You are speculating that someone is going to design a custom chip with the same dimensions, same labels and IDs, and same functionality as the real chip and then infiltrate the Purism supply chain in Carlsbad, California just to put those ersatz chips on one of the boards in L5USA? Do you realize how much work that would involve and the kind of budget that would take? Who would do that for a phone in development that has to be charged twice a day, has only shipped 2600 units so far, and is mostly used by Linux geeks?
Even if you distrust Synopsys’s proprietary DDR4 timing blob stored on the separate 2MB SPI NOR Flash chip, then you can pull that blob off the chip and compare it with the code that you can download from NXP’s web site (you have to register with an email address that looks like a company email, so not Google, Yahoo, Hotmail, etc.). You can run md5sum on the two blobs to ensure that your phone has the same Synopsys blob as NXP’s web site.
Synopsys (which has a $53 billion market cap) and NXP (which has a $58 billion market cap) are not going to risk their businesses by providing spyware, because if news ever got out that they were doing this, they would lose serious customers and their stock value would plummet. Synopsys and NXP sell to companies that do understand the technical details, and Synopsys understands that it would lose most of its business to competitors like Cadence, Mentor Graphics or Agnisys, if it were selling them spyware. Look at what happened to Supermicro’s stock when it lost 40% of its value, dropping from $20.61 to $12.40 within a day after the Bloomberg story of Chinese spy chips being inserted in its motherboards. The stock value recovered, because no evidence was produced to back up the Bloomberg story, so most businesses that buy Supermicro motherboards decided that the story was false. However, if the story had been verified, Supermicro would have been destroyed as a business. Look at how much effort Cisco has put into lobbying the US government to stop the NSA from intercepting its equipment during shipping, and that was a highly targeted program, that probably only effected a tiny number of its customers. Google and Apple have millions of clueless customers, so they aren’t nearly as vulnerable as businesses like Synopsys and NXP that supply tech companies, yet look at how Google and Apple started implementing encryption everywhere to hinder NSA spying after the Snowden revelations publicized the fact that the NSA had access to their servers.
Aside from the fact that I don’t think that Synopsys and NXP would risk putting spyware into the DDR4 timing code, I also don’t think that it is technically possible to do any serious spying at that point in the boot procedure, when almost nothing is functioning, so you would have to include all the code to bring up the critical systems and operate them–in other words, you would need a mini-operating system and the drivers for some communication device and a TCP/IP stack, and I don’t think that is even possible with just 2MB of code. No matter how good your obfuscation techniques are, you can’t hide an entire OS of that size, and someone looking at the binary with a hex editor can see that something isn’t right, plus they will see communications from the WiFi or cellular modem and will wonder why there is communications during bootup and start analyzing the traffic.
The only scenario that makes sense to me is if the DDR4 timing blob inserts some code that will be executed after u-boot finishes the bootup, but that means altering code stored on the eMMC, and then having that code disappear, so it is not detectable later.
At this point, the L5 doesn’t have verified boot, so I’m not going to say that it is impossible for the DDR4 timing blob to insert code in the boot files which aren’t currently encrypted, but there is no reason for Synopsys, NXP or Purism to want to risk destroying their businesses to do this. The only way that this makes any sense is if you are talking about some 3rd party hacking both the NXP and Purism servers to change their copies of the blob and altering their published checksums, so that none of this is detectable.
Honestly, I don’t think you are talking about anything that will ever happen in the real world, because this is beyond anything that a normal hacker can do, and it requires the resources of a state actor with deep pockets (like the USA’s NSA, Britain’s GCHQ, Israel’s Unit 8200 or China’s 3PLA). There is far easier hardware for state spy agencies to target, and at this point, I doubt that many people high on their target lists are actually using the L5.
Even if Purism were secretly collaborating with the NSA, it would be very hard for Purism to do any spying with the L5, because so much of the software can be verified and altered by the user, and there is so much isolation of the individual components. The L5 uses six separate chips in place of an integrated mobile SoC like a Snapdragon, and components like the RS9116, BM818 and Teseo-LIV3F really can’t get at the memory of the code being execute by PureOS or any other distro that you chose to install. Remember that Purism has made over 150 commits to mainline Linux to support the L5’s hardware, you so that in the future it will be possible to install any distro on the L5 (which was one of the original crowdfunding goals), so that you don’t have to trust Purism’s OS.
Purism would also be destroyed as a business if someone discovered that Purism was deliberately putting backdoors in the L5, and I imagine that many of the people working there would also quit on principal. I have no idea what Purism pays its developers on the L5, but I’m pretty sure that they are all taking pay cuts to work at Purism, and it has to be a very stressful job, so they do it because they believe in the mission, and those sort of idealistic people won’t stay if they discover something nefarious at Purism.
If Purism had some nefarious plot to hide backdoors on the L5, it wouldn’t have hired people like Guido Gunther (Debian dev), Adrien Plazas (GTK/GNOME dev), Mohammed Sadiq (GTK/GNOME dev), Alexander Mikhaylenko (GNOME dev), Evangelos R. Tzaras (Mobian dev), Angus Ainslie (former Openmoko distro maintainer), Sebastian Krzyszkowiak (long-time Linux phone hacker for FSO and Neo900) and Nicole Faerber (founder of the GPE Palmtop Environment), because these are people who have long histories of working on this stuff and they have public reputations to protect, so they are likely to publicly expose any backdoors that they find on the phone.
Let’s assume in 5 years that the software will be good enough that there really are significant people using the L5 (foreign government officials, corporate leaders, criminals, political leaders, dissidents, etc.) that spy agencies want to monitor. At that point, it may make sense to worry about nefarious firmware or spy chips being inserted in the M.2 boards for the RS9116 WiFi/BT or BM818 modem, since those parts are essentially black boxes and they handle communications, but I think the far more likely target would be cracking PureOS’s security, rather than messing with firmware or hardware spy chips. In 5 years time, the L5 will probably have verified boot, good sandboxing of apps with Flatpack+bubblewrap and more secure communications based on encryption keys stored on an OpenPGP card, so it will be harder to compromise. By that point, Ubuntu Touch will be ported to the L5, which already provides verified boot and good sandboxing. Furthermore, it is likely that the community will have found other M.2 cards that can be used in place of the M.2 cards that the L5 uses, so you can switch out the WiFi/BT and cellular modems if you don’t trust Purism.
Aside from the PinePhone, there is no other phone on the market that allows you to recompile everything in the OS and to verify how it works down to the schematics, except the PinePhone. However, the PP’s hardware kill switches are not useful in my opinion, you can’t turn off its sensors with physical switches, and you can’t replace its modem and WiFi/BT, it doesn’t have a smartcard reader for an OpenPGP card, and it isn’t paying programmers to work on its security.
