Is this a developer build on Linux, developer build on device, or a canary channel release?

Dev build on Eve

I can repro on linux-chromeos on ToT 095fe82673b4

1. System tray > Network > wifi2_PSK
2. Can't type into password field.

I tried reverting two recent system dialog CLs (https://chromium-review.googlesource.com/c/1366793 and https://chromium-review.googlesource.com/c/1367086) but the problem persists.

I wonder if something is broken with IME.

I'm going to flip SingleProcessMash off by default, since this would be a dev-blocker and there's a dev channel release scheduled for next Tuesday.

msw, can you look into the underlying issue?

Sure, i'll take a look

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/908a62bf055bca31313d18dcc31617241df80b69

commit 908a62bf055bca31313d18dcc31617241df80b69
Author: James Cook <jamescook@chromium.org>
Date: Fri Jan 04 23:26:30 2019

Revert "chromeos: Enable SingleProcessMash (in-proc window service) by default"

This reverts commit badd8a0fab485b41aee70b4ed463868b7e218359.

Reason for revert: Breaks the Wi-Fi network password entry field
 http://crbug.com/919183 

Original change's description:
> chromeos: Enable SingleProcessMash (in-proc window service) by default
> 
> Turns on the window service (code in //services/ws) and uses it to back
> aura and views (code in //ui/aura/mus and //ui/views/mus). Runs the
> window service in-process.
> 
> A label "SingleProcessMash enabled" will appear on the desktop
> wallpaper to let developers and QA know the feature is enabled.
> The label will be removed in a separate CL before this feature
> goes to dev channel. Otherwise there are no visible changes.
> 
> Flips the former "single_process_mash_*" bot configs to
> "non_single_process_mash_*" and runs them with the feature disabled.
> This should keep the old code paths working until we're sure this
> feature will stick.
> 
> TBR=sky@chromium.org
> 
> BUG=918537
> TEST=single_process_mash_* bots
> 
> Change-Id: I58a30fd81ea12645b1aabc049c7e697ef1d9e72a
> Reviewed-on: https://chromium-review.googlesource.com/c/1393467
> Reviewed-by: Jun Mukai <mukai@chromium.org>
> Reviewed-by: James Cook <jamescook@chromium.org>
> Commit-Queue: James Cook <jamescook@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#619508}

TBR=jamescook@chromium.org,mukai@chromium.org,sky@chromium.org

# Not skipping CQ checks because original CL landed > 1 day ago.

Bug: 918537,  919183 
Change-Id: Ic3b5f0dd5b941d6383fd7290a0a55b60eedce30d
Reviewed-on: https://chromium-review.googlesource.com/c/1396702
Reviewed-by: James Cook <jamescook@chromium.org>
Reviewed-by: Jun Mukai <mukai@chromium.org>
Commit-Queue: James Cook <jamescook@chromium.org>
Cr-Commit-Position: refs/heads/master@{#620103}
[modify] https://crrev.com/908a62bf055bca31313d18dcc31617241df80b69/testing/buildbot/chromium.chromiumos.json
[modify] https://crrev.com/908a62bf055bca31313d18dcc31617241df80b69/testing/buildbot/chromium.memory.json
[modify] https://crrev.com/908a62bf055bca31313d18dcc31617241df80b69/testing/buildbot/test_suite_exceptions.pyl
[modify] https://crrev.com/908a62bf055bca31313d18dcc31617241df80b69/testing/buildbot/test_suites.pyl
[modify] https://crrev.com/908a62bf055bca31313d18dcc31617241df80b69/ui/base/ui_base_features.cc

It's weird that key presses aren't being processed for input, but I'm able to paste text into the field with Ctrl-V.
Also, event routing looks okay from the WindowService side, they're going to the deepest window (RenderWidgetHostViewAura):

