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

  1. Running in kernel is worse than running as root.
    161
  2. Aren't these are intended for data center where customers lease VMs or some other slice ? AMD has encrypted RAM....
    129
  3.  @cassandrasibley228  well, some of those xeon server processors end up in high-end workstations. I suspect higher end workstations might have take a mid road between core count and individual core performance.
    61
  4. ​ @LowLevelTV you are scholar and a gentleman, sir. Though we disagree, I am impressed at you reaction to a mistake.
    46
  5. Torvalds doesn't complain about C and it gets used a lot.
    34
  6.  @AnarexicSumo  no. you cannot, apparently. the extreme limit of all technology is an attempt to fill a need for the customers that use it. The customers that use chips with 100s of cores are data centers and compute utilities. Learn about 'em if you want to, but also learn why they were developed, it's part of the story. Then again... Maybe saying this is like saying rocketship a are designed for astronauts, which isn't quite right... It's more like astronauts are trained to fly rocketships for the purpose of space exploration and experimentation. We all benefit from that but it's immediately apparent that you're probably not getting a rocketship anytime soon. If you do get a rocketship sooner rather than later, well then I'll be your copilot.
    30
  7. I recommend that people start with C BECAUSE you need to be aware of how memory works. Avoiding it because beginner developers might write unsafe code just delays the inevitablity of using a linter to identify where there's a security issue.
    29
  8. Oooh. I know this one. Its the function epilogue. Functions get compiled into some bits and the function calling conventions have a prologue that stuffs the current program counter pushed to the stack. Return just pops that value back when the function returns. This is why recursive function calls can overrun the stack
    17
  9. I think the C language will riegn supreme. Rust will be a side note in history like Eiffel and Algol.
    11
  10. Yeah, took me a few weeks to figure this out. You can make a career outta the stuff you're trying to learn, keep it up! These cortex-m cores have dozens of hardware interrupts.
    6
  11.  @dahahaka  well, depends. Running inside the kernel could cause a kernel panic and crash the whole system, running as root just causes a segfault
    5
  12. Good explanation on BJT transistors, but we use CMOS transistors. CMOS is fundamentally different mechanism that use similar materials but BJTs are current controlled devices while CMOS are voltage controlled devices. CMOS is a refactor optimized for on/off state. BJTs are good for analog operation amplifiers, radios, etc...
    5
  13. C is an ancient language many attempt, but few master. Remember well that your data structures are key to successfully implementing the algorithms.
    4
  14. Lemme guess, you ran a linter over the code to find all these seven bugs. Why does a whole new language need to enforce something that static code analysis can solve.
    4
  15. C plus a decent linter and fixing such warnings results in memory safe. You don't really NEED a whole new language to do this. Just use better static analysis.
    4
  16. yes it's hard, but running io stuff on a thread is great no matter c or assembly. It feels so good when you get it to work!
    4
  17. Those poor hardware guys... Are people I know. Be nice, please. They mean well.
    4
  18. Void pointer... its a pointer to... something.
    3
  19. Well, should I suggest getting enough random number for 256 bits, then run SHA-256 on it? That might obscure it a little more...
    3
  20. You can define such symbols in your linker script, memory management in constrained systems usually requires lots of predefined locations in a real memory map. I worked with a team to copy several critical functions to SRAM section of a constrained system, most of the program flash was on a NOR chip on a QSPI bus. We defined a specific bank of SRAM to be an instruction cache and used half of it to store functions for interrupts, and the other half was a LRU cache for several more functions. I guess it was kind of a dynamic linker with a shared object with no one to really share it with, though I heard later on that they implemented a tasking system that also ran out of the same space as the IRQ handlers. Many of those functions had to be fixed for reentrant requirements, lots of bugs, but the linter we had made this easy.
    3
  21. Awe ... Of course you think its over hyped. You're a rust apologist.
    3
  22.  @bakedbeings  I didn't even understand his comment. Still confused... I usually disagree with Capt Rusty on this channel so I thought I would throw him a bone.
