Comments by "Michael Lenczewski" (@kayakMike1000) on "LaurieWired" channel.

  1. I think a picture would have helped. There's a whole world in here that goes way beyond MMU and virtual memory. Context switching comes to mind...
    5
  2. ​ @Tomahawkist_ I don't think she really knows java bytecode from a decompiled object file.
    2
  3. Recursion is somewhat frowned on in embbedded programing.
    2
  4. So.... Reverse engineering some code? Like decompiling the binary? I hope you like assembly.
    2
  5. ​ @bobbobson6290  You could practice a lot and get better at writing code. What do you hope to do in Cyber security?
    2
  6. You skipped microcode.
    2
  7. Largely a solved problem, it's a skills challenge.
    1
  8. All cybersecurity experts should do a full implementation of the rijndael encryption algorithm in portable C. Compile that code so it runs on a microcontroller or other small chip. Then that expert should try to side channel their own code.
    1
  9. Here I thought you were going to go over context switching...
    1
  10. I love the modularity. Its kinda fun to figure out how the modules could work using logic gates
    1
  11. Not knowing how much memory your program needs seems kinda weird...
    1
  12. Magic? No.
    1
  13. Virtual memory.... Hmm. I wonder if i could program in virtual memory paging into a tiny OS context switcher.... I guess the hardware support in the MMU saves pointer arithmetic with every memory access. Wait... Now i am thinking the context switch needs to know what page the current process needs.
    1
  14. Why c++? Plain ol' proedural C seems like it would be pretty good.
    1
  15. Why do this at all when there's compiler to do it for you? Is it just like mountains, you climb yhe mountains because they are there?
    1
  16. Why ARMv7? I thought raspi was aarch64
    1
  17. Based on what you're saying, I don't think you know what the effe you're talking about.
    1
  18. I wrote a deterministic memory allocator in C, for microcontrollers with tons of SRAM. Unused blocks were all saved in a linked list. Allocating memory popped blocks off the top of the list, but only if they were within the size requested. I also updated a journal that could track who was using which blocks... Sorta like the other side of the svc call. I was imagining a tiny OS that ran little apps or could perhaps plugin to the kernel somehow. Requesting memory needed an ID so a misbehaving app could be cleaned up. I was also considering configuring the MPU (if available) so any access requests outside the requested allocation could be trapped. I think you could get at least 8 different regions setup that way...
    1