Hacker Newsnew | past | comments | ask | show | jobs | submit | harry8's commentslogin

Bold move to make the argument that Texas doesn't matter...

>... convenient same-day shipping

...replacing a significant part of the economy that paid its taxes with amazon who don't leading to budget cuts and so on.

That's the argument. Agree, disagree, don't engage, whatever, but what you've written here is

squints

...nonsense.

Sometimes people try to keep the forum sensible, according to the house rules, maybe its a losing battle already lost.


Amazon workers don’t pay income taxes?

The vast majority of Federal income comes from income taxes, mostly from high earners.

Also: https://fred.stlouisfed.org/series/FCTAX


Amazon does pay taxes though. Quite a lot, in fact.

Does every large company have a big server with some h100s and a couple of open weight models for just such an occasion. Especially if you don’t have the luxury of being able to wait or postpone eg under cyber attack?

Only until NVIDIA starts cryptography verification and block anything not from US closed models, which should be anytime soon

I hope that doesn’t happen. There is a good alternative: if MOE models work for you, then Apple Silicon does a decent job at inference using whatever harnesses you like.

Just get an OpenRouter or Vercel account. Your option aren't pay Anthropic or buy hardware yourself, there are hundreds of providers out there.

Claude has us locked in right now while they subsidize cheap usage to get us subscribed and dependent on their monthly plans.

It depends on what "us" means. I am switched off of Claude almost entirely now except for background things where I want to abuse my subsidized account. Otherwise, OpenRouter + open weight models are my daily drivers. It takes a bunch of work but I'm doing it for roughly a max20 account cost, a bit more, but it's not like the difference is huge. They are less capable but I've modified my workflow to mitigate that as much as possible, and I can still do things when Anthropic goes down.

What models are you using? I was satisfied with deepseek v4.1 while ai was still using my brain, before being time-pressured into shipping something I don't understand and now I feel I need stronger models to steer it. Kimi K3 or the closed models through openrouter is too expensive compared to subsidized plans though so I'm using a $20 sub for light reviews/consultations.

As temporary backup, one can use claude models hosted on GCP - my understanding is that those don't depend on platform.claude.com and should be available even if main claude platform is dead

There is an absolute tonne of competition.

No you missed the point.

This kind of morality is between you and your Rabbi. We are allowed to look down on it all we like. It has no relevance to law or keeping your job because if you allow this kind of thing to have relevance it never ends. Rank selfishness comes out in so many ways.

"You got a parking fine, your social conscience is clearly lacking..."


It does end, because people lose their jobs for office affairs all the time and rarely if ever for parking tickets. There’s no moral inconsistency there, not all bad behavior is equivalent.

number of people who lose their jobs for infidelity divided by the number of people who do it.

Hard to estimate it precisely but less than 1% is likely to be accurate, no?


What percentage of people who get parking tickets do you think lose their jobs as a result?

Not many, because we have principles of that kind of morality not being applicable. Double parking beside fire hydrants isn't universally celebrated though, is it? I look down on that kind of thing, you might be the same yourself.

If someone is willing to break their most sacred vows to their loved ones, why should I, a relative stranger, think they will keep their word to me?

You are absolutely entitled to that opinion.

You do not have to judge someone for a parking fine just because you judge someone for cheating on their wife.

> Him being married is relevant: he opened himself to extortion/blackmail.

Do you really want to apply that to congress, the judiciary, the police. Let's face it anyone in any position of trust, which turns out to be everyone in the end because homeless, unemployed I could be a witness to something.

So we just outlawed private consensual behavior of individuals. (Not outlawed, you just lose your right to keep your job or whatever, yeah sure right...)

Doesn't seem like an american Freedom to me but what would I know?


Yes I especially want that to apply to congress, judiciary, and police, even more so than corporate CEOs! People in positions of higher responsibility, should, you know... be more responsible. Police and people in power should break laws less often, and be more trustworthy, and less selfish than average. They are in a position that requires a higher standard.

Handmaiden's tale.

This reminds me of when people complain about a lack “freedom of speech” but what they actually want is freedom from the consequences of their speech.

Plenty of jobs come with personal life restrictions. Work nights. Work on an oil rig. Work on an aircraft carrier in the ocean for months at a time. Everyone is free to not take those jobs. I don’t see why a “no affairs” clause for a CEO should be all that different. If you don’t consent to the rules, take a different job.


> Do you really want to apply that to congress, the judiciary, the police. Let's face it anyone in any position of trust, which turns out to be everyone in the end because homeless, unemployed I could be a witness to something.

