On 11/15/2012 04:43 AM, Eric Blake wrote:
The assumption here is that any kernel lacking cpu/present also lacks
hotplug, and therefore the cpuNN will be consecutive and we aren't going
to miss anything. You removed that assumption from the comment, but I
think it is important to leave in.
ACK. I'm firing up my RHEL 5 VM to test this before I push with this
comment change, but it looks sane.
FWIW: I also run a 64-bit CentOS 5.8 and it supports hotplugging
despite the lack of cpu/present. Only cpu/cpu0 cannot be
hotplugged (no cpu/cpu0/online pseudofile).
Still the cpu/cpuNN directories are consecutive as they are present
even if the respective CPUs are offline.
--
Mit freundlichen Grüßen/Kind Regards
Viktor Mihajlovski
IBM Deutschland Research & Development GmbH
Vorsitzender des Aufsichtsrats: Martin Jetter
Geschäftsführung: Dirk Wittkopp
Sitz der Gesellschaft: Böblingen
Registergericht: Amtsgericht Stuttgart, HRB 243294