is anyone here familiar with creating new sections in a linker script for something like uncached memory? I can probably create the section with something like this:

```txt

MEMORY
{
  /* Zedboard: 512 MB DDR3. Only use 63 MB for now, should be plenty for a bare-metal app.
  Leave 1 MB of memory which will be configured as uncached device memory by the MPU. This is
  recommended for something like DMA descriptors. */
  CODE(rx) : ORIGIN = 0x00100000, LENGTH = 63M
  UNCACHED(rx): ORIGIN = 0x4000000, LENGTH = 1M
}

REGION_ALIAS("DATA", CODE);

SECTIONS
{
  /* Initialized data in uncached memory */
  .data.uncached : ALIGN(4) {
    _sdata_uncached = .;
    *(.data.uncached .data.uncached.*);
    _edata_uncached = .;
  } > UNCACHED AT > CODE

  /* Zero-initialized uncached memory */
  .bss.uncached (NOLOAD) : ALIGN(4) {
    _sbss_uncached = .;
    *(.bss.uncached .bss.uncached.*);
    _ebss_uncached = .;
  } > UNCACHED
}
```
 but I am not sure how to handle uncached data and bss handling. I suppose the best thing is to do that in the user application then which requires data structures in uncached memory? Or using `MaybeUninit<u8>`? I looked whether https://github.com/rust-embedded/cortex-m/blob/master/cortex-m-rt/link.x.in had special handling for something like this, but did not find anything.