On Mon, 2012-09-24 at 09:15 +0200, Eric Botcazou wrote: > > A while ago I've tried doing that for SH (implementing bswap16 with RTL > > combine). It was like an explosion of patterns, because combine would > > try out a lot of things depending on the surrounding code around the > > actual bswap16. In the end I decided to drop that stuff for the most > > part. > > OK, thanks for your feedback. Note that you won't get the benefit of the > optimization for SH if you don't define a bswaphi2 pattern.
Sure sure. I'll resume my efforts after Christophe's patch is in. > > > BTW, the built-in documentation says: > > > > Built-in Function: int16_t __builtin_bswap16 (int16_t x) > > Built-in Function: int32_t __builtin_bswap32 (int32_t x) > > > > However, it seems the result is always unsigned for those. > > Yes, builtins.def contains: > > DEF_GCC_BUILTIN (BUILT_IN_BSWAP16, "bswap16", BT_FN_UINT16_UINT16, > ATTR_CONST_NOTHROW_LEAF_LIST) > DEF_GCC_BUILTIN (BUILT_IN_BSWAP32, "bswap32", BT_FN_UINT32_UINT32, > ATTR_CONST_NOTHROW_LEAF_LIST) > DEF_GCC_BUILTIN (BUILT_IN_BSWAP64, "bswap64", BT_FN_UINT64_UINT64, > ATTR_CONST_NOTHROW_LEAF_LIST) > > The documentation indeed needs to be fixed. > OK, I'll check it out. Cheers, Oleg
