[PATCH 02/17] mips: namespace pollution - mem_... -> __mem_... in io.h

[PATCH 02/17] mips: namespace pollution - mem_... -> __mem_... in io.h

[Al Viro]

Date: 1134015174 -0500

A pile of internal functions use only inside mips io.h has names starting
with mem_... and clashing with names in drivers; renamed to __mem_....

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

---

 include/asm-mips/io.h |   32 ++++++++++++++++----------------
 1 files changed, 16 insertions(+), 16 deletions(-)

290f10ae4230ef06b71e57673101b7e70c1b29a6
diff --git a/include/asm-mips/io.h b/include/asm-mips/io.h
index a9fa125..6c0aae5 100644
--- a/include/asm-mips/io.h
+++ b/include/asm-mips/io.h
@@ -56,38 +56,38 @@
  * variations of functions: non-prefixed ones that preserve the value
  * and prefixed ones that preserve byte addresses.  The latters are
  * typically used for moving raw data between a peripheral and memory (cf.
- * string I/O functions), hence the "mem_" prefix.
+ * string I/O functions), hence the "__mem_" prefix.
  */
 #if defined(CONFIG_SWAP_IO_SPACE)
 
 # define ioswabb(x)		(x)
-# define mem_ioswabb(x)		(x)
+# define __mem_ioswabb(x)	(x)
 # ifdef CONFIG_SGI_IP22
 /*
  * IP22 seems braindead enough to swap 16bits values in hardware, but
  * not 32bits.  Go figure... Can't tell without documentation.
  */
 #  define ioswabw(x)		(x)
-#  define mem_ioswabw(x)	le16_to_cpu(x)
+#  define __mem_ioswabw(x)	le16_to_cpu(x)
 # else
 #  define ioswabw(x)		le16_to_cpu(x)
-#  define mem_ioswabw(x)	(x)
+#  define __mem_ioswabw(x)	(x)
 # endif
 # define ioswabl(x)		le32_to_cpu(x)
-# define mem_ioswabl(x)		(x)
+# define __mem_ioswabl(x)	(x)
 # define ioswabq(x)		le64_to_cpu(x)
-# define mem_ioswabq(x)		(x)
+# define __mem_ioswabq(x)	(x)
 
 #else
 
 # define ioswabb(x)		(x)
-# define mem_ioswabb(x)		(x)
+# define __mem_ioswabb(x)	(x)
 # define ioswabw(x)		(x)
-# define mem_ioswabw(x)		cpu_to_le16(x)
+# define __mem_ioswabw(x)	cpu_to_le16(x)
 # define ioswabl(x)		(x)
-# define mem_ioswabl(x)		cpu_to_le32(x)
+# define __mem_ioswabl(x)	cpu_to_le32(x)
 # define ioswabq(x)		(x)
-# define mem_ioswabq(x)		cpu_to_le32(x)
+# define __mem_ioswabq(x)	cpu_to_le32(x)
 
 #endif
 
@@ -417,7 +417,7 @@ __BUILD_MEMORY_SINGLE(bus, bwlq, type, 1
 									\
 __BUILD_MEMORY_PFX(__raw_, bwlq, type)					\
 __BUILD_MEMORY_PFX(, bwlq, type)					\
-__BUILD_MEMORY_PFX(mem_, bwlq, type)					\
+__BUILD_MEMORY_PFX(__mem_, bwlq, type)					\
 
 BUILDIO_MEM(b, u8)
 BUILDIO_MEM(w, u16)
@@ -430,7 +430,7 @@ BUILDIO_MEM(q, u64)
 
 #define BUILDIO_IOPORT(bwlq, type)					\
 	__BUILD_IOPORT_PFX(, bwlq, type)				\
-	__BUILD_IOPORT_PFX(mem_, bwlq, type)
+	__BUILD_IOPORT_PFX(__mem_, bwlq, type)
 
