A version of Librem 5 without any modem

Hardware Librem 5
21

@ruff and @Caliga, I am not exactly sure what you are saying about the functionality provided by the modem.

The modem, for example the Sierra MC7455 thats being used in the latest report, runs its own OS (Linux+some proprietary bits as said by Nicole(?) elsewhere), implements the 2G, 3G, 4G protocols, is capable of using GPS and so on. Its not just something which converts bit streams into RF signals. Sure, thats one part but it does a whole lot more which is not transparent to the application processor. If the application processor knew about every bit that’s being transmitted/received by the RF antenna then we wouldn’t have the issues we have now (stuff like CarrierIQ, SilentSMS,…)

Take a look at the SIM900 modules. With just an Arduino controlling it by sending appropriate AT commands I can send/receive SMS, make/receive calls, connect to the internet via GPRS. All the work is being done by the modem. I am sure MC7455 has this capability. At the least, the data sheet shows pins for audio. However I am unfamiliar with the QMI protocol that it offers and how audio/video/GPS data is fed through that protocol. So maybe the main CPU will have more control.

I know how GPS works. I don’t know what exactly you think I implied about the modem. What was concerning was the GPS function provided by MC7455. If one had to turn on the modem just to use GPS then it would have been a huge problem. But I read elsewhere that they are going to use u-blox. So I didn’t mention anything about it.

It seems clear that phone carriers aren’t going away. In the traditional sense, they might but they will still be needed for data connectivity. By this I am talking about the current distinction between phone calls, SMS, packet data. With LTE this is slightly blurred and in the future it might all be just data. But phone carriers and modems will still be needed for the connectivity. You aren’t going to have WiFi everywhere. But like I said before, I would like some friction while using this connectivity.

Based on the input by several people, I now understand that even though a modemless version is simpler to design, since Librem5(with a modem) is what is currently funded, it would be additional work and will end up delaying Librem5. Note that I was never asking for delaying Librem5. However, I can not take a soldering iron/heat gun to a $600 phone. Like @patch pointed out, I hope that they include the modem in some modular fashion so that it can be completely removed/replaced by another/upgraded in some way. After some freedom is achieved, one could put that specific modem back in. Even if the modem has to be soldered down can it be on some sort of a daughter board? Some phones already have separate boards for charging, display, actual RF chip, SIM and microSD slot, etc. Have they said anything about how modular the phone is going to be? A modular design would also allow easier repairs, replacements and upgrades which would be really great considering the current price tag.

Regarding goals, this is what Nicole’s words translate to me: it will be many years before any sort of freedom is achieved with regards to the modem. I am not saying that the team won’t work for it but I see no indicators showing that it will happen anytime soon. OTOH it looks like they can free up the rest of the stuff in a year or so and that is what interests me.

@prog-amateur Are there any details available about the dev kit? Building stuff myself isn’t a problem and I don’t mind bulkiness. So if the dev kit is modular then it would work well. I also see that it will be available sooner.

22

The trouble is, network providers may well require inclusion of such features before they allow a modem to connect to their network. In fact, I’ve always believed it is probably not realistically possible to produce a cellular modem with freed firmware which is strictly still allowed to connect to the network, since they may argue that users could disable or impair important functionality with adverse effects to the network. Their network, their rules. The modem/baseband is effectively the network provider’s domain, even though it is physically owned by the person who owns the phone and installed inside the phone owner’s property. Changing that will take cooperation from providers.

Within the current framework of the way cellular networks work, I think the best we can achieve is to push all of the ugly cellular-network-interfacing modem stuff out into a separate module and consider the interface between the modem module and our CPU to be the boundary between the untrusted cellular network and our more trustworthy local device. I don’t think anyone here is in disagreement about that. The modem will be a separate module that’s only loosely coupled to the main CPU.

The central argument of this thread, then, seems to be whether the modem module should be physically integrated into the phone while maintaining separation, or whether the module should be part of an entirely separate (and optional) device.

Personally I think the usefulness of the Librem 5 is quite closely tied to making the compromise of including the modem inside the device. However, I understand that there are other possible use cases in which an integrated non-free modem is an ugly and unnecessary blemish on the face of an otherwise pristine device!

I think the first release of the Librem 5 will necessarily include the modem. That’s what was offered and pledged for, and I think it’s important to focus on getting that one thing done properly before introducing further complexity.

However, if there was sufficient demand, I don’t see why they couldn’t subsequently manufacture some phones with the modem unpopulated, even if it is normally soldered down. It’s the sort of thing where simply omitting the part would do the job. For the software, it would be no different than a phone with the modem kill switch permanently in the “kill” position. The main complication would be inventory management (as mentioned by prog-amateur). Purism has done this sort of thing with laptop features, but those have been things that would be added to the base configuration, not taken away from it, so offering a modem-free Librem 5 might be more difficult than, say, offering laptops with a TPM.

