SlideShare a Scribd company logo
Harold G. Tiedeman, Jr., Technical Fellow, Rockwell Collins
Paul J. Parkinson, Principal Systems Architect, Wind River
Plan With Confidence: Route to a Successful
DO-178C Multi-Core Certification
Moderator:
John McHale, OpenSystems Media
Speakers:
Agenda
 Housekeeping
 Presentation
 Questions and Answers
 Wrap-up
© 2018 WIND RIVER. ALL RIGHTS RESERVED.
PLAN WITH CONFIDENCE:
ROUTE TO A SUCCESSFUL
DO-178C MULTI-CORE
CERTIFICATION
Harold G. Tiedeman, Jr.
Technical Fellow, Rockwell Collins
Paul Parkinson
Principal Systems Architect, Wind River
Web Seminar, September 27, 2018
4 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
 Current state of the art in civil certified processors
– Single core, complex system-on-chip
– Single core active designs
 Partitioned operating systems
– Support multi-DAL software
– Mature and in service, both civil and military systems
 Next logical step: full multi-core with all cores active
– Realize performance and SWaP benefits
– Must prove safety and deterministic behavior
 Never before accomplished
INTRODUCTION
5 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
WHAT IS UNIQUE ABOUT MULTI-CORE PROCESSORS?
 Provide increased performance
potential
 Challenges for application isolation
and determinism
 Variation in suitability of designs
for use in avionics
 Shared resources provide multi-
core interference potential
 Analysis of behaviour and worst-
case execution timing (WCET) is
more complex
 Processor design information may
not be available for hardware
analysis
Notional Multi-core Processor Architecture
with Shared L2 Cache
Core #1
L1 I-Cache L1 D-Cache
Core #2
L1 I-Cache L1 D-Cache
L2 Cache
Coherency Module
6 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
RELATIONSHIPS AND INDUSTRY INVOLVEMENT
 Engage OEMs, civil, and military certification
authorities
– Start early to help them understand issues
– Agree how to address each one
 Use proprietary relationships
– Operating system (OS) vendors
– System-on-chip (SoC) vendors
 Aligned with the EASA multi-core certification review
item (CRI) and FAA certification authorities software
team CAST-32A position paper
 Engage internal subject matter experts to make the
leap to multi-core systems
 Use of certification authorities and certification
liaisons to help guide efforts
7 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
SELECTING THE RIGHT MULTI-CORE PROCESSOR:
MULTICORE FOR AVIONICS (MCFA)
 Rockwell Collins MCFA involvement
– Founding member and executive board member
– Multiple individuals engaged from both
Commercial and Government Systems businesses
 Processor assessments conducted by
Rockwell Collins
– Reviews conducted at supplier’s site
– Based on a set of questions and artifacts
established by MCFA
– Provides insight into vendor’s processes for
development, production, and quality
– Freescale/NXP QorIQ P-Series assessment
completed summer 2011
– Freescale/NXP QorIQ T-Series assessment
completed summer 2016
 FSL became part of NXP Dec 2015
NXP public presentation on MCFA
https://community.nxp.com/docs/DOC-331635
8 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
SELECTING THE RIGHT MULTI-CORE PROCESSOR:
PROCESSING PLATFORM DEFINITION
 Multi-core common processing resource (MCPR)
– Common reusable logical and schematic design
– Can be easily laid out repeatedly as necessary for various form factors
 Includes
– Multi-core processor
– System memory
– DAL A hardware monitor FPGA and software monitors
– Ethernet networking
– Basic I/O (serial and discretes)
– Non-volatile storage (NAND, NOR, nvROM)
 4x existing product line performance
9 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
SELECTING THE RIGHT MULTI-CORE PROCESSOR:
PROCESSOR SELECTION
 Freescale Quad-Core T2080 QorIQ SoC has been
chosen for the MCPR
– Process can be iterative as deep dives occur
– Switch from previous plan to use the NXP P-series
 T2080 provides significant processing advantage
– E6500 Core improves integer and floating point
processing (2x improvement)
– About 5.5x DMIPS performance over baseline
processor
 Higher-performance internal SoC interconnect
fabric significantly reduces interference
 Additional I/O controllers provide better mitigation
options to reduce interference
 Performance enhancements, i.e., AltiVec™ and
dual threading at no added power
NXP QorIQ® P4080 Processor (Image: NXP)
NXP QorIQ® T2080 Processor (Image: NXP)
Peripheral Access
Management Unit0
128 KB Backside
L2 Cache
Power Architecture
e500mc Core
32 KB D-Cache 32 KB I-Cache
1024 KB CoreNet
Platform Cache
1024 KB CoreNet
Platform Cache
64-bit DDR2/3 Memory
Controller with ECC
64-bit DDR2/3 Memory
Controller with ECC
CoreNet Coherency Fabric
PAMU PAMU PAMU PAMU
Security Fuse Processor
Security Monitor
2x USB 2.0 w/ULPI
eLBC
Power Management
SD/MMC
2x DUART
1x 12C
SPI, GPIO
SEC
4.0
PME
2.0
Queue
Mgr.
