On Tue, 17 Mar 2026 at 21:09, Joseph Myers <[email protected]> wrote: > > On Tue, 17 Mar 2026, Jonathan Wakely wrote: > > > Anyway, I've reverted the patch now. We will have to figure out how to > > deal with funky targets. If va_list does not have external linkage on > > those targets, then it's already undefined to use it in C++, so really > > that needs to be fixed in those targets. Or we hack workarounds into > > the C++ front end for these system headers. > > Maybe the targets need to set up __va_list_tag in some different way then, > since i386 / rs6000 / s390 also have __va_list_tag but don't seem to have > this issue (and I don't know whether stormy16 or xtensa have it or not > since glibc doesn't support those architectures).
I opened these for the va_list definitions on those targets: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124565 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124566
