark Brown          2011-10-28  457   * Returns 0 on success or an errno on failure.
f8beab2bb Mark Brown          2011-10-28  458   *
f8beab2bb Mark Brown          2011-10-28  459   * In order for this to be efficient the chip really should use a
f8beab2bb Mark Brown          2011-10-28  460   * register cache.  The chip driver is responsible for restoring the
f8beab2bb Mark Brown          2011-10-28  461   * register values used by the IRQ controller over suspend and resume.
f8beab2bb Mark Brown          2011-10-28  462   */
f8beab2bb Mark Brown          2011-10-28  463  int regmap_add_irq_chip(struct regmap *map, int irq, int irq_flags,
b026ddbbd Mark Brown          2012-05-31  464  			int irq_base, const struct regmap_irq_chip *chip,
f8beab2bb Mark Brown          2011-10-28  465  			struct regmap_irq_chip_data **data)
f8beab2bb Mark Brown          2011-10-28  466  {
f8beab2bb Mark Brown          2011-10-28  467  	struct regmap_irq_chip_data *d;
4af8be67f Mark Brown          2012-05-13  468  	int i;
f8beab2bb Mark Brown          2011-10-28  469  	int ret = -ENOMEM;
bc998a730 Bartosz Golaszewski 2018-12-07  470  	int num_type_reg;
16032624f Stephen Warren      2012-07-27  471  	u32 reg;
7b7d1968e Guo Zeng            2015-09-17  472  	u32 unmask_offset;
f8beab2bb Mark Brown          2011-10-28  473  
e12892070 Xiubo Li            2014-05-19  474  	if (chip->num_regs <= 0)
e12892070 Xiubo Li            2014-05-19  475  		return -EINVAL;
e12892070 Xiubo Li            2014-05-19  476  
f01ee60ff Stephen Warren      2012-04-09  477  	for (i = 0; i < chip->num_irqs; i++) {
f01ee60ff Stephen Warren      2012-04-09  478  		if (chip->irqs[i].reg_offset % map->reg_stride)
f01ee60ff Stephen Warren      2012-04-09  479  			return -EINVAL;
f01ee60ff Stephen Warren      2012-04-09  480  		if (chip->irqs[i].reg_offset / map->reg_stride >=
f01ee60ff Stephen Warren      2012-04-09  481  		    chip->num_regs)
f01ee60ff Stephen Warren      2012-04-09  482  			return -EINVAL;
f01ee60ff Stephen Warren      2012-04-09  483  	}
f01ee60ff Stephen Warren      2012-04-09  484  
4af8be67f Mark Brown          2012-05-13  485  	if (irq_base) {
f8beab2bb Mark Brown          2011-10-28  486  		irq_base = irq_alloc_descs(irq_base, 0, chip->num_irqs, 0);
f8beab2bb Mark Brown          2011-10-28  487  		if (irq_base < 0) {
f8beab2bb Mark Brown          2011-10-28  488  			dev_warn(map->dev, "Failed to allocate IRQs: %d\n",
f8beab2bb Mark Brown          2011-10-28  489  				 irq_base);
f8beab2bb Mark Brown          2011-10-28  490  			return irq_base;
f8beab2bb Mark Brown          2011-10-28  491  		}
4af8be67f Mark Brown          2012-05-13  492  	}
f8beab2bb Mark Brown          2011-10-28  493  
f8beab2bb Mark Brown          2011-10-28  494  	d = kzalloc(sizeof(*d), GFP_KERNEL);
f8beab2bb Mark Brown          2011-10-28  495  	if (!d)
f8beab2bb Mark Brown          2011-10-28  496  		return -ENOMEM;
f8beab2bb Mark Brown          2011-10-28  497  
eeda1bd69 lixiubo             2015-11-20  498  	d->status_buf = kcalloc(chip->num_regs, sizeof(unsigned int),
f8beab2bb Mark Brown          2011-10-28  499  				GFP_KERNEL);
f8beab2bb Mark Brown          2011-10-28  500  	if (!d->status_buf)
f8beab2bb Mark Brown          2011-10-28  501  		goto err_alloc;
f8beab2bb Mark Brown          2011-10-28  502  
eeda1bd69 lixiubo             2015-11-20  503  	d->mask_buf = kcalloc(chip->num_regs, sizeof(unsigned int),
f8beab2bb Mark Brown          2011-10-28  504  			      GFP_KERNEL);
f8beab2bb Mark Brown          2011-10-28  505  	if (!d->mask_buf)
f8beab2bb Mark Brown          2011-10-28  506  		goto err_alloc;
f8beab2bb Mark Brown          2011-10-28  507  
eeda1bd69 lixiubo             2015-11-20  508  	d->mask_buf_def = kcalloc(chip->num_regs, sizeof(unsigned int),
f8beab2bb Mark Brown          2011-10-28  509  				  GFP_KERNEL);
f8beab2bb Mark Brown          2011-10-28  510  	if (!d->mask_buf_def)
f8beab2bb Mark Brown          2011-10-28  511  		goto err_alloc;
f8beab2bb Mark Brown          2011-10-28  512  
a43fd50dc Mark Brown          2012-06-05  513  	if (chip->wake_base) {
eeda1bd69 lixiubo             2015-11-20  514  		d->wake_buf = kcalloc(chip->num_regs, sizeof(unsigned int),
a43fd50dc Mark Brown          2012-06-05  515  				      GFP_KERNEL);
a43fd50dc Mark Brown          2012-06-05  516  		if (!d->wake_buf)
a43fd50dc Mark Brown          2012-06-05  517  			goto err_alloc;
a43fd50dc Mark Brown          2012-06-05  518  	}
a43fd50dc Mark Brown          2012-06-05  519  
bc998a730 Bartosz Golaszewski 2018-12-07  520  	num_type_reg = chip->type_in_mask ? chip->num_regs : chip->num_type_reg;
bc998a730 Bartosz Golaszewski 2018-12-07  521  	if (num_type_reg) {
bc998a730 Bartosz Golaszewski 2018-12-07  522  		d->type_buf_def = kcalloc(num_type_reg,
7a78479fd Laxman Dewangan     2015-12-22  523  					  sizeof(unsigned int), GFP_KERNEL);
7a78479fd Laxman Dewangan     2015-12-22  524  		if (!d->type_buf_def)
7a78479fd Laxman Dewangan     2015-12-22  525  			goto err_alloc;
7a78479fd Laxman Dewangan     2015-12-22  526  
bc998a730 Bartosz Golaszewski 2018-12-07  527  		d->type_buf = kcalloc(num_type_reg, sizeof(unsigned int),
7a78479fd Laxman Dewangan     2015-12-22  528  				      GFP_KERNEL);
7a78479fd Laxman Dewangan     2015-12-22  529  		if (!d->type_buf)
7a78479fd Laxman Dewangan     2015-12-22  530  			goto err_alloc;
7a78479fd Laxman Dewangan     2015-12-22  531  	}
7a78479fd Laxman Dewangan     2015-12-22  532  
7ac140ec4 Stephen Warren      2012-08-01  533  	d->irq_chip = regmap_irq_chip;
ca142750f Stephen Warren      2012-08-01  534  	d->irq_chip.name = chip->name;
a43fd50dc Mark Brown          2012-06-05  535  	d->irq = irq;
f8beab2bb Mark Brown          2011-10-28  536  	d->map = map;
f8beab2bb Mark Brown          2011-10-28  537  	d->chip = chip;
f8beab2bb Mark Brown          2011-10-28  538  	d->irq_base = irq_base;
022f926a2 Graeme Gregory      2012-05-14  539  
022f926a2 Graeme Gregory      2012-05-14  540  	if (chip->irq_reg_stride)
022f926a2 Graeme Gregory      2012-05-14  541  		d->irq_reg_stride = chip->irq_reg_stride;
