Frequent crash (diagnostic assert) in [@ mozilla::nsDisplayItem::GetOldListIndex] in chatgpt.com
Categories
(Core :: Web Painting, defect, P2)
Tracking
()
| Tracking | Status | |
|---|---|---|
| firefox-esr115 | --- | unaffected |
| firefox-esr128 | --- | unaffected |
| firefox138 | --- | unaffected |
| firefox139 | --- | disabled |
| firefox140 | + | fixed |
People
(Reporter: aryx, Assigned: emilio)
References
(Blocks 2 open bugs, Regression)
Details
(5 keywords, Whiteboard: [viewtransitions:m1])
Crash Data
Attachments
(1 file)
43 reports from 26 installs of Firefox 139.0a1 with build IDs 20250425212424 and 20250426091604 on all desktop OS. Only few crash reporters shared the url but it was always ChatGPT. I could not reproduce with a basic query.
Push log in which the regression has likely started.
Emilio: Is this a regression from enabling view transitions in bug 1950759?
Crash report: https://crash-stats.mozilla.org/report/index/7e8055c4-2d10-4fb3-bdef-0c7530250426
MOZ_CRASH Reason:
Item found was in the wrong list! type 20 (outer type was 21 at depth 6, now is 21)
Top 10 frames:
0 xul.dll MOZ_Crash(char const*, int, char const*) mfbt/Assertions.h:381
0 xul.dll mozilla::nsDisplayItem::GetOldListIndex(mozilla::nsDisplayList*, unsigned int... layout/painting/nsDisplayList.h:2218
0 xul.dll mozilla::MergeState::HasMatchingItemInOldList(mozilla::nsDisplayItem*, mozill... layout/painting/RetainedDisplayListBuilder.cpp:633
0 xul.dll mozilla::MergeState::ProcessItemFromNewList(mozilla::nsDisplayItem*, mozilla:... layout/painting/RetainedDisplayListBuilder.cpp:460
0 xul.dll mozilla::RetainedDisplayListBuilder::MergeDisplayLists(mozilla::nsDisplayList... layout/painting/RetainedDisplayListBuilder.cpp:835
1 xul.dll mozilla::MergeState::MergeChildLists(mozilla::nsDisplayItem*, mozilla::nsDisp... layout/painting/RetainedDisplayListBuilder.cpp:509
1 xul.dll mozilla::MergeState::ProcessItemFromNewList(mozilla::nsDisplayItem*, mozilla:... layout/painting/RetainedDisplayListBuilder.cpp:481
1 xul.dll mozilla::RetainedDisplayListBuilder::MergeDisplayLists(mozilla::nsDisplayList... layout/painting/RetainedDisplayListBuilder.cpp:835
2 xul.dll mozilla::MergeState::MergeChildLists(mozilla::nsDisplayItem*, mozilla::nsDisp... layout/painting/RetainedDisplayListBuilder.cpp:509
2 xul.dll mozilla::MergeState::ProcessItemFromNewList(mozilla::nsDisplayItem*, mozilla:... layout/painting/RetainedDisplayListBuilder.cpp:481
| Assignee | ||
Comment 1•6 months ago
|
||
Seems likely (thus nightly only).
Jason, ni?ing just in case you have a test case since I suspect fuzzers will hit it relatively soon
Comment 2•6 months ago
|
||
Have the fuzzers been trained to produce view transitions testcases?
| Assignee | ||
Comment 3•6 months ago
|
||
(In reply to Timothy Nikkel (:tnikkel) from comment #2)
Have the fuzzers been trained to produce view transitions testcases?
Afaiui yeah (I checked a while ago and I got told that once the pref gets turned on they should start hitting it).
Comment 4•6 months ago
|
||
Emilio, is this not the same issue as bug 1936080?
Comment 5•6 months ago
|
||
If not, I have another testcase the produces the nearly the same assertion mentioned in comment 0.
Hit MOZ_CRASH(Item found was in the wrong list! type 20 (outer type was 21 at depth 2, now is 2)) at /builds/worker/checkouts/gecko/layout/painting/nsDisplayList.h:2221
| Assignee | ||
Comment 6•6 months ago
|
||
Hmm, not quite, we want something that requires document.startViewTransition(), otherwise it's likely not to be the same root cause...
Comment 7•6 months ago
|
||
(In reply to Jason Kratzer [:jkratzer] from comment #4)
Emilio, is this not the same issue as bug 1936080?
The testcase in that bug uses overflow-clip-box: content-box which is not something we ever plan on enabling for content for users.
Comment 8•6 months ago
|
||
:emilio do you plan on landing a fix for this? Or, should we disable it for now since it is a top crash in nighty?
Comment 9•6 months ago
|
||
I can reliably reproduce this crash in Nightly 140.0a1 (2025-04-29) on macOS: log into chatgpt.com and start a new chat.
| Assignee | ||
Comment 10•6 months ago
|
||
Would be great to have a reproducible test-case here but will try to debug without it.
Comment 11•6 months ago
|
||
Pernosco session incoming...
Comment 12•6 months ago
|
||
A Pernosco session is available here: https://pernos.co/debug/n3RdGY8BGsGLfS7uWHzbog/index.html
Comment 13•5 months ago
|
||
The bug is linked to a topcrash signature, which matches the following criteria:
- Top 10 desktop browser crashes on nightly
- Top 10 AArch64 and ARM crashes on nightly
For more information, please visit BugBot documentation.
| Assignee | ||
Comment 14•5 months ago
|
||
Since they create a stacking context.
| Assignee | ||
Updated•5 months ago
|
Comment 15•5 months ago
|
||
Updated•5 months ago
|
Comment 17•5 months ago
|
||
| bugherder | ||
Updated•5 months ago
|
Description
•