x86: Correct behaviour of irq affinity
authorRusty Russell <rusty@rustcorp.com.au>
Tue, 24 Mar 2009 21:17:19 +0000 (14:17 -0700)
committerIngo Molnar <mingo@elte.hu>
Wed, 25 Mar 2009 17:48:29 +0000 (18:48 +0100)
commite06b1b56f9bfcc91e1f175fe8d8bf3e35dafa080
tree7bb1a3a9e38bb8af4ead2177573c17356e16563d
parentf56e5034121c4911a155ba907076ab920754626d
x86: Correct behaviour of irq affinity

Impact: get correct smp_affinity as user requested

The effect of setting desc->affinity (ie. from userspace via sysfs) has
varied over time.  In 2.6.27, the 32-bit code anded the value with
cpu_online_map, and both 32 and 64-bit did that anding whenever a cpu
was unplugged.

2.6.29 consolidated this into one routine (and fixed hotplug) but
introduced another variation: anding the affinity with cfg->domain.

We should just set it to what the user said - if possible.

(cpu_mask_to_apicid_and already takes cpu_online_mask into account)

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
LKML-Reference: <49C94DDF.2010703@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
arch/x86/kernel/apic/io_apic.c