Re: [PATCH net-next 6/6] selftests: net: exercise page pool reporting via netlink

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

 



Jakub Kicinski <kuba@xxxxxxxxxx> writes:

> Add a Python test for the basic ops.
>
>   # ./net/nl_netdev.py
>   KTAP version 1
>   1..3
>   ok 1 nl_netdev.empty_check
>   ok 2 nl_netdev.lo_check
>   ok 3 nl_netdev.page_pool_check
>   # Totals: pass:3 fail:0 xfail:0 xpass:0 skip:0 error:0
>
> Signed-off-by: Jakub Kicinski <kuba@xxxxxxxxxx>
> ---
>  tools/testing/selftests/net/lib/py/nsim.py |  7 ++
>  tools/testing/selftests/net/nl_netdev.py   | 79 +++++++++++++++++++++-
>  2 files changed, 84 insertions(+), 2 deletions(-)
>
> diff --git a/tools/testing/selftests/net/lib/py/nsim.py b/tools/testing/selftests/net/lib/py/nsim.py
> index 94aa32f59fdb..1fd50a308408 100644
> --- a/tools/testing/selftests/net/lib/py/nsim.py
> +++ b/tools/testing/selftests/net/lib/py/nsim.py
> @@ -28,6 +28,13 @@ from .utils import cmd, ip
>          self.dfs_dir = "%s/ports/%u/" % (nsimdev.dfs_dir, port_index)
>          ret = ip("-j link show dev %s" % ifname, ns=ns)
>          self.dev = json.loads(ret.stdout)[0]
> +        self.ifindex = self.dev["ifindex"]
> +
> +    def up(self):
> +        ip("link set dev {} up".format(self.ifname))
> +
> +    def down(self):
> +        ip("link set dev {} down".format(self.ifname))

Yeah, what I meant by integration pain with LNST the other day was
basically this. You end up rewriting the iproute2 API in Python.
But it is what it is.

>      def dfs_write(self, path, val):
>          self.nsimdev.dfs_write(f'ports/{self.port_index}/' + path, val)
> diff --git a/tools/testing/selftests/net/nl_netdev.py b/tools/testing/selftests/net/nl_netdev.py
> index 2b8b488fb419..afc510c044ce 100755
> --- a/tools/testing/selftests/net/nl_netdev.py
> +++ b/tools/testing/selftests/net/nl_netdev.py
> @@ -1,7 +1,8 @@
>  #!/usr/bin/env python3
>  # SPDX-License-Identifier: GPL-2.0
>  
> -from lib.py import ksft_run, ksft_pr, ksft_eq, ksft_ge, NetdevFamily
> +import time
> +from lib.py import ksft_run, ksft_pr, ksft_eq, ksft_ge, NetdevFamily, NetdevSimDev, ip
>  
>  
>  def empty_check(nf) -> None:
> @@ -15,9 +16,83 @@ from lib.py import ksft_run, ksft_pr, ksft_eq, ksft_ge, NetdevFamily
>      ksft_eq(len(lo_info['xdp-rx-metadata-features']), 0)
>  
>  
> +def page_pool_check(nf) -> None:
> +    with NetdevSimDev() as nsimdev:
> +        nsim = nsimdev.nsims[0]
> +
> +        # No page pools when down
> +        nsim.down()
> +        pp_list = nf.page_pool_get({}, dump=True)
> +        pp_list = [pp for pp in pp_list if pp.get("ifindex") == nsim.ifindex]

This combo of page_pool_get / filter looks like it should be in a helper.

> +        ksft_eq(len(pp_list), 0)
> +
> +        # Up, empty page pool appears
> +        nsim.up()
> +        pp_list = nf.page_pool_get({}, dump=True)
> +        pp_list = [pp for pp in pp_list if pp.get("ifindex") == nsim.ifindex]
> +        ksft_ge(len(pp_list), 0)
> +        refs = sum([pp["inflight"] for pp in pp_list])
> +        ksft_eq(refs, 0)
> +
> +        # Down, it disappears, again
> +        nsim.down()
> +        pp_list = nf.page_pool_get({}, dump=True)
> +        pp_list = [pp for pp in pp_list if pp.get("ifindex") == nsim.ifindex]
> +        ksft_eq(len(pp_list), 0)
> +
> +        # Up, allocate a page
> +        nsim.up()
> +        nsim.dfs_write("pp_hold", "y")
> +        pp_list = nf.page_pool_get({}, dump=True)
> +        refs = sum([pp["inflight"] for pp in pp_list if pp.get("ifindex") == nsim.ifindex])
> +        ksft_ge(refs, 1)
> +
> +        # Now let's leak a page
> +        nsim.down()
> +        pp_list = nf.page_pool_get({}, dump=True)
> +        pp_list = [pp for pp in pp_list if pp.get("ifindex") == nsim.ifindex]
> +        ksft_eq(len(pp_list), 1)
> +        refs = sum([pp["inflight"] for pp in pp_list if pp.get("ifindex") == nsim.ifindex])

The second filtering is unnecessary.

> +        ksft_eq(refs, 1)
> +        undetached = [pp for pp in pp_list if "detach-time" not in pp]

Maybe call this attached to be in sync with the next hunk?

> +        ksft_eq(len(undetached), 0)
> +
> +        # New pp can get created, and we'll have two
> +        nsim.up()
> +        pp_list = nf.page_pool_get({}, dump=True)
> +        pp_list = [pp for pp in pp_list if pp.get("ifindex") == nsim.ifindex]
> +        attached = [pp for pp in pp_list if "detach-time" not in pp]
> +        undetached = [pp for pp in pp_list if "detach-time" in pp]

detached.

> +        ksft_eq(len(attached), 1)
> +        ksft_eq(len(undetached), 1)
> +
> +        # Free the old page and the old pp is gone
> +        nsim.dfs_write("pp_hold", "n")
> +        # Freeing check is once a second so we may need to retry
> +        for i in range(50):
> +            pp_list = nf.page_pool_get({}, dump=True)
> +            pp_list = [pp for pp in pp_list if pp.get("ifindex") == nsim.ifindex]
> +            if len(pp_list) == 1:
> +                break
> +            time.sleep(0.05)
> +        ksft_eq(len(pp_list), 1)

Yeah, I don't know if busywait / slowwait sort of a helper is practical
to write in Python. No idea how to go about it. But the bash experience
shows it would useful fairly often.

> +        # And down...
> +        nsim.down()
> +        pp_list = nf.page_pool_get({}, dump=True)
> +        pp_list = [pp for pp in pp_list if pp.get("ifindex") == nsim.ifindex]
> +        ksft_eq(len(pp_list), 0)
> +
> +        # Last, leave the page hanging for destroy, nothing to check
> +        # we're trying to exercise the orphaning path in the kernel
> +        nsim.up()
> +        nsim.dfs_write("pp_hold", "y")
> +
> +
>  def main() -> None:
>      nf = NetdevFamily()
> -    ksft_run([empty_check, lo_check], args=(nf, ))
> +    ksft_run([empty_check, lo_check, page_pool_check],
> +             args=(nf, ))
>  
>  
>  if __name__ == "__main__":





[Index of Archives]     [Linux Wireless]     [Linux Kernel]     [ATH6KL]     [Linux Bluetooth]     [Linux Netdev]     [Kernel Newbies]     [Share Photos]     [IDE]     [Security]     [Git]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux ATA RAID]     [Samba]     [Device Mapper]

  Powered by Linux