Re: [PATCH v1 bpf-next 5/8] selftests/bpf: Factor out load_path and defines from test_sock_addr

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



> The newer prog_tests/sock_addr.c was created when adding AF_UNIX support. It has
> a very similar setup as the older test_sock_addr.c and the intention was to
> finally retire test_sock_addr.c. e.g. It also has "load_fn loadfn" but is done
> with skeleton, the program is also attached to cgroup...etc.

Thanks, I wasn't sure what the preferred direction would be here. I
saw some discussion
in the AF_UNIX support patch series about prog_tests being the
preferred approach and
there being plans to migrate the old test_sock_addr.c tests to the new format.

I suspect the changes mentioned here (load done with skeleton, etc.)
and helpers needed
to support the tests in my patch series may also be helpful in
retiring test_sock_addr.c.

> Instead of adding a new sock_addr_kern.c in patch 7, it probably will be easier
> to add the kernel socket tests into the existing prog_tests/sock_addr.c.

Originally it felt awkward to try to merge these tests with
prog_tests/sock_addr.c, because
while the goals are similar the flow is different with
loading/unloading a kernel module to
drive socket operations. I'll revisit this and see if I can merge
everything into
prog_tests/sock_addr.c. Switching to kfuncs as mentioned earlier might
make this simpler
as well.

> Also setup the netns and veth in the prog_tests/sock_addr.c instead of calling
> out the test_sock_addr.sh (which should also go away eventually), there are
> examples in prog_tests/ (e.g. mptcp.c).

Ack. Sure thing.

-Jordan




[Index of Archives]     [Linux Samsung SoC]     [Linux Rockchip SoC]     [Linux Actions SoC]     [Linux for Synopsys ARC Processors]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]


  Powered by Linux