Skip to contentSkip to conversation

Last batch of W=1 warning fixes in MFD

PATCH In review Lee Jones 4 participants 21 messages Jun 29, 2020 – Jul 6, 2020

Subsystems (3)
lore
Share & tools
lore

Lee Jones

author
#0 lore raw

cover Last batch of W=1 warning fixes in MFD

Attempting to clean-up W=1 kernel builds, which are currently overwhelmingly riddled with niggly little warnings.

Lee Jones (5):
mfd: si476x-cmd: Add missing documentation for

    si476x_cmd_fm_rds_status()'s arg 'report'

mfd: lm3533-ctrlbank: Cap BRIGHTNESS_MAX to 127 since API uses u8 as
carrier
mfd: rave-sp: Fix mistake in 'struct rave_sp_deframer's kerneldoc
mfd: sprd-sc27xx-spi: Fix divide by zero when allocating register
offset/mask
mfd: axp20x-i2c: Do not define 'struct acpi_device_id' when
!CONFIG_ACPI

 drivers/mfd/axp20x-i2c.c      | 2 ++
 drivers/mfd/lm3533-ctrlbank.c | 2 +-
 drivers/mfd/rave-sp.c         | 2 +-
 drivers/mfd/si476x-cmd.c      | 2 ++
 drivers/mfd/sprd-sc27xx-spi.c | 2 +-
 5 files changed, 7 insertions(+), 3 deletions(-)
Signature · 7 lines
2.25.1


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
author
#1 lore raw

PATCH 1/5 mfd: si476x-cmd: Add missing documentation for si476x_cmd_fm_rds_status()'s arg 'report'

Kerneldoc syntax is used, but not complete.

Descriptions are required for all arguments.

Fixes the following W=1 build warning:

drivers/mfd/si476x-cmd.c:907: warning: Function parameter or member 'report' not described in 'si476x_core_cmd_fm_rds_status'

Cc Andrey Smirnov <[email protected]>
---
 drivers/mfd/si476x-cmd.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/mfd/si476x-cmd.c +2 -0
index 0a40c5cb751db..d15b3e7833692 100644
@@ -892,6 +892,8 @@ EXPORT_SYMBOL_GPL(si476x_core_cmd_fm_seek_start);
  *                rest RDS data contains the last valid info received
  * @mtfifo: if set the command clears RDS receive FIFO
  * @intack: if set the command clards the RDSINT bit.
+ * @report: - all signal quality information retured by the command
+ *           (if NULL then the output of the command is ignored)
  *
  * Function returns 0 on success and negative error code on failure
  */
Signature · 7 lines
2.25.1


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
author
#2 lore raw

PATCH 2/5 mfd: lm3533-ctrlbank: Cap BRIGHTNESS_MAX to 127 since API uses u8 as carrier

In reply to Lee Jones

Since its conception in 2012 brightness has been artificially capped at 127 since the variable carrying the value is u8. We could go to the trouble of changing the whole API (crossing 3 different subsystems), but clearly this hasn't bothered anyone in the best part of a decade.

Simply, cap BRIGHTNESS_MAX to 127 instead (for now at least).

Fixes the following W=1 warning(s):

drivers/mfd/lm3533-ctrlbank.c: In function ‘lm3533_ctrlbank_set_brightness’:
drivers/mfd/lm3533-ctrlbank.c:98:10: warning: comparison is always false due to limited range of data type [-Wtype-limits] 98 | if (val > LM3533_##_NAME##_MAX) | ^
drivers/mfd/lm3533-ctrlbank.c:125:1: note: in expansion of macro ‘lm3533_ctrlbank_set’
125 | lm3533_ctrlbank_set(brightness, BRIGHTNESS);
| ^~~~~~~~~~~~~~~~~~~

Cc Johan Hovold <[email protected]>
---
 drivers/mfd/lm3533-ctrlbank.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mfd/lm3533-ctrlbank.c +1 -1
index 34fba06ec7057..348ce67453092 100644
@@ -17,7 +17,7 @@
 #define LM3533_MAX_CURRENT_MAX		29800
 #define LM3533_MAX_CURRENT_STEP		800
 
-#define LM3533_BRIGHTNESS_MAX		255
+#define LM3533_BRIGHTNESS_MAX		127  /* Capped by API - could be up to 255 */
 #define LM3533_PWM_MAX			0x3f
 
 #define LM3533_REG_PWM_BASE		0x14
Signature · 7 lines
2.25.1


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#3 lore raw

Re: [PATCH 2/5] mfd: lm3533-ctrlbank: Cap BRIGHTNESS_MAX to 127 since API uses u8 as carrier

Quoted 6 lines · Lee Jones

Since its conception in 2012 brightness has been artificially capped at 127 since the variable carrying the value is u8. We could go to the trouble of changing the whole API (crossing 3 different subsystems), but clearly this hasn't bothered anyone in the best part of a decade.

Simply, cap BRIGHTNESS_MAX to 127 instead (for now at least).

Hmm. This patch is clearly broken and would contrary to the claim be introducing an artificial cap at half brightness. u8 can hold the max brightness value 255 just fine.

