> This solution is silly for at least the following reasons: > > - If the async queue gets long, we'll run out of allocatable memory. > - We don't batch the TLB flushing or worker wakeups at all. > - We drop FPI flags and skip the pcplists. While fiddling with this code I just noticed that, in my enthusiasm for stripping the logic to the bare functional minimum, I went overboard and totally removed the logic to wake up the kworker. Anyway, this patch is stupid regardless for the reasons above, but just noting this down in case anyone does read it and wonder how it can possibly work. It just leaks memory. If anyone wants to try out this RFC and needs their system to stay alive for a while, let me know and I'll prepare a branch with this issue fixed.
