Hacker News Reader

Comment by djsumdog | Hacker News Reader

Exactly this. ARM is not a platform. It's some SoC where each manufacture solders some random pins to random chips. Few manufactures publish device tress and only the dead Windows phones have UEFI+ARM (with their locked bootloaders. Fuck you MS).

If ARM manufactures want to get serious about devices for enthusiasts, hobbits and Linux users, they need more UEFI+ARM solutions that can boot plain old Linux or Windows 10 ARM.

View on HN

> locked bootloaders

Easily unlockable with WPInternals :)

> only the dead Windows phones

Marvell/SolidRun MACCHIATObin, SoftIron Overdrive, even the old Gigabyte boards with APM X-Gene have both U-Boot and EDK2 (TianoCore) firmwares available. I'm not even talking about big servers (ThunderX) :)

Heck, U-Boot itself can run EFI binaries, with network access for netbooting even — and it's good enough for FreeBSD/aarch64 to not even consider non-EFI booting methods.

I wouldn't put SolidRun in that list. Their hardware is garbage. I wasted money on this thing:

https://penguindreams.org/blog/review-clearfog-pro/

Their support is terrible and they only have a 14 day refund window.

I have a SolidRun CuBox running 24h/7d for about 6 years and the thing is a beauty. I agree that the company could be better though - they stopped providing updates very quickly, and abandoned their apt repository. I looked at newer models but their prices were way too high in a market where buying a Pi will mean getting less headaches as well as saving money. I don’t think their focus is on NUC/plugs anymore.

just out of curiosity...

if it's an eval board, then do they provide schematics which are typical of other eval board producers?

Having the schematic might aid in getting things working assuming the MAC <-> PHY circuitry supports what you want to do.

I've always been surprised that nobody's tried to weld ARM SoCs and the ACPI standard together yet. (How would that even work? I have no idea. Maybe the SoC would have a standardized PMC embedded in it?)

> Fuck you MS

To be fair I've yet to see a good ARM manufacturer and they'd all deserve a "fuck you".

Rockchip, Marvell, AMD (barely an ARM manufacturer, but SoftIron sells them), Cavium (mostly on super-expensive servers though)

Marvell? really?

Pretty sure Rockchip has violated Linux's GPL. Rest of who you listed I've never seen/used in real life.

I think they did, but I'll take GPL violations and (possibly leaked?) publicly available documentation over GPL-compliant source code dumps (which are next-to-useless for actually understanding the hardware anyway.)

> possibly leaked

Not leaked, entirely official: http://opensource.rock-chips.com ("Hardware Support" section)

And instead of dumps, they have proper git commits! https://github.com/rockchip-linux/kernel/commits/release-4.4

And lots of upstream linux commits with @rock-chips.com addresses.

Interesting. The docs I found before were marked "Rockchip Confidential". But the ones on their site are not complete either --- the fact that only "Part1" of the TRM is there suggests that there is a 2nd part (and possibly more.)

Yeah, I've seen the "confidential" mark on the same documents (or older versions) — these were published on board vendors' sites (e.g. Pine64's).

Not sure what would be in a "secret 2nd part"… the only thing I can think of that's missing from "Part1" is the display system. But there is a driver in mainline Linux for that https://github.com/torvalds/linux/blob/master/drivers/gpu/dr...

Maybe a very long time ago. Or you're confusing them with Allwinner.

Rockchip has the source of all the things on their GitHub, and a ton of commits in mainline Linux.

You've never heard of amd?

Reply on news.ycombinator.com