Could EU force hardware manufacturers to make working drivers for Linux?

https://old.reddit.com/r/linux/comments/16wv53f/could_the_eu_force_hardware_manufacturers_to_make/

by opengears • 3 years ago

197 356 3 years ago

old.reddit.com

jandrese4• 3 years ago

More likely they would require companies to publish specs suitable for driver authors. I’ve always thought that companies treating their register configuration and I/O offsets as a trade secret is ridiculous.

smoldesu19• 3 years ago

This sounds much more reasonable to me. Banning unnecessary hardware DRM and requiring specs to get published would be a much more effective way to stimulate the market. Forcing manufacturers to publish Linux drivers would probably just end up in low-effort support on either side. The last person I want writing my Linux drivers is someone who doesn't care.

jandrese4• 3 years ago

Yeah, you know it would end up with the manufacturers just releasing a binary driver for exactly one point release of the kernel and calling that box checked. Might even just be some NDIS thing where they wrap the Windows driver to get it mostly working.

msm_• 3 years ago

This. Linked post is a misconception of how EU actually work. They will never "hardcode" something like "linux drivers" into the law. But, in theory, they could write a law like "If you want to sell your device in the EU, you must publish a freely available technical spec with enough level of detail to write a functioning driver".

adrian_b• 3 years ago

This would not even be something completely new.

It would just revert the market for computer peripherals and interfaces to how it was before the launch of MS Windows 95.

In the MS-DOS days everything was provided with detailed technical information, so you could write your own device drivers, for any operating system.

This ended in 1995, after which most hardware vendors stopped providing documentation and replaced it with MS Windows device drivers.

bubblethink• 3 years ago

And the ability to use such a driver. Hell, even the spec wouldn't be such a strong requirement as people can reverse them. The issue is usually the firmware and signature checks.

bpye• 3 years ago

This would be my expectation too. Though I’d think you’d need to go one further - hardware manuals AND a license to redistribute any necessary firmware binaries - the latter has been an issue with Nouveau for example.

In fact I think manuals are probably better in a lot of cases, OEM drivers are routinely terrible.

suprjami• 3 years ago

Intel publish datasheets which describe their network interface cards to this level. It's fantastic, literally anybody can write a driver for an Intel NIC for any OS. You can also debug existing drivers when you get strange hardware issues, which is part of my job. No other networking vendor does this.

gjsman-10002• 3 years ago

Drivers… I don’t know if I could sign on for that for the following reasons:

1. The burden it would put on smaller manufacturers and companies (every little regulation adds up), especially difficult considering the comparative lack of qualified developers

2. Drivers != Quality, Upstreamable Drivers. Making the judgements of Linux maintainers legally binding is a bad idea. Case in point: Apple has Linux drivers internally, they’re just not complete or upstreamable. You would be forced to fight the Linux maintainers (and the power trips they already have!) in order to legally sell your product.

3. If we can force companies to support Linux, why not force all websites to support Firefox? Why not force all desktop programs like Adobe to support Linux? Etc…

4. What about poor FreeBSD? Serenity OS? That guy who still loves OS/2?

ohdannyboy• 3 years ago

Don't forget TempleOS, lest you be religiously intolerant.

yjftsjthsd-h8• 3 years ago

3. If we can force companies to support Linux, why not force all websites to support Firefox? Why not force all desktop programs like Adobe to support Linux? Etc…

...Although I agree that there are questions about required effort and breadth of support and burden to smaller companies, every one of those sound like wonderful outcomes to me.

exabrial• 3 years ago

Please no.

The idea is rooted in idealism, but absolute crap will be produced in reality.

WesolyKubeczek• 3 years ago

Look at the realtek drivers made by realtek. Reconsider.

doikor• 3 years ago

Why should EU give preference to Linux over any other operating system?

A much more EU "style" directive would be to force the companies to release enough of the specs so anyone can write the driver if they want to.

mnd999• 3 years ago

A much more EU style directive would be to force all operating systems to use the Linux 5.1 kernel.

diego_sandoval• 3 years ago

A much more EU style directive would be to force all operating systems to ask on every boot: Do you consent for this OS to store files on your hard drive?

amelius3• 3 years ago

That's also fine.

loup-vaillant24• 3 years ago

They don't necessarily have to: there's another, more comprehensive hammer they (or the US for that matter) could use: disallow vertical integration.

Force hardware vendors to only sell hardware. If any software that can be changed remotely or by the user is off limits. Even firmware. Conversely, software vendors can only sell (or freely distribute) software. For vendors from abroad, don't go extra-territorial, just force them to chose: either they only sell hardware in the EU, or they only sell software. Intel and microsoft would have no problem. NVDIA might complain very loudly. Apple would likely have to split itself.

Now the complicated part is how to define a company. We don't want a single company to just split itself into 2 legal entities that work so closely together they might as well be the same company.

