Unable to send 128 bytes of data using the pingpong example

I ported the pingpong example to the STM32F103 with the initial configuration : SF=7 BR=125k CR = 4/5. I can send and receive 64 bytes of data.
If I change the #define BUFFER_SIZE 64 to #define BUFFER_SIZE 128, only the master side sends data successfully and the slave side cannot receive the data.
The lora program I wrote myself is able to send and receive 128 bytes of data normally.The flow of the program is: configure fifo, configure send, block and wait for DIO0 pin to pull high.
I’m not quite sure what causes this difference. If there are details of the pingpong example that I have not noticed, please let me know.