    3
  23. Ooooo this is so cool. Stack smashing was a thing that might be related.
    3
  24. Drink the Kool-Aid!
    3
  25. The functional wizard has spoken. (Pay no attention to the man behind the curtain)
    3
  26.  @insertoyouroemail  I feel a strange desire to obey you...
    2
  27. We kinda take what we know for granted. Or granite, I really don't remember how that idiom goes.
    2
  28. Learn without an IDE. you gotta know how your compiler and linker works.
    2
  29. Well, there's always random cosmic particles.
    2
  30. And C is for men.
    2
  31. I've been there once or twice!
    2
  32. Magic numbers BAD. use a macro, damn it!
    2
  33. I think if you used the SHA-256 of the random number, it would be a lot less trivial to predict. I dunno, good random numbers have certain characteristics. An RNG could generate "1234" ( which is the same code for my luggage) but it's unlikely...
    2
  34. Why do you dehumanize people for being financially successful? Billionaires are people too? There was a group of German National Socialists that would say similiar things about the Jewish people back in the 1930s. Its not cool to discriminate against anyone. There WAS a group of people called the Kulaks which were associated with wealth, they were successful farmers who grew the food in Russia. as they were successful, they had stuff that other people envied and ultimately, they were reviled the same way you complain about billionaires. Some kulak gets a new tractor, it was just that dirty kulak trying to increase his profits. During a revolutionary time in Russia, the Kulaks were all executed and the farms were transfered to the proletariates from the revolutionaries. well, there was a famine so bad, somehting like 40 million people died. So... examine your conscious, son. You're on a slippery slope.
    2
  35. You can totally do OOP in plain C. Its just ugly.
    2
  36. Good idea.
    2
  37.  @Stopinvadingmyhardware  I second that.
    2
  38. We refined those rocks and jammed dope into them, then hit it with lightning....
    1
  39. I think the "finFET" stuff that intell figured out was pretty important, I think they all do it that way now.
    1
  40. I am going to see your security researcher and raise you some practical skills issue researcher.
    1
  41. C++ needs a system for exceptions and the "new" keyword. malloc doesn't count because its not a keyword, its a function call to something in stdlib.
    1
  42. Yes but can Rust self host itself? Can the Rust compiler and env be written in Rust?
    1
  43. Elegant solutions or syntax sugar?
    1
  44. _sbrk...
    1
  45. Learn c. You can read more code.
    1
  46. People should use a linter.
    1
  47. ​ @dynfoxx  you have never used any static analysis tools. There's really no magical code that's in a compiler that can't be implemented in a a static analysis tool. It's braindead easy to implement lifetimes on memory locations by scanning through the abstract symbol tree
    1
  48. ​ @dynfoxx  also C++ does have object lifetimes and a garbage collector, but this is generally craptacular and pales in comparison to just managing it yourself.
    1
  49. My recommendation: avoid inheritence. Embrace composition. Abstract classes for interface inheritence isnt really inheritance. Its implementing an interface, though i think java has explicit syntax for this.
    1
  50. I wouldn't trust the compiler to do that, and even if it were guaranteed by your current compiler, that code may not be portable to other compilers.
    1
  51.  @bozingo  functional code NOT horrible. 😂 😂. Ha ha ha ha ..
    1
  52. Alternatively, you could learn how to write C well enough so you don't corrupt memory. Static analysis could help establish enough evidence for memory safety.
    1
  53. It's not so much that rust runs on everything, you just have a toolchain that compiles to whatever janky BS that is AVR.
    1
  54. Plain C is supposed to be snake case!
    1
  55. No. All you need is C.
    1
  56. ​ @everyhandletaken  ha ha oxidation... Rust. Well played sir
    1
  57. Rust still sucks. What does the whitehouse actually know?
    1
  58. Why can't the ideas in rust be implemented for C through some sort of linting or other static analysis?
    1
  59. ØMQ is a facade then? It kinda sounds like a Plan9 channel.
