Locked bootloaders?
So now even the PI has now become enshitified.
Next step. You can remove to bootloader restrictions for just $/£/€ 9.99 per month
Raspberry Pi OS has gained a new desktop layout, while surging memory prices have revived a row over firmware that restricts RAM upgrades. Although Raspberry Pi doesn't emphasize version numbers in its materials, the updated operating system's splash screen identifies it as version 6.3, dated September 2026. Its long-running …
So now even the PI has now become enshitified.
No, the RPI foundation are trying to protect their reputation and prevent dodgy resellers selling mislabelled kit.
What would your solution be?
(I will confessed I may have missed the paragraph describing how all the 3rd parties are advertising them openly as post factory modified, and providing the support and guarantees to cover any problems this might cause going forward?)
As was mentioned in the article, if you're skilled enough to perform the mod, you're skilled enough to downgrade your firmware.
Next step. You can remove to bootloader restrictions for just $/£/€ 9.99 per month
Get annoyed about that if it happens, but otherwise given the reasons stated it feels a bit slippery slope. ;)
> You need to be both skilled and determined to even try – it means desoldering the Pi's ball-grid-array system-on-a-chip and the RAM underneath, and then resoldering it. This is not trivial stuff.
Er, No Shit..
Is there anyone here who would attempt this? I've done a fair bit of tiny SMT soldering and I certainly wouldn't..
I bet lots of people here are perfectly capable of doing that.
Although all my carrer has been on the SW part of things, in high school and uni I tinkered a lot with electronics, having even designed and fabricated a couple of PCBs and soldered all components.
Soldering those buggers was a challenge, but practise makes perfect :)
Soldering those buggers was a challenge ...
Reminds me of when I soldered* an extra 4Mb(!), piggybacked to the OEM 2Mb roach inside my very dear HP200LX.
I was 26 years younger but scared shitless ...
The real issue was to get the right/compatible chip, not easy to find.
* all credits for the hack and well written instructions go to a pair of very knowldgeable chaps: Michel Bel and Rudy Moore, well known in HP200LX circles at the time.
.
I was 26 years younger but scared shitless ...
Hey, you had the edge on me. I remember shitting bricks when I upgraded the Kickstart ROM on my A500 - and that was merely pulling a DIP chip out of its socket and installing a different one.
The feeling of relief when it powered on afterwards, and greeted me with that ooh-shiny Kickstart 2.0 splash screen rather than the dreary KS1.3 one, was better than sex. Not that 16-year-old-me had a basis for comparison, of course...
~40 years ago...I read about, but never upgraded the memory of my early Atari 520ST from 512k to 1M - IIRC, it meant getting hold of another set of (16) DRAM chips and piggy-back soldering them onto the existing chips (all but the chip select lines?). Then solder extra decoupling capacitors and wire up the chip select pins which were bent upwards so as not to make contact with the equivalent pin on the lower chip, onto somewhere else on the motherboard. Later boards were a bit easier as they had an unpopulated bank
If you have the right kit, hot plates, reflow station, hot air gun and the correct flux surface mount stuff can be easier to than conventional soldering.
It doesn't require any real manual skill - just the right kit. It's a damn sight easier than all the 99AWG specially solder resistant wires you use for cryogenics
Still not sure I would want to do a RPi myself though.
> Easier to _do_ I guess?
This exactly. Easy to do, but hard to redo as you will no longer have the pre-placed solder balls. BGA _soldering_ is perfectly feasible, but BGA _rework_ is very difficult. For a BGA on the underside of another BGA, I can't imagine the success rate being very high even for an expert with all the right tools.
> If you have the right kit, hot plates, reflow station, hot air gun and the correct flux surface mount stuff can be easier to than conventional soldering.
Easier to _do_ I guess?
That's interesting. I've never dared try, myself, but this is encouraging.
There's plenty of videos on Youtube of people doing it, and like many skilled people doing things making it look easy. (Youtube adding a TB of storage to a macbook neo for example).
Also, I like that you can literally buy solder balls, that really tickles me. :D (Some people seem to use paste, some people just have tiny balls).
However, I have a friend who's an electronics engineer by trade and a pretty good solderer. He's tried several times to do BGA stuff and not had any success so I'm not convinced it's as easy as some of the vids make it look.
