| Message ID | 1483569671-1462-3-git-send-email-keith.busch@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Wed, Jan 04, 2017 at 05:41:07PM -0500, Keith Busch wrote: > The offline CPUs need to assigned to something incase they come online > later, otherwise anyone using the mapping for things other than affinity > will have blank entries for that online CPU. I don't really like the idea behind it. Back when we came up with this code I had some discussion with Thomas if we should do the assignment only for online CPUs, or maybe for all possible CPUs. Except for some big iron with physical node hotplug the difference usually is just that some nodes have been temporarily offlined. So maybe we need to bite the bullet and move the irq and block code to consider all possible cpus. -- To unsubscribe from this list: send the line "unsubscribe linux-block" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
>> The offline CPUs need to assigned to something incase they come online >> later, otherwise anyone using the mapping for things other than affinity >> will have blank entries for that online CPU. > > I don't really like the idea behind it. Back when we came up with > this code I had some discussion with Thomas if we should do the > assignment only for online CPUs, or maybe for all possible CPUs. > > Except for some big iron with physical node hotplug the difference > usually is just that some nodes have been temporarily offlined. So > maybe we need to bite the bullet and move the irq and block code > to consider all possible cpus. I tend to agree. Would be great if we could have a cpuhp_register() for external users that would get ops + state mask structure to act on (this case would need CPUHP_OFFLINE, CPUHP_ONLINE). We do have something similar for mmu (mmu_notifier). -- To unsubscribe from this list: send the line "unsubscribe linux-block" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/kernel/irq/affinity.c b/kernel/irq/affinity.c index b25dce0..2367531 100644 --- a/kernel/irq/affinity.c +++ b/kernel/irq/affinity.c @@ -129,6 +129,14 @@ irq_create_affinity_masks(int nvecs, const struct irq_affinity *affd) } done: + /* + * Assign offline CPUs to first mask in case they come online later. A + * driver can rerun this from a cpu notifier if they want a more optimal + * spread. + */ + cpumask_andnot(nmsk, cpu_possible_mask, cpu_online_mask); + irq_spread_init_one(masks, nmsk, cpumask_weight(nmsk)); + put_online_cpus(); /* Fill out vectors at the end that don't need affinity */
The offline CPUs need to assigned to something incase they come online later, otherwise anyone using the mapping for things other than affinity will have blank entries for that online CPU. Signed-off-by: Keith Busch <keith.busch@intel.com> --- kernel/irq/affinity.c | 8 ++++++++ 1 file changed, 8 insertions(+)