Do that, and you'll get much better than device drivers for one free OS. You'll get the necessary specs required to make it work on all OSes. Even better, the user-facing hardware interface will start to matter, and there will be some selection pressure to drive the more complex ones, or the non-standard ones, out of the market. (Won't be ideal, I can see a particular over-complex architecture win out, similar to x86, but at least there won't be that many left, so writing a driver for most devices will actually be possible).

gjsman-10002• 3 years ago

Sounds great - there’s just one problem: The desktop applications will still almost entirely be written for Windows. Or MacOS. Or Android.

So yay, you can buy a device… and 99% of people will install the same OS on it that it would have originally come with anyway. What’s the point here?

You’d have to say desktop apps must support all competing operating systems. But, hah, good luck with that.

loup-vaillant24• 3 years ago

The hope is that hardware interfaces harmonise to the point where writing a new OS becomes actually possible. Right now it's not, there simply is too much diversity, and that's before we talk about all the quirks in hardware APIs vendors currently hide under the software rug. Then maybe, just maybe, we could start some real competition on the OS space.

Even if that doesn't happen I would be happy with the end of vendor lock-in.

askonomm9• 3 years ago

Well that world would suck immensely, with horrible integration between hardware and software like exists right now outside of the Apple world. Case in point: can you copy text on your phone, and paste in your computer (and vice versa), seamlessly as if it were the same device, without any configuration, hacks or programs needed? Things like that are amazing and I'd never give up. And I'm sure most people wouldn't want to give this up either just because Linux fanboys want to.

smoldesu19• 3 years ago

Case in point: can you copy text on your phone, and paste in your computer (and vice versa), seamlessly as if it were the same device, without any configuration, hacks or programs needed?

Circumstantially, yes. I am a KDE user though, on GNOME you do have to install a little app for it.

oynqr• 3 years ago

With those preconditons, it doesn't even work on Apple devices.

ta8903• 3 years ago

KDEConnect lets you sync your phone keyboard, even with Windows. And Apple could let you sync your clipboard to non-iOS devices too, they just block arbitrary features unless you buy into their entire ecosystem.

tincholio• 3 years ago

Not only copy/paste, but have notifications shared both ways, use your computer as a phone laptop, use the phone as a trackpad/mouse, and much more. It just works.

monocasa• 3 years ago

Why would this be an issue? Under the proposed model, the vendor for phone software can be the same vendor for desktop software.

mschuster9111• 3 years ago

Well that world would suck immensely, with horrible integration between hardware and software like exists right now outside of the Apple world.

Why? The hardware people at companies like NVIDIA or Apple already have to write detailed documentation for their stuff anyway so that the software people can do their side.

Assuming they do a good enough job at writing documentation and testing their hardware, it should not be a problem for Linux developers to write appropriate drivers. Hell there are a very few select people able to write high quality drivers even with zero documentation (see Asahi Linux).

The only ones in trouble would be the vendors of crap SoCs who ship extremely buggy hardware and make up for the lack of QA in software quirks instead.

gray_-_wolf• 3 years ago

Well that world would suck immensely

Maybe.

Case in point

But this does not really support your statement no? What you described seems to me SW only functionality no? And if in requires any special HW (BT? I have no idea how it works under the hood), why does it need tight integration? I do not see a technical reason why Android and Windows could not provide the same feature out of the box, despite neither the Google nor Microsoft making the HW.

delta_p_delta_x2• 3 years ago

Case in point: can you copy text on your phone, and paste in your computer (and vice versa), seamlessly as if it were the same device, without any configuration, hacks or programs needed?

I am guessing that AirDrop, Sidecar, etc have absolutely nothing to do with hardware-software integration, and have ≥ Layer 3 implementations. KDE Connect and Microsoft My Phone are competitors.

lnxg33k12• 3 years ago

Really your example serves as disproving your point, without the hardware walled garden there could be nothing to prevent apple to distribute the software to copy-paste from-to multiple devices

the_third_wave3• 3 years ago

Yes, that works if I want it to, e.g by running KDE Connect (which works outside of KDE). Works on Linux, Android, Windows and even on Apple-things in case you need to escape the walled garden...

jksk61• 3 years ago

Literally never needed to copy text from smartphone to pc (or viceversa). At best anyone needs to sync between photos, docs or stuff like that. For that you just need a sync on the cloud like Dropbox, Drive, Box, and so on (also never heard of bluetooth for small files?)

moreover, apple's software is generally very buggy.

michaelmrose6• 3 years ago

What is KDE Connect? Which incidentally runs equally well on other desktops and even windows.

https://userbase.kde.org/KDEConnect

Also gsconnect can provide integration with gnome.

https://extensions.gnome.org/extension/1319/gsconnect/

KDE Connect is a project that enables all your devices to communicate with each other. Here are a few things KDE Connect can do:

    Receive your phone notifications on your desktop computer and reply to messages

    Control music playing on your desktop from your phone

    Use your phone as a remote control for your desktop

    Run predefined commands on your PC from connected devices. See the list of example commands for more details.


