I don't know who needs to hear this, but: A Linux process can voluntarily restrict the files it (and any children it spawns) may access (and how) using the landlock(7) API.
Also, prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0) prevents a process from gaining privileges by executing a setuid program.
This seems like a great way to lock down worker processes that parse untrusted user input. If set up correctly, even if an attacker manages to take over, the attacker is limited in what they can do because they might not be allowed to write or execute any files, for example.
#linux #landlock
Remote
Oriel Jutty 
@barubary@infosec.exchange
Indoor European. Knows #regex. Writes #code (in #C or #Haskell or #Perl or #JavaScript or #bash). Not a fan of racism, sexism, transphobia, or bigotry. Almost done thinking.
Kompatibel mit handelsüblichen Klemmbausteinen.
242 Followers
68 Following
50 Posts
Joined November 06, 2022
Exit vim:
:q
Verified:
Pronouns:
he, his, him, him
Goose value:
71
OPSEC:
100%
Open post
Replying to
@flyingsaceur@ioc.exchange @rujasu@kind.social Now I'm wondering if the name is a reference to Gradius, which maybe was supposed to be Gladius (because Japanese)?
1
0
1
0
Open post
Replying to
@nuintari@mastodon.bsd.cafe The list part can be written fully left-to-right: ($bar->baz->@*)
Whether that's an improvement is in the eye of the beholder.
1
1
0
0
Open post
Replying to
@raktheundead@techhub.social @maybe_mbk@mastodon.social @Cregg@gamedev.lgbt @zzt@mas.to @gamingonlinux@mastodon.social That's the one, but I prefer to refer to "Knoll's law of media accuracy", which came first:
Everything you read in the newspapers is absolutely true except for the rare story of which you happen to have firsthand knowledge.
My issue with the "Gell-Mann amnesia effect" is that the term was coined in a speech by a climate denialist where he argued that you no one knows the future for certain, so all speculation is baseless and pointless, and the media is frequently wrong about all kinds of things. His examples included Y2K and global warning:
We need to start remembering that everybody who said that Y2K wasn't a real problem was either shouted down, or kept off the air. The same thing is true now of issues like species extinction and global warming.
3
1
0
0
Open post
Replying to
@troed@swecyb.com How do you "license" code which you did not write and of which you are not the copyright holder?
11
1
0
0
Open post
Replying to
@bruce@darkmoon.social ... digging deeper: Running PERLDOCDEBUG=2 perldoc strict (and then hitting q to exit less) on my machine shows a tmp file being written, followed by a call to /usr/bin/less < /tmp/XXXXXXXXXX. Is your /tmp directory OK and readable/writeable? Or is perldoc using a different directory on your machine?
3
2
0
0
Open post
Replying to
@mjd@mathstodon.xyz
asking it to solve a not-too-difficult but open problem of game theory, and to write a paper about it.
This it did
Facts not in evidence. It apparently produced a paper, but he doesn't say it solved the problem.
3
0
0
0
Open post
Replying to
@mjd@mathstodon.xyz @vnikolov@ieji.de Ah. Looks like "Lodz University of Technology" https://p.lodz.pl/en is the official English version of "Politechnika Łódzka" https://p.lodz.pl/
2
0
0
0
Open post
Replying to
@bruce@darkmoon.social Are PERLDOC and/or PERLDOC_PAGER set in your environment?
1
3
0
0
Open post
Replying to
@bruce@darkmoon.social Weird. I've never seen that one. Do perldoc -m strict and perldoc -lm strict produce the same error?
1
4
0
0
Open post
Replying to
@marocddot@hachyderm.io Not necessarily sans CPAN: https://metacpan.org/dist/App-FatPacker/view/bin/fatpack can bundle pure-perl dependencies into your script file for deployment.
1
1
0
0
Open post
Replying to
@isotopp@infosec.exchange Nein. Die Halluzinationen gelangen nicht in den Code; die Halluzinationen sind der Code. Das ist ja gerade das lustige an LLMs: 100% des Outputs sind Halluzination, weil das das einzige ist, was die Maschine produzieren kann.
1
0
0
0
Open post
Replying to
@isotopp@infosec.exchange
"Halluzinationen" – also Abweichungen zwischen dem, was man erwartet und was das Modell produziert
Das ist m.M.n. ein Definitionsfehler. "Halluzinationen" ist all das, was ein Sprachmodell ohne Zusammenhang zur Realität (oder zumindest einem Modell der Realität), quasi im Vakuum, produziert (fabuliert?). Das kann dann im Ergebnis zufällig richtig sein (also mit der Realität übereinstimmen), aber der Prozess war trotzdem halluzinatorisch.
1
2
0
0
Open post
Replying to
@mjd@mathstodon.xyz Probably a member of the same cult that decided Lisp should have declare, declaim, and proclaim statements, all different. Or =, eq, eql, equal, and equalp.
#lisp
1
1
0
0
Open post
Replying to
1
0
0
0
Open post
Replying to
@hmwilker@social.tchncs.de @feorag@wandering.shop Yes, assuming we're talking about the pastry: https://de.wikipedia.org/wiki/Schweinsohr_%28Geb%C3%A4ck%29
1
1
0
0
Open post
Replying to
@feorag@wandering.shop Not to be confused with https://de.wikipedia.org/wiki/Teetz_(Kyritz)
1
0
0
0
Open post
https://invent.kde.org/svin shows "This user is blocked" for me. Does this mean svin has been banned from that Gitlab instance?
#kde
0
0
0
0
Open post
Replying to
@Thayer@mastodon.social Finally. Bias automation at scale. Just what I've always wanted.