[193106:193106:0104/154259.851511:ERROR:window_tree.cc(217)] SendEventToClient window=FrameSinkId(4294967295, 14) event_type=ET_KEY_PRESSED event_id=19446103
[193106:193106:0104/154300.291816:ERROR:debug_commands.cc(134)] RootWindow 0:
RootWindow-0 (0x17c1423a3c40) type=0 visible 0,0 1366x768
   ScreenRotationContainer (0x17c142431c40) type=0 visible 0,0 1366x768
      WallpaperContainer (0x17c142431e00) type=0 visible 0,0 1366x768
         WallpaperView (0x17c1430ad1c0) type=2 visible 0,0 1366x768
      NonLockScreenContainersContainer (0x17c1424af000) type=0 visible 0,0 1366x768
         AppListTabletModeContainer (0x17c1424af700) type=0 visible 0,0 1366x768 [snapped]
         UnparentedControlContainer (0x17c1424af8c0) type=0 0,0 1366x768
         DefaultContainer (0x17c1424afa80) type=0 visible 0,0 1366x768 [snapped]
         AlwaysOnTopContainer (0x17c1424afc40) type=0 visible 0,0 1366x768 [snapped]
             [proxy] id=FrameSinkId(4294967295, 10) ui/views/window/ClientView (0x17c1448dcc40) type=1 WindowStateType::DEFAULT [active] visible 427,100 512x512 [snapped]
                [proxy] id=FrameSinkId(4294967295, 11) DesktopNativeWidgetAura - content window (0x17c1448d68c0) type=0 visible 0,0 512x512
                   [proxy] id=FrameSinkId(4294967295, 13) NativeViewHostAuraClip (0x17c1448d9380) type=0 visible 0,32 512x480
                      [proxy] id=FrameSinkId(4294967295, 12) WebContentsViewAura (0x17c1448df700) type=0 visible 0,0 512x480
                         [proxy] id=FrameSinkId(4294967295, 14) RenderWidgetHostViewAura (0x17c1448d98c0) type=0 [focused] visible 0,0 512x480 ax_tree_id=4D4800BBD075A3B973F6A3ACF9A947D8
         AppListContainer (0x17c1424afe00) type=0 visible 0,0 1366x768 [snapped]

Worth noting:
* This does not affect other WebUI (e.g. Settings is fine, even in a window).
* These dialogs use SystemWebDialogDelegate [1] which primarily sets 'keep_on_top' = true. (|parent| is typically null) and calls chrome::ShowWebDialogWithParams [2].

[1] https://cs.chromium.org/chromium/src/chrome/browser/ui/webui/chromeos/system_web_dialog_delegate.cc?l=120
[2] https://cs.chromium.org/chromium/src/chrome/browser/ui/views/chrome_web_dialog_view.cc?g=0&l=47

Good call thinking to try Settings... that has a similar RenderWidgetHostViewAura event target:
Also, the Tab key works to cycle focus within the wifi dialog. I'll look at the RenderWidgetHostViewAura now.

[193106:193106:0104/154915.400572:ERROR:window_tree.cc(217)] SendEventToClient window=FrameSinkId(4294967295, 41) event_type=ET_KEY_PRESSED event_id=17308981
[193106:193106:0104/154915.746435:ERROR:debug_commands.cc(134)] RootWindow 0:
RootWindow-0 (0x17c1423a3c40) type=0 visible 0,0 1366x768
   ScreenRotationContainer (0x17c142431c40) type=0 visible 0,0 1366x768
      WallpaperContainer (0x17c142431e00) type=0 visible 0,0 1366x768
         WallpaperView (0x17c1430ad1c0) type=2 visible 0,0 1366x768
      NonLockScreenContainersContainer (0x17c1424af000) type=0 visible 0,0 1366x768
         AppListTabletModeContainer (0x17c1424af700) type=0 visible 0,0 1366x768 [snapped]
         UnparentedControlContainer (0x17c1424af8c0) type=0 0,0 1366x768
         DefaultContainer (0x17c1424afa80) type=0 visible 0,0 1366x768 [snapped]
             [proxy] id=FrameSinkId(4294967295, 38) BrowserFrame (0x17c14496f540) type=1 WindowStateType::DEFAULT [active] visible 88,16 1100x696 [snapped]
                [proxy] id=FrameSinkId(4294967295, 39) DesktopNativeWidgetAura - content window (0x17c1425fca80) type=0 visible 0,0 1100x696
                   [proxy] id=FrameSinkId(4294967295, 42) NativeViewHostAuraClip (0x17c1430ad8c0) type=0 visible 0,32 1100x664
                      [proxy] id=FrameSinkId(4294967295, 40) WebContentsViewAura (0x17c142e8de00) type=0 visible 0,0 1100x664
                         [proxy] id=FrameSinkId(4294967295, 41) RenderWidgetHostViewAura (0x17c14366d700) type=0 [focused] visible 0,0 1100x664 ax_tree_id=E708DA0A4FA4BC5F92705CB062370A7B
             [proxy] id=FrameSinkId(4294967295, 43) StatusBubble (0x17c144915700) type=2 WindowStateType::DEFAULT 87,691 366x22 [snapped]
                [proxy] id=FrameSinkId(4294967295, 44) DesktopNativeWidgetAura - content window (0x17c1449298c0) type=0 0,0 366x22
         AlwaysOnTopContainer (0x17c1424afc40) type=0 visible 0,0 1366x768 [snapped]

