On Mon, Jul 17, 2023 at 09:02:07AM +0200, Dmitry Vyukov wrote:
Hi Geraldo,
What exactly is cryptic in the report? Is there anything that can be done to make it less cryptic?
Hi again, Dmitry.
Perhaps also a bad choice of words. Cryptic borders on the undecipharable while esoteric is the more proper word here. Those kernel hackers with esoteric C and assembly skills like Takashi Iwai or you will quickly infer that a kfree() is missing in such and such scope.
In my other message, I meant to say that such esoteric knowledge is barely possessed by a novice kernel hacker, and they end up adding noise to the lists specially if they are involved in the patch acceptance process, specially as author of the patch, which I'm neither in this case.
Now, if somebody were to apply LLMs to the build and checker bots and actually get to a point where they were getting good patch propositions from the machine rather than a bunch of hallucinations, that would be quite the feat. It's only a faint dream right now, but you did specifically ask for the "vision" :)
Thank you, Geraldo Nascimento