[Bug tree-optimization/69376] [6 Regression] wrong code at -Os and above on x86_64-linux-gnu

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69376

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #5)
> Created attachment 37433 [details]
> patch
> 
> Patch that I have tested but which exposes another latent issue?

That patch LGTM.

> FAIL:
> 21_strings/basic_string/requirements/exception/generation_prohibited.cc ex
> ecution test
> FAIL: ext/vstring/requirements/exception/generation_prohibited.cc execution
> test
> 
> 
> on x86_64 both -m32 and -m64 with allocating gigabytes of memory and
> eventually throwing bad_alloc ().  -fno-tree-pre recovers from this, so does
> -fno-tree-vrp
> (as expected), so does disabling only vrp2 (or vrp1).

Are you going to debug those, or shall I have a look?