Project ACRN

The ACRN Project made a splash at the Open Source Summit/Embedded Linux Conference Europe/OpenIoT Summit in October in Edinburgh, UK. In addition to being featured in an ongoing demo in the Intel booth at the conference, project architect Eddie Dong presented ACRN: A Big Little Hypervisor for IoT Developmentto a packed room with 112 attendees.

ACRN continues to be very well-received in automotive, industrial, and other IoT industries as an IoT-friendly lightweight hypervisor with a very small codebase. The project is concerned with making the hypervisor easier to certify for Functional Safety, which was of particular interest at this event due to the attendance of European automotive electronics companies. ACRN offers flexibility by allowing guest operating systems to share devices between VMs, or to create completely isolated hardware partitions. The ACRN demo showed that ACRN performs extremely well and provides support for Linux-based guest operating systems, with RTOS and other alternatives on the roadmap.

ACRN is a fully open source project hosted by the Linux Foundation. For more information about ACRN, lightweight hypervisor technology, and the potential for collaboration on your project, visit https://projectacrn.org.

ACRN has released version 0.2 (see the release notes), a new updated revision of the project with many added features and fixed bugs. The project team encourages you to learn more about ACRN and try it out, and also to join the weekly technical call.

The major new and updated features are summarized in the release notes, along with bugs fixed and known issues. These are the new features in 0.2:

VT-x, VT-d support: Based on Intel VT-x virtualization technology, ACRN emulates a virtual CPU with core partition and simple schedule. VT-d provides hardware support for isolating and restricting device accesses to the owner of the partition managing the device. It allows assigning I/O devices to a VM, and extending the protection and isolation properties of VMs for I/O operations.

PIC/IOAPIC/MSI/MSI-X/PCI/LAPIC: ACRN hypervisor supports virtualized APIC-V/EPT/IOAPIC/LAPIC functionality.

Ethernet: ACRN hypervisor supports virtualized Ethernet functionality. Ethernet Mediator is executed in the Service OS and provides packet forwarding between the physical networking devices (Ethernet, Wi-Fi, etc.) and virtual devices in the Guest VMs (also called “User OS”). Virtual Ethernet device could be shared by Linux, Android, and Service OS guests for regular (i.e. non-AVB) traffic. All hypervisor para-virtualized I/O is implemented using the VirtIO specification Ethernet pass-through.

Storage (eMMC): ACRN hypervisor supports virtualized non-volatile R/W storage for the Service OS and Guest OS instances, supporting VM private storage and/or storage shared between Guest OS instances.

USB (xDCI): ACRN hypervisor supports virtualized assignment of all USB xHCI and/or xDCI controllers to a Guest OS from the platform.

USB Mediator (xHCI and DRD): ACRN hypervisor supports a virtualized USB Mediator.

CSME: ACRN hypervisor supports a CSME to a single Linux, Android, or RTOS guest or the Service OS even when in a virtualized environment.

WiFi: ACRN hypervisor supports the passthrough assignment of the WiFi subsystem to the IVI, enables control of the WiFi as an in-vehicle hotspot for 3rd party devices, provides 3rd party device applications access to the vehicle, and provides access of 3rd party devices to the TCU provided connectivity.

IPU (MIPI-CS2, HDMI-in): ACRN hypervisor supports passthrough IPU assignment to Service OS or guest OS, without sharing.

Bluetooth: ACRN hypervisor supports bluetooth controller passthrough to a single Guest OS (IVI).

GPU – Preemption: GPU Preemption is one typical automotive use case which requires the system to preempt GPU resources occupied by lower priority workloads. This is done to ensure performance of the most critical workload can be achieved. Three different schedulers for the GPU are involved: i915 UOS scheduler, Mediator GVT scheduler, and i915 SOS scheduler.

GPU – display surface sharing via Hyper DMA: Surface sharing is one typical automotive use case which requires that the SOS accesses an individual surface or a set of surfaces from the UOS without having to access the entire frame buffer of the UOS. Hyper DMA Buffer sharing extends the Linux DMA buffer sharing mechanism where one driver is able to share its pages with another driver within one domain.

