On Wed, Sep 22, 2021 at 09:47:59PM +0200, Peter Zijlstra wrote: > On Wed, Sep 22, 2021 at 09:33:43PM +0200, Peter Zijlstra wrote: > > > Anyway, lemme see if I get your proposal; lets say the counter starts at > > 0 and is in kernel space. > > > > 0x00(0) - kernel > > 0x02(2) - user > > 0x04(0) - kernel > > > > So far so simple, then NMI on top of that goes: > > > > 0x00(0) - kernel > > 0x03(3) - kernel + nmi > > 0x04(0) - kernel > > 0x06(2) - user > > 0x09(1) - user + nmi > > 0x0a(2) - user > > > > Which then gives us: > > > > (0) := kernel > > (1) := nmi-from-user > > (2) := user > > (3) := nmi-from-kernel > > > > Which should work I suppose. But like I said above, I'd be happier if > > this counter would live in context_tracking rather than RCU. > > Furthermore, if we have this counter, the we can also do things like: > > seq = context_tracking_seq_cpu(that_cpu); > if ((seq & 3) != USER) > // nohz_fail, do something > set_tsk_thread_flag(curr_task(that_cpu), TIF_DO_SOME_WORK); > if (seq == context_tracking_seq_cpu(that_cpu)) > // success!! > > To remotely set pending state. Allowing yet more NOHZ_FULL fixes, like, > for example, eliding the text_poke IPIs. Nice! There have been several instances where I thought that the extra state would help RCU, but each time there turned out to be a simpler way to get things done. Or that it eventually turned out that RCU didn't need to care about the difference between idle and nohz_full userspace. Thanx, Paul
