10 ms·
One of the major major major issues is the hardware itself. Let's jump back to the late 90s. You want to try Linux. You might partition your disk, or build a d
by djsumdog 10y ago
One of the major major major issues is the hardware itself.
Let's jump back to the late 90s. You want to try Linux. You might partition your disk, or build a different machine and get a KVM or use an old laptop or add a second hard drive. In all cases, you can just install a Linux distro and see if it works. Maybe Ethernet or sound or something doesn't work. You could try a different distro with a newer kernel, maybe compile your own kernel with the modules you need, etc. But no matter what, it would at least boot (usually).
Intel x86/64 is a standard architecture. EFI makes the booting process even easier (although some motherboards still fuck this up; but for the most part fiddling with efibootmgr and turning off TPM will work. If not you can always turn on legacy BIOS and use the MBR). ARM is not a standard architecture. It's a spec that is sold to companies who all make their own SoC templates. There are things like Device Tree Config that make some of this more standard, but it's still far from x86/64. Just read Torvalds rants on ARM to learn more.
Even if you make some standard Android builds with all the drivers for all phones in a massive 20MB initrd, a lot of kernels are fundamentally different with the way bus connections are defined. Manufactures take forever to release their kernel code, often under threats by people for GPL violations, and even then they are filled with tons of binary blobs and shims that connect their .o files so kernel calls.
Stuff like Plasma looks amazing. It only has like two phones it supports. If you look at any android rom project, you see totally different builds for every single device. If Intel had been like this, Linux would have died off in the 90s, for both servers and desktops.
Device manufactures depend on you buying new phones. They cannot support phones after two years or else it kills their profit margins. Planned obsolescence like this leads to waste (don't kid yourself, e-waste recycling gets shipping to Africa where kids remove components. A small percentage of phones actually get recycled).
TL;DR Mobile phones are based on unmaintainable and non-standard hardware (intentionally). I wrote a post on this a while back:
http://penguindreams.org/blog/android-fragmentation/ http://penguindreams.org/blog/android-fragmentation/
- mtgx 10y ago> If Intel had been like this, Linux would have died off in the 90s, for both servers and desktops. That's the thing. You said "Intel". Intel is not an architecture. It's a company - that happens to dominate the x86 market. Do you think the situation would be the same if there were 10 different x86 chip companies with at least 5% market share each? If we wanted everything to just work, as it does on "Intel", then we should probably wish for a Qualcomm monopoly with 90% of the market. And hope it's profitable enough that they can support their own chips for more than 18 months. I can only describe Intel's chips as a rip-off for the most part (and it's only gotten worse in recent years). I mean, they're now selling Atom chips for $160 (about 4-5x more than a similar performance ARM chip) to notebook manufacturers, with very little difference in performance between them and the older Atom for smartphones. So I would certainly hope Intel's chips would be better supported. Although I actually don't think its Atom chips are as well supported as its Core chips. Now, not having a single company completely dominate the ARM ecosystem is certainly a "problem" from this point of view. However, I still mostly blame Google, who mainly just wants its ads to run well on these devices, which seem to do that rather well in the current situation, as well as OEMs who want to be "different" and are probably even hostile to the idea of allowing say an LG ROM run on a Samsung phone. I would try to fix this Google/OEM situation first and aim my complaints at them.
- AlphaSite 10y agoWell, there are 3.
- djsumdog 10y agoIntel, AMD and... VIA?
- nickpsecurity 10y agoVIA, thanks to Centaur, was first with the low power plus had PadLock Engine with TRNG and crypto acceleration. Their ARTIGO's made nice little boxes for $300.
- userbinator 10y agoDo you think the situation would be the same if there were 10 different x86 chip companies with at least 5% market share each? In the 90s, that's actually what happened. https://en.wikipedia.org/wiki/List_of_x86_manufacturers https://en.wikipedia.org/wiki/List_of_x86_manufacturers
- wtallis 10y agoTracing the history of those various competitors is really interesting. Some (Transmeta) failed completely in a relatively straightforward manner and just disappeared. At the opposite end of the spectrum, NexGen became AMD's future when AMD's in-house design was a flop. Parts of Cyrix ended up with VIA and parts with AMD. Almost all of the big semiconductor companies either owned one of these x86 design firms for a time, or sold those chips under license.
- pcwalton 10y agoI can't think of any reasonable definition of "standard" that admits x86 but not ARM. They're equally closed, proprietary ISAs. It's the IBM PC architecture, not x86, that is the de facto standard. It's reasonable to prefer that to the relatively recent, and poorly adopted, device tree stuff that you see in the embedded ARM world. But let's not ding the ARM CPUs themselves for that.
- oneplane 10y agoThat's just the thing, there is no 'ARM CPU'. There is a ARM ISA, and some Core spec's, not a real 'ARM CPU'. Maybe a complete ARM SoC design you can buy, but nothing like an interchangable CPU, let alone socketed with multiple vendors. If it's ARM-based, it's probably an embedded, closed, proprietary system with zero design need for 'compatibility'. With x86-based boards, you know there will at least be some sort of standard BIOS of EFI interface, a set of busses, a minimal number of devices that will behave and work the same on almost all systems. With ARM, there is no such thing, except for most parts of the ISA, and that's simply not enough.
