A BPF_F_CPU update that creates a [lru_]percpu_hash element writes the
named CPU's slot and leaves the others holding the recycled element's
values, so a lookup of the new key returns a deleted key's per-cpu
values.

Patch 1 zero-fills the other CPUs.  Patch 2 adds the selftest: the
existing cpu_flag subtests always prime a key with BPF_F_ALL_CPUS
first, so the create path is not covered today.

v1: https://lore.kernel.org/bpf/20260920093153.439743-1-donggeunyoo.kernel@gmail.com/
v2: https://lore.kernel.org/bpf/20260923000801.1764758-1-donggeunyoo.kernel@gmail.com/

Changes in v3:
 - patch 1: key init_cpu on BPF_F_CPU rather than on onallcpus (Leon
   Hwang)
 - patch 1: re-flow the paragraph naming pcpu_copy_value() (BPF CI AI
   review)
 - patch 2: pin the thread across the delete and the create, and name a
   CPU other than the pinned one, so the BPF_F_NO_PREALLOC arm does not
   depend on staying put (BPF CI AI review)

Changes in v2:
 - patch 1: cover the BPF_F_CPU entry condition in the block comment
   above pcpu_init_value() (BPF CI AI review, Alexei Starovoitov)
 - patch 2: skip the new subtests instead of failing them on a
   uniprocessor machine (Sashiko AI review)

test_progs -t percpu_alloc and fourteen neighboring map tests, x86_64
under QEMU/KVM, 4 vCPUs:

  without patch 1   31/112 PASSED, 1/3 FAILED
  with patch 1      32/115 PASSED, 0/0 FAILED

Donggeun Yoo (2):
  bpf: Zero-fill other CPUs when BPF_F_CPU creates a per-cpu hash
    element
  selftests/bpf: Test per-cpu initialization of a BPF_F_CPU created
    element

 kernel/bpf/hashtab.c                          |  11 +-
 .../selftests/bpf/prog_tests/percpu_alloc.c   | 105 ++++++++++++++++++
 2 files changed, 112 insertions(+), 4 deletions(-)

-- 
2.53.0