[RFC,v1,0/3] isolation: limit msix vectors based on housekeeping CPUs

Message

Nitesh Narayan Lal Sept. 9, 2020, 3:08 p.m. UTC

This is a follow-up posting for "[v1] i40e: limit the msix vectors based on
housekeeping CPUs" [1] (It took longer than expected for me to get back to
this).


Issue
=====
With the current implementation device drivers while creating their MSIX
vectors only takes num_online_cpus() into consideration which works quite well
for a non-RT environment, but in an RT environment that has a large number of
isolated CPUs and a very few housekeeping CPUs this could lead to a problem.
The problem will be triggered when something like tuned will try to move all
the IRQs from isolated CPUs to the limited number of housekeeping CPUs to
prevent interruptions for a latency sensitive workload that will be runing on
the isolated CPUs. This failure is caused because of the per CPU vector
limitation.


Proposed Fix
============
In this patch-set, the following changes are proposed:
- A generic API num_housekeeping_cpus() which is meant to return the available
  housekeeping CPUs in an environment with isolated CPUs and all online CPUs
  otherwise.
- i40e: Specifically for the i40e driver the num_online_cpus() used in 
  i40e_init_msix() to calculate numbers msix vectors is replaced with the above
  defined API. This is done to restrict the number of msix vectors for i40e in
  RT environments.
- pci_alloc_irq_vector(): With the help of num_housekeeping_cpus() the max_vecs
  passed in pci_alloc_irq_vector() is restricted only to the available
  housekeeping CPUs only in an environment that has isolated CPUs. However, if
  the min_vecs exceeds the num_housekeeping_cpus(), no change is made to make
  sure that a device initialization is not prevented due to lack of
  housekeeping CPUs.



Reproducing the Issue
=====================
I have triggered this issue on a setup that had a total of 72 cores among which
68 were isolated and only 4 were left for housekeeping tasks. I was using
tuned's realtime-virtual-host profile to configure the system. In this
scenario, Tuned reported the error message 'Failed to set SMP affinity of IRQ
xxx to '00000040,00000010,00000005': [Errno 28] No space left on the device'
for several IRQs in tuned.log due to the per CPU vector limit.


Testing
=======
Functionality:
- To test that the issue is resolved with i40e change I added a tracepoint
  in i40e_init_msix() to find the number of CPUs derived for vector creation
  with and without tuned's realtime-virtual-host profile. As per expectation
  with the profile applied I was only getting the number of housekeeping CPUs
  and all available CPUs without it.

Performance:
- To analyze the performance impact I have targetted the change introduced in 
  pci_alloc_irq_vectors() and compared the results against a vanilla kernel
  (5.9.0-rc3) results.

  Setup Information:
  + I had a couple of 24-core machines connected back to back via a couple of
    mlx5 NICs and I analyzed the average bitrate for server-client TCP and UDP
    transmission via iperf. 
  + To minimize the Bitrate variation of iperf TCP and UDP stream test I have
    applied the tuned's network-throughput profile and disabled HT.
 Test Information:
  + For the environment that had no isolated CPUs:
    I have tested with single stream and 24 streams (same as that of online
    CPUs).
  + For the environment that had 20 isolated CPUs:
    I have tested with single stream, 4 streams (same as that the number of
    housekeeping) and 24 streams (same as that of online CPUs).

 Results:
  # UDP Stream Test:
    + There was no degradation observed in UDP stream tests in both
      environments. (With isolated CPUs and without isolated CPUs after the
      introduction of the patches).
  # TCP Stream Test - No isolated CPUs:
    + No noticeable degradation was observed.
  # TCP Stream Test - With isolated CPUs:
    + Multiple Stream (4)  - Average degradation of around 5-6%
    + Multiple Stream (24) - Average degradation of around 2-3%
    + Single Stream        - Even on a vanilla kernel the Bitrate observed for
                             a TCP single stream test seem to vary
                             significantly across different runs (eg. the %
                             variation between the best and the worst case on
                             a vanilla kernel was around 8-10%). A similar
                             variation was observed with the kernel that
                             included my patches. No additional degradation
                             was observed.

Since the change specifically for pci_alloc_irq_vectors is going to impact
several drivers I have posted this patch-set as RFC. I would be happy to
perform more testing based on any suggestions or incorporate any comments to
ensure that the change is not breaking anything.

[1] https://lore.kernel.org/patchwork/patch/1256308/ 

Nitesh Narayan Lal (3):
  sched/isolation: API to get num of hosekeeping CPUs
  i40e: limit msix vectors based on housekeeping CPUs
  PCI: Limit pci_alloc_irq_vectors as per housekeeping CPUs

 drivers/net/ethernet/intel/i40e/i40e_main.c |  3 ++-
 include/linux/pci.h                         | 16 ++++++++++++++
 include/linux/sched/isolation.h             |  7 +++++++
 kernel/sched/isolation.c                    | 23 +++++++++++++++++++++
 4 files changed, 48 insertions(+), 1 deletion(-)