    Check your phones battery level from the desktop

    Ring your phone to help finding it

    Share files and links between devices

    Browse your phone from the desktop

    Control the desktop's volume from the phone

Also you can read and reply to messages on your Android phone by navigating to messages.google.com without installing anything

Firefox and I presume chrome let you push tabs from one machine to the other or with forefox simply pull up a list of tabs currently open on the other device.

Unified remote also gives you a great remote for our multimedia pc.

Basically every email service abd chat app has good integration between desktop and phone.

Mpd and mafa lets me easily listen to my music

Calibre and calibre companion let's me easily sync my books

I doubt one person making these things would make them better

kanbara3• 3 years ago

terrible idea.

people buy apple products precisely because of how integrated they are. this idealism and desire to force companies to support everything (as much as i want gaming on mac/linux and drivers for things supported) is the single most dangerous idea that would completely ruin the user experience for millions.

loup-vaillant24• 3 years ago

this idealism and desire to force companies to support everything

I'm sorry, where did I proposed we do that? As far as I am aware my proposal would only force Apples to split itself in 2: a hardware company that sells top notch hardware (so top notch in fact that even the hardware interfaces are as clean and slick as an iPad's case), and a software company that sells a top-notch integrated software suite on whatever hardware they chose to support (and no other hardware).

This idea that Apple's hardware is somehow better because they don't publish their spec sounds incoherent to be honest. As is the idea that their software is better because it runs on company hardware. If there's still a company selling good hardware, one can still write good software for that particular hardware.

So… forcing hardware companies to support all software? Nope, though publishing their spec does automatically provide some level of support for arbitrary software. Forcing software companies to support all hardware? No more than they do now, possibly even less: hardware vendors will be incentivised to harmonise their respective interfaces to appeal to a maximum number of software vendors, so software vendors will be incentivised to support a relatively limited set of interfaces.

ClumsyPilot4• 3 years ago

people buy apple products

People used to buy cocaine toothache drops and give them to children to get them to shut up. People used to send babies by post. Just because people like something doesn't mean it doesn't have terrible effects. [1]

single most dangerous idea

I mean, sure after guns, drugs, superbugs, global nuclear war, climate change, super-plague produced by synthetic biology...

1 - https://www.irishtimes.com/life-and-style/health-family/coca...

TheRealPomax2• 3 years ago

rather than "disallow vertical integration", start with "disallow lock-out". By all means, vertically integrate as much as you want but it should not be illegal for someone to take your hardware and use it for another purpose that you have no say in, or write and run software using your hardware without you having a say in that.

cassianoleal• 3 years ago

Yeah, someone on that Reddit thread said it right - force the companies to publish documentation for the interfaces. Let the hackers write the drivers without painful, slow and error-prone reverse engineering.

loup-vaillant24• 3 years ago

It would be more prudent indeed to just force hardware vendors to release the specs, and it would be even more prudent still to just disallow lock-out. I would start there like you suggest.

I do suspect however that we'd miss out on many benefits if we don't go all the way the way to disallowing vertical integration eventually. It's also something I rarely see spoken of.

scarface_7482• 3 years ago

I know no other way to say this. But this idea is just dumb. The entire value proposition would be worse for everyone. Software and hardware integration is what makes a product better.

yjftsjthsd-h8• 3 years ago

Software and hardware integration is what makes a product better.

... Except for the times when it makes things worse.

unix_fan2• 3 years ago

This would be terrible for specialized hardware, such as assistive technology. Open source software doesn’t really exist for it because there is simply no interest.

loup-vaillant24• 3 years ago

We could carve out exceptions for specific niches. Ugly, but could work.

Who says the software has to be open? It would be better for sure, but companies may be interested in selling software for assistive technologies.

Interest may go up if the barrier to entry lowers. Undocumented hardware is quite the turn off.

PaulDavisThe1st• 3 years ago

You'd have to define "hardware" in a rather interesting way for that to have any hope of working.

Consider any sort of user-facing device (say, an electronic toy, or a music performance device, or for that matter, even a phone). These are all aggregates of hardware made by other companies. Yet they are also hardware. Being unable to develop such devices along with the firmware that runs them would make it more or less impossible to develop them at all. So where does "hardware" stop and "device" start?

loup-vaillant24• 3 years ago

Let's say one wants to make hardware made of 3 pieces of hardware that talk to each other. They may need to write firmware to do that. At a first approximation, I would say one of the following should apply:

Either the firmware can be modified by the user, or remotely, in which case it should not be shipped.

Or it is an implementation detail of the communication between the 3 pieces of hardware, not reasonably modifiable without returning the whole thing to the factory, in which case that firmware is actually part of the 3-piece hardware, and can be shipped.

chme2• 3 years ago

I get you in principle, but the company should still be allowed to sell hardware with some default software installed, because if I had to search for someone on the internet that wrote software for my specific smart TV, washing machine or whatever this becomes much more painful.

I would rather see a 'right to repair' extended to software bundled. Any device or software that can connect to a network or can be interfaced with in any way is prone to security (or any other) bugs, which the customer should have the right to fix themselves, because they bought it, and in order to do so, the company needs to hand out the source code and documentation for all their devices.

