[Bug rtl-optimization/113390] [14/15/16 Regression] ICE: in model_update_limit_points_in_group, at haifa-sched.cc:1986 with -O2 --param=max-sched-region-insns=200 --param=max-sched-extend-regions-iters=2

[avinashd at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113390

Avinash Jayakar <avinashd at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |avinashd at gcc dot gnu.org

--- Comment #9 from Avinash Jayakar <avinashd at gcc dot gnu.org> ---
Similar issue happened in rs6000 backend as reported in PR123144. I see some
problem in the model recomputation. 
The model_schedule has instruction only within the given basic block, but the
ready list may contain inter block instructions. Due to this there is a problem
in bookkeeping of the register pressure data, where the max_pressure drops
below zero (the register is marked as dead by seeing the instruction that is
not there in the model).
Should the register pressures be updated even if the insn is not in the model?