On Tue Dec 12 22:44:57 2023 +0000, Alexandre Julliard wrote:
When this happens, it can be received within the small range of
instructions where %fs has been restored but we have not yet switched to the syscall stack, or the other way around in the return path. Is there a reason that we can't do these things in the right order then?
It might not be possible to restore %fs *after* switching to the user stack as some of the return paths use `iret` to do that?