Quoted 8 lines

Fixes the following W=1 warning(s):

drivers/mfd/lm3533-ctrlbank.c: In function ‘lm3533_ctrlbank_set_brightness’:
drivers/mfd/lm3533-ctrlbank.c:98:10: warning: comparison is always false due to limited range of data type [-Wtype-limits] 98 | if (val > LM3533_##_NAME##_MAX) | ^
drivers/mfd/lm3533-ctrlbank.c:125:1: note: in expansion of macro ‘lm3533_ctrlbank_set’
125 | lm3533_ctrlbank_set(brightness, BRIGHTNESS);
| ^~~~~~~~~~~~~~~~~~~

This warning is benign. The same macro is used to defined two function where in one case the max value coincides with U8_MAX so that the sanity check becomes redundant.

Quoted 11 lines
Cc Johan Hovold <[email protected]>
---
 drivers/mfd/lm3533-ctrlbank.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mfd/lm3533-ctrlbank.c +0 -0
index 34fba06ec7057..348ce67453092 100644

drivers/mfd/lm3533-ctrlbank.c line 23

Quoted 8 lines
drivers/mfd/lm3533-ctrlbank.c +1 -1
@@ -17,7 +17,7 @@
 #define LM3533_MAX_CURRENT_MAX		29800
 #define LM3533_MAX_CURRENT_STEP		800
 
-#define LM3533_BRIGHTNESS_MAX		255
+#define LM3533_BRIGHTNESS_MAX		127  /* Capped by API - could be up to 255 */
 #define LM3533_PWM_MAX			0x3f
 
 #define LM3533_REG_PWM_BASE		0x14

Johan

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

author
#4 lore raw

Re: [PATCH 2/5] mfd: lm3533-ctrlbank: Cap BRIGHTNESS_MAX to 127 since API uses u8 as carrier

Quoted 11 lines · Johan Hovold
Quoted 6 lines · Lee Jones

Since its conception in 2012 brightness has been artificially capped at 127 since the variable carrying the value is u8. We could go to the trouble of changing the whole API (crossing 3 different subsystems), but clearly this hasn't bothered anyone in the best part of a decade.

Simply, cap BRIGHTNESS_MAX to 127 instead (for now at least).

Hmm. This patch is clearly broken and would contrary to the claim be introducing an artificial cap at half brightness. u8 can hold the max brightness value 255 just fine.

Yes, of course it can. Senior moment on my account.

Quoted 9 lines
Quoted 8 lines

Fixes the following W=1 warning(s):

drivers/mfd/lm3533-ctrlbank.c: In function ‘lm3533_ctrlbank_set_brightness’:
drivers/mfd/lm3533-ctrlbank.c:98:10: warning: comparison is always false due to limited range of data type [-Wtype-limits] 98 | if (val > LM3533_##_NAME##_MAX) | ^
drivers/mfd/lm3533-ctrlbank.c:125:1: note: in expansion of macro ‘lm3533_ctrlbank_set’
125 | lm3533_ctrlbank_set(brightness, BRIGHTNESS);
| ^~~~~~~~~~~~~~~~~~~

Quoted 3 lines

This warning is benign. The same macro is used to defined two function where in one case the max value coincides with U8_MAX so that the sanity check becomes redundant.

A benign warning, as most W=1 warnings are, is still a warning.

So how do you propose we fix it?

Is 255 a valid and used brightness level?

If so, how do you feel about:

  /* Avoid 'always false' check '(u8) > 255' */
  if (LM3533_##_NAME##_MAX != 0xff && val > LM3533_##_NAME##_MAX)
          return -EINVAL;        
Quoted 21 lines
Quoted 11 lines
Cc Johan Hovold <[email protected]>
---
 drivers/mfd/lm3533-ctrlbank.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mfd/lm3533-ctrlbank.c +0 -0
index 34fba06ec7057..348ce67453092 100644

drivers/mfd/lm3533-ctrlbank.c line 23

Quoted 8 lines
drivers/mfd/lm3533-ctrlbank.c +1 -1
@@ -17,7 +17,7 @@
 #define LM3533_MAX_CURRENT_MAX		29800
 #define LM3533_MAX_CURRENT_STEP		800
 
-#define LM3533_BRIGHTNESS_MAX		255
+#define LM3533_BRIGHTNESS_MAX		127  /* Capped by API - could be up to 255 */
 #define LM3533_PWM_MAX			0x3f
 
 #define LM3533_REG_PWM_BASE		0x14

Johan

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

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#5 lore raw

Re: [PATCH 2/5] mfd: lm3533-ctrlbank: Cap BRIGHTNESS_MAX to 127 since API uses u8 as carrier

Quoted 25 lines · Lee Jones
Quoted 11 lines · Johan Hovold
Quoted 6 lines · Lee Jones

Since its conception in 2012 brightness has been artificially capped at 127 since the variable carrying the value is u8. We could go to the trouble of changing the whole API (crossing 3 different subsystems), but clearly this hasn't bothered anyone in the best part of a decade.

Simply, cap BRIGHTNESS_MAX to 127 instead (for now at least).

Hmm. This patch is clearly broken and would contrary to the claim be introducing an artificial cap at half brightness. u8 can hold the max brightness value 255 just fine.

Yes, of course it can. Senior moment on my account.

