1790665 - Add some diagnostic asserts to `PresShell::MarkFramesInSubtreeApproximatelyVisible`

Status: RESOLVED FIXED

(Core :: Layout, task)

Description

[Daniel Holbert [:dholbert]]

The crashes in bug 1708808 seem to be hitting garbage data (mix of null, 0xfff..., and near-null addresses).

Some appear to be "true" null derefs, which we can at least assert about, which we can perhaps use to gather a bit more information if the assertions fail.

Let's add a couple diagnostic asserts to MarkFramesInSubtreeApproximatelyVisible to validate that child-frame pointers in that function are non-null, to make assumptions a bit more explicit and perhaps get a bit closer to understanding what's going on.

Comment 1

[Daniel Holbert [:dholbert]]

Attachment: Bug 1790665: Add a diagnostic assert to PresShell::MarkFramesInSubtreeApproximatelyVisible. r?tnikkel

This API only should never receive a null frame pointer (and child frames
should never be null); but we occasionally get crash reports in this function
where a null frame pointer seems to be involved. This patch's diagnostic
asserts will help us catch these crashes slightly earlier and more directly,
and might help us reason about what's going on a bit better.

Comment 2

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/5816896351a6
Add a diagnostic assert to PresShell::MarkFramesInSubtreeApproximatelyVisible. r=tnikkel

Comment 3

[Daniel Holbert [:dholbert]]

Attachment: Bug 1790665 followup: Add another diagnostic assert to PresShell::MarkFramesInSubtreeApproximatelyVisible. rs=tnikkel

(This assert should be trivially satisfied, but nonetheless we get crash
reports that suggest we have null pointers here, so let's see...)

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/ea0aad9868d4
followup: Add another diagnostic assert to PresShell::MarkFramesInSubtreeApproximatelyVisible. rs=tnikkel

Comment 5

[Marian-Vasile Laza]

https://hg.mozilla.org/mozilla-central/rev/5816896351a6
https://hg.mozilla.org/mozilla-central/rev/ea0aad9868d4

Comment 6

[Timothy Nikkel (:tnikkel)]

First crash in a build with these asserts
https://crash-stats.mozilla.org/report/index/4ff8b3ed-b415-41bd-9685-f06460221002
EXCEPTION_ACCESS_VIOLATION_EXEC at 0x65746e49, so not one of the near null crashes.