https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90458
--- Comment #2 from Jeffrey A. Law <law at redhat dot com> --- So this is an interaction between stack-clash protection and SEH. I'm not *at all* familiar with SEH, though obviously I know a bit about stack clash. In general on x86 the compiler handles stack allocation (and probing when stack clash protection is enabled). However, on Windows targets that stuff is actually handled by calls to __chkstk_ms. One could easily argue that stack-clash-protection should be a NOP or generate an error in a Windows environment.
