one last annoying question ^^

again: shared libraries

whenever i call a function of a shared library, the called function needs to work on the shared library's data.

thus: do all exported shared library functions do somthing like:

Code:

some_shared_function {
  mov somewhere, ds // back up the current data segment
  mov ax, (the data segment of the library) // prepare the shared libraries data   segment for loading
  lds ax ; load the data segment of the shared library into ds

  (usuall calculations and stuff)

  // restore the old data segment
  mov ax, somewhere
  lds ax ; restore old data segment
  ret // return to the caller
}

or how do shared library functions access their data?

have you ever thought that the compiler/Os do that for you?

well im asking HOW it is done
not WHO does it for me

cause in my assembly book it just says: "when you want to access another data segment, simple use lds to load a segment selector it into the ds register, or use a segment override prefix"
ok, got that.

then when looking what the inter-segment-call-instruction does it basically sais:
intersegment-calls require a logical address, thus a segment selector and an offset,
then cs is set to the segment selector, and eip is set to the offset.

so performing a far call would set cs to another code segment, but ds is left unchanged. thus if i call a shared library function, that function would work on *my* data, and not *its* data, since ds "points" to my data segment.

so am i right that in the prolog code of shared library functions one action is to load the appropriate data segment into ds?

the reason why i try to understand that is cause i want to write a bytecode interpreter but i would like to have shared library support in mind. thus i have to understand whats going on