QNX: The PlayBook OS

A year and a half ago RIM jumped on the opportunity to purchase an OS vendor called QNX (pronounced Que-NIX, like UNIX with a Q in front). RIM saw a growing gap between its BlackBerry OS and what Apple/Google were working on and realized it needed something new to effectively keep up with the joneses. The age old debate between build vs. buy kicked in to high gear and RIM took the route with less internal risk: acquire QNX.

QNX's Neutrino 6.5 OS is the basis of the BlackBerry Tablet OS running on the PlayBook. QNX features a very small (by modern standards) microkernel of around 100K lines of code. By comparison the modern day Linux kernel is around 14M lines of code. QNX argues that by limiting the scope of the kernel it can ensure greater stability and less vulnerability to bugs in the code. QNX is functionally a modern OS, however everything beyond the base kernel is supplied in the form of separate, self contained services. Even drivers are excluded from the microkernel.

QNX is, as a result, a very modular operating system. Additional features are added simply by bundling extra services, however for specific markets you don't incur any complexity or security penalty as unneeded services are simply turned off.

Because it's so modular, QNX has no issues being used in everything from the PlayBook to high end Cisco routers. Obviously the requirements of a tablet are very different from a router, so RIM actually implemented an updated version of QNX 6.5 into the PlayBook with some tablet specific features. The QNX team worked on improving media playback and GPU performance in QNX, features that will eventually make their way into QNX 6.6.

Another benefit of the modularity of QNX is that each service runs effectively sandboxed. A crash within a single service won't bring down the whole OS. Restarting a single crashed service can be done quickly. A small kernel should also be able to boot quicker, however the PlayBook itself has a longer boot time than Honeycomb because the entire OS image is validated using a crytographic hash on boot up. As a result of this image validation, the PlayBook will always boot into a known secure environment. Should the image validation fail, the PlayBook will typically install a previous known-good copy of the kernel stored on a separate partition.

Communication between services/processes happens via a structured messaging system. This is of course one of the benefits of a microkernel OS: inter process communication is usually quite good, because it has to be.

In QNX every task or thread has a priority. While a handful of priorities are reserved for system level events (to prevent an app from taking priority over refreshing the screen for example), everything else is defined by the caller. QNX argues that unlike in monolithic non-realtime OSes, task/thread priorities are nearly always respected here. Inevitably you'll have a number of tasks that have the same priority, and there the scheduler will just round robin between them. The one guarantee QNX offers is that any task with a higher priority will execute in accordance with that priority. This is ultimately what makes QNX a realtime OS.

Large monolithic OSes often give you the same promises, however QNX argues that they don't always hold true to them. There's always some system process that's interrupting things or a runaway task that prevents your screen from updating as quickly as you need it to. These days with hefty multi-core CPU architectures, hiding scheduling latency due to a system process interrupting something else isn't too difficult. On tablets/smartphones it's more of a problem given limited CPU resources, but ultimately it'll diminish there as well. There is arguably a power efficiency benefit here but at a high level that's a difficult thing to measure.

The QNX OS itself has been ported to everything from x86 to PowerPC. Although the PlayBook uses an ARM based OMAP 4430 today, it looks like RIM is pretty open to moving to other microarchitectures should the need arise.

File System

On the PlayBook RIM implements the latest version of the QNX file system. The setup supports 64-bit LBAs (effectively no limit on single file size given current NAND capacities) and 1K file names.

File system performance is difficult to measure at this point given that we're dealing with pretty low performance NAND as well in devices like the PlayBook.

A Functional Bezel TI's OMAP 4430
POST A COMMENT

78 Comments

View All Comments

  • jjj - Wednesday, April 13, 2011 - link

    In the final words it would be worth reminding readers that it has no SD card slot, IMO a fundamental feature for phones/tablets nowdays. Reply
  • Azethoth - Thursday, April 14, 2011 - link

    That is a curious statement. What do you want lots of memory for?

    I can see memory being better for a phone that you listen to music from as more memory = more of your (compressed) library can fit on it. Personally I only sync particular playlists to my phone / iPad anyway.

    As for other stuff, well apps just do not consume a large amount of space. For my iPad 2 I went with the smallest memory size. The larger size I have on my original, er I mean on my sister's "new" iPad, was just a waste for me.
    Reply
  • Chloiber - Thursday, April 14, 2011 - link

    Because it's an easy way to share things and upgrade your memory if you need more. I won't pay 100$ for 16GB of NAND flash (which cost's like 15$). Reply
  • jjj - Thursday, April 14, 2011 - link

    For a device that can shoot and play 1080p video 16GB-64GB of storage is very little (and anything above 16GB is way too costly) Then there are also photos,music,apps that maybe soon will be actually able to do things and become bigger,it is after all a computing device and even if smartphones/tablets are in their infancy we can still hope that they mature sooner rather than later. Reply
  • BuffyzDead - Thursday, April 14, 2011 - link

    "The screen is too small to comfortably read in portrait mode and even in landscape things can get a bit cramped." Steve Jobs already warned the entire industry on this point. It's too small to be a successful tablet. Everyone has portability in their Smartphone.

