mbox series

[ovs-dev,v11,00/11] tests: Add system-traffic.at tests to check-offloads.

Message ID 167577861530.1166983.2484474802211392495.stgit@ebuild.local
Headers show
Series tests: Add system-traffic.at tests to check-offloads. | expand

Message

Eelco Chaudron Feb. 7, 2023, 2:03 p.m. UTC
This series makes it possible to include system-traffic.at tests into
"make check-offloads" tests.

The last patch of the series explains which tests are still not passing
and might need some more work.

I'll try to work on the remaining failing test cases or find someone
who can work on them.

These tests where executed on a Fedora37 machine with the kernel
6.1.5-200.fc37.x86_64 installed.

v11:
  - Lowered log level for "recirc_id sharing not supported" message.
    As this is not an error, but should be debug, like all other
    EOPNOTSUPP, related log messages.
v10:
  - Added missing userspace/kernel macro's.
v9:
  - Exclude "recirc_id sharing not supported" warning from the log.
  - Reworked to use a skip based macro rather than skip list.
  - Fixes some spelling.
  - Removed patches for issues no longer existing when using the
    latest kernel/OVS.
v8:
  - Re-based on top of latest OVS master.
v7:
  - Removed left over merge comment, and re-run all tests.
v6:
  - Added ACKs from v5
  - Changed 'netdev-offload-tc: If the flow has not been used, report
    it as such.' to also work on hardware offloaded flows.
v5:
  - Include all patches, v4 went out with missing two patches :(
v4:
  - Fix rename from system-traffic.at to sym-traffic.at in patch 11
v3:
  - Fixed missing MACRO's in patches 4, 6 and 10.
v2:
  - Fix commit message on last patch
  - Moved handling of system-traffic.at tests to a separate file
    system-offloads.at
  - Re-based to the latest ovs master branch
  - Added Roi's ACKs

Eelco Chaudron (11):
      tests: Include working system-traffic tests into the system-offloads-testsuite.
      test: Do not use MPLS implicit null label in test cases.
      netdev-offload-tc: Fix tc conntrack force commit support.
      test: Flush datapath when changing rules on the fly.
      netdev-offload-tc: Conntrack ALGs are not supported with tc.
      test: tc does not support conntrack timeout, skip the related test.
      test: Fix 'conntrack - Multiple ICMP traverse' for tc case.
      odp-util: Make odp_flow_key_from_flow__ nlattr order the same as the kernel.
      netdev-offload-tc: If the flow has not been used, report it as such.
      tests: Fix reading of OpenFlow byte counters in GRE test cases.
      tests: Comment currently failing TC system-traffic tests.


 Documentation/howto/tc-offload.rst        | 11 +++
 lib/netdev-offload-tc.c                   | 19 ++++-
 lib/odp-util.c                            | 21 +++---
 lib/tc.c                                  | 14 +++-
 tests/automake.mk                         |  3 +-
 tests/dpif-netdev.at                      | 28 ++++----
 tests/mcast-snooping.at                   |  4 +-
 tests/nsh.at                              | 10 +--
 tests/odp.at                              | 84 +++++++++++------------
 tests/ofproto-dpif.at                     | 30 ++++----
 tests/packet-type-aware.at                | 22 +++---
 tests/pmd.at                              |  2 +-
 tests/system-kmod-macros.at               |  5 ++
 tests/system-offloads-testsuite-macros.at | 69 +++++++++++++++++++
 tests/system-offloads-testsuite.at        |  3 +
 tests/system-traffic.at                   | 37 ++++++----
 tests/system-userspace-macros.at          |  5 ++
 tests/tunnel-push-pop-ipv6.at             |  2 +-
 tests/tunnel-push-pop.at                  |  2 +-
 tests/tunnel.at                           |  2 +-
 20 files changed, 249 insertions(+), 124 deletions(-)
 create mode 100644 tests/system-offloads-testsuite-macros.at