This is almost certainly an issue with IME.

I am not sure how IME currently works with mus, but when it was first being worked on, the design-doc for that work was this: https://docs.google.com/document/d/1LNxH7fRpQcuNzKxzMs4kKl4xnO-2Wa_ocZkNsr9vfdo/edit#heading=h.92cxd2mu4hww

Key presses to the wifi dialog are marked |stopped_propagation| and never reach InsertChar in InputMethodChromeOS::PostProcessUnfilteredKeyPressEvent (whereas functional browser window key presses do hit InsertChar): https://cs.chromium.org/chromium/src/ui/base/ime/input_method_chromeos.cc?rcl=69644b69c5482983eafdbc1528e71a797ed21f79&l=475

It looks like that bool comes from the event's stopped_propagation field, set sometime during WindowTreeHost::DispatchKeyEventPostIME:
https://cs.chromium.org/chromium/src/ui/aura/window_tree_host.cc?rcl=cb494102ab73c46eb0bcc3b49f8a1e4c6bbeb015&l=266
The bool is plumbed via ui::internal::InputMethodDelegate::CallDispatchKeyEventPostIMEAck and TextInputClientImpl/RemoteTextInputClient::DispatchKeyEventPostIME.

Key presses in the wifi dialog and browser windows both follow this rough path to get there (lots of hops, tricky to trace!):
WindowTreeClient::OnWindowInputEvent -> InputMethodMus::DispatchKeyEvent -> InputMethodBridge::ProcessKeyEvent -> InputMethodChromeOS::DispatchKeyEvent -> InputMethodBase::DispatchKeyEventPostIME -> RemoteTextInputClient::DispatchKeyEventPostIME -> WindowTreeHost::DispatchKeyEventPostIME -> InputMethodChromeOS::PostProcessUnfilteredKeyPressEvent -> InputMethodMus::ProcessKeyEventCallback.

Similarly, wifi dialog key events are handled, while browser events are *not*, as per InputMethodMus::ProcessKeyEventCallback:
https://cs.chromium.org/chromium/src/ui/aura/mus/input_method_mus.cc?rcl=69644b69c5482983eafdbc1528e71a797ed21f79&l=242

More investigation needed...

I'm not familiar with the SSID field. Is it webui or a View's textfield?

If a renderer has focus it is given the event first. If the renderer consumes the event, then processing in ash won't happen.

Wi-Fi network connect dialog is webui.

(Pretty much all system UI dialogs are webui these days.)

The "handled/processed" flag is set by WebDialogView::HandleKeyboardEvent [1] from RenderWidgetHostImpl::OnKeyboardEventAck [2]. The comment on WebDialogView::HandleKeyboardEvent says it is a simplified BrowserView::HandleKeyboardEvent. Maybe we should use UnhandledKeyboardEventHandler, or figure out a proper way to repost.

