Jarkko Sakkinen
Posts
6410Following
352Followers
5701. https://github.com/puu-os/bootc
2. https://github.com/puu-os/installer
Jarkko Sakkinen
jarkko
Edited 2 days ago
I got one idea for dealing with AI kernel patches. It's not full all encompassing solution but would at least improve the situation.
If contributors were required to explain either in the cover letter or alternatively before diffstate (after "---") how they used AI assistance while writing the kernel patc, the barrier for any possible misconduct would grow.
This would also reduce mass spamming of LLM generated unchecked patches.
For maintainers this should be opt-in but recommended when possible.
#linux #kernel
If contributors were required to explain either in the cover letter or alternatively before diffstate (after "---") how they used AI assistance while writing the kernel patc, the barrier for any possible misconduct would grow.
This would also reduce mass spamming of LLM generated unchecked patches.
For maintainers this should be opt-in but recommended when possible.
#linux #kernel
0
0
1
Jarkko Sakkinen
jarkko
One more LLM rant (and I don't claim to have answers to any of these).
I'm not sure what value LWN contributor charts serve in the post-LLM world (except tested-by and reviewed-by).
I'm not sure what value LWN contributor charts serve in the post-LLM world (except tested-by and reviewed-by).
0
0
1
Jarkko Sakkinen
jarkko
Edited 3 days ago
I don't know for sure but I presume that there are already kernel contributors with no background and no understanding and they pass on whatever LLM finds to LKML.
I'm not sure how this phenomenon should be addressed. For me it would feel quite unfair if this is equal to well considered contribution, and it decreases the motivation for future talent to do such contributions as effort does not pay.
I'm not sure how this phenomenon should be addressed. For me it would feel quite unfair if this is equal to well considered contribution, and it decreases the motivation for future talent to do such contributions as effort does not pay.
0
0
1
Jarkko Sakkinen
jarkko
Edited 3 days ago
We need probably sooner or later better instructions for dealing with LLM assisted patches.
E.g., what to do if the fix itself is alright but commit messages have a continuous trend of being quite bad. I.e. you clearly can observe the patch being pure LLM dump but possibly still a legit fix for critical bug.
This quite defacto dilemma situation. At least for me. We need a more detailed guide on LLMs.
I don't this was the purpose of a patch being LLM assisted because there is zero involvement by the so called author. It's also a risk for kernel because author's own sanity check is our first layer of protection against malicious changes. I make mistakes too and it is now only up to me to explain the patch to myself and acking based on that, making me in a way both reviewer and the actual virtual author in a way.
#linux #kernel #ai #llm
E.g., what to do if the fix itself is alright but commit messages have a continuous trend of being quite bad. I.e. you clearly can observe the patch being pure LLM dump but possibly still a legit fix for critical bug.
This quite defacto dilemma situation. At least for me. We need a more detailed guide on LLMs.
I don't this was the purpose of a patch being LLM assisted because there is zero involvement by the so called author. It's also a risk for kernel because author's own sanity check is our first layer of protection against malicious changes. I make mistakes too and it is now only up to me to explain the patch to myself and acking based on that, making me in a way both reviewer and the actual virtual author in a way.
#linux #kernel #ai #llm
2
0
1
Jarkko Sakkinen
jarkko
Sent a PR to pi-background-tasks providing Landstrip integration: https://github.com/ismailsaleekh/pi-background-tasks/pull/41
0
0
0
Jarkko Sakkinen
jarkko
Edited 4 days ago
Decomps and recomps of games are just uninteresting. I don't feel anything. Nobody will ever try them.
It is so easy to buy games these days that nobody really pirates them anymore. And decomp/recomp slop is factors worse than even grabbing a pirated torrent. Laws aside, at least pirated torrent is honest stealing ;-)
It is so easy to buy games these days that nobody really pirates them anymore. And decomp/recomp slop is factors worse than even grabbing a pirated torrent. Laws aside, at least pirated torrent is honest stealing ;-)
0
0
1
Jarkko Sakkinen
jarkko
I need to play with landstrip from command-line and get an idea how improve ergonomics. I'd like to use it for some other applications too and not just coding agents.
At least I found a bug i.e.
$ npx @landstrip/landstrip --help
npm notice run npx
npm notice run 'landstrip' --help
OS-level sandbox runner
Usage: landstrip [OPTIONS] <COMMAND>
Commands:
run Run a program inside the sandbox
policy Inspect a policy without running a program
doctor Verify that the current OS sandbox is operational
help Print this message or the help of the given subcommand(s)
Options:
--debug Enable debug logs
-h, --help Print help
-V, --version Print version
But:
npx @landstrip/landstrip
npm notice run npx
npm notice run 'landstrip'
{"kind":"usage","code":"USAGE_ERROR","message":"OS-level sandbox runner\n\nUsage: landstrip [OPTIONS] <COMMAND>\n\nCommands:\n run Run a program inside the sandbox\n policy Inspect a policy without running a program\n doctor Verify that the current OS sandbox is operational\n help Print this message or the help of the given subcommand(s)\n\nOptions:\n --debug Enable debug logs\n -h, --help Print help\n -V, --version Print version\n"}
This is easy to fix tho and quite harmless.
