I agree with you that rtl_verify_edges() shouldn't be changed. While I am not sure about purge_dead_edges() since it is not something one looks at when bringing a new flag. But it hasn't caused problems so far, so I can't necessitate changing it either...
On Tue, Jul 22, 2014 at 8:16 PM, Jeff Law <l...@redhat.com> wrote: > On 07/16/14 12:41, Yi Yang wrote: >> >> There are a few if statements in cfgrtl.c that are very fragile in the >> sense that introducing an irrelevant edge flag breaks things. >> >> This patch rewrites them to avoid such breakage. >> >> -- >> >> 2014-07-16 Yi Yang <ahyan...@google.com> >> >> gcc: >> * cfgrtl.c (rtl_verify_edges, purge_dead_edges): Rewrite certain >> if statements. > > One could easily argue that 'fragile' in this context is what we actually > want. Ie, if someone introduces a new edge flag, they ought to be looking > at how that impacts verification. > > I strongly prefer to have developers explicitly handle any new edge flags in > the verification code. If they are truly irrelevant for verification, it's > easy enough to filter them out in the appropriate verification routines. > > Jeff
