Posts
4817
Following
319
Followers
489
Linux kernel hacker and maintainer etc.

OpenPGP: 3AB05486C7752FE1
@kernellogger Looking at this thread, I get the feeling that maybe some of the comments are a bit over the top, no matter what has happened.

Personally I think that whatever goes in any possible mailing list I don't personally hate anyone. I don't mind meeting any person IRL. It's just for me personally plain shit posting that needs to be addressed and call it a day. Of course it is totally unprofessional and lame but not worth of this scale of group wanking.

I don't really even know this person, or use bachefs but as a person with some empathy I can only imagine how bad he might feel if he ever comes to is senses, and as an optimist I I'd like to phrase that "as he comes to his senses" :-)

What Linus toned his response as a decent person should, at least in my world view. He even described Kent as a smart person.
1
0
0
Sitten kun on tämä Purra vielä, joka vertailee itseänsä ilman huumoria tai itseironiaa, johonkin Risto Rytiin, ja jonka ainoa CV-merkintä taitaa olla PS:n puoluetoimisto, niin eihän tämä kokonaisuus kovin hyvältä näytä. Ei sillä ole käsittääkseni oikein mitään relevanttia talousosaamista, ja sitten oma käsitys taidoista on lievästi sanottuna ylioptimistinen.

Karmeetahan tätä on sivusta seurata...
0
0
1
@kernellogger sometimes i feel that we should have a "butthurt award" for kernel maintainers (inspired by darwin award) ;-)
0
0
3
@dngrs True,I agree, and like in my case I get the job done with that other tool. It's exactly a nice to have feature as far as I'm concerned.
0
0
0
@dngrs thanks but nope, more like wondering what keeps https://github.com/rust-lang/rust/issues/90957 open... i've been occasionally following this since 2022-23 when it was being considered for Enarx so there is nothing new to me. Just wondering what keeps the issue open up to date, since it has been a while.

It was not critical issue then and likely not now as global_asm!() allows to both write inline multi-line snippet of free form assembly code, or even include it from a file (refer to documentation). For me it would be at most cleaner syntax for common use cases (and better match with C's naked functions).
1
0
0
@rolle Sen käsitykset pienituloisista ainakin vilpittömän oloisia, täysin paikkaansapitämättömiä ja samalla jäätävän ylimielisiä (Petteri-setä tulee vähän ojentamaan köyhiä ihmisiä oikeille urilleen, jotka ei ymmärrä omaa parastaan).

Tämmöisissä hyvissä tuloluokissa Petrin asennemaailma on valitettavasti ihan jopa mainstriimiä. Kun ei tunneta kuin just jotain turkulaisia vapaamuurareita, niin tollainen tukari siitä voi vaan seurata :-)
1
0
2
So what is going ATM with naked functions and Rust? Apparently it is still unstable. I don't really see what it could accomplish what could not be accomplished with global_asm!() macro tho.

Just looking into stuff related to my new job and also reasoning about Enarx, which I'm (still) interested to re-burnish once getting more into developing Rust again in my day job after 1.5 years break.

They share similarities as both delegate I/O to host and calculate in the guest and have predefined address space.

#rust #rustlang #antiguachicken
1
0
0
Problems do not arise when let's say daemon logs "1024 MiB". Then can be quite safely assumed the aligment with ISO/IEC 80000-13.

However, if the same daemon emits "1024 MB" instead, I need to download the full source code and study it enough to be sure what is what.

Obviously this mess applies also to hardware.

It is a cardinal sin to break backwards compatibility just plain randomly. Should be dead obvious for any legit engineer that instead it would have been better to invent a new unit for base-1000 sizes, instead of corrupting base-1024 sizes.
0
0
1
@karinjiri it's sick, agreed, but it is no nearly as bad as introduction of mibibytes and changing sizes as being ambiguous entities :-)
1
0
0
Edited 11 months ago
Whoever thinks that getting cute by replacing precise timestamps with "1 hour ago", "3 weeks ago", "6 months ago", "recently" etc is a good idea:

STOP IT!

I am particularly glaring at you, github! I don't know if you did it first, but so many other people copied this blight on humanity from you.
3
13
0
@thejpster Cool, and pretty clever work! Just thought that a research group matching country, size-to-fit OS and using Rust so at least three correlation points :-)

Heh, that "nobody's using XXX" always does the triggering job for me :-) Actually, if I consider any tech conference, I'm *most* interested on presentations where someone is doing something with tools that nobody's using. That's why I e.g. went to see a Servo presentation back in LinuxCon EU 2014. I went to see it exactly because nobody was using Rust, and thus it grabbed my attention. Popularity is not really a recommended way to approach tech really.
0
0
0
@thejpster Why anyone would even care if any possible tool is used by anyone else? In commercial setting, anyone else usually means competitors, so an overlooked tool is merely a competitive advantage. No logic in that comment tbh as the outcome is for the most part irrelevant :-)

There's a really cool hard RTOS framework from Sweden focused on Cortext-M, and more recently ported to RISC-V, based on hardware priority based scheduling. Wonder if they are basing on that: https://rtic.rs/2/book/en/. Met the creator randomly about a year ago (was visiting at the local university in Tampere, FI), and he gave a great presentation about their research groups work. Was quite amazed.

Its task creation and other shenanigans are essentially a small domain language based on Rust macros and end outcome is bare-metal blobs.
1
0
0
Edited 11 months ago
Looking at statistics and error ranges and not much can be said about outcome:

https://ig.ft.com/us-elections/2024/polls/

Neither do I believe that world ends to either outcome.

We've have had in Finland some crazy and/or drunkard dictator who wants to destroy the planet leading the neighbor country for most of the post-war history, and usually terms have been "up-until-being-a-corpse", not four years.

Not going to tell, which of our neighbors I'm talking about, just in order not to point any fingers :-) For most it would be better probably to put more focus on elections where you have the voting privilege in the first place.
0
0
0
@ljs @vbabka ya i agree 😀
1
0
2
@ljs @vbabka right, so you are having a conversation about the chickens in Antigua while I preach about the chickens, got it :-)
1
0
2
Edited 11 months ago
Phew, all biometric and also ThinkShield and ThinkCloud shenanigans disabled from ThinkPad BIOS.

I never use biometric anywhere because in the end of the day it is just obfuscated clear text.

#biometric #infosec
0
0
1
@vbabka @ljs I might spam like a meth head but *definitely* not like a mm masterminds ;---------------------------------------------) Learning from the best.

[picture courtesy of kernel mastodon login page]
1
0
1
Obviously this has some security advantages too I guess. Given partitioning lot's of cross-VM side-channel scenarios are ruled out. I don't think you can e.g. use any Spectre derivative between two guests in Jailhouse (academic guess, I have not read its source code properly).
0
0
0
Edited 11 months ago
I don't like hypervisors where guest is like an extended entity of a process (/dev/kvm).

Or like I get that at a data center where you lease resources but actually the best possible scheme running VM's at home is partitioning.

Good example is creating a VM running Windows with optimal perf, which requires tons of knowledge about even things like huge pages and how to control them and stuff like that. All that complexity comes from melding the process be a bit like a partition of resources instead.

So when this came up, I thought that this is exciting exactly for home use:

https://github.com/siemens/jailhouse

I discovered it, if I recall correctly, OpenSource Summit or Plumbers when the band wagon was at LA 2018. Totally made sense for me because it kind of does of the shelf the best defaults for home use.

I wonder what happened to this project, is it in upstream or doing any progress? Not around a machine with my dev stuff to check this so thus not checking myself :-)
1
1
2
Show older