022f926a2 Graeme Gregory      2012-05-14  542  	else
022f926a2 Graeme Gregory      2012-05-14  543  		d->irq_reg_stride = 1;
022f926a2 Graeme Gregory      2012-05-14  544  
7a78479fd Laxman Dewangan     2015-12-22  545  	if (chip->type_reg_stride)
7a78479fd Laxman Dewangan     2015-12-22  546  		d->type_reg_stride = chip->type_reg_stride;
7a78479fd Laxman Dewangan     2015-12-22  547  	else
7a78479fd Laxman Dewangan     2015-12-22  548  		d->type_reg_stride = 1;
7a78479fd Laxman Dewangan     2015-12-22  549  
67921a1a6 Markus Pargmann     2015-08-21  550  	if (!map->use_single_read && map->reg_stride == 1 &&
a7440eaa9 Mark Brown          2013-01-03  551  	    d->irq_reg_stride == 1) {
549e08a0a lixiubo             2015-11-20  552  		d->status_reg_buf = kmalloc_array(chip->num_regs,
549e08a0a lixiubo             2015-11-20  553  						  map->format.val_bytes,
549e08a0a lixiubo             2015-11-20  554  						  GFP_KERNEL);
a7440eaa9 Mark Brown          2013-01-03  555  		if (!d->status_reg_buf)
a7440eaa9 Mark Brown          2013-01-03  556  			goto err_alloc;
a7440eaa9 Mark Brown          2013-01-03  557  	}
a7440eaa9 Mark Brown          2013-01-03  558  
f8beab2bb Mark Brown          2011-10-28  559  	mutex_init(&d->lock);
f8beab2bb Mark Brown          2011-10-28  560  
f8beab2bb Mark Brown          2011-10-28  561  	for (i = 0; i < chip->num_irqs; i++)
f01ee60ff Stephen Warren      2012-04-09  562  		d->mask_buf_def[chip->irqs[i].reg_offset / map->reg_stride]
f8beab2bb Mark Brown          2011-10-28  563  			|= chip->irqs[i].mask;
f8beab2bb Mark Brown          2011-10-28  564  
f8beab2bb Mark Brown          2011-10-28  565  	/* Mask all the interrupts by default */
f8beab2bb Mark Brown          2011-10-28  566  	for (i = 0; i < chip->num_regs; i++) {
f8beab2bb Mark Brown          2011-10-28  567  		d->mask_buf[i] = d->mask_buf_def[i];
16032624f Stephen Warren      2012-07-27  568  		reg = chip->mask_base +
16032624f Stephen Warren      2012-07-27  569  			(i * map->reg_stride * d->irq_reg_stride);
36ac914ba Xiaofan Tian        2012-08-30  570  		if (chip->mask_invert)
a71411dbf Michael Grzeschik   2017-06-23  571  			ret = regmap_irq_update_bits(d, reg,
36ac914ba Xiaofan Tian        2012-08-30  572  					 d->mask_buf[i], ~d->mask_buf[i]);
7b7d1968e Guo Zeng            2015-09-17  573  		else if (d->chip->unmask_base) {
7b7d1968e Guo Zeng            2015-09-17  574  			unmask_offset = d->chip->unmask_base -
7b7d1968e Guo Zeng            2015-09-17  575  					d->chip->mask_base;
a71411dbf Michael Grzeschik   2017-06-23  576  			ret = regmap_irq_update_bits(d,
7b7d1968e Guo Zeng            2015-09-17  577  					reg + unmask_offset,
7b7d1968e Guo Zeng            2015-09-17  578  					d->mask_buf[i],
7b7d1968e Guo Zeng            2015-09-17  579  					d->mask_buf[i]);
7b7d1968e Guo Zeng            2015-09-17  580  		} else
a71411dbf Michael Grzeschik   2017-06-23  581  			ret = regmap_irq_update_bits(d, reg,
0eb46ad0c Mark Brown          2012-08-01  582  					 d->mask_buf[i], d->mask_buf[i]);
f8beab2bb Mark Brown          2011-10-28  583  		if (ret != 0) {
f8beab2bb Mark Brown          2011-10-28  584  			dev_err(map->dev, "Failed to set masks in 0x%x: %d\n",
16032624f Stephen Warren      2012-07-27  585  				reg, ret);
f8beab2bb Mark Brown          2011-10-28  586  			goto err_alloc;
f8beab2bb Mark Brown          2011-10-28  587  		}
2753e6f82 Philipp Zabel       2013-07-22  588  
2753e6f82 Philipp Zabel       2013-07-22  589  		if (!chip->init_ack_masked)
2753e6f82 Philipp Zabel       2013-07-22  590  			continue;
2753e6f82 Philipp Zabel       2013-07-22  591  
2753e6f82 Philipp Zabel       2013-07-22  592  		/* Ack masked but set interrupts */
2753e6f82 Philipp Zabel       2013-07-22  593  		reg = chip->status_base +
2753e6f82 Philipp Zabel       2013-07-22  594  			(i * map->reg_stride * d->irq_reg_stride);
2753e6f82 Philipp Zabel       2013-07-22  595  		ret = regmap_read(map, reg, &d->status_buf[i]);
2753e6f82 Philipp Zabel       2013-07-22  596  		if (ret != 0) {
2753e6f82 Philipp Zabel       2013-07-22  597  			dev_err(map->dev, "Failed to read IRQ status: %d\n",
2753e6f82 Philipp Zabel       2013-07-22  598  				ret);
2753e6f82 Philipp Zabel       2013-07-22  599  			goto err_alloc;
2753e6f82 Philipp Zabel       2013-07-22  600  		}
2753e6f82 Philipp Zabel       2013-07-22  601  
d32334333 Alexander Shiyan    2013-12-15  602  		if (d->status_buf[i] && (chip->ack_base || chip->use_ack)) {
2753e6f82 Philipp Zabel       2013-07-22  603  			reg = chip->ack_base +
2753e6f82 Philipp Zabel       2013-07-22  604  				(i * map->reg_stride * d->irq_reg_stride);
a650fdd94 Guo Zeng            2015-09-17  605  			if (chip->ack_invert)
a650fdd94 Guo Zeng            2015-09-17  606  				ret = regmap_write(map, reg,
a650fdd94 Guo Zeng            2015-09-17  607  					~(d->status_buf[i] & d->mask_buf[i]));
a650fdd94 Guo Zeng            2015-09-17  608  			else
2753e6f82 Philipp Zabel       2013-07-22  609  				ret = regmap_write(map, reg,
2753e6f82 Philipp Zabel       2013-07-22  610  					d->status_buf[i] & d->mask_buf[i]);
2753e6f82 Philipp Zabel       2013-07-22  611  			if (ret != 0) {
2753e6f82 Philipp Zabel       2013-07-22  612  				dev_err(map->dev, "Failed to ack 0x%x: %d\n",
2753e6f82 Philipp Zabel       2013-07-22  613  					reg, ret);
2753e6f82 Philipp Zabel       2013-07-22  614  				goto err_alloc;
2753e6f82 Philipp Zabel       2013-07-22  615  			}
2753e6f82 Philipp Zabel       2013-07-22  616  		}
f8beab2bb Mark Brown          2011-10-28  617  	}
f8beab2bb Mark Brown          2011-10-28  618  
40052ca0c Stephen Warren      2012-08-01  619  	/* Wake is disabled by default */
40052ca0c Stephen Warren      2012-08-01  620  	if (d->wake_buf) {
40052ca0c Stephen Warren      2012-08-01  621  		for (i = 0; i < chip->num_regs; i++) {
40052ca0c Stephen Warren      2012-08-01  622  			d->wake_buf[i] = d->mask_buf_def[i];
40052ca0c Stephen Warren      2012-08-01  623  			reg = chip->wake_base +
40052ca0c Stephen Warren      2012-08-01  624  				(i * map->reg_stride * d->irq_reg_stride);
