[BUG] when subscribe the attributes, Chip-tool will crash

[nicelyjust]

Reproduction steps

SOE:MEIZU Android 7
when subscribe the attributes，
val endpointId = ChipPathId.forId(endpoint) val clusterId = ChipPathId.forWildcard() val attributeId = ChipPathId.forWildcard() val minInterval = 1 // seconds val maxInterval = 4 // seconds val attributePath = ChipAttributePath.newInstance(endpointId, clusterId, attributeId) val eventPath = ChipEventPath.newInstance(endpointId, clusterId, attributeId) TLog.d(TAG,"attributePath: [${attributePath}]") chipClient.chipDeviceController.subscribeToPath(...)
then app will crash!
A art/runtime/indirect_reference_table.cc:132] JNI ERROR (app bug): local reference table overflow (max=512) A art/runtime/indirect_reference_table.cc:132] local reference table dump: A art/runtime/indirect_reference_table.cc:132] Last 10 entries (of 512): A art/runtime/indirect_reference_table.cc:132] 511: 0x1386cad8 java.lang.String "{"value":[]}" A art/runtime/indirect_reference_table.cc:132] 510: 0x132a8570 byte[] (2 elements) A art/runtime/indirect_reference_table.cc:132] 509: 0x1375cc58 java.util.ArrayList A art/runtime/indirect_reference_table.cc:132] 508: 0x1375ccb8 chip.devicecontroller.model.ChipAttributePath A art/runtime/indirect_reference_table.cc:132] 507: 0x13904f00 java.lang.Class<chip.devicecontroller.model.NodeState> A art/runtime/indirect_reference_table.cc:132] 506: 0x1375cd90 chip.devicecontroller.model.AttributeState A art/runtime/indirect_reference_table.cc:132] 505: 0x13868de0 java.lang.Class<chip.devicecontroller.model.AttributeState> A art/runtime/indirect_reference_table.cc:132] 504: 0x13cc2ab8 java.lang.String "chip/devicecontr... (42 chars) A art/runtime/indirect_reference_table.cc:132] 503: 0x136506d0 java.lang.String "chip/devicecontr... (45 chars) A art/runtime/indirect_reference_table.cc:132] 502: 0x13904600 java.lang.Class<chip.devicecontroller.model.ChipAttributePath> A art/runtime/indirect_reference_table.cc:132] Summary: A art/runtime/indirect_reference_table.cc:132] 223 of java.lang.Class (11 unique instances) A art/runtime/indirect_reference_table.cc:132] 90 of java.lang.String (89 unique instances) A art/runtime/indirect_reference_table.cc:132] 77 of java.lang.Long (77 unique instances) A art/runtime/indirect_reference_table.cc:132] 40 of chip.devicecontroller.model.ChipAttributePath (40 unique instances) A art/runtime/indirect_reference_table.cc:132] 39 of chip.devicecontroller.model.AttributeState (39 unique instances) A art/runtime/indirect_reference_table.cc:132] 16 of java.util.ArrayList (16 unique instances) A art/runtime/indirect_reference_table.cc:132] 15 of java.lang.Integer (15 unique instances) A art/runtime/indirect_reference_table.cc:132] 6 of java.lang.Boolean (6 unique instances) A art/runtime/indirect_reference_table.cc:132] 2 of chip.devicecontroller.model.NodeState (2 unique instances) A art/runtime/indirect_reference_table.cc:132] 2 of java.util.HashMap (2 unique instances) A art/runtime/indirect_reference_table.cc:132] 1 of chip.devicecontroller.ChipStructs$DescriptorClusterDeviceTypeStruct A art/runtime/indirect_reference_table.cc:132] 1 of byte[] (2 elements) A art/runtime/indirect_reference_table.cc:132]

Bug prevalence

always

GitHub hash of the SDK that was being used

8db0d65

Platform

android

Platform Version(s)

Android 7 cpu:qualcomm 625

Anything else?

native crash.txt

[nicelyjust]

https://stackoverflow.com/questions/59680110/crash-error-jni-error-local-reference-table-overflow-max-512 , maybe that is why Google Matter supports a minimum of version 8.0

[yunhanw-google]

@nicelyjust could you trying the latest master code, and see whether this issue can be reproduced?

[yunhanw-google]

@nicelyjust are you in slack? could you find me in yunhanw-google in slack?
Could you also try this fix #29175, the local reference issue spreads in multiple place, this PR should help ChipAttributePath entry in local reference tag per your log.

[nicelyjust]

@yunhanw-google Thank for your response.I'll try it later, it needs to be compiled

[yunhanw-google]

@nicelyjust please hold on a little bit for testing. just root caused the leak points, and would create the pull request in the morning, we would also have another 2 or 3 pull request to fix other located leaking local reference points.

[yunhanw-google]

@nicely just are you in slack?

[nicelyjust]

@yunhanw-google yes,I'm. Already sent you a message on Slack