One of these days I will try myself. (My gut feeling is you really have to get over the 'I'm making everything too hot' urge).
If you can tie your shoes without assistance, you can do it easily.
It's basically matter of melting the solder.
Apply flux, blow hot air, poke the chip from time to time. If it moves, lift it. Half of the job done.
Clean the pads with isopropyl alcohol. Apply flux, get solder wick and clean the remaining solder.
Clean the pads again. Apply flux.
Place new memory chip on the board. Blow hot air. After a while, poke it. If it comes back to place. Stop hot air. Job done, if not, continue until it does.
It's probably less than 5 minutes jobbie.
You might mess up a couple of times, so better buy some boards and chips to practice first. Plenty of videos on YouTube as well.
> But why, per the article, do you need to desolder the SoC?
It is entirely likely that the author -- that's me -- got it wrong; I didn't check. In previous Pis the RAM chip was sandwiched with the SoC but it may not be on the Pi 5, and I think all the current fuss is about the Pi 5...
All my Pi 5s are in cases and I did not take the time to disassemble one to check, either.
That's for a normal soldering job. Embedded SOCs have a CPU and memory sandwiched / stacked together package on package. You would have to have a fantastically steady hand, heating place and probably a jig / template of some kind to apply fresh paste, heat everything and combine them back together. Not only do you have to seat the memory back on the CPU, but the CPU back on the motherboard.
Many may attempt it, few will succeed.
This typically requires professional level gear, although there may be a few gifted individuals who can do it with a jury rigged toaster heating element and Radio Shack level components.
Per the title, may be a good project for der8auer.
It requires the correct tools, but China is your friend in this regard. Low quality hot air blowers and heating mats are readily available.
I personally would not trust either of those.... things that get hot should probably not be bought on a budget.
The problem with this though is if you want to upgrade, the availability of cheap stuff has meant a lot of things now have 'cheap' and 'expensive', with few options in between.
I wanted a de soldering pump recently. I could either spent 70-80 quid for a no-name brand, or 300 quid on importing the cheapest professional level one I could get. There was no middle ground. :(
...smacks of Apple's walled garden. Combine that with the price (not their fault), then its starts to feel like a betrayal of the ethics of entry level. You can buy an old Windows system for less cash, complete with a monitor.
Surely the solution is to produce a slimmer OS that uses less RAM, but is file and peripheral compatible. And allow people to expand it as they see fit, as we did in the old days. More slots, fewer rules.
You can have the taskbar at the side, but there's no gui switch for it. You need to edit the labwc/rc.xml.
The x11 is there but deprecated so the changes are not aimed at it.
People started selling raspberry Pis with faked RAM and so they had to do something about it. Whingers gonna whinge. Whingers that don't have to field support and complaints about the problems it causes.
Exactly. Perhaps just a non-writeable value set in the SMBIOS (or whatever equivalent it has) on the assembly line, such that on power-on it reports something like "Pi 5, 8192MB RAM (original: 2048MB RAM)". That way those few people who've upgraded theirs have no problems, and buyers have a means to know that they might have been scammed.
> , and buyers have a means to know that they might have been scammed.
Funny how you adopt corporate scaremongering. Once it becomes common language = 3rd party upgraded RAM = scam, the corporations will laugh all the way to the bank.
That's a very odd way to frame it, and not what's happening at all.
Vendors are claiming to be selling an 8GB Pi when it's a 2GB Pi upgraded by them.
If you're selling that as a 2GB Pi that has been upgraded, and being open about that, and offering whatever warranty and support your customers need then it's not a scam.
However if you're passing off your upgraded 2G pi as an 8G pi then that's deceptive, you're selling a product as something it isn't. You're also possibly opening up the pi foundation to unnecessary support when customers believe they've bought one thing and in fact have bought another. (at the minimum you're selling a product where an important component has at least had a lot more thermal stress than it should've had).
I am curious to know why this is a scam. Are ther other hardware differences between a 2GiB model and an 8GiB model besides the memory capacity? If teh memory capacity has been increased by replacing the RAM, and that RAM is usable, then surely there is no practical difference between a board that originally had 8GiB RAM and one that has had its 2 GiB chip replaced with an 8GiB chip. I'm assuming that the replacement still has to be compatible in terms of capabilities (i.e. clock speed, etc.) or it just isn't going to work. Is there some subtlety I'm missing?
