Raspberry Pi computers require a piece of non-free software to boot — the infamous raspi-firmware package. But for almost as long as there has been a Raspberry Pi to talk of (this year it turns 10 years old!), there have been efforts to get it to boot using only free software. How is it progressing?
Turns out a lot better than expected.
Hah, I was wondering if anybody besides clever was working on that. Good to see that he’s still making progress.
In my opinion more should have been done in the last decade to de-blob Raspberry Pi. Doesn’t really feel this is a priority for them. If they would invest more money in this area. Then likely they would make big progress. On the other hand i do suspect they could lose partners doing that. As reading on how a single person is working on some area. And doing that more or less as a hobby project. It’s like saying we will get there. In the next century. Maybe. But there is no real interest for that as we are perfectly OK with the blobs. All in all Raspberry Pi please do more in this regard. Invest more money and employ more people to work on it full time.
Ahh, yes… That’s what the Raspberry Pi needs most: A replacement for a perfectly functional piece of firmware that boots the computer and isn’t bothering anyone, not -say- a working OpenGL driver that can be confidently enabled by default (so Raspberry Pi users can use the all those OpenGL application packages that exist for Desktop Linux).
Yes, I know that the Raspberry Pi officially supports OpenGL ES, but that’s not what most Desktop Linux apps use.
At some point, the FOSS community should understand they will never be able to purge all proprietary software. The HDMI monitor they connect their Raspberry Pi to most likely runs proprietary firmware too. A smarter move would be to focus on the things that matter, aka replacements for proprietary software that can create lock-in (via network effects) or fixing broken stuff.
“Fanaticism consists in redoubling your effort when you have forgotten your aim”
~George Santayana
Therefore your suggestion goes in the direction of Raspberry Pi should become a software company. They should provide FOSS alternatives to proprietary software that can create lock-in. Instead of making sure you can actually boot their computer without a lock-in. Sounds a bit strange doesn’t it? Your suggestion.
No, I mean providing FOSS alternatives that don’t require lock-in to proprietary alternatives that do. And either the FOSS community or the Raspbery Pi company to fix broken stuff like the OpenGL driver for Raspberry Pi,
Exactly. It just baffles me that you wrote that you in a way object working on a FOSS alternative to a blob. An alternative that would enable you to boot their computer without the blob and the lock-in. That is after a decade and tens of millions of units sold. You still can’t even boot their computer without a blob and a lock-in. Not cool.
Such as? I can’t think of any proprietary client software which would run on a RPi that people could reasonably replicate.
That mostly seems like a job for very expensive server.
kurkosdr,
The problem is that closed code holds back indy & 3rd party development and innovation. I’ve often encountered problems and limitations resulting from proprietary code and firmware. You may not appreciate 3rd party needs because proprietary blobs are fine for you, but it doesn’t negate that there are those who think outside the box and see the greater potential of things. Not everyone likes being confined to the whims of proprietary manufacturers.
kurkosdr,
Although you might be right in terms of priorities (implementing a missing feature vs re-implementing an existing one), there are still some concerns.
First, this is a hobby project after all. And people do love tinkering with stuff. And Raspberry is made for tinkerers.
Second, as the demo projects demonstrate, there is more functionality possible from the firmware than what is provided by the default one.
That being said, I think Open GL is not the right target at the moment. The current efforts are better spent on making Vulkan more stable, since it is already in a better state:
https://www.phoronix.com/scan.php?page=news_item&px=V3DV-Vulkan-RPi-2022
For example running Quake 3 @100+ fps:
https://www.linuxadictos.com/en/raspberry-pi-vulkan-quake-iii-100fps.html
Not to mention that with a fully functional Vulkan driver, you get OpenGL virtually for free with Zink.
They know the aim. The aim is to de-blob the RPi.
Fixing broken stuff and creating replacements for proprietary software, would be replacing a binary firmware blob. The RPi is a hardware project.
There are other projects which are working with on software.
Someone has to try or it won’t happen.
> a perfectly functional piece of firmware that boots the computer and isn’t bothering anyone
Yeah, no. Believe me, enough people, myself included, have run into enough problems with that firmware. If I had had the chance to look through the code to see what was wrong with it, I’d have done that, but I couldn’t, because it’s closed source.
What it’s actually odd is why Broadcom just don’t open source the current firmware. It’s not like it is usable on anything but their own products.
This looks more a issue of corporate culture than a technical or intellectual property issue.
It’s this fairly obsolete tech on top of everything? I thought part of the appeal, for Broadcom anyway, was the RPi was using up stock of old chips.
It continually amazes me how hardware companies don’t want people to actually use their products. I get using professional services as a profit center and artificial segmentation via software (Nvidia, Intel?), but wouldn’t they get more money if people bought more of their chips? x86 seems to be doing pretty well with their open, -ish, platform.
That’s the very reason that these fine Taiwanese and Chinese companies are cornering more and more of the semiconductor industry and chip design.
I found that dealing with established companies from USA and Europe to be a major pain in the ass. Some are learning (Microchip and STM for example), but lots are so ingrained with a 80s/90s mentality that is shocking.
Want to use their products? Well….
– first you must figure out how to purchase their development board, that may require you to send a mail to a commercial department, that will make you answer 2000 different questions. Doing personal project? Screw you, we only deal with companies with portfolios or “university partnerships” for educational deals (you know… only people at universities learn new stuff, right?! right!?!!?!).
– Then, after they decide that you are worth of their marvelous product, they will send you a offer to purchase a 6000 dollars development board, with a lead time of 5 months.
– You want to place a order? Their commercial will first send to you a pile of legal doc called “NDA” wrote in the most arcane legal language possible to sign, and you have to wait a week or two for it be reviewed.
– The order is placed.You wait the lead time, it arrives. Ok? No. Now you must deal with their “development support” team. Given they make as hard as they can get to easily retrieve information about the chip. Either they will hide all technical documentation (that may be very incomplete) behind a login wall that you must first request a access or send to you by email. Community support? Don’t make me laugh, they don’t like their “customers” talking to each other. You might find some forum with user attempting to use their products if it is popular enough, but don’t dare do mention it while dealing with their support.
– Finally you got a product. Now you must get a OEM, get your vendor to accept delivering to that OEM after some deal and lots of legal documents. They may also request to see the code of your software and review your practices to ensure that no pesk leaks of their libraries or proprietary code happens.
While from a Chinese or Taiwanese company:
– First, open Aliexpress/banggood/whaterver.
– Second, buy the dev kit. Usually will cost a 100 dollars or so.
– Third, use.
– Forth, do you have a product? just ask your OEM to place a volume purchase of the component and pay for it.
That’s it. And with this guerrilla strategy, they are taking the world, given that innovation happens with the small guy working at his spare time, and with such low entry barrier, anyone can develop hardware now.
And that’s how Rockchips, Allwinners, unisocs and mediateks are all over the place.
That’s cool. Thanks for the info!
Indeed. One of my little projects is figuring out how to build carrier boards.