On Sat, Dec 19, 2009 at 10:43:31PM +0000, Richard Henderson wrote: > It has been compile tested, but nothing else. > > I'm moderately surprised at the implementation of brcond. It's > missing the BLTZ, BLEZ, BGTZ, BGEZ branches. It's could be > improved with the SLTI and SLTIU insns used here with setcond.
The MIPS TCG support is really new, and the goal was to make it working, not to get it as optimized as possible. Anyway I doubt it will make a big difference in speed. > Indeed, I suspect that a good implementation of brcond would be > to take care of the cases implementable directly by branches, > and then call into setcond to take care of the rest. I don't > plan to do that. > > > r~ > > > > Richard Henderson (3): > tcg: Add tcg_swap_cond. > tcg-mips: Implement setcond, setcond2. > tcg-mips: Implement movcond. > > tcg/mips/tcg-target.c | 296 > ++++++++++++++++++++++++++++++++++++++++++++++++- > tcg/tcg.h | 8 ++ > 2 files changed, 302 insertions(+), 2 deletions(-) > > > > -- Aurelien Jarno GPG: 1024D/F1BCDB73 aurel...@aurel32.net http://www.aurel32.net
