byteorder: allow arch to opt to use GCC intrinsics for byteswapping

Since GCC 4.4, there have been __builtin_bswap32() and __builtin_bswap16()
intrinsics. A __builtin_bswap16() came a little later (4.6 for PowerPC,
48 for other platforms).

By using these instead of the inline assembler that most architectures
have in their __arch_swabXX() macros, we let the compiler see what's
actually happening. The resulting code should be at least as good, and
much *better* in the cases where it can be combined with a nearby load
or store, using a load-and-byteswap or store-and-byteswap instruction
(e.g. lwbrx/stwbrx on PowerPC, movbe on Atom).

When GCC is sufficiently recent *and* the architecture opts in to using
the intrinsics by setting CONFIG_ARCH_USE_BUILTIN_BSWAP, they will be
used in preference to the __arch_swabXX() macros. An architecture which
does not set ARCH_USE_BUILTIN_BSWAP will continue to use its own
hand-crafted macros.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Acked-by: H. Peter Anvin <hpa@linux.intel.com>

1 files changed, 19 insertions, 0 deletions

diff --git a/arch/Kconfig b/arch/Kconfig
index 366ec06a5185..c31416b10586 100644
--- a/arch/Kconfig
+++ b/arch/Kconfig
@@ -112,6 +112,25 @@ config HAVE_EFFICIENT_UNALIGNED_ACCESS
           See Documentation/unaligned-memory-access.txt for more
           information on the topic of unaligned memory accesses.
 
+config ARCH_USE_BUILTIN_BSWAP
+       bool
+       help
+         Modern versions of GCC (since 4.4) have builtin functions
+         for handling byte-swapping. Using these, instead of the old
+         inline assembler that the architecture code provides in the
+         __arch_bswapXX() macros, allows the compiler to see what's
+         happening and offers more opportunity for optimisation. In
+         particular, the compiler will be able to combine the byteswap
+         with a nearby load or store and use load-and-swap or
+         store-and-swap instructions if the architecture has them. It
+         should almost *never* result in code which is worse than the
+         hand-coded assembler in <asm/swab.h>.  But just in case it
+         does, the use of the builtins is optional.
+
+         Any architecture with load-and-swap or store-and-swap
+         instructions should set this. And it shouldn't hurt to set it
+         on architectures that don't have such instructions.
+
 config HAVE_SYSCALL_WRAPPERS
         bool