https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81842
--- Comment #15 from Florian Weimer <fw at gcc dot gnu.org> --- This is all very strange. How have extended makecontext for x86 AVX2/AVX-512 support? The CPU context needs to be stored somewhere, after all. I find it difficult to believe that there is no space left anyway to store the shadow stack pointer information.