Hacker News Reader

Comment by swiley | Hacker News Reader

It’s just as closed as the typical laptop (better in many ways, some newer laptops have some nasty stuff in the bios) I think there might be 2 WiFi cards that even have open source firmware and the firmware isn’t complete enough to be usable for most people.

I’m not sure who would be surprised by this, I feel like anyone that cares this much would be paying enough attention to know what’s going on.

View on HN

It is, however, a lot more open than most ARM Single board computers and ARM phones. Most ARM Phones/SBCs cannot run mainline Linux, and due to closed drivers, will likely never be able to. You in addition have Phones with locked bootloaders, so you can't even load your own firmware in it!

This gets us into the situation we see today: even if you have a Phone that you can run your own firmware on it, it has a lifetime because the vendor has little financial intentive to do so.

Even the most open Android Phones (Pixels) have about a 3 year lifetime:

https://support.google.com/nexus/answer/4457705

On an iPhone, you are completely at the mercy of Apple for updates (for which they have a much better track record than Android).

Meanwhile, my Thinkpad x200 was released in 2008 and continues to be a supported just fine, with no end in sight. I hate to say, in contrast to even my Novena, is much better, as my Novena sits in my desk because it does not run mainline Linux, and I can't even get it to run properly on an updated Linux 4.4 Kernel that I tried to compile (due to no support).

So the Pinephone (and Librem 5) shooting to ensure that they can run unmodified Mainline Linux is a huge win for openness and longetivity in a Phone.

There's been some great work on mainline support for the Pixel 3XL / Snapdragon 845: https://gitlab.com/postmarketOS/pmaports/issues/153#note_207...

That's good to see! If I remember right, the last Google Phone with mainline support was a Nexus 5.

The Nexus 5 did not have mainline support while current; it ran a patched kernel like just about every other AOSP/Android build. Mainline support is now being provided by the postmarketOS folks for the Nexus 5, and could be achievable for other devices.

Ahh, thanks for the correction.

> On an iPhone, you are completely at the mercy of Apple for updates

Hm, isn't it theoretically possible to use the checkm8 exploit to boot your own code? Is the problem just that writing your own OS for the iPhone is difficult, or is there anything actually blocking you once you have code execution in the boot ROM?

Boot, and then what? Are there open drivers to the iPhone peripherals? Last I checked, there weren't: only having access to the CPU is not entirely satisfactory.

That's the point of the article: being capable of running code on main CPU is but the first step, drivers and firmware for all the other parts are far more problematic.

Uh, no? This is much more free-er than typical laptop, be it recent or old.

The vast majority of laptops have closed-source BIOS. The vast majority of laptops have components with closed-source firmware with access to RAM (Intel ME comes to mind), which includes BIOS (because it keeps running after boot).

Also, only modern laptops have IO-MMUs to prevent PCI-Express devices from accessing the RAM directly. (PinePhone simply doesn't use PCI-E)

And I don't think there is /any/ x86 laptop on which we know how to do DRAM initialization without closed-source blobs.

> I don't think there is /any/ x86 laptop on which we know how to do DRAM initialization without closed-source blobs

Pre-SandyBridge 2009 era stuff?

Modern desktop CPUs require blobs to boot (Intel FSP / AMD AGESA), Intel laptops (except Chromebooks) are especially bad because Boot Guard. Anything with Allwinner/Rockchip/NXP/Marvell SoCs is not nearly as closed.

How are chrome books more free? Don’t they require everything to be signed (with support for disabling user space checks at the cost of a loud threat from the firmware during start up?)

It's not a "loud threat", and it's only if you keep the stock coreboot. With a debug cable, you can flash anything you want, I have a fully custom coreboot+edk2 build with cleaned ME and no signature checks anywhere.

The librem phone seems a lot more open, and the zerophone is also very interesting.

I'd say both are about the same. The Librem 5 physically puts the Modem and Wifi/BT modules on a physical card, where the PinePhone make it all into a single board. I believe both have the same physical/logical seperation (i.e. Modem and Wifi/BT are on a Bus versus integrated directly into the CPU), and both support a hardware disable due to the physical seperation.

Both Phones have to load firmware into the Modem and WiFi/BT.

For all practical purposes, it looks to me that there's no difference between soldered and slotted when they both have a physical switch to disable them.

In theory you can change the modem/wifi in the Librem, that may be more "open" but you pay for it in size.

the librem isn't more open than the PinePhone at all. The only difference is that they put the modem on a separate card… but that doesn't make it any more open at all. What matters, given the fact that there just isn't any open modem out there, is the fact that the modem is kept isolated from the SoC in a way that doesn't allow the modem to access anything the SoC doesn't explicitly give it.

ESP8266 is not free, not secure, but open and fully usable(if used in its originally intended purpose)

Is it fully open? I thought the lower-level WLAN firmware was only available as binary blobs there too.

Reply on news.ycombinator.com