RFR: 8224852: JVM crash on watched field access from native code [v8]

[Alex Menkov]

On Sat, 25 Oct 2025 02:00:48 GMT, Leonid Mesnik <lmesnik at openjdk.org> wrote:

>> The field access/modification events set interp only mode and compiled frame is not expected. However JNI might call `post_field_access_by_jni` while the last java frame is compiled. 
>> 
>> 1) The thread switched to interponly mode while it is in JNI code. The deoptimization is triggered but each frame is really changed only execution returns to it.  So last java frame was not executed and thus is still compiled. 
>> 2) The JNI accessed field from the thread where field events are not enabled. So the `post_field_access_by_jni` is called in threads in interp_only mode. 
>> 
>> The original example doesn't reproduce issue because of JDK changes and I don't know of it is 1) or 2)I. I implemented regression test for both problems. 
>> 
>> The location should be zero for JNI access.
>
> Leonid Mesnik has updated the pull request incrementally with one additional commit since the last revision:
> 
>   counter are updated

The fix looks good.
There are some minor notes about the test

test/hotspot/jtreg/serviceability/jvmti/events/FieldAccess/FieldEventsFromJNI/libFieldEventsFromJNI.cpp line 33:

> 31: jvmtiEnv* jvmti_env;
> 32: 
> 33: // The event counters are used to check events from differen threads.

Suggestion:

// The event counters are used to check events from different threads.

test/hotspot/jtreg/serviceability/jvmti/events/FieldAccess/FieldEventsFromJNI/libFieldEventsFromJNI.cpp line 44:

> 42: static const char* ACCESS_METHOD_NAME = "enableEventsAndAccessField";
> 43: static const char* MODIFY_FIELD_NAME  = "modifyField";
> 44: static const char* MODIFY_FIELD_VALUE = "modifyFieldValue";

Not used (TBH I don't see any value in checking ACCESS_FIELD_VALUE)

test/hotspot/jtreg/serviceability/jvmti/events/FieldAccess/FieldEventsFromJNI/libFieldEventsFromJNI.cpp line 116:

> 114: Agent_OnLoad(JavaVM *vm, char *options, void *reserved) {
> 115:   jvmtiEnv *jvmti = nullptr;
> 116:   jint res = vm->GetEnv((void **) &jvmti, JVMTI_VERSION_21);

Suggestion:

  jint res = vm->GetEnv((void **)&jvmti, JVMTI_VERSION_21);

test/hotspot/jtreg/serviceability/jvmti/events/FieldAccess/FieldEventsFromJNI/libFieldEventsFromJNI.cpp line 122:

> 120:   jvmtiError err = JVMTI_ERROR_NONE;
> 121:   jvmtiCapabilities capabilities;
> 122:   (void) memset(&capabilities, 0, sizeof (capabilities));

Suggestion:

  (void)memset(&capabilities, 0, sizeof(capabilities));

test/hotspot/jtreg/serviceability/jvmti/events/FieldAccess/FieldEventsFromJNI/libFieldEventsFromJNI.cpp line 128:

> 126:   check_jvmti_error(err, "AddCapabilities");
> 127:   jvmtiEventCallbacks callbacks;
> 128:   (void) memset(&callbacks, 0, sizeof (callbacks));

Suggestion:

  (void)memset(&callbacks, 0, sizeof(callbacks));

test/hotspot/jtreg/serviceability/jvmti/events/FieldAccess/FieldEventsFromJNI/libFieldEventsFromJNI.cpp line 131:

> 129:   callbacks.FieldAccess = &cbFieldAccess;
> 130:   callbacks.FieldModification = &cbFieldModification;
> 131:   err = jvmti->SetEventCallbacks(&callbacks, (int) sizeof (jvmtiEventCallbacks));

Suggestion:

  err = jvmti->SetEventCallbacks(&callbacks, (jint)sizeof(jvmtiEventCallbacks));

test/hotspot/jtreg/serviceability/jvmti/events/FieldAccess/FieldEventsFromJNI/libFieldEventsFromJNI.cpp line 160:

> 158:   check_jvmti_error(err, "SetFieldAccessWatch");
> 159: 
> 160:   jstring jname = (jstring)jni->GetObjectField(self, fieldToRead);

The name is misleading. it's the field value, not name

test/hotspot/jtreg/serviceability/jvmti/events/FieldAccess/FieldEventsFromJNI/libFieldEventsFromJNI.cpp line 167:

> 165:   check_jvmti_error(err, "SetEventNotificationMode");
> 166: 
> 167:   const char* name_str = jni->GetStringUTFChars(jname, nullptr);

`value_str`?

test/hotspot/jtreg/serviceability/jvmti/events/FieldAccess/FieldEventsFromJNI/libFieldEventsFromJNI.cpp line 174:

> 172: 
> 173:   if (access_cnt != numOfEventsExpected) {
> 174:     fatal(jni, "The field access count is incorrect.");

For error analysis it would be useful to log actual and expected values

test/hotspot/jtreg/serviceability/jvmti/events/FieldAccess/FieldEventsFromJNI/libFieldEventsFromJNI.cpp line 207:

> 205: 
> 206:   if (modify_cnt != numOfEventsExpected) {
> 207:     fatal(jni, "The field access count should be 1.");

numOfEventsExpected can be 0
Would be useful to log actual and expected values

-------------

PR Review: https://git.openjdk.org/jdk/pull/27584#pullrequestreview-3396914324
PR Review Comment: https://git.openjdk.org/jdk/pull/27584#discussion_r2476050208
PR Review Comment: https://git.openjdk.org/jdk/pull/27584#discussion_r2476017561
PR Review Comment: https://git.openjdk.org/jdk/pull/27584#discussion_r2476030439
PR Review Comment: https://git.openjdk.org/jdk/pull/27584#discussion_r2476030963
PR Review Comment: https://git.openjdk.org/jdk/pull/27584#discussion_r2476031549
PR Review Comment: https://git.openjdk.org/jdk/pull/27584#discussion_r2476033084
PR Review Comment: https://git.openjdk.org/jdk/pull/27584#discussion_r2476043111
PR Review Comment: https://git.openjdk.org/jdk/pull/27584#discussion_r2476044112
PR Review Comment: https://git.openjdk.org/jdk/pull/27584#discussion_r2476045472
PR Review Comment: https://git.openjdk.org/jdk/pull/27584#discussion_r2476049693