[flatland] Handle fence overflow in flatland_connection.cc #53366
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
filmil
force-pushed
the
fix-flatland-scheduling
branch
2 times, most recently
from
066b1a3
to
5cac8fd
Compare
uysalere
reviewed
Jun 13, 2024
filmil
force-pushed
the
fix-flatland-scheduling
branch
2 times, most recently
from
4fee0e0
to
ddbffb0
Compare
|
@uysalere PTAL - I merged the code of |
filmil
force-pushed
the
fix-flatland-scheduling
branch
2 times, most recently
from
45f0f34
to
2da2d1e
Compare
filmil
force-pushed
the
fix-flatland-scheduling
branch
from
2da2d1e
to
4c0c89f
Compare
uysalere
approved these changes
Jun 13, 2024
flatland_connection.cc used to allow an arbitrary number of acquire and release fences to be scheduled for each frame. Sadly, Fuchsia has a limitation of (1) the number of total handles that can be sent per a FIDL call, but also (2) the Flatland protocol only supports sending up to 16 fences per each fence type. Now, normally there should be very few scheduled fences per frame. But if frames get skipped, we could amass many fences which would then crash our attempts to send all of them to the Flatland `Present` endpoint. This change introduces two fence multiplexer, which allow us to signal more than 16 fences per type, at a performance penalty. We expect to be able *not* to crash the FIDL subsystem using this approach, and may even be able to hobble along for a bit, until the fences issue is hopefully self-resolved. That said, this issue seems to indicate there are frame scheduling problems elsewhere. But this is a fairly straightforward change to make without affecting the rest of the flatland code or integration, so we opt to do that first. Issue: #150136
filmil
force-pushed
the
fix-flatland-scheduling
branch
from
4c0c89f
to
acca647
Compare
chinmaygarde
approved these changes
Jun 13, 2024
|
@chinmaygarde Can you merge this for me? I am not a Flutter committer. Thanks! |
chinmaygarde
added
the
autosubmit
|
Happy to endorse. Your call. go/flutter-onboarding. |
I don't think I'm at a point where flutter onboarding makes sense for me. Thanks for the pointer, however. |
engine-flutter-autoroll
added a commit
to engine-flutter-autoroll/flutter
that referenced
this pull request
Jun 17, 2024
engine-flutter-autoroll
added a commit
to engine-flutter-autoroll/flutter
that referenced
this pull request
Jun 17, 2024
engine-flutter-autoroll
added a commit
to engine-flutter-autoroll/flutter
that referenced
this pull request
Jun 17, 2024
engine-flutter-autoroll
added a commit
to engine-flutter-autoroll/flutter
that referenced
this pull request
Jun 18, 2024
auto-submit bot
pushed a commit
to flutter/flutter
that referenced
this pull request
Jun 18, 2024
…150399) flutter/engine@a4f266f...1c4e5e2 2024-06-17 flar@google.com [DisplayList] Create DrawDashedLine for paragraph code (flutter/engine#53411) 2024-06-17 chinmaygarde@google.com [Impeller] Update Android CPU profiling instructions. (flutter/engine#53437) 2024-06-17 jonahwilliams@google.com [Impeller] move draw vertices to dl unittests. (flutter/engine#53400) 2024-06-17 chinmaygarde@google.com [Impeller] Link CPU profiling docs from the main README. (flutter/engine#53435) 2024-06-17 skia-flutter-autoroll@skia.org Roll Dart SDK from 1ce6b4d54247 to 51bbba0da7d3 (1 revision) (flutter/engine#53432) 2024-06-17 jonahwilliams@google.com [Impeller] Move drawAtlas golden tests to display list. (flutter/engine#53398) 2024-06-17 fmil@google.com [flatland] Handle fence overflow in flatland_connection.cc (flutter/engine#53366) 2024-06-17 skia-flutter-autoroll@skia.org Roll Skia from 0dda1054f543 to 2b9bc16df969 (1 revision) (flutter/engine#53431) If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/flutter-engine-flutter-autoroll Please CC chinmaygarde@google.com,rmistry@google.com,zra@google.com on the revert to ensure that a human is aware of the problem. To file a bug in Flutter: https://github.com/flutter/flutter/issues/new/choose To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
flatland_connection.cc used to allow an arbitrary number of acquire and release fences to be scheduled for each frame.
Sadly, Fuchsia has a limitation of (1) the number of total handles that can be sent per a FIDL call, but also (2) the Flatland protocol only supports sending up to 16 fences per each fence type.
Now, normally there should be very few scheduled fences per frame. But if frames get skipped, we could amass many fences which would then crash our attempts to send all of them to the Flatland
Presentendpoint.This change introduces two fence multiplexer, which allow us to signal more than 16 fences per type, at a performance penalty. We expect to be able not to crash the FIDL subsystem using this approach, and may even be able to hobble along for a bit, until the fences issue is hopefully self-resolved.
That said, this issue seems to indicate there are frame scheduling problems elsewhere. But this is a fairly straightforward change to make without affecting the rest of the flatland code or integration, so we opt to do that first.
Issues: #150136
Pre-launch Checklist
///).