8295102: Always load @lambda-form-invoker lines from default classlist #10639

iklam · 2022-10-10T22:18:33Z

The @lambda-form-invoker lines in the default classlist ($JAVA_HOME/lib/classlist) control what lambda-form invoker methods are added to these classes that are generated during a CDS dump (see here):

java.lang.invoke.Invokers$Holder
java.lang.invoke.DirectMethodHandle$Holder
java.lang.invoke.DelegatingMethodHandle$Holder
java.lang.invoke.LambdaForm$Holder

The list of these invoker methods is generated as part of the JDK build process. They are used for invoking MethodHandles of commonly-used types, so that we can avoid generating custom LambdaForm classes.

When a CDS archive is created with a non-default classlist, (e.g., -XX:SharedClassListFile=foo.classlist), such a classlist may not contain all the @lambda-form-invoker lines as in the default classlist. As a result, at runtime, more custom LambdaForm classes may be generated than necessary.

The solution is to always load the @lambda-form-invoker lines from the default classlist, to make sure the CDS image has all the commonly-used invoker methods.

Progress

Change must be properly reviewed (1 review required, with at least 1 Reviewer)
Change must not contain extraneous whitespace
Commit message must refer to an issue

Issue

JDK-8295102: Always load @lambda-form-invoker lines from default classlist

Reviewers

Claes Redestad (@cl4es - Reviewer) ⚠️ Review applies to 2e6af772
Calvin Cheung (@calvinccheung - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk pull/10639/head:pull/10639
$ git checkout pull/10639

Update a local copy of the PR:
$ git checkout pull/10639
$ git pull https://git.openjdk.org/jdk pull/10639/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 10639

View PR using the GUI difftool:
$ git pr show -t 10639

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/10639.diff

bridgekeeper · 2022-10-10T22:23:01Z

👋 Welcome back iklam! 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 · 2022-10-10T22:25:31Z

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

hotspot-runtime

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 · 2022-10-11T02:20:16Z

Webrevs

dholmes-ora · 2022-10-11T03:23:01Z

Seems to me this is edging towards a scheme where the user-defined classlist is to be used to augment the default one, rather than being a complete replacement. Either the classlist should be the one-true-source for the classes to include in the archive, or else you need a way to express that you want to use it in union with the default list (to some extent), but not just (seemingly arbitrarily) decide that a special subset of classes should always be included even if missing from the classlist.

iklam · 2022-10-11T06:10:23Z

Seems to me this is edging towards a scheme where the user-defined classlist is to be used to augment the default one, rather than being a complete replacement. Either the classlist should be the one-true-source for the classes to include in the archive, or else you need a way to express that you want to use it in union with the default list (to some extent), but not just (seemingly arbitrarily) decide that a special subset of classes should always be included even if missing from the classlist.

This PR doesn't add extra classes into the archive. It just adds methods to the four Holder classes listed in the PR's description. In fact, these 4 classes (as stored in the JDK's module image) are generated at JDK build time to contain commonly used invoker methods. See GenerateLinkOptData.gmk.

The intention of the @lambda-form-invoker lines in the classlist is to add extra invoker methods into these holder classes (according to the app's profile) to avoid runtime generation of LambdaForm classes. Therefore, when dumping the archive, these 4 classes are regenerated and populated with methods specified by the @lambda-form-invoker lines.

However, the application's profile may use only a small number of invoker methods. As a result, if we just depend on the app's profile, the regenerated Holder classes, which are stored in the CDS archive, will be missing some invoker methods that are in JDK's module image.

These missing methods cause the following problems

Excessive LambdaForm classes may be generated at runtime if the app executes code outside of its profile
It will cause complication in future CDS optimizations (e.g., JDK-8293336)

The default classlist (also generated by GenerateLinkOptData.gmk) contains the @lambda-form-invoker lines to regenerate the exact same set of methods as in the JDK's module image (this is checked by the new test case). By always loading these lines, we can guarantee that when a Holder class is stored in the CDS archive, it will not be missing any of the methods that are in the modules image.

iklam · 2022-10-11T06:13:04Z

More background info:

dholmes-ora · 2022-10-11T06:26:32Z

Sorry this was an aspect of archive generation that I was not aware of - and I still don't really grok the full picture here.

