On 17.06.2016 18:09, Grygorii Strashko wrote:

On 06/17/2016 06:31 PM, Marcin Niestroj wrote:

quoted

On 17.06.2016 16:42, Grygorii Strashko wrote:

quoted

On 06/17/2016 05:05 PM, Marcin Niestroj wrote:

quoted

On 16.06.2016 15:03, Grygorii Strashko wrote:

quoted

On 06/16/2016 02:41 PM, Marcin Niestroj wrote:

quoted

Add support for handling IRQs: power button, AC and USB power state
changes. Mask and interrupt bits are shared within one register, which
prevents us to use regmap_irq implementation. New irq_domain is
created in
order to add interrupt handling for each tps65217's subsystem. IRQ
resources have been added for charger subsystem to be able to notify
about
AC and USB state changes.

Signed-off-by: Marcin Niestroj <redacted>
---
  drivers/mfd/Kconfig          |   1 +
  drivers/mfd/tps65217.c       | 194
+++++++++++++++++++++++++++++++++++++++++--
  include/linux/mfd/tps65217.h |  11 +++
  3 files changed, 198 insertions(+), 8 deletions(-)

[...]

quoted

quoted

quoted

+{
+    int ret;
+
+    mutex_init(&tps->irq_lock);
+
+    /* Mask all interrupt sources */
+    tps->irq_mask = (TPS65217_INT_RESERVEDM | TPS65217_INT_PBM
+            | TPS65217_INT_ACM | TPS65217_INT_USBM);
+    tps65217_reg_write(tps, TPS65217_REG_INT, tps->irq_mask,
+            TPS65217_PROTECT_NONE);
+
+    tps->irq_domain = irq_domain_add_linear(tps->dev->of_node,
+        TPS65217_NUM_IRQ, &tps65217_irq_domain_ops, tps);
+    if (!tps->irq_domain) {
+        dev_err(tps->dev, "Could not create IRQ domain\n");
+        return -ENOMEM;
+    }
+
+    ret = devm_request_threaded_irq(tps->dev, irq, NULL,
+                    tps65217_irq_thread,
+                    IRQF_TRIGGER_RISING | IRQF_ONESHOT,

Are there any reasons why IRQ trigger type specified here explicitly?

Not really. I have configured it that way, it worked and I forgot about
it when preparing patches. Could you give some hint here?

It's better to get it from DT and in case of DT boot it will - the real
IRQ trigger type may depends on board.

So what I understand, I need to remove IRQF_TRIGGER_RISING here.
Addidional flags will be passed by 'irq_flags' in
"interrupts = <irq_num irq_flags>" in DT, right?

Indeed. And usually it is not "Addidional flags" - its mandatory cell for
 the most of IRQ controllers.

The problem with hard-coded IRQ trigger type values in code is that on
different boards IRQ can be wired on different way - to the GIC/INTC,
to SoC GPIO, to GPIO expanders and .. And they can support different sets
of allowed IRQ triggering types. More over, on some boards IRQ signal can be
inverted, for example :P

So, It's more generic to not hard-code it if your driver is expected to be
used only with DT.
Also as per TRM http://www.ti.com/lit/ds/symlink/tps65217.pdf
nINT - interrupt output (active low, open drain). Pin is pulled low if an interrupt bit is set. The
       output goes high after the bit causing the interrupt in register INT has been read.
       The interrupt sources can be masked in register INT, so no interrupt is generated when the
       corresponding interrupt bit is set

Thanks very much for making this clear to me.

I have just send new patch version with fixes from your comments.

-- 
Regards,
Marcin Niestroj