[Bug rtl-optimization/93264] [10 Regression] ICE in cfg_layout_redirect_edge_and_branch_force, at cfgrtl.c:4522

[rsandifo at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93264

rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> changed:

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

--- Comment #4 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #3)
> In general it's a bad idea to try go "back" to CFG layout mode and the fix is
> to not do that.  Which probably means scheduling pass_sms earlier and indeed
> then best before pass_partition_blocks.  I don't see any reason to _not_ do
> this, the current pipeline is
> 
>       NEXT_PASS (pass_partition_blocks);
>       NEXT_PASS (pass_outof_cfg_layout_mode);
>       NEXT_PASS (pass_split_all_insns);
>       NEXT_PASS (pass_lower_subreg3);
>       NEXT_PASS (pass_df_initialize_no_opt);
>       NEXT_PASS (pass_stack_ptr_mod);
>       NEXT_PASS (pass_mode_switching);
>       NEXT_PASS (pass_match_asm_constraints);
>       NEXT_PASS (pass_sms);
>       NEXT_PASS (pass_live_range_shrinkage);
>       NEXT_PASS (pass_sched);
>       NEXT_PASS (pass_early_remat);
>       NEXT_PASS (pass_ira);
>       NEXT_PASS (pass_reload);
>       NEXT_PASS (pass_postreload);
>       PUSH_INSERT_PASSES_WITHIN (pass_postreload)
> 
> and while I understand sms wants to run as close to RA as possible the
> passes between its current position and partitioning should not mess
> with the schedule (mode switching might insert code I think).

Yeah, it ought to be better to do mode switching first.  But I think
the more important ones are:

       NEXT_PASS (pass_split_all_insns);
       NEXT_PASS (pass_lower_subreg3);

Scheduling should happen on the split form of insns rather than the
unsplit form.  lower_subreg should also improve "schedulability".