Raf delivery: run rafs for all pipeline if tick received for any.#33395
Raf delivery: run rafs for all pipeline if tick received for any.#33395jdm merged 2 commits intoservo:mainfrom
Conversation
|
🔨 Triggering try run (#10790452232) for Linux WPT |
49003c1 to
c5e5c16
Compare
|
|
c5e5c16 to
ebb481f
Compare
9725b09 to
63f1835
Compare
|
🔨 Triggering try run (#10792054606) for Linux WPT |
|
Test results for linux-wpt-layout-2020 from try job (#10792054606): Flaky unexpected result (15)
Stable unexpected results that are known to be intermittent (13)
|
|
✨ Try run (#10792054606) succeeded. |
63f1835 to
1557c7f
Compare
|
🔨 Triggering try run (#10793278423) for Linux WPT |
|
Test results for linux-wpt-layout-2020 from try job (#10793278423): Flaky unexpected result (22)
Stable unexpected results that are known to be intermittent (11)
|
|
✨ Try run (#10793278423) succeeded. |
|
Interesting
|
I think this can happen because the test recursively requests rafs: and we prioritize the "update the rendering task" versus other tasks, including the timeout used to signal that the test is finished. So this can result in many runs of the prioritized update the rendering task until finally the script-thread runs the sequential timeout. Will try to address in a commit. |
|
🔨 Triggering try run (#10810966797) for Linux WPT |
|
Test results for linux-wpt-layout-2020 from try job (#10810966797): Flaky unexpected result (17)
Stable unexpected results that are known to be intermittent (15)
|
|
✨ Try run (#10810966797) succeeded. |
|
🔨 Triggering try run (#10814358552) for Linux WPT |
|
Test results for linux-wpt-layout-2020 from try job (#10814358552): Flaky unexpected result (26)
Stable unexpected results that are known to be intermittent (9)
|
|
✨ Try run (#10814358552) succeeded. |
… any Signed-off-by: gterzian <2792687+gterzian@users.noreply.github.com>
6a51741 to
84ef307
Compare
Signed-off-by: gterzian <2792687+gterzian@users.noreply.github.com>
84ef307 to
b3e56dc
Compare
jdm
left a comment
There was a problem hiding this comment.
Looks like a reasonable solution!
|
This seems to have caused an intermittent failure: #33413 |
The first commits ensures the ordering of running rafs between parent and child docs, by running rafs for all pipeline if a tick was received for any(see #33251).
The second commits prevents a sequence of "update the rendering" tasks from running one after the other and starving other script tasks(see #33395 (comment)).
For the first commit there is a TLA spec modelling the idea at https://gist.github.com/gterzian/1f297aa22393ea8604ad14ca6a5cfff6
./mach build -ddoes not report any errors./mach test-tidydoes not report any errors/html/webappapis/update-rendering/child-document-raf-order.html#33028 (GitHub issue number if applicable)