8252372: Check if cloning is required to move loads out of loops in PhaseIdealLoop::split_if_with_blocks_post() #3689

rwestrel · 2021-04-26T09:46:18Z

Sinking data nodes out of a loop when all uses are out of a loop has
several issues that this attempts to fix.

1- Only non control uses are considered which makes little sense (why
not sink if the data node is an argument to a call or a returned
value?)

2- Sinking of Loads is broken because of the handling of
anti-dependence: the get_late_ctrl(n, n_ctrl) call returns a control
in the loop because it takes all uses into account.

3- For data nodes for which a control edge can't be set, commoning of
clones back in the loop is prevented with:
_igvn._worklist.yank(x);
which gives no guarantee

This patch tries to address all issues:

1- it looks at all uses, not only non control uses

2- anti-dependences are computed for each use independently

3- Cast nodes are used to pin clones out of loop

2- requires refactoring of the PhaseIdealLoop::get_late_ctrl()
logic. While working on this, I noticed a bug in anti-dependence
analysis: when the use is a cfg node, the code sometimes looks at uses
of the memory state of the cfg. The logic uses the use of the cfg
which is a projection of adr_type identical to the cfg. It should
instead look at the use of the memory projection.

The existing logic for sinking loads calls clear_dom_lca_tags() for
every load which seems like quite a waste. I added a
_dom_lca_tags_round variable that's or'ed with the tag_node's _idx. By
incrementing _dom_lca_tags_round, new tags that don't conflict with
existing ones are produced and there's no need for
clear_dom_lca_tags().

For anti-dependence analysis to return a correct result, early control
of the load is needed. The only way to get it at this stage, AFAICT,
is to compute it by following the load's input until a pinned node is
reached.

The existing logic pins cloned nodes next to their use. The logic I
propose pins them right out of the loop. This could possibly avoid
some redundant clones. It also makes some special handling for corner
cases with loop strip mining useless.

For 3-, I added extra Cast nodes for float types. If a chain of data
nodes are sunk, the new logic tries to keep a single Cast for the
entire chain rather than one Cast per node.

Progress

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

Issue

JDK-8252372: Check if cloning is required to move loads out of loops in PhaseIdealLoop::split_if_with_blocks_post()

Reviewers

Tobias Hartmann (@TobiHartmann - Reviewer) ⚠️ Review applies to 1d90bb1
Vladimir Kozlov (@vnkozlov - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.java.net/jdk pull/3689/head:pull/3689
$ git checkout pull/3689

Update a local copy of the PR:
$ git checkout pull/3689
$ git pull https://git.openjdk.java.net/jdk pull/3689/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 3689

View PR using the GUI difftool:
$ git pr show -t 3689

Using diff file

Download this PR as a diff file:
https://git.openjdk.java.net/jdk/pull/3689.diff

bridgekeeper · 2021-04-26T09:46:53Z

👋 Welcome back roland! 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 · 2021-04-26T09:48:32Z

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

hotspot-compiler

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 · 2021-04-26T09:53:49Z

Webrevs

TobiHartmann · 2021-04-27T09:13:04Z

Hi Roland,

I didn't look at this in detail yet but gave it a quick run through our testing. I'm seeing many of the following failures with the jdk/incubator/vector/ tests:

assert(cast != __null) failed: must have added a cast to pin the node

Current CompileTask:
C2:  48324 7090    b        ShortMaxVectorTests::getShortMaxVectorTests (1916 bytes)