9442490a0 Mark Brown          2013-01-04  625  
9442490a0 Mark Brown          2013-01-04  626  			if (chip->wake_invert)
a71411dbf Michael Grzeschik   2017-06-23  627  				ret = regmap_irq_update_bits(d, reg,
9442490a0 Mark Brown          2013-01-04  628  							 d->mask_buf_def[i],
9442490a0 Mark Brown          2013-01-04  629  							 0);
9442490a0 Mark Brown          2013-01-04  630  			else
a71411dbf Michael Grzeschik   2017-06-23  631  				ret = regmap_irq_update_bits(d, reg,
9442490a0 Mark Brown          2013-01-04  632  							 d->mask_buf_def[i],
40052ca0c Stephen Warren      2012-08-01  633  							 d->wake_buf[i]);
40052ca0c Stephen Warren      2012-08-01  634  			if (ret != 0) {
40052ca0c Stephen Warren      2012-08-01  635  				dev_err(map->dev, "Failed to set masks in 0x%x: %d\n",
40052ca0c Stephen Warren      2012-08-01  636  					reg, ret);
40052ca0c Stephen Warren      2012-08-01  637  				goto err_alloc;
40052ca0c Stephen Warren      2012-08-01  638  			}
40052ca0c Stephen Warren      2012-08-01  639  		}
40052ca0c Stephen Warren      2012-08-01  640  	}
40052ca0c Stephen Warren      2012-08-01  641  
bc998a730 Bartosz Golaszewski 2018-12-07  642  	if (chip->num_type_reg && !chip->type_in_mask) {
7a78479fd Laxman Dewangan     2015-12-22  643  		for (i = 0; i < chip->num_irqs; i++) {
7a78479fd Laxman Dewangan     2015-12-22 @644  			reg = chip->irqs[i].type_reg_offset / map->reg_stride;
7a78479fd Laxman Dewangan     2015-12-22 @645  			d->type_buf_def[reg] |= chip->irqs[i].type_rising_mask |
7a78479fd Laxman Dewangan     2015-12-22 @646  					chip->irqs[i].type_falling_mask;
7a78479fd Laxman Dewangan     2015-12-22  647  		}
7a78479fd Laxman Dewangan     2015-12-22  648  		for (i = 0; i < chip->num_type_reg; ++i) {
7a78479fd Laxman Dewangan     2015-12-22  649  			if (!d->type_buf_def[i])
7a78479fd Laxman Dewangan     2015-12-22  650  				continue;
7a78479fd Laxman Dewangan     2015-12-22  651  
7a78479fd Laxman Dewangan     2015-12-22  652  			reg = chip->type_base +
7a78479fd Laxman Dewangan     2015-12-22  653  				(i * map->reg_stride * d->type_reg_stride);
7a78479fd Laxman Dewangan     2015-12-22  654  			if (chip->type_invert)
a71411dbf Michael Grzeschik   2017-06-23  655  				ret = regmap_irq_update_bits(d, reg,
7a78479fd Laxman Dewangan     2015-12-22  656  					d->type_buf_def[i], 0xFF);
7a78479fd Laxman Dewangan     2015-12-22  657  			else
a71411dbf Michael Grzeschik   2017-06-23  658  				ret = regmap_irq_update_bits(d, reg,
7a78479fd Laxman Dewangan     2015-12-22  659  					d->type_buf_def[i], 0x0);
7a78479fd Laxman Dewangan     2015-12-22  660  			if (ret != 0) {
7a78479fd Laxman Dewangan     2015-12-22  661  				dev_err(map->dev,
7a78479fd Laxman Dewangan     2015-12-22  662  					"Failed to set type in 0x%x: %x\n",
7a78479fd Laxman Dewangan     2015-12-22  663  					reg, ret);
7a78479fd Laxman Dewangan     2015-12-22  664  				goto err_alloc;
7a78479fd Laxman Dewangan     2015-12-22  665  			}
7a78479fd Laxman Dewangan     2015-12-22  666  		}
7a78479fd Laxman Dewangan     2015-12-22  667  	}
7a78479fd Laxman Dewangan     2015-12-22  668  
4af8be67f Mark Brown          2012-05-13  669  	if (irq_base)
4af8be67f Mark Brown          2012-05-13  670  		d->domain = irq_domain_add_legacy(map->dev->of_node,
4af8be67f Mark Brown          2012-05-13  671  						  chip->num_irqs, irq_base, 0,
4af8be67f Mark Brown          2012-05-13  672  						  &regmap_domain_ops, d);
4af8be67f Mark Brown          2012-05-13  673  	else
4af8be67f Mark Brown          2012-05-13  674  		d->domain = irq_domain_add_linear(map->dev->of_node,
4af8be67f Mark Brown          2012-05-13  675  						  chip->num_irqs,
4af8be67f Mark Brown          2012-05-13  676  						  &regmap_domain_ops, d);
4af8be67f Mark Brown          2012-05-13  677  	if (!d->domain) {
4af8be67f Mark Brown          2012-05-13  678  		dev_err(map->dev, "Failed to create IRQ domain\n");
4af8be67f Mark Brown          2012-05-13  679  		ret = -ENOMEM;
4af8be67f Mark Brown          2012-05-13  680  		goto err_alloc;
f8beab2bb Mark Brown          2011-10-28  681  	}
f8beab2bb Mark Brown          2011-10-28  682  
09cadf6e0 Valentin Rothberg   2015-02-11  683  	ret = request_threaded_irq(irq, NULL, regmap_irq_thread,
09cadf6e0 Valentin Rothberg   2015-02-11  684  				   irq_flags | IRQF_ONESHOT,
f8beab2bb Mark Brown          2011-10-28  685  				   chip->name, d);
f8beab2bb Mark Brown          2011-10-28  686  	if (ret != 0) {
eed456f93 Mark Brown          2013-03-19  687  		dev_err(map->dev, "Failed to request IRQ %d for %s: %d\n",
eed456f93 Mark Brown          2013-03-19  688  			irq, chip->name, ret);
4af8be67f Mark Brown          2012-05-13  689  		goto err_domain;
f8beab2bb Mark Brown          2011-10-28  690  	}
f8beab2bb Mark Brown          2011-10-28  691  
72a6a5df2 Krzysztof Kozlowski 2014-03-13  692  	*data = d;
72a6a5df2 Krzysztof Kozlowski 2014-03-13  693  
f8beab2bb Mark Brown          2011-10-28  694  	return 0;
f8beab2bb Mark Brown          2011-10-28  695  
4af8be67f Mark Brown          2012-05-13  696  err_domain:
4af8be67f Mark Brown          2012-05-13  697  	/* Should really dispose of the domain but... */
f8beab2bb Mark Brown          2011-10-28  698  err_alloc:
7a78479fd Laxman Dewangan     2015-12-22  699  	kfree(d->type_buf);
7a78479fd Laxman Dewangan     2015-12-22  700  	kfree(d->type_buf_def);
a43fd50dc Mark Brown          2012-06-05  701  	kfree(d->wake_buf);
f8beab2bb Mark Brown          2011-10-28  702  	kfree(d->mask_buf_def);
f8beab2bb Mark Brown          2011-10-28  703  	kfree(d->mask_buf);
f8beab2bb Mark Brown          2011-10-28  704  	kfree(d->status_buf);
a7440eaa9 Mark Brown          2013-01-03  705  	kfree(d->status_reg_buf);
f8beab2bb Mark Brown          2011-10-28  706  	kfree(d);
f8beab2bb Mark Brown          2011-10-28  707  	return ret;
f8beab2bb Mark Brown          2011-10-28  708  }
f8beab2bb Mark Brown          2011-10-28  709  EXPORT_SYMBOL_GPL(regmap_add_irq_chip);
f8beab2bb Mark Brown          2011-10-28  710  