S3: ACRN hypervisor supports S3 feature, partially enabled in LaaG.

Demonstration of Next Generation E-Cockpit System on AliOS

Themed “Empower Digital China”, the  4 day Computing Conference 2018 opened this week in Hangzhou, China. There were over 170 leading-edge industry forums, gathering 450 companies and more than 60,000 elite IT professionals attending. The companies and guests exchanged ideas in the hot fields of cloud computing, big data, AI, etc.

Intel’s exhibition area at Computing Conference 2018

As strategic partner of the Computing Conference 2018, Intel has also joined this digital carnival to showcase its capabilities for the data era, targeting brand new development for cloud computing based on Intel Architecture. At the Conference, Intel announced its latest cooperation with Alibaba in the fields of edge computing development, hybrid cloud solutions, and infrastructure for e-commerce applications.

At the Shenzhen Summit in March this year, both parties joined hands to launch the next-generation electronic cockpit solution based on Intel’s architecture and AliOS, aiming to provide auto manufacturers and primary distributors with diversified options of integrated hardware-software meeting the demands for system integration, to enrich local ecology and to accelerate product development & continuous innovation. At this conference in Hangzhou, the two companies teamed up again to demonstrate the AliOS Internet Car System based on Intel Apollo Lake platform.

AliOS Internet Car System based on Intel Apollo Lake platform

In this demonstration, AliOS presented the future-oriented Internet-based auto operating system, the first integrated platform for the auto service ecology and the largest open platform of intelligent hardware, boasting the first online sensible map and multi-modal in-vehicle interactive mode. In this joint effort, AliOS and Intel built whole-vehicle intelligence based on the next-generation Intel Atom Processor A3900 Series, which is a car-level chip family, taking advantage of the powerful computing capacity of the chip processor and the rich ecology of AliOS to introduce more digital capabilities to cockpits. These capabilities, including the lightweight virtualization of ACRN hypervisor, AI-accelerated real-time lane line detection, 3D HD 360-degree environment, traveling in teams, and payment without exiting the car, can fully meet the application requirements in different scenarios.

We have been busy working on project ACRN since its launch in March earlier this year. We are now excited to share the details with you on our latest release of ACRN – version 0.1, on July 17, 2018. This release comes packed with an exciting list of new features, tools, and support for additional hardware.

We are happy to announce that, in addition to the original Intel Apollo Lake NUC we launched with, ACRN now supports the UP Squared board, an x86 maker board based on the Intel Apollo Lake platform and well suited for IoT, industrial, automotive, and digital signage applications. This will allow for easier testing and development, and code contributions back to the project!

One of the more exciting new features to highlight is the graphics GPU sharing feature we’ve enabled. GVT-g Virtual Graphics support is now available in this version of ACRN. GVT-g virtual graphic support lets Service OS and User OS applications run GPU workloads simultaneously with minimal overhead. This helps to ensure that both the Service OS (SOS) and the User OS (UOS) can benefit from the full physical GPU capabilities. The developer is able to present an immersive user experience by running GPU intensive workloads, such as 2D and 3D media decoding, encoding, and transcoding, directly on the (virtualized) GPU while offloading the CPU to run other critical safety applications.   

GVT-g: Mediated Pass-Through Architecture

Another important feature is the enablement of the Virtio virtualization standard, allowing the Service OS and User OS applications to share physical LAN network and storage devices using popular standard APIs. This is important because it is an industry standard for I/O virtualization where performance-critical devices sharing is enabled. By adopting this virtio specification, we can reuse many frontend virtio drivers already available in a Linux-based User OS, dramatically reducing the development effort for the frontend drivers.

VirtIO Architecture

A big thank you all ACRN community members participating in the project – all contributions, requirements, testing, validating, and documentation helps us, and ensures that the project continues to grow and meet our community’s needs.

To learn more about project ACRN please see our Getting Started Guides, join the mailing list, and visit our website!

