From: Patrick Robb <probb@iol.unh.edu>
To: Dean Marx <dmarx@iol.unh.edu>
Cc: luca.vizzarro@arm.com, yoan.picchi@foss.arm.com,
Honnappa.Nagarahalli@arm.com, paul.szczepanek@arm.com,
dev@dpdk.org
Subject: Re: [PATCH v1] dts: fix devbind initialization bug
Date: Sun, 15 Jun 2025 23:21:12 -0400 [thread overview]
Message-ID: <CAJvnSUC3U2rUPfa0wAcUtTcwPdidcX=qgRpQf=k3ORBKCB0_vw@mail.gmail.com> (raw)
In-Reply-To: <20250612201220.614724-1-dmarx@iol.unh.edu>
[-- Attachment #1: Type: text/plain, Size: 4811 bytes --]
On Thu, Jun 12, 2025 at 4:12 PM Dean Marx <dmarx@iol.unh.edu> wrote:
> Rearrange the topology and DPDK setup/teardown calls during
> test runs to ensure the devbind script is not called
> while the DPDK tmp directory doesn't exist.
>
> Fixes: 4cef16f1f0a4 ("dts: improve port handling")
>
> Signed-off-by: Dean Marx <dmarx@iol.unh.edu>
> ---
> dts/framework/test_run.py | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/dts/framework/test_run.py b/dts/framework/test_run.py
> index cff0085317..4287028b4b 100644
> --- a/dts/framework/test_run.py
> +++ b/dts/framework/test_run.py
> @@ -344,8 +344,8 @@ def next(self) -> State | None:
>
> test_run.ctx.sut_node.setup()
> test_run.ctx.tg_node.setup()
> - test_run.ctx.topology.setup()
> test_run.ctx.dpdk.setup()
> + test_run.ctx.topology.setup()
> test_run.ctx.tg.setup(test_run.ctx.topology)
>
> self.result.ports = test_run.ctx.topology.sut_ports +
> test_run.ctx.topology.tg_ports
> @@ -433,8 +433,8 @@ def next(self) -> State | None:
> """Next state."""
> self.test_run.ctx.shell_pool.terminate_current_pool()
> self.test_run.ctx.tg.teardown()
> - self.test_run.ctx.dpdk.teardown()
> self.test_run.ctx.topology.teardown()
> + self.test_run.ctx.dpdk.teardown()
> self.test_run.ctx.tg_node.teardown()
> self.test_run.ctx.sut_node.teardown()
> self.result.update_teardown(Result.PASS)
> --
> 2.49.0
>
>
Thanks this makes sense in principle but I think there is an additional
issue, which I saw when I just tested your patch on 2 XL710 NICs connected
both on the Braum server.
Here is the execution order from the testrun
1. testrun init creates DPDKRuntimeEnvironment with _node=sutnote like:
dpdk_runtime_env = DPDKRuntimeEnvironment(config.dpdk, sut_node,
dpdk_build_env)
which is then added to ctx.
2. TestRunSetup calls test_run.ctx.dpdk.setup() which
calls _prepare_devbind_script() which adds the devbind_script_path to
_node.main_session, but _node in this case is the sut node. I don't believe
there is the code path for adding devbind_script_path to the tg node
main_session, so later at topology.setup() we call self._setup_ports("sut")
which runs fine, and then error on self._setup_ports("tg") because again
the tg node main session is lacking the devbind_script_path attribute. I
think you will need to address this second part for your patch to be
complete. In any case, let's chat in the morning and figure out what a v2
of this bugfix can be.
See this stacktrace:
2025/06/16 02:59:03 - test_run_setup - dts.braum-intel-40g-SUT - INFO -
Sending: 'lshw -quiet -json -C network'
2025/06/16 02:59:04 - test_run_setup - dts.braum-intel-40g-TG - INFO -
Sending: 'lshw -quiet -json -C network'
2025/06/16 02:59:05 - test_run_setup - dts - ERROR - A CRITICAL ERROR has
occurred during test run setup. Unrecoverable state reached, shutting down.
2025/06/16 02:59:05 - test_run_setup - dts - ERROR - An unexpected error
has occurred.
Traceback (most recent call last):
File "/dpdk/dts/framework/runner.py", line 81, in run
test_run.spin()
File "/dpdk/dts/framework/test_run.py", line 242, in spin
next_state = self.state.handle_exception(e)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/dpdk/dts/framework/test_run.py", line 240, in spin
next_state = self.state.next()
^^^^^^^^^^^^^^^^^
File "/dpdk/dts/framework/test_run.py", line 348, in next
test_run.ctx.topology.setup()
File "/dpdk/dts/framework/testbed_model/topology.py", line 130, in setup
self._setup_ports("tg")
File "/dpdk/dts/framework/testbed_model/topology.py", line 154, in
_setup_ports
self._bind_ports_to_drivers(node, ports, "kernel")
File "/dpdk/dts/framework/testbed_model/topology.py", line 205, in
_bind_ports_to_drivers
node.main_session.bind_ports_to_driver(ports, driver_name)
File "/dpdk/dts/framework/testbed_model/linux_session.py", line 187, in
bind_ports_to_driver
f"{self.devbind_script_path} -b {driver_name} --force
{ports_pci_addrs}",
^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.12/functools.py", line 995, in __get__
val = self.func(instance)
^^^^^^^^^^^^^^^^^^^
File "/dpdk/dts/framework/testbed_model/linux_session.py", line 212, in
devbind_script_path
raise InternalError("Accessed devbind script path before setup.")
framework.exception.InternalError: Accessed devbind script path before
setup.
2025/06/16 02:59:05 - test_run_setup - dts - INFO - Moving from stage
'test_run_setup' to stage 'post_run'.
2025/06/16 02:59:05 - post_run - dts - INFO - DTS execution has ended.
[-- Attachment #2: Type: text/html, Size: 5897 bytes --]
next prev parent reply other threads:[~2025-06-16 3:26 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-06-12 20:12 Dean Marx
2025-06-16 3:21 ` Patrick Robb [this message]
2025-06-16 18:35 ` [PATCH v2] " Dean Marx
2025-06-16 18:38 ` Dean Marx
2025-06-17 0:17 ` Patrick Robb
2025-06-17 0:51 ` Patrick Robb
2025-06-17 15:18 ` Luca Vizzarro
2025-06-18 0:21 ` Patrick Robb
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAJvnSUC3U2rUPfa0wAcUtTcwPdidcX=qgRpQf=k3ORBKCB0_vw@mail.gmail.com' \
--to=probb@iol.unh.edu \
--cc=Honnappa.Nagarahalli@arm.com \
--cc=dev@dpdk.org \
--cc=dmarx@iol.unh.edu \
--cc=luca.vizzarro@arm.com \
--cc=paul.szczepanek@arm.com \
--cc=yoan.picchi@foss.arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).