- pcwalton 10y agox86 chips are themselves almost complete SoCs at this point [1]. That's why you can buy things like Compute Sticks, after all. It's true that Intel is by far the largest supplier of x86 chips, and so you get some level of "standardization" with things like Intel HD Audio. But that's more the effect of a monopoly than anything else. Competition breeds fragmentation, but given the choice I'd rather have competition. [1]: https://www.techpowerup.com/img/15-08-18/77a.jpg https://www.techpowerup.com/img/15-08-18/77a.jpg
- oneplane 10y agoWhile true, it wasn't always that way. The IBM PC with the BIOS was basicaly one of the first layers you'd have to be compatible with, and up to UEFI systems with CSM for BIOS emulation, it's been supported for about 30 years. There was a time where the chipset did a lot more than it does now. Right now, it's the PCH or whatever it's called, and you even have those SoC's where you don't even need that anymore. You used to have the possibility of a standard BIOS with a chipset from someone else (SiS for example) and audio from someone else, as well as PCI controllers, PICs, RTCs, HBAs, UARTs, and CPU's. Or have an Intel chipset with a non-Intel CPU, also possible. You could even make a complete x86 system with no Intel chip inside (pun intended). Right now, that's something you probably can only do with an AMD system and that's about it.
- zzzcpan 10y agoOr maybe the problem is in Linux itself not being well suited to run manufacturers' blobs independently from the rest of the kernel?
- subway 10y agoIt's far worse than that. If it were simply a matter of running mfg's blob's, this would be an easy problem. Instead manufacturers fork the kernel, then patch the hell out of it (often in horrific-never-going-to-be-accepted-upstream ways).
- euyyn 10y agoI'm not familiar with any of the specific reasons why manufacturers fork the kernel. But an obvious question to be addressed is: Can the kernel be modified, so they can achieve their use cases without needing to fork it? By making it more modular in some parts, or adding hooks they can attach to...
- 91bananas 10y agoSame wonder here, what are the reasons they do this?
- TD-Linux 10y agoIt's quicker and easier than submitting any patches upstream. They don't care about support past product release. Usually the SoC vendor provides the forked version, and the manufacturer forks that again, so they are twice removed from upstream. If they ever upgrade Android versions, they port all of their patches onto a new fork from the SoC vendor.
- euyyn 10y agoThe question is not "why don't they upstream the changes". Rather why do they have to fork to achieve what they need.
- ocdtrekkie 10y agoHow has Windows Mobile solved this? Dev preview is available for all phones, updates are pushed immediately from Microsoft regardless of the manufacturer. Firmware with drivers are released separately from the OS. All on the same ARM architecture.
- my123 10y agoWindows on ARM uses UEFI and ACPI, it can even boot with zero drivers on Cortex A15-class(architectural timer) and later HW. Of course, you only have USB and eMMC in such a config, the driver model is also stable.
- golf1052 10y agoMaybe because they require hardware manufacturers to use a known set of processors (see https://msdn.microsoft.com/en-us/library/windows/hardware/dn915086(v=vs.85).aspx#section_2.0_-_minimum_hardware_requirements_for_windows_10_mobile https://msdn.microsoft.com/en-us/library/windows/hardware/dn...)
- my123 10y agoGetting Windows go boot on the completely fictuous qemu -M virt platform for both arm and arm64 was very easy, USB and SCSI worked after less than a week of fixing Qemu emulation bugs (GIC).
- oneplane 10y agoYou're basically patching the hardware to work with immutable software (as you won't have the windows sources). At the same time, the QEMU ARM virtual system and the Microsoft designated ARM SoCs are perhaps closely compatible whereas it would be very different with other ARM implementations.
- my123 10y agoOnly GIC (interrupt controller) bugs needed to be fixed. The GIC is standard since the Cortex A9. On the A9, you have to use an HAL extension (declared in ACPI) for the timer(a sample is available on MS' Git repo). On the A7/A15 and later, Windows uses the GTDT ACPI table only, so nothing is needed :)
- 0xFFC 10y ago>Device manufactures depend on you buying new phones. They cannot support phones after two years or else it kills their profit margins. This is the reason after using Nexus devices for 5 year, I will buy an iPhone and I will never ever look at Android again. My girlfriend switched from Galaxy note 4 to iPhone last year, and she is amazed how much iPhone experience is better than high-end Android experience. 2 year only update? Give me a fucking break, this total rip off. If I remember correctly even iPhone 5 does get update(and remember iPhone 5 is 5 year old phone, which will get new version of OS, not just security update, this alone is more than lifespan of Nexus or pixel device). Let alone the fact phone processors passed the threshold, phone produced after 2015 are almost as powerful as desktop from ~2005. My point is , if you don't play intensive games or any specific use case, an above ordinary phone would run browser,mail client, chat apps for more than 4 year without any issue. But companies trying to minimize lifespan just because of their own profit. And my answer to them would be my middle finger. They are totally screwing up customers.