 BUILDIO_IOPORT(b, u8)
 BUILDIO_IOPORT(w, u16)
@@ -464,7 +464,7 @@ static inline void writes##bwlq(volatile
 	const volatile type *__addr = addr;				\
 									\
 	while (count--) {						\
-		mem_write##bwlq(*__addr, mem);				\
+		__mem_write##bwlq(*__addr, mem);			\
 		__addr++;						\
 	}								\
 }									\
@@ -475,7 +475,7 @@ static inline void reads##bwlq(volatile 
 	volatile type *__addr = addr;					\
 									\
 	while (count--) {						\
-		*__addr = mem_read##bwlq(mem);				\
+		*__addr = __mem_read##bwlq(mem);			\
 		__addr++;						\
 	}								\
 }
@@ -488,7 +488,7 @@ static inline void outs##bwlq(unsigned l
 	const volatile type *__addr = addr;				\
 									\
 	while (count--) {						\
-		mem_out##bwlq(*__addr, port);				\
+		__mem_out##bwlq(*__addr, port);				\
 		__addr++;						\
 	}								\
 }									\
@@ -499,7 +499,7 @@ static inline void ins##bwlq(unsigned lo
 	volatile type *__addr = addr;					\
 									\
 	while (count--) {						\
-		*__addr = mem_in##bwlq(port);				\
+		*__addr = __mem_in##bwlq(port);				\
 		__addr++;						\
 	}								\
 }
-- 
0.99.9.GIT

Re: [PATCH 02/17] mips: namespace pollution - mem_... -> __mem_... in io.h

[Maciej W. Rozycki]

On Wed, 8 Feb 2006, Al Viro wrote:

> A pile of internal functions use only inside mips io.h has names starting
> with mem_... and clashing with names in drivers; renamed to __mem_....

 Then the corresponding ones with no "mem_" prefix (these for the PCI I/O 
port space) should be prefixed with "__" for consistency as well.

  Maciej

Re: [PATCH 02/17] mips: namespace pollution - mem_... -> __mem_... in io.h

[Al Viro]

On Wed, Feb 08, 2006 at 11:01:52AM +0000, Maciej W. Rozycki wrote:
> On Wed, 8 Feb 2006, Al Viro wrote:
> 
> > A pile of internal functions use only inside mips io.h has names starting
> > with mem_... and clashing with names in drivers; renamed to __mem_....
> 
>  Then the corresponding ones with no "mem_" prefix (these for the PCI I/O 
> port space) should be prefixed with "__" for consistency as well.

Huh???

Things like outb(), etc. *are* public; mem_... ones are not. 

Re: [PATCH 02/17] mips: namespace pollution - mem_... -> __mem_... in io.h

[Maciej W. Rozycki]

On Wed, 8 Feb 2006, Al Viro wrote:

> >  Then the corresponding ones with no "mem_" prefix (these for the PCI I/O 
> > port space) should be prefixed with "__" for consistency as well.
> 
> Huh???
> 
> Things like outb(), etc. *are* public; mem_... ones are not. 

 I mean if we rename e.g. mem_ioswabb() to __mem_ioswabb(), then we should 
rename ioswabb() to __ioswabb() as well.  Sorry for not having been clear 
enough, but I have assumed it is obvious.

  Maciej

Re: [PATCH 02/17] mips: namespace pollution - mem_... -> __mem_... in io.h

[Al Viro]

On Wed, Feb 08, 2006 at 04:21:45PM +0000, Maciej W. Rozycki wrote:
> On Wed, 8 Feb 2006, Al Viro wrote:
> 
> > >  Then the corresponding ones with no "mem_" prefix (these for the PCI I/O 
> > > port space) should be prefixed with "__" for consistency as well.
> > 
> > Huh???
> > 
> > Things like outb(), etc. *are* public; mem_... ones are not. 
> 
>  I mean if we rename e.g. mem_ioswabb() to __mem_ioswabb(), then we should 
> rename ioswabb() to __ioswabb() as well.  Sorry for not having been clear 
> enough, but I have assumed it is obvious.

In principle that would be nice, but...  Take a look at those macros.
We can do that, but it would mean #define readb __readb, etc.  Since
really nasty clashes are in mem_inb() et.al. (let's face it, coming up
with one of those is far more likely than using ioswabb() for driver-internal
purposes) I've stopped at that.  Can do a followup switching to __ioswab...
and adding defines compensating for changes in visible symbols, but IMO
that's a separate patch...