[PATCH v2] net: smc91x: Fix pointer types

Thorsten Blum posted 1 patch 1 year, 9 months ago
There is a newer version of this series
drivers/net/ethernet/smsc/smc91x.h | 16 ++++++++--------
1 file changed, 8 insertions(+), 8 deletions(-)
[PATCH v2] net: smc91x: Fix pointer types
Posted by Thorsten Blum 1 year, 9 months ago
Use void __iomem pointers as parameters for mcf_insw() and mcf_outsw()
to align with the parameter types of readw() and writew().

Use lp->base instead of ioaddr when calling SMC_outsw(), SMC_outsb(),
SMC_insw(), and SMC_insb() to retain its type across macro boundaries
and to fix the following warnings reported by kernel test robot:

drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
drivers/net/ethernet/smsc/smc91x.c:483:17: sparse: warning: incorrect type in argument 1 (different address spaces)
drivers/net/ethernet/smsc/smc91x.c:483:17: sparse:    expected void *a
drivers/net/ethernet/smsc/smc91x.c:483:17: sparse:    got void [noderef] __iomem *

Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202405160853.3qyaSj8w-lkp@intel.com/
Acked-by: Nicolas Pitre <nico@fluxnic.net>
---
Changes in v2:
- Use lp->base instead of __ioaddr as suggested by Andrew Lunn. They are
  essentially the same, but using lp->base results in a smaller diff
- Remove whitespace only changes as suggested by Andrew Lunn
- Preserve Acked-by: Nicolas Pitre tag (please let me know if you
  somehow disagree with the changes in v2)
---
 drivers/net/ethernet/smsc/smc91x.h | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/drivers/net/ethernet/smsc/smc91x.h b/drivers/net/ethernet/smsc/smc91x.h
index 45ef5ac0788a..a1523fb503a3 100644
--- a/drivers/net/ethernet/smsc/smc91x.h
+++ b/drivers/net/ethernet/smsc/smc91x.h
@@ -142,14 +142,14 @@ static inline void _SMC_outw_align4(u16 val, void __iomem *ioaddr, int reg,
 #define SMC_CAN_USE_32BIT	0
 #define SMC_NOWAIT		1
 
-static inline void mcf_insw(void *a, unsigned char *p, int l)
+static inline void mcf_insw(void __iomem *a, unsigned char *p, int l)
 {
 	u16 *wp = (u16 *) p;
 	while (l-- > 0)
 		*wp++ = readw(a);
 }
 
-static inline void mcf_outsw(void *a, unsigned char *p, int l)
+static inline void mcf_outsw(void __iomem *a, unsigned char *p, int l)
 {
 	u16 *wp = (u16 *) p;
 	while (l-- > 0)
@@ -1034,7 +1034,7 @@ static const char * chip_ids[ 16 ] =  {
 			void __iomem *__ioaddr = ioaddr;		\
 			if (__len >= 2 && (unsigned long)__ptr & 2) {	\
 				__len -= 2;				\
-				SMC_outsw(ioaddr, DATA_REG(lp), __ptr, 1); \
+				SMC_outsw(lp->base, DATA_REG(lp), __ptr, 1); \
 				__ptr += 2;				\
 			}						\
 			if (SMC_CAN_USE_DATACS && lp->datacs)		\
@@ -1042,12 +1042,12 @@ static const char * chip_ids[ 16 ] =  {
 			SMC_outsl(__ioaddr, DATA_REG(lp), __ptr, __len>>2); \
 			if (__len & 2) {				\
 				__ptr += (__len & ~3);			\
-				SMC_outsw(ioaddr, DATA_REG(lp), __ptr, 1); \
+				SMC_outsw(lp->base, DATA_REG(lp), __ptr, 1); \
 			}						\
 		} else if (SMC_16BIT(lp))				\
-			SMC_outsw(ioaddr, DATA_REG(lp), p, (l) >> 1);	\
+			SMC_outsw(lp->base, DATA_REG(lp), p, (l) >> 1);	\
 		else if (SMC_8BIT(lp))				\
-			SMC_outsb(ioaddr, DATA_REG(lp), p, l);	\
+			SMC_outsb(lp->base, DATA_REG(lp), p, l);	\
 	} while (0)
 
 #define SMC_PULL_DATA(lp, p, l)					\
@@ -1080,9 +1080,9 @@ static const char * chip_ids[ 16 ] =  {
 			__len += 2;					\
 			SMC_insl(__ioaddr, DATA_REG(lp), __ptr, __len>>2); \
 		} else if (SMC_16BIT(lp))				\
-			SMC_insw(ioaddr, DATA_REG(lp), p, (l) >> 1);	\
+			SMC_insw(lp->base, DATA_REG(lp), p, (l) >> 1);	\
 		else if (SMC_8BIT(lp))				\
-			SMC_insb(ioaddr, DATA_REG(lp), p, l);		\
+			SMC_insb(lp->base, DATA_REG(lp), p, l);		\
 	} while (0)
 
