The bare-metal linking model focuses on the conventional embedded market where the whole program, possibly including a Real-Time Operating System (RTOS), is linked in one pass.
The linker can make very few assumptions about the memory map of a
bare-metal system. Therefore, you must use the scatter-loading mechanism if you want more
precise control. Scatter-loading allows different regions in an image memory map to be
placed at addresses other than at their natural address. Such an image is a relocatable
image, and the linker must adjust program addresses and resolve references to external
symbols.
By default, the linker attempts to resolve all the relocations statically.
However, it is also possible to create a position-independent or relocatable image. Such an
image can be executed from different addresses and have its relocations resolved at load or
run-time. You can use a dynamic model to create relocatable images. A position-independent
image does not require a dynamic model.
With the bare-metal model, you can:
Identify the regions that can be relocated or are position-independent
using a scatter file or command-line options.
Identify the symbols that can be imported and exported using a steering
file.
You can use --scatter=file with this model.
You can use the following options when scatter-loading is not used:
Arm’s Privacy Policy has been updated. By continuing to use our site, you consent to Arm’s Privacy Policy. Please review our Privacy Policy to learn more about our collection, use and transfers of your data.