Buffer
Mgr.
Frame Manager
Parse, Classify,
Distribute
10 G
1 G 1 G
1 G 1 G
Frame Manager
Parse, Classify,
Distribute
10 G
1 G 1 G
1 G 1 G
Real-Time Debug
2x DMA
PCIe
PCIe
PCIe
sRIO
sRIO
Watchpoint
Cross Trigger
Perf.
Monitor
Trace
Aurora
18 Lanes, 5 GHz SerDes
Core Complex (CPU, L2 and Frontside CoreNet Platform Cache)
P4080 and P4081 Only
Networking Elements
P4080 and P4040 Only
Accelerators and Memory Control
Basic Peripherals and Interconnect
Peripheral Access
Management Unit
2 MB Banked L2
521 KB
Platform Cache
64-bit DDR2/3
Memory Controller
Coherency Fabric
PAMU PAMU PAMU
Security Fuse Processor
Security Monitor
IFC
Power Management
SDXC/eMMC
2x DUART
4x 12C
SPI, GPIO
2x USB 2.0 + PHY
DCE
1.0
PME
2.1
Security
5.2
(XoR,
CRC)
RMan
Frame Manager
Parse, Classify, Distribute
hIGIG/+
4x 1 / 2.5 / 10 G
DCB
2x 1 / 2.5 G
Real-Time Debug
Watchpoint
Cross Trigger
Perf.
Monitor
Trace
Aurora
Core Complex (CPU, L2, L3 Cache)
Accelerators and Memory Control
Basic Peripherals and Interconnect
Networking Elements
Pre-fetch
32 KB
D-Cache
32 KB
I-Cache
32 KB
D-Cache
32 KB
I-Cache
32 KB
D-Cache
32 KB
I-Cache
32 KB
D-Cache
32 KB
I-Cache
PowerTM
E6500
T1 T2
PowerTM
E6500
T1 T2
PowerTM
E6500
T1 T2
PowerTM
E6500
T1 T2
Peripheral Access Management Unit
Queue
Mgr.
Buffer
Mgr.
8ch
DMA
PCIe
PCIe
PCIe
PCIe
sRIO
sRIO
8ch
DMA
8ch
DMA
SATA2.0
SATA2.0
8-Lane 1- GHz SerDes 8-Lane 1- GHz SerDes
10 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
REQUIREMENTS FOR A SAFETY-CRITICAL
OPERATING PLATFORM
 Open architecture reduces system integration costs
 Preserves investment in existing applications
Multiple Guest OS
 Consolidates onto a single platform, reducing size, weight,
and power (SWaP) requirements
Consolidation
 Enables flexible configuration using efficient hardware
virtualization support
Virtualization Support
 Enables hosting of portable POSIX®, ARINC 653, and
FACE™ applications
Open Standards
Flexible Business
Model
 Supports product line approach and multiple customers
without additional licensing overhead
Safety Certification
 Low perceived DO-178C certification risk
 Supports incremental certification
11 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
SELECTING THE RIGHT OS PLATFORM FOR
MULTI-CORE CERTIFICATION
 Wind River safety-critical track
record:
– 2000: VxWorks Cert for federated
applications
– 2002: VxWorks 653 for IMA
applications
 Integrated modular avionics
– Enables many federated systems
to be hosted on a common
computing platform
– Reduces SWaP
 VxWorks 653
– Supports ARINC 653, POSIX®,
VxWorks, and Ada applications
– Provides COTS DO-178
certification packages
– First OS to achieve conformance
certification for FACE™ OSS
Safety Base Profile
VxWorks and VxWorks 653 cumulative design wins
(federated and IMA)
Cumulative Federated Cumulative IMA Projects
12 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
SELECTING THE RIGHT OS PLATFORM FOR
MULTI-CORE CERTIFICATION
 High-level design goals for VxWorks 653 Multi-core Edition
• COTS RTCA DO-178C DAL A certification evidence
• Support of multiple DALs on multiple cores
• Perform static configuration and enforcement as per ARINC 653
• Perform fault isolation and containment (health monitors)
• Enable IMA role-based development and delivery as per RTCA DO-297
• Robust partitioning of application and OS environments for ease of updates
and reduced certification burden
13 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
VxWorks 653 Application Executive
Multi-Core Hardware Hardware Virtualization Support
Board Support Architecture Support
Avionics Bus (MIL-STD-1553B, ARINC 429, ARINC 664, SAE AS6802…)
XML Configuration Data
Core 1Core 0 Core 2 Core 3
Flight
Management
Application
DAL B
VxWorks
Partition OS
Flight
Display
Application
DAL A
ARINC 653
Guest OS
Sensor
Intelligence
Application
DAL C
VxWorks
Partition OS
New
Applications
DAL E
Linux
Guest OS
Legacy
Applications
DAL E
Legacy
Guest OS
VxWorks 653 Multi-Core Edition system architecture
Virtual Machines
Virtualized partition Guest Operating Systems and applications
Enables consolidation of applications onto common platform
Reduces program migration and lifecycle costs compared to multiple LRUs
Type-1 Hypervisor-based module operating system
Provides lightweight supervision of multiple cores
Utilizes full hardware virtualization assistance
Implements ARINC 653 health management
Independent Build Link and Load (IBLL)
Supports role-based development as per DO-297