Also about vendor writing Linux drivers, you probably don't want them too, there are already vendor board support packages (BSP) out there that enabled their hardware on Linux, but those are overwhelmingly abysmal in quality.

What you rather want is for vendors to upstream their driver into the kernel, so that it is properly integrated, doesn't break the hardware of others and has a high quality, but that process is not something you can regulate and force.

thomastjeffery12• 3 years ago

Of course your hardware would come with software installed! That software would simply be written by a separate company.

We already do this! HP didn't write Windows, but they sure as hell sell laptops with Windows preinstalled.

PrimeMcFly19• 3 years ago

Trying to leverage right to repair to apply to software, as a way to demand source for commercial apps is a huge stretch.

vegetablepotpie• 3 years ago

I like this idea because it’s incredibly American. If we’re going to go full on free-market, we might as well live up to those principles.

There is precedent for this idea, in 1948 the Supreme Court forced the film studios to give up exhibition (theaters) with the paramount decision. This enabled the rise of independent film, and audiences benefited from more choice.

If you try to do this legislatively the Apples and the NVidias of the world will hire an army of lobbyists and storm Capitol Hill. But it might be to their benefit. After the breakup of Standard Oil, John Rockefeller became richer because competition forced out the less efficient portions of his old business from the market.

It’s true that it could temporarily degrade the user experience of some people, but that’s a small price to pay. People who give up liberty for safety, deserve neither liberty nor safety.

scarface_7482• 3 years ago

There are so many flawed analogies here it’s hard to know where to start. How will it make my everyday experience better as an Apple customer without the integration of software and hardware between my Mac, iPads Watch, Airpods and phone?

Windows laptops are worse in every vector than MacBooks. No consumer is asking for this.

So you want to degrade the user experience for “freedom” by giving the government more power???

thomastjeffery12• 3 years ago

We live with such an extreme amount of vertical integration today that people actually believe it is benefiting them.

Every time I see an Amazon delivery truck, part of me dies inside.

osigurdson• 3 years ago

In practice, what fines would hardware manufacturers be held liable for for any bugs in their specifications? How do we conclude that, yes, a hardware manufacturer hasn’t lived up to the regulation? Perhaps the operating system is “using it wrong”. Do governments have entire departments running tests on these devices?

loup-vaillant24• 3 years ago

That's in part why I didn't propose that we force manufacturers to specify their products, but simply make sure that if they don't, nobody would want to use their product. Who would buy a CPU they don't know the ISA of?

Forcing the separation between software and hardware companies may be more stringent, but it may be simpler to enforce, and likely even more effective at making sure the hardware is humanely specified than direct regulation would be.

jacquesm• 3 years ago

I really like this. Other than an upvote I would very much like to express my support for this idea and I hope someone with connections to the halls of power passes it on, there is a lot of merit to this. It would immediately result in hardware manufacturers having to document each and every input and output of their hardware if they expect any sales at all and it would also do an end run around various monopolies.

Check out the idiocy around Broadcomm's WiFi/Bluetooth drivers on the Mac, it's absolute insanity, there are so many of them that to install Linux on one of the machines with such a chipset you need to copy your drivers from the Mac operating system (if you still have it installed...) in order to give Linux the ability to load firmware onto the devices so that they can work. Three different files iirc.

tbrownaw2• 3 years ago

either they only sell hardware in the EU, or they only sell software. Intel and microsoft would have no problem.

Intel has microcode.

Microsoft has Xbox which I'd guess would have issues getting game releases if excluded from a decent-sized market.

Also. My phone (hardware) came with an OS (software) already installed. As do other pre-assembled computers.

loup-vaillant24• 3 years ago

Microcode can be non-configurable.

I'm willing to sacrifice the business model of game consoles.

I hate OEM deals, it just propagates the current monopolies and oligopolies. I'm okay with forcing everyone to buy their phone or laptop separately from the OS.

Someone• 3 years ago

Force hardware vendors to only sell hardware. If any software that can be changed remotely or by the user is off limits. Even firmware. Conversely, software vendors can only sell (or freely distribute) software.

So, I would go buy a smartphone at a shop, and then buy an OS to run on it at another shop? Or would it be allowed for a third party to integrate the two and sell me a working phone?

Also, would “any software that can be changed remotely or by the user is off limits” mean Intel/Samsung/… couldn’t sell CPUs with microcode installed, or even write microcode? It may be hard to legally define where the border between “software that’s part of the hardware” and “software.” lies.

You'll get the necessary specs required to make it work on all OSes

I don’t see how that follows. Software company C could pay hardware company H to build hardware according to specs they give, with the specs under NDA, possibly even forbidding H to sell that hardware design to others than C.

I think banning trade secrets (https://www.wipo.int/tradesecrets/en/tradesecrets_faqs.html) on the interface between software and hardware, rather than splitting companies into hardware and software parts is the way to achieve your goals.

loup-vaillant24• 3 years ago

So, I would go buy a smartphone at a shop, and then buy an OS to run on it at another shop?