calvinccheung · 2022-10-11T23:30:51Z

test/hotspot/jtreg/runtime/cds/appcds/TestCommon.java

+        if (a.length > b.length) {
+            throw new RuntimeException("Output mismatch on line :" + max
+                                       + ", a=" + a[max]
+                                       + ", b=<none>");
+
+        }
+        if (a.length < b.length) {
+            throw new RuntimeException("Output mismatch on line :" + max
+                                       + ", a=<none>"
+                                       + ", b=" + b[max]);
+        }


Should these checks be done before the "for" loop at line 756? Otherwise, could one of the array be indexed out-out-bound?

Thanks for finding the bug. I have fixed it in version 2e6af77. I also tested locally with this code:

String[] a = {"1", "2"}, b = {"1"}; TestCommon.linesMustMatch(a, b);

which correctly produces this error:

java.lang.RuntimeException: Output mismatch on line 1: a=2, b=<none>

cl4es

This patch ensures that the methods in the LF holder classes at least contain the set of methods as generated by jlink at build, which seems to me to be reasonable to hold as an invariant to avoid surprises at runtime. (Does resolving a LF from a holder does trigger emitting a @lambda-form-invoker line when generating the classlist?)

openjdk · 2022-10-12T09:40:24Z

@iklam 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:

8295102: Always load @lambda-form-invoker lines from default classlist

Reviewed-by: redestad, ccheung

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 46 new commits pushed to the master branch:

1961e81: 8289509: Improve test coverage for XPath Axes: descendant, descendant-or-self, following, following-sibling
5699041: 8294772: Remove workaround in os::dll_address_to_library_name
d125265: 8277970: Test jdk/sun/security/ssl/SSLSessionImpl/NoInvalidateSocketException.java fails with "tag mismatch"
0475c34: 8292386: jvmti/thread/SuspendThread/suspendthrd03 failed with "FAIL: Status is 2"
760a260: 8295213: Run GHA manually with user-specified make and configure arguments
8402260: 8295198: Update more openjdk.java.net => openjdk.org URLs
c357b59: 8295211: Fix autoconf 2.71 warning "AC_CHECK_HEADERS: you should use literals"
8607842: 8295205: Add jcheck whitespace checking for markdown files
cb62f1c: 8295218: New KeepAliveTest.java has invalid copyright notice
9cf6651: 8294238: ZGC: Move CLD claimed mark clearing
... and 36 more: https://git.openjdk.org/jdk/compare/eb90c4fc0479379c8c1452afca8f37746c762e18...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.

calvinccheung

Updates look good. Thanks.

iklam · 2022-10-12T17:09:25Z

This patch ensures that the methods in the LF holder classes at least contain the set of methods as generated by jlink at build, which seems to me to be reasonable to hold as an invariant to avoid surprises at runtime. (Does resolving a LF from a holder does trigger emitting a @lambda-form-invoker line when generating the classlist?)

The @lambda-form-invoker lines are generated in MethodHandleStatics.java::traceLambdaForm() when LambdaForms are being resolved:

jdk/src/java.base/share/classes/java/lang/invoke/MethodHandleStatics.java

Lines 122 to 131 in 5699041

    
           static void traceLambdaForm(String name, MethodType type, Class<?> holder, MemberName resolvedMember) { 
        
               if (TRACE_RESOLVE) { 
        
                   System.out.println("[LF_RESOLVE] " + holder.getName() + " " + name + " " + 
        
                           shortenSignature(basicTypeSignature(type)) + 
        
                           (resolvedMember != null ? " (success)" : " (fail)")); 
        
               } 
        
               if (CDS.isDumpingClassList()) { 
        
                   CDS.traceLambdaFormInvoker("[LF_RESOLVE]", holder.getName(), name, shortenSignature(basicTypeSignature(type))); 
        
               } 
        
           }

This is the same place where GenerateLinkOptData.gmk gathers information for generating the Holder classes for the JDK's module image.

