8275715: D3D pipeline processes multiple PaintEvent at initial drawing #6064

masyano · 2021-10-21T12:19:43Z

Could you please review the 8275715 bug fixes?

I think D3DScreenUpdateManager posts unnecessary PaintEvent during processing PaintEvent. When the validate method is called from createGraphics, repaintPeerTarget should not be called.

Progress

Change must not contain extraneous whitespace
Commit message must refer to an issue
Change must be properly reviewed

Issue

JDK-8275715: D3D pipeline processes multiple PaintEvent at initial drawing

Reviewers

Phil Race (@prrace - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.java.net/jdk pull/6064/head:pull/6064
$ git checkout pull/6064

Update a local copy of the PR:
$ git checkout pull/6064
$ git pull https://git.openjdk.java.net/jdk pull/6064/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 6064

View PR using the GUI difftool:
$ git pr show -t 6064

Using diff file

Download this PR as a diff file:
https://git.openjdk.java.net/jdk/pull/6064.diff

bridgekeeper · 2021-10-21T12:21:08Z

👋 Welcome back myano! A progress list of the required criteria for merging this PR into master will be added to the body of your pull request. There are additional pull request commands available for use with this pull request.

openjdk · 2021-10-21T12:23:13Z

@masyano The following label will be automatically applied to this pull request:

client

When this pull request is ready to be reviewed, an "RFR" email will be sent to the corresponding mailing list. If you would like to change these labels, use the /label pull request command.

mlbridge · 2021-10-21T12:27:29Z

Webrevs

01: Full - Incremental (26dcc7b4)
00: Full (fcf36233)

mrserb · 2021-10-23T19:37:26Z

test/jdk/sun/java2d/DirectX/MultiPaintEventTest/MultiPaintEventTest.java

+
+public class MultiPaintEventTest extends Canvas {
+
+    private int count = 0;


Some synchronization is needed, the field is updated on EDT and checked on the main thread.

I fixed a test as it was pointed out to me.

mrserb · 2021-10-23T19:38:35Z

test/jdk/sun/java2d/DirectX/MultiPaintEventTest/MultiPaintEventTest.java

+            if (test.getCount() > 1) {
+                throw new RuntimeException("Processed unnecessary paint().");
+            }
+        } catch (InterruptedException ex) {


Do not ignore an exception

I fixed a test as it was pointed out to me.

mrserb · 2021-10-23T19:40:59Z

src/java.desktop/windows/classes/sun/java2d/d3d/D3DScreenUpdateManager.java

@@ -452,7 +452,7 @@ public void run() {
                        } finally {
                            rq.unlock();
                        }
-                    } else if (!validate(sd)) {
+                    } else if (!validate(sd, true)) {


The comment below says that "the surface will also trigger a repaint", will it be possible we will trigger it inside validate here and then later inside "replaceSurfaceDataLater"?

Yes, this run() is called on "D3D Screen Updater" thread. It is reasonable that a new PaintEvent is posted when SurfaceData is replaced on this thread. I would limit posting new PaintEvent via createGraphics() only.

Probably I should clarify my question, you added a parameter to the validate method and pass the "true" so the "validate" method will post a paint event, but just a few lines below there is a comment that the next code line "sd.getPeer().replaceSurfaceDataLater();" also will post an event. Is the comment outdated, or we will post two of them?

Neither. replaceSurfaceDataLater() is called when validate() fails to post a PaintEvent. So only one of them will be posted.

@mrserb Could you please review and reply to the above comment?

@mrserb Could you please check this pull request?

This is a pretty important block of code and we are about to fork JDK 18 so it should be looked at after that fork.

@mrserb @prrace JDK 18 has been forked, so please resume the review of this fix.

@mrserb @prrace Could you please review the above comments?

masyano · 2021-11-22T09:17:47Z

@mrserb The fixes in JDK-8270874 have been integrated, so please review this PR again.

masyano · 2021-11-29T11:26:52Z

@mrserb Could you please reply to the above comment?

masyano · 2021-12-03T10:23:04Z

@mrserb Could you please check this pull request?

masyano · 2022-01-05T09:51:41Z

@mrserb Could you please check this pull request?

masyano · 2022-01-12T06:50:33Z

@mrserb How long will it take to review this fix?

masyano · 2022-01-17T10:09:55Z

@mrserb Could you please review this PR again?

masyano · 2022-01-26T07:12:19Z

@mrserb Could you tell me the confirmation status of this PR?

masyano · 2022-02-01T08:00:57Z

@mrserb @prrace Could you tell me what I need to do to merge this PR fix?

prrace · 2022-02-06T21:46:37Z

The problem I have with this is that this being in createGraphics() isn't obviously relevant to me.
The repaintPeerTarget() is called only if the surface is lost which means it needs repainting.
Or does surface lost here perhaps mean "surface has not yet been created", because we aren't
yet showing the window contents at all and we really should wait for the first paint event ?

masyano · 2022-02-10T07:39:16Z

@prrace createGraphics() is called to create surface during "the first paint event" processing. And validate() posts new PaintEvent with repaintPeerTarget() when creates surface.

