Look at the Silicon Labs (formerly Cygnal) part lineup. They have many parts wherein small blocks of on-board FLASH memory can be written by the application program at run time. The application code that writes to the FLASH is able to simultaneously be running out of other FLASH blocks.

Note that not all microcontrollers offer this nice feature. Some require that the FLASH be written from software algorithm code that is running from a special onchip ROM during the time FLASH is accessed. Some other microcontrollers will require that the software code that does the FLASH access be code that has been moved to RAM. This latter method is more common on non-8051 type microcontrollers that use a common address space for code and data memory.

I have previously published here an example loader routine that contains in-application FLASH writing code for the Cygnal C8051F126 processor. Here is a link to that code should you find it interesting:

C8051F126 Loader Code

Michael Karas