smukherjee19

Good findings, but the findings are so AI-generated and hard to read… Feels like I am being thrown numbers without a coherent story, and even worse, with beautiful animations which only serves as distraction… This looks very similar to the diary entries my LLM writes when working on a coding project.

Stereotyping mathematicians, well played sir.

I wonder if this paper has been peer-reviewed at a decent conference/journal.

OP here. Thank you so much for taking the time to write it!

This is so much easier to understand, and I also learnt a few things along the way, like the zero page mechanism, munmap having to notify the MMU about the invalid TLB and more.

Really appreciate it!

Unfortunately, the kernel bug would need to be understood to be patched, and the developed patch itself would need to be reviewed before being accepted, which seems... very hard, to put it mildly, using the nails-on-chalkboard LLM spew report.

Thanks for the explanation. I should have looked up each term a bit more.

So "backing" is the underlying physical memory the page maps to.

So if I were to make sense of that piece of slop, is this what it would be?

"In one thread, a page fault happens during a store operation. The physical memory address is the proper address, which is immediately (around 10 assembly instructions later) replaced by the 0x00000000 memory address by another thread due to race condition, resulting in a crash when the original thread tries to read that page again."

I tried reading and gave up at the "Headline"...

Quoting from the bug analysis:

>Headline. The crash is real and reproducible. With musl instrumentation we pin the in-process mechanism precisely: a thread's own store to a freshly- faulted anonymous page becomes invisible to that same thread's reload ~10 instructions later, because the page's backing is replaced mid-function. A pagemap read at the instant of the fault shows the backing is the kernel's zero page. A captured core dump confirms the crash site with matched virtual addresses. The mechanism localizes to the interaction between the per-VMA-lock anonymous-fault fast path and a concurrent munmap's TLB shootdown. A source-level review of Linux 7.0.12 identifies a specific race in that interaction, and a git comparison across v6.19/v7.0/v7.1/mainline identifies the v7.0-introduced change on the munmap-teardown side that widens it.

"a thread's own store to a freshly- faulted anonymous page becomes invisible to that same thread's reload ~10 instructions later, because the page's backing is replaced mid-function." -> ??? What's a "backing" of a page? Freshly-faulted? Fresh fries? "~10 instructions later"?

"A pagemap read at the instant of the fault shows the backing is the kernel's zero page." -> Backing?

"The mechanism localizes to the interaction between the per-VMA-lock anonymous-fault fast path and a concurrent munmap's TLB shootdown." -> How can a mechanism "localize"?

And more.

This is... words strung together. Nothing more. I wonder how people read and make sense of this.

In case people have forgotten what real technical writing looks like, here's a sample (I am not the author): https://yifan.lu/2019/01/11/the-first-f00d-exploit/

Clickbait and aggressive title, shallow article. I read it and not worth clicking.

That's weird. I use Gmail regularly and did not experience anything like that; it works just the way it did for the past 10+ years. And auto categorization still works just fine, and no AI summarizing my email or trying to write my drafts. I wonder what difference we have in settings.

Also, not to be disrespectful to the OP, but seems quite an... overblown reaction. To each their own, though.

Just a guess, with Macbook Neo releasing, M$ likely got scared of mass exodus to MacOS (the average PC user can now consider a Macbook and it likely wipes the floor with all Windows laptops out there) and decided to throw in a damage control piece.

Yes, especially if people living in the city. I have known Japanese people who can’t speak English well but can read technical CS papers and understand well enough to give a summary and presentation in Japanese.

Just keep in mind they are usually very good in reading, okayish in listening, and kinda needs work on speaking. But that’s expected. If you live a daily life in Japan like the Japanese, you barely need to speak English, or hear it, if at all. Even the foreign staff at the convenience store speak Japanese good enough for them to carry on their duties.

I clicked "Accept" because I didn't want to deal with it later, especially given the new terms, and Claude training on new chats and resuming existing chats from September 28th,

but the "Accept" and "Not Now" are for the new terms of use.

So by toggling off and clicking "Accept", it will accept the toggled version. You can check the setting after accepting it at: https://claude.ai/settings/data-privacy-controls It's the "Help Improve Claude" button.

Guy actually went full-in. Amazing.

Agree on the formatting. Also, not using LaTeX, when it is pretty much a standard in this field.

Is there any way to access the models without:

- Linking the chats with my personal account - Having Anthropic train the model with my data?

Like, having the knowledge of the model with the privacy of local LLMs?

I personally don't use any AI Code Assistant, but I did find the below video a level-headed and nice analysis of the subject, where the author uses 4 different AIs to build an HTTP server in Python.

Kinda different from your specific use case, but should give some hints on which one would serve you best, and is an interesting watch:

https://www.youtube.com/watch?v=-ONQvxqLXqE

Sorry to hear how Yelp treated the OP...

Given how many stories of greed and throwing people under the bus for money I hear nowadays, we might already be living in a dystopia.

As someone who moved to Japan over 8 years ago, the observations make sense!

Also, if you're considering "living" in Japan (which is a totally different beast from visiting as a tourist), here's a hypothetical scenario. See how you feel the first time you read it.

You, alongwith a group of friends/group tour members are walking down the street towards the station to go to your next (sightseeing) destination in Tokyo. Assume none of your friends/group tour members are ill-meaning (as in, they don't behave like bad tourists, littering the streets, being loud because they want to etc.).

Now, assume that the group you are with is being louder than other locals around you, possibly because they are in good spirits after a good meal or sightseeing experience. What is your way of thinking:

a) Do you realize that your group is being louder than other locals around you?

b) If you realize so, do you still not feel anything and keep walking with the group?

c) Or do you feel uncomfortable that your group is being louder than others around you, and want to distance yourself away from your group?

If you're more inclined towards c), then your way of thinking is closer to the Japanese people.

Source: had the same experience when I visited Japan over 10 years ago for the first time, and I was so uncomfortable it felt like torture just walking with the same group.

Japan, for better or worse, is "unique" in many ways even today, including how people think and behave. It's not for everyone, but if you think the article's observations and the scenario above are something you like and want to be a part of, you'll likely not have too much of a hard time adapting to life here.

Also, if time and energy permits, please, learn the language. It'll expand your world so much here.

I have to agree. I don't mean to disrespect the existing app, it looks really good and well-polished, but... as someone new to the MacOS ecosystem, $150 feels highly pricey. It's as costly as a Windows / Office license!

At the same time, there are apps like Mindkit [0], which seem to do similar things as this, but with more features, but which cost 3000 JPY(20 USD) lifetime.

Personally, the only time I can justify 150 USD for a purchase of an app is when it has as many features or is as impactful as Word/Excel/Powerpoint combined.

No disrespect to the people who made it, good luck with the app!

[0]: https://apps.apple.com/us/app/mindkit-keep-things-organized/...