    1
  60. Does this trigger a hard fault like IRQ 3 in arm cortex-m0+?
    1
  61. Well... Technically... Interpretation vs decoding in the CPU architecture is a wee bit of a gray area. You could design any architecture to be this way...
    1
  62. No... No... Do fixed point math. Edit: Oh yeah man. You got it man. Godbolt FTW.
    1
  63. Hardware makes randomness pretty good. Need hardware module to measure entropy in the environment. Hmm.... What if you could get entropy based on the weather report, direction or the wind, or I dunno just put an antenna that listens to static?
    1
  64. ​ @y00t00b3r  400 pounds of prefrontal cortex. I am just a brain in a jar of nutrient solution.
    1
  65. Ew. I think i will just use better pointer hygiene
    1
  66. Magic number, the number 42 is. Linting your code you shall, errors avoid you will.
    1
  67. Also... Recursive functions are not really deterministic, which is probably something important in an embedded system, you often want predictable behavior
    1
  68. F
    1
  69. How dare you make this so much simpler! You must maintain the dark arts of firmware!!!
    1
  70. Magic numbers inline bad. Use macro.
    1
  71. Oh. I watched this already....
    1
  72. Hum... I am working on the same thing, though I am targeting a few different chips... Sorta built-in portability. So... Lemme get your opinion. The OS needs to set up each process stack (and initial context frame), each process heap, schedule each process, and provide mechanism for IPC?
    1
  73. CMake is the devil incarnate! Meson is better.
    1
  74. F
    1
  75. First transistors were made of germanium, not silicon.
    1
  76. Rust is silly.
    1
  77.  @9SMTM6  well sure, FETs are more fundamental than CMOS, but the complimentary part is an essential to the design. In most logical like a flip flop, there's equal numbers of PMOS and NMOS components. I think there's at least 4 or 5 pairs in a CMOS flipflop for a total of 8 or 10 transistors per bit. Wikipedia is cool, but take it with a pound of salt. The complimentary part is important to make these "gates" reliable, energy efficient, switch fast, avoid latching, etc. etc. CMOS is the best refactor optimized for binary operations. It's way more than just a manufacturing technique. But technically, you're correct. It's more like FET is important.
    1
  78. Look up tables? Edit. Nope... I was wrong...
    1
  79. Yup. I learned the same thing in my physics degree. Experience matters more.
    1
  80. These restrictions are actually pretty durn good for embedded programing.
    1
  81. ​@William Drum infinite recursion loops are a problem. I recall an XML parser that would run into memory issues loading an XML with a billion lols. I think that was a heap issue, not a stack smashing issue, but it was an example of recursion being somewhT bad.
    1
  82. I brought this in for show and tell at work... They adopted it for a piece of legacy firmware.
    1
  83. Parallelllilize.
    1
  84. X86 isn't arm.
    1
  85. Design your own cpu based on the riscv architecture
    1
  86. Spinnaker has thousands of ARM968E custom CPU cores. I think each has a bank of tightly coupled memory for each core.
    1
  87. Magic numbers kinda suck too, but that's a mistake in all languages. Also... What's up with that noop? Are there not proper timers? Also ... Screw make. Get Meson and Ninja. Much much better
    1
  88. Linux is a great application that runs pretty well on my windows PC.
    1
  89. Hmm... Unlikely encrypted. They licenced an arm core for their ASIC. There's a possibility they implemented encryption in hardware.
    1
  90. Someone recently found a alpha64 windows2000 install. It was never released, but they wrote it. They even found a compiler forit.
    1
  91. Why? They are just null terminated sequences of bytes. It's a fairly easy data structure, one of the easiest.
    1
  92. Now do OPAQUE pointers!
    1