xdp-dump: Use perf-ring bulking to speedup performance about xdp-tools

[netoptimizer]

Maybe @olsajiri or @acmel can help us with the C-code API for flushing/emptying the perf-ring buffer on exit?

[acmel]

Em Mon, May 25, 2020 at 05:06:28AM -0700, Jesper Dangaard Brouer escreveu:

Maybe @olsajiri or @acmel can help us with the C-code API for flushing/emptying the perf-ring buffer on exit?

Have you looked at: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/tools/lib/perf/Documentation/examples/sampling.c Its part of libperf, Jiri even added documentation at: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/tools/lib/perf/Documentation/ perf itself can be used as an example on using it, with 'tools/perf/builtin-record.c' just consuming the buffers as fast as possible to dump it into a perf.data file, while 'builtin-top.c' and 'builtin-trace.c' will parse the events and act on them as they are read from the mmap ring buffer. - Arnaldo

…

-- You are receiving this because you were mentioned. Reply to this email directly or view it on GitHub: #28 (comment)

-- - Arnaldo

[chaudron]

@acmel thanks for adding the above, I did not notice it until I sort of solved it for my implementation.

xdpdump is using libbpf's perf_buffer__xxx() APIs for doing this, which does not use the libperf. To keep using this API I'll will sent a patch upstream to just flush/process the buffers without waiting for a poll wake up.

[olsajiri]

from quick look I did not see ring buffer 'flush' interface in libbpf perf buffer API,
you might need to add it.. or switch to libperf ;-)

[chaudron]

from quick look I did not see ring buffer 'flush' interface in libbpf perf buffer API,
you might need to add it.. or switch to libperf ;-)

Just did https://lore.kernel.org/bpf/159048487929.89441.7465713173442594608.stgit@ebuild/T/#u
maybe we can switch later...

[tohojo]

Let's have this depend on #42 for backwards compatibility