:::::: The code at line 644 was first introduced by commit
:::::: 7a78479fd2acd25db7ecd1744d76f6841ec8a257 regmap: irq: add support for configuration of trigger type

:::::: TO: Laxman Dewangan <[email protected]>
:::::: CC: Mark Brown <[email protected]>

---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation

#2 lore raw
Quoted 21 lines · kbuild test robot

Hi Matti,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on regmap/for-next]
[also build test ERROR on next-20181218]
[cannot apply to v4.20-rc7]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]

url:    https://github.com/0day-ci/linux/commits/Matti-Vaittinen/regmap-regmap-irq-gpio-max77620-add-level-irq-support/20181218-225844
base:   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap.git for-next
config: x86_64-randconfig-x006-201850 (attached as .config)
compiler: gcc-7 (Debian 7.3.0-1) 7.3.0
reproduce:
        # save the attached .config to linux build tree
        make ARCH=x86_64 

All errors (new ones prefixed by >>):

drivers/base/regmap/regmap-irq.c: In function 'regmap_add_irq_chip':

Quoted 1 line
Quoted 1 line

drivers/base/regmap/regmap-irq.c:644:24: error: 'const struct regmap_irq' has no member named 'type_reg_offset'; did you mean 'reg_offset'?

Quoted 8 lines
       reg = chip->irqs[i].type_reg_offset / map->reg_stride;
                           ^~~~~~~~~~~~~~~
                           reg_offset
Quoted 1 line
Quoted 1 line

drivers/base/regmap/regmap-irq.c:645:41: error: 'const struct regmap_irq' has no member named 'type_rising_mask'

       d->type_buf_def[reg] |= chip->irqs[i].type_rising_mask |
                                            ^
Quoted 1 line
Quoted 1 line

drivers/base/regmap/regmap-irq.c:646:19: error: 'const struct regmap_irq' has no member named 'type_falling_mask'

         chip->irqs[i].type_falling_mask;

---

Quoted 11 lines
Quoted 11 lines

Version 3 of this patch is intended to be functionally identical to v2. This patch is rebased on top of a tree which contains changes:
"regmap: irq: handle HW using separate rising/falling edge interrupts" from Bartosz Golaszewski and the change
"regmap: regmap-irq: Remove default irq type setting from core" (proposed here):
https://lore.kernel.org/lkml/[email protected]/

There should not be direct dependency to "regmap: regmap-irq: Remove default irq type setting from core" though. Patch was also tested to apply cleany on regmap-tree.

I forgot that the block of code the commit "regmap: regmap-irq: Remove default irq type setting from core" removes do use the old type specifiers whcih this patch changes. So even though this patch applies cleanly on tree which does not include "regmap: regmap-irq: Remove default irq type setting from core" - it does not mean there is no dependency. This was my brain fart. There is dependency. "regmap: regmap-irq: Remove default irq type setting from core" should be applied prior this patch.

Should I combine these two patches as a series (and resend them) or what is the correct way to note the dependency?

Br,
Matti Vaittinen

maintainer
#3 lore raw
Quoted 10 lines · Matti Vaittinen

I forgot that the block of code the commit "regmap: regmap-irq: Remove default irq type setting from core" removes do use the old type specifiers whcih this patch changes. So even though this patch applies cleanly on tree which does not include "regmap: regmap-irq: Remove default irq type setting from core" - it does not mean there is no dependency. This was my brain fart. There is dependency. "regmap: regmap-irq: Remove default irq type setting from core" should be applied prior this patch.