These kinds of paths to blackmail/extortion are literally what government background checks pressure test when hiring employees. So, yes, we do want to apply those to people in positions of trust.

I'm not clear why you claim this extends to everyone including the homeless and unemployed, that seems wrong to me. You can't blackmail a homeless person into violating national security.


But if you allow them privilege denied the great and the good what message does that send "to the children" ?

Are you really claiming the homeless don't matter?

Even if you can do all these things then you must deny them the ability to testify in court because they could be blackmailed.

If you don't see how this goes I probably can't help you. Read a diversity of sci-fi? Read what life was like in the soviet union?


Not the person you replied to, but

> Are you really claiming the homeless don't matter?

Towards being blackmailed against national security interests? Yeah I’d say the risk of that is low. The GP didn’t claim anything else


Jensen is not as far as any of us are aware, a spy.

Perhaps trying to make sure there aren't morality based repercussions takes the sting out of future blackmail attempts on other successful businessfolk would be a better strategy for national security than, I dunno, briefing the New York Times and trying to make it as miserable as possible?

But sure maybe the CIA is just woefully incompetent, disorganised and accountability free in pursuing this national security goal of blackmail resistance. Or perhaps this is not the best theory as to what is going on.

There's a really, really good reason we keep this kind of moral judgement private, "between you and your rabbi" as Feynmann put it.


dude you better believe if you have a clearance and they find out you have an affair you're losing your clearance

exceptions of course for the commander in chief


Yeah, right, the /only/ exception.

Jensen's job does not rely on having security clearance. There's no evidence he has security clearance is there?


I don't get what you're meaning. I don't think he had a clearance. You mentioned a bunch of government people including some people with clearances and claimed it was legislating private life. I pointed to an example where this is legislated already and most people consider it normal and reasonable.

Has anyone stood up the claim she's a spy yet?

Or is the evidence right over there next to the evidence of WMD?

I know some people believe anonymous govt officials but I kinda don't so if there's actual evidence, let's see it, or note the the doubt legitimately exists.

It seems a very odd thing to be briefing newspapers about rather than the very many other, seemingly more useful ways of using that information if it is true. How does everyone feel about the CIA briefing the New York Times in general to seemingly get a CEO removed? Who takes responsibility for making that decision?


>Remember when you were able to take over any Instagram account by asking the helpful support AI to send a password reset to an email address of your choice?

Wow, first I've heard of that. Damages must have been immense for that level of incompetence occasioning financial harm. Or do those laws only apply if you can't hire very expensive lawyers to blame a computer?


So I tried this.

Qwen-3.8-27B using qwen-code via ollama on a desktop machine with 64G ram and a GeForce RTX 3060 with 12G vram.

I used the owntone-server codebase downloaded from github.

/init

It is unbelievably slow. Ok, sure.

It has hallucinated a macro "DLOG" from its read_file output(!!!) and so tied itself in knots over it because it can't find the macro that is wholly absent from the codebase. Hours later its still going and shows no prospect of writing anything to QWEN.md

How the hell are my results so very different to what you're claiming? What have I done wrong here?

I haven't got anywhere near planning anything with it, much less have it implement something, at any quality. I mean it's just so bad.

Edit: Asked claude which suggested a Modelfile

   FROM qwen3.8:27b
   PARAMETER num_ctx 16384
   PARAMETER num_thread 12
ollama create qwen38-fast -f Modelfile

I'll give that a go. I wonder if you guys just have a bunch of H100s or something like this.


Update: failed abysmally.

    x[API Error: 500 no user query found in messages] (Press Ctrl+Y to retry) 
Which means not enough context using PARAMETER num_ctx 32768

I wonder what hardware you need to use Qwen-3.8-27b locally at all..


So dumb question maybe:

Why tf do we have options in the header at all if you ca drop everything that does, be secure and everything still works?


Fair question. IP options in the header are generally not used due to performance degredation in parsing extended headers in hardware based packet filters. So they are part of the spec but generally not used due to this. I have worked in CDNs and large internet facing networks for over 20 years and we implicitly drop any IP headers over 20 bytes (meaning they have options set).

There are a few parts of TCP/IP that are basically vestigial on the modern internet.

Follow up question, maybe even dumber:

Why the actual F%$K haven't they been deprecated and packets using them dropped by literally every tcp/ip stack if their only function is to get you f%$king pwned?

That might be a very dumb question for all I know. Maybe i shouldn't be cranky. It all seems like there are some very blase attitudes to this kind of thing, that should not even exist anymore, creating massive security holes, with zero upside tradeoff.

The seeming asymmetry offends me.

