Re: [PATCH bpf-next v3 0/2] bpftool: Consume ring buffer maps with event_pipe

[Tianyi Chen <diannaaav@gmail.com>]

Hi CI team,

Could you rerun the failing x86_64 GCC 15 configuration for series
1162498 and, if it fails again, compare the unpatched baseline?

The only failing subtest was rhash/test_rhash_iter: key_sum was 2553
instead of 2080, and elem_count was 78 instead of 64. All 17
bpftool_ringbuf subtests passed in that same job:
https://github.com/kernel-patches/bpf/actions/runs/34557186977/job/103134336101

The x86_64 GCC 15 ASAN job passed both RHASH and ring-buffer tests:
https://github.com/kernel-patches/bpf/actions/runs/34557186977/job/103134793523

This series changes bpftool, documentation and its selftests; it does
not change the RHASH implementation or tests. For a local baseline
check, I built test_progs from unpatched bpf-next and ran
./test_progs -t rhash/test_rhash_iter 200 times. The selected subtest
passed all 200 runs without skips. This used an x86-64 KVM guest with
a bpf/master Linux 7.3-rc2 kernel built with GCC 13 and LLVM 20 BPF
objects, so it does not reproduce the CI toolchain or configuration.
The RHASH implementation and two relevant test files match the CI
pre-merge baseline.

Source inspection suggests deferred table resizing could lead the
iterator to revisit entries, but I have not established that as this
failure's cause. A rerun and baseline comparison would help distinguish
an intermittent RHASH issue from a regression in this series.

Thanks,
Tianyi