Copilot commented on code in PR #18793:
URL: https://github.com/apache/iotdb/pull/18793#discussion_r4197134675


##########
iotdb-client/service-rpc/src/main/java/org/apache/iotdb/rpc/TCompressedElasticFramedTransport.java:
##########
@@ -59,10 +59,14 @@ protected void closeAllocatedBuffers() {
 
   @Override
   protected void readFrame() throws TTransportException {
+    // Discard the previous frame's budget before reading the next frame 
header and payload.
+    resetMessageSizeAndConsumedBytes();
     underlying.readAll(i32buf, 0, 4);
     int size = TFramedTransport.decodeFrameSize(i32buf);
     validateFrame(size);
     readBuffer.fill(underlying, size);
+    // Bind subsequent protocol reads to the current compressed frame size.
+    resetMessageSizeAndConsumedBytes(size);

Review Comment:
   The compressed path has the same limit mismatch: this reset validates the 
compressed frame size against the underlying endpoint's default 
`maxMessageSize`, even when this transport was constructed with a larger 
`thriftMaxFrameSize`. SSL/blocking endpoints using Thrift's 100 MiB default 
will reject otherwise-valid larger compressed frames here. Align the underlying 
endpoint configuration with IoTDB's configured frame limit before wrapping it.



##########
iotdb-client/service-rpc/src/test/java/org/apache/iotdb/rpc/TElasticFramedTransportTest.java:
##########
@@ -19,18 +19,85 @@
 
 package org.apache.iotdb.rpc;
 
+import org.apache.thrift.TConfiguration;
 import org.apache.thrift.transport.TByteBuffer;
+import org.apache.thrift.transport.TMemoryBuffer;
 import org.apache.thrift.transport.TTransportException;
 import org.junit.Test;
 
 import java.nio.ByteBuffer;
 import java.nio.charset.StandardCharsets;
+import java.util.Arrays;
 
+import static org.junit.Assert.assertArrayEquals;
 import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.fail;
 
 public class TElasticFramedTransportTest {
 
+  @Test
+  public void testReadFramesResetMessageSize() throws TTransportException {
+    byte[] firstFrame = {1, 2};
+    byte[] secondFrame = {3, 4, 5, 6, 7, 8};
+    ByteBuffer framedData =
+        ByteBuffer.allocate(8 + firstFrame.length + secondFrame.length)
+            .putInt(firstFrame.length)
+            .put(firstFrame)
+            .putInt(secondFrame.length)
+            .put(secondFrame)
+            .flip();
+    TConfiguration configuration =
+        
TConfiguration.custom().setMaxMessageSize(10).setMaxFrameSize(10).build();
+    TMemoryBuffer underlying = new TMemoryBuffer(configuration, 10);

Review Comment:
   This regression test does not prove that frame budgets are reset: in Thrift 
0.25, `TMemoryBuffer.read` checks each requested length but does not decrement 
the remaining message budget. If both new reset calls are removed, every 
individual read here is still at most 10 bytes and the test passes. Use a 
budget-consuming underlying transport so the test fails against the pre-fix 
implementation.
   
   This issue also appears on line 87 of the same file.



##########
iotdb-client/service-rpc/src/main/java/org/apache/iotdb/rpc/TElasticFramedTransport.java:
##########
@@ -194,10 +194,14 @@ public int read(byte[] buf, int off, int len) throws 
TTransportException {
   }
 
   protected void readFrame() throws TTransportException {
+    // Discard the previous frame's budget before reading the next frame 
header and payload.
+    resetMessageSizeAndConsumedBytes();
     underlying.readAll(i32buf, 0, 4);
     int size = TFramedTransport.decodeFrameSize(i32buf);
     validateFrame(size);
     readBuffer.fill(underlying, size);
+    // Bind subsequent protocol reads to the current frame size.
+    resetMessageSizeAndConsumedBytes(size);

Review Comment:
   `resetMessageSizeAndConsumedBytes(size)` rejects sizes above the underlying 
transport's `maxMessageSize`, but this wrapper's `thriftMaxFrameSize` is 
configured independently. Blocking server sockets and SSL client sockets are 
created with Thrift's 100 MiB default, while IoTDB clients may permit frames up 
to 512 MiB, so a valid frame between those limits is read and then fails here 
with `MESSAGE_SIZE_LIMIT`. Configure the endpoint message budget consistently 
(including the 4-byte header) before wrapping it, or otherwise reconcile the 
two limits before resetting.



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to