On the one hand it increases the attack surface and so the security risk. On the other, nothing. Yeah, let's take /that/ tradeoff on behalf of the entire internet.


Because TCP/IP is used for more than just the internet. For pretty much all of these weird features someone is using them somewhere on their own networks. (And these things are frequently blocked on the internet or on firewalls).

> Because TCP/IP is used for more than just the internet.

So I'm so stupid & ignorant of security I think I would define "Internet tcp/ip" which conformant stacks block this rubbish by default like they do any other non-conforming packet. Thus reducing the attack surface essentially with no tradeoff. Weird uses can use customised lwip or whatever for these uses where we don't know the who, the what or the why (or perhaps even the IF).

Seems like nobody is shipping these firewall rules by default on internet routers. We get the vulnerability and as a tradeoff we get nothing of any value to us. Who do we thank for that?

What is the rationale, the justification for not paring away the garbage that does nothing but create vulnerabilities?


The internet is a loose connection of people and organisations that is very difficult to coordinate. A lot of people do block these packets, which is why they're basically vestigial, but it's not something that's really worth spending co-ordination effort on when it's something that's easily implementable by those who want to and there still isn't universal support for things like standards that prevent BGP hijacking.

Why? See the story we’re commenting on maybe?

Obvious and sensible engineering at a protocol level seems very worthwhile to me.


I agree new releases of TCP/IP stacks could ship with IP options disabled by default, and allow the system owner to turn on the ones they really need.

Question:

Every consumer router-wirelessAP I've seen runs linux.

I never see security upgrades for the kernels they use. Not one. Not ever.

Is this something that should be a cause for concern? Even if you don't run a consumer-grade tplink, netgear, asus, whatever router yourself.

Update: The context which I should have said is that I think LLMs are giving script kids nation state level tools and expertise. This known bug is always less interesting than the next twenty.


  Every consumer router-wirelessAP I've seen runs linux.
  I never see security upgrades for the kernels they use. Not one. Not ever.
Turris [1] has been meticulously updating their routers for years.

[1] https://www.turris.com


Good to know and to be introduced to Turris.

10G looks a little less consumer-y. I'd imagine cisco, arista, etc do security updates if you want to go down that path.


Turris routers are still consumer products. SOHO, to be precise. ;)

AFAIK, none of them supports WPA Enterprise, which I'd consider necessary for enterprise-grade equipment.

OTOH, they are Open Source and Open Hardware, are still pretty powerful, and are much cheaper than enterprise HW.

PS. The proper link to their site in English is https://www.turris.com/en/ .


500+ Euros, 10G, Linux containers, that just is not targeted at consumers. My guess the number of people who have one of these in their home is very few indeed.

We're talking about the 500 million? A billion? more? Households with a consumer internet wirelessAP/Router with a public IP. Turris may be wonderful and i wish you the best of luck but in its present form it appears to be no part of the solution, sadly.


Enthusiasts and self-hosters, or just demanding users, are consumers too.

There are different kinds of consumer products. Some are simplistic and some have lots of features. Some are cheap and some are not. Some are slow and some are fast. Some are mass-produced and some are not.

For instance, there are Asus ROG routers with 25G, Samba, network drive, WireGuard etc. for over €700 and they're still consumer products. And I'm pretty sure those are not used in that many households either.

All of the Turris' product pages mention home usage. They also mention small business and office, which is expected for SOHO products.

To be frank, until today I thought that SOHO is a category of consumer products, but apparently it is not, so it looks like they target both.

I agree that this kind of devices is out of reach for many people, but it was not the topic until now.

I'm not affiliated with Turris in any way, so no need to wish me luck on this matter, but thanks anyway!


Mosty probably: not. Traffic passes throught the ASIC, usually it is not processed on CPU or custom code. Unlike typical web servers, that reduces the attack surface as there aren't software parsers (a common place where vulnerabilities hide). Additionally the OS may have additional protection measures, such as signing each binary, signing the booting image and limiting which syscalls can be done by any binary.


> Additionally the OS may have additional protection measures, such as signing each binary, signing the booting image and limiting which syscalls can be done by any binary.

I'd love for that to be true. Unfortunately, a lot of consumer routers are fairly straightforward Linux systems with no particular attempts at hardening.


but he has a point that if the ASIC is doing the TCP header offload, wouldn't this particular CVE not be applicable?

I'm not sure. Maybe it would; maybe the IP options would make the packet bypass offload and get processed by the kernel. Either way, I wouldn't want to rely on hardware NAT offloading as a security feature.

The bug in question is not mitigated by any userspace actions (like signing binaries or boot images) or even by limiting syscalls, and all it takes is a listening socket on the device from either direction (inside or outside the network) to get root since router-destined packets are always processed by the CPU.


