On Wed, 20 Jan 2021 14:34:15 GMT, Severin Gehwolf <[email protected]> wrote:
>> This patch adds some explicit capacity for local refs. New regression test
>> fails prior and passes after the patch.
>>
>> Thoughts?
>
> Severin Gehwolf has updated the pull request with a new target base due to a
> merge or a rebase. The incremental webrev excludes the unrelated changes
> brought in by the merge/rebase. The pull request contains six additional
> commits since the last revision:
>
> - Merge branch 'master' into JDK-8258836-check-jni-mbeanserver
> - Actually assign the variable returned from PopLocalFrame
> - Merge test files into one
> - Adress review feedback from dholmes
> - Merge branch 'master' into JDK-8258836-check-jni-mbeanserver
> - 8258836: JNI local refs exceed capacity getDiagnosticCommandInfo
There are serveral places where EXCEPTION_CHECK_AND_FREE is called while there
is a pending PushLocalFrame (or 2). Corresponding PopLocalFrames are needed if
there was an exception. The error handling has already proven somewhat tricky
and error prone. This just makes it worse. I'm not sure if there is a good
solution to this, or just continue adding PopLocalFrames in more places.
src/jdk.management/share/native/libmanagement_ext/DiagnosticCommandImpl.c line
194:
> 192:
> dcmd_info_array[i].num_arguments);
> 193: if (args == NULL) {
> 194: (*env)->PopLocalFrame(env, NULL);
You actually need 2 PopLocalFrame calls here. One to match the PushLocalFrame
inside this loop, and one to match the PushLocalFrame outside the loop.
src/jdk.management/share/native/libmanagement_ext/DiagnosticCommandImpl.c line
218:
> 216: args);
> 217: if (obj == NULL) {
> 218: (*env)->PopLocalFrame(env, NULL);
And 2 here also.
-------------
PR: https://git.openjdk.java.net/jdk/pull/2130
