1307237 - 11,600 instances of "'!IsSelectionValid()'" and associated warnings emitted from widget/ContentCache.cpp during debug testing

Status: NEW

(Core :: Widget, defect)

Description

[Eric Rahm [:erahm]]

> 5835 WARNING: '!IsSelectionValid()', file widget/ContentCache.cpp, line 651

This warning [1] shows up in the following test suites:

>   2724 - Mochitest e10s 2
>   1398 - Mochitest e10s 5
>   1296 - Mochitest e10s 4
>    162 - Mochitest e10s DevTools Browser Chrome dt4
>    126 - Mochitest e10s 3
>     66 - Mochitest e10s Browser Chrome bc7
>     36 - Mochitest e10s Browser Chrome bc4
>     27 - Mochitest e10s 1

It shows up in 35 tests. A few of the most prevalent:

>   1800 - [e10s] dom/html/test/forms/test_input_range_rounding.html
>   1080 - [e10s] layout/base/tests/test_scroll_snapping.html
>    960 - [e10s] layout/forms/test/test_bug935876.html
>    576 - [e10s] dom/html/test/forms/test_input_range_key_events.html
>    171 - [e10s] layout/base/tests/test_reftests_with_caret.html
>    126 - [e10s] dom/html/test/forms/test_input_radio_radiogroup.html
>    126 - [e10s] dom/tests/mochitest/bugs/test_bug265203.html
>    120 - [e10s] layout/forms/test/test_bug563642.html
>    120 - [e10s] layout/forms/test/test_bug365410.html
>    108 - [e10s] devtools/client/inspector/test/browser_inspector_highlighter-eyedropper-events.js

[1] https://hg.mozilla.org/mozilla-central/annotate/955840bfd3c2/widget/ContentCache.cpp#l651

Comment 1

[Eric Rahm [:erahm]]

This correlates 1:1 with a IPC warning as well and appears to be e10s only. First we see "WARNING: '!aEvent.mSucceeded'" and then "WARNING: '!IsSelectionValid()'".

Masayuki I think this is generally related to code you're interested in.

> 5835 WARNING: '!aEvent.mSucceeded', file dom/ipc/TabParent.cpp, line 2246

This warning [1] shows up in the following test suites:

>   2724 - Mochitest e10s 2
>   1398 - Mochitest e10s 5
>   1296 - Mochitest e10s 4
>    162 - Mochitest e10s DevTools Browser Chrome dt4
>    126 - Mochitest e10s 3
>     66 - Mochitest e10s Browser Chrome bc7
>     36 - Mochitest e10s Browser Chrome bc4
>     27 - Mochitest e10s 1

It shows up in 35 tests. A few of the most prevalent:

>   1800 - [e10s] dom/html/test/forms/test_input_range_rounding.html
>   1080 - [e10s] layout/base/tests/test_scroll_snapping.html
>    960 - [e10s] layout/forms/test/test_bug935876.html
>    576 - [e10s] dom/html/test/forms/test_input_range_key_events.html
>    171 - [e10s] layout/base/tests/test_reftests_with_caret.html
>    126 - [e10s] dom/html/test/forms/test_input_radio_radiogroup.html
>    126 - [e10s] dom/tests/mochitest/bugs/test_bug265203.html
>    120 - [e10s] layout/forms/test/test_bug563642.html
>    120 - [e10s] layout/forms/test/test_bug365410.html
>    108 - [e10s] devtools/client/inspector/test/browser_inspector_highlighter-eyedropper-events.js

[1] https://hg.mozilla.org/mozilla-central/annotate/955840bfd3c2/dom/ipc/TabParent.cpp#l2246

Comment 2

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Probably, a bug of TextInputHandler (or its related classes) is detected.

I guess that before preparing selection cache, TextInputHandler receives some notifications which causes to query the contents.

Comment 3

[Hiroyuki Ikezoe (:hiro)]

I see this warning while running browser/base/content/test/performance/browser_windowopen_reflows.js on Linux.

Comment 4

[Hiroyuki Ikezoe (:hiro)]

I am seeing this warning and "'!mContentCache.HandleQueryContentEvent( aEvent, textInputHandlingWidget)', file /home/hiro/opt/dom/ipc/BrowserParent.cpp:304" on a test case attached in bug 1692705 during pressing arrow keys.

Is this warning still worth spewing?

Comment 5

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

(In reply to Hiroyuki Ikezoe (:hiro) from comment #4)

I am seeing this warning and "'!mContentCache.HandleQueryContentEvent( aEvent, textInputHandlingWidget)', file /home/hiro/opt/dom/ipc/BrowserParent.cpp:304" on a test case attached in bug 1692705 during pressing arrow keys.

Is this warning still worth spewing?

It's odd. The test case does not have editable content, but somebody tries to retrieve the selection? (But I guess that it's valid case that there is no selection range.)

Comment 6

[Hiroyuki Ikezoe (:hiro)]

It comes from https://searchfox.org/mozilla-central/rev/01adc17c9a41d9f7975de170acc78634bd743609/dom/ipc/BrowserParent.cpp#2018.

Comment 7

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Thanks. It means that the handling has a bug. If scrollable element has focus for scrolling without selection ranges, this does not return correct writing mode for the key navigation.