[Bug target/84521] aarch64: Frame-pointer corruption with __builtin_setjmp/__builtin_longjmp and -fomit-frame-pointer

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84521

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P1                          |P3
            Summary|[8 Regression] aarch64:     |aarch64: Frame-pointer
                   |Frame-pointer corruption    |corruption with
                   |with                        |__builtin_setjmp/__builtin_
                   |__builtin_setjmp/__builtin_ |longjmp and
                   |longjmp and                 |-fomit-frame-pointer
                   |-fomit-frame-pointer        |

--- Comment #24 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
With the change of the default this actually isn't a regression anymore, so not
a release blocker.  That doesn't mean a simple fix, e.g. __builtin_setjmp
forcing frame pointer even if not otherwise needed, can't be accepted for GCC8
(as an exception), but it is unlikely right time for finding all issues with
the builtin and trying to fix all of them.  That can be done in GCC9.