[1] https://cs.chromium.org/chromium/src/ui/views/controls/webview/web_dialog_view.cc?rcl=62a4f4357b9c18e4abc5b31ba007590c54d43ba1&l=296

[2] https://cs.chromium.org/chromium/src/content/browser/renderer_host/render_widget_host_impl.cc?rcl=9842ec7f5ec1658d3530ac717f284828f4c9772c&l=2184

Thanks for finding a fix so quickly, Xiyuan.
Passing this issue your way per http://crrev.com/c/1397853

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/6d5f163cd9aeb9fdc9113081b3243f852b7882ec

commit 6d5f163cd9aeb9fdc9113081b3243f852b7882ec
Author: Xiyuan Xia <xiyuan@chromium.org>
Date: Mon Jan 07 18:17:31 2019

SPM: Fix WebDialogView key event not working

WebDialogView uses UnhandledKeyboardEventHandler for unhandled
key events from renderer instead of repost and marking them as
handled.

Bug:  919183 
Change-Id: I753f1892a26764c1bece8b1558985cf31dac7cb3
Reviewed-on: https://chromium-review.googlesource.com/c/1397853
Reviewed-by: Michael Wasserman <msw@chromium.org>
Commit-Queue: Xiyuan Xia <xiyuan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#620382}
[modify] https://crrev.com/6d5f163cd9aeb9fdc9113081b3243f852b7882ec/ui/views/controls/webview/web_dialog_view.cc
[modify] https://crrev.com/6d5f163cd9aeb9fdc9113081b3243f852b7882ec/ui/views/controls/webview/web_dialog_view.h

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/4564429c16a892a4eb67f10b4182fba810b9ae36

commit 4564429c16a892a4eb67f10b4182fba810b9ae36
Author: Xiyuan Xia <xiyuan@chromium.org>
Date: Mon Jan 07 23:48:09 2019

Remove NativeWidgetPrivate::RepostNativeEvent

Remove after fix for  http://crbug.com/919183  because it is
no longer needed.

Bug:  919183 
Change-Id: Iccfff820eb6ddb91618c8a96378176361596ca95
Reviewed-on: https://chromium-review.googlesource.com/c/1399421
Reviewed-by: Scott Violet <sky@chromium.org>
Commit-Queue: Xiyuan Xia <xiyuan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#620534}
[modify] https://crrev.com/4564429c16a892a4eb67f10b4182fba810b9ae36/ui/views/widget/desktop_aura/desktop_native_widget_aura.cc
[modify] https://crrev.com/4564429c16a892a4eb67f10b4182fba810b9ae36/ui/views/widget/desktop_aura/desktop_native_widget_aura.h
[modify] https://crrev.com/4564429c16a892a4eb67f10b4182fba810b9ae36/ui/views/widget/native_widget_aura.cc
[modify] https://crrev.com/4564429c16a892a4eb67f10b4182fba810b9ae36/ui/views/widget/native_widget_aura.h
[modify] https://crrev.com/4564429c16a892a4eb67f10b4182fba810b9ae36/ui/views/widget/native_widget_mac.h
[modify] https://crrev.com/4564429c16a892a4eb67f10b4182fba810b9ae36/ui/views/widget/native_widget_mac.mm
[modify] https://crrev.com/4564429c16a892a4eb67f10b4182fba810b9ae36/ui/views/widget/native_widget_private.h

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/12ee0285b653e848581602a6df24e09a95a274d5

commit 12ee0285b653e848581602a6df24e09a95a274d5
Author: Xiyuan Xia <xiyuan@chromium.org>
Date: Wed Jan 09 00:22:50 2019

Add WebDialogBrowserTest.TextInputViaKeyEvent

Test that key event updates text input in WebDialogView.

Bug:  919183 
Change-Id: I9f1bff1debeae17b5044599bed3f2d5411766fde
Reviewed-on: https://chromium-review.googlesource.com/c/1399702
Commit-Queue: Xiyuan Xia <xiyuan@chromium.org>
Reviewed-by: Michael Wasserman <msw@chromium.org>
Cr-Commit-Position: refs/heads/master@{#620959}
[modify] https://crrev.com/12ee0285b653e848581602a6df24e09a95a274d5/chrome/browser/ui/views/web_dialog_view_browsertest.cc

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/1aa56c256936f433f6e9393e844b33fda6bd7bec

commit 1aa56c256936f433f6e9393e844b33fda6bd7bec
Author: James Cook <jamescook@chromium.org>
Date: Wed Jan 09 15:53:46 2019

Reland "chromeos: Enable SingleProcessMash (in-proc window service) by default"

This reverts commit 908a62bf055bca31313d18dcc31617241df80b69.