Should I combine these two patches as a series (and resend them) or what is the correct way to note the dependency?

It's fine as-is.

#4 lore raw
In reply to Matti Vaittinen

Hi Matti,

Quoted 2 lines · Matti Vaittinen

Add level active IRQ support to regmap-irq irqchip. Change breaks existing regmap-irq type setting. Convert the existing drivers which

Indeed it does.

Quoted 10 lines

use regmap-irq with trigger type setting (gpio-max77620) to work with this new approach. So we do not magically support level-active IRQs on gpio-max77620 - but add support to the regmap-irq for chips which support them =)

We do not support distinguishing situation where HW supports rising and falling edge detection but not both. Separating this would require inventing yet another flags for IRQ types.

This is now upstream as commit 1c2928e3e3212252 ("regmap:
regmap-irq/gpio-max77620: add level-irq support"), and breaks da9063-rtc on the Renesas Koelsch board:

genirq: Setting trigger mode 8 for irq 157 failed
(regmap_irq_set_type+0x0/0x140)

    da9063-rtc da9063-rtc: Failed to request ALARM IRQ 157: -524
    da9063-rtc: probe of da9063-rtc failed with error -524
Quoted 37 lines

---

Version 3 of this patch is intended to be functionally identical to v2. This patch is rebased on top of a tree which contains changes:
"regmap: irq: handle HW using separate rising/falling edge interrupts" from Bartosz Golaszewski and the change
"regmap: regmap-irq: Remove default irq type setting from core" (proposed here):
https://lore.kernel.org/lkml/[email protected]/

There should not be direct dependency to "regmap: regmap-irq: Remove default irq type setting from core" though. Patch was also tested to apply cleany on regmap-tree.

Same statement regarding testing applies - gpio-max77620 are only tested to compile. All real testing would be _HIGHLY_ appreciated.

 drivers/base/regmap/regmap-irq.c | 35 ++++++++++-----
 drivers/gpio/gpio-max77620.c     | 96 ++++++++++++++++++++++++++--------------
 include/linux/regmap.h           | 27 ++++++++---
 3 files changed, 110 insertions(+), 48 deletions(-)
drivers/base/regmap/regmap-irq.c +1 -1
index 8b216b2e2c19..31d23c9a5ae7 100644
@@ -199,7 +199,7 @@ static void regmap_irq_enable(struct irq_data *data)
        const struct regmap_irq *irq_data = irq_to_regmap_irq(d, data->hwirq);
        unsigned int mask, type;
 
-       type = irq_data->type_falling_mask | irq_data->type_rising_mask;
+       type = irq_data->type.type_falling_val | irq_data->type.type_rising_val;
 
        /*
         * The type_in_mask flag means that the underlying hardware uses
@@ -234,27 +234,42 @@ static int regmap_irq_set_type(struct irq_data *data, unsigned int type)
        struct regmap_irq_chip_data *d = irq_data_get_irq_chip_data(data);

drivers/base/regmap/regmap-irq.c static int regmap_irq_set_type(struct irq_data *data, unsigned int type) · line 241

Quoted 10 lines
drivers/base/regmap/regmap-irq.c +4 -3
        struct regmap *map = d->map;
        const struct regmap_irq *irq_data = irq_to_regmap_irq(d, data->hwirq);
-       int reg = irq_data->type_reg_offset / map->reg_stride;
+       int reg;
+       const struct regmap_irq_type *t = &irq_data->type;
 
-       if (!(irq_data->type_rising_mask | irq_data->type_falling_mask))
-               return 0;
+       if ((t->types_supported & type) != type)
+               return -ENOTSUPP;

Given types_supported defaults to zero, I think this breaks every existing setup using REGMAP_IRQ_REG().

Gr{oetje,eeting}s,

Geert

Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
#5 lore raw

Hello Geert,

Sorry for waiting - I just opened my computer after the holidays.

Quoted 20 lines · Geert Uytterhoeven

Hi Matti,

Quoted 2 lines · Matti Vaittinen

Add level active IRQ support to regmap-irq irqchip. Change breaks existing regmap-irq type setting. Convert the existing drivers which

Indeed it does.

Quoted 10 lines

use regmap-irq with trigger type setting (gpio-max77620) to work with this new approach. So we do not magically support level-active IRQs on gpio-max77620 - but add support to the regmap-irq for chips which support them =)

We do not support distinguishing situation where HW supports rising and falling edge detection but not both. Separating this would require inventing yet another flags for IRQ types.

Quoted 8 lines

This is now upstream as commit 1c2928e3e3212252 ("regmap:
regmap-irq/gpio-max77620: add level-irq support"), and breaks da9063-rtc on the Renesas Koelsch board:

genirq: Setting trigger mode 8 for irq 157 failed
(regmap_irq_set_type+0x0/0x140)

    da9063-rtc da9063-rtc: Failed to request ALARM IRQ 157: -524
    da9063-rtc: probe of da9063-rtc failed with error -524

This is strange as I do not see any type setting support code in drivers/mfd/da9063-irq.c. The type setting registers are neither specified in static const struct regmap_irq_chip da9063l_irq_chip nor in static const struct regmap_irq_chip da9063_irq_chip. Hence I don't understand how the da9063 could have been supporting IRQ type setting in first place.

Quoted 43 lines
Quoted 43 lines

---

Version 3 of this patch is intended to be functionally identical to v2. This patch is rebased on top of a tree which contains changes:
"regmap: irq: handle HW using separate rising/falling edge interrupts" from Bartosz Golaszewski and the change
"regmap: regmap-irq: Remove default irq type setting from core" (proposed here):
https://lore.kernel.org/lkml/[email protected]/

There should not be direct dependency to "regmap: regmap-irq: Remove default irq type setting from core" though. Patch was also tested to apply cleany on regmap-tree.

Same statement regarding testing applies - gpio-max77620 are only tested to compile. All real testing would be _HIGHLY_ appreciated.

 drivers/base/regmap/regmap-irq.c | 35 ++++++++++-----
 drivers/gpio/gpio-max77620.c     | 96 ++++++++++++++++++++++++++--------------
 include/linux/regmap.h           | 27 ++++++++---
 3 files changed, 110 insertions(+), 48 deletions(-)
drivers/base/regmap/regmap-irq.c +3 -2
index 8b216b2e2c19..31d23c9a5ae7 100644
@@ -199,7 +199,7 @@ static void regmap_irq_enable(struct irq_data *data)
        const struct regmap_irq *irq_data = irq_to_regmap_irq(d, data->hwirq);
        unsigned int mask, type;
 
-       type = irq_data->type_falling_mask | irq_data->type_rising_mask;
+       type = irq_data->type.type_falling_val | irq_data->type.type_rising_val;
 
        /*
         * The type_in_mask flag means that the underlying hardware uses
@@ -234,27 +234,42 @@ static int regmap_irq_set_type(struct irq_data *data, unsigned int type)
        struct regmap_irq_chip_data *d = irq_data_get_irq_chip_data(data);
        struct regmap *map = d->map;
        const struct regmap_irq *irq_data = irq_to_regmap_irq(d, data->hwirq);
-       int reg = irq_data->type_reg_offset / map->reg_stride;
+       int reg;
+       const struct regmap_irq_type *t = &irq_data->type;
 
Quoted 7 lines

drivers/base/regmap/regmap-irq.c static int regmap_irq_set_type(struct irq_data *data, unsigned int type) · line 241

Quoted 4 lines
drivers/base/regmap/regmap-irq.c +2 -2
-       if (!(irq_data->type_rising_mask | irq_data->type_falling_mask))
-               return 0;
+       if ((t->types_supported & type) != type)
+               return -ENOTSUPP;

Given types_supported defaults to zero, I think this breaks every existing setup using REGMAP_IRQ_REG().

