arm64/io: add constant-argument check

In some configurations __const_iowrite32_copy() does not get inlined
and gcc runs into the BUILD_BUG():

In file included from <command-line>:
In function '__const_memcpy_toio_aligned32',
    inlined from '__const_iowrite32_copy' at arch/arm64/include/asm/io.h:203:3,
    inlined from '__const_iowrite32_copy' at arch/arm64/include/asm/io.h:199:20:
include/linux/compiler_types.h:487:45: error: call to '__compiletime_assert_538' declared with attribute error: BUILD_BUG failed
  487 |         _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__)
      |                                             ^
include/linux/compiler_types.h:468:25: note: in definition of macro '__compiletime_assert'
  468 |                         prefix ## suffix();                             \
      |                         ^~~~~~
include/linux/compiler_types.h:487:9: note: in expansion of macro '_compiletime_assert'
  487 |         _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__)
      |         ^~~~~~~~~~~~~~~~~~~
include/linux/build_bug.h:39:37: note: in expansion of macro 'compiletime_assert'
   39 | #define BUILD_BUG_ON_MSG(cond, msg) compiletime_assert(!(cond), msg)
      |                                     ^~~~~~~~~~~~~~~~~~
include/linux/build_bug.h:59:21: note: in expansion of macro 'BUILD_BUG_ON_MSG'
   59 | #define BUILD_BUG() BUILD_BUG_ON_MSG(1, "BUILD_BUG failed")
      |                     ^~~~~~~~~~~~~~~~
arch/arm64/include/asm/io.h:193:17: note: in expansion of macro 'BUILD_BUG'
  193 |                 BUILD_BUG();
      |                 ^~~~~~~~~

Move the check for constant arguments into the inline function to ensure
it is still constant if the compiler decides against inlining it, and
mark them as __always_inline to override the logic that sometimes leads
to the compiler not producing the simplified output.

Note that either the __always_inline annotation or the check for a
constant value are sufficient here, but combining the two looks cleaner
as it also avoids the macro. With clang-8 and older, the macro was still
needed, but all versions of gcc and clang can reliably perform constant
folding here.

Fixes: ead79118 ("arm64/io: Provide a WC friendly __iowriteXX_copy()")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/20240604210006.668912-1-arnd@kernel.orgSigned-off-by: Will Deacon <will@kernel.org>

arch/arm64/include/asm/io.h

@@ -153,8 +153,9 @@ extern void __memset_io(volatile void __iomem *, int, size_t);
  * emit the large TLP from the CPU.
  */
 
-static inline void __const_memcpy_toio_aligned32(volatile u32 __iomem *to,
-						 const u32 *from, size_t count)
+static __always_inline void
+__const_memcpy_toio_aligned32(volatile u32 __iomem *to, const u32 *from,
+			      size_t count)
 {
 	switch (count) {
 	case 8:
@@ -196,24 +197,22 @@ static inline void __const_memcpy_toio_aligned32(volatile u32 __iomem *to,
 
 void __iowrite32_copy_full(void __iomem *to, const void *from, size_t count);
 
-static inline void __const_iowrite32_copy(void __iomem *to, const void *from,
-					  size_t count)
+static __always_inline void
+__iowrite32_copy(void __iomem *to, const void *from, size_t count)
 {
-	if (count == 8 || count == 4 || count == 2 || count == 1) {
+	if (__builtin_constant_p(count) &&
+	    (count == 8 || count == 4 || count == 2 || count == 1)) {
 		__const_memcpy_toio_aligned32(to, from, count);
 		dgh();
 	} else {
 		__iowrite32_copy_full(to, from, count);
 	}
 }
+#define __iowrite32_copy __iowrite32_copy
 
-#define __iowrite32_copy(to, from, count)                  \
-	(__builtin_constant_p(count) ?                     \
-		 __const_iowrite32_copy(to, from, count) : \
-		 __iowrite32_copy_full(to, from, count))
-
-static inline void __const_memcpy_toio_aligned64(volatile u64 __iomem *to,
-						 const u64 *from, size_t count)
+static __always_inline void
+__const_memcpy_toio_aligned64(volatile u64 __iomem *to, const u64 *from,
+			      size_t count)
 {
 	switch (count) {
 	case 8:
@@ -255,21 +254,18 @@ static inline void __const_memcpy_toio_aligned64(volatile u64 __iomem *to,
 
 void __iowrite64_copy_full(void __iomem *to, const void *from, size_t count);
 
-static inline void __const_iowrite64_copy(void __iomem *to, const void *from,
-					  size_t count)
+static __always_inline void
+__iowrite64_copy(void __iomem *to, const void *from, size_t count)
 {
-	if (count == 8 || count == 4 || count == 2 || count == 1) {
+	if (__builtin_constant_p(count) &&
+	    (count == 8 || count == 4 || count == 2 || count == 1)) {
 		__const_memcpy_toio_aligned64(to, from, count);
 		dgh();
 	} else {
 		__iowrite64_copy_full(to, from, count);
 	}
 }
-
-#define __iowrite64_copy(to, from, count)                  \
-	(__builtin_constant_p(count) ?                     \
-		 __const_iowrite64_copy(to, from, count) : \
-		 __iowrite64_copy_full(to, from, count))
+#define __iowrite64_copy __iowrite64_copy
 
 /*
  * I/O memory mapping functions.