XML system configuration and DO-178C qualified XML-to-binary compiler
Minimises impact of change
Significantly reduces cost of incremental certification
SELECTING THE RIGHT OS PLATFORM
FOR MULTI-CORE CERTIFICATION
14 © 2018 WIND RIVER. ALL RIGHTS RESERVED.14 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
COMPLYING WITH FAA CAST-32A
MULTI-CORE OBJECTIVES
 Initial lack of FAA and EASA formal policy on
multi-core processors
 Publication of FAA CAST-32A multi-core objectives in 2016
 Rockwell Collins and Wind River collaborative approach
 Positive feedback on approach from multiple certification agencies
15 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
COMPLYING WITH FAA CAST-32A
MULTI-CORE OBJECTIVES
Objective Why It’s Important
MCP_Planning_1: Plans identify MCP software
architecture (including dynamic and IMA aspects).
 The plans provide the overall system
design context for the certifying authority.
MCP_Planning_2: Plans provide a high-level
description of how MCP shared resources and
dynamic features will be used and how the applicant
intends to allocate and verify the use of shared
resources.
 Planning shows that the applicant has
thought through critical multi-core issues
that could impact the execution of the
application software.
MCP_Resource_Usage_1: The applicant has
determined and documented the MCP configuration
settings.
 MCPs are extremely complex and likely
have configuration settings that could
negatively impact system safety.
MCP_Resource_Usage_2: The applicant has
planned, developed, documented, and verified a
means that ensures that in the event of any of the
critical configuration settings of the MCP being
inadvertently altered, an appropriate means of
mitigation is specified.
 Configuration settings could be
inadvertently modified by software errors
or single event upsets in ways that result
in undefined behavior if not mitigated.
16 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
COMPLYING WITH FAA CAST-32A
MULTI-CORE OBJECTIVES
Objective Why It’s Important
MCP_Resource_Usage_3: The applicant has
identified the interference channels and has verified the
means of mitigation of the interference.
 Interference channels are a source of
jitter and performance degradation and
may have a significant, negative impact
on the determinism of the processing
system.
MCP_Resource_Usage_4: The applicant has
identified, allocated, and verified that the available
resources of the MCP and of its interconnect are
sufficient to meet the demands of the integrated
software.
 MCP shared resources could be
oversubscribed by the full collection of
hosted software, resulting in degradation
of expected functionality.
MCP_Software_1: There is verification that all the
hosted software components function correctly and
have sufficient time to complete their execution when
all the hosted software is executing in the intended final
configuration.
 Software applications running
simultaneously on different cores impact
each other’s execution timing and need
to be integrated together to understand
the impacts to operational behavior.
17 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
COMPLYING WITH FAA CAST-32A
MULTI-CORE OBJECTIVES
Objective Why It’s Important
MCP_Software_2: Verification that the data and
control coupling is correct during software requirement-
based testing.
 Data and control coupling across cores is
more complex than coupling across
partitions on a single-core processor and
may result in unintended behavior if not
verified to be correct.
MCP_Error_Handling_1: Identification of the effects of
failures that may occur within the MCP and plan,
design, implement, and verify means by which to detect
and handle those failures in a fail-safe manner.
 The high level of integration of device
functions and peripheral interfaces
typically found within multi-core SoC
designs drives a need for more built-in-
test and monitoring for desired behavior.
MCP_Accomplishment_Summary_1: The applicant
has summarized in their SAS, HAS, or other deliverable
documentation how they have met each of the
objectives of this document.
 Provides a reference for the evidence
developed to build assurance that the
MCP system design is appropriate for
safety-critical use.
18 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
MULTI-CORE CERTIFICATION APPROACH—
EXECUTION FLOW
 Much of the multi-core–specific
certification effort executes in
parallel with