Use sun.misc.IOUtils new API readAllBytes()

[JasonFengJ9]

Use sun.misc.IOUtils new API readAllBytes()

Replace IOUtils.readFully(is, -1, true) with IOUtils.readAllBytes(is) to avoid IOException: length cannot be negative: -1.

[ci skip] because the new API readAllBytes(is) is not available at extension repo openj9 branch.

Notes:

1. Though readNBytes(is, Integer.MAX_VALUE) has same result, readAllBytes(is) seems better choice;
2. API doc within sun.misc.IOUtils specifies @since 1.9 or @since 11 though this is a dedicated Java 8 extension rep https://github.com/ibmruntimes/openj9-openjdk-jdk8/blob/e6095d9adfe3c072d483a7df3fcdf1eeec76044b/jdk/src/share/classes/sun/misc/IOUtils.java#L284-L302
   @andrew-m-leonard any insights?

Reviewer: @DanHeidinga

Fixes #8308
Fixes #8309

Signed-off-by: Jason Feng fengj@ca.ibm.com

[DanHeidinga]

lgtm

[DanHeidinga]

Jenkins test sanity zlinux jdk8

[DanHeidinga]

@JasonFengJ9 Can you port the change to the 0.18 branch?

[DanHeidinga]

FYI - I'm not sure the PR build will pass as it depends on the openjdk import

[DanHeidinga]

2. API doc within sun.misc.IOUtils specifies @since 1.9 or @since 11 though this is a dedicated Java 8 extension rep

This is due to the changes being a backport. I wouldn't worry about incorrect since tags in the upstream code

[pshipton]

Started testing with the openj9-staging branch here https://ci.eclipse.org/openj9/view/Pipelines/job/Pipeline-Release-Build/144/

[JasonFengJ9]

@DanHeidinga created #8315 for 0.18 branch.

[JasonFengJ9]

This PR made change to https://github.com/eclipse/openj9/blob/master/jcl/src/java.base/share/classes/java/lang/invoke/SecurityFrameInjector.java
@pshipton the component label is comp:vm instead of comp:test, is that correct?

[DanHeidinga]

@JasonFengJ9 Does this test need to be changed as well? https://github.com/eclipse/openj9/blob/2f295ea91cdd97ff37eb5d164fc63014ea718bc7/test/functional/Java8andUp/src/org/openj9/test/regression/Cmvc194280.java#L47

[pshipton]

Note the extended test failures in https://ci.eclipse.org/openj9/job/Test_openjdk8_j9_extended.functional_s390x_linux_OpenJDK8/1/

[JasonFengJ9]

raw = IOUtils.readFully(stream, 1024, false);

This works as before because of the back compatibility implemented by
https://github.com/ibmruntimes/openj9-openjdk-jdk8/blob/e6095d9adfe3c072d483a7df3fcdf1eeec76044b/jdk/src/share/classes/sun/misc/IOUtils.java#L284-L312

[JasonFengJ9]

[ERR] Caused by: java.io.IOException: length cannot be negative: -1
 [ERR] 	at sun.misc.IOUtils.readFully(IOUtils.java:305)
 [ERR] 	at java.lang.invoke.SecurityFrameInjector$1.run(SecurityFrameInjector.java:100)
 [ERR] 	... 10 more

This seems same error.

[JasonFengJ9]

https://ci.eclipse.org/openj9/job/Test_openjdk8_j9_extended.functional_s390x_linux_OpenJDK8/1/

This appears the initial build triggered by JDK8 Acceptance pipeline.
There is no failure in https://ci.eclipse.org/openj9/job/Pipeline_Build_Test_JDK8_s390x_linux/1274/ however it has only sanity.functional.

@pshipton could you launch another extended.functional test build based on this PR?

[pshipton]

Started extended.functional https://ci.eclipse.org/openj9/view/Pipelines/job/Pipeline-Release-Build/147/

[pshipton]

Although I merged this, I expect it has broken jdk14 and jdk(next). Hopefully the readAllBytes() API will be delivered there shortly, otherwise we'll have to use the processor to handle these streams differently.

Nm, I forgot this was specific to jdk8.