Good point, by default you would. Perhaps allowing some third party to bundle the two could be a nice convenience, but then we must make absolutely sure that they never sell the bundle for less than the separate parts, that would be anti-competitive.

Also, I'm sick of paying for a Windows copy I never use. I'd very much like to see the OEM deals Microsoft make with laptop vendors just die.

Also, would “any software that can be changed remotely or by the user is off limits” mean Intel/Samsung/… couldn’t sell CPUs with microcode installed, or even write microcode?

Depends whether that microcode can be changed by the user or remotely, or if doing so basically means returning the CPU to the factory. If it can conveniently be changed, that's software, and the actual interface to the CPU is not the final ISA, but how to encode that micro-code to make the final ISA. If however the microcode is etched or locked, then it's part of the hardware.

Another example would be an FPGA implementation of a RISC-V core: if the bitstream is locked into the NVCM it can no longer be modified, and thus counts as hardware. But if it's in the easily rewritable SPI flash, then it's software, and it can't be bundled with the FPGA itself.

While I'm not absolutely certain my border is the right one, I believe it's at least one that can be fairly easily delimited.

Software company C could pay hardware company H to build hardware according to specs they give, with the specs under NDA, possibly even forbidding H to sell that hardware design to others than C.

Correct, some trade secrets need to be banned. Even further, I would void all ISA related patents and copyright. Specifically, cloning the functionality of any piece of hardware should be allowed, similar to how the IBM PC was cloned.

Note thought that splitting the activities is still worth it: when software is bundled with the hardware, we still need exploitable specs. Banning trade secrets doesn't force vendors to disclose them, and even if they do disclose them, nothing forces them to make especially readable beyond regulatory mandate, and nothing forces them to simplify their hardware interfaces. Exposing hardware interfaces to the competition however may provide very strong incentive to improve not only the hardware, but it's programming interfaces as well.

thomastjeffery12• 3 years ago

You know, computers that aren't smartphones do exist..

Nearly every non-Apple PC on the planet is sold with Windows preinstalled, yet Microsoft didn't manufacture them.

amelius3• 3 years ago

Yes, and I'd also like hardware and software companies to not have access to my data. E.g. Google should be split into hardware, software, and data companies.

gigatexal2• 3 years ago

What if we allow hardware manufacturers to create software to run their hardware but open the schematics and make the hardware hackable such that others can run software they want? That seems like a decent compromise because there’s very little margin in hardware alone and if you force hardware makers to go that route pretty soon there won’t be any because they’ll all have gone out of business.

loup-vaillant24• 3 years ago

Hardware manufacturers can’t all go out of business. We need hardware at some point. Some might, but what I’m proposing would drive up costs for everyone. Most likely, prices will just go up a little bit.

Allowing hardware manufacturers to ship their own software, even if that software must be free, even if the hardware specs must be released, will still allow them to ship crappy hardware interfaces and compensate that with bloated software. I’m also afraid some manufacturer would try to game the system, with, say, tweaking the instruction set of every new GPU so third party drivers can’t really keep up.

stainablesteel2• 3 years ago

interesting premise, but i ultimately think this would be worse for the tech by forcing them to play in a limited playing field that doesn't have parallel in other businesses, does it?

loup-vaillant24• 3 years ago

Oh but it does have parallel in other businesses. Splitting oil companies being one of the oldest examples. Rockefeller reportedly became richer as a result, so it wasn't even a socialist move.

Another feature of forbidding vertical integration is that is prevents companies from competing with their own customers.

crote2• 3 years ago

This is completely impossible with any hardware made in the last 20 years or so.

The problem is that you cannot make a hard distinction between hardware and software. An Intel CPU simply isn't going to run without microcode, and having to get rid of microcode would easily set back chip design by a decade.

Even something as trivial as a USB cable isn't pure hardware anymore. It has a programmable eMarker in order to advertise its capabilities, so it is also a combination of hardware and software. Everything but completely trivial chips has some kind of computation embedded in it, requiring some firmware.

Chips are complicated and weird. Pretty much nobody writes code which directly interacts with hardware. Even people who write code for embedded microprocessors will use an SDK provided by the chip manufacturer to smooth out the nastiest details. How's that supposed to work when the hardware vendor can no longer provide any software?

And then there's the question of what exactly constitutes "firmware". Is code in a separate programmable flash chip okay if they hardwire the Write Protect pin? What if the exact same code is stored in programmable space in the MCU itself? Is it fine to place that code in One-Time-Programmable memory? Bake it into the chip during wafer manufacturing? If anything, being able to update the firmware in the field is better for the consumer when it comes to fixing bugs.

loup-vaillant24• 3 years ago

The problem is that you cannot make a hard distinction between hardware and software. An Intel CPU simply isn't going to run without microcode, and having to get rid of microcode would easily set back chip design by a decade.

Yes you can make the distinction, and no it doesn't require getting rid of the microcode: https://news.ycombinator.com/item?id=37731829

Simply put, if the microcode can be updated by the user or remotely by Intel, then it count as "software", and Intel should stop shipping it. But if those update are either impossible or require a factory recall, then it's "hardware", and Intel can happily continue to ship it.

