[Bug tree-optimization/100263] [11/12 Regression] Wrong removal of statement in copyprop3

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100263

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
And, just to verify it, I've compiled the testcase with r11-39 with a
breakpoint on pass_expand::execute and set global_options.x_optimize = 0 there,
and resulting testcase passes.
So most likely things go wrong during RTL optimizations.