Reason for revert: Text input and omnibox issues are fixed.

Original change's description:
> Revert "chromeos: Enable SingleProcessMash (in-proc window service) by default"
>
> This reverts commit badd8a0fab485b41aee70b4ed463868b7e218359.
>
> Reason for revert: Breaks the Wi-Fi network password entry field
>  http://crbug.com/919183 
>
> Original change's description:
> > chromeos: Enable SingleProcessMash (in-proc window service) by default
> >
> > Turns on the window service (code in //services/ws) and uses it to back
> > aura and views (code in //ui/aura/mus and //ui/views/mus). Runs the
> > window service in-process.
> >
> > A label "SingleProcessMash enabled" will appear on the desktop
> > wallpaper to let developers and QA know the feature is enabled.
> > The label will be removed in a separate CL before this feature
> > goes to dev channel. Otherwise there are no visible changes.
> >
> > Flips the former "single_process_mash_*" bot configs to
> > "non_single_process_mash_*" and runs them with the feature disabled.
> > This should keep the old code paths working until we're sure this
> > feature will stick.
> >
> > TBR=sky@chromium.org
> >
> > BUG=918537
> > TEST=single_process_mash_* bots
> >
> > Change-Id: I58a30fd81ea12645b1aabc049c7e697ef1d9e72a
> > Reviewed-on: https://chromium-review.googlesource.com/c/1393467
> > Reviewed-by: Jun Mukai <mukai@chromium.org>
> > Reviewed-by: James Cook <jamescook@chromium.org>
> > Commit-Queue: James Cook <jamescook@chromium.org>
> > Cr-Commit-Position: refs/heads/master@{#619508}
>
> TBR=jamescook@chromium.org,mukai@chromium.org,sky@chromium.org
>
> # Not skipping CQ checks because original CL landed > 1 day ago.
>
> Bug: 918537,  919183 
> Change-Id: Ic3b5f0dd5b941d6383fd7290a0a55b60eedce30d
> Reviewed-on: https://chromium-review.googlesource.com/c/1396702
> Reviewed-by: James Cook <jamescook@chromium.org>
> Reviewed-by: Jun Mukai <mukai@chromium.org>
> Commit-Queue: James Cook <jamescook@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#620103}

TBR=jamescook@chromium.org,mukai@chromium.org,sky@chromium.org

# Not skipping CQ checks because original CL landed > 1 day ago.

Bug: 918537,  919183 
Change-Id: Ie8ec5125558124d756e76d8fdef20e9b3c982e53
Reviewed-on: https://chromium-review.googlesource.com/c/1399296
Reviewed-by: James Cook <jamescook@chromium.org>
Commit-Queue: James Cook <jamescook@chromium.org>
Cr-Commit-Position: refs/heads/master@{#621156}
[modify] https://crrev.com/1aa56c256936f433f6e9393e844b33fda6bd7bec/testing/buildbot/chromium.chromiumos.json
[modify] https://crrev.com/1aa56c256936f433f6e9393e844b33fda6bd7bec/testing/buildbot/chromium.memory.json
[modify] https://crrev.com/1aa56c256936f433f6e9393e844b33fda6bd7bec/testing/buildbot/test_suite_exceptions.pyl
[modify] https://crrev.com/1aa56c256936f433f6e9393e844b33fda6bd7bec/testing/buildbot/test_suites.pyl
[modify] https://crrev.com/1aa56c256936f433f6e9393e844b33fda6bd7bec/ui/base/ui_base_features.cc

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/3ddd534adc33cada51560d862d209f8a3abb1451

commit 3ddd534adc33cada51560d862d209f8a3abb1451
Author: James Cook <jamescook@chromium.org>
Date: Sat Jan 12 00:12:06 2019

Revert "chromeos: Enable SingleProcessMash (in-proc window service) by default""

This reverts commit 1aa56c256936f433f6e9393e844b33fda6bd7bec.