Stack: [0x00007f13688f8000,0x00007f13689f9000],  sp=0x00007f13689f3250,  free space=1004k
Native frames: (J=compiled Java code, A=aot compiled Java code, j=interpreted, Vv=VM code, C=native code)
V  [libjvm.so+0x13293b4]  PhaseIdealLoop::try_sink_out_of_loop(Node*)+0x8d4
V  [libjvm.so+0x132969c]  PhaseIdealLoop::split_if_with_blocks_post(Node*)+0x5c
V  [libjvm.so+0x132a311]  PhaseIdealLoop::split_if_with_blocks(VectorSet&, Node_Stack&)+0x201
V  [libjvm.so+0x131d2ad]  PhaseIdealLoop::build_and_optimize(LoopOptsMode)+0x121d
V  [libjvm.so+0xa4193a]  PhaseIdealLoop::optimize(PhaseIterGVN&, LoopOptsMode)+0x2da
V  [libjvm.so+0xa3d1f6]  Compile::Optimize()+0x586
V  [libjvm.so+0xa40946]  Compile::Compile(ciEnv*, ciMethod*, int, bool, bool, bool, bool, DirectiveSet*)+0x2276
V  [libjvm.so+0x861cfa]  C2Compiler::compile_method(ciEnv*, ciMethod*, int, bool, DirectiveSet*)+0x1ea
V  [libjvm.so+0xa50a59]  CompileBroker::invoke_compiler_on_method(CompileTask*)+0xfb9
V  [libjvm.so+0xa517c8]  CompileBroker::compiler_thread_loop()+0x5a8
V  [libjvm.so+0x18bc4d1]  JavaThread::thread_main_inner()+0x271
V  [libjvm.so+0x18c3f10]  Thread::call_run()+0x100
V  [libjvm.so+0x159a26e]  thread_native_entry(Thread*)+0x10e

jdk/incubator/vector/ShortMaxVectorTests.java with -ea -esa -XX:CompileThreshold=100 -XX:+UnlockExperimentalVMOptions -Xcomp

rwestrel · 2021-04-27T13:03:28Z

I didn't look at this in detail yet but gave it a quick run through our testing. I'm seeing many of the following failures with the jdk/incubator/vector/ tests:

Thanks for running the tests. I will work on the failures.

rwestrel · 2021-04-27T15:53:50Z

That issue should be fixed. I added a CastVV node for vectors.

TobiHartmann · 2021-05-03T06:23:46Z

Thanks, testing looks good now. I need some more time to review this though.

TobiHartmann

This is hard to review but looks reasonable to me. Performance and correctness testing also looks good.

TobiHartmann · 2021-05-11T09:07:53Z

src/hotspot/share/opto/loopopts.cpp

-  if (u_loop->_child) {
-    if (useblock == u_loop->_head && u_loop->_head->is_OuterStripMinedLoop()) {
-      return u_loop->_head->in(LoopNode::EntryControl);
+Node* PhaseIdealLoop::place_near_use(Node* useblock, IdealLoopTree* loop) const {


Maybe the name and comment should be adjusted since we no longer place it next to the use but right outside of the loop.

Right. Updated.

openjdk · 2021-05-11T09:08:53Z

@rwestrel 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:

8252372: Check if cloning is required to move loads out of loops in PhaseIdealLoop::split_if_with_blocks_post()

Reviewed-by: thartmann, kvn

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 94 new commits pushed to the master branch:

fe33343: 8256304: should MonitorUsedDeflationThreshold be experimental or diagnostic
8f10c5a: 8267190: Optimize Vector API test operations
94cfeb9: 8256155: Allow multiple large page sizes to be used on Linux
ec8a809: 8267119: switch expressions lack support for deferred type-checking
4ba7613: 8267332: xor value should handle bounded values
ee2651b: 8203359: Container level resources events
b5d32bb: 8260690: JConsole User Guide Link from the Help menu is not accessible by keyboard
e48d7d6: 8264218: Public method javax.swing.JMenu.setComponentOrientation() has no spec
9eaa4af: 8267056: tools/jpackage/share/RuntimePackageTest.java fails with NoSuchFileException
e094f3f: 8266856: Make element void
... and 84 more: https://git.openjdk.java.net/jdk/compare/f4227879b0504bc4656bd84153b13b04c007aa66...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.

rwestrel · 2021-05-17T08:08:14Z

This is hard to review but looks reasonable to me. Performance and correctness testing also looks good.

Thanks for reviewing it.

vnkozlov

This looks reasonable to me.
Did we got performance results for it?