Quoted 9 lines
Quoted 8 lines

Fixes the following W=1 warning(s):

drivers/mfd/lm3533-ctrlbank.c: In function ‘lm3533_ctrlbank_set_brightness’:
drivers/mfd/lm3533-ctrlbank.c:98:10: warning: comparison is always false due to limited range of data type [-Wtype-limits] 98 | if (val > LM3533_##_NAME##_MAX) | ^
drivers/mfd/lm3533-ctrlbank.c:125:1: note: in expansion of macro ‘lm3533_ctrlbank_set’
125 | lm3533_ctrlbank_set(brightness, BRIGHTNESS);
| ^~~~~~~~~~~~~~~~~~~

Quoted 5 lines
Quoted 3 lines

This warning is benign. The same macro is used to defined two function where in one case the max value coincides with U8_MAX so that the sanity check becomes redundant.

A benign warning, as most W=1 warnings are, is still a warning.

Not every warning needs to be addressed, there's a reason some of these are hidden behind W=1 or higher.

Quoted 3 lines

So how do you propose we fix it?

Is 255 a valid and used brightness level?

Yes.

Quoted 5 lines

If so, how do you feel about:

  /* Avoid 'always false' check '(u8) > 255' */
  if (LM3533_##_NAME##_MAX != 0xff && val > LM3533_##_NAME##_MAX)
          return -EINVAL;        

I'm afraid that's not sufficient to shut the compiler up.

I'll send you patch expanding these accessors instead. Having exported functions implemented by macros is particularly nice (hard to grep for etc). There are a couple of more sets of control-bank registers that could potentially have shared the implementation and which motivated the use of macros, but it does not seem very likely that we'll be adding those anytime soon anyway.

Johan

author
#6 lore raw

PATCH 3/5 mfd: rave-sp: Fix mistake in 'struct rave_sp_deframer's kerneldoc

In reply to Lee Jones

Argument 'received' was incorrectly named by its struct type 'completion' instead of its name.

Fixes the following W=1 warning:

drivers/mfd/rave-sp.c:107: warning: Function parameter or member 'received' not described in 'rave_sp_reply'

Cc Andrey Vostrikov <[email protected]>
Cc Nikita Yushchenko <[email protected]>
Cc Andrey Smirnov <[email protected]>
---
 drivers/mfd/rave-sp.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mfd/rave-sp.c +1 -1
index 26c7b63e008a8..abaab541df19d 100644
@@ -96,7 +96,7 @@ struct rave_sp_deframer {
  * @data:	Buffer to store reply payload in
  * @code:	Expected reply code
  * @ackid:	Expected reply ACK ID
- * @completion: Successful reply reception completion
+ * @received:   Successful reply reception completion
  */
 struct rave_sp_reply {
 	size_t length;
Signature · 7 lines
2.25.1


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
author
#7 lore raw

PATCH 4/5 mfd: sprd-sc27xx-spi: Fix divide by zero when allocating register offset/mask

In reply to Lee Jones

Since ddata->irqs[] is already zeroed when allocated by devm_kcalloc() and dividing 0 by anything is still 0, there is no need to re-assign ddata->irqs[i].* values. Instead, it should be safe to begin at 1.

This fixes the following W=1 warning:

drivers/mfd/sprd-sc27xx-spi.c:255 sprd_pmic_probe() debug: sval_binop_unsigned: divide by zero

Cc Orson Zhai <[email protected]>
Cc Baolin Wang <[email protected]>
Cc Chunyan Zhang <[email protected]>
---
 drivers/mfd/sprd-sc27xx-spi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mfd/sprd-sc27xx-spi.c +1 -1
index c305e941e435c..694a7d429ccff 100644
@@ -251,7 +251,7 @@ static int sprd_pmic_probe(struct spi_device *spi)
 		return -ENOMEM;
 
 	ddata->irq_chip.irqs = ddata->irqs;
-	for (i = 0; i < pdata->num_irqs; i++) {
+	for (i = 1; i < pdata->num_irqs; i++) {
 		ddata->irqs[i].reg_offset = i / pdata->num_irqs;
 		ddata->irqs[i].mask = BIT(i % pdata->num_irqs);
 	}
Signature · 7 lines
2.25.1


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#8 lore raw

Re: [PATCH 4/5] mfd: sprd-sc27xx-spi: Fix divide by zero when allocating register offset/mask

Quoted 21 lines · Lee Jones

Since ddata->irqs[] is already zeroed when allocated by devm_kcalloc() and dividing 0 by anything is still 0, there is no need to re-assign ddata->irqs[i].* values. Instead, it should be safe to begin at 1.

This fixes the following W=1 warning:

drivers/mfd/sprd-sc27xx-spi.c:255 sprd_pmic_probe() debug: sval_binop_unsigned: divide by zero

Cc Orson Zhai <[email protected]>
Cc Baolin Wang <[email protected]>
Cc Chunyan Zhang <[email protected]>
---
 drivers/mfd/sprd-sc27xx-spi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mfd/sprd-sc27xx-spi.c +0 -0
index c305e941e435c..694a7d429ccff 100644

drivers/mfd/sprd-sc27xx-spi.c static int sprd_pmic_probe(struct spi_device *spi) · line 257

Quoted 8 lines
drivers/mfd/sprd-sc27xx-spi.c +1 -1
@@ -251,7 +251,7 @@ static int sprd_pmic_probe(struct spi_device *spi)
 		return -ENOMEM;
 
 	ddata->irq_chip.irqs = ddata->irqs;
-	for (i = 0; i < pdata->num_irqs; i++) {
+	for (i = 1; i < pdata->num_irqs; i++) {
 		ddata->irqs[i].reg_offset = i / pdata->num_irqs;
 		ddata->irqs[i].mask = BIT(i % pdata->num_irqs);
 	}

This doesn't look right either.

