Re: [PATCH i-g-t] tests/intel-ci/xe-sriov-vf.blocklist.txt: Blocklist for SR-IOV VF BAT runs

[Michal Wajdeczko <michal.wajdeczko@intel.com>]

On 17.07.2024 17:30, Marcin Bernatowicz wrote:
> Blocklist tests not applicable for Virtual Function (VF) BAT runs.

is this 'VF BAT run' about running IGTs in VM, or it's something different?

> The list extends xe.blocklist.txt to enable xe-fast-feedback.testlist
> execution on VF device.

maybe it's just a outdated doc, but README [1] says that blocklist is
only applied to *FULL* not *BAT* runs:

"blacklist.txt

This file contains regular expressions (one per line) for tests that are
not to be executed in full suite test rounds."

[1]
https://gitlab.freedesktop.org/drm/igt-gpu-tools/-/blob/master/tests/intel-ci/README?ref_type=heads

> 
> Cc: Adam Miszczak <adam.miszczak@linux.intel.com>
> Cc: Lukasz Laguna <lukasz.laguna@intel.com>
> Cc: Jakub Kolakowski <jakub1.kolakowski@intel.com>
> Cc: Michal Wajdeczko <Michal.Wajdeczko@intel.com>
> Cc: Kamil Konieczny <kamil.konieczny@linux.intel.com>
> Signed-off-by: Marcin Bernatowicz <marcin.bernatowicz@linux.intel.com>
> ---
>  tests/intel-ci/meson.build               |  1 +
>  tests/intel-ci/xe-sriov-vf.blocklist.txt | 21 +++++++++++++++++++++
>  2 files changed, 22 insertions(+)
>  create mode 100644 tests/intel-ci/xe-sriov-vf.blocklist.txt
> 
> diff --git a/tests/intel-ci/meson.build b/tests/intel-ci/meson.build
> index f16243cbb..59323eda9 100644
> --- a/tests/intel-ci/meson.build
> +++ b/tests/intel-ci/meson.build
> @@ -11,6 +11,7 @@ intelci_files = [
>    'xe.blocklist.txt',
>    'xe-fast-feedback.testlist',
>    'xe-fast-feedback-chamelium-only.testlist',
> +  'xe-sriov-vf.blocklist.txt',
>  ]
>  
>  install_data(sources : intelci_files, install_dir : datadir)
> diff --git a/tests/intel-ci/xe-sriov-vf.blocklist.txt b/tests/intel-ci/xe-sriov-vf.blocklist.txt
> new file mode 100644
> index 000000000..dae20bc6c
> --- /dev/null
> +++ b/tests/intel-ci/xe-sriov-vf.blocklist.txt

not a CI expert, but if there is:

	xe-fast-feedback.testlist

then maybe corresponding blocklist should be named:

	xe-fast-feedback.vf.blocklist

(and btw, I'm wondering why blocklists have .txt extension while
testlists correctly don't have it)



> @@ -0,0 +1,21 @@
> +##################################################################
> +# Block list for VF xe-fast-feedback runs
> +##################################################################
> +# The module is already loaded when VF is created
> +##################################################################
> +igt@xe_module_load

if tests are executed inside the VM then testing a module reload is
something that we want to have, no?

> +##################################################################
> +# Not applicable on VF
> +##################################################################
> +igt@fbdev

some devices don't have display (like ATSM) and likely don't have
dedicated .blocklist for them, so maybe it's fine that these tests will
SKIP (unless there will be separate BAT.display run introduced, where we
can also run kms tests)

> +igt@sriov_basic
> +igt@kms.*
> +igt@xe_gt_freq
> +igt@xe_live_ktest

only individual live-testcases might be N/A for VFs, but in general we
want to be able to execute live-tests on the VF

see [2] how this will be improved soon

[2] https://patchwork.freedesktop.org/patch/604990/?series=136163&rev=2

> +igt@xe_pm_residency
> +##################################################################
> +# Under investigation

what does it mean?
is it broken test or broken VF driver?
do we track those failures anywhere?

> +##################################################################
> +igt@xe_debugfs@gt
> +igt@xe_evict_ccs@evict-overcommit-simple
> +igt@xe_evict_ccs@evict-overcommit-parallel-nofree-samefd