Skip to content

RFC Implementing self IPI with NMI destination

Cyrill Gorcunov lkml 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 RFC Latest2 patches 2 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/2x86, apic: Get rid of some IPI wrappersarch/x86/kernel/apic/apic_flat_64.c arch/x86/kernel/apic/bigsmp_32.c arch/x86/kernel/apic/es7000_32.c +2 more+12 -430 replies
2/2x86, apic: Allow to send self NMI ipiarch/x86/include/asm/ipi.h arch/x86/kernel/apic/ipi.c arch/x86/kernel/apic/probe_64.c +3 more+26 -50 replies

Reviews trailers per patch

No Reviewed-by, Acked-by or Tested-by trailers for v1 yet.

Read the v1 thread

Cover letter [rfc -tip 0/2] Implementing self IPI with NMI destination

Here is self-IPI with NMI destination implemented. Don was testing
perf NMI's and we found that we don't have a simple call for self-nmi IPI.

So there is two way of implementing such facility:

1) Make .send_IPI_self to check for NMI_VECTOR in arguments and convert this call internally to NMI IPI with self in destination.

2) Make a separate .send_IPI_self_nmi function and call it only if needed.

The series implements 1st choise but I tend to think that 2nd might
be more preferred since it would not bring overhead into .send_IPI_self and would make code a bit less messed.

Anyway, this series is not for merging yet but rather to collect
complains about this approach. And to be fair we seem don't have x2apic machine at moment so x2apic mode is not tested at all.

In short -- please check and review this two patches. Anything
is welcome ;)

Cyrill