 #endif  /* _SMC91X_H_ */
-- 
2.45.0
Re: [PATCH v2] net: smc91x: Fix pointer types
Posted by Andrew Lunn 1 year, 9 months ago
On Thu, May 16, 2024 at 05:56:12PM +0200, Thorsten Blum wrote:
> Use void __iomem pointers as parameters for mcf_insw() and mcf_outsw()
> to align with the parameter types of readw() and writew().
> 
> Use lp->base instead of ioaddr when calling SMC_outsw(), SMC_outsb(),
> SMC_insw(), and SMC_insb() to retain its type across macro boundaries
> and to fix the following warnings reported by kernel test robot:
> 
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
> drivers/net/ethernet/smsc/smc91x.c:483:17: sparse: warning: incorrect type in argument 1 (different address spaces)
> drivers/net/ethernet/smsc/smc91x.c:483:17: sparse:    expected void *a
> drivers/net/ethernet/smsc/smc91x.c:483:17: sparse:    got void [noderef] __iomem *
> 
> Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
> Reported-by: kernel test robot <lkp@intel.com>
> Closes: https://lore.kernel.org/oe-kbuild-all/202405160853.3qyaSj8w-lkp@intel.com/
> Acked-by: Nicolas Pitre <nico@fluxnic.net>

Reviewed-by: Andrew Lunn <andrew@lunn.ch>

You could add a follow up patch which removes the 
void __iomem *__ioaddr = ioaddr; lines and uses lp->base.
The code will then be more uniform.

	Andrew
Re: [PATCH v2] net: smc91x: Fix pointer types
Posted by Nicolas Pitre 1 year, 9 months ago
On Thu, 16 May 2024, Andrew Lunn wrote:

> You could add a follow up patch which removes the 
> void __iomem *__ioaddr = ioaddr; lines and uses lp->base.
> The code will then be more uniform.

Beware, It is sometimes overridden with __ioaddr = lp->datacs.


Nicolas
[PATCH v3] net: smc91x: Fix pointer types
Posted by Thorsten Blum 1 year, 9 months ago
Use void __iomem pointers as parameters for mcf_insw() and mcf_outsw()
to align with the parameter types of readw() and writew() to fix the
following warnings reported by kernel test robot:

drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
drivers/net/ethernet/smsc/smc91x.c:483:17: sparse: warning: incorrect type in argument 1 (different address spaces)
drivers/net/ethernet/smsc/smc91x.c:483:17: sparse:    expected void *a
drivers/net/ethernet/smsc/smc91x.c:483:17: sparse:    got void [noderef] __iomem *

Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202405160853.3qyaSj8w-lkp@intel.com/
Acked-by: Nicolas Pitre <nico@fluxnic.net>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
---
Changes in v2:
- Use lp->base instead of __ioaddr as suggested by Andrew Lunn. They are
 essentially the same, but using lp->base results in a smaller diff
- Remove whitespace only changes as suggested by Andrew Lunn
- Preserve Acked-by: Nicolas Pitre tag (please let me know if you
 somehow disagree with the changes in v2 or v3)

Changes in v3:
- Revert changing the macros as this is unnecessary. Neither the types
  nor the __iomem attributes get lost across macro boundaries
- Preserve Reviewed-by: Andrew Lunn tag (please let me know if you
  somehow disagree with the changes in v3)
---
 drivers/net/ethernet/smsc/smc91x.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/smsc/smc91x.h b/drivers/net/ethernet/smsc/smc91x.h
index 45ef5ac0788a..38aa4374e813 100644
--- a/drivers/net/ethernet/smsc/smc91x.h
+++ b/drivers/net/ethernet/smsc/smc91x.h
@@ -142,14 +142,14 @@ static inline void _SMC_outw_align4(u16 val, void __iomem *ioaddr, int reg,
 #define SMC_CAN_USE_32BIT	0
 #define SMC_NOWAIT		1
 
-static inline void mcf_insw(void *a, unsigned char *p, int l)
+static inline void mcf_insw(void __iomem *a, unsigned char *p, int l)
 {
 	u16 *wp = (u16 *) p;
 	while (l-- > 0)
 		*wp++ = readw(a);
 }
 