1 Like
23

Cards may run whatever they want, it doesn’t mean they will have access to host. Unless host runs software which bridges foreing os to hosts. HBA cards for ages running own OS for instance.

Again, QMI is a protocol of communication between host and foreing firmware (whether it is simplistic uOS or linux). What will it be able to see on hosts depends on what host will allow it to see, but for that you need to have control over hosts software (no blobs).
Let’s try it this way.
Take example of intel ME. It runs full-blown OS. But does it need to? No. Does it provide any added value? To corporate user - maybe, in some corner cases, but risk outweights the value. To home user? Not at all.
ME though runs on host, so software isolation from host is not sufficient as it’s hooked into hardware. Similarly like Qualcomm SoCs with integrated baseband. iMX though does not have integrated baseband, so whatever runs on foreign chip it cannot inject itself to host, unless host allows it.

1 Like
24

shh…:sushing_face: let sleeping dogs lie :wink:
I think there is no immediate risk of regulation on this, as it is very niche at the moment.
In the long run, the situation will probably be very similar for wifi, gsm, cable and phone line.

I remembered there was some uproar on wifi legislation and FLOSS / DD-WRT some time back (just the first search hit, I don’t know what the current state is).
It is of course legitimate to have legislation on which frequencies may be used and how much power a sender may use (and these differ between countries, that’s why you should select proper countries for wifi drivers).
If that should really become a problem, I think we will see solutions that split the firmware in two parts. The infrastructure-critical one (very small, ideally) can still be open source, but has to be signed.

There is also legislation that goes the other way. Recently, Germany guaranteed users the right to choose their end-device (DSL modem/router). It’s always about the current tide of politics I guess…

Currently, the cell phone infrastructure is in absolutely no way tamper proof. Everybody can just carry a “cell tower” in their pocket and the phones arround him will happily connect! Some about that here: 26C3: GSM, SRSLY? (again, just the first best link I came up with)
So I think they can hardly argue that free software poses additional risks :wink:

[EDIT] Sidenote: according to that video, open source software that implements the GSM protocols and emulates a base station was available almost 10 years ago. So, firmware sounds doable.

Now, on “Their network, their rules”: Of course, the firmware has has to be standards compliant. So it will probably (have to?) process a ‘silent SMS’, but I don’t know if legislation exists that prohibts the firmware to let the driver/software/user know that such an event happend.
Should such kind of legislation happen (US, Europe), please let’s finally realize that we don’t live in free societies… But even then you might find some way around and replace the original firmware after shipme…:speak_no_evil:
:gun: :gun: :gun: :male_detective::male_detective::male_detective:
this user does not exist - go away - tell nobody what you did not see here

3 Likes
25

The Developer kit is available for $299 here (scroll-down) :
https://puri.sm/shop/librem-5/

Caution : I don’t know if such modem is sold as a module, and if the pre-ordering is still possible.

26

You might be right about “freed” firmware. However if anyone could verify the binary firmware with an open source code and open toolchain, it might be sufficient. PR issues would mean that they would fix bugs and remove all spyware.

This is exactly how I feel. The modem “pollutes” Librem5 and since it isn’t that important to me I would rather not have it. I also know quite well that the version that has been funded promises a modem and the backers don’t feel as I do.

@ruff I was responding to what you wrote earlier about baseband providing JUST RF. CarrierIQ, SilentSMS and who knows what else tells us that its just not the case. Having access to the host is irrelevant if the modem can do compromising stuff on its own. If SIM900 can deal with calls on its own, MC7455 should definitely be capable of doing so. Do you/anyone else know how audio is dealt with by MC7455? What control does the host have? This is what I was talking about with regards to QMI.

@Caliga With Wifi, just regulating the Tx power and RF band would suffice but modems connect to cell towers and more regulation would be required. As you say, open source part of FLOSS might be the only feasible part. Is the OSS mentioned either Osmocom or FreeCalypso?

Also we don’t live in as free a society as one would think. The US House and Senate have legalized mass surveillance. China, Russia are “bad” from the beginning. India is getting there with its biometric id project spreading tentacles to every part of life. Its good that EU has GPDR but is there nothing going on behind the screens?

27

The option to pre-order is there but its not clear what the development mainboard will look like. Do the Purism team reply to email inquiries or should I create a thread about this?

28

Have a look at this recent interview with Todd. It is my understanding that the dev-kit hardware shall be as close to the final hardware as possible. We will hear more about that in March/April.

1 Like