Hypervisors are pretty complex pieces of software, and that means new ones don’t appear on the scene very often — although a few new hypervisors designed for container environments have hit the headlines in the last year or so.

But alongside containers, there is one other area of technology that has been touted to become huge in the next few years, and that’s the Internet of Things (IoT), which includes the potentially mega-huge Industrial Internet of Things (IIoT). (To get some idea of scale, worldwide IIoT spending was estimated at about $20 billion in 2012, but it is forecast to reach $500 billion by 2020.)

With that in mind, the arrival on the scene of ACRN (pronounced “Acorn”) should not come as a huge surprise. What is ACRN? Put simply, it is an open source reference hypervisor that has been built to meet the unique needs of embedded IoT development. It was announced by the Linux Foundation in late March.

Read Article »

By Hongbo Wang, Cherry Wang and Xiaolan Wang on behalf of The ACRN Meetup Team

Executive Summary

The first ACRN Community meet-up was held on June 14, 2018 in Shanghai, hosted by Intel. There was a total of 118 attendees with representative from various companies & industry segments, e.g. Shanghai GM, Mentor, Archemind, Huawei, Bosch, Ruijie, and more to name a few.

The meetup was a great event to bring the community up-to-speed on ACRN usage, roadmap & hypervisor architecture overview, focused sessions on Virtio & GPU sharing, and how to build/bring-up & debug ACRN. Attendees were very engaged and there were lots of questions during the sessions.

The team prepared five Apollo Lake UP2 boards as lucky draw prizes that were given out to the developers in attendance. We are looking forward to their feedback and contributions back on these boards. We also shared pictures real-time via the ACRN WeChat Group and weibo, then published meet-up materials and video recording so those who weren’t able to come could also study the materials offline.

The meetup allowed us to learn of the strong interest in using ACRN for various IoT usages from the community. We also received immediate positive feedback from the attendees, and received 2 patches from the community within 48 hours of the meet-up.

Agenda and Speakers

The full-day Meetup covered ACRN usage models, open source project governance model, features list and roadmap discussion, hypervisor architecture, Virtio framework, GPU sharing, featured a hands-on session for build/bring-up/debug ACRN and showcased three ACRN demos. Attendees were very engaged and there were lots of Q&As during the sessions.

Demos and Hands-on

1. LaaG on Apollo Lake NUC: SOS (Clear Linux) + 1 * Guest OS (Clear Linux)
2. AaaG on Kabylake desktop: SOS (Clear Linux) + 3 * Guest Android (Android O)
3. LaaG hands-on on Apollo Lake UP2 boards

Q&A and Lucky Draw

There were good questions throughout the sessions. At the end, we collected feedback from the attendees and gave away five UP2 boards as lucky draw to the developers in attendance.

Sample Feedback Comments

• “I appreciate the good organization of the ACRN Hypervisor Meetup. A new beautiful and exiting world was shown to us on the seminar. Wish I can catch the developing of this VM technology on Embedded Systems”
• “Thank you very much for organizing this great event to have community members, customers and Intel team to share, brainstorming and collaborate! Great event and appreciate everybody’s contribution and sharing!!”

More On-Site Photos

Special thanks To (Not the Full Attendee List)

Speakers:

• Aung, Diana
• Chen, Conghui
• Jiang, Fei
• Li, Susie
• Ren, Jack
• Wen, Denny Z
• Wang, Hongbo
• Yang, Ailin/Wang, Nianhui

Demos and Hands-On: ACRN

• Wang, Nianhui
• Yan, Like
• Yang, Ailin
• Zou, Terry

Event Team/Volunteers:

• Gu, Queenie
• Wang, Xiaolan
• Wang, Cherry
• Zhuang, Lena
• Zhai, Edwin
• Yan, Like

This article was produced by The Linux Foundation with contributions from Eddie Dong, Principle Engineer of Intel Open Source Center.

As the Internet of Things has grown in scale, IoT developers are increasingly expected to support a range of hardware resources, operating systems, and software tools/applications. This is a challenge given many connected devices are size-constrained. Virtualization can help meet these broad needs, but existing options don’t offer the right mix of size, flexibility, and functionality for IoT development.
ACRN

ACRN™ is different by design. Launched at Embedded Linux Conference 2018, ACRN is a flexible, lightweight reference hypervisor, built with real-time and safety-criticality in mind and optimized to streamline embedded development through an open source platform.

» Read More

This article was first released on TMTPost (钛媒体), a Chinese media/news outlet focusing on technology, media and telecom industries on April 27, 2018 in Mandarin. This is the translated version of the news article.

When the automobile is seen as a mobility tool with more intelligent and connected attributes, developing a model with premium experiences is no longer simply the business of automobile manufacturers. A growing number of technology companies and suppliers are beginning to shift their mindset, adding more internet-enabled features and experiences.

During the 2018 AutoChina 2018 Show in Beijing, a unique Maserati car was showcased at Pavilion E4 in Harman’s booth. Unique features of this car adopted Harman’s latest “intelligent cockpit” solution based on Intel’s innovative technology, a result of the three-parties’ collaboration.
Harman’s new-generation “intelligent cockpit” uses a large QLED screen that extends from the center of the vehicle, all the way to the front passenger door. Though not as big as the 64-inch screen demonstrated by Byton, the point is–this suite of the solution developed by Harman allows customization for the needs of drivers and passengers.

The large QLED screen that extends from the center of the vehicle, all the way to the passenger door

In specific driving scenarios, two users are able to use separate screens without affecting each other. For example, when the driver is using it to access navigation features, the passenger is able to simultaneously use it to access entertainment features such as watching a movie or playing games, without occupying the same screen, which is very convenient.

Knob switch with embedded OLED screen

The interactive design includes three centrally located knob in the center console to control the functions of air temperature, time, and access to voice assistants respectively. Each knob switch features an embedded OLED screen that displays the desired function. Nowadays, the physical keys for interactive interfaces of automobiles are giving way to an increasing number of virtual ones.

Additionally, this control system comes with a curved OLED screen located just below the dashboard, allowing users to have access to A/C controls, driver settings, and other various vehicle options.

A close up of Haman intelligent cockpit based on Intel chip

From a technical perspective, Haman’s new generation of “intelligent cockpit” demonstrated on the Maserati model is powered by the Intel® Atom™ processor Apollo Lake and the virtualization software, ACRN.

As a processor designed specially for car experiences, Intel® Atom™ processor Apollo Lake features powerful computing power and workload integration functions, and multiple high definition video output interfaces. The processor can support in-vehicle infotainment system/cockpit experience, a digital instrument cluster, rear-seat entertainment and advanced driving assist system;

ACRN is a flexible and lightweight virtualization software recently unveiled by Linux Foundation. With most of the engineering and code contributions from Intel, the project caters to the use case of automobile products. ACRN is positioned for efficient real-time and safety-critical features and allows continuous optimization of functional design through an open-source platform.

ACRN is comprised of two main components: the hypervisor software and its device model, which support the sharing of a broad range of I/O devices. Intel’s rich experience in virtualization technology played a critical role in the initial development of this hypervisor solution.

In the transformation of future mobility, Intel has accurately judged the market direction of intelligent cockpits, introduced appropriate product planning, and joined hands with a number of world-renowned companies such as Tesla, BMW, GM, FCA, Toyota, as well as Neusoft and SAIC in China to co-develop intelligent cockpits and other platform technologies, as well as modular automotive-grade products.

In addition, to accelerate the application of intelligent technologies to automobiles, Intel has partnered with multiple Tier 1 suppliers of auto parts and OEMs in China. Intel’s broad and abundant market resources, will undoubtedly speed up the timetable for mass production.

Harman and Intel will play their respective roles in promoting the localization of “intelligent cockpits”. It is reported that Haman’s new-generation “intelligent cockpit” has been gradually adopted by local Chinese car manufacturers.