Reason for revert: Various window and input related issues, see
blockers of crbug.com/918537

Original change's description:
> Reland "chromeos: Enable SingleProcessMash (in-proc window service) by default"
> 
> This reverts commit 908a62bf055bca31313d18dcc31617241df80b69.
> 
> Reason for revert: Text input and omnibox issues are fixed.
> 
> Original change's description:
> > Revert "chromeos: Enable SingleProcessMash (in-proc window service) by default"
> >
> > This reverts commit badd8a0fab485b41aee70b4ed463868b7e218359.
> >
> > Reason for revert: Breaks the Wi-Fi network password entry field
> >  http://crbug.com/919183 
> >
> > Original change's description:
> > > chromeos: Enable SingleProcessMash (in-proc window service) by default
> > >
> > > Turns on the window service (code in //services/ws) and uses it to back
> > > aura and views (code in //ui/aura/mus and //ui/views/mus). Runs the
> > > window service in-process.
> > >
> > > A label "SingleProcessMash enabled" will appear on the desktop
> > > wallpaper to let developers and QA know the feature is enabled.
> > > The label will be removed in a separate CL before this feature
> > > goes to dev channel. Otherwise there are no visible changes.
> > >
> > > Flips the former "single_process_mash_*" bot configs to
> > > "non_single_process_mash_*" and runs them with the feature disabled.
> > > This should keep the old code paths working until we're sure this
> > > feature will stick.
> > >
> > > TBR=sky@chromium.org
> > >
> > > BUG=918537
> > > TEST=single_process_mash_* bots
> > >
> > > Change-Id: I58a30fd81ea12645b1aabc049c7e697ef1d9e72a
> > > Reviewed-on: https://chromium-review.googlesource.com/c/1393467
> > > Reviewed-by: Jun Mukai <mukai@chromium.org>
> > > Reviewed-by: James Cook <jamescook@chromium.org>
> > > Commit-Queue: James Cook <jamescook@chromium.org>
> > > Cr-Commit-Position: refs/heads/master@{#619508}
> >
> > TBR=jamescook@chromium.org,mukai@chromium.org,sky@chromium.org
> >
> > # Not skipping CQ checks because original CL landed > 1 day ago.
> >
> > Bug: 918537,  919183 
> > Change-Id: Ic3b5f0dd5b941d6383fd7290a0a55b60eedce30d
> > Reviewed-on: https://chromium-review.googlesource.com/c/1396702
> > Reviewed-by: James Cook <jamescook@chromium.org>
> > Reviewed-by: Jun Mukai <mukai@chromium.org>
> > Commit-Queue: James Cook <jamescook@chromium.org>
> > Cr-Commit-Position: refs/heads/master@{#620103}
> 
> TBR=jamescook@chromium.org,mukai@chromium.org,sky@chromium.org
> 
> # Not skipping CQ checks because original CL landed > 1 day ago.
> 
> Bug: 918537,  919183 
> Change-Id: Ie8ec5125558124d756e76d8fdef20e9b3c982e53
> Reviewed-on: https://chromium-review.googlesource.com/c/1399296
> Reviewed-by: James Cook <jamescook@chromium.org>
> Commit-Queue: James Cook <jamescook@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#621156}

TBR=jamescook@chromium.org,mukai@chromium.org,sky@chromium.org

# Not skipping CQ checks because original CL landed > 1 day ago.

Bug: 918537,  919183 
Change-Id: I3ba65c42be2d1b9747dac544cd73cd2c17f4f000
Reviewed-on: https://chromium-review.googlesource.com/c/1407777
Reviewed-by: James Cook <jamescook@chromium.org>
Commit-Queue: James Cook <jamescook@chromium.org>
Cr-Commit-Position: refs/heads/master@{#622233}
[modify] https://crrev.com/3ddd534adc33cada51560d862d209f8a3abb1451/testing/buildbot/chromium.chromiumos.json
[modify] https://crrev.com/3ddd534adc33cada51560d862d209f8a3abb1451/testing/buildbot/chromium.memory.json
[modify] https://crrev.com/3ddd534adc33cada51560d862d209f8a3abb1451/testing/buildbot/test_suite_exceptions.pyl
[modify] https://crrev.com/3ddd534adc33cada51560d862d209f8a3abb1451/testing/buildbot/test_suites.pyl
[modify] https://crrev.com/3ddd534adc33cada51560d862d209f8a3abb1451/ui/base/ui_base_features.cc