|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [Xen-devel] Support for guest per-vcpu hard and soft affinity masks
Dario,
Are the vcpu hard and soft affinity masks getting set for guest vcpus?
I'm running xen built from your numa/per-vcpu-affinity-v5 branch.
I added the following code to...
sched_credit.c in function csched_vcpu_insert
sched_credit2.c in function runq_assign
if ( cpumask_full(vc->cpu_hard_affinity) )
printk("JTW - hard affinity map IS full, dom %d\n",
vc->domain->domain_id);
else
printk("JTW - hard affinity map is NOT full, dom %d\n",
vc->domain->domain_id);
if ( cpumask_full(vc->cpu_soft_affinity) )
printk("JTW - soft affinity map IS full, dom %d\n",
vc->domain->domain_id);
else
printk("JTW - soft affinity map is NOT full, dom %d\n",
vc->domain->domain_id);
I started up two domains; one is pinned to 0-3 and the other used
automatic placement which xl vcpu-list says set its soft affinity to
8-15.
xl vcpu-list
------------
Name ID VCPU CPU State Time(s) Affinity (Hard / Soft)
VM_MatrixMult 1 0 3 -b- 4.2 0-3 / all
VM_Quicksort 2 0 1 -b- 4.1 all / 8-15
Got the same output running credit and credit2 ...
------
(XEN) JTW - hard affinity map IS full, dom 1
(XEN) JTW - soft affinity map IS full, dom 1
(XEN) JTW - hard affinity map IS full, dom 2
(XEN) JTW - soft affinity map IS full, dom 2
I expected the hard map for 1 to be not full, and the soft map for 2
to be not full.
If they are getting set later, when does that happen? Or am I using a
branch that doesn't yet have this functionality? Sorry if I missed
something in the documentation or patch descriptions.
Thank you,
Justin
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel
|
![]() |
Lists.xenproject.org is hosted with RackSpace, monitoring our |