Skip to contentSkip to conversation

Fix IRQ issue by setting IRQ_DISABLE_UNLAZY flag

PATCH Superseded Masahiro Honda 3 participants 8 messages Apr 20, 2023 – May 1, 2023

Subsystems (2)

v3 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 Fix IRQ issue by setting IRQ_DISABLE_UNLAZY flag

The Sigma-Delta ADCs supported by this driver can use SDO as an interrupt line to indicate the completion of a conversion. However, some devices cannot properly detect the completion of a conversion by an interrupt. This is for the reason mentioned in the following commit.

commit e9849777d0e2 ("genirq: Add flag to force mask in

                      disable_irq[_nosync]()")

A read operation is performed by an extra interrupt before the completion of a conversion. This patch fixes the issue by setting IRQ_DISABLE_UNLAZY flag.

---
v3:
- Remove the Kconfig option.
v2: https://lore.kernel.org/linux-iio/[email protected]/
- Rework commit message.
- Add a new entry in the Kconfig.
- Call irq_clear_status_flags(irq, IRQ_DISABLE_UNLAZY) when freeing the IRQ.
v1: https://lore.kernel.org/linux-iio/[email protected]/

 drivers/iio/adc/ad_sigma_delta.c | 25 ++++++++++++++++++++-----
 1 file changed, 20 insertions(+), 5 deletions(-)
drivers/iio/adc/ad_sigma_delta.c +20 -5
index d8570f620..215ecbedb 100644
@@ -565,6 +565,14 @@ int ad_sd_validate_trigger(struct iio_dev *indio_dev, struct iio_trigger *trig)
 }
 EXPORT_SYMBOL_NS_GPL(ad_sd_validate_trigger, IIO_AD_SIGMA_DELTA);
 
+static void ad_sd_free_irq(void *sd)
+{
+	struct ad_sigma_delta *sigma_delta = sd;
+
+	irq_clear_status_flags(sigma_delta->spi->irq, IRQ_DISABLE_UNLAZY);
+	free_irq(sigma_delta->spi->irq, sigma_delta);
+}
+
 static int devm_ad_sd_probe_trigger(struct device *dev, struct iio_dev *indio_dev)
 {
 	struct ad_sigma_delta *sigma_delta = iio_device_get_drvdata(indio_dev);
@@ -584,11 +592,18 @@ static int devm_ad_sd_probe_trigger(struct device *dev, struct iio_dev *indio_de
 	init_completion(&sigma_delta->completion);
 
 	sigma_delta->irq_dis = true;
-	ret = devm_request_irq(dev, sigma_delta->spi->irq,
-			       ad_sd_data_rdy_trig_poll,
-			       sigma_delta->info->irq_flags | IRQF_NO_AUTOEN,
-			       indio_dev->name,
-			       sigma_delta);
+	irq_set_status_flags(sigma_delta->spi->irq, IRQ_DISABLE_UNLAZY);
+	ret = request_irq(sigma_delta->spi->irq,
+			  ad_sd_data_rdy_trig_poll,
+			  sigma_delta->info->irq_flags | IRQF_NO_AUTOEN,
+			  indio_dev->name,
+			  sigma_delta);
+	if (ret) {
+		irq_clear_status_flags(sigma_delta->spi->irq, IRQ_DISABLE_UNLAZY);
+		return ret;
+	}
+
+	ret = devm_add_action_or_reset(dev, ad_sd_free_irq, sigma_delta);
 	if (ret)
 		return ret;
 
Signature · 1 line
2.34.1
Reviewed-by on v3 1/1
#1 lore raw
Quoted 8 lines · Masahiro Honda

The Sigma-Delta ADCs supported by this driver can use SDO as an interrupt line to indicate the completion of a conversion. However, some devices cannot properly detect the completion of a conversion by an interrupt. This is for the reason mentioned in the following commit.

commit e9849777d0e2 ("genirq: Add flag to force mask in disable_irq[_nosync]()")

Quoted 6 lines

A read operation is performed by an extra interrupt before the completion of a conversion. This patch fixes the issue by setting IRQ_DISABLE_UNLAZY flag.

