HP & ASUS Rollout Their ARM-Powered Laptops

[r_a_trip]

The "secure boot" feature you're thinking of is specific to intel UEFI. These ARM machines do not use UEFI.

I was under the impression that the Windows RT versions used UEFI on ARM. Of course UEFI isn't a must. Even so, most bootloaders on ARM aren't that much different. If MS wants, they can lock it.

[DrYak]

I was under the impression that the Windows RT versions used UEFI on ARM. Of course UEFI isn't a must. Even so, most bootloaders on ARM aren't that much different. If MS wants, they can lock it.

Indeed you're right. Basically Windows can't boot on anything but UEFI. UEFI was a requirement to get Windows on a device.
But the exact requirement differed regarding UEFI on x86 and ARM.
We'll have to dig through old phoronix articles, but I think that :

- for x86 SecureBoot wasn't mandatory, and a mechanism to add custom signing keys to secure boot is mandatory.
(So basically, you can boot a standard Linux EFI payload (or loader), you can boot a specific shim that is signed with microsoft's key which will in turn load a Linux EFI payload (or loader), or you can install your key and boot your own home-signed Linux EFI payload/loader).
Probably mostly due to Microsoft fearing to be hit with the anti-trust-hammer, once again.

- for ARM SecureBoot was mandatory, and no word about custom keys.
So it might be possible to only boot microsoft-signed payloads.
Or it might be possible to add your own keys, depending on the whim of the manufacturer.
(Which seems to be standard practice in the smartphone market, mostly due to legals still considering them as single purpose devices and not as pocket computers)

On the other hand, back with x86, distributions managed to have microsoft sign a shim. So it might be possible they'll get again a ARM shim signed by microsoft.

Plus why it is a "big problem" if a driver is reverse engineered?

Because community devs doing the reverse engineering don't have magic wands and there's only so much that they can do without having access to actual documentation but having to rely instead on guessing using dumps ? (In general, small problem will take much longer to get ironed out)
Because there are no longterm devs on a payroll specially dedicated to this driver, but instead only community devs doing bits of improvement here and there when they have some spare time left from their normal food-paying day jobs ?

Just look at nouveau - one of the hugest reverse engineering effort - that still sort-of works if you happen to have a chipset that matches one of those that the devs can test on, etc. but there are still weird bugs creeping on.
Vs. amdgpu which basically works, and if it doesn't (like some of the recent chips during the early time after their release) there actual devs paid to fix the problems ASAP.

I don't want to denigrate the efforts of the nouveau team. They are doing wonders given the circumstances.
But they are severly limited in what they can do due to lack of resources.

Actually, Qualcomm is teaming up with AMD to get that modem in Ryzen Mobile laptops. They won't probably have 22 hour battery life, but they will be much faster and more opensource friendly.

So the guys who eventually added a hidden ARM core to their Ryzen CPUs to run some signed blob frimware with ring -3 access to everything, in order not to be left out of the Intel ME/AMT shitfest, are teaming with the guys who make ARM chipsets where the baseband modem acts as the chipset's north-bridge while running blobs that legally have to be written only by special license-holder ?
So you're piling closed blobs upon closed blobs ? What could possibly go wrong...

I guess this is some kind of wine like system.

No. Wine Is Not an Emulator - that's where they got their name from. Wine is just making a win32 API accessible to application. And Windows 10 already supports this API simply by being windows.
What they use is something like QEMU User Mode : they have a special emulator for running an app written in a foreign instruction set (x86 apps on ARM CPUs) and then they pass the system calls to the actual system.

[coder111]

Right. Is the BIOS/Bootloader "secure" and locked to only boot Windows, as with other ARM Windows devices? And there have been other Windows ARM devices (some Surface versions).

If so- I'm not touching any of those devices with a 10 foot pole. BTW- this is not the first time ARM notebooks are being made. Last time it happened (~2007/2008, started with whole netbook and Eee PC movement, and there were some models running Linux)- Microsoft shut down the whole thing by strongarming/threatening OEMs. I guess they didn't have Windows version ready to run on Arm back in the day...

[robclark]

The "secure boot" feature you're thinking of is specific to intel UEFI. These ARM machines do not use UEFI.

It's a safe bet these arm laptops will be using UEFI. That is what the windows 10 IOT version thing is using on dragonboard 410c. And it is what qcom (and everyone else) is using on there server chips.

[starshipeleven]

Right. Is the BIOS/Bootloader "secure" and locked to only boot Windows, as with other ARM Windows devices?

Probably. Although phones (more close relatives to this device's SoC) are long since unlocked. (2015 and the guy updated the tool like 5 days ago so it's still 100% fine on newer stuff too) https://forum.xda-developers.com/win...nlock-t3257483

[schmidtbag]

Not that long ago I was wishing for a decent ARM laptop. The NovaGo seems to be everything I want, though just a tad expensive. Drop the price by $100 and I'm sold. But, I would definitely like to know the Linux functionality before committing. Windows being pre-loaded on this is discouraging. Unfortunately Asus really likes to suck up to MS. HP does too but they're less blatant about it IMO.

[pal666]

can a generic installer be made for ARM that would work on all ARM devices?

it can be made, but it is not made yet. current generic installer would work on (regularly growing)subset of all arm devices

[pal666]

That's what I was thinking, and afraid of. If Windows 10 ARM takes off, and more computer manufacturers take note and start making more ARM-based computers over x86/x64, would that not make it more difficult for an average Linux user to install their favorite distro on one of these? Instead of having a generic installer, you'd need to get (or make) a pre-built image for your specific hardware.

lol, do you think they pre-build windows 10 for each board?
windows 10 is much less flexible than linux, if windows 10 can boot there, so can (generic)linux

[pal666]

"Good community and culture" is far from being sufficient to maintain good quality builds for many platforms

is that why windows does not have any quality builds for many platforms?

[pal666]

In my opinion, the big problem with Linux on ARM is the most of the graphics drivers are closed source (or reverse engineered). The exception is the the "VC4" (Open GLES 2.1 ) on the Raspberry Pi 2 / 3. There is also work on the VC5 (OpenGL & Vulkan).

while freedreno is reverse engineered, it is probably best arm opengl driver atm(vc4 has much less features because its hardware has much less features)