All traffic? In particular, I'd expect new connection establishment to go through a software path, looking up port forwarding and firewall rules etc. and creating hardware forwarding table entries if applicable.


This is one of the biggest benefits of using OpenWRT. They update the entire stack like every other month or so you’re never on an ancient kernel or utils.


The downside is it warns me if I update things will probably break and to take a backup, I've always been wary of updating my openwrt router because of that.


No kidding make a backup. The warning is sensible.

I haven’t had problems upgrading in at least 4 years. There’s a screen where you click a button to auto-upgrade.

Just get it installed.


I checked yesterday. Updates were not ready yet. I took the chance to upgrade from 21 to 25 anyway.


It's a bit of bad luck tbh; they just made a point release a few days prior to the bug being announced, with kernel .112


very few WIFI 7 SOCs support openwrt

This is why I use a fanless PC as my internal router. I have complete control and get timely kernel updates from a responsible distro.


Ubiquity routers and APs do. Also schedulable backups.

Granted they are pro-sumer devices, not strictly consumer.


According to its logs, my bargain basement Huawei 4G router from 2019 got an update last week.

I hope that most of these devices actually do update themselves automatically, and people just aren't noticing. (Whether the updates are any good is another matter!)


It got an update, but from whom?


My UDM-Pro had an update available yesterday. I don't know if it's related to the CVE or the shit eMudhra CA.


This is a good point, thanks. We know one does which beats none.


You may want to try to look at Gl.inet routers. I'm not sure yet what is their modus operandi, but at least Flint 2 has both their official firmware (with Mediatek's SDK, I guess) and a vanilla Openwrt version -- on their company page.

Their features (Glinet's mesh) are not compatible but it does seem better than just drop support totally.

There also seems to be some involvement in the upbringing of openwrt support -- so it's not a hands-off approach, but on the other hand I'm only looking at Flint 2 firmware/OS development.

Additionally, their Flint 3e uses Qualcomm so the route it takes may be different -- though I hope not!!!

Now, if we are talking about my ISP router/modem....... I'd rather not think about it.

PS: Gl.inet also has a router with fiber optic. Not sure how that would work if I wanted to throw my ISP's router/modem out the window.


Google Wifi used to be good about that. I even got a notification for when my AP was being EOL'd and would no longer receive automatic firmware updates. I don't know if Google is still on the ball there. The Nest acquisition was confusing. Have the Google Wifi and Nest Wifi teams been merged?

But FWIW I run my AP in bridge mode. Wifi drivers and chip firmware seem to be riddled with exploits. I don't want my router & firewall on the same system as the AP.


eero updates its APs OTA, they run linux. Even stuff sold > 5 years ago is still receiving updates.

full disclosure: they're my employer


Eero needs to make sure that they're the only ones exfiltrating data from their products. :^)


While it's certainly not the right product for the openwrt crowd wanting to flash their own firmware and look under the covers, I wouldn't work here if we were obviously doing such things to the degree your comment implies.

My home router is managed by my ISP. They can, and do, remotely upgrade routers in the middle of the night.


Is it a cause of concern? Yes. Will you get hacked? Most likely not.

Most consumer routers run ancient stacks with all kinds of weird proprietary vendor patches that are tied to a specific kernel version. Looking at you Qualcomm!

However they generally also drop all traffic by default so from the WAN side they look like a black hole. I guess that's "safe" enough for most consumers.

For fun, take a look at the open PRs in the OpenWRT repos. Those people are doing some amazing work but it's also frustrating to see them struggle with things that _could_ be easy if the vendor made some slightly better choices.

I would strongly prefer something like OpenWRT though. Call me paranoid but after the last few weeks I'm increasingly less comfortable with connecting my devices to a blackbox.


I know the context here is consumer routers, but for those reading who might be running a server, or a homelab, or anything that does accept connections from outside, I want to make it clear that "most likely not" turns into "likely trivial if targeted" for servers.


Can you give examples of kernel bugs that can be targeted on an otherwise locked down system? Let's say something running just LAMP and all other ports closed off with a software firewall. Even the one in this article seems very tricky to actually exploit.

There are plenty of out of support systems still in the wild and while I'm not arguing they are completely secure I was under the impression that kernel vulnerabilities generally make a minor exploit in something else worse (chained priv escalation for example) and generally are difficult to exploit on their own. If I'm mistaken I'd love to know


Any TCP packet (with IP options) you can send to a Linux kernel that doesn't get DROP'd by netfilter makes you technically vulnerable.

That includes the one port your LAMP stack uses.

