Microsoft Announces Linux-Powered "Azure Sphere" IoT Platform

[stormcrow]

How does one extinguish GPL licensed code

Patents. You can look at the code all you want, but you can't legally reuse it in the US when it's implementing a patented feature. There's no provisions for patent licensing in the GPL v2 that the Linux kernel uses.

I'm less concerned about the patenting problem than I am the privacy problem associated with anything that's pushing "The Cloud", and security of such devices. There's nothing in the articles I've read that actually solves IoT security woes. It's a lot of lip service to a "Secure OS" and pointing at "Windows' security features".

People buy IoT items because they're cheap and are "good enough" for what they want them for: baby monitors, security cameras, etc. Most people don't even know that these items should be updated, and MFGs don't encourage people to update, rather they want people buying another device. So even if Microsoft has a "forced update" feature in their IoT devices, how long will that actually last? 2 years? 3? Will they still be updating it 6 years down the line? If they don't, then assuming only 1 million devices sold each year and a 10% annual retention rate after the first two years, you've still got 100,000 devices being orphaned each year if they keep to the industry norm of a 2 year planned obsolesense for each released device. If they use a paid subscription system for this, it'll be an even worse problem because most people want stuff cheap on a one time charge. You might get a few people to pony up, but most will not, or not for the long haul.

Assuming their IoT firmware uses patented features like exFAT, no one else can support them without risk of patent infringement. So far the only royalty free offers I've seen are for the hardware designs for manufacturers, not software.

[stormcrow]

Unless Microsoft uses strong key-only communications between a client and device, it won't solve the problem of weak and default passwords on such devices, either.

[jacob]

It reminds me the good old time of Hotmail story where when Microsoft decided to move the service on windows, it crashed so bad that they had to go back to Linux...

IIRC it was FreeBSD, not Linux. But don't forget that was back in the NT 4.x days. Today's Windows 2012/2016 are actually a capable and reliable server OS.

[jacob]

Microsoft has actually leap-frogged Linux in various ways (The one that people tend to think of most is the ability to recover from a graphics driver crash or change graphics drivers without a restart) and has put quite a bit of security hardening into the Windows kernel. (Partly to compensate for all of the legacy API baggage that's specific to Win32 APIs.)

I don't know if I would put the gfx driver issue under the security umbrella but yes, it's definitely one advantage of the Windows kernel over Linux. The saddest part is that apparently even Wayland doesn't allow it. But from a strict security point of view, all I can think of is that MS decided to use ACLs from the start, whereas Linux carried over the moronic "UGO" permissions model from Unix (and even today, its ACL support still sucks). It's obviously not a "Windows security innovation" though, ACLs have been used all the way back to MULTICS.

Other than that, I don't see any security feature in Windows that Linux doesn't already do better. Its hardening possibilities go way beyond what most other OSes, particularly Windows, can offer.

[stormcrow]

IIRC it was FreeBSD, not Linux. But don't forget that was back in the NT 4.x days. Today's Windows 2012/2016 are actually a capable and reliable server OS.

Mix of FreeBSD and Solaris. Windows NT v4.x couldn't manage it at the time. Microsoft's executives wanted a major success story to buck their (deserved) reputation for being a toy software company with the Big Iron crowd. What they got was unmistakable proof that they really weren't in the same league as the work horse (Unix/IBM) OSes even on the same hardware. It took some years and a new release (Server 2000) before NT Server was even close to being able to handle Hotmail's compute loads.

[jacob]

Mix of FreeBSD and Solaris. Windows NT v4.x couldn't manage it at the time. Microsoft's executives wanted a major success story to buck their (deserved) reputation for being a toy software company with the Big Iron crowd. What they got was unmistakable proof that they really weren't in the same league as the work horse (Unix/IBM) OSes even on the same hardware. It took some years and a new release (Server 2000) before NT Server was even close to being able to handle Hotmail's compute loads.

It was during that period that Microsoft also ran its "Get The Facts" propaganda about how much better Windows NT was than Linux. It all came to a crashing halt when someone noticed that the Get The Facts website itself ran on... Linux. That must have been the Microsoft FUD machine's finest moment.

[jacob]

Make sure you agree to Microsoft's vision of what democracy is. Microsoft world police and champions of social justice.

By clicking on OK you agree with the terms of Microsoft Democracy 2018 EULA, including but not limited to, forfeiting your rights to vote, run for an elective office, protest or form an opposition party.

[RomuloP]

I don't know if I would put the gfx driver issue under the security umbrella but yes, it's definitely one advantage of the Windows kernel over Linux. The saddest part is that apparently even Wayland doesn't allow it. But from a strict security point of view, all I can think of is that MS decided to use ACLs from the start, whereas Linux carried over the moronic "UGO" permissions model from Unix (and even today, its ACL support still sucks). It's obviously not a "Windows security innovation" though, ACLs have been used all the way back to MULTICS.

Other than that, I don't see any security feature in Windows that Linux doesn't already do better. Its hardening possibilities go way beyond what most other OSes, particularly Windows, can offer.

If they really had pioneered it in Windows I believe it should have been a Windows kernel... Honestly they still don't have anything similar to seccomp for example... Anyway, the ability of recovering the GPU drive is great on desktops, this is true, but not on servers, you don't want it to reset your driver because as intentioned a CUDA program dominates your GPU.

[stormcrow]

It was during that period that Microsoft also ran its "Get The Facts" propaganda about how much better Windows NT was than Linux. It all came to a crashing halt when someone noticed that the Get The Facts website itself ran on... Linux. That must have been the Microsoft FUD machine's finest moment.

Yeah, I remember that campaign. I also remember the Red Hat parody of Window's need for frequent restarts (including NT) with the mouse has moved please restart your computer ad. Microsoft generally deserved most of the vitriol directed at it in those days, and there's plenty of that ill will still lingering. I'm cautiously optimistic about Microsoft's embrace of open source, but I also know a Leopard won't change is spots in a single generation. Microsoft has a horde of software patents it can use to cause a huge amount of hurt to open source efforts. The fact that they haven't made the patented exFAT royalty free (recently adopted by SDXC standards for any card over 32 GB) is worrying.

[stormcrow]

If they really had pioneered it in Windows I believe it should have been a Windows kernel... Honestly they still don't have anything similar to seccomp for example... Anyway, the ability of recovering the GPU drive is great on desktops, this is true, but not on servers, you don't want it to reset your driver because as intentioned a CUDA program dominates your GPU.

Uh, reasonably sure you'd want it on servers too, because if the *driver* has crashed anything running on those driver managed GPUs is likely corrupted as well. This is why you save your computational state as frequently as performance limits allow. It'll let you pick up where you were from before the crash, assuming something in your code isn't what's causing the crash. What you *don't* want is a driver crash bringing the whole system down which has been the traditional error state in Windows for most of its lifetime.