On Wed, Dec 09, 2020 at 02:33:04PM +0800, Shung-Hsi Yu wrote:
> Hi Thomas,
>
> On Tue, Nov 10, 2020 at 09:56:27PM +0100, Thomas Gleixner wrote:
> > The real problem is irqbalanced aggressively exhausting the vector space
> > of a _whole_ socket to the point that there is not a single vector left
> > for serial. That's the problem you want to fix.
>
> I believe this warning also gets triggered even when there's _no_ vector
> exhaustion.
>
> This seem to happen when the IRQ's affinity mask is set (wrongly) to CPUs on
> a different NUMA node (e.g. cpumask_of_node(1) when the irqd->irq == 0).
typo ^^^^^^^^^^^^^^
Should be "affinity mask set to cpumask_of_node(1) when
irq_data_get_node(irqd) == 0".
Shung-Hsi
> $ lscpu
> ...
> NUMA node0 CPU(s): 0-25,52-77
> NUMA node1 CPU(s): 26-51,78-103
>
> $ cat /sys/kernel/debug/tracing/trace
> ...
> irqbalance-1994 [017] d... 74.912799: irq_matrix_alloc: bit=33 cpu=26
> online=1 avl=198 alloc=3 managed=1 online_maps=104 global_avl=20687,
> global_rsvd=341, total_alloc=217
> irqbalance-1994 [017] d... 74.912802: vector_alloc: irq=4 vector=33
> reserved=0 ret=0
> irqbalance-1994 [017] d... 74.912804: vector_update: irq=4 vector=33
> cpu=26 prev_vector=33 prev_cpu=7
> irqbalance-1994 [017] d... 74.912805: vector_config: irq=4 vector=33
> cpu=26 apicdest=0x00000040
> <idle>-0 [007] d.h. 74.970733: vector_free_moved: irq=4 cpu=7
> vector=33 is_managed=0
> <idle>-0 [007] d.h. 74.970738: irq_matrix_free: bit=33 cpu=7
> online=1 avl=200 alloc=1 managed=1 online_maps=104 global_avl=20687,
> global_rsvd=341, total_alloc=217
> ...
> (agetty)-3004 [047] d... 81.731231: vector_deactivate: irq=4
> is_managed=0 can_reserve=1 reserve=0
> (agetty)-3004 [047] d... 81.738035: vector_clear: irq=4 vector=33
> cpu=26 prev_vector=0 prev_cpu=7
> (agetty)-3004 [047] d... 81.738040: irq_matrix_free: bit=33 cpu=26
> online=1 avl=199 alloc=2 managed=1 online_maps=104 global_avl=20689,
> global_rsvd=341, total_alloc=215
> (agetty)-3004 [047] d... 81.738046: irq_matrix_reserve:
> online_maps=104 global_avl=20689, global_rsvd=342, total_alloc=215
> (agetty)-3004 [047] d... 81.766739: vector_reserve: irq=4 ret=0
> (agetty)-3004 [047] d... 81.766741: vector_config: irq=4 vector=239
> cpu=0 apicdest=0x00000000
> (agetty)-3004 [047] d... 81.777152: vector_activate: irq=4
> is_managed=0 can_reserve=1 reserve=0
> (agetty)-3004 [047] d... 81.777157: vector_alloc: irq=4 vector=0
> reserved=1 ret=-22
> ----------------------------------------> irq_matrix_alloc() failed with
> EINVAL because the cpumask
> passed in is empty, which is a
> result of affmask being
> (ff,ffffc000,000fffff,fc000000)
> and cpumask_of_node(node)
> being
>
> (00,00003fff,fff00000,03ffffff).
>
> (agetty)-3004 [047] d... 81.789349: irq_matrix_alloc: bit=33 cpu=1
> online=1 avl=199 alloc=2 managed=1 online_maps=104 global_avl=20688,
> global_rsvd=341, total_alloc=216
> (agetty)-3004 [047] d... 81.789351: vector_alloc: irq=4 vector=33
> reserved=1 ret=0
> (agetty)-3004 [047] d... 81.789353: vector_update: irq=4 vector=33
> cpu=1 prev_vector=0 prev_cpu=26
> (agetty)-3004 [047] d... 81.789355: vector_config: irq=4 vector=33
> cpu=1 apicdest=0x00000002
> ----------------------------------------> "irq 4: Affinity broken due to
> vector space exhaustion."
> warning shows up
>
> (agetty)-3004 [047] d... 81.900783: irq_matrix_alloc: bit=33 cpu=26
> online=1 avl=198 alloc=3 managed=1 online_maps=104 global_avl=20687,
> global_rsvd=341, total_alloc=217
> (agetty)-3004 [047] d... 82.053535: vector_alloc: irq=4 vector=33
> reserved=0 ret=0
> (agetty)-3004 [047] d... 82.053536: vector_update: irq=4 vector=33
> cpu=26 prev_vector=33 prev_cpu=1
> (agetty)-3004 [047] d... 82.053538: vector_config: irq=4 vector=33
> cpu=26 apicdest=0x00000040
