Hi Everyone,
I am working on the K64 and I have some questions about the use of SRAM. I already consulted the links below, but I still have some ambiguities:
1-When i have done a memcpy of a buffer from SRAM_U to SRAM_L the system crashed.
2-The Cortex M4 ref Manual mentionned that the SRAM section starts from the adress 0x20000000 and can have 0.5GB as size:
So why we don't use a unique sram block of 256Kb starting from 0x20000000 ?
3-Can i use the totality of the SRAM_U and SRAM_L ? There's no risks? (apart misaligned variable declaration of course)
Links:
https://community.nxp.com/message/844720?commentID=844720#comment-844720
https://mcuoneclipse.com/2013/07/10/freertos-heap-with-segmented-kinetis-k-sram/
https://community.nxp.com/thread/336126
https://community.nxp.com/thread/384917
Best Regards.
Hi all,
1. Besides what Mark said, it is also important to make sure that any object is not stored by overlapping both SRAM sections (just like it is mentioned in the discussion from the community you checked: https://community.nxp.com/thread/437160?commentID=844720#comment-844720). This last issue doesn't happen if you use MCUXpresso (the IDE helps you fix the issue).
2a. In the following table from the k64 Reference Manual you can check that indeed it can only be used up to 256k. The memory map you used refers to how much memory the M4 could handle, not how much it has:
2b. Due to design reasons, you cannot use a unique SRAM block of 256Kb starting from 0x20000000; the k64 is designed with a Harvard architecture and is set with SRAM_L accesible by the code bus and SRAM_U accessible by the system bus (these buses access the memories through the addresses set for the k64).
3. You can indeed use the totality of the SRAM_U and SRAM_L as long as you remember the precautions you must take that were mentioned by Mark and myself.
I hope this information can help you.
Ricardo Delsordo
