RFR: 8266764: [REDO] JDK-8255493 Support for pre-generated java.lang.invoke classes in CDS dynamic archive [v5]
Yumin Qi
minqi at openjdk.java.net
Fri May 14 15:41:20 UTC 2021
> Hi, please review
> This is REDO for jdk-8255493, the integration of the fix caused ParallelLambdaLoadTest.java failed.
> The old PR description:
> ------------- old ------------
> When do dynamic dump, the pre-generated lambda form classes from java.lang.invoke are not stored in dynamic archive. The patch will regenerate the four holder classes,
> "java.lang.invoke.Invokers$Holder",
> "java.lang.invoke.DirectMethodHandle$Holder",
> "java.lang.invoke.DelegatingMethodHandle$Holder",
> "java.lang.invoke.LambdaForm$Holder"
> which will include the versions in static archive and new loaded functions all together and stored in dynamic archive. New test case added.
> ------------ new -------------
> The new fix (incremental)
> 1) Added a lock to protect LambdaFormInvokers::_lambdaform_lines, at dynamic dump case, there are potentially concurrent issue the list is added from multiple threads at same time regeneration of holder class is reading from the list.
> 2) Added a new function to MetaspaceShared, MetaspaceShared::regenerate_lambdaforminvokers_holders which calls into LambdaFormInvokers::LambdaFormInvokers::regenerate_holder_classes, and handle exceptions, this way even the regeneration fails, no affect to normal dump process.
> 3) Move the call to regenerate holder classes to MetaspaceShared::link_and_cleanup_shared_classes, this is before java shutdown hook execution, since after that, it is not safe to call into java.
> 4) Since change in 3), jcmd to dynamic dump, we need call MetaspaceShared::regenerate_lambdaforminvokers_holders before do dumping.
>
> Tests: tier1,tier2,tier3,tier4
> local test on runtime/cds
> local test ParallelLambdaLoadTest.java in loops without crash.
>
> Thanks
> Yumin
Yumin Qi has updated the pull request incrementally with one additional commit since the last revision:
New added DynamicArchive::prepare_for_dynamic_dumping_at_exit should be CDS only
-------------
Changes:
- all: https://git.openjdk.java.net/jdk/pull/3953/files
- new: https://git.openjdk.java.net/jdk/pull/3953/files/c2236e8d..bcbbc749
Webrevs:
- full: https://webrevs.openjdk.java.net/?repo=jdk&pr=3953&range=04
- incr: https://webrevs.openjdk.java.net/?repo=jdk&pr=3953&range=03-04
Stats: 4 lines in 2 files changed: 4 ins; 0 del; 0 mod
Patch: https://git.openjdk.java.net/jdk/pull/3953.diff
Fetch: git fetch https://git.openjdk.java.net/jdk pull/3953/head:pull/3953
PR: https://git.openjdk.java.net/jdk/pull/3953
More information about the hotspot-runtime-dev
mailing list