- eon1 10y ago"..remember iPhone 5 is 5 year old phone, which will get new version of OS, not just security update" Yes, and remember that Apple intentionally writes OS updates that will run slower on their older devices, thus achieving almost the same thing but slightly less blatantly.
- rrdharan 10y agoIt maybe the case that OS updates run slower on older devices but I think it's a pretty strong statement to claim that this is "intentional", at least in the sense that you seem to be implying. Do you have some strong evidence of this? Amusingly it's actually being tested in court (see the class action referred to in the link below): http://www.cbronline.com/news/mobility/devices/apple-accused-of-sabotaging-old-iphones-with-upgrades-but-how-can-users-protect-them-4944274 http://www.cbronline.com/news/mobility/devices/apple-accused...
- macintux 10y agoSome updates actually make older devices go faster, anecdotally.
- niftich 10y agoYour points are valid but this clusterfuck doesn't excuse Android: both Microsoft and Apple were able to deliver working ARM-based systems not subject to these issues: Microsoft by mandating UEFI, Apple by reasons unknown to me, but presumably helped by the fact that they're a vertical. As a few people have noted, it's not "Intel x86/64" that's standard, they just happen to be a dominant player right now. The original architecture is the "IBM PC", and its clones that are "IBM PC-compatible", that all descend from the same BIOS lineage. IBM became dismayed how other companies were profiting from its architecture, and released the incompatible PS/2 architecture instead [1], which fizzled, and its good parts -- like 3.5" floppies, VGA, and PS/2 connectors -- were adopted by the clones. In the 1990s, Intel and Microsoft rose to prominence and began to exert more control and dictate the direction of the architecture with specs like 'PC System Design Guide' [2], and they and other vendors produced several improvements (ACPI, USB, PCIe, AHCI, SATA, etc.) which were adopted by broad community consensus, pushing the architecture forward. ARM is different, because they make their money licensing their cores, so historically they had no interest in cultivating a general-purpose, "vendor consensus" platform. And Android's owner, Google, just wanted to get its services like Google Search, Google Maps, Google Mail, Google Now into as many people's hands and pockets as they could -- and they succeeded, despite the platform's haphazard nature. Samsung knows that Google and the handset manufacturers are allies of convenience. Samsung is hard at work trying to cook up an OS and ecosystem for its hardware that's not contingent on Google's continual involvement, even allying with Microsoft to bring .NET to Tizen after its previous attempts saw no developer uptake. Meanwhile, Google knows that continued press coverage of Android's security situation is hurting the platform. It's trying to re-frame the partnership it has with hardware makers, to design more hardware in-house, and to lean on its partners only for contract manufacturing -- a shift from the Nexus days, incompletely executed with the first Pixel phone, but likely more to follow. This will enable Google to control more of the software on the phone. [1] http://www.mcamafia.de/pdf/pdfref.htm http://www.mcamafia.de/pdf/pdfref.htm [2] http://tech-insider.org/windows/research/1997/0711.html http://tech-insider.org/windows/research/1997/0711.html
- bitmapbrother 10y ago>Your points are valid but this clusterfuck doesn't excuse Android: both Microsoft and Apple were able to deliver working ARM-based systems not subject to these issues. Then how do you explain all of the Windows Phones, from Microsoft and their OEM partners, that were stuck on the previous versions of the OS that are receiving no updates whatsoever? >Samsung is hard at work trying to cook up an OS and ecosystem for its hardware that's not contingent on Google's continual involvement, even allying with Microsoft to bring .NET to Tizen after its previous attempts saw no developer uptake. Let's see, if Microsoft wasn't able to make .NET work on their phones what makes you think Samsung has a chance with a platform that has even less marketshare and mindshare? Samsung is just doing what Samsung always does. Incidentally, Google also joined the .NET foundation for what it's worth.
- avar 10y ago> Let's jump back to the late 90s. You want to > try Linux [...] get a KVM [...]. KVM as in Kernel-based Virtual Machine? Don't you have the late 90s confused with the late 00s? The first Linux release with KVM was in 2007, KVM came out of an Israeli startup around 2005.
- jupiter2 10y agoLate response: I'm having a hard time understanding your reasoning here. A less generic comparison is Apple. Apple builds it's own devices and supports them. If I buy an ASUS Tablet, I'm not expecting them to provide security updates to Lenova or Samsung Tablets. I am expecting them to support their own devices. They have everything they need - they built the devices themselves. Similarly, if I buy an phone that's been purposely tweaked by AT&T, working with Samsung - I expect that they are more than able to provide security updates but are simply unwilling to do so. They have all the tools (drivers, access, etc) they need. You're looking at the general Android landscape. It's a question of holding the device makers accountable. Providing security updates for devices you build yourself, with drivers you developed yourself (or through a partner) isn't an issue - it's pure (litigable) greed.