Default MRP_RETRY_INTERVAL_SENDER_BOOST to 0 ms (#27555)

src/messaging/ReliableMessageMgr.cpp

@@ -261,9 +261,7 @@ System::Clock::Timestamp ReliableMessageMgr::GetBackoff(System::Clock::Timestamp
     }
 #endif
 
-#ifdef CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST
     mrpBackoffTime += CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST;
-#endif
 
     return mrpBackoffTime;
 }

src/messaging/ReliableMessageProtocolConfig.h

@@ -162,8 +162,10 @@ namespace chip {
 #ifndef CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST
 #if CHIP_ENABLE_OPENTHREAD && !CHIP_DEVICE_LAYER_TARGET_LINUX
 #define CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST (500_ms)
+#else
+#define CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST (0_ms)
 #endif
-#endif // CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST && !CHIP_DEVICE_LAYER_TARGET_LINUX
+#endif // CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST
 
 /**
  *  @brief

src/messaging/tests/TestAbortExchangesForFabric.cpp

@@ -206,11 +206,10 @@ void CommonCheckAbortAllButOneExchange(nlTestSuite * inSuite, TestContext & ctx,
         // trigger a MRP failure due to timing out waiting for an ACK.
         //
         auto waitTimeout = System::Clock::Milliseconds32(1000);
-#ifdef CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST
         // Account for the retry delay booster, so that we do not timeout our IO processing before the
         // retransmission failure is triggered.
         waitTimeout += CHIP_CONFIG_RMP_DEFAULT_MAX_RETRANS * CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST;
-#endif
+
         ctx.GetIOContext().DriveIOUntil(waitTimeout, [&]() { return false; });
     }
     else

src/messaging/tests/TestReliableMessageProtocol.cpp

@@ -58,17 +58,13 @@ using TestContext = Test::LoopbackMessagingContext;
 
 const char PAYLOAD[] = "Hello!";
 
-#ifdef CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST
-// When the CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST is set, the stack
-// operates under the assumption of a high latency network (like Thread),
-// so it adds extra delays to avoid spurious retransmits.
+// The CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST can be set to non-zero value
+// to boost the retransmission timeout for a high latency network like Thread to
+// avoid spurious retransmits.
 //
 // This adds extra I/O time to account for this. See the documentation for
 // CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST for more details.
 constexpr auto retryBoosterTimeout = CHIP_CONFIG_RMP_DEFAULT_MAX_RETRANS * CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST;
-#else
-constexpr auto retryBoosterTimeout = System::Clock::Milliseconds32(0);
-#endif
 
 class MockAppDelegate : public UnsolicitedMessageHandler, public ExchangeDelegate
 {

src/protocols/secure_channel/tests/TestPASESession.cpp

@@ -279,7 +279,8 @@ void SecurePairingHandshakeTestCommon(nlTestSuite * inSuite, void * inContext, S
 
     while (delegate.mMessageDropped)
     {
-        chip::test_utils::SleepMillis(100);
+        // Wait some time so the dropped message will be retransmitted when we drain the IO.
+        chip::test_utils::SleepMillis((100_ms + CHIP_CONFIG_MRP_RETRY_INTERVAL_SENDER_BOOST).count());
         delegate.mMessageDropped = false;
         ReliableMessageMgr::Timeout(&ctx.GetSystemLayer(), ctx.GetExchangeManager().GetReliableMessageMgr());
         ctx.DrainAndServiceIO();