First, the loop is never executed if num_irqs is zero.

Second, the current code looks bogus too as reg_offset is always set to zero and mask to BIT(i)...

Johan

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

author
#9 lore raw

Re: [PATCH 4/5] mfd: sprd-sc27xx-spi: Fix divide by zero when allocating register offset/mask

Quoted 33 lines · Johan Hovold
Quoted 21 lines · Lee Jones

Since ddata->irqs[] is already zeroed when allocated by devm_kcalloc() and dividing 0 by anything is still 0, there is no need to re-assign ddata->irqs[i].* values. Instead, it should be safe to begin at 1.

This fixes the following W=1 warning:

drivers/mfd/sprd-sc27xx-spi.c:255 sprd_pmic_probe() debug: sval_binop_unsigned: divide by zero

Cc Orson Zhai <[email protected]>
Cc Baolin Wang <[email protected]>
Cc Chunyan Zhang <[email protected]>
---
 drivers/mfd/sprd-sc27xx-spi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mfd/sprd-sc27xx-spi.c +0 -0
index c305e941e435c..694a7d429ccff 100644

drivers/mfd/sprd-sc27xx-spi.c static int sprd_pmic_probe(struct spi_device *spi) · line 257

Quoted 8 lines
drivers/mfd/sprd-sc27xx-spi.c +1 -1
@@ -251,7 +251,7 @@ static int sprd_pmic_probe(struct spi_device *spi)
 		return -ENOMEM;
 
 	ddata->irq_chip.irqs = ddata->irqs;
-	for (i = 0; i < pdata->num_irqs; i++) {
+	for (i = 1; i < pdata->num_irqs; i++) {
 		ddata->irqs[i].reg_offset = i / pdata->num_irqs;
 		ddata->irqs[i].mask = BIT(i % pdata->num_irqs);
 	}

This doesn't look right either.

Quoted 1 line

First, the loop is never executed if num_irqs is zero.

The point of the patch is that 0 entries are never processed.

However, what I appear to have overlooked is that BIT(0 % x) is not 0, it's 1.

Quoted 2 lines

Second, the current code looks bogus too as reg_offset is always set to zero and mask to BIT(i)...

Heh. I wonder if/how this was tested.

I'm going to wait to hear from the authors before attempting to fix this again.

Baolin, Could you please clarify this for us please?

--
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services

Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

maintainer
#10 lore raw

Re: [PATCH 4/5] mfd: sprd-sc27xx-spi: Fix divide by zero when allocating register offset/mask

Quoted 36 lines · Lee Jones
Quoted 33 lines · Johan Hovold
Quoted 21 lines · Lee Jones

Since ddata->irqs[] is already zeroed when allocated by devm_kcalloc() and dividing 0 by anything is still 0, there is no need to re-assign ddata->irqs[i].* values. Instead, it should be safe to begin at 1.

This fixes the following W=1 warning:

drivers/mfd/sprd-sc27xx-spi.c:255 sprd_pmic_probe() debug: sval_binop_unsigned: divide by zero

Cc Orson Zhai <[email protected]>
Cc Baolin Wang <[email protected]>
Cc Chunyan Zhang <[email protected]>
---
 drivers/mfd/sprd-sc27xx-spi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mfd/sprd-sc27xx-spi.c +0 -0
index c305e941e435c..694a7d429ccff 100644

drivers/mfd/sprd-sc27xx-spi.c static int sprd_pmic_probe(struct spi_device *spi) · line 257

Quoted 8 lines
drivers/mfd/sprd-sc27xx-spi.c +1 -1
@@ -251,7 +251,7 @@ static int sprd_pmic_probe(struct spi_device *spi)
            return -ENOMEM;
 
    ddata->irq_chip.irqs = ddata->irqs;
-   for (i = 0; i < pdata->num_irqs; i++) {
+   for (i = 1; i < pdata->num_irqs; i++) {
            ddata->irqs[i].reg_offset = i / pdata->num_irqs;
            ddata->irqs[i].mask = BIT(i % pdata->num_irqs);
    }

This doesn't look right either.

Quoted 6 lines
Quoted 1 line

First, the loop is never executed if num_irqs is zero.

The point of the patch is that 0 entries are never processed.

However, what I appear to have overlooked is that BIT(0 % x) is not 0, it's 1.

Yes.

Quoted 3 lines
Quoted 3 lines

Second, the current code looks bogus too as reg_offset is always set to zero and mask to BIT(i)...

Now the result is correct, since all PMIC irq mask bits are in one register now, which means the reg_offset is always 0 can work well. But I think the logics still can be improved if our PMIC irq numbers are larger than 32 in future.

Quoted 7 lines

Heh. I wonder if/how this was tested.

I'm going to wait to hear from the authors before attempting to fix this again.

Baolin, Could you please clarify this for us please?

Yes, see above comments.

