That's interesting. Are there many instances where you would prefer malloc during real-time processing compared to pre-allocated arrays? I used FreeRTOS' heap_4.c [0] to run malloc-based test code on systems without native malloc, but I wouldn't consider using it post-initialization [and only used it in test code], so I didn't really care about performance.
[0] https://www.freertos.org/a00111.html