Current stack trace is here:

[1] java.awt.event.PaintEvent. (PaintEvent.java:106)
[2] sun.awt.PaintEventDispatcher.createPaintEvent (PaintEventDispatcher.java:80)
[3] sun.awt.windows.WComponentPeer.postPaintIfNecessary (WComponentPeer.java:892)
[4] sun.awt.windows.WComponentPeer.handlePaint (WComponentPeer.java:886)
[5] sun.java2d.d3d.D3DScreenUpdateManager.repaintPeerTarget (D3DScreenUpdateManager.java:283)
[6] sun.java2d.d3d.D3DScreenUpdateManager.validate (D3DScreenUpdateManager.java:492)
[7] sun.java2d.d3d.D3DScreenUpdateManager.createGraphics (D3DScreenUpdateManager.java:260)
[8] sun.awt.windows.WComponentPeer.getGraphics (WComponentPeer.java:646)
[9] java.awt.Component.getGraphics (Component.java:3,128)
[10] sun.awt.RepaintArea.paint (RepaintArea.java:227)
[11] sun.awt.windows.WComponentPeer.handleEvent (WComponentPeer.java:371)
[12] java.awt.Component.dispatchEventImpl (Component.java:5,056)
[13] java.awt.Container.dispatchEventImpl (Container.java:2,325)
[14] java.awt.Window.dispatchEventImpl (Window.java:2,775)
[15] java.awt.Component.dispatchEvent (Component.java:4,827)
[16] java.awt.EventQueue.dispatchEventImpl (EventQueue.java:772)

Created surface is normally handled in this processing PaintEvent, so I think there is no need to call repaintPeerTarget(). Also ScreenUpdateManager class which is used for GDI rendering (and parent class of D3DScreenUpdateManager) does not post new PaintEvent at createGraphics(). This validate() is called from createGraphics(), so I hope to reduce to create another PaintEvent at this route only.

prrace · 2022-02-14T20:12:39Z

Ok. Please confirm you have run all automated headful jtreg tests on a system with D3D enabled
and that you have also done manual testing (using SwingSet2 and J2Ddemo) and no problems are observed ?

masyano · 2022-02-17T07:32:04Z

I tryed jtreg tests with -a -k:headful :jdk_desktop, then all tests are passed. And all graphical tests are rendered correctly. Also SwingSet2 and J2Ddemo works fine.

masyano · 2022-02-24T06:13:38Z

@prrace Could you please check the above comments and reply?

masyano · 2022-03-02T06:21:15Z

@prrace How long will it take for you to reply to the comment?

openjdk · 2022-03-04T23:12:18Z

@masyano This change now passes all automated pre-integration checks.

ℹ️ This project also has non-automated pre-integration requirements. Please see the file CONTRIBUTING.md for details.

After integration, the commit message for the final commit will be:

8275715: D3D pipeline processes multiple PaintEvent at initial drawing

Reviewed-by: prr

You can use pull request commands such as /summary, /contributor and /issue to adjust it as needed.

At the time when this comment was updated there had been 1595 new commits pushed to the master branch:

bae0d5e: 8236907: JTable added to nested panels does not paint last visible row
b0028a4: 8272853: improve JavadocTester.runTests
e07fd39: 8281181: Do not use CPU Shares to compute active processor count
9c817d3: 8015854: [macosx] JButton's HTML ImageView adding unwanted padding
733c790: 8282081: java.time.DateTimeFormatter: wrong definition of symbol F
f9f9c0a: 8252769: Warn in configure if git config autocrlf has invalid value
603050b: 8282661: [BACKOUT] ByteBufferTest.java: replace endless recursion with RuntimeException in void ck(double x, double y)
5247153: 8282615: G1: Fix some includes
a584c90: 8282573: ByteBufferTest.java: replace endless recursion with RuntimeException in void ck(double x, double y)
d5e8e52: 8282532: Allow explicitly setting build platform alongside --openjdk-target
... and 1585 more: https://git.openjdk.java.net/jdk/compare/c41ce6d159e59a8c05dbeacde2d2612b58733d46...master

As there are no conflicts, your changes will automatically be rebased on top of these commits when integrating. If you prefer to avoid this automatic rebasing, please check the documentation for the /integrate command for further details.