    "Apple's A5 still has a much faster GPU" .and the playbook isn't even shipping yet

    "App launches are unfortunately a bit high latency. .....whole process takes a couple of seconds but it feels longer than firing up similar iOS or Honeycomb apps." just throw more CPU at the problem. You know, down the road.

    "With no email or calendar apps, the PlayBook doesn't have a whole lot to notify you of. Presently the only notifications the PlayBook will deliver have to do with remaining battery capacity." LOL at this one.

    I predict Now, this thing will never sell in volume. Even improved versions down the road won't sell.
    3 reasons:
    1) It's too small.
    2) The User Experience does not even come close to that of the iPad1
    3) NO APPS

    Yes, while there are company's that may force this down their employees throat, that is not where the growth of tablet use is coming from, in enterprise.

    It's coming 100% from employees wanting to use their iPad's in the work environment.

    You finish with:
    "there are still more revolutions that will take place between now and when the mobile market finally matures"

    Ask yourself, Honestly, is there ANYTHING about this "experiment for RIM"
    that has an inkling of REVOLUTION ????
    Reply
  • Azethoth - Thursday, April 14, 2011 - link

    The proper quote is:
    "While Apple and Google are clearly out to a substantial lead, there are still more revolutions that will take place between now and when the mobile market finally matures. I'm not saying that Apple or Google won't end up on top, I'm just saying that it's not guaranteed they will either."
    and
    "The PlayBook is a reasonable experiment for RIM, but I need to see more to really recommend the tablet."

    See how in context Anand makes sense? Rather than claiming the ridiculous: "this RIM tablet is a revolution", Anand is merely saying that this is an immature industry. Everyone fully expects actual game changing revolutions in this area in the future.

    Search this site for Anand's excellent follow up to the iPad 2 release that asks: "How do I as a blogger use a tablet to create [text] content". There were some responses about maybe covers that double as keyboards. There was wishful thinking about voice input maturing real soon now. Mostly there is a need for some kind of interface revolution before a tablet can become a reasonable answer for a blogger on the go. These are the revolutions Anand needs in a tablet.
    Reply
  • BuffyzDead - Thursday, April 14, 2011 - link

    Correct !
    I wholeheartedly agree with Anand's review and conclusion:
    There are no guarantee's what the future may hold and he can not recommend this Playbook.

    1) I wanted to add that this Playbook will in fact be a total failure. Time will prove me right or wrong. Again, I predict Total Failure.

    2) I wanted to point out how RIM, as is evident by this Playbook version 1,
    has demonstrated Zero in it's capability to provide ANY REVOLUTION.
    At most, it's an outright attempt to copy or emulate (poorly) what the iPad REVOLUTION IS.

    BTW,
    If you think that "designing a tablet" to cater to "the blogger on the go" is a measure of success, then I pray for RIM's sake, you are not on their design team.

    Anand has repeatedly pointed out how a tablet might just not be for him, in general.
    I have maintained for the past 15 months that the iPad's true REVOLUTION, is that it CREATED a NEW MARKET.

    ALL of Apples competitors are playing catchup & copycat to cater to that NEW MARKET, which the iPad CREATED.
    Reply
  • SandmanWN - Thursday, April 14, 2011 - link

    For real... lay off the coffee and Jobs shlong. Reply
  • melgross - Thursday, April 14, 2011 - link

    For real, it's the only tablet so far that sold more than a small number. As many others have pointed out, so far the tablet market is really the iPad market. Other manufacturers have to prove that they can sell a large number of devices. We know that the Tab, is not a real tablet by Google's standards, and that it sold in much smaller numbers than the number shipped to retailers and cell companies. The Xoom is assumed to have managed about 100,000 sales, and what else has there been that seriously competes?

    Now, the Playbook, which has been criticized by those in the industry for having poor battery life, and problems with the software before release, despite RIM,s denials, is proving, from all the reviews I've now read today, to be having all of those problems just days before release. Pogue has stated that RIM is feverishly sending out updated on a daily basis. That's not good.

    Other tablets won't arrive for at least a couple more months.

    So what does the market really consist of now?
    Reply
  • melgross - Thursday, April 14, 2011 - link

    It's hard to say. There are a lot of writers who have said that they type just fine on the iPad's virtual keyboard. Everyone's different. I remember a lot of people complaining about the first iPhone's keyboard, but since then, many, if not most new smartphones have no physical keyboards anymore, so people are getting used to them.

    I'm typing on my iPad2 now. The only complaints about the way Apple set it up are that I think that too much space is wasted on the two large numeric keyboard call up keys at the bottom of the keyboard, much of which could have been used for other functions, such as an "@" key, for instance. Otherwise, it's fine. On a 7" screen, typing for longer periods will be more problematical.
    Reply

Log in

Don't have an account? Sign up now