Can SRAM1, SRAM2, SRAM3 be treated as one continuous memory region?

[inojosh]

(This is regarding STM32H75x)

SRAM1, SRAM2, and SRAM3 are separate in the memory map, but they are continuous - there are no gaps between them:

 

Can they be treated as continuous with respect to the linker, for example:

RAM_D2    (xrw)    : ORIGIN = 0x30000000,   LENGTH = 288K

And then place some arbitrary data in that region?

Like, can we place an array that is 288k in that region and use it as a heap, or will bad things happen at the region where SRAM1 ends / SRAM2 starts, etc?

[inojosh]

I feel like there is a simple answer to this but I haven't been able to find it...

[Tesla DeLorean]

From the processors perspective they look/appear contiguous.

From and DMA/BUS perspective they might be treated differently, you'd have to review in the context of the bus matrix diagrams

Clocks should also be enabled.

[alister]

To use as heap, they're contiguous and the linker doesn't care what they, read https://sourceware.org/binutils/docs/ld/MEMORY.html#MEMORY.

  SRAM123 (rw) : ORIGIN = 0x30000000, LENGTH = 0x48000

Otherwise find your part's RM and other resources, e.g. AN4891, at https://www.st.com/content/st_com/en/search.html.

[Pavel A.]

If this is relevant for your application, test whether multi-byte writes over the boundary of the RAMs don't cause trouble.

For example: *(uint32_t *)(0x3003FFFF) = 0x12345678; *(uint32_t *)(0x3001FFFE) = 0x12345678;

-- pa

[inojosh]

Thanks everyone.

One interesting thing: I did not realize that clocks had to be enabled for these SRAM blocks, and I was somehow using them just fine without those clocks enabled.

I am not sure how... but I have added the clock init in.