On Wed, Aug 16, 2017 at 8:47 AM, Dibyendu Majumdar <mobile@xxxxxxxxxxxxxxx> wrote: > > That is only if you want a minimal SSA right? Do you need one? Again I > suggest you try out Luc's implementation which appears to work fine - > and it doesn't do this extra pass I believe. If it does not do this extra pass then it will leave redundant phi node around. Base on my understanding of the paper. Yes, I would like to have minimal phi node SSA. Other than the code size, the extra redundant phi node will have impact on later stage of the optimization. It make the define and usage chain less clear, having the middle man phi node might or might not be needed. Chris -- To unsubscribe from this list: send the line "unsubscribe linux-sparse" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
