[Bug ipa/58721] [4.9 Regression] The subroutine perdida is no longer inlined in fatigue.f90

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58721

--- Comment #8 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> ... bug was introduced by r202567 ...

I have checked that and fixing it with r205586 restores the timing of r202566.
The reason why this is not seen on recent revisions is r203167 which introduces
yet another parameter 'builtin-expect-probability' with a default value 90.
Setting this parameter to 92 allows the run time for fatigue2 to go from 78.2s
to 29.1s (26.6s with r202566).

It does not look reasonable to control the inlining with at least three
parameters: 'large-function-growth', 'max-inline-insns-auto', and
'builtin-expect-probability'. In top of that their documentation is scarce.