On Fri, Dec 15, 2017 at 12:32 AM, Nicholas Piggin [off-list ref] wrote:

On Thu, 14 Dec 2017 23:16:26 +1100
Balbir Singh [off-list ref] wrote:

quoted

On Thu, Dec 14, 2017 at 12:51 PM, Nicholas Piggin [off-list ref] wrote:

quoted

quoted

quoted

I can't call smp_send_nmi_ipi due to the nmi_ipi_busy_count and
I'm worried about calling a stale nmi_ipi_function via the
system_reset_exception path, if we are OK with it, I can revisit
the code path

You shouldn't get a stale one, that would also be a bug -- we
have to cope with NMIs coming in at any time that are triggered
externally (not by smp_send_nmi_ipi), so if you see any bugs
there those need to be fixed separately.

Yes, I think it's a bug, nothing clears nmi_ipi_function (from what
I can see), so when the next NMI comes in and goes into
pnv_system_reset_exception
it'll execute the stale handler.

The CPU won't be in the nmi_ipi_pending_mask though, so it shouldn't
get that far. You could add a bit of paranoia to clear the function
pointer I suppose, but AFAIKS it's not needed.

Yep your right, but these things are so subtle :) I will as paranoia
cleanup nmi_ipi_function, but I'll add that as a TODO

Balbir Singh.