Le 03/08/2021 à 18:54, Nathan Lynch a écrit :
Laurent Dufour <lduf...@linux.ibm.com> writes:
V5:
- Rework code structure
- Reintroduce the capability to reuse other node's ids.
OK. While I preferred v4, where we would fail an add rather than allow
CPU IDs to appear to "travel" between nodes, this change is a net
improvement.
Reviewed-by: Nathan Lynch <nath...@linux.ibm.com>
Thanks Nathan,
Regarding the reuse of other nodes free CPU ids, with this patch the kernel does
it best to prevent that. Instead of failing adding new CPUs, I think it's better
to reuse free CPU ids of other nodes, otherwise, only a reboot would allow the
CPU adding operation to succeed.
Laurent.