Comments

Ilya Maximets Feb. 9, 2023, 12:45 p.m. UTC | #1
On 2/7/23 15:03, Eelco Chaudron wrote:
> This series makes it possible to include system-traffic.at tests into
> "make check-offloads" tests.
> 
> The last patch of the series explains which tests are still not passing
> and might need some more work.
> 
> I'll try to work on the remaining failing test cases or find someone
> who can work on them.
> 
> These tests where executed on a Fedora37 machine with the kernel
> 6.1.5-200.fc37.x86_64 installed.
> 
> v11:
>   - Lowered log level for "recirc_id sharing not supported" message.
>     As this is not an error, but should be debug, like all other
>     EOPNOTSUPP, related log messages.
> v10:
>   - Added missing userspace/kernel macro's.
> v9:
>   - Exclude "recirc_id sharing not supported" warning from the log.
>   - Reworked to use a skip based macro rather than skip list.
>   - Fixes some spelling.
>   - Removed patches for issues no longer existing when using the
>     latest kernel/OVS.
> v8:
>   - Re-based on top of latest OVS master.
> v7:
>   - Removed left over merge comment, and re-run all tests.
> v6:
>   - Added ACKs from v5
>   - Changed 'netdev-offload-tc: If the flow has not been used, report
>     it as such.' to also work on hardware offloaded flows.
> v5:
>   - Include all patches, v4 went out with missing two patches :(
> v4:
>   - Fix rename from system-traffic.at to sym-traffic.at in patch 11
> v3:
>   - Fixed missing MACRO's in patches 4, 6 and 10.
> v2:
>   - Fix commit message on last patch
>   - Moved handling of system-traffic.at tests to a separate file
>     system-offloads.at
>   - Re-based to the latest ovs master branch
>   - Added Roi's ACKs
> 
> Eelco Chaudron (11):
>       tests: Include working system-traffic tests into the system-offloads-testsuite.
>       test: Do not use MPLS implicit null label in test cases.
>       netdev-offload-tc: Fix tc conntrack force commit support.
>       test: Flush datapath when changing rules on the fly.
>       netdev-offload-tc: Conntrack ALGs are not supported with tc.
>       test: tc does not support conntrack timeout, skip the related test.
>       test: Fix 'conntrack - Multiple ICMP traverse' for tc case.
>       odp-util: Make odp_flow_key_from_flow__ nlattr order the same as the kernel.
>       netdev-offload-tc: If the flow has not been used, report it as such.
>       tests: Fix reading of OpenFlow byte counters in GRE test cases.
>       tests: Comment currently failing TC system-traffic tests.
> 
> 
>  Documentation/howto/tc-offload.rst        | 11 +++
>  lib/netdev-offload-tc.c                   | 19 ++++-
>  lib/odp-util.c                            | 21 +++---
>  lib/tc.c                                  | 14 +++-
>  tests/automake.mk                         |  3 +-
>  tests/dpif-netdev.at                      | 28 ++++----
>  tests/mcast-snooping.at                   |  4 +-
>  tests/nsh.at                              | 10 +--
>  tests/odp.at                              | 84 +++++++++++------------
>  tests/ofproto-dpif.at                     | 30 ++++----
>  tests/packet-type-aware.at                | 22 +++---
>  tests/pmd.at                              |  2 +-
>  tests/system-kmod-macros.at               |  5 ++
>  tests/system-offloads-testsuite-macros.at | 69 +++++++++++++++++++
>  tests/system-offloads-testsuite.at        |  3 +
>  tests/system-traffic.at                   | 37 ++++++----
>  tests/system-userspace-macros.at          |  5 ++
>  tests/tunnel-push-pop-ipv6.at             |  2 +-
>  tests/tunnel-push-pop.at                  |  2 +-
>  tests/tunnel.at                           |  2 +-
>  20 files changed, 249 insertions(+), 124 deletions(-)
>  create mode 100644 tests/system-offloads-testsuite-macros.at
> 

Thanks, Eelco, Roi and Simon!

I tested this set on rhel9 and ubuntu 22.04.  Both seems to work fine.

There are some failures in the testsuite in general: ubuntu fails the
meter tests in 100% of cases and rhel9 fails them occasionally as well
but for some other reason.  However, these failures are not connected
to a current patch set, tests are failing without it too.

Tests from patch 10/11 are more or less stable on both of these systems
for me, but, yes, they do fail once in 10-ish runs.

I fixed the incorrect indentation in patch 05/11 and the missing 'Fixes:'
in patch 09/11.  With that, applied to master.

Patches 3, 5 and 9 additionally backported down to 2.17.