Not really. The type setting support is not too old or widely used in regmap_irq. If the type_base and num_type_reg in struct regmap_irq_chip have not been given the type setting has not been supported. And the macro REGMAP_IRQ_REG() has not been initializing those fields - they must have been explicitly set by the driver. Only in-tree driver which really used the regmap_irq type-setting was gpio-max77620 (if my grepping did not go totally wrong). Is your version of drivers/mfd/da9063-irq.c same I can see in
https://elixir.bootlin.com/linux/v4.20/source/drivers/mfd/da9063-irq.c ?

I will try to see if the regmap_irq code has just silently ignored irq type setting requests if type setting information has not been populated by driver. May that has been changed by my patch. I wonder what would be the correct action if there is no type-setting information in struct regmap_irq_chip - and if type setting irq callbacks are called.

Br,
Matti Vaittinen

Signature · 4 lines
Matti Vaittinen
ROHM Semiconductors

~~~ "I don't think so," said Rene Descartes.  Just then, he vanished ~~~
#6 lore raw

Hello All,

Quoted 30 lines · Matti Vaittinen
Quoted 20 lines · Geert Uytterhoeven

Hi Matti,

Quoted 2 lines · Matti Vaittinen

Add level active IRQ support to regmap-irq irqchip. Change breaks existing regmap-irq type setting. Convert the existing drivers which

Indeed it does.

Quoted 10 lines

use regmap-irq with trigger type setting (gpio-max77620) to work with this new approach. So we do not magically support level-active IRQs on gpio-max77620 - but add support to the regmap-irq for chips which support them =)

We do not support distinguishing situation where HW supports rising and falling edge detection but not both. Separating this would require inventing yet another flags for IRQ types.

Quoted 8 lines

This is now upstream as commit 1c2928e3e3212252 ("regmap:
regmap-irq/gpio-max77620: add level-irq support"), and breaks da9063-rtc on the Renesas Koelsch board:

genirq: Setting trigger mode 8 for irq 157 failed
(regmap_irq_set_type+0x0/0x140)

    da9063-rtc da9063-rtc: Failed to request ALARM IRQ 157: -524
    da9063-rtc: probe of da9063-rtc failed with error -524
Quoted 57 lines

This is strange as I do not see any type setting support code in drivers/mfd/da9063-irq.c. The type setting registers are neither specified in static const struct regmap_irq_chip da9063l_irq_chip nor in static const struct regmap_irq_chip da9063_irq_chip. Hence I don't understand how the da9063 could have been supporting IRQ type setting in first place.

Quoted 50 lines
Quoted 37 lines

---

Version 3 of this patch is intended to be functionally identical to v2. This patch is rebased on top of a tree which contains changes:
"regmap: irq: handle HW using separate rising/falling edge interrupts" from Bartosz Golaszewski and the change
"regmap: regmap-irq: Remove default irq type setting from core" (proposed here):
https://lore.kernel.org/lkml/[email protected]/

There should not be direct dependency to "regmap: regmap-irq: Remove default irq type setting from core" though. Patch was also tested to apply cleany on regmap-tree.

Same statement regarding testing applies - gpio-max77620 are only tested to compile. All real testing would be _HIGHLY_ appreciated.

 drivers/base/regmap/regmap-irq.c | 35 ++++++++++-----
 drivers/gpio/gpio-max77620.c     | 96 ++++++++++++++++++++++++++--------------
 include/linux/regmap.h           | 27 ++++++++---
 3 files changed, 110 insertions(+), 48 deletions(-)
drivers/base/regmap/regmap-irq.c +1 -1
index 8b216b2e2c19..31d23c9a5ae7 100644
@@ -199,7 +199,7 @@ static void regmap_irq_enable(struct irq_data *data)
        const struct regmap_irq *irq_data = irq_to_regmap_irq(d, data->hwirq);
        unsigned int mask, type;
 
-       type = irq_data->type_falling_mask | irq_data->type_rising_mask;
+       type = irq_data->type.type_falling_val | irq_data->type.type_rising_val;
 
        /*
         * The type_in_mask flag means that the underlying hardware uses
@@ -234,27 +234,42 @@ static int regmap_irq_set_type(struct irq_data *data, unsigned int type)
        struct regmap_irq_chip_data *d = irq_data_get_irq_chip_data(data);

drivers/base/regmap/regmap-irq.c static int regmap_irq_set_type(struct irq_data *data, unsigned int type) · line 241

Quoted 10 lines
drivers/base/regmap/regmap-irq.c +4 -3
        struct regmap *map = d->map;
        const struct regmap_irq *irq_data = irq_to_regmap_irq(d, data->hwirq);
-       int reg = irq_data->type_reg_offset / map->reg_stride;
+       int reg;
+       const struct regmap_irq_type *t = &irq_data->type;
 
-       if (!(irq_data->type_rising_mask | irq_data->type_falling_mask))
-               return 0;
+       if ((t->types_supported & type) != type)
+               return -ENOTSUPP;

Given types_supported defaults to zero, I think this breaks every existing setup using REGMAP_IRQ_REG().

Right. Now I see what you mean. Original code did:

	if (!(irq_data->type_rising_mask | irq_data->type_falling_mask))
		return 0;

Eg, even when the driver was not able to perform the type-setting this failure was silently ignored, right. So doing:

       if ((t->types_supported & type) != type)
               return 0;

would be functionally equal. It feels like utterly wrong thing to do (to me) - if driver is written to work with edge or level active interrupts - and if the irq controller is not supporting this - then we should warn the user. Just silently ignoring this sounds like asking for irq storm or missed interrupts - but maybe I just don't get this =)

I'll send a patch with

	if (!(irq_data->type_rising_mask | irq_data->type_falling_mask))
		return 0;

in order to not break existing functionality - but it feels plain wrong to me.

Quoted 8 lines

Br,
Matti Vaittinen

Signature · 4 lines
Matti Vaittinen
ROHM Semiconductors