➡️ To integrate this PR with the above commit message to the master branch, type /integrate in a new comment.

masyano · 2022-03-06T23:52:44Z

/integrate

openjdk · 2022-03-06T23:54:02Z

Going to push as commit 415bf44.
Since your change was applied there have been 1599 commits pushed to the master branch:

974ef55: 8282617: sun.net.www.protocol.https.HttpsClient#putInKeepAliveCache() doesn't use a lock while dealing with "inCache" field
bc42e7c: 8282382: Report glibc malloc tunables in error reports
52278b8: 8282694: ProblemList runtime/CommandLine/VMDeprecatedOptions.java
c459f8f: 8282142: [TestCase] compiler/inlining/ResolvedClassTest.java will fail when --with-jvm-features=-compiler1
bae0d5e: 8236907: JTable added to nested panels does not paint last visible row
b0028a4: 8272853: improve JavadocTester.runTests
e07fd39: 8281181: Do not use CPU Shares to compute active processor count
9c817d3: 8015854: [macosx] JButton's HTML ImageView adding unwanted padding
733c790: 8282081: java.time.DateTimeFormatter: wrong definition of symbol F
f9f9c0a: 8252769: Warn in configure if git config autocrlf has invalid value
... and 1589 more: https://git.openjdk.java.net/jdk/compare/c41ce6d159e59a8c05dbeacde2d2612b58733d46...master

Your commit was automatically rebased without conflicts.

openjdk · 2022-03-06T23:54:14Z

@masyano Pushed as commit 415bf44.

💡 You may see a message that your pull request was closed with unmerged commits. This can be safely ignored.

dholmes-ora · 2022-03-07T00:12:41Z

@masyano the new test has an invalid copyright notice: missing comma after 2021, which is causing a tier 1 build failure in our CI.

masyano · 2022-03-07T00:17:43Z

@dholmes-ora Thank you for notice. Should I create a new pull request to correct the copyright?

dholmes-ora · 2022-03-07T00:18:10Z

New bug filed: JDK-8282713 - which will need a new PR. Thanks.

masyano · 2022-03-07T00:56:17Z

I opened new PR at #7716.

prrace · 2022-03-07T02:26:25Z

FWIW this should not have been pushed without a 2nd reviewer (meaning two approvals).
Please review the code review policies section here https://openjdk.java.net/groups/client-libs/
where it says
The standard requirement in Java SE is for one (or more) reviewer prior to code freeze and two (or more) reviewers thereafter. The Java Client Library Group has standardized on two reviewers with few exceptions.

You pushed with only one approval. Bear that in mind !

8275715: D3D pipeline processes multiple PaintEvent at initial drawing

fcf3623

openjdk bot added the rfr Pull request is ready for review label Oct 21, 2021

openjdk bot added the client client-libs-dev@openjdk.org label Oct 21, 2021

mrserb reviewed Oct 23, 2021

View reviewed changes

8275715: D3D pipeline processes multiple PaintEvent at initial drawing

26dcc7b

mrserb mentioned this pull request Nov 10, 2021

8270874: JFrame paint artifacts when dragged from standard monitor to HiDPI monitor #6339

Closed

3 tasks

prrace approved these changes Mar 4, 2022

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Mar 4, 2022

openjdk bot added the integrated Pull request has been integrated label Mar 6, 2022

openjdk bot closed this Mar 6, 2022

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Mar 6, 2022


		public class MultiPaintEventTest extends Canvas {

		private int count = 0;

8275715: D3D pipeline processes multiple PaintEvent at initial drawing #6064

8275715: D3D pipeline processes multiple PaintEvent at initial drawing #6064

Conversation

masyano commented Oct 21, 2021 • edited by openjdk bot Loading

Progress

Issue

Reviewers

Reviewing

bridgekeeper bot commented Oct 21, 2021

openjdk bot commented Oct 21, 2021

mlbridge bot commented Oct 21, 2021 • edited Loading

Webrevs

mrserb Oct 23, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mrserb Oct 28, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

masyano commented Nov 22, 2021

masyano commented Nov 29, 2021

masyano commented Dec 3, 2021

masyano commented Jan 5, 2022

masyano commented Jan 12, 2022

masyano commented Jan 17, 2022

masyano commented Jan 26, 2022

masyano commented Feb 1, 2022

prrace commented Feb 6, 2022

masyano commented Feb 10, 2022

prrace commented Feb 14, 2022

masyano commented Feb 17, 2022

masyano commented Feb 24, 2022

masyano commented Mar 2, 2022

openjdk bot commented Mar 4, 2022 • edited Loading

masyano commented Mar 6, 2022

openjdk bot commented Mar 6, 2022

openjdk bot commented Mar 6, 2022

dholmes-ora commented Mar 7, 2022

masyano commented Mar 7, 2022

dholmes-ora commented Mar 7, 2022

masyano commented Mar 7, 2022

prrace commented Mar 7, 2022

masyano commented Oct 21, 2021 •

edited by openjdk bot

Loading

mlbridge bot commented Oct 21, 2021 •

edited

Loading

mrserb Oct 23, 2021 •

edited

Loading

mrserb Oct 28, 2021 •

edited

Loading

openjdk bot commented Mar 4, 2022 •

edited

Loading