---

LGTM:

maintainer
#2 lore raw
In reply to Masahiro Honda
Quoted 30 lines · On Thu, 20 Apr 2023 19:23:16 +0900 Masahiro Honda <[email protected]> wrote:

The Sigma-Delta ADCs supported by this driver can use SDO as an interrupt line to indicate the completion of a conversion. However, some devices cannot properly detect the completion of a conversion by an interrupt. This is for the reason mentioned in the following commit.

commit e9849777d0e2 ("genirq: Add flag to force mask in

                      disable_irq[_nosync]()")

A read operation is performed by an extra interrupt before the completion of a conversion. This patch fixes the issue by setting IRQ_DISABLE_UNLAZY flag.

---
v3:
- Remove the Kconfig option.
v2: https://lore.kernel.org/linux-iio/[email protected]/
- Rework commit message.
- Add a new entry in the Kconfig.
- Call irq_clear_status_flags(irq, IRQ_DISABLE_UNLAZY) when freeing the IRQ.
v1: https://lore.kernel.org/linux-iio/[email protected]/

 drivers/iio/adc/ad_sigma_delta.c | 25 ++++++++++++++++++++-----
 1 file changed, 20 insertions(+), 5 deletions(-)
drivers/iio/adc/ad_sigma_delta.c +0 -0
index d8570f620..215ecbedb 100644

drivers/iio/adc/ad_sigma_delta.c int ad_sd_validate_trigger(struct iio_dev *indio_dev, struct iio_trigger *trig) · line 574

Quoted 10 lines
drivers/iio/adc/ad_sigma_delta.c +7 -0
@@ -565,6 +565,14 @@ int ad_sd_validate_trigger(struct iio_dev *indio_dev, struct iio_trigger *trig)
 }
 EXPORT_SYMBOL_NS_GPL(ad_sd_validate_trigger, IIO_AD_SIGMA_DELTA);
 
+static void ad_sd_free_irq(void *sd)
+{
+	struct ad_sigma_delta *sigma_delta = sd;
+
+	irq_clear_status_flags(sigma_delta->spi->irq, IRQ_DISABLE_UNLAZY);
+	free_irq(sigma_delta->spi->irq, sigma_delta);
+}

Don't fuse the two operations unwinding like this. Just register a callback that only does the irq_clear_status_flags immediately after setting them. Then leave the orginally devm_request_irq call alone. If it fails, the devm cleanup will deal with the irq_clear_status_flag for you.

It almost never makes sense for a single devm call to unwind more than one function call in a driver.

Otherwise this looks fine to me,

Thanks,

Jonathan

Quoted 28 lines
drivers/iio/adc/ad_sigma_delta.c +13 -5
+
 static int devm_ad_sd_probe_trigger(struct device *dev, struct iio_dev *indio_dev)
 {
 	struct ad_sigma_delta *sigma_delta = iio_device_get_drvdata(indio_dev);
@@ -584,11 +592,18 @@ static int devm_ad_sd_probe_trigger(struct device *dev, struct iio_dev *indio_de
 	init_completion(&sigma_delta->completion);
 
 	sigma_delta->irq_dis = true;
-	ret = devm_request_irq(dev, sigma_delta->spi->irq,
-			       ad_sd_data_rdy_trig_poll,
-			       sigma_delta->info->irq_flags | IRQF_NO_AUTOEN,
-			       indio_dev->name,
-			       sigma_delta);
+	irq_set_status_flags(sigma_delta->spi->irq, IRQ_DISABLE_UNLAZY);
+	ret = request_irq(sigma_delta->spi->irq,
+			  ad_sd_data_rdy_trig_poll,
+			  sigma_delta->info->irq_flags | IRQF_NO_AUTOEN,
+			  indio_dev->name,
+			  sigma_delta);
+	if (ret) {
+		irq_clear_status_flags(sigma_delta->spi->irq, IRQ_DISABLE_UNLAZY);
+		return ret;
+	}
+
+	ret = devm_add_action_or_reset(dev, ad_sd_free_irq, sigma_delta);
 	if (ret)
 		return ret;
 
#3 lore raw

Hi Jonathan,

Quoted 44 lines · Jonathan Cameron
Quoted 41 lines · On Thu, 20 Apr 2023 19:23:16 +0900 Masahiro Honda <[email protected]> wrote:

The Sigma-Delta ADCs supported by this driver can use SDO as an interrupt line to indicate the completion of a conversion. However, some devices cannot properly detect the completion of a conversion by an interrupt. This is for the reason mentioned in the following commit.

commit e9849777d0e2 ("genirq: Add flag to force mask in disable_irq[_nosync]()")

A read operation is performed by an extra interrupt before the completion of a conversion. This patch fixes the issue by setting IRQ_DISABLE_UNLAZY flag.