…not be affected by -Xlog options

calvinccheung

Test changes look good.

iklam · 2022-10-12T23:59:27Z

Thanks to @calvinccheung and @cl4es for the review.
/integrate

openjdk · 2022-10-13T00:01:14Z

Going to push as commit 90fb9a0.
Since your change was applied there have been 47 commits pushed to the master branch:

ac19414: 8294994: Update Jarsigner and Keytool i18n tests to validate i18n compliance
1961e81: 8289509: Improve test coverage for XPath Axes: descendant, descendant-or-self, following, following-sibling
5699041: 8294772: Remove workaround in os::dll_address_to_library_name
d125265: 8277970: Test jdk/sun/security/ssl/SSLSessionImpl/NoInvalidateSocketException.java fails with "tag mismatch"
0475c34: 8292386: jvmti/thread/SuspendThread/suspendthrd03 failed with "FAIL: Status is 2"
760a260: 8295213: Run GHA manually with user-specified make and configure arguments
8402260: 8295198: Update more openjdk.java.net => openjdk.org URLs
c357b59: 8295211: Fix autoconf 2.71 warning "AC_CHECK_HEADERS: you should use literals"
8607842: 8295205: Add jcheck whitespace checking for markdown files
cb62f1c: 8295218: New KeepAliveTest.java has invalid copyright notice
... and 37 more: https://git.openjdk.org/jdk/compare/eb90c4fc0479379c8c1452afca8f37746c762e18...master

Your commit was automatically rebased without conflicts.

openjdk · 2022-10-13T00:01:32Z

@iklam Pushed as commit 90fb9a0.

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

8295102: Always load @lambda-form-invoker lines from default classlist

c3cd3fa

openjdk bot added the hotspot-runtime hotspot-runtime-dev@openjdk.org label Oct 10, 2022

fixed build

2fd5057

iklam marked this pull request as ready for review October 11, 2022 02:15

openjdk bot added the rfr Pull request is ready for review label Oct 11, 2022

Fixed open/test/hotspot/jtreg:hotspot_appcds_dynamic tests

54d9e24

calvinccheung reviewed Oct 11, 2022

View reviewed changes

@calvinccheung comment: fixed bug in TestCommon.linesMustMatch

2e6af77

cl4es approved these changes Oct 12, 2022

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Oct 12, 2022

calvinccheung approved these changes Oct 12, 2022

View reviewed changes

rewrote test to write output to a file instead of stdout, so it will …

fada121

…not be affected by -Xlog options

calvinccheung approved these changes Oct 12, 2022

View reviewed changes

openjdk bot added the integrated Pull request has been integrated label Oct 13, 2022

openjdk bot closed this Oct 13, 2022

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Oct 13, 2022

8295102: Always load @lambda-form-invoker lines from default classlist #10639

8295102: Always load @lambda-form-invoker lines from default classlist #10639

Uh oh!

Conversation

iklam commented Oct 10, 2022 • edited by openjdk bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Progress

Issue

Reviewers

Reviewing

Uh oh!

bridgekeeper bot commented Oct 10, 2022

Uh oh!

openjdk bot commented Oct 10, 2022

Uh oh!

mlbridge bot commented Oct 11, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

dholmes-ora commented Oct 11, 2022

Uh oh!

iklam commented Oct 11, 2022

Uh oh!

iklam commented Oct 11, 2022

Uh oh!

dholmes-ora commented Oct 11, 2022

Uh oh!

calvinccheung Oct 11, 2022

Choose a reason for hiding this comment

Uh oh!

iklam Oct 12, 2022

Choose a reason for hiding this comment

Uh oh!

cl4es left a comment

Choose a reason for hiding this comment

Uh oh!

openjdk bot commented Oct 12, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

calvinccheung left a comment

Choose a reason for hiding this comment

Uh oh!

iklam commented Oct 12, 2022

Uh oh!

calvinccheung left a comment

Choose a reason for hiding this comment

Uh oh!

iklam commented Oct 12, 2022

Uh oh!

openjdk bot commented Oct 13, 2022

Uh oh!

openjdk bot commented Oct 13, 2022

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

4 participants

iklam commented Oct 10, 2022 •

edited by openjdk bot

Loading

mlbridge bot commented Oct 11, 2022 •

edited

Loading

openjdk bot commented Oct 12, 2022 •

edited

Loading