Memory Mapping

• read and write operations on memory are done with one instruction … so how??? can a kernel intercept such operations and update a file acordingly

  • we have the same problem, with how do me access device memory buffers for Buffer Management and Allocation (computer graphics)

• after hours i finally found the answer here: https://wiki.osdev.org/Paging#Paging_Tricks

  • when the user accesess a byte of the mmap’d file, a Page Fault occurs … the OS then loads the bytes for this full Page (memory) into a newly allocated memory page, then adds entries in the Page Table for this Page (memory), then resumes the userspace program
  • but how does this work now with mapping buffers from other devices????
    • when you request a full Page (memory) of that devices buffer, you have it in two locations and get a massive syncronisation problem…