Bug 212077 – [Flatpak SDK] Add bubblewrap ... wrapper

• Component: Tools / Tests
• Status: RESOLVED FIXED

[Philippe Normand]

Allowing us to use flatpak run now.

[Philippe Normand]

Created attachment 399740 [details]
Patch

[Carlos Alberto Lopez Perez]

Comment on attachment 399740 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=399740&action=review

> Tools/ChangeLog:8
> +        The bind-mounts are now handled through webkit-bwrap.

It looks to me some binds mounts are gone with this new way of doing things.

> Tools/flatpak/flatpakutils.py:-661
> -                           "--bind-mount=/run/shm=/dev/shm",

I don't see this now bind-mounted with the new setup. Why is that?

> Tools/flatpak/flatpakutils.py:-668
> -                           "--bind-mount=/run/systemd/journal=/run/systemd/journal",

neither this, isn't it needed?

> Tools/flatpak/flatpakutils.py:-715
> -                    flatpak_command.append("--bind-mount={uid_doc_path}={uid_doc_path}".format(uid_doc_path=uid_doc_path))

I also don't see this mounted

[Philippe Normand]

That's correct, flatpak run sets those bind-mounts already, while flatpak build doesn't.

[Philippe Normand]

Comment on attachment 399740 [details]
Patch

I couldn't reproduce the api-gtk failures here.

[EWS]

Committed r261909: <https://trac.webkit.org/changeset/261909>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 399740 [details].

[Radar WebKit Bug Importer]

<rdar://problem/63437366>

[Diego Pino]

Reverted r261909 for reason:

execution of LayoutTests returns syntax error in the bots (GTK, WPE)

Committed r261917: <https://trac.webkit.org/changeset/261917>

[Philippe Normand]

python ./Tools/Scripts/run-webkit-tests --no-build --no-show-results --no-new-test-results --clobber-old-results --builder-name 'GTK Linux 64-bit Release (Tests)' --build-number 13781 --buildbot-worker gtk-linux-slave-6 --master-name webkit.org --buildbot-master build.webkit.org --report https://results.webkit.org --exit-after-n-crashes-or-timeouts 50 --exit-after-n-failures 500 --release --layout-tests-directory ./Tools/BuildSlaveSupport/build.webkit.org-config/public_html/dashboard/Scripts/tests --gtk --results-directory layout-test-results/dashboard-layout-test-results --debug-rwt-logging
 in dir /home/slave/webkitgtk/gtk-linux-64-release-tests/build (timeout 1200 secs)
 watching logfiles {}
 argv: ['python', './Tools/Scripts/run-webkit-tests', '--no-build', '--no-show-results', '--no-new-test-results', '--clobber-old-results', '--builder-name', 'GTK Linux 64-bit Release (Tests)', '--build-number', '13781', '--buildbot-worker', 'gtk-linux-slave-6', '--master-name', 'webkit.org', '--buildbot-master', 'build.webkit.org', '--report', 'https://results.webkit.org', '--exit-after-n-crashes-or-timeouts', '50', '--exit-after-n-failures', '500', '--release', '--layout-tests-directory', './Tools/BuildSlaveSupport/build.webkit.org-config/public_html/dashboard/Scripts/tests', '--gtk', '--results-directory', 'layout-test-results/dashboard-layout-test-results', '--debug-rwt-logging']
 using PTY: True
sh: 1: Syntax error: "(" unexpected
program finished with exit code 0
elapsedTime=0.892059

[Philippe Normand]

Created attachment 399954 [details]
Patch

[Carlos Alberto Lopez Perez]

Comment on attachment 399954 [details]
Patch

what is the change done compared with r261909?

[Philippe Normand]

No change :) I'd like to reland as-is, making sure the exec bit is set on webkit-bwrap this time.

[Carlos Alberto Lopez Perez]

(In reply to Philippe Normand from comment #11)
> No change :) I'd like to reland as-is, making sure the exec bit is set on
> webkit-bwrap this time.

I'm confused. Your previous patch already set it as executable.

