1741776 - Hit MOZ_CRASH(bug: this is an unexpected case - please open a bug and talk to #gfx team!) at gfx/wr/webrender/src/spatial_tree.rs:957

Status: VERIFIED FIXED

(Core :: Graphics: WebRender, defect, P3)

Description

[Tyson Smith [:tsmith]]

Attachment: testcase.html

Found while fuzzing m-c 20211109-1bedbf784ec9 (--enable-address-sanitizer --enable-fuzzing)

Hit MOZ_CRASH(bug: this is an unexpected case - please open a bug and talk to #gfx team!) at gfx/wr/webrender/src/spatial_tree.rs:957

#0 0x7f7106321360 in MOZ_Crash /builds/worker/workspace/obj-build/dist/include/mozilla/Assertions.h:256:3
#1 0x7f7106321360 in RustMozCrash /gecko/mozglue/static/rust/wrappers.cpp:18:3
#2 0x7f7106321296 in mozglue_static::panic_hook::h183adc4d73b027cc /gecko/mozglue/static/rust/lib.rs:91:9
#3 0x7f7106320615 in core::ops::function::Fn::call::h2c49d8cefb0980e2 /builds/worker/fetches/rust/library/core/src/ops/function.rs:70:5
#4 0x7f7107e99587 in std::panicking::rust_panic_with_hook::hd83d5a96a789e1d3 (/home/worker/builds/m-c-20211109093712-fuzzing-asan-opt/libxul.so+0x18407587)
#5 0x7f7104dde5e8 in std::panicking::begin_panic::_$u7b$$u7b$closure$u7d$$u7d$::h5cffcd3c09905dfb /builds/worker/fetches/rust/library/std/src/panicking.rs:544:9
#6 0x7f7104db36d9 in std::sys_common::backtrace::__rust_end_short_backtrace::h1a31b5ce359ddaac /builds/worker/fetches/rust/library/std/src/sys_common/backtrace.rs:141:18
#7 0x7f70f4586f9c in std::panicking::begin_panic::hf2a51ed4961b214a /builds/worker/fetches/rust/library/std/src/panicking.rs:543:12
#8 0x7f710503ed4e in webrender::spatial_tree::SpatialTree::get_relative_transform_with_face::h8626d9d16120c196 /gecko/gfx/wr/webrender/src/spatial_tree.rs:957:9
#9 0x7f7105024cc0 in webrender::space::SpaceMapper$LT$F$C$T$GT$::set_target_spatial_node::h9a8da5cafc0dbf05 /gecko/gfx/wr/webrender/src/space.rs:73:29
#10 0x7f710513e233 in webrender::picture::PicturePrimitive::propagate_bounding_rect::hd4a52d6d699d4e94 /gecko/gfx/wr/webrender/src/picture.rs:6353:13
#11 0x7f710515361a in webrender::picture_graph::PictureGraph::propagate_bounding_rects::h0aca73798e40e340 /gecko/gfx/wr/webrender/src/picture_graph.rs:136:17
#12 0x7f710509ee3b in webrender::frame_builder::FrameBuilder::build_layer_screen_rects_and_cull_layers::hca85e550d09a2fb9 /gecko/gfx/wr/webrender/src/frame_builder.rs:389:9
#13 0x7f710509ee3b in webrender::frame_builder::FrameBuilder::build::hcfceb1ae26a610c2 /gecko/gfx/wr/webrender/src/frame_builder.rs:584:9
#14 0x7f71051ad082 in webrender::render_backend::Document::build_frame::h025c3bb47ae09797 /gecko/gfx/wr/webrender/src/render_backend.rs:453:25
#15 0x7f71051dcd97 in webrender::render_backend::RenderBackend::update_document::h1efe661c88788063 /gecko/gfx/wr/webrender/src/render_backend.rs:1357:41
#16 0x7f71051c7e9f in webrender::render_backend::RenderBackend::prepare_transactions::hec4e32534e742cd1 /gecko/gfx/wr/webrender/src/render_backend.rs:1206:28
#17 0x7f71051c7e9f in webrender::render_backend::RenderBackend::process_api_msg::h4ea01fe812ce4221 /gecko/gfx/wr/webrender/src/render_backend.rs:1058:17
#18 0x7f7104db6b36 in webrender::render_backend::RenderBackend::run::hcd5c20e118240471 /gecko/gfx/wr/webrender/src/render_backend.rs:722:21
#19 0x7f7104db6b36 in webrender::renderer::Renderer::new::_$u7b$$u7b$closure$u7d$$u7d$::hc32fd3dfe39adfde /gecko/gfx/wr/webrender/src/renderer/mod.rs:1325:13
#20 0x7f7104db6b36 in std::sys_common::backtrace::__rust_begin_short_backtrace::h707c1c3cd5edb57f /builds/worker/fetches/rust/library/std/src/sys_common/backtrace.rs:125:18
#21 0x7f7104e0bcde in std::thread::Builder::spawn_unchecked::_$u7b$$u7b$closure$u7d$$u7d$::_$u7b$$u7b$closure$u7d$$u7d$::hd8f10fa91d3a21d4 /builds/worker/fetches/rust/library/std/src/thread/mod.rs:481:17
#22 0x7f7104e0bcde in _$LT$core..panic..unwind_safe..AssertUnwindSafe$LT$F$GT$$u20$as$u20$core..ops..function..FnOnce$LT$$LP$$RP$$GT$$GT$::call_once::hb8fb6945e3d0be79 /builds/worker/fetches/rust/library/core/src/panic/unwind_safe.rs:271:9
#23 0x7f7104e0bcde in std::panicking::try::do_call::h347864793df2690f /builds/worker/fetches/rust/library/std/src/panicking.rs:403:40
#24 0x7f7104e0bcde in std::panicking::try::he1f3d83b25eca1a7 /builds/worker/fetches/rust/library/std/src/panicking.rs:367:19
#25 0x7f7104e0bcde in std::panic::catch_unwind::h4f630a3d4d41bb39 /builds/worker/fetches/rust/library/std/src/panic.rs:129:14
#26 0x7f7104e0bcde in std::thread::Builder::spawn_unchecked::_$u7b$$u7b$closure$u7d$$u7d$::h88c49cf0f29b96f2 /builds/worker/fetches/rust/library/std/src/thread/mod.rs:480:30
#27 0x7f7104e0bcde in core::ops::function::FnOnce::call_once$u7b$$u7b$vtable.shim$u7d$$u7d$::h6e87c90d8c20f292 /builds/worker/fetches/rust/library/core/src/ops/function.rs:227:5
#28 0x7f7107ea0102 in std::sys::unix::thread::Thread::new::thread_start::h3b1213720f18b702 std.ac3adaa7-cgu.2
#29 0x7f7114147608 in start_thread /build/glibc-eX1tMB/glibc-2.31/nptl/pthread_create.c:477:8
#30 0x7f7113d0f292 in __clone /build/glibc-eX1tMB/glibc-2.31/misc/../sysdeps/unix/sysv/linux/x86_64/clone.S:95

Comment 1

[Tyson Smith [:tsmith]]

https://crash-stats.mozilla.org/report/index/b01100a4-d21c-4cfe-ab9e-10de30211118

Comment 2

[Tyson Smith [:tsmith]]

A Pernosco session is available here: https://pernos.co/debug/bZ3fZF4LgUiQVo7D3lBTqQ/index.html

Comment 3

[Bugmon [:jkratzer for issues]]

Bugmon Analysis
Verified bug as reproducible on mozilla-central 20211118034937-f172e803d80c.
The bug appears to have been introduced in the following build range:

Start: 3339b501299addf81885d685c1ae1c139e10831b (20210927222840)
End: 72cb687968f86105ec9220c788b4af83240a2297 (20210928000602)
Pushlog: https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=3339b501299addf81885d685c1ae1c139e10831b&tochange=72cb687968f86105ec9220c788b4af83240a2297

Comment 4

[Miko Mynttinen]

Seems like someone actually did as the message requested. :)

Comment 5

[Glenn Watson [:gw]]

I can reproduce the assertion failure. I suspect it's a bug in how Gecko is producing display lists, but I don't understand the test case very well - so I have some questions!

If I disable that assert, run the page, then save page, and remove the JS, it looks like:

<html>
  <head>
    <meta http-equiv="content-type" content="text/html; charset=UTF-8">
    <style>
      #a {
        rotate: 1deg 1 0 44;
        filter: drop-shadow(81px 6px 0px -moz-mac-menushadow);
      }
    </style>
  </head>
  <body>
    <select id="a" multiple="multiple">
      <option id="b">x
        <content id="c">x
          <ol id="d" style="position: fixed;">x</ol>
        </content>
      </option>
    </select>
  </body>
</html>

If I then load that page, it doesn't trigger the assert. Looking in devtools, the content element is missing. I'm assuming this is because it's not a legal element to have within an option tag, would that be right?

If that's the case, should Gecko be preventing JS from appending the content element to the option, maybe that's the correct fix?

I suspect what's happening is that the JS is producing a DOM structure that's not expected, and that the Gecko DL building code is then producing an invalid display list, since it doesn't expect to see a fixed position element within a select.

NI'ing a couple people who might be able to answer those questions.

Comment 6

[Emilio Cobos Álvarez (:emilio)]

We shouldn't be creating frames for random elements inside <select> as per this code, so yeah that might be the first place to dig into.

Comment 7

[Glenn Watson [:gw]]

Un-assigning for now, Emilio will take a look when time allows. Feel free to re-assign to me if it does still look like a gfx/wr issue.

Comment 8

[Emilio Cobos Álvarez (:emilio)]

So on a debug build I get this assert. So yeah at the very least we're creating a malformed display-list.

Comment 9

[Emilio Cobos Álvarez (:emilio)]

Attachment: Bug 1741776 - Make sure <option> is a containing-block for positioned descendants. r=#layout-reviewers

We have other code to make sure that options themselves are in-flow, so
we really want abspos boxes not to escape the select element.

Other browsers don't allow having out-of-flow option children at all
(they seem to force descendants to be in-flow). However that seems
fairly more complicated to implement, and I don't think it particularly
matters how we deal with this as long as we deal with it in a
sorta-reasonable way.

Comment 10

[BugBot [:suhaib / :marco/ :calixte]]

:emilio, since this bug contains a bisection range, could you fill (if possible) the regressed_by field?
For more information, please visit auto_nag documentation.

Comment 11

[Emilio Cobos Álvarez (:emilio)]

Technically bug 1732596, but it likely caused issues before...

Comment 12

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/1f2d586e5968
Make sure <option> is a containing-block for positioned descendants. r=miko

Comment 13

[Web Platform Test Sync Bot (Matrix: #interop:mozilla.org)]

Created web-platform-tests PR https://github.com/web-platform-tests/wpt/pull/31721 for changes under testing/web-platform/tests

Comment 14

[Atila Butkovits]

Backed out for causing failures at test_bug858459.html.

Backout link: https://hg.mozilla.org/integration/autoland/rev/7902135e6c944c55062ebc721e049b88d5f507af

Push with failures: https://treeherder.mozilla.org/jobs?repo=autoland&selectedTaskRun=Ybdo5P78T8S5cOj6abEEjg.0&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel&revision=1f2d586e5968cea4c31a646177bfc65a521bae22

Failure log: https://treeherder.mozilla.org/logviewer?job_id=358994495&repo=autoland&lineNumber=7901

Comment 15

[Web Platform Test Sync Bot (Matrix: #interop:mozilla.org)]

Upstream PR was closed without merging

Comment 16

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/4a51ad7f61fb
Make sure <option> is a containing-block for positioned descendants. r=miko

Comment 17

[Iulian Moraru]

https://hg.mozilla.org/mozilla-central/rev/4a51ad7f61fb

Comment 18

[Bugmon [:jkratzer for issues]]

Bugmon Analysis
Verified bug as fixed on rev mozilla-central 20211125043756-0bfe7aadbc81.
Removing bugmon keyword as no further action possible. Please review the bug and re-add the keyword for further analysis.

Comment 19

[Web Platform Test Sync Bot (Matrix: #interop:mozilla.org)]

Upstream PR merged by moz-wptsync-bot