Or, the attacker might be originating from the other side of a device, like within a LAN. Malicious employees/house guests/what-have-you could target devices from inside your network where most devices (not just servers) don't have as much protection and often have other open ports accessible, like file sharing/SMB, SSH, SMTP, RDP, databases etc.


Does the bug we're talking about not count as an example for some reason?


The comment addressed this

Don't a lot of consumer routers allow for remote connection for tech help trouble-shooting?

How about forwarding a port for wireguard or ssh?


Those are usually not enabled by default.

AFAIK all the backdoors which have been found in some models are only accessible from the LAN side, and if you have an attacker inside the network, it's already too late.


> if you have an attacker inside the network, it's already too late.

It shouldn't be though. Just because my smart lightbulb has an RCE and is isolated on its own VLAN with the router doesn't mean it should be able to attack the router and escalate to access my devices on other VLANs. It should only be able to do damage limited within its own vlan.


> if you have an attacker inside the network, it's already too late.

Disagree, I think it's still a good idea to try to mitigate e.g. root escalations via this bug from devices within your network as well, as they could become compromised at any moment via other means and then promote themselves to root.


It's a massive risk, and only a matter of time. Especially for hardware that's no longer updated firmware wise.

I'm thinking of what to do myself.

The ghost of android releasing 6 month old patches every month is going to come home to roost.


I have a MikroTik router at home, and they have already patched this [1]

[1] https://mikrohub.io/mikrotik/changelog/long-term/7.24.6


A lot of them run BSD, not linux, which wasn't affected


BSD never ever has security issues so kernel updates are never necessary?

That would be a bold claim and one I've never heard before. Hard not to be suspicious of it.

This security flaw is unlikely to be the only security flaw that ever happens to a kernel used by routers.

The point: Consumer routers do not seem to get updates. Ever.

Is that a worry or what?


Google tells me almost none of them run BSD and nothing I've ever heard of, much less seen.

But what do I know about it beyond that?


BSD is very popular for open router ecosystems, like pfSense, opnSense.

Some more commercial ones (pfsense is commercial) use it too like juniper junos.


We’re talking about consumer router/wireless APs here.


Any wireless AP is going to run Linux because BSD has had a tough time with anything better than 802.11n

Any router/switch is going to be running an ancient Linux kernel, which isn't patched. EdgeOS for example, won't be patched. The old Brocade/Ruckus switches I got here, won't have this patched. I'm holding my breath for my Unifi APs.


Last consumer router that ran any kind of BSD based firmware was probably Apple Airport and that was a long time ago.


could be intentional, even


> BSD never ever has security issues so kernel updates are never necessary?

Of course it does. But it was not affected in this particular case.


Same for Flock cameras, running a 2017 kernel.


i admit they don't enumerate their patches well, but the major residential router vendors do regular updates , usually indicated with "security fixes" .

unfortunately most lack SSH access or a real ability to test the kernel patch version.


Nah I wouldn't worry about I'm sure they're fine


Good thing everything on the lan side is fully patched all the time. God bless those vigilant smart device manufacturers, especially the tv guys


most of the routers are patched by ISPs directly. also: they are usually not x86 architecture - so not all exploits are possible.


Never seen an ISP patch a router. Ever.

MIPS and Arm have no security issues ever so never need patching which is an x86 only thing? Or maybe this comment is no more relevant than pointing out not all Arm patches apply to x86?

LLMs are giving script kids nation state level tools and expertise.


I'm on Comcast in USA, I BYOD for the router and it's a Netgear C3500 (I think). It gets new firmware pushed to it from Comcast. Can't even load your own manually.


BYOD = bring your own device? And yet your isp pushes updates to it?

Defintely seems potentially better than no updates at all.

How do you know if you have a wide open CVE they're ignoring, like it's claimed ubuntu are doing?


> BYOD = bring your own device? And yet your isp pushes updates to it?

Perhaps better characterized as "buy your own device". Cable ISPs will only support a limited set of modems on their network, and they're fully managed by the ISP, both in terms of software and configuration.


Correct, and they are limited in features. I'm going to swap out this combo-modem+WiFi and get a dedicated -modem and then have a different device as my "edge" router/NAT


So comcast in the USA do it. Unclear how well. Assuming very well, what portion of the consumer market's home connections are covered? In the usa? In the world? This is the general issue beyond today's unpatched bug.

I've had probably at least about 5 routers supplied by 5 different ISPs over the past couple of decades and not one of them has ever got a security update. Is that an usual experience?


even more concerning would be the ones who run closed source os but never update.

at least with an open source os there are more paths forward.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: