Hit the Open Road with Automotive Grade Linux – Walt Miner, The Linux Foundation

Hit the Open Road with Automotive Grade Linux – Walt Miner, The Linux Foundation


all right good morning everybody my name is Walt miner I am the community and development manager for automotive-grade Linux you can follow me on twitter at at v– star walt you’ll see a picture of everybody here in the room you could pick yourselves out I’ve been with the Linux Foundation and automotive-grade linux about three plus years now and my most of my message today is about collaboration AGL I think you’ll find is one of the friendliest projects for working on with developers we have a lot of developers who pitch in and help out you’ll you’ll find that we were always more than happy to help people come onto the project and we really love working together we had a an all number meeting last week in Dresden about two hours away from here by train we had about 150 developers come together everybody got a lot of work done had a lot of really good meetings and really helped advance the project so that’s really the the message of today our tagline is collaborating to build the car of the future through rapid innovation and hope to show you some of what that’s all about so automotive-grade linux we’re nonprofit or an open source project based in the with with the linux foundation itself and we’re really working on rapid innovation of vehicle software throughout the vehicle we’re working to build a single platform for the entire industry we’re an OEM you know someone who makes cars or a tier 1 supplier to that OEM can take a GL and have about 70 to 80 percent of a starting point for a production project our mantra is code first you know as with any open source project commit early commits often and really we aim to reduce fragmentation in the automotive industry by combining the best of various open-source projects that have been formed to work in the automotive industry over the last 10 to 15 years we’re working to develop an ecosystem of course of developers suppliers and expertise all using that single platform like I said AGL is really the the only industry the only only organization trying to address all the software in the car Dan caoxi who’s the executive director of automotive grade Linux likes to say if it’s Linux in the car it should be based on a GL where we’ve been focusing the last few years on entertainment and ivi systems for the vehicle we’re branching out into instrument cluster telematics heads-up display we’re putting effort into functional safety over the next year to get the gets Asil B and ISO to six to plus qualified so we can eventually add advanced driver assistance or a TAS Private Eye products and autonomous driving to our portfolio and like I said AGL is a code first organization so we don’t spend a lot of time on specifications but that doesn’t mean we don’t spend a lot of time on documentation but specifications as we’ve seen another in the automotive industry initiative specifications have led to fragmentation as different companies end up implementing those those specifications in different ways and there’s been numerous examples of this in the automotive industry through the years so just to kind of give some examples of collaboration within the project this is the list of the top 25 get committers so far this year on an automotive grade Linux I really ran these statistics on October 1st so we’ve had a total of two thousand seven hundred sixty five total commits this year to a GL repositories on our master branch this doesn’t include our release branches that we’ve had 60 individual committers have commit code this year 27 individual companies have committed code this year and you can see some people take our committer aleikum it off and mantra to heart and they really do they really do that others who may have fewer commits on this list are still contributing a great deal of code they just don’t tend to commit as as often as other people by company I listed every company that’s committed this year you can see it’s a pretty wide variety of companies that have committed to the project both tier ones tier two Zoe M’s is fees some other numbers over the last we’ve been running the AGA GL what we call the unified code base project for the last two plus years now in that two and a half years or two plus years we’ve had 81 individuals unique individuals commit code 33 different companies and interestingly seven individuals not affiliated with a member company have committed code as well so people who just saw what we were doing and wanted to contribute saw something that they they wanted to help out on and then there’s another measure this year and this number is actually a little little stale because we’ve done a release since I did this but this year we’ve closed 429 individual JIRA issues so those we use JIRA for project management and for bug tracking so that’s a mix of defects new features improvements things like that that we’ve tracked throughout this year so we’re making a lot of really good progress we release code twice a year we make more releases twice a year our members didn’t like when I was just calling it you know version 1 version 2 they insisted that I come up with some kind of naming system so after a lot of thought I put I came up a fish I spent a lot of time researching fish in alphabetical order and and of course appropriate adjectives this is a lot of my time spent on this so our next group we had our last release was a daring dab that was in July we’ve a few patch releases since then our upcoming release in December for that’s targeting the Consumer Electronics Show is going to be called electric eel and you can see we released the daring dab release in July and since then we’ve done two patch releases over one oh no – we apply it on at least two more through this year and into next year after CES electric eel we’re targeting for the middle of the middle of December the schedule and I’ve got a lot I uploaded the slides earlier today so you can go to the the schedule the schedule webpage and you can get a you get all the slides if you’d like I have all hyperlinks to – a lot of the different topics on here so you can see on our wiki page I keep the schedule up to date I tried to update it every couple weeks as new things happen and you’ll find more much more detail than this you’ll find that you know the opening dates and closing dates for our merge windows and things like that so I just wanted to give a quick overview of what we’re working on in terms of the architecture we hgl is we have this thing we’re calling it we call the unified code base we’ve got these releases twice a year we’re based on the Yocto project so with every octo release we incorporate their release about three months three months after they make their initial release we switch over our master branch to using their release and then we take that and we spend about six months working on that adding new features stabilizing the bsp etc so that our release then is is about a nine month old up pocky release from Yocto project and you’ll see here we we get the the octo project includes our our Linux kernel and BSP and device drivers and some of the basic services then we add we add this with these this reference code here in blue so things like audio control sound sound vehicle to cloud we have an AP of an HDL application framework that includes a service binder concept that I’ll get into in a second and we also include security and some transport mechanisms AGL also includes some reference applications up here that you can use to start playing with to create your own apps and what I find whenever I go to a show like this or to the AGL or member meeting last week is that companies come to me with all these really cool applications that I hadn’t even thought of and they’re often using all of this all these bits and taking these reference applications as something to start off with and they go off and they create something pretty cool with it so there were some really good examples of that once again at in Dresden last week of people using using a GL to create STL applications or smart device link all sorts of so all sorts of audio applications that we’ve seen coming out of out of smaller companies so it’s really interesting to see what people do with this thing so we have this binder binding functionality built into our or a GL application framework you see here that each of the binders that are interconnecting are connected to each other and there’s security built in there’s a security concept there’s a security context built into each binder so that your applications as they call these api’s then we ensure that only the applications that are allowed to use those api’s or other services that are allowed to use those api’s can do that our security context our security is all built on smack and a lot of that work has was based on a lot of the application framework work was based on the Tizen application framework and a lot of the guys who work at on that Tizen application framework are now working on the AGL application framework and using a lot of the lessons learned there too help improve the AGL app framework of the AGL binder binding mechanism so again I included a link we have a really good documentation site Doc’s that automotive dot Linux org so I included a link here to the application framework binder information so you can learn how to build your own also this week I don’t recall if it’s today or tomorrow Scott Murray and Matt Porter will have a talk about building your own service binders in AGL and some of the experience that they’ve had in building the Bluetooth and audio and radio tuner binders so there’s some really good information available there as well one thing I should point out about this is when I talk about cool applications that people are doing Lee on an IV from consult codes also giving a talk later today I guess he’s built a robotic controller using Raspberry Pi based on the AGL app framework in the AGL UCB so that’s another interesting talk you can learn where people have actually made use of a GL we’ve also got an SDK available for you app developers or people who want to develop new services on a GL here’s a link here to the documentation and how to how to download the app the SDK and get it up and running it’s basically it’s a docker image we’re trying to eliminate hosts dependency issues though still tend to come up but with our with our platform developers who aren’t using the SDK what we had found early in a GL was a lot of UI excuse me a lot of Yocto knowledge was needed for app developers and we really wanted to eliminate the need for app developers to have to learn a lot of Yocto and allow a lot about the build system so what was our so I just I presented this to our advisory board last week we our roadmap for this year just want to run through the roadmap for this year what we accomplished and what we’re looking to accomplish next year and really the thing to keep in mind is for anything we’re looking to accomplish coming up this year next year and Beyond hgl has you know people we have people that were funding for development but also a lot of people come to us and work on projects that they’re just interested in because Lee I think leanest or volt said you know the best way to get involved in an open-source project is to find a niche that you need to scratch and so really Eddie you know we’ve got we’ve had people add code to the to the distribution based on something that they wanted to work on so by no means are we limited to just doing these kinds of things in the next year or two so like I said we’ve released our – we had our daring dab release in July we really planned to release and we’re on track to release electric eel for CES the our vision of the beginning of the year was we would have the HDL apt frame or complete it’s really hard to you know it’s really hard to quantify to complete so you know we’re on basically we’re on track with the AGL the framework is usable we’ve got people developing apps we’ve got people using the SDK so I’ll call that a win for this year we want to have a GL reference apps available in both cute five which is I’ve been our primary target and html5 we’re a little behind on the html5 side it’s looking more like that’ll be the beginning of next year all API is available as app framework service binders well again how do you quantify all I’ll say that we’re in progress we’ve got a good number that are already available and a good number that are currently being developed and we should see all of them done pretty soon instrument cluster telematics other profiles like I said we want to be in all parts of the vehicle so we’ve had to do some rework of our our build system of our Yocto layers to enable headless profiles in order to do some more enablement of lower end radios and things like that so the profiles are ready to go they’re in our build system there’s actually of a specific JIRA issue spec 145 you can follow along all of the trials and tribulations of the group working on that they’re basically ready to go and will be turned on and validated more after CES at this point we don’t want to risk the EE release with fooling around with the build system or at this point in time a complete set of documentation well again how do you quantify complete I’d say we have our newly launched the dock site at the end very end of last year the documentation there is it’s in progress it’s always evolving we need to do a better job of versioning the documentation for the various releases but I think we’ve got a pretty good start there we also wanted to do more continuous integration automated testing daily weekly event based builds for testing and publishing test results so we’ve got a lot of activity in our continuous integration automated test group if you go we use garrett i’ve got a lot i’ve got a lot of links to developer resources and the last few slides here that you can take a look at including to our garrett site and our various expert groups and you’ll find that all of our all of our Garrett submit garrett submissions are built by our CI system there’s some level of validation being done automated testing being done each of the submissions you’ll see that we’ve got for our our latest electric eel release candidate we publish test results for the various boards so we’re definitely on track for this and I think by March of next year we’ll really have a very solid system and we’re also we’re also working with a number of outside open source projects like lava lake Fuego like CIP the civil infrastructure pride checked and coordinating our efforts with those guys so we can all jointly make a bigger impact in this area vision for next year really we’re seeing a lot of interest in major content providers to bring bring that content on Sage onto the AGL platform so hopefully in the next year we’ll see people like Spotify and Amazon and Comcast’s and streaming providers in the u.s. coming on board to AGL and making their apps available to the AGL community we’re seeing a lot of interest in this area already we want we’ve had a number of OEMs that are interested in porting legacy apps to AGL so basically making sure we’ve got all that enablement in place and start seeing the first of those apps being ported in 2018 to two releases for next year funky flounder and grumpy guppy spent a lot of time with these names functional safety advancing the farce functional safety initiative we’ve got a lot of different ideas on how to accomplish this and we just need to invest some time and some some resources into getting this done our board and a lot of our members we use leaves cute cute five four four all of our most of our graphics back-end our graphics toolkit so our board and our members really want to see us move to more of a true open-source toolkit there’s some concern I mean q5 is a great tool don’t get me wrong there’s some concern about the costs and production so they still want us to continue to explore with other consumer electronics projects and a true open source toolkit this is a much bigger project than just a GL though so we need to work with other projects to figure out how or what we’re going to do we’ve been working I didn’t mention it in the vision for this year but we’ve been working on on a new home screen and window managers services for hgl we we’re getting the first version of that into electric eel I was a lot of discussion last week on exactly how to do that when we were in Dresden basically want to complete that in the first half of next year have that have that all in the system and have all of our legacy apps ported have all the new apps running that chromium based html5 apps like I said earlier as we move into running html5 based apps we need something to run the backend of that so we’re working with agalya on get making chromium available to the AGL community continue to evolve our app framework improving security lessons learned from the work that we do this year we’re always doing post mortems on on with our app developers to try to figure out what we can improve working with our security team to figure out what we can improve Tizen recently released their version 4.0 of their framework their app framework so what can we take from their security work that they’ve done also improving user management and then using C groups and namespace and and new work that’s been done in system D especially in the latest version v2 35 for using dynamic use dynamic users for managing power managing memory and CPUs and things like that another another part of our vision for next year having alcohol at a sustainable audio solution we’ve been using some pieces of the Geneva which is another open-source project using some parts of the Geneva Audio Manager we found that parts of that are kind of inappropriate for app level api’s and for managing the actual devices it can be very complicated to write plugins to their audio manager so we’ve got a lot of work in progress the we call it the HDL advanced Audio architecture or for a again I come up with these really cool acronyms for a and having a video player reference app vehicle to cloud services there’s a lot of you know there’s been for many many years in the US especially but also in Europe a lot of telematics services which nobody would have thought of calling vehicle to cloud ten years ago but now that’s what we call it so having a telematics architecture and a vehicle vehicle to a standardized vehicle to cloud architecture is very much interesting and we’re forming a new expert group in this area over the next month speech to text text to speech services not that we would have a speech engine integrated into a GL but at a pet what we would have a common API where different speech providers could then plug in their engine and our apps could then develop to those specific api’s that we provide we have an ongoing reference hardware system architecture effort being led by some of our OEMs we have ten OEMs in a GL Bay pretty much all the Japanese OEMs plus mercedes-benz here in Europe Jaguar Land Rover here in Europe and Ford in the US and they’ve been they’ve been working on defining what the system architecture of the future is in vehicle for not only ivi but also telematics and a DES and things like that so they’ve released the initial version of that document and we’ll be continuing to review that and figure out the takeaways from that document that we can use for incorporating that into the software so just a quick overview we have these things I really don’t like the name we have these things called expert groups but let’s call them working groups you don’t necessarily need to be an expert to be involved in one of them we have six active expert groups right now with a vehicle to cloud expert group coming online very shortly at framework and security connectivity continuous integration automated test etc so just a quick overview our expert groups when we formed them we wanted we really only sweetleaf stuck to this principle throughout the last four or five years we wanted them we didn’t want to create a bunch of expert groups just to have say we had complete coverage throughout the vehicle we really wanted a critical mass so that we would have a minimum of five to ten people who were participating in these expert groups and and working on these issues and so what you’ll see as I go through these is some of the some of the definitions of these expert groups get a little big and we’re not really really haven’t split them out unless there’s been a again one of these critical masses that are willing to take on the work for these expert groups so you’ll see the application framework and security expert group takes on a lot of different things that you might not think of immediately when you think of either of those topics and some people don’t even think of application framework and security is going together so the app framework and security expert group tasks they’ve made namespace and C groups are now available to use in hgl but we really need to work on exactly how we’re going to make use of those so the enablement is there we’ve talked about doing some resource management recently the system that we’re using system D to third beat version 232 which is the latest Yocto version we know it until the version 235 has some really interesting features in there for or dynamic users and things like that as well as additional features in the namespace and cgroups areas so we really want to up Rev to that as soon as possible and start making use of that I talked about Tyson Florida already our SDK we’ve we’ve done a lot of work with the SDK we’ve created app templates so that the app developers have something they can start with learning how to create widgets service binders with a developer guide all that’s available now with the SDK and and this group isn’t is tasked with owning that more things we’ll be working on in the next year or the in the upcoming year including identity and user management so that you can bring a RFID card and it’ll recognize you and it’ll change the Preferences on the ivi system to yours you know you’ll you’ll have your Facebook account you’ll have your Spotify playlist you’ll have whatever it is that comes with your identity key management for app installation band and making sure that the policies are all correct and sayonara consent management things like that are all upcoming on our docket for this group the connectivity expert group was one of those that was again had a very broad remit and recently we like I said we’ve split that out some of these out to the to a new vehicle to cloud expert group so but the connectivity group I think now has a fairly reasonable focus which includes vehicle connectivity can most glynn things like that network and vehicle firewalls Bluetooth Wi-Fi NFC and smart device link so they’ve actually got a fairly reasonable amount of amount of tasks on their plate compared to some of the others they’ve been working on vehicle messaging and a vehicle messaging app framework binder so that we can abstract the signals coming in from the vehicle whether they come from kin or they come from GPIO or they come from anywhere else the applications don’t have to care about that bluetooth we’ve got binders available for Bluetooth we’re refactoring some of the binders now based on some of the lessons learned we’ve got a telephony binder that’s in progress an NFC binder that’s in progress a Wi-Fi binder that’s already available so a lot of work that’s been done by this group already a continuous integration automated test they’ve been integrating Lava with Fuego so we can do remote remote board testing for our for our automated testing they’ve really done a lot of work with improving the our Garrett Jenkins Lava workflow over the last few months they’ve created a web interface for test results so now that you can for any given board any given build you can drill down into what tests have been run which tests have pat passed which tests have failed taking a look at the log file for any failed test things like that we’ve been working with up streaming our Fork of Fuego we have a concept of the lab in a box so that companies or individuals will be able to take our test environment and our test cases and move them inside their firewall and run the tests inside their firewall without having to a lot of companies are a little iffy about punching holes in their firewall so we want to take care of that eventually we want to add some static code analysis tools that’s been on our plate for a while but we haven’t had a lot of time to look at that yet you eye and graphics expert expert group has a really big set of tasks that might seem haphazard as well but they’re looking at the what you the basically what you would consider for UI and graphics which would be compositor layer manager window manager GPU interface things like that but we also bundled multimedia audio display audio manager media we’ve got a media reference media player and media manager so that’s all bundled into this you Iying graphics expert group as well so they’ve been working on updating everything to waylynn 2.0 refactoring the home screen and the window manager updating the window manager for better secondary display support as well as updating it to include the ability to have a multiple apps running owning parts of the parts of the display and running their parts of the window better focus management for out-of-focus applications we do kind of a not a great job