The same logic applies to the USB cable, and even FPGA: if the bitstream is locked into the NVCM, that's hardware. If it's merely written in the rewritable SPI flash, that's software. It's not a property of the bitstream itself, it's a property of how it is stored.

Obviously the cost of micro-code bugs will sharply increase. That's kind of a feature though: now hardware vendors must be really careful.

olliej• 3 years ago

So your saying apple would have to stop making operating systems?

Or are you saying Qualcomm would be required to not provide drivers?

This is a super dumb take, like beyond stupid, and I get that it’s directed at apple.

Here’s the thing: since day 1 apple has been a hardware company, and what makes hardware good or bad is often the software that drives runs it. That goes for every product apple makes.

The idea that you can remove one half of that from the other is beyond stupid.

thomastjeffery12• 3 years ago

The same people who are currently working on hardware for Apple would be working for a new company that makes hardware, and the new Apple Software Co. would continue to work with those people just as closely as they always have.

The only difference is that Microsoft Windows, Linux kernel devs, etc. would also be able to work with that hardware company to get their OS working on that hardware.

The hardware company would not be allowed to keep secrets between itself and Apple.

loup-vaillant24• 3 years ago

* So your saying apple would have to stop making operating systems?*

No. That's only 1 of 3 alternatives, the other 2 being stop making hardware, or split in 2.

Or are you saying Qualcomm would be required to not provide drivers?

Yes. They'd have to provide an actual manual for a change.

This is a super dumb take, like beyond stupid, and I get that it’s directed at apple.

It's not. Many more would have to rethink their entire business model.

KirillPanov2• 3 years ago

Now the complicated part is how to define a company. We don't want a single company to just split itself into 2 legal entities

HN routinely seems unaware that this is a totally solved problem in the tax law of virtually every industrialized country.

In the US, for example, the phrase is "Common Control": https://www.law.cornell.edu/cfr/text/26/1.414(c)-2

All sorts of modern tax schemes would have truck-sized loopholes without a rigorous definition of Common Control. It's already in the tax code of every industrialized country (possibly under a different name). All you have to do is reference it. The necessary definitions have been in the tax codes since the 1980s.

I'm kind of annoyed at how often this "but but but two companies with overlapping owners yadda yadda" comes up. Go ahead try that on the IRS and see how it works out for you.

loup-vaillant24• 3 years ago

I should have thought of those issues, thank you. I’ll look up this "Common Control".

pasabagi• 3 years ago

It's a cool idea, but where do you draw the line? Where do you place something like microcode? Depending on how you design a computer, you can have software almost all the way down.

loup-vaillant24• 3 years ago

https://news.ycombinator.com/item?id=37731829

I draw the line at "can it be updated without returning to the factory?"

If microcode can be updated by the user (or remotely by Intel), it’s software. If it’s immutable then its hardware. If an FPGA bitstream (say a RISC-V soft core) is locked into the FPGA’s NVCM, then it can no longer be modified, and it counts as hardware. If however it is merely stored in the SPI flash and can be modified, then it’s software.

doctor_radium• 3 years ago

An alternate paradigm? 20 years ago I mused over a law called the "Operating System Fairness Act", designed to level the playing field, mostly for smaller OS's. It's nowhere as revolutionary as separating hw development from sw development, but may thus be more realistic. One clause (of maybe six) concerned "Forced licensing of must-have drivers, algorithms, codecs, file formats, etc. to any viable OS, and for a price that market can bear".

At the time I was thinking of technologies like the Unisys GIF patent, Flash, various audio and video codecs, and peripheral drivers. If not having a technology makes an OS vitally less useful and appealing to a user then there's potentially a path to developer documentation. Who decides? I think the current system for DMCA exemptions works pretty well (noting that I despise the DMCA), so maybe a branch of the US Copyright Office or another government's similar branch? A viable OS is one with a notable following and which can use the technology. If I write RadiumOS for myself, that doesn't count. Likewise Widevine for the Atari 800 is out. But Linux anything would, the BSD family, and Amiga, BeOS/Haiku, etc. would have a fair shot. Making a free, public developers kit is always better, but offering an NDA to a developer representing each OS family would likely be common. No, it wouldn't be an instantaneous process, yes, there would be paperwork and maybe a lawyer involved, but the burden has to be bearable for each OS/developer. If you're Apple wanting the MS Office file formats, MS is allowed to charge a not insignificant amount, but if you're a free, open source OS, then the cost would be either free or nearly so. Not every vendor in a field would be cracked open. If AMD is very friendly to developers, then Nvidia could keep their video drivers proprietary. With all the OSS available today vs. 20 years ago, something like this is maybe a bit less useful, but IMHO still a significant step forward, and a hedge against companies suddenly taking their specs private.

sylware• 3 years ago

If the hardware is not too insanely complex, until there is a maintained and properly written open source driver with public hardware programming documentation for some OS, a working linux driver will probably follow if this hardware has a pertinent meaning.