Signature · 6 lines
Baolin Wang

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#11 lore raw

Re: [PATCH 4/5] mfd: sprd-sc27xx-spi: Fix divide by zero when allocating register offset/mask

Quoted 37 lines · Baolin Wang
Quoted 36 lines · Lee Jones
Quoted 33 lines · Johan Hovold
Quoted 21 lines · Lee Jones

Since ddata->irqs[] is already zeroed when allocated by devm_kcalloc() and dividing 0 by anything is still 0, there is no need to re-assign ddata->irqs[i].* values. Instead, it should be safe to begin at 1.

This fixes the following W=1 warning:

drivers/mfd/sprd-sc27xx-spi.c:255 sprd_pmic_probe() debug: sval_binop_unsigned: divide by zero

Cc Orson Zhai <[email protected]>
Cc Baolin Wang <[email protected]>
Cc Chunyan Zhang <[email protected]>
---
 drivers/mfd/sprd-sc27xx-spi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mfd/sprd-sc27xx-spi.c +0 -0
index c305e941e435c..694a7d429ccff 100644

drivers/mfd/sprd-sc27xx-spi.c static int sprd_pmic_probe(struct spi_device *spi) · line 257

Quoted 8 lines
drivers/mfd/sprd-sc27xx-spi.c +1 -1
@@ -251,7 +251,7 @@ static int sprd_pmic_probe(struct spi_device *spi)
            return -ENOMEM;
 
    ddata->irq_chip.irqs = ddata->irqs;
-   for (i = 0; i < pdata->num_irqs; i++) {
+   for (i = 1; i < pdata->num_irqs; i++) {
            ddata->irqs[i].reg_offset = i / pdata->num_irqs;
            ddata->irqs[i].mask = BIT(i % pdata->num_irqs);
    }

This doesn't look right either.

Quoted 3 lines
Quoted 3 lines
Quoted 1 line

First, the loop is never executed if num_irqs is zero.

The point of the patch is that 0 entries are never processed.

So what's the problem? There's no division by zero here.

And what compiler are you using, Lee? Seems broken.

Quoted 7 lines
Quoted 2 lines
Quoted 2 lines

Second, the current code looks bogus too as reg_offset is always set to zero and mask to BIT(i)...

Now the result is correct, since all PMIC irq mask bits are in one register now, which means the reg_offset is always 0 can work well. But I think the logics still can be improved if our PMIC irq numbers are larger than 32 in future.

The code is still bogus as pointed out above. Why do you bother to divide by num_irqs at all?

And what have you guys been smoking? ;)

Johan

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

maintainer
#12 lore raw

Re: [PATCH 4/5] mfd: sprd-sc27xx-spi: Fix divide by zero when allocating register offset/mask

Quoted 50 lines · Johan Hovold
Quoted 37 lines · Baolin Wang
Quoted 36 lines · Lee Jones
Quoted 33 lines · Johan Hovold
Quoted 21 lines · Lee Jones

Since ddata->irqs[] is already zeroed when allocated by devm_kcalloc() and dividing 0 by anything is still 0, there is no need to re-assign ddata->irqs[i].* values. Instead, it should be safe to begin at 1.

This fixes the following W=1 warning:

drivers/mfd/sprd-sc27xx-spi.c:255 sprd_pmic_probe() debug: sval_binop_unsigned: divide by zero

Cc Orson Zhai <[email protected]>
Cc Baolin Wang <[email protected]>
Cc Chunyan Zhang <[email protected]>
---
 drivers/mfd/sprd-sc27xx-spi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mfd/sprd-sc27xx-spi.c +0 -0
index c305e941e435c..694a7d429ccff 100644

drivers/mfd/sprd-sc27xx-spi.c static int sprd_pmic_probe(struct spi_device *spi) · line 257

Quoted 8 lines
drivers/mfd/sprd-sc27xx-spi.c +1 -1
@@ -251,7 +251,7 @@ static int sprd_pmic_probe(struct spi_device *spi)
            return -ENOMEM;
 
    ddata->irq_chip.irqs = ddata->irqs;
-   for (i = 0; i < pdata->num_irqs; i++) {
+   for (i = 1; i < pdata->num_irqs; i++) {
            ddata->irqs[i].reg_offset = i / pdata->num_irqs;
            ddata->irqs[i].mask = BIT(i % pdata->num_irqs);
    }

This doesn't look right either.

Quoted 3 lines
Quoted 3 lines
Quoted 1 line

First, the loop is never executed if num_irqs is zero.

The point of the patch is that 0 entries are never processed.

So what's the problem? There's no division by zero here.

And what compiler are you using, Lee? Seems broken.

Quoted 3 lines
Quoted 2 lines
Quoted 2 lines

Second, the current code looks bogus too as reg_offset is always set to zero and mask to BIT(i)...

Quoted 7 lines
Quoted 4 lines

Now the result is correct, since all PMIC irq mask bits are in one register now, which means the reg_offset is always 0 can work well. But I think the logics still can be improved if our PMIC irq numbers are larger than 32 in future.

The code is still bogus as pointed out above. Why do you bother to divide by num_irqs at all?

