On Fri, May 17, 2019 at 6:42 PM Vladislav Ivanishin <v...@ispras.ru> wrote: > > Richard Biener <richard.guent...@gmail.com> writes: > > > On Tue, May 14, 2019 at 3:58 PM Vladislav Ivanishin <v...@ispras.ru> wrote: > >> > >> Hi! > >> > >> The split_critical_edges() function has multiple uses and it seems, a > >> portion of its code was added to work only when called from tree-ssa-pre > >> but right now it is executed regardless of the caller. > >> > >> The below patch survives bootstrap and regression testing on > >> x86_64-pc-linux-gnu. Does it make sense? > > > > Yeah. Please rename the in_pre_p parameter to for_edge_insertion_p > > since that is what it ensures. Note that the use in tree-ssa-sink.c > > also requires this since it looks for places to sink code to. Similar > > the use in tree-tail-merge.c (where I'm not sure why it needs split > > critical edges at all...). > > > > So, it seems more safe to have the default of the parameter to true, > > or rather have no default but adjust all few cases effectively only > > changing the one before uninit. > > > > Better (source) documentation would be using an overload, > > split_edges_for_insertion? > > Thanks for the feedback. > > Here is a safer version of the patch. > > > Bootstrap and regression tests are running. OK if they succeed?
OK. Thanks, Richard. > -- > Vlad
