[Bug middle-end/122686] [16 Regression] 521.wrf_r build hang on GNR with -Ofast -march=native -flto after patch set r16-5247...r16-5259 in ranger during expand

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

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122686

--- Comment #25 from Richard Biener <rguenth at gcc dot gnu.org> ---
Created attachment 62894
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=62894&action=edit
LTO IR testcase

With the attached "testcase" you can reproduce the hang (and debug it) on trunk
around r16-5548-gb207d9d1836ea1 (at some point the LTO IL will become invalid).
To reproduce:

./lto1 -quiet -dumpbase ./wrf_r.ltrans113.ltrans -march=x86-64-v3 -g0 -Ofast
-Ofast -version -fno-openacc -fno-pie -fcf-protection=none -fno-openmp -fltrans
./wrf_r.ltrans113.o -o ./wrf_r.ltrans113.ltrans.s

and observe that ranger_cache::propagate_cache does not finish.