ACK: [SRU][B/F/H/I/J][PATCH 0/1] Fix for selftests/net/reuseport_bpf_numa (lp#1867570)

[Tim Gardner]

Acked-by: Tim Gardner <tim.gardner at canonical.com>

On 11/8/21 4:25 AM, Kleber Sacilotto de Souza wrote:
> BugLink: https://bugs.launchpad.net/bugs/1867570
> 
> [ Impact ]
> The reuseport_bpf_numa testcase in selftests/net fails on some PowerPC systems
> with:
> 
>   # selftests: net: reuseport_bpf_numa
>   # ---- IPv4 UDP ----
>   # send node 0, receive socket 0
>   # libnuma: Warning: Cannot read node cpumask from sysfs
>   # ./reuseport_bpf_numa: failed to pin to node: No such file or directory
>   not ok 3 selftests: net: reuseport_bpf_numa # exit=1
> 
> [ Fix ]
> In some platforms the numa node numbers are not necessarily consecutive, so when
> reuseport_bpf_numa iterates on the nodes from 0 to the max available node some
> of those will not be available. The fix is to change the testcase so it checks
> whether the node is available before trying to send/receive from it.
> 
> [ Test ]
> Run reuseport_bpf_numa from selftests/net in one of the affected test nodes.
> 
> [ Where Problems Could Occur ]
> This is a very simple fix which relies on the libnuma exported functions and
> bitmaps to check whether a numa node number is available on the system. If
> something is not right on how this check is made or on the libnuma code itself
> only the reuseport_bpf_numa testcase will be affected by reporting false
> negatives or having a reduced test coverage.
> 
> Kleber Sacilotto de Souza (1):
>    selftests/net: Fix reuseport_bpf_numa by skipping unavailable nodes
> 
>   tools/testing/selftests/net/reuseport_bpf_numa.c | 4 ++++
>   1 file changed, 4 insertions(+)
> 

-- 
-----------
Tim Gardner
Canonical, Inc