Is there some subtlety I'm missing?
Yes. Per the original coverage, the reason that the RaspPi Foundation did this is that some unscrupulous vendors were upgrading, to variable standards, lower-memory boards with the RAM equivalent of that sewer-grease that's used in China. Hence people who bought reworked "8GB" boards were rolling the dice; it might work perfectly, or it might flake out in interesting and unpredictable ways.
Which is why I suggested a boot-time indicator of the board's original, as-shipped config, so that users who thought their brand-new 8GB board might be faulty could know to ask questions of the reseller, not the RPF.
I don't see a probem when a person buys a PC with an extra stick of RAM installed at the store. I also don't see a problem when someone buys a car with a dealer-installed security system. Such practices are standard for value-added resellers and can benefit everyone.
I don't like firmware updates from a printer manufacturers that block third-party ink. I don't like encryption used in farm machinery to prevent repairing a tractor, operating systems that refuse to execute programs that weren't purchased through the company store, and firmware updates that brick Pi's previously field-upgraded with more RAM.
While many companies do this, when monopolistic power is used to monetise customers by decreasesing the value of already purchased products, this trades my goodwill for distrust.
I don't see a probem when a person buys a PC with an extra stick of RAM installed at the store.
But you know it's installed at the store, and the store will support and warranty that for you.
Put it another way... if you bought an 8GB Pi online, and it had problems, would you return it to the seller or the RPi foundation? It's the people doing the latter (believing their hardware is defective) that's causes the problem.
While many companies do this, when monopolistic power is used to monetise customers by decreasesing the value of already purchased products, this trades my goodwill for distrust.
The Raspberry Pi foundation is not in any way a monopoly. (Hell, they'll even sell you the SOC if you want to design and build your own). This is a manufacturer trying to protect their bottom line, it's not a monopoly price gouging.
"People started selling raspberry Pis with faked RAM"
But was it "faked RAM" or just alternatively sourced and they don't like not getting the profit?
A Pi 5 1 GB board costs $45, Pi 5 16 GB costs $305. If someone can upgrade 1 GB to 16 GB for less than $260 then why shouldn't they be allowed to?
They could have set a warranty bit when upgraded RAM was detected to remove their support obligations. Instead they chose to brick the boards.
So is this benefitting the user or protecting the revenue stream?
>So is this benefitting the user or protecting the revenue stream?
I imagine they get more than enough revenue stream from the industrial compute module - commercially it's probably not worth selling individual consumer RPis today.
What they are protecting is the brand. If knock-off / dodgy upgraded RPis start getting common enough, then RaspBerryPi seems no safer than buying a DingleBerry SBC from Aliexpress or an SD Card from Amazon - then the whole point of a single sourced, trusted, well-supported machine for kids and schools is lost.
Individual boards are also a gateway to the compute modules - get people familiar with the system then when they need to build it into a machine they have the obvious choice.
If all you have is a hammer, everything looks like a nail. If you just have experience with RPis, every problem can be solved with RPis.
Imagine if PC manufacturers were locking RAM upgrades so you couldn't buy 8GB one and then expand later, just in case you put faulty RAM.
Nonsense.
It is pure unadulterated corporate greed with a heap of gaslighting.
Regulators should come down on Raspberry PI with a ton of bricks for this anti-consumer move.
Imagine if PC manufacturers were locking RAM upgrades so you couldn't buy 8GB one and then expand later, just in case you put faulty RAM.
Straw man. How many people would try to expand by first removing the surface mounted SoC and then the surface mounted RAM devices underneath?
Even the skilled people with the right equipment can't do it reliably and the fallout always comes back to Cambridge, not Shenzhen.
The question isn't about the soldering skills of Raspberry Pi owners. People also don't add an additional DIMM to their HP and Dell desktops themselves nor change the oil in their own cars. These things are done by a repair shop or third-party contractor.
Field upgraded RAM does not affect warranty claims, because the upgrade is obvious and warranty void unless a country has weird consumer protection rules. Maybe that's the point.
Otherwise, all I see happening is bad publicity made worse by the price increases.