0
0
0
0
Open post
Replying to
@jernej__s@infosec.exchange @David_Kelly_SF@mstdn.ca Hah, that page messes with all sorts of input events: Key presses, mouse clicks, touch events, ... all because (according to the source code):
'You are not allowed to copy content or view source'
and
"Content is protected !!"
0
0
0
0
Open post
Replying to
@foolishowl@social.coop Sounds reasonable to me. For private stuff I generally use v5.36 at minimum, plus whatever newer features I find convenient. For things I publish I tend to be more conservative, but I don't go out of my way to stay compatible with a 10 year old version of perl. If things happen to work, great; if not, no big deal.
(Also, writing for anything before v5.20 is pointless unless you have some ancient systems you absolutely must support.)
0
0
0
0
Open post
Open post
@cpansec@fosstodon.org How do I get incorrect/misleading information in a CVE fixed?
Specifically, CVE-2026-13577: "[...] silently falls back to a built-in rand-derived session id when both Math::Random::ISAAC::XS and Crypt::URandom are unavailable" should be "silently falls back to a built-in rand-derived session id when either Math::Random::ISAAC::XS or Crypt::URandom are unavailable."
Or perhaps better: "silently falls back to a built-in rand-derived session id unless both Math::Random::ISAAC::XS and Crypt::URandom are available."
0
0
0
0
Open post
Open post
Replying to
@PorkBBQ@dice.camp @nuintari@mastodon.bsd.cafe Postfix dereferencing first appeared in perl 5.20, released in May 2014. It's more than 12 years old at this point.
(It officially left "experimental" status in perl 5.24, released in May 2016. Still more than a decade.)
0
0
0
0
Open post
Replying to
@simontatham@hachyderm.io I've had similar thoughts. I think recursive algorithms come most naturally when you have a recursive problem/data structure, like a tree.
So I think good introductory examples would be recursively searching a directory for all .png files or eval'ing an expression tree in a calculator.
0
0
0
0
Open post
RE: https://infosec.exchange/@wdormann/116965837513850094
TIL installing SharePoint updates does not update SharePoint.
Open quoted post
Open quoted post
Quoting
Looks like somebody [posted an exploit for CVE-2026-50522] (patched in July)(https://gist.githubusercontent.com/testanull/0868e02d81d57d6c59a91261969f7f81/raw/4d16304047e525057d732401f718e20fa830eb0a/SharePoint%2520SE%2520p2o%2520PoC.ps1)
While I can reproduce it on SharePoint Subscription Edition, I didn't get it to work on SharePoint 2016 or 2019. But I could also attribute that to me having no clue how to properly install SharePoint.
Here it is tweaked to run calc.exe as the SharePoint user (sp_farm in my case), as that's a touch more noticeable than it making a web request.
And remember, kids, just installing SharePoint update bits does not update SharePoint. You have to run SharePoint Products Configuration Wizard after installing the updates, which luckily every SharePoint admin knows.

0
0
0
0
Open post
Replying to
@pozorvlak@mathstodon.xyz Yeah, sure. It's not the climate collapse, not the rise of fascism, not the threat of worldwide war, no: The really scary thing is the statistical word completion machine (that someone hooked up to a shell and gave an internet connection).
0
0
0
0
Open post
RE: https://mathstodon.xyz/@pozorvlak/116968236641276698
Somehow this commentary is very funny to me. (And I think it absolutely was a marketing stunt.)
0
1
0
0
Open post
Replying to
@mjd@mathstodon.xyz "What I’ve checked of it so far checks out." I understand this to mean he did some spot checks, but didn't verify the results himself.
See also his reply to you: "at least historically before AI was as good as it is today and I'd want to do a bit more work to ensure all the results are correct -- but I did check some of them at least".
0
0
0
0
Open post
Open post
Replying to
@ChristineMalec@mstdn.ca Out of interest, does your screen reader stumble over things like emoji, quotation marks, and all-caps words in alt text?
0
1
0
0
Open post
RE: https://spectre.asta.lgbt/@aud/statuses/01M218903KJRMAKC42CV0Y5WYE
Huh. I think I've changed. I may actually care more about people now. (Also, so far I've successfully managed to avoid being in any sort of leadership position.)
0
0
0
0
Open post
0
0
0
0
Open post
On the command line, sometimes I want to tell a tool (e.g. grep) to look in all directories except for one or two. This can be done with shell wildcards/globs.
All entries in the current directory:
grep -r 'honk honk' *
Subdirectories only:
grep -r 'honk honk' */
All subdirectories except for _build:
grep -r 'honk honk' !(_build)/
All subdirectories except for _build and cache:
grep -r 'honk honk' !(_build|cache)/
#bash
0
0
0
0
Open post
Open post
Open post
Replying to
@rrwo@infosec.exchange Is this exploitable if the base string ends with a slash, like rewrite => 'https://example.com/'
0
1
0
0
Open post
I simply wouldn't get addicted to cocaine. I'm too intelligent.
0
0
0
0
Open post
Replying to
@offseq@infosec.exchange Unlike what this website claims, you need to install both Math::Random::ISAAC::XS and Crypt::URandom to not be vulnerable.
0
0
0
0
Open post
Replying to
@mjd@mathstodon.xyz Reminds me of https://web.archive.org/web/20041108121656/http://www.research.ibm.com/people/d/dfb/papers/Bacon01Concurrent.pdf which contains a few proofs of correctness (of the concurrent version of their algorithm), but their basic single-threaded algorithm has a use-after-free bug.
0
0
0
0