[Philippe Normand]

How do you know?

[Philippe Normand]

I've rebased the patch mostly for Nikolas though. I'll need to test it again before landing in any case.

[Carlos Alberto Lopez Perez]

(In reply to Philippe Normand from comment #13)
> How do you know?

I just checked out r261909 locally and checked that the file was executable.

[Philippe Normand]

With svn or git/git-svn?

[Carlos Alberto Lopez Perez]

(In reply to Philippe Normand from comment #16)
> With svn or git/git-svn?

With git. There should be no difference.

You can also check  https://trac.webkit.org/browser/webkit/trunk/Tools/flatpak/webkit-bwrap?rev=261909 <-- it says "Property svn:executable set to *"

[Carlos Alberto Lopez Perez]

(In reply to Carlos Alberto Lopez Perez from comment #17)
> (In reply to Philippe Normand from comment #16)
> > With svn or git/git-svn?
> 
> With git. There should be no difference.
> 
> You can also check 
> https://trac.webkit.org/browser/webkit/trunk/Tools/flatpak/webkit-
> bwrap?rev=261909 <-- it says "Property svn:executable set to *"

Or with command:
  svn diff -c 261909  https://svn.webkit.org/repository/webkit/trunk/

[Philippe Normand]

Well I can't reproduce this issue anywhere and since the error message doesn't give much details, I can only guess here. Without relanding this patch I won't be able to debug this.

[Philippe Normand]

Created attachment 400034 [details]
Patch

[Carlos Alberto Lopez Perez]

(In reply to Philippe Normand from comment #19)
> Well I can't reproduce this issue anywhere and since the error message
> doesn't give much details, I can only guess here. Without relanding this
> patch I won't be able to debug this.

Ok. Then Feel free to land any follow-up fixes unreviewed.(In reply to Philippe Normand from comment #19)
> Well I can't reproduce this issue anywhere and since the error message
> doesn't give much details.

I can. You need to run the command like the bot does.

python ./Tools/Scripts/run-webkit-tests --no-build --no-show-results --no-new-test-results --clobber-old-results --builder-name "GTK Linux 64-bit Release (Tests)" --build-number 13781 --buildbot-worker gtk-linux-slave-6 --master-name webkit.org --buildbot-master build.webkit.org --report https://results.webkit.org --exit-after-n-crashes-or-timeouts 50 --exit-after-n-failures 500 --release --gtk --results-directory layout-test-results --debug-rwt-logging

sh: 1: Syntax error: "(" unexpected




And the issue its still happening with this last version of the patch

It looks it is caused by the parameters: --builder-name "GTK Linux 64-bit Release (Tests)"

The parenthesis in "(Tests)" causes This. If the parameters are passed to a shell maybe they need to be quoted/escaped properly

[Carlos Alberto Lopez Perez]

(In reply to Philippe Normand from comment #19)
> Well I can't reproduce this issue anywhere and since the error message
> doesn't give much details, I can only guess here. Without relanding this
> patch I won't be able to debug this.

I can. You need to run the command like the bot does.

python ./Tools/Scripts/run-webkit-tests --no-build --no-show-results --no-new-test-results --clobber-old-results --builder-name "GTK Linux 64-bit Release (Tests)" --build-number 13781 --buildbot-worker gtk-linux-slave-6 --master-name webkit.org --buildbot-master build.webkit.org --report https://results.webkit.org --exit-after-n-crashes-or-timeouts 50 --exit-after-n-failures 500 --release --gtk --results-directory layout-test-results --debug-rwt-logging

sh: 1: Syntax error: "(" unexpected



And the issue its still happening with this last version of the patch

It looks it is caused by the parameters: --builder-name "GTK Linux 64-bit Release (Tests)"

The parenthesis in "(Tests)" causes This. If the parameters are passed to a shell maybe they need to be quoted/escaped properly

[Philippe Normand]

🤯

[Philippe Normand]

Created attachment 400043 [details]
Patch

[EWS]

Committed r262057: <https://trac.webkit.org/changeset/262057>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 400043 [details].