> So is this benefitting the user or protecting the revenue stream?
¿Porque no los dos?
Why not both?
It's both blocking unauthorised tampering with something amazingly fiddly and thus easy to get wrong, _and_ protecting their revenues. I don't think RPL is a hugely profitable company. I saw £166M mentioned in this thread. That is a peppercorn amount compared to the OpenAI style outfits, and they are peddling nothing but lies and electronic plagiarism machines.
#1 -- Context matters here, folks!
#2 -- remember you don't need to run Linux on a Pi. Run RISC OS or 9front and a 1 or 2 GB Pi is perfectly usable. Hell, run Alpine on it with no GUI.
... the BBC Micro - designed and built by Acorn and what I'd describe as the spiritual ancestor to the Pi - came in two main variants - a Model A with 16kb of RAM and a Model B with 32Kb of RAM.
If you bought the "A" and later decided you wanted to upgrade you sort-of could... The A and B models (intentionally) shared the same motherboard, but the A excluded the second set of 8 (2Kb) RAM chips, along with some interface chips. If you wanted to, you could either solder in 8 more RAM chips directly... or you could add a DIL/DIP sockets and then simply press your RAM in to those mounting frames. I've seen both approaches work equally well.
Apparently this whole "teaching kids about computers" no longer means what I thought it meant.
Sigh.
I remember owning an Model A.
I bought an upgrade kit from Magenta Electronics, 8 RAM chips. Open the micro up and on my issue 1 board it had the DIL sockets already installed. So, inserted all the chips, double checked orientation, moved the link from positions 1-2 to 2-3 to indicated 32K installed.
Powered up - success, 32K appears.
Mode 0
Hires - excellent, 30 seconds later - screen blanks...... ???
Turns out my beeb was very early and the ULA controlling the video needed to be replaced for some reason.
Sent the machine back for a warranty repair, and they replaced the ULA *and* removed the OS 0.1 which was on four EPROMS, and replaced with a ROM. So I had to pay another £10 to upgrade to MOS 1.2 when issued.
Eventually soldered all the I/O ports in, added 6522 VIA and dual floppy 8271/1770 controller.
Then added a Solidisk 65C02 upgrade daughter board - with a frequency selector switch. Then moved the BASIC ROM to it - flicked the switch to "high speed" and the beeb locked up.
Seems that the BASIC ROM was too slow for the higher frequency, and was advised to copy it to a faster EPROM.
Happy days!
So many years ahead of it's time, though... I had a "Saturday job" and was lucky enough to get myself a model B... then the plinth that included a pair of double-sided 5.25 inch diskette drives with the little toggle switches to jump between 40 and 80 tracks per side... one of those weird RGB monitors that looked like a near-perfect cube until you picket one up and rapidly discovered all the weight was with the glass display...
I went with the Acorn official 65C02 "Second Processor" with that extremely sketchy ribbon cable - the one that you had to carefully fold twice to get it to reach and even then they cut it so darned short [for signal run issues] that you had maybe 1mm of play... But the killer difference was that the second processor CPU clocked in at 3MHz, not the 2MHz of the mainboard 6502, so with a slightly improve internals design you got 50-60% speed uplift right out of the box.
But running Hi-Basic in the slightly faster RAM of the 2nd Processor made quite a huge difference to performance. And not only did you get 44Kb of program/variable space, all of that was irrespective of display mode, because of course the memory-mapped display was handled directly by the main machine.
And "Second Processor Elite" kicked a$$...
Happy days...
In the early days of the BBC, there were problems with the supply of ULAs. I suspect that marginal ULAs that had problems with 32K of memory were still used in Model A's to allow Acorn to ship at least some systems. I think that it had to run at a higher bit frequency serialising the screen memory for the higher graphics modes. I also know that early BEEBs did not have heat sinks on the ULA, something that they fitted sometime later because they ran hot. The fast RAM chips were also in short supply.
I waited 6 months from the date I ordered it (pretty much on the first day that they were accepting orders) for my model B to arrive because of the various supply issues. As it turned out, my system had a faulty 6522 driving the user (and printer) port for the longest time. It worked with most things like joysticks and printers, right up to when I tried putting a SD card reader in the system that was driven through the user port (one of the timers appeared dodgy).
> I remember owning an Model A.