Right, no need to divide by num_irqs, can be simplified as below. Lee, care to resend your patch with simplifying the code? Or you want me to send a patch?

drivers/mfd/sprd-sc27xx-spi.c +2 -5
index 33336cde4724..2ed5f3a4e79c 100644
@@ -250,10 +250,8 @@ static int sprd_pmic_probe(struct spi_device *spi)
                return -ENOMEM;
 
        ddata->irq_chip.irqs = ddata->irqs;
-       for (i = 0; i < pdata->num_irqs; i++) {
-               ddata->irqs[i].reg_offset = i / pdata->num_irqs;
-               ddata->irqs[i].mask = BIT(i % pdata->num_irqs);
-       }
+       for (i = 0; i < pdata->num_irqs; i++)
+               ddata->irqs[i].mask = BIT(i);
 
-- 

Baolin Wang

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

author
#13 lore raw

Re: [PATCH 4/5] mfd: sprd-sc27xx-spi: Fix divide by zero when allocating register offset/mask

Quoted 67 lines · Baolin Wang
Quoted 50 lines · Johan Hovold
Quoted 37 lines · Baolin Wang
Quoted 36 lines · Lee Jones
Quoted 33 lines · Johan Hovold
Quoted 21 lines · Lee Jones

Since ddata->irqs[] is already zeroed when allocated by devm_kcalloc() and dividing 0 by anything is still 0, there is no need to re-assign ddata->irqs[i].* values. Instead, it should be safe to begin at 1.

This fixes the following W=1 warning:

drivers/mfd/sprd-sc27xx-spi.c:255 sprd_pmic_probe() debug: sval_binop_unsigned: divide by zero

Cc Orson Zhai <[email protected]>
Cc Baolin Wang <[email protected]>
Cc Chunyan Zhang <[email protected]>
---
 drivers/mfd/sprd-sc27xx-spi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mfd/sprd-sc27xx-spi.c +0 -0
index c305e941e435c..694a7d429ccff 100644

drivers/mfd/sprd-sc27xx-spi.c static int sprd_pmic_probe(struct spi_device *spi) · line 257

Quoted 8 lines
drivers/mfd/sprd-sc27xx-spi.c +1 -1
@@ -251,7 +251,7 @@ static int sprd_pmic_probe(struct spi_device *spi)
            return -ENOMEM;
 
    ddata->irq_chip.irqs = ddata->irqs;
-   for (i = 0; i < pdata->num_irqs; i++) {
+   for (i = 1; i < pdata->num_irqs; i++) {
            ddata->irqs[i].reg_offset = i / pdata->num_irqs;
            ddata->irqs[i].mask = BIT(i % pdata->num_irqs);
    }

This doesn't look right either.

Quoted 3 lines
Quoted 3 lines
Quoted 1 line

First, the loop is never executed if num_irqs is zero.

The point of the patch is that 0 entries are never processed.

So what's the problem? There's no division by zero here.

And what compiler are you using, Lee? Seems broken.

Quoted 3 lines
Quoted 2 lines
Quoted 2 lines

Second, the current code looks bogus too as reg_offset is always set to zero and mask to BIT(i)...

Quoted 7 lines
Quoted 4 lines

Now the result is correct, since all PMIC irq mask bits are in one register now, which means the reg_offset is always 0 can work well. But I think the logics still can be improved if our PMIC irq numbers are larger than 32 in future.

The code is still bogus as pointed out above. Why do you bother to divide by num_irqs at all?

Right, no need to divide by num_irqs, can be simplified as below. Lee, care to resend your patch with simplifying the code? Or you want me to send a patch?

drivers/mfd/sprd-sc27xx-spi.c +0 -0
index 33336cde4724..2ed5f3a4e79c 100644

drivers/mfd/sprd-sc27xx-spi.c static int sprd_pmic_probe(struct spi_device *spi) · line 254

Quoted 9 lines
drivers/mfd/sprd-sc27xx-spi.c +2 -4
@@ -250,10 +250,8 @@ static int sprd_pmic_probe(struct spi_device *spi)
                return -ENOMEM;
 
        ddata->irq_chip.irqs = ddata->irqs;
-       for (i = 0; i < pdata->num_irqs; i++) {
-               ddata->irqs[i].reg_offset = i / pdata->num_irqs;
-               ddata->irqs[i].mask = BIT(i % pdata->num_irqs);
-       }
+       for (i = 0; i < pdata->num_irqs; i++)
+               ddata->irqs[i].mask = BIT(i);

I'm happy to resend with your Suggested-by.

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

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
author
#14 lore raw

PATCH 1/1 mfd: sprd-sc27xx-spi: Fix-up bogus IRQ register offset and mask setting

In reply to Lee Jones

'i / pdata->num_irqs' always equates to 0 and 'BIT(i % pdata->num_irqs)' always ends up being BIT(i) here, so make that clearer in the code. If the code base needs to support more than 32 IRQs in the future, this will have to be reworked, but lets just keep it simple for as long as we can.