-static inline void mcf_outsw(void *a, unsigned char *p, int l)
+static inline void mcf_outsw(void __iomem *a, unsigned char *p, int l)
 {
 	u16 *wp = (u16 *) p;
 	while (l-- > 0)
-- 
2.45.0
Re: [PATCH v3] net: smc91x: Fix pointer types
Posted by Andrew Lunn 1 year, 9 months ago
On Fri, May 17, 2024 at 12:30:05AM +0200, Thorsten Blum wrote:
> Use void __iomem pointers as parameters for mcf_insw() and mcf_outsw()
> to align with the parameter types of readw() and writew() to fix the
> following warnings reported by kernel test robot:
> 
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
> drivers/net/ethernet/smsc/smc91x.c:483:17: sparse: warning: incorrect type in argument 1 (different address spaces)
> drivers/net/ethernet/smsc/smc91x.c:483:17: sparse:    expected void *a
> drivers/net/ethernet/smsc/smc91x.c:483:17: sparse:    got void [noderef] __iomem *
> 
> Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
> Reported-by: kernel test robot <lkp@intel.com>
> Closes: https://lore.kernel.org/oe-kbuild-all/202405160853.3qyaSj8w-lkp@intel.com/
> Acked-by: Nicolas Pitre <nico@fluxnic.net>
> Reviewed-by: Andrew Lunn <andrew@lunn.ch>
> ---
> Changes in v2:
> - Use lp->base instead of __ioaddr as suggested by Andrew Lunn. They are
>  essentially the same, but using lp->base results in a smaller diff
> - Remove whitespace only changes as suggested by Andrew Lunn
> - Preserve Acked-by: Nicolas Pitre tag (please let me know if you
>  somehow disagree with the changes in v2 or v3)
> 
> Changes in v3:
> - Revert changing the macros as this is unnecessary. Neither the types
>   nor the __iomem attributes get lost across macro boundaries
> - Preserve Reviewed-by: Andrew Lunn tag (please let me know if you
>   somehow disagree with the changes in v3)

This fixes the warning, but we still have the macro accessing things
not passed to them. If you are going to brother to fix the warnings,
it would also be good to fix the bad practice. Please make a patchset
to do this.

It would also be good if you read:

https://www.kernel.org/doc/html/latest/process/maintainer-netdev.html

	Andrew
Re: [PATCH v3] net: smc91x: Fix pointer types
Posted by Thorsten Blum 1 year, 9 months ago
On 17. May 2024, at 00:51, Andrew Lunn <andrew@lunn.ch> wrote:
> On Fri, May 17, 2024 at 12:30:05AM +0200, Thorsten Blum wrote:
>> Use void __iomem pointers as parameters for mcf_insw() and mcf_outsw()
>> to align with the parameter types of readw() and writew() to fix the
>> following warnings reported by kernel test robot:
>> 
>> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
>> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
>> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
>> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
>> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
>> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
>> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse: warning: incorrect type in argument 1 (different address spaces)
>> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    expected void *a
>> drivers/net/ethernet/smsc/smc91x.c:590:9: sparse:    got void [noderef] __iomem *
>> drivers/net/ethernet/smsc/smc91x.c:483:17: sparse: warning: incorrect type in argument 1 (different address spaces)
>> drivers/net/ethernet/smsc/smc91x.c:483:17: sparse:    expected void *a
>> drivers/net/ethernet/smsc/smc91x.c:483:17: sparse:    got void [noderef] __iomem *
>> 
>> Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
>> Reported-by: kernel test robot <lkp@intel.com>
>> Closes: https://lore.kernel.org/oe-kbuild-all/202405160853.3qyaSj8w-lkp@intel.com/
>> Acked-by: Nicolas Pitre <nico@fluxnic.net>
>> Reviewed-by: Andrew Lunn <andrew@lunn.ch>
>> ---
>> Changes in v2:
>> - Use lp->base instead of __ioaddr as suggested by Andrew Lunn. They are
>> essentially the same, but using lp->base results in a smaller diff
>> - Remove whitespace only changes as suggested by Andrew Lunn
>> - Preserve Acked-by: Nicolas Pitre tag (please let me know if you
>> somehow disagree with the changes in v2 or v3)
>> 
>> Changes in v3:
>> - Revert changing the macros as this is unnecessary. Neither the types
>>  nor the __iomem attributes get lost across macro boundaries
>> - Preserve Reviewed-by: Andrew Lunn tag (please let me know if you
>>  somehow disagree with the changes in v3)
> 
> This fixes the warning, but we still have the macro accessing things
> not passed to them. If you are going to brother to fix the warnings,
> it would also be good to fix the bad practice. Please make a patchset
> to do this.

I would prefer to submit another patch to fix the macros. I submitted v3
because the patch description for v2 was wrong (type information or
attributes don't get lost across macro boundaries) and the macro changes
are unnecessary to fix the warnings.

I should never have changed the macros, but after first adding __iomem
to mcf_insw() and mcf_outsw(), I kept getting the same errors and looked
for the problem in the SMC_* macros. I probably didn't do a clean build
or forgot to save my changes and just refactored the macros as a side
effect.

> It would also be good if you read:
> 
> https://www.kernel.org/doc/html/latest/process/maintainer-netdev.html

Will do.

Thanks,
Thorsten
Re: [PATCH v3] net: smc91x: Fix pointer types
Posted by Thorsten Blum 1 year, 8 months ago
On 17. May 2024, at 01:21, Thorsten Blum <thorsten.blum@toblux.com> wrote:
> On 17. May 2024, at 00:51, Andrew Lunn <andrew@lunn.ch> wrote:
>> It would also be good if you read:
>> 
>> https://www.kernel.org/doc/html/latest/process/maintainer-netdev.html
> 
> Will do.

Reading this was helpful and I learned that:
- net-next is closed during a merge window
- patches should be prefixed with the tree name

I'll submit the patch that cleans up ioaddr from all SMC_* macros when
net-next is open again.

Thanks,
Thorsten