Side note: just before I emigrated to Czechia I was given a stack of BBC Micros.
Mostly Model Bs.
Snag 1: I didn't have the room and was packing the house to leave.
Snag 2: They'd already been offered to TNMOC at Bletchley. Term: if they say yes, I have to give them over.
Turns out one was a Model A.
And a totally un-upgraded Model A.
Apparently _everyone_ upgraded their Model A. This was stock. Damned thing was worth nearly a 5-digit amount of GBP. I thought I could fit a new kitchen.
Turns out, the Museum _really wanted that one_ but only that one...
*Sigh*
"I have to ask myself what it was all for. Is this the future we were promised? If so, I'm starting to regret my part in it."
I don't understand Mike Thompson's opinion. You cannot regret being "part" of something because economic events (RAM shortages and AI) impacts it years later. Mr. Thompson's Raspbian already has it's good and rightful place in history.
I don't think it's his legacy so much as the goal was to attract kids into computers as a career by giving them a cheap computer to play with.
The cheap computer got expensive, for reasons out of their control, but the career we lured kids into pursuing is also disappearing for the same reason.
> Your name ends up being dragged into the mud along with the rest of it.
Hang on. Retain a sense of proportion here.
Raspbian was great _and it still is_. It's the lightest-weight cleanest GUI Debian variant, and it still runs on a Model 1 Pi from 14 years ago. That's quite an achievement and nothing to be ashamed of.
What is saddening is the rise of AI. You teach kids to code, they grow up, get a job, and prompt a bot to code for them, badly... and the resulting bot industry makes a top-end Pi go from £30 to £300.
That is not the Pi's fault, or Raspbian, or RPL, or any of this.
Keep the story straight.
Any one of you in the comments saying "but bots do good code now!" or "it makes my job so much quicker!" -- it's your fault.
I don't think Mr Thompson's legacy has been "shitted upon", this is just OTT:
- Economics is currently affecting prices.
- Older models are effectivity the same price as launch.
- The Pi has involved as it needed to do.
- AI is changing the industry.
Unless you have specific/obscure views, the Raspberry PI and Raspbian is still written into computing history.
I'm struggling to fine a reason for regret here. Maybe I need to be for fanatical.
On the subject of UI changes to reflect the different experience of children and young people with technology: have a look at a recent Casio scientific calculator. The general approach is now menu driven to a much greater extent than previously, with the need to go quite a few menus deep to reach some commonly used functions. I suspect this reflects the deep menus with smaller numbers of choices at each level on the common mobile phone UIs.
The cheapy own brand basic scientific calculators retain the older UI with almost all functions reached by buttons.
PS: while I'm being off topic for this thread, I may as well mention that it appears to no longer be able to browse The Register using a text mode web browser. Including eww on emacs.
Icon: am I just being a stick in the mud?
Probably pissing in the wind but here goes.
1. Not Raspberry Pi Foundation, what was Raspberry Pi Ltd and is now the PLC. Foundation has nothing to do with the HW.
2. Only certain models in the Zero range have the SoC and the memory combined in one package, and only one particular revision of the main SBCs has the RAM underneath. The vast majority of Pi SBC have RAM soldered on the top of the board separate from the SoC.
3. This firmware change was made 2 years ago, and has only just been noticed, so I suspect its not a problem for the vast majority of users.
4. There is no issue with right to repair - replace the chip with the same size chip, and it will work.
5. As someone who deal with tech questions at Pi, Yes, we do get complaints in from people who have not been told there is a RAM swap and find their boards stop working. So this is a real problem. There are more complaints from this than there are complaints from makers trying to do the RAM swap themselves.
6. Pi have passed on the price rises from the RAMpocalyps. because the price of RAM has gone up byu at Least a factor of 10.You cannot even buy a 16GB RAM chip for the cost what we used the sell the entire Pi500+ at.
7. Raspbian is only used for the 32 bit devices now. The 64 bit OS is a slight modification of Debian, called Raspberry Pi OS, which uses the Debian repos, not Raspbian.
8. The firmware change has nothing to do with the IPO, it just happened at about the same time, like a lot of other stuff did.
9. We still sell a lot of kit to the consumer market - we certainly do not ignore it. Remember, everyone pays the same price.