DPDK patches and discussions
 help / color / mirror / Atom feed
* [RFC PATCH 0/1] Initial Implementation For Jumbo Frames
@ 2024-05-24 18:06 Nicholas Pratte
  2024-05-24 18:06 ` [RFC PATCH 1/1] Initial Implementation For Jumbo Frames Test Suite Nicholas Pratte
  2024-05-24 18:13 ` [RFC PATCH 0/1] Initial Implementation For Jumbo Frames Nicholas Pratte
  0 siblings, 2 replies; 4+ messages in thread
From: Nicholas Pratte @ 2024-05-24 18:06 UTC (permalink / raw)
  To: paul.szczepanek, jspewock, luca.vizzarro, thomas, juraj.linkes,
	bruce.richardson, Honnappa.Nagarahalli, probb, yoan.picchi
  Cc: dev, Nicholas Pratte

The following is a rough design for the implementation of the jumbo
frames test suite in new DTS. The test suite uses the same
testing methodology from the test suite implementation in old
DTS. In doing so, much of the logic in this implementation has been
stripped away for the sake of maintaining an OS-agnostic design
philosopgy. Thus, there are some concerns here that need further
discussion.

All test cases behave accordingly on lab hardware. However, issues with
the testpmd shell, all of which will be fixed with the upcoming context
manager, prevent the test suite from functioning properly if all test
cases are run sequentially. Thus, for testing, each test case needs to
be run individually. So, expect issues if attempting to test.

Old DTS implements some basic logic that detects 1GB NICs, specifically
when testing jumbo frame packets greater than the specified MTU length.
This is because, according to a commit message from 2016, 1GB NICs will
automatically adjust their size to +4 bytes, meaning that when setting
an MTU of 9000, testpmd adjust the MTU size to 9004. To compensate, some
logic was inserted in the old suite to adjust packet sizes accordingly.
Given that Juraj's capabilities patch is in development, it would be
possible to set a restriction on this test suite to remove the support
of 1GB NICs, which could be determined from testpmd outputs; it could
also be possible that this problem is entirely null and void today,
essentially allowing us to forget about it.

The current implementation is not developed with both the capabilities
patch and the params patch in mind; however, the current design can and
will be refactored to do so.

Nicholas Pratte (1):
  Initial Implementation For Jumbo Frames Test Suite

 dts/framework/config/conf_yaml_schema.json |   3 +-
 dts/framework/test_suite.py                |   8 +
 dts/tests/TestSuite_jumboframes.py         | 210 +++++++++++++++++++++
 3 files changed, 220 insertions(+), 1 deletion(-)
 create mode 100644 dts/tests/TestSuite_jumboframes.py

-- 
2.44.0


^ permalink raw reply	[flat|nested] 4+ messages in thread
* [RFC PATCH 0/1] Initial Implementation For Jumbo Frames
@ 2024-05-24 18:36 Nicholas Pratte
  0 siblings, 0 replies; 4+ messages in thread
From: Nicholas Pratte @ 2024-05-24 18:36 UTC (permalink / raw)
  To: thomas, jspewock, juraj.linkes, luca.vizzarro, yoan.picchi,
	bruce.richardson, Honnappa.Nagarahalli, probb, paul.szczepanek
  Cc: dev, Nicholas Pratte

The following is a rough design for the implementation of the jumbo
frames test suite in new DTS. The test suite uses the same
testing methodology from the test suite implementation in old
DTS. In doing so, much of the logic in this implementation has been
stripped away for the sake of maintaining an OS-agnostic design
philosopgy. Thus, there are some concerns here that need further
discussion.

All test cases behave accordingly on lab hardware. However, issues with
the testpmd shell, all of which will be fixed with the upcoming context
manager, prevent the test suite from functioning properly if all test
cases are run sequentially. Thus, for testing, each test case needs to
be run individually. So, expect issues if attempting to test.

Old DTS implements some basic logic that detects 1GB NICs, specifically
when testing jumbo frame packets greater than the specified MTU length.
This is because, according to a commit message from 2016, 1GB NICs will
automatically adjust their size to +4 bytes, meaning that when setting
an MTU of 9000, testpmd adjust the MTU size to 9004. To compensate, some
logic was inserted in the old suite to adjust packet sizes accordingly.
Given that Juraj's capabilities patch is in development, it would be
possible to set a restriction on this test suite to remove the support
of 1GB NICs, which could be determined from testpmd outputs; it could
also be possible that this problem is entirely null and void today,
essentially allowing us to forget about it.

The current implementation is not developed with both the capabilities
patch and the params patch in mind; however, the current design can and
will be refactored to do so.

Nicholas Pratte (1):
  Initial Implementation For Jumbo Frames Test Suite

 dts/framework/config/conf_yaml_schema.json |   3 +-
 dts/tests/TestSuite_jumboframes.py         | 210 +++++++++++++++++++++
 2 files changed, 212 insertions(+), 1 deletion(-)
 create mode 100644 dts/tests/TestSuite_jumboframes.py

-- 
2.44.0


^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2024-05-24 18:36 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-05-24 18:06 [RFC PATCH 0/1] Initial Implementation For Jumbo Frames Nicholas Pratte
2024-05-24 18:06 ` [RFC PATCH 1/1] Initial Implementation For Jumbo Frames Test Suite Nicholas Pratte
2024-05-24 18:13 ` [RFC PATCH 0/1] Initial Implementation For Jumbo Frames Nicholas Pratte
2024-05-24 18:36 Nicholas Pratte

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).