---
v3:
 - Remove the Kconfig option.
v2:
https://lore.kernel.org/linux-iio/[email protected]/
 - Rework commit message.
 - Add a new entry in the Kconfig.
 - Call irq_clear_status_flags(irq, IRQ_DISABLE_UNLAZY) when freeing the IRQ.
v1:
https://lore.kernel.org/linux-iio/[email protected]/

 drivers/iio/adc/ad_sigma_delta.c | 25 ++++++++++++++++++++-----

 1 file changed, 20 insertions(+), 5 deletions(-)
drivers/iio/adc/ad_sigma_delta.c +0 -0
index d8570f620..215ecbedb 100644
@@ -565,6 +565,14 @@ int ad_sd_validate_trigger(struct iio_dev *indio_dev,
 struct iio_trigger *trig)
  }
  EXPORT_SYMBOL_NS_GPL(ad_sd_validate_trigger, IIO_AD_SIGMA_DELTA);

drivers/iio/adc/ad_sigma_delta.c +3 -0
+static void ad_sd_free_irq(void *sd)
+{
+       struct ad_sigma_delta *sigma_delta = sd;
Quoted 8 lines

drivers/iio/adc/ad_sigma_delta.c int ad_sd_validate_trigger(struct iio_dev *indio_dev, struct iio_trigger *trig) · line 574

Quoted 4 lines
drivers/iio/adc/ad_sigma_delta.c +4 -0
+
+       irq_clear_status_flags(sigma_delta->spi->irq, IRQ_DISABLE_UNLAZY);
+       free_irq(sigma_delta->spi->irq, sigma_delta);
+}

Don't fuse the two operations unwinding like this.  Just register a callback that only
does the irq_clear_status_flags immediately after setting them.  Then leave

I was the one to propose fusing them together because I thought that we could have issues by clearing the flag after calling free_irq(). After looking again at the IRQ code, I can see that it is not up to free_irq() to free the allocated irq_descs (that might only happen when unmapping the virq) which means we should be fine doing the normal way.

That said, looking at the only users that care to clear this flag, it looks like they do it before calling free_irq(). Hence, I'm not sure if there's anything subtle going on. In fact, looking at this line:

https://elixir.bootlin.com/linux/latest/source/kernel/irq/manage.c#L1909

I'm not so sure we actually need to clear the flag as for these devices, we should only have one consumer/action per IRQ. Anyways, probably for correctness we should still explicitly clear it?

- Nuno Sá

maintainer
#4 lore raw
Quoted 56 lines · On Mon, 24 Apr 2023 11:09:34 +0200 Nuno Sá <[email protected]> wrote:

Hi Jonathan,

Quoted 44 lines · Jonathan Cameron
Quoted 41 lines · On Thu, 20 Apr 2023 19:23:16 +0900 Masahiro Honda <[email protected]> wrote:

The Sigma-Delta ADCs supported by this driver can use SDO as an interrupt line to indicate the completion of a conversion. However, some devices cannot properly detect the completion of a conversion by an interrupt. This is for the reason mentioned in the following commit.

commit e9849777d0e2 ("genirq: Add flag to force mask in disable_irq[_nosync]()")

A read operation is performed by an extra interrupt before the completion of a conversion. This patch fixes the issue by setting IRQ_DISABLE_UNLAZY flag.

---
v3:
 - Remove the Kconfig option.
v2:
https://lore.kernel.org/linux-iio/[email protected]/
 - Rework commit message.
 - Add a new entry in the Kconfig.
 - Call irq_clear_status_flags(irq, IRQ_DISABLE_UNLAZY) when freeing the IRQ.
v1:
https://lore.kernel.org/linux-iio/[email protected]/

 drivers/iio/adc/ad_sigma_delta.c | 25 ++++++++++++++++++++-----

 1 file changed, 20 insertions(+), 5 deletions(-)
drivers/iio/adc/ad_sigma_delta.c +0 -0
index d8570f620..215ecbedb 100644
@@ -565,6 +565,14 @@ int ad_sd_validate_trigger(struct iio_dev *indio_dev,
 struct iio_trigger *trig)
  }
  EXPORT_SYMBOL_NS_GPL(ad_sd_validate_trigger, IIO_AD_SIGMA_DELTA);

drivers/iio/adc/ad_sigma_delta.c +3 -0
+static void ad_sd_free_irq(void *sd)
+{
+       struct ad_sigma_delta *sigma_delta = sd;
Quoted 8 lines

drivers/iio/adc/ad_sigma_delta.c int ad_sd_validate_trigger(struct iio_dev *indio_dev, struct iio_trigger *trig) · line 574

