[Bug ada/81361] [8 regression] broken exception handling at -O2

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81361

--- Comment #10 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> (1) The failure reported in pr82141 does not need -O2:
> 
> ada/rts% /opt/gcc/build_w/gcc/xgcc -B/opt/gcc/build_w/gcc/ -c g-exptty.adb
> 
> raised RTSFIND.RE_NOT_AVAILABLE : rtsfind.adb:851

Yes, it does and compiling rtsfind.adb at -O1 is the workaround.

> (2) I see similar acats failures (see
> https://gcc.gnu.org/ml/gcc-testresults/2017-07/msg02582.html), IMO they are
> due to revision r249930 ("fix" for darwin_function_switched_text_sections).

AFAICS this patch is only a formattig fix...

> (3) The "reduced" test in pr82141 compiles
> 
> % /opt/gcc/build_w/gcc/xgcc -B/opt/gcc/build_w/gcc/ -c p.adb

Compile at -O1 and execute, then compile at -O2 and execute.