almatabata• 3 years ago

As others have pointed out why single out linux? Why not BSD? And second which linux? 4.14? 5.10? 6.2? Can they release it for linux 2.7 and call it a day afterwards? Supporting linux requires effort and money. How would you go about defining "reasonable effort" at supporting linux. If it breaks every month? every year? Enforcing this simply would turn out to become a nightmare. I think a better approach would look like what jandrese said in this thread. Force them to publish the specs so other companies or individuals can write open source drivers for the hardware.

skirge• 3 years ago

What I want should be a right and someone forced to provide it to me? Cool, until I'm the one forced to do something, then it's exploitation.

noobermin2• 3 years ago

I've swung over the last few years from thinking it's manufacturers' fault to realising that most of the fault here lies with the linux project itself. They're the ones who have normalised constant churn and thus needing to keep drivers for hardware, sometimes many years old, up to date year after year or even month after month. After seeing the churn mentality affect everything mainly spreading from web dev circles to the rest of the ecosystem, it's hard not to identify that trend as having been the norm in the kernel for decades.

I've only come to realise that blaming manufacturers for failing to keep up to date with a constantly moving target was totally unfair.

mschuster9111• 3 years ago

I've only come to realise that blaming manufacturers for failing to keep up to date with a constantly moving target was totally unfair.

The intention of keeping the kernel moving and unstable at the inner layer is intentional: it is a negative incentive to make people at least publish documentation, if not outright publish driver source code, in the open.

The userspace-facing layer of Linux has remained very VERY stable over the decades, no matter the consequences [1].

[1] https://archive.fosdem.org/2016/schedule/event/design_linux_...

eep_social• 3 years ago

Sorry but your claim is that Linux is the driver of churn?! That’s a wild thought.

Hard for me to imagine how Linux could be more responsible that the incentives of capitalism, eg planned obsolescence, the race to the bottom, and consumer demand for more and better cameras, storage, etc.

noobermin2• 3 years ago

I didn't claim that it birthed it but that the same issue I see in the unsustainability in software today is literally the norm in the kernel. If it's bad in one domain, perhaps it's reasonable to question its value in the other.

pjmlp7• 3 years ago

Yes, ideology has a price.

ylyn• 3 years ago

They're the ones who have normalised constant churn and thus needing to keep drivers for hardware, sometimes many years old, up to date year after year or even month after month.

That's exactly why you should mainline your code. Then people refactoring other subsystems will fix your code for you.

TacticalCoder• 3 years ago

Could EU force hardware manufacturers to make working drivers for Linux?
Why are these companies like intel, Razer, nvidia or AMD..

The question makes no sense. As one of the Reddit comment says: Intel and AMD are among the biggest contributors to the Linux kernel.

The real-world is pretty much powered by millions of Linux machines running on Intel or AMD hardware (for the most part). Try replacing that with Windows servers and their "working drivers" and then we talk.

nextos• 3 years ago

What could work well is forcing manufacturers to release hardware documentation that is sufficiently detailed to allow developing drivers without reverse engineering.

Perhaps forcing them to do so a couple of years after hardware release, so that it does not interfere too much with trade secrets.

This would avoid perfectly functional hardware to go to the landfill because of no further software support, and it would also prevent incredibly time-consuming reverse engineering efforts.

Vilian• 3 years ago

what about releasing the source code of drivers/firmware after the official support ended

pjmlp7• 3 years ago

Azure, and all the games being emulated on Proton as means to make SteamDeck a viable product.

0xDEF• 3 years ago

A lot of the innovation in the GPU space and competition between Nvidia and AMD happens at the driver level.

This could be an

jraph• 3 years ago

I don't think the law being Linux-specific would be right.

But making them release comprehensive documentation / specs and forbidding them from requiring signed firmware would be something.

phendrenad2• 3 years ago

Of course they could. Will they? Probably not. Should they? Maybe. Do I want them to? Hell yeah. Let's light this firecracker.

jokethrowaway• 3 years ago

This is why society is screwed.

All these people don't understand the dangers of socialism. History is bound to repeat itself.

The cost of the EU forcing everyone to adopt a standard is that next time a new manufacturer need to enter the market, it will have to cater to the need of the 1% (of which I am part, I'd love to have drivers for linux for everything).

Who is going to benefit from this regulation? The existing players who can afford to support linux: it will be peanuts for them and it will make or break a new broke manufacturer.

amelius3• 3 years ago

We can make it so that it only applies to the really big or top 5 (say) manufacturers.

h4ckerle• 3 years ago

I actually agree with your argument, but what does any of this have to do with who owns the means of production? Because that's everything socialism is about.

the_biot• 3 years ago

being forced to make drivers that work equally in linux, windows and even macOS

That is almost a recipe for drivers built on hardware abstraction layers (HALs), so vendors can use literally the same code on every platform. That, of course, results in a bunch of unnecessary HAL code being added -- unnecessary from the OS point of view, which already has perfectly good hooks for all the stuff a driver needs. That's the reason Linux does not accept HALs into the mainline kernel.