Best regards, Ilya Maximets.
Eelco Chaudron Feb. 9, 2023, 1:26 p.m. UTC | #2
On 9 Feb 2023, at 13:45, Ilya Maximets wrote:

> On 2/7/23 15:03, Eelco Chaudron wrote:
>> This series makes it possible to include system-traffic.at tests into
>> "make check-offloads" tests.
>>
>> The last patch of the series explains which tests are still not passing
>> and might need some more work.
>>
>> I'll try to work on the remaining failing test cases or find someone
>> who can work on them.
>>
>> These tests where executed on a Fedora37 machine with the kernel
>> 6.1.5-200.fc37.x86_64 installed.
>>
>> v11:
>>   - Lowered log level for "recirc_id sharing not supported" message.
>>     As this is not an error, but should be debug, like all other
>>     EOPNOTSUPP, related log messages.
>> v10:
>>   - Added missing userspace/kernel macro's.
>> v9:
>>   - Exclude "recirc_id sharing not supported" warning from the log.
>>   - Reworked to use a skip based macro rather than skip list.
>>   - Fixes some spelling.
>>   - Removed patches for issues no longer existing when using the
>>     latest kernel/OVS.
>> v8:
>>   - Re-based on top of latest OVS master.
>> v7:
>>   - Removed left over merge comment, and re-run all tests.
>> v6:
>>   - Added ACKs from v5
>>   - Changed 'netdev-offload-tc: If the flow has not been used, report
>>     it as such.' to also work on hardware offloaded flows.
>> v5:
>>   - Include all patches, v4 went out with missing two patches :(
>> v4:
>>   - Fix rename from system-traffic.at to sym-traffic.at in patch 11
>> v3:
>>   - Fixed missing MACRO's in patches 4, 6 and 10.
>> v2:
>>   - Fix commit message on last patch
>>   - Moved handling of system-traffic.at tests to a separate file
>>     system-offloads.at
>>   - Re-based to the latest ovs master branch
>>   - Added Roi's ACKs
>>
>> Eelco Chaudron (11):
>>       tests: Include working system-traffic tests into the system-offloads-testsuite.
>>       test: Do not use MPLS implicit null label in test cases.
>>       netdev-offload-tc: Fix tc conntrack force commit support.
>>       test: Flush datapath when changing rules on the fly.
>>       netdev-offload-tc: Conntrack ALGs are not supported with tc.
>>       test: tc does not support conntrack timeout, skip the related test.
>>       test: Fix 'conntrack - Multiple ICMP traverse' for tc case.
>>       odp-util: Make odp_flow_key_from_flow__ nlattr order the same as the kernel.
>>       netdev-offload-tc: If the flow has not been used, report it as such.
>>       tests: Fix reading of OpenFlow byte counters in GRE test cases.
>>       tests: Comment currently failing TC system-traffic tests.
>>
>>
>>  Documentation/howto/tc-offload.rst        | 11 +++
>>  lib/netdev-offload-tc.c                   | 19 ++++-
>>  lib/odp-util.c                            | 21 +++---
>>  lib/tc.c                                  | 14 +++-
>>  tests/automake.mk                         |  3 +-
>>  tests/dpif-netdev.at                      | 28 ++++----
>>  tests/mcast-snooping.at                   |  4 +-
>>  tests/nsh.at                              | 10 +--
>>  tests/odp.at                              | 84 +++++++++++------------
>>  tests/ofproto-dpif.at                     | 30 ++++----
>>  tests/packet-type-aware.at                | 22 +++---
>>  tests/pmd.at                              |  2 +-
>>  tests/system-kmod-macros.at               |  5 ++
>>  tests/system-offloads-testsuite-macros.at | 69 +++++++++++++++++++
>>  tests/system-offloads-testsuite.at        |  3 +
>>  tests/system-traffic.at                   | 37 ++++++----
>>  tests/system-userspace-macros.at          |  5 ++
>>  tests/tunnel-push-pop-ipv6.at             |  2 +-
>>  tests/tunnel-push-pop.at                  |  2 +-
>>  tests/tunnel.at                           |  2 +-
>>  20 files changed, 249 insertions(+), 124 deletions(-)
>>  create mode 100644 tests/system-offloads-testsuite-macros.at
>>
>
> Thanks, Eelco, Roi and Simon!
>
> I tested this set on rhel9 and ubuntu 22.04.  Both seems to work fine.
>
> There are some failures in the testsuite in general: ubuntu fails the
> meter tests in 100% of cases and rhel9 fails them occasionally as well
> but for some other reason.  However, these failures are not connected
> to a current patch set, tests are failing without it too.
>
> Tests from patch 10/11 are more or less stable on both of these systems
> for me, but, yes, they do fail once in 10-ish runs.
>
> I fixed the incorrect indentation in patch 05/11 and the missing 'Fixes:'
> in patch 09/11.  With that, applied to master.
>
> Patches 3, 5 and 9 additionally backported down to 2.17.

Thanks Ilya and Simon for the reviews!

> Best regards, Ilya Maximets.