This fixes the following W=1 warning:

drivers/mfd/sprd-sc27xx-spi.c:255 sprd_pmic_probe() debug: sval_binop_unsigned: divide by zero

Cc Orson Zhai <[email protected]>
Cc Chunyan Zhang <[email protected]>
Cc Johan Hovold <[email protected]>
drivers/mfd/sprd-sc27xx-spi.c +2 -4
index c305e941e435c..4a1a61e1a86ea 100644
@@ -251,10 +251,8 @@ static int sprd_pmic_probe(struct spi_device *spi)
                return -ENOMEM;
 
        ddata->irq_chip.irqs = ddata->irqs;
-       for (i = 0; i < pdata->num_irqs; i++) {
-               ddata->irqs[i].reg_offset = i / pdata->num_irqs;
-               ddata->irqs[i].mask = BIT(i % pdata->num_irqs);
-       }
+       for (i = 0; i < pdata->num_irqs; i++)
+               ddata->irqs[i].mask = BIT(i);
 
        ret = devm_regmap_add_irq_chip(&spi->dev, ddata->regmap, ddata->irq,
                                       IRQF_ONESHOT | IRQF_NO_SUSPEND, 0,
maintainer Reviewed-by on v1 1/1
#15 lore raw

Re: [PATCH] mfd: sprd-sc27xx-spi: Fix-up bogus IRQ register offset and mask setting

Quoted 8 lines · Lee Jones

'i / pdata->num_irqs' always equates to 0 and 'BIT(i % pdata->num_irqs)' always ends up being BIT(i) here, so make that clearer in the code. If the code base needs to support more than 32 IRQs in the future, this will have to be reworked, but lets just keep it simple for as long as we can.

This fixes the following W=1 warning:

Quoted 7 lines

drivers/mfd/sprd-sc27xx-spi.c:255 sprd_pmic_probe() debug: sval_binop_unsigned: divide by zero

Cc Orson Zhai <[email protected]>
Cc Chunyan Zhang <[email protected]>
Cc Johan Hovold <[email protected]>

Thanks.

Quoted 18 lines
drivers/mfd/sprd-sc27xx-spi.c +2 -4
index c305e941e435c..4a1a61e1a86ea 100644
@@ -251,10 +251,8 @@ static int sprd_pmic_probe(struct spi_device *spi)
                return -ENOMEM;
 
        ddata->irq_chip.irqs = ddata->irqs;
-       for (i = 0; i < pdata->num_irqs; i++) {
-               ddata->irqs[i].reg_offset = i / pdata->num_irqs;
-               ddata->irqs[i].mask = BIT(i % pdata->num_irqs);
-       }
+       for (i = 0; i < pdata->num_irqs; i++)
+               ddata->irqs[i].mask = BIT(i);
 
        ret = devm_regmap_add_irq_chip(&spi->dev, ddata->regmap, ddata->irq,
                                       IRQF_ONESHOT | IRQF_NO_SUSPEND, 0,
Signature · 1 line
Baolin Wang
author
#16 lore raw

PATCH 5/5 mfd: axp20x-i2c: Do not define 'struct acpi_device_id' when !CONFIG_ACPI

In reply to Lee Jones

Since ACPI_PTR() is used to NULLify the value when !CONFIG_ACPI, struct axp20x_i2c_acpi_match becomes defined by unused.

This squashes the current W=1 kernel builds warning:

drivers/mfd/axp20x-i2c.c:82:36: warning: ‘axp20x_i2c_acpi_match’ defined but not used [-Wunused-const-variable=]

Cc Chen-Yu Tsai <[email protected]>
Cc Carlo Caione <[email protected]>
---
 drivers/mfd/axp20x-i2c.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/mfd/axp20x-i2c.c +2 -0
index 14f9df74f855c..3dd650125c239 100644
@@ -79,6 +79,7 @@ static const struct i2c_device_id axp20x_i2c_id[] = {
 };
 MODULE_DEVICE_TABLE(i2c, axp20x_i2c_id);
 
+#if IS_ENABLED(CONFIG_ACPI)
 static const struct acpi_device_id axp20x_i2c_acpi_match[] = {
 	{
 		.id = "INT33F4",
@@ -87,6 +88,7 @@ static const struct acpi_device_id axp20x_i2c_acpi_match[] = {
 	{ },
 };
 MODULE_DEVICE_TABLE(acpi, axp20x_i2c_acpi_match);
+#endif
 
 static struct i2c_driver axp20x_i2c_driver = {
 	.driver = {
Signature · 7 lines
2.25.1


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Acked-by on v1 5/5
#17 lore raw

Re: [PATCH 5/5] mfd: axp20x-i2c: Do not define 'struct acpi_device_id' when !CONFIG_ACPI

Adding Hans to the list as he's the one that deals with all the x86 platforms that use this series of chips.

Quoted 20 lines · Lee Jones

Since ACPI_PTR() is used to NULLify the value when !CONFIG_ACPI, struct axp20x_i2c_acpi_match becomes defined by unused.

This squashes the current W=1 kernel builds warning:

drivers/mfd/axp20x-i2c.c:82:36: warning: ‘axp20x_i2c_acpi_match’ defined but not used [-Wunused-const-variable=]

Cc Chen-Yu Tsai <[email protected]>
Cc Carlo Caione <[email protected]>
---
 drivers/mfd/axp20x-i2c.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/mfd/axp20x-i2c.c +0 -0
index 14f9df74f855c..3dd650125c239 100644

drivers/mfd/axp20x-i2c.c static const struct i2c_device_id axp20x_i2c_id[] = { · line 83

Quoted 5 lines
drivers/mfd/axp20x-i2c.c +1 -0
@@ -79,6 +79,7 @@ static const struct i2c_device_id axp20x_i2c_id[] = {
 };
 MODULE_DEVICE_TABLE(i2c, axp20x_i2c_id);
 
+#if IS_ENABLED(CONFIG_ACPI)
 static const struct acpi_device_id axp20x_i2c_acpi_match[] = {

I'd rather use "__maybe_unused" if possible to at least get a compile check, and also because "ACPI_PTR NULLifies the value" might not be well known to people not working on ACPI-based platforms.