~~~ "I don't think so," said Rene Descartes.  Just then, he vanished ~~~
Signature · 4 lines
Matti Vaittinen
ROHM Semiconductors

~~~ "I don't think so," said Rene Descartes.  Just then, he vanished ~~~
#7 lore raw
Quoted 60 lines · Matti Vaittinen

Hello All,

Quoted 19 lines · Matti Vaittinen
Quoted 9 lines · Geert Uytterhoeven

Hi Matti,

Quoted 2 lines · Matti Vaittinen

Add level active IRQ support to regmap-irq irqchip. Change breaks existing regmap-irq type setting. Convert the existing drivers which

Indeed it does.

Quoted 8 lines

This is now upstream as commit 1c2928e3e3212252 ("regmap:
regmap-irq/gpio-max77620: add level-irq support"), and breaks da9063-rtc on the Renesas Koelsch board:

genirq: Setting trigger mode 8 for irq 157 failed
(regmap_irq_set_type+0x0/0x140)

    da9063-rtc da9063-rtc: Failed to request ALARM IRQ 157: -524
    da9063-rtc: probe of da9063-rtc failed with error -524
Quoted 22 lines

This is strange as I do not see any type setting support code in drivers/mfd/da9063-irq.c. The type setting registers are neither specified in static const struct regmap_irq_chip da9063l_irq_chip nor in static const struct regmap_irq_chip da9063_irq_chip. Hence I don't understand how the da9063 could have been supporting IRQ type setting in first place.

Quoted 15 lines

Review static int regmap_irq_set_type(struct irq_data *data, unsigned int type) · line 241

Quoted 12 lines
(quoted diff) +4 -3
@@ -234,27 +234,42 @@ static int regmap_irq_set_type(struct irq_data *data, unsigned int type)
        struct regmap_irq_chip_data *d = irq_data_get_irq_chip_data(data);
        struct regmap *map = d->map;
        const struct regmap_irq *irq_data = irq_to_regmap_irq(d, data->hwirq);
-       int reg = irq_data->type_reg_offset / map->reg_stride;
+       int reg;
+       const struct regmap_irq_type *t = &irq_data->type;
 
-       if (!(irq_data->type_rising_mask | irq_data->type_falling_mask))
-               return 0;
+       if ((t->types_supported & type) != type)
+               return -ENOTSUPP;

Given types_supported defaults to zero, I think this breaks every existing setup using REGMAP_IRQ_REG().

Right. Now I see what you mean. Original code did:

	if (!(irq_data->type_rising_mask | irq_data->type_falling_mask))
		return 0;

Eg, even when the driver was not able to perform the type-setting this failure was silently ignored, right. So doing:

       if ((t->types_supported & type) != type)
               return 0;

would be functionally equal. It feels like utterly wrong thing to do (to me) - if driver is written to work with edge or level active interrupts - and if the irq controller is not supporting this - then we should warn the user. Just silently ignoring this sounds like asking for irq storm or missed interrupts - but maybe I just don't get this =)

Quoted 5 lines

I'll send a patch with

	if (!(irq_data->type_rising_mask | irq_data->type_falling_mask))
		return 0;

in order to not break existing functionality - but it feels plain wrong to me.

Geert, I did send this patch yesterday. It's here if you wish to try it:
https://lore.kernel.org/patchwork/patch/1027963/

Last night - just when I was about to get some sleep - it stroke me. I think the correct thing to do would be leaving the irq_set_type to NULL for those IRQ chips which do not support type setting. If we do that, then the irq core will take care of situations where user requests type setting but the chip does not support it. Which means the regmap-irq would be no different from any other irq chip where type setting is not supported.

/// irrelevant ///
I guess you know the moment of "Hypnagogia" when you are comfortably at bed your head feels all dizzy and world is starting to faint away... And just a second later you are fully awake thinking of a possible solution - and definitely not able to sleep anymore :/

https://www.reddit.com/r/ProgrammerHumor/comments/93eq0e/everytime/
/// irrelevant ends ///

I just took a peek in
kernel/irq/manage.c - and found following:

int __irq_set_trigger(struct irq_desc *desc, unsigned long flags)

{
	struct irq_chip *chip = desc->irq_data.chip;
	int ret, unmask = 0;

	if (!chip || !chip->irq_set_type) {
		/*
		 * IRQF_TRIGGER_* but the PIC does not support multiple
		 * flow-types?
		 */
		pr_debug("No set_type function for IRQ %d (%s)\n",
			 irq_desc_get_irq(desc),
			 chip ? (chip->name ? : "unknown") : "unknown");
		return 0;
	}

...

so at the moment the IRQ core is also just spilling out the warning and then returning zero. But at least we would get the warning from core if the irq-chip does not support type config.

So at the cost of removing "const" from regmap_irq_chip we could do:

drivers/base/regmap/regmap-irq.c +2 -2
index 2a031743f31f..b6aef50ab378 100644
@@ -298,12 +298,11 @@ static int regmap_irq_set_wake(struct irq_data *data, unsigned int on)
 	return 0;
 }
 
-static const struct irq_chip regmap_irq_chip = {
+static struct irq_chip regmap_irq_chip = {
 	.irq_bus_lock		= regmap_irq_lock,
 	.irq_bus_sync_unlock	= regmap_irq_sync_unlock,
 	.irq_disable		= regmap_irq_disable,
 	.irq_enable		= regmap_irq_enable,
-	.irq_set_type		= regmap_irq_set_type,
 	.irq_set_wake		= regmap_irq_set_wake,
 };
 
@@ -610,6 +609,7 @@ int regmap_add_irq_chip(struct regmap *map, int irq, int irq_flags,
 
 	num_type_reg = chip->type_in_mask ? chip->num_regs : chip->num_type_reg;
 	if (num_type_reg) {
+		regmap_irq_chip.irq_set_type = regmap_irq_set_type;
 		d->type_buf_def = kcalloc(num_type_reg,
 					  sizeof(unsigned int), GFP_KERNEL);
 		if (!d->type_buf_def)

Mark, Geert, what do you think? (And maybe same for the .irq_set_wake - but I did omit this as I have never looked at the wake functionality before).

Br,
Matti Vaittinen

Signature · 4 lines
Matti Vaittinen
ROHM Semiconductors

~~~ "I don't think so," said Rene Descartes.  Just then, he vanished ~~~
maintainer
#8 lore raw
Quoted 7 lines · Matti Vaittinen

Last night - just when I was about to get some sleep - it stroke me. I think the correct thing to do would be leaving the irq_set_type to NULL for those IRQ chips which do not support type setting. If we do that, then the irq core will take care of situations where user requests type setting but the chip does not support it. Which means the regmap-irq would be no different from any other irq chip where type setting is not supported.