always with that they’re working on watham for internode display protocol so that you can have apps running and uh on your ivi system and displaying parts of the parts of a window in the instrument cluster it’s a pretty standard feature these days in most in a lot of vehicles improving audio management and configuration we talked about that a little bit already with the 4a architecture adding speech services and chromium of course so we’re looking to bring this vehicle to cloud expert group online in the next month or so including we need to need to take a look at the overall use cases tal maddox personalization and to end authentication things like that define a reference application work with some back-end providers on how to standardize what the backend expects and then identify any reference applications that we want to develop in that expert group so as I wrap up here I just got a bunch of links to our our developer resources we have a getting started guide on our wiki links to our documentation site links to our JIRA the getting the getting started guide will also give you information on how to sign up for you needed an LF ID Lynch foundation ID to access our sites but most of you probably already have one of those so you can you can get the access pretty quickly we have pre-built binaries and source tarball is available on our basically through this it’s on our wiki page with all the release notes release notes are available on our wiki page which also includes the links to the guitar balls and the the binaries like I said we use Garrett for our workflow and and get obviously on the back end so I have about five minutes left oh one other thing to point out we do like I said in the spirit of collaboration we have any number of face-to-face meetings with our developers you can see this is just this is the list of what we’ve had this year so not only at our all member meetings in which we have twice a year we had them in Tokyo and Dresden this year but also various various sites hosted by companies within AGL for there’s usually specific topics we have a automotive discussions mail list you can sign up for whenever there’s a new face to face meeting we make an announcement through the the mail list anybody is free to sign up and attend these Sam house San Jose was focused on audio manager and window manager then various topics throughout the year and we have our next one coming up in Yokohama in November so five minutes left any questions one over there yeah that’s an optional yeah we have I guess I didn’t show this slide for the we have a virtuals so the question was about hypervisor in the architecture slide I showed the AGL running on a hypervisor and the question was do we have a particular hypervisor we have a virtualization expert group if you go to the wiki you can take a look at that they’ve been exploring a number of different hypervisors Michelle Paulino of virtual open systems leads that group and he submitted patches now for AGL to run on KVM but they’re also looking at jailhouse and Zen and some others so one at one I’m I should have mentioned this one of the topics in the next month or so is having that expert group kind of take a step back and explore what the real use cases are for hypervisors in automotive and create a white paper and then create create a white paper that will release early next year and then use that to drive the work of the virtualization expert group through that throughout 2018 so if anybody is interested in participating in that effort I’ll be sending out a call for participation next week or maybe I guess this week on the AGL mail list to get people working on that yes that’s a great question I don’t have the answer oh I’m sorry he we took he asked he says I were talking about we I had in my slides I talked about an upcoming ace will be a certification how are we going to accomplish that with Linux that’s a great question and we’re going to that’s what we need to figure out it’s a big big topic and it really is is a is there a hypervisor there that we ice used just to just isolate Linux is there some kind of scaled-down version of Linux fully documented that’s that’s what we’re working what we need to figure out what we’re doing can you repeat the beginning part the comment was we wouldn’t have we wouldn’t have any luck getting into a cluster without that kind of certification yes I agree any other questions can you repeat that one more time great question I should have mentioned that so Toyota announced the question was are there any car manufacturers that have adopted a GL into their products yet Toyota announced at the automotive Linux summit earlier this year that AGL is running on the 2018 Toyota Camry and will be migrating to the rest of their vehicle platforms throughout the next few years and we expect other car manufacturers will quickly follow any other questions I can’t hear you we have the question was do we have a hardware reference platform we we’re running basically all of our code is available on a number of the eval boards hardware Hardware eval boards from Rena sauce from Intel TI the Bayou board what we want to do is is use the system reference hardware system architecture team to drive a to a more automotive looking hardware architecture where we’re to where we’re truly have the the startup startup from say NAND flash as opposed to an SD card and have some some of the real wake up circuits in there so we’re working towards that but right now we are available on about eight or nine different boards and you can find all those on the wiki page any other questions if thirty seconds left No okay well oh sorry how many oh yeah so are there other more the car manufacturers in Toyota there’s a total of 10 manufacturers that are involved with AGL most of the Japanese eco system Toyota Honda Nissan Subaru Suzuki I’m sure I’m forgetting somebody Daimler here in Europe Jaguar Land Rover Ford so there’s quite a few and il Mazda as well in in Japan so I expect that they’ll be migrating we expect that they’ll be migrating their their platforms to AGL very very shortly ok then if you have any questions and you see me around don’t be afraid to ask there’s a number of really good AGL talks this week as well that you can attend and you can ask them questions thank you very much [Applause]


Leave a Reply

Your email address will not be published. Required fields are marked *