1772074 - spi-nw xpcshell tests seems to run with network.process.enabled=true and network.process.enabled=false

Status: RESOLVED FIXED

(Firefox Build System :: Task Configuration, task)

Description

[Olli Pettay [:smaug][[e-mail]]]

https://searchfox.org/mozilla-central/rev/301e9545863b14ff85aa110fabd4274da4593f9d/taskcluster/ci/test/variants.yml#119-128

and I think the false is coming from:
https://searchfox.org/mozilla-central/rev/3419858c997f422e3e70020a46baae7f0ec6dacc/taskcluster/gecko_taskgraph/transforms/test/other.py#750

Current m-c
https://treeherder.mozilla.org/logviewer?job_id=379855608&repo=mozilla-central
[task 2022-06-01T04:54:27.575Z] executing ['/builds/worker/bin/test-linux.sh', '--setpref=network.process.enabled=true', '--setpref=network.http.network_access_on_socket_process.enabled=true', '--setpref=network.ssl_tokens_cache_enabled=true', '--xpcshell-suite=xpcshell', '--setpref=media.peerconnection.mtransport_process=false', '--setpref=network.process.enabled=false', '--setpref=layers.d3d11.enable-blacklist=false', '--download-symbols=true']

Comment 1

[Andrew Halberstadt [:ahal]]

Great catch!

Looks like this is happening because that transform in the second link isn't taking into account composite variants. The variant here is fission+socketprocess_networking.

Comment 2

[Kershaw Chang [:kershaw]]

I have a patch for this. The patch below can fix this issue.

--- a/taskcluster/gecko_taskgraph/transforms/test/other.py
+++ b/taskcluster/gecko_taskgraph/transforms/test/other.py
@@ -742,6 +742,8 @@ def ensure_spi_disabled_on_all_but_spi(config, tasks):
             has_setpref
             and variant != "socketprocess"
             and variant != "socketprocess_networking"
+            and variant != "socketprocess_networking+no-fission"
+            and variant != "socketprocess_networking+fission"
         ):
             task["mozharness"]["extra-options"].append(

Comment 3

[Kershaw Chang [:kershaw]]

https://treeherder.mozilla.org/jobs?repo=try&revision=082d9f3f50329ffff88b38dcd282c66acd01ecc9&selectedTaskRun=F-TbxlevQEaM11EWztbTUQ.0

Comment 4

[Kershaw Chang [:kershaw]]

:ahal, if this is the right approach, I am happy to submit my patch and you can review it. Thanks.

Comment 5

[Andrew Halberstadt [:ahal]]

Hey, thanks for the patch! Sorry, I already started working on this with a test. I think it would be better to solve this in the general case anyway so we don't run into it again if we decide to add another composite variant involving socketprocess. I'll have it up shortly :)

Comment 6

[Andrew Halberstadt [:ahal]]

Attachment: Bug 1772074 - [gecko_taskgraph] Add a test for the 'ensure_spi_disabled_on_all_but_spi' transform

Comment 7

[Andrew Halberstadt [:ahal]]

Attachment: Bug 1772074 - [gecko_taskgraph] Simplify logic in 'ensure_spi_disabled_on_all_but_spi' transform

Depends on D147871

Comment 8

[Andrew Halberstadt [:ahal]]

Attachment: Bug 1772074 - [gecko_taskgraph] Ensure 'ensure_spi_disabled_on_all_but_spi' transform properly handles composite variants

Depends on D147872

Comment 9

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/7e5a8329ee6c
[gecko_taskgraph] Add a test for the 'ensure_spi_disabled_on_all_but_spi' transform r=taskgraph-reviewers,jcristau
https://hg.mozilla.org/integration/autoland/rev/03f59a8a0d25
[gecko_taskgraph] Simplify logic in 'ensure_spi_disabled_on_all_but_spi' transform r=taskgraph-reviewers,jcristau
https://hg.mozilla.org/integration/autoland/rev/f3ab8e564c47
[gecko_taskgraph] Ensure 'ensure_spi_disabled_on_all_but_spi' transform properly handles composite variants r=taskgraph-reviewers,jcristau

Comment 10

[bszekely]

https://hg.mozilla.org/mozilla-central/rev/7e5a8329ee6c
https://hg.mozilla.org/mozilla-central/rev/03f59a8a0d25
https://hg.mozilla.org/mozilla-central/rev/f3ab8e564c47