What often happens on Linux is vendors will then move their HAL and drivers into userspace, interfacing with the kernel via a shim instead. This has the side effect of no longer needing to open-source the driver code at all, since it's not linked into the kernel. Look how that works out!

In case of Android, it was literally Google that did the HAL/shim work, giving vendors a pass on open sourcing or mainlining their drivers.

yjftsjthsd-h8• 3 years ago

What often happens on Linux is vendors will then move their HAL and drivers into userspace, interfacing with the kernel via a shim instead. This has the side effect of no longer needing to open-source the driver code at all, since it's not linked into the kernel. Look how that works out!

It doesn't stop anyone else from writing better drivers (i.e. the thing they already do) but decouples things so that people can at least update their kernel while keeping their drivers. So, win-win.

chx• 3 years ago

facepalm

Whoever asked this have not considered for one second what would it mean in a way that can be put into legislation to "make drivers that work equally in linux, windows and even macOS". This is an incredibly, incredibly difficult topic. How would you phrase and likely measure "equality" here? I have no clue how to answer this even in layman terms. Perhaps someone would need to create and maintain a test suite covering all class of peripherals covered by the legislation and mandate this test suite passes. But even that would not cover performance. Much good does it to you if the video card tests pass unaccelerated.

And then the way this question is put forward also shows this person is not at all familiar with how the EU works. The EU does not have laws in the very first place. It does not. But that aside, the amount of study and coordination that goes into creating or amending an existing directive is just monumental. You would need a strong, compelling need to go through a multi year process, costing many millions of euros. In this case, the need was crystal clear: "these new obligations will lead to more re-use of chargers and will help consumers save up to 250 million euro a year on unnecessary charger purchases. Disposed of and unused chargers account for about 11 000 tonnes of e-waste annually in the EU". How many EU consumers are even affected? Because mobile phone chargers, these days, affect everyone (above the age of three or some such).

Here's a quote from the relevant USB C legislation:

Stakeholder consultations

The following consultation activities were conducted between May 2019 and April 2021 in order to assess potential areas for revision and the impacts of the suggested policy option in various areas:

– an inception impact assessment (2018-2019) targeted citizens, consumer associations, non-governmental organisations (NGOs), manufacturers’ associations, and individual manufacturers;

– a public consultation (2019) targeted member states, citizens, consumer associations, NGOs, manufacturers’ associations, and individual manufacturers;

– two consumer surveys (2019 and 2021) targeted citizens;

– a stakeholders survey (2020-2021) targeted Member States, citizens, consumer associations, and manufacturers;

– targeted interviews (2021) targeted consumer associations, environmental associations, market surveillance authorities, NGOs, manufacturers’ associations, and manufacturers;

– expert group meetings targeted consumer associations, Member States, market surveillance authorities, NGOs, manufacturers’ associations, and manufacturer

And all of that was to survey compelling to use an existing, well understood, already ubiquitous standard.

webmobdev9• 3 years ago

Remember how old electrical appliances used to come with full circuit diagrams to help repair them? That's one way to go about this is - force hardware manufacturers to provide complete device and technical specification (e.g. https://datasheets.raspberrypi.com/rp2040/rp2040-datasheet.p... (PDF)) for every hardware they manufacture. This should enable any competent system developers to create the drivers for it, for any software system. (After all, realistically, we cannot force hardware manufacturers to create drivers for every OS in the world).

One major objection we can expect is that sometimes hardware manufacturers deliberately cripple their products through their drivers. This enables them to sell a cheaper version, that is crippled, and a costlier one that isn't. One example of this is Intel and AMD manufacturing a quad-core processor, but selling the same processor as dual-core and quad-core (remember how AMD allowed you to "unlock" extra cores on their processors?). I think NVIDIA also limits some of their graphic card with their drivers, to sell the same hardware at different prices.

danielEM• 3 years ago

Any "vendor lock in" practices should be considered as monopoly practices. And anti-monopoly laws are already here in EU. Just a matter of interpretation of what falls to monopoly.

gumballindie• 3 years ago

That would be amasing, and should happen. If you sell hardware you should include all documentation, code and support needed to use it.

viktorcode• 3 years ago

There's absolutely no legal basis for it. EC technically may risk it and then lose in court.

userbinator• 3 years ago

I'd rather they force them to release documentation instead, documentation which they certainly already have, and which hardware manufacturers used to freely provide.

If they're worried about IP, that's what the patent system is for.

KirillPanov2• 3 years ago

Two words: regulatory capture.

poulpy123• 3 years ago

The USA could also do some work...

devinprater• 3 years ago

I can get behind this. It'd definitely be great to see VoiceOver, the screen reader on Mac, get some competition. Maybe then blind Mac users could do their work on Mac, even in Salesforce and Google Docs and Sheets and such, and not hear "Safari not responding" at the slightest sign of overused JavaScript. There's a reason most blind people either use Windows, or don't use a desktop or laptop at all.

Could EU force hardware manufacturers to make working drivers for Linux?

Related Stories