On Wed, Oct 20, 2021 at 09:57:58AM -1000, Linus Torvalds wrote: > On Wed, Oct 20, 2021 at 7:44 AM Eric W. Biederman <ebiederm@xxxxxxxxxxxx> wrote: > > > > + force_sig(SIGKILL); > > I wonder if SIGFPE would be a more intuitive thing. > > Doesn't really matter, this is a "doesn't happen" event anyway, but > that was just my reaction to reading the patch. I think SIGKILL makes more sense unless there's a way the process could handle the resulting SIGFPE and recover. I'd actually like to see the lazy allocation of FPU state just removed (the amount of space saved is tiny relative to the complexity cost and the negative aspects of unrecoverable late failure) but for now let's just go with this. Rich
