[PATCH -mm 2/6] powerpc: convert little-endian bitops macros to static inline functions

(This patch is intended to be folded into the patch in -mm:
powerpc-introduce-little-endian-bitops.patch)

The little-endian bitops on powerpc are written as preprocessor
macros with the cast to "unsigned long *".
This means that even non-pointers will be accepted without an error, and
that is a Very Bad Thing.

This converts the little-endian bitops macros to static inline functions
with proper prototypes.

Suggested-by: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <redacted>
Cc: linuxppc-dev@lists.ozlabs.org
---
 arch/powerpc/include/asm/bitops.h |   43 +++++++++++++++++++++++++------------
 1 files changed, 29 insertions(+), 14 deletions(-)

diff --git a/arch/powerpc/include/asm/bitops.h b/arch/powerpc/include/asm/bitops.h
index fe67024..2e56187 100644
--- a/arch/powerpc/include/asm/bitops.h
+++ b/arch/powerpc/include/asm/bitops.h

@@ -288,20 +288,35 @@ static __inline__ int test_bit_le(unsigned long nr,
 	return (tmp[nr >> 3] >> (nr & 7)) & 1;
 }
 
-#define __set_bit_le(nr, addr) \
-	__set_bit((nr) ^ BITOP_LE_SWIZZLE, (unsigned long *)(addr))
-#define __clear_bit_le(nr, addr) \
-	__clear_bit((nr) ^ BITOP_LE_SWIZZLE, (unsigned long *)(addr))
-
-#define test_and_set_bit_le(nr, addr) \
-	test_and_set_bit((nr) ^ BITOP_LE_SWIZZLE, (unsigned long *)(addr))
-#define test_and_clear_bit_le(nr, addr) \
-	test_and_clear_bit((nr) ^ BITOP_LE_SWIZZLE, (unsigned long *)(addr))
-
-#define __test_and_set_bit_le(nr, addr) \
-	__test_and_set_bit((nr) ^ BITOP_LE_SWIZZLE, (unsigned long *)(addr))
-#define __test_and_clear_bit_le(nr, addr) \
-	__test_and_clear_bit((nr) ^ BITOP_LE_SWIZZLE, (unsigned long *)(addr))
+static inline void __set_bit_le(int nr, void *addr)
+{
+	__set_bit(nr ^ BITOP_LE_SWIZZLE, addr);
+}
+
+static inline void __clear_bit_le(int nr, void *addr)
+{
+	__clear_bit(nr ^ BITOP_LE_SWIZZLE, addr);
+}
+
+static inline int test_and_set_bit_le(int nr, void *addr)
+{
+	return test_and_set_bit(nr ^ BITOP_LE_SWIZZLE, addr);
+}
+
+static inline int test_and_clear_bit_le(int nr, void *addr)
+{
+	return test_and_clear_bit(nr ^ BITOP_LE_SWIZZLE, addr);
+}
+
+static inline int __test_and_set_bit_le(int nr, void *addr)
+{
+	return __test_and_set_bit(nr ^ BITOP_LE_SWIZZLE, addr);
+}
+
+static inline int __test_and_clear_bit_le(int nr, void *addr)
+{
+	return __test_and_clear_bit(nr ^ BITOP_LE_SWIZZLE, addr);
+}
 
 #define find_first_zero_bit_le(addr, size) \
 	find_next_zero_bit_le((addr), (size), 0)

-- 
1.7.3.4

On Thu, 2011-01-27 at 22:56 +0900, Akinobu Mita wrote:

(This patch is intended to be folded into the patch in -mm:
powerpc-introduce-little-endian-bitops.patch)

The little-endian bitops on powerpc are written as preprocessor
macros with the cast to "unsigned long *".
This means that even non-pointers will be accepted without an error, and
that is a Very Bad Thing.

This converts the little-endian bitops macros to static inline functions
with proper prototypes.

No objection to the powerpc variant of the patches. What is the status
with the wholes series tho ? Does it looks like its going to be
accepted ? Do you expect my Ack and will merge the whole thing at once ?
Does it break bisection unless it's merged as one single giant patch ?

Cheers,
Ben.

2011/2/7 Benjamin Herrenschmidt [off-list ref]:

On Thu, 2011-01-27 at 22:56 +0900, Akinobu Mita wrote:

quoted

(This patch is intended to be folded into the patch in -mm:
powerpc-introduce-little-endian-bitops.patch)

The little-endian bitops on powerpc are written as preprocessor
macros with the cast to "unsigned long *".
This means that even non-pointers will be accepted without an error, and
that is a Very Bad Thing.

This converts the little-endian bitops macros to static inline functions
with proper prototypes.

No objection to the powerpc variant of the patches. What is the status
with the wholes series tho ? Does it looks like its going to be
accepted ? Do you expect my Ack and will merge the whole thing at once ?

The whole series now seems acceptable since I fixed two issues
that Linus found annoying. (the naming and the change of prototype)

Please give your ack if it is OK.

I should have fixed them quickly so that the series went upstream
in the last merge windows. But I couldn't because I spent some time
fixing and compile testing for a bisection hole.

Does it break bisection unless it's merged as one single giant patch ?

I think there is no known problem that breaks bisectability by
this patch series.