TobiHartmann · 2021-05-19T07:36:52Z

Yes, I've linked it to the bug.

rwestrel · 2021-05-26T09:18:49Z

@vnkozlov thanks for the review

rwestrel · 2021-05-26T09:19:00Z

/integrate

openjdk · 2021-05-26T09:21:01Z

@rwestrel Since your change was applied there have been 141 commits pushed to the master branch:

0394416: 8267468: Rename refill waster counters in ThreadLocalAllocBuffer
b33b8bc: 8267750: Incomplete fix for JDK-8267683
ac36b7d: 8267452: Delegate forEachRemaining in Spliterators.iterator()
d0d2ddc: 8267651: runtime/handshake/HandshakeTimeoutTest.java times out when dumping core
a98e476: 8267311: vmTestbase/gc/gctests/StringInternGC/StringInternGC.java eventually OOMEs
5aa45f2: 8267403: tools/jpackage/share/FileAssociationsTest.java#id0 failed with "Error: Bundler "Mac PKG Package" (pkg) failed to produce a package"
c20ca42: 8267691: Change table to obsolete CriticalJNINatives in JDK 18, not 17
e751b7b: 8267683: rfc7301Grease8F value not displayed correctly in SSLParameters javadoc
0b77359: 8224243: Add implSpec's to AccessibleObject and seal Executable
594d454: 8267574: Dead code in HtmlStyle/HtmlDocletWriter
... and 131 more: https://git.openjdk.java.net/jdk/compare/f4227879b0504bc4656bd84153b13b04c007aa66...master

Your commit was automatically rebased without conflicts.

Pushed as commit 9d305b9.

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

fix

40f1ffe

extra comments

3bb62a6

openjdk bot added the rfr Pull request is ready for review label Apr 26, 2021

openjdk bot added the hotspot-compiler hotspot-compiler-dev@openjdk.org label Apr 26, 2021

Merge branch 'master' into JDK-8252372

58a4ef6

CastVV

1d90bb1

TobiHartmann approved these changes May 11, 2021

View changes

openjdk bot added the ready Pull request is ready to be integrated label May 11, 2021

rwestrel added 2 commits May 17, 2021

Merge branch 'master' into JDK-8252372

aaf7f98

Tobias' review

19a01e9

vnkozlov reviewed May 19, 2021

View changes

vnkozlov approved these changes May 21, 2021

View changes

openjdk bot closed this May 26, 2021

openjdk bot added integrated Pull request has been integrated and removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels May 26, 2021

8252372: Check if cloning is required to move loads out of loops in PhaseIdealLoop::split_if_with_blocks_post() #3689

8252372: Check if cloning is required to move loads out of loops in PhaseIdealLoop::split_if_with_blocks_post() #3689

rwestrel commented Apr 26, 2021 •

edited by openjdk bot

bridgekeeper bot commented Apr 26, 2021

openjdk bot commented Apr 26, 2021

mlbridge bot commented Apr 26, 2021 •

edited

TobiHartmann commented Apr 27, 2021

rwestrel commented Apr 27, 2021

rwestrel commented Apr 27, 2021

TobiHartmann commented May 3, 2021

TobiHartmann left a comment

TobiHartmann May 11, 2021

rwestrel May 17, 2021

openjdk bot commented May 11, 2021 •

edited

rwestrel commented May 17, 2021

vnkozlov left a comment

TobiHartmann commented May 19, 2021

rwestrel commented May 26, 2021

rwestrel commented May 26, 2021

openjdk bot commented May 26, 2021

8252372: Check if cloning is required to move loads out of loops in PhaseIdealLoop::split_if_with_blocks_post() #3689

8252372: Check if cloning is required to move loads out of loops in PhaseIdealLoop::split_if_with_blocks_post() #3689

Conversation

rwestrel commented Apr 26, 2021 • edited by openjdk bot

Progress

Issue

Reviewers

Reviewing

bridgekeeper bot commented Apr 26, 2021

openjdk bot commented Apr 26, 2021

mlbridge bot commented Apr 26, 2021 • edited

Webrevs

TobiHartmann commented Apr 27, 2021

rwestrel commented Apr 27, 2021

rwestrel commented Apr 27, 2021

TobiHartmann commented May 3, 2021

TobiHartmann left a comment

Choose a reason for hiding this comment

TobiHartmann May 11, 2021

Choose a reason for hiding this comment

rwestrel May 17, 2021

Choose a reason for hiding this comment

openjdk bot commented May 11, 2021 • edited

rwestrel commented May 17, 2021

vnkozlov left a comment

Choose a reason for hiding this comment

TobiHartmann commented May 19, 2021

rwestrel commented May 26, 2021

rwestrel commented May 26, 2021

openjdk bot commented May 26, 2021

rwestrel commented Apr 26, 2021 •

edited by openjdk bot

mlbridge bot commented Apr 26, 2021 •

edited

openjdk bot commented May 11, 2021 •

edited