Either way,

Quoted 11 lines
drivers/mfd/axp20x-i2c.c +1 -1
        {
                .id = "INT33F4",
@@ -87,6 +88,7 @@ static const struct acpi_device_id axp20x_i2c_acpi_match[] = {
        { },
 };
 MODULE_DEVICE_TABLE(acpi, axp20x_i2c_acpi_match);
+#endif
 
 static struct i2c_driver axp20x_i2c_driver = {
        .driver = {
--
Quoted 2 lines

2.25.1

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

author Applied
#18 lore raw

Re: [PATCH 5/5] mfd: axp20x-i2c: Do not define 'struct acpi_device_id' when !CONFIG_ACPI

Quoted 27 lines · Chen-Yu Tsai

Adding Hans to the list as he's the one that deals with all the x86 platforms that use this series of chips.

Quoted 23 lines · Lee Jones

Since ACPI_PTR() is used to NULLify the value when !CONFIG_ACPI, struct axp20x_i2c_acpi_match becomes defined by unused.

This squashes the current W=1 kernel builds warning:

drivers/mfd/axp20x-i2c.c:82:36: warning: ‘axp20x_i2c_acpi_match’ defined but not used [-Wunused-const-variable=]

Cc Chen-Yu Tsai <[email protected]>
Cc Carlo Caione <[email protected]>
---
 drivers/mfd/axp20x-i2c.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/mfd/axp20x-i2c.c +0 -0
index 14f9df74f855c..3dd650125c239 100644
@@ -79,6 +79,7 @@ static const struct i2c_device_id axp20x_i2c_id[] = {
 };
 MODULE_DEVICE_TABLE(i2c, axp20x_i2c_id);
 
Quoted 6 lines

drivers/mfd/axp20x-i2c.c static const struct i2c_device_id axp20x_i2c_id[] = { · line 83

Quoted 2 lines
drivers/mfd/axp20x-i2c.c +1 -0
+#if IS_ENABLED(CONFIG_ACPI)
 static const struct acpi_device_id axp20x_i2c_acpi_match[] = {

I'd rather use "__maybe_unused" if possible to at least get a compile check, and also because "ACPI_PTR NULLifies the value" might not be well known to people not working on ACPI-based platforms.

FYI, I've gone back to this patch, as it seems to be the common census across the rest of the kernel, due to the extra space saving.

The code will get compile checks when compiled with different architectures, and I'm not sure I understand the last point you made.

Quoted 3 lines

Either way,

I've applied this to the patch. Thanks.

Quoted 11 lines
Quoted 11 lines
drivers/mfd/axp20x-i2c.c +1 -0
        {
                .id = "INT33F4",
@@ -87,6 +88,7 @@ static const struct acpi_device_id axp20x_i2c_acpi_match[] = {
        { },
 };
 MODULE_DEVICE_TABLE(acpi, axp20x_i2c_acpi_match);
+#endif
 
 static struct i2c_driver axp20x_i2c_driver = {
        .driver = {
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
author
#19 lore raw

PATCH v2 1/1 mfd: axp20x-i2c: Tell the compiler that ACPI functions may not be used

In reply to Lee Jones

... as is the case when !CONFIG_ACPI.

This squashes the current W=1 kernel builds warning:

drivers/mfd/axp20x-i2c.c:82:36: warning: ‘axp20x_i2c_acpi_match’ defined but not used [-Wunused-const-variable=]

Cc Chen-Yu Tsai <[email protected]>
Cc Carlo Caione <[email protected]>
drivers/mfd/axp20x-i2c.c +1 -1
index 14f9df74f855c..7d08c06de7afc 100644
@@ -79,7 +79,7 @@ static const struct i2c_device_id axp20x_i2c_id[] = {
 };
 MODULE_DEVICE_TABLE(i2c, axp20x_i2c_id);
 
-static const struct acpi_device_id axp20x_i2c_acpi_match[] = {
+static const struct acpi_device_id __maybe_unused axp20x_i2c_acpi_match[] = {
        {
                .id = "INT33F4",
                .driver_data = AXP288_ID,
Acked-by on v2 1/1
#20 lore raw

Re: [PATCH v2] mfd: axp20x-i2c: Tell the compiler that ACPI functions may not be used

Quoted 10 lines · Lee Jones

... as is the case when !CONFIG_ACPI.

This squashes the current W=1 kernel builds warning:

drivers/mfd/axp20x-i2c.c:82:36: warning: ‘axp20x_i2c_acpi_match’ defined but not used [-Wunused-const-variable=]

Cc Chen-Yu Tsai <[email protected]>
Cc Carlo Caione <[email protected]>
Quoted 14 lines
drivers/mfd/axp20x-i2c.c +1 -1
index 14f9df74f855c..7d08c06de7afc 100644
@@ -79,7 +79,7 @@ static const struct i2c_device_id axp20x_i2c_id[] = {
 };
 MODULE_DEVICE_TABLE(i2c, axp20x_i2c_id);
 
-static const struct acpi_device_id axp20x_i2c_acpi_match[] = {
+static const struct acpi_device_id __maybe_unused axp20x_i2c_acpi_match[] = {
        {
                .id = "INT33F4",
                .driver_data = AXP288_ID,