Quoted 4 lines
drivers/iio/adc/ad_sigma_delta.c +4 -0
+
+       irq_clear_status_flags(sigma_delta->spi->irq, IRQ_DISABLE_UNLAZY);
+       free_irq(sigma_delta->spi->irq, sigma_delta);
+}  

Don't fuse the two operations unwinding like this.  Just register a callback that only
does the irq_clear_status_flags immediately after setting them.  Then leave

Quoted 5 lines

I was the one to propose fusing them together because I thought that we could have issues by clearing the flag after calling free_irq(). After looking again at the IRQ code, I can see that it is not up to free_irq() to free the allocated irq_descs (that might only happen when unmapping the virq) which means we should be fine doing the normal way.

Ah. I'd missed the ordering. If that had been valid (and I think you are correct that it is not required) then a comment to make that clear would be necessary.

Usual case of: When doing something non obvious with ordering, say why.

Quoted 10 lines

That said, looking at the only users that care to clear this flag, it looks like they do it before calling free_irq(). Hence, I'm not sure if there's anything subtle going on. In fact, looking at this line:

https://elixir.bootlin.com/linux/latest/source/kernel/irq/manage.c#L1909

I'm not so sure we actually need to clear the flag as for these devices, we should only have one consumer/action per IRQ. Anyways, probably for correctness we should still explicitly clear it?

Good question... Looks to me like a driver shouldn't be clearing this flag itself, but it's probably harmless in most cases.

I'd drop the clear of the status flag, perhaps adding a comment that the irq core does it for us.

Jonathan

Quoted 3 lines

- Nuno Sá

#5 lore raw

Hi all,

Quoted 82 lines · Jonathan Cameron
Quoted 56 lines · On Mon, 24 Apr 2023 11:09:34 +0200 Nuno Sá <[email protected]> wrote:

Hi Jonathan,

Quoted 41 lines · Jonathan Cameron
Quoted 38 lines · On Thu, 20 Apr 2023 19:23:16 +0900 Masahiro Honda <[email protected]> wrote:

The Sigma-Delta ADCs supported by this driver can use SDO as an interrupt line to indicate the completion of a conversion. However, some devices cannot properly detect the completion of a conversion by an interrupt. This is for the reason mentioned in the following commit.

commit e9849777d0e2 ("genirq: Add flag to force mask in

                      disable_irq[_nosync]()")

A read operation is performed by an extra interrupt before the completion of a conversion. This patch fixes the issue by setting IRQ_DISABLE_UNLAZY flag.

---
v3:
- Remove the Kconfig option.
v2:
https://lore.kernel.org/linux-iio/[email protected]/
- Rework commit message.
- Add a new entry in the Kconfig.
- Call irq_clear_status_flags(irq, IRQ_DISABLE_UNLAZY) when freeing the
IRQ.
v1:
https://lore.kernel.org/linux-iio/[email protected]/

 drivers/iio/adc/ad_sigma_delta.c | 25 ++++++++++++++++++++-----
 1 file changed, 20 insertions(+), 5 deletions(-)
drivers/iio/adc/ad_sigma_delta.c +0 -0
index d8570f620..215ecbedb 100644
@@ -565,6 +565,14 @@ int ad_sd_validate_trigger(struct iio_dev *indio_dev,
 struct iio_trigger *trig)
 }
 EXPORT_SYMBOL_NS_GPL(ad_sd_validate_trigger, IIO_AD_SIGMA_DELTA);
 
Quoted 11 lines

drivers/iio/adc/ad_sigma_delta.c int ad_sd_validate_trigger(struct iio_dev *indio_dev, struct iio_trigger *trig) · line 574

Quoted 7 lines
drivers/iio/adc/ad_sigma_delta.c +7 -0
+static void ad_sd_free_irq(void *sd)
+{
+       struct ad_sigma_delta *sigma_delta = sd;
+
+       irq_clear_status_flags(sigma_delta->spi->irq, IRQ_DISABLE_UNLAZY);
+       free_irq(sigma_delta->spi->irq, sigma_delta);
+}

Don't fuse the two operations unwinding like this. Just register a callback that only
does the irq_clear_status_flags immediately after setting them. Then leave

Quoted 5 lines

I was the one to propose fusing them together because I thought that we could have issues by clearing the flag after calling free_irq(). After looking again at the IRQ code, I can see that it is not up to free_irq() to free the allocated irq_descs (that might only happen when unmapping the virq) which means we should be fine doing the normal way.

Ah. I'd missed the ordering. If that had been valid (and I think you are correct that it is not required) then a comment to make that clear would be necessary.

Usual case of: When doing something non obvious with ordering, say why.

Quoted 10 lines

That said, looking at the only users that care to clear this flag, it looks like they do it before calling free_irq(). Hence, I'm not sure if there's anything subtle going on. In fact, looking at this line:

https://elixir.bootlin.com/linux/latest/source/kernel/irq/manage.c#L1909

I'm not so sure we actually need to clear the flag as for these devices, we should only have one consumer/action per IRQ. Anyways, probably for correctness we should still explicitly clear it?

Quoted 13 lines

Good question... Looks to me like a driver shouldn't be clearing this flag itself, but it's probably harmless in most cases.

I'd drop the clear of the status flag, perhaps adding a comment that the irq core does it for us.

Jonathan

Quoted 3 lines

- Nuno Sá

I'll remove the callback and just call irq_set_status_flags() with a comment.

Thanks,

Masahiro

#6 lore raw
In reply to Jonathan Cameron

Hi Jonathan,

Quoted 50 lines · Jonathan Cameron
Quoted 30 lines · On Thu, 20 Apr 2023 19:23:16 +0900 Masahiro Honda <[email protected]> wrote:

The Sigma-Delta ADCs supported by this driver can use SDO as an interrupt line to indicate the completion of a conversion. However, some devices cannot properly detect the completion of a conversion by an interrupt. This is for the reason mentioned in the following commit.

commit e9849777d0e2 ("genirq: Add flag to force mask in

                      disable_irq[_nosync]()")

A read operation is performed by an extra interrupt before the completion of a conversion. This patch fixes the issue by setting IRQ_DISABLE_UNLAZY flag.

---
v3:
- Remove the Kconfig option.
v2: https://lore.kernel.org/linux-iio/[email protected]/
- Rework commit message.
- Add a new entry in the Kconfig.
- Call irq_clear_status_flags(irq, IRQ_DISABLE_UNLAZY) when freeing the IRQ.
v1: https://lore.kernel.org/linux-iio/[email protected]/

 drivers/iio/adc/ad_sigma_delta.c | 25 ++++++++++++++++++++-----
 1 file changed, 20 insertions(+), 5 deletions(-)
drivers/iio/adc/ad_sigma_delta.c +0 -0
index d8570f620..215ecbedb 100644

drivers/iio/adc/ad_sigma_delta.c int ad_sd_validate_trigger(struct iio_dev *indio_dev, struct iio_trigger *trig) · line 574

Quoted 10 lines
drivers/iio/adc/ad_sigma_delta.c +7 -0
@@ -565,6 +565,14 @@ int ad_sd_validate_trigger(struct iio_dev *indio_dev, struct iio_trigger *trig)
 }
 EXPORT_SYMBOL_NS_GPL(ad_sd_validate_trigger, IIO_AD_SIGMA_DELTA);
 
+static void ad_sd_free_irq(void *sd)
+{
+     struct ad_sigma_delta *sigma_delta = sd;
+
+     irq_clear_status_flags(sigma_delta->spi->irq, IRQ_DISABLE_UNLAZY);
+     free_irq(sigma_delta->spi->irq, sigma_delta);
+}

Don't fuse the two operations unwinding like this. Just register a callback that only does the irq_clear_status_flags immediately after setting them. Then leave the orginally devm_request_irq call alone. If it fails, the devm cleanup will deal with the irq_clear_status_flag for you.

Quoted 9 lines

It almost never makes sense for a single devm call to unwind more than one function call in a driver.

Otherwise this looks fine to me,

Thanks,

Jonathan

I understand. I'll fix it.

Thanks,

#7 lore raw

Hi Jonathan,

Quoted 3 lines · Masahiro Honda

I understand. I'll fix it.

Sorry, I made a mistake.
I would like to wait for a conclusion.

Regards,