At least I found a bug i.e.
$ npx @landstrip/landstrip --help
npm notice run npx
npm notice run 'landstrip' --help
OS-level sandbox runner
Usage: landstrip [OPTIONS] <COMMAND>
Commands:
run Run a program inside the sandbox
policy Inspect a policy without running a program
doctor Verify that the current OS sandbox is operational
help Print this message or the help of the given subcommand(s)
Options:
--debug Enable debug logs
-h, --help Print help
-V, --version Print version
But:
npx @landstrip/landstrip
npm notice run npx
npm notice run 'landstrip'
{"kind":"usage","code":"USAGE_ERROR","message":"OS-level sandbox runner\n\nUsage: landstrip [OPTIONS] <COMMAND>\n\nCommands:\n run Run a program inside the sandbox\n policy Inspect a policy without running a program\n doctor Verify that the current OS sandbox is operational\n help Print this message or the help of the given subcommand(s)\n\nOptions:\n --debug Enable debug logs\n -h, --help Print help\n -V, --version Print version\n"}
This is easy to fix tho and quite harmless.
0
0
0
Jarkko Sakkinen
jarkko
Edited 4 days ago
I think that in many IT jobs engineering managers should start to make sure that employees use ENOUGH time for tasks.
This type of HR policy would mitigate many issues related to coding agents.
E.g., if there would be a minimum required spent time for a work task, the extra time can be used for:
1. Doing manual alternatives.
2. Generating alternatives.
3. Cleaning up the slop i.e. from slop into AI assisted code.
This type of HR policy would mitigate many issues related to coding agents.
E.g., if there would be a minimum required spent time for a work task, the extra time can be used for:
1. Doing manual alternatives.
2. Generating alternatives.
3. Cleaning up the slop i.e. from slop into AI assisted code.
0
0
1
Jarkko Sakkinen
jarkko
When I've been talking about NVLink in the context of Puu OS I've actually meant NVIDIA ConnectX-7 and QSFP ports :-) They have no relation.
0
0
0
Jarkko Sakkinen
jarkko
Edited 5 days ago
I promised to test (in some post or comment) authenticated hibernate patch set if it gets resent so now I guess is time deliver that promise as @mjg59 provided a fresh version for us.
[1] https://lore.kernel.org/linux-integrity/20261008132532.1155166-1-matthewg@nvidia.com/
EDIT: s/encrypted/authenticated/
1
0
0
Jarkko Sakkinen
jarkko
"Making robots more like humans is infinitely better than viewing humans as robots." -Yann LeCun (today)
0
0
1
Jarkko Sakkinen
jarkko
Great I finally did it. I separated my inbox and kernel.org. I used subscript before. Should have done this ages ago :-)
I use kapsi.fi account for this as that inbox has no other use ATM.
I also scraped MEGA and Fastmail and moved my personal stuff all-in Google's death star
I use kapsi.fi account for this as that inbox has no other use ATM.
I also scraped MEGA and Fastmail and moved my personal stuff all-in Google's death star
1
0
0
Jarkko Sakkinen
jarkko
An alternative way to merge Github PRs:
for x in `seq 197 202`; do git fetch origin refs/pull/$x/head && (git cherry-pick -s FETCH_HEAD || git cherry-pick --abort) && gh pr close $x; done
Got fed up with noise that gh pr merge does and came up with this.
for x in `seq 197 202`; do git fetch origin refs/pull/$x/head && (git cherry-pick -s FETCH_HEAD || git cherry-pick --abort) && gh pr close $x; done
Got fed up with noise that gh pr merge does and came up with this.
0
0
0
Jarkko Sakkinen
jarkko
Microsoft has other software too than just Windows e.g., Office, VSCode and what not. They are all exposed to the same risk as Adobe products.
This is the price of not doing regulation early on.
This is the price of not doing regulation early on.
0
0
1
Jarkko Sakkinen
jarkko
Doing a few new OpenCode 2 updates for Landstrip. I only use Pi if anything so it requires an extra motivation to try the new OC :-)
0
0
0
Jarkko Sakkinen
jarkko
Puu OS is still way too broken for general consumption but it is on the right tracks as per architecture i.e., core choices enable maximum robustness to change anything over many "real" distributions. With bootc any distribution packaging gets only in the way and that friction has been eliminated from the get go (probably the most important decision).
0
0
0
Jarkko Sakkinen
jarkko
Edited 6 days ago
Having lately received quite a lot of AI assisted patches I really enjoyed this discussion with Serge about my RFC path set: https://lore.kernel.org/keyrings/20260924055521.1981957-1-jarkko@kernel.org/
No rush, good old times ;-)
#linux #kernel
No rush, good old times ;-)
#linux #kernel
0
0
2
Jarkko Sakkinen
jarkko
It's good that Adobe got hit as only when large corporations do, legistlation starts to move.
Who knows maybe we get vibe coded version of Windows 11 ;-) That would be morally right in many ways.
Who knows maybe we get vibe coded version of Windows 11 ;-) That would be morally right in many ways.
1
0
2