Yes, this is the best fix - let the framework handle things properly. We'll need a second set of operations and to select which to use based on having type information but that's fine.

Quoted 1 line

So at the cost of removing "const" from regmap_irq_chip we could do:

...

Quoted 3 lines

Mark, Geert, what do you think? (And maybe same for the .irq_set_wake - but I did omit this as I have never looked at the wake functionality before).

We need a separate struct as otherwise if there's multiple devices with regmap irq_chip implementations then they'll collide with each other but otherwise I like this approach (or we could copy the irq_chip struct when registering and then modify which is going to scale a bit better - you're probably right that we need to do the same thing for the wake configuration. I'll still look at applying your patch as a temporary fix though.

#9 lore raw
Quoted 16 lines · Mark Brown
Quoted 7 lines · Matti Vaittinen

Last night - just when I was about to get some sleep - it stroke me. I think the correct thing to do would be leaving the irq_set_type to NULL for those IRQ chips which do not support type setting. If we do that, then the irq core will take care of situations where user requests type setting but the chip does not support it. Which means the regmap-irq would be no different from any other irq chip where type setting is not supported.

Yes, this is the best fix - let the framework handle things properly. We'll need a second set of operations and to select which to use based on having type information but that's fine.

Quoted 1 line

So at the cost of removing "const" from regmap_irq_chip we could do:

Quoted 8 lines

...

Quoted 3 lines

Mark, Geert, what do you think? (And maybe same for the .irq_set_wake - but I did omit this as I have never looked at the wake functionality before).

We need a separate struct as otherwise if there's multiple devices with regmap irq_chip implementations then they'll collide with each other

Right. I must admit I didn't notice this! I was about to make a nasty error there...

Quoted 2 lines

otherwise I like this approach (or we could copy the irq_chip struct when registering and then modify which is going to scale a bit better -

I am really not a fan of dynamic allocation - I'd rather had static structs with different set of operations. But I admit I can't think of a sane system where we would have more than few regmap_irq controllers so memory consumption of allocating new structs is hardly an issue here.

Quoted 3 lines

you're probably right that we need to do the same thing for the wake configuration. I'll still look at applying your patch as a temporary fix though.

Thanks Mark. I try to cook a patch with copying of struct irq_chip still at this week but I wont rush it (I have some other topics under work) as the regression should be fixed by the other patch.

Br,
Matti Vaittinen

Signature · 4 lines
Matti Vaittinen
ROHM Semiconductors

~~~ "I don't think so," said Rene Descartes.  Just then, he vanished ~~~
#10 lore raw
Quoted 23 lines · Matti Vaittinen
Quoted 22 lines · Mark Brown
Quoted 7 lines · Matti Vaittinen

Last night - just when I was about to get some sleep - it stroke me. I think the correct thing to do would be leaving the irq_set_type to NULL for those IRQ chips which do not support type setting. If we do that, then the irq core will take care of situations where user requests type setting but the chip does not support it. Which means the regmap-irq would be no different from any other irq chip where type setting is not supported.

Yes, this is the best fix - let the framework handle things properly. We'll need a second set of operations and to select which to use based on having type information but that's fine.

Quoted 1 line

So at the cost of removing "const" from regmap_irq_chip we could do:

...

Quoted 3 lines

Mark, Geert, what do you think? (And maybe same for the .irq_set_wake - but I did omit this as I have never looked at the wake functionality before).

Quoted 6 lines
Quoted 2 lines

We need a separate struct as otherwise if there's multiple devices with regmap irq_chip implementations then they'll collide with each other

Right. I must admit I didn't notice this! I was about to make a nasty error there...

Looking at the code I think it just copies the struct anyway, basically using it as a template so I think this should be fine.

Quoted 8 lines
Quoted 3 lines

you're probably right that we need to do the same thing for the wake configuration. I'll still look at applying your patch as a temporary fix though.

Thanks Mark. I try to cook a patch with copying of struct irq_chip still at this week but I wont rush it (I have some other topics under work) as the regression should be fixed by the other patch.

Just to check that is this patch here:

https://lore.kernel.org/lkml/[email protected]/

Just want to check what will be applied so I know it will fix the regression I am seeing as well.

Thanks,
Charles

#11 lore raw
Quoted 27 lines · Charles Keepax
Quoted 26 lines · Matti Vaittinen
Quoted 24 lines · Mark Brown
Quoted 7 lines · Matti Vaittinen

Last night - just when I was about to get some sleep - it stroke me. I think the correct thing to do would be leaving the irq_set_type to NULL for those IRQ chips which do not support type setting. If we do that, then the irq core will take care of situations where user requests type setting but the chip does not support it. Which means the regmap-irq would be no different from any other irq chip where type setting is not supported.

Yes, this is the best fix - let the framework handle things properly. We'll need a second set of operations and to select which to use based on having type information but that's fine.

Quoted 1 line

So at the cost of removing "const" from regmap_irq_chip we could do:

...

Quoted 3 lines

Mark, Geert, what do you think? (And maybe same for the .irq_set_wake - but I did omit this as I have never looked at the wake functionality before).

We need a separate struct as otherwise if there's multiple devices with regmap irq_chip implementations then they'll collide with each other

Quoted 6 lines
Quoted 3 lines

Right. I must admit I didn't notice this! I was about to make a nasty error there...

Looking at the code I think it just copies the struct anyway, basically using it as a template so I think this should be fine.

Rigth. It seems to be:

d->irq_chip = regmap_irq_chip;
not:
d->irq_chip = &regmap_irq_chip;

and

struct regmap_irq_chip_data {
        struct mutex lock;
        struct irq_chip irq_chip;

        struct regmap *map;
	...
};

not

struct regmap_irq_chip_data {
        struct mutex lock;
        struct irq_chip *irq_chip;

        struct regmap *map;
	...
};
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

you're probably right that we need to do the same thing for the wake configuration. I'll still look at applying your patch as a temporary fix though.

I won't touch the wake thing (at least not yet) as I am not familiar with it. Is it Ok to change it with another patch later?

Quoted 6 lines
Quoted 5 lines

Thanks Mark. I try to cook a patch with copying of struct irq_chip still at this week but I wont rush it (I have some other topics under work) as the regression should be fixed by the other patch.

Quoted 6 lines

Just to check that is this patch here:

https://lore.kernel.org/lkml/[email protected]/

Just want to check what will be applied so I know it will fix the regression I am seeing as well.

Yep. That's the patch Mark did apply.

Quoted 3 lines

Thanks,
Charles

Signature · 4 lines
Matti Vaittinen
ROHM Semiconductors

~~~ "I don't think so," said Rene Descartes.  Just then, he vanished ~~~