It Turns Out Windows Unconditionally Reserves The First 1MB Of RAM, Linux Was Just Late To Do So

[polarathene]

AI wonder if this was the reason Lenovo laptops use to break just by booting Linux: https://www.omgubuntu.co.uk/2017/12/...vo-laptop-bios

I bet it's what caused me grief putting linux on an Acer 2019Q3 model I got last year. After the install I remember the BIOS being really glitchy and at one point the thing seemed bricked and unbootable (couldn't access the BIOS either, just blank screen on power on with no activity/response). Eventually after N boot attempts or battery drain, it did seem to "reset" itself and become consistent at booting and accessing the BIOS properly, not quite sure what resolved it into that state but it was a weird experience.

[skeevy420]

Any motherboard manufacturers to avoid? If they can't fix their shit, they certainly don't deserve my money.

I've had good results with my Gigabyte B550M DS3H. There's also an AC model with built-in wifi. I opted for the wifi free edition because I have ethernet running to my PC. Only had it for around 5 months but I've been happy with it so far.

The problem you'll find if you ask that question enough is that everyone will report that every brand sucks so the best you can do is ask about what luck people are having with their current hardware.

About my only complaint is on a cold boot it can be a PITA to access the UEFI....my keyboard takes for fucking ever to initialize so I have to spam whatever I'm trying to do because, on cold boot, there's a split second where both the keyboard is initialized and the initial boot screen will take input. That's not my motherboard's fault but it's an annoyance none-the-less.

If you have an RGB keyboard that does bullshit when it initializes....well.....that might be an issue with any motherboard that initializes quickly and not just mine. I need a Goldilocks motherboard. Doesn't initialize too fast; doesn't initialize too slow; initializes just right. Or a better keyboard....

[sophisticles]

The correct way would be for both Windows and Linux to refuse to boot if the BIOS corrupts the memory. That would lead to an actual fix of the root cause....

I don't know what's sadder, that you feel it's better to not have a working computer in the event the BIOS corrupts a small 1mb portion of memory or that 7 people upvoted your comment indicating agreement.

I would think it's much more preferable for the OS to take steps to ensure that the system boots so that an end user can find a fix to the problem.

How would you guys go about fixing a system where the BIOS corrupts the ram if both Windows and Linux refused to boot?

And of course Windows does it better, there's a reason why MS is worth billions of dollars and one of the world's most valuable companies and there are maybe 4 Linux based companies that actually make a profit.

[cyring]

Hope this patch won't prevent drivers to remap those first megabyte addresses ...

[torsionbar28]

there's a reason why MS is worth billions of dollars and one of the world's most valuable companies and there are maybe 4 Linux based companies that actually make a profit.

IBM, Google, Facebook, Amazon?

[Termy]

I would think it's much more preferable for the OS to take steps to ensure that the system boots so that an end user can find a fix to the problem.

Of course it's too late now, but if both windows and linux would do it "the right way" and the system wouldn't boot, the hardware-vendor would be forced to write a proper firmware. That's the "fix" he mentioned...

[mangeek]

Windows: Doing it correctly 11 years ago what Linux only does now.

Of course this makes sense to do, but Windows hasn't been running on bazillions of devices with 8-32MB RAM for the last twenty years.

[darkoverlordofdata]

Just in time - Intel announced last year they are phasing out bios. But this is still needed for AMD and legacy hardware. You can dump bios today - just switch to ARM. The M1 looks pretty good....

[torsionbar28]

Of course it's too late now, but if both windows and linux would do it "the right way" and the system wouldn't boot, the hardware-vendor would be forced to write a proper firmware. That's the "fix" he mentioned...

Yes that would drive technical improvement of the system firmware. But it doesn't help Windows sales. Microsoft rose to desktop dominance by making it easy for the system vendors to pre-install their OS. One way of doing that is by being compatible with everything. Microsoft working around the various vendors' bios and firmware bugs, rather than kicking it back to the vendor, is a big incentive for the vendor to go with a Microsoft OS. The end result is a long list of buggy crap firmware implementations, and Windows OS being able to run without issue on all of them. So I understand why they did this. I don't like it - but I understand it.

Remember that the profit margins on PC sales are pretty slim, and the refresh cycles are fast paced. There's simply not time or budget to rework and fine tune the BIOS implementations. It's more important to get product out the door, than to have it be really polished and clean. Again, I don't like it, but I understand why they do it this way.

This is one reason I only use server / workstation grade hardware in my PC builds. Supermicro motherboards are my go-to. There's more budget in the enterprise hardware space to do things the right way, and less tolerance for buggy implementations. That's not to say they're flawless, but they are certainly better than the low-budget family peecee stuff.

[willmore]

I read the original announcement wrong, it seems. I though this change was to *eliminate* the reservation of the first MB or DRAM on x86/x86_64 (clearly not an issue on other archs). And I though "Wow, that seems like a risky bet for a tiny amount of memory. I hope this doesn't cause tons of subtle bugs due to crappy BIOS/UEFI/firmware." Glad to see I misread it. Can't believe we've been doing it this way all this time. Once memory went over 1GB, this seems like an obvious low hanging fruit of stability.