Consolidate devtools frame state bits into nsIFrame
Categories
(DevTools :: Inspector, task)
Tracking
(firefox79 fixed)
Tracking | Status | |
---|---|---|
firefox79 | --- | fixed |
People
(Reporter: bradwerth, Assigned: bradwerth)
References
Details
Attachments
(1 file)
(deleted),
text/x-phabricator-request
|
Details |
Devtools uses frame state bits to trigger the generation of "computed info" on some reflows. The computed info structures are used by devtools to inspect details of the layout process. Two of these bits are NS_STATE_FLEX_GENERATE_COMPUTED_VALUES
and NS_STATE_GRID_GENERATE_COMPUTED_VALUES
. These bits should be combined into a single bit value and moved to nsIFrame.h, which has more space for frame bits.
Assignee | ||
Comment 1•4 years ago
|
||
This patch takes the flags NS_STATE_GRID_GENERATE_COMPUTED_VALUES and
NS_STATE_FLEX_GENERATE_COMPUTED_VALUES and combines them into a single new
flag that is defined in nsIFrame.h, with accessors.
Comment 3•4 years ago
|
||
Backed out for crashes on nsGridContainerFrame
backout: https://hg.mozilla.org/integration/autoland/rev/9767e198bdee8308d820a3747f9df81ebfcfe7a1
push with other related crashes: https://treeherder.mozilla.org/#/jobs?repo=autoland&group_state=expanded&revision=bf91e9e8455eda80299902401c930f085e85b002&searchStr=reftest&selectedTaskRun=AbPg8gltRBC3Argz1k1bRw-0
failure log: https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=305723690&repo=autoland&lineNumber=1977
Comment 4•4 years ago
|
||
Also crashes on [@ InvalidArrayIndex_CRASH(unsigned long, unsigned long)] on multiple different files e.g. https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=305725392&repo=autoland&lineNumber=14995
Assignee | ||
Comment 5•4 years ago
|
||
Something is tripping the bitfield. I'm having trouble debugging it since it's tough to break on a bitfield memory change. I'll figure it out.
Assignee | ||
Comment 6•4 years ago
|
||
The test that reproduces the crash for me locally is layout/generic/crashtests/1278461-2.html.
Assignee | ||
Comment 7•4 years ago
|
||
(In reply to Brad Werth [:bradwerth] from comment #5)
Something is tripping the bitfield. I'm having trouble debugging it since it's tough to break on a bitfield memory change. I'll figure it out.
Nope, I just forgot to initialize the new bit in the constructor. Fix inbound.
Comment 9•4 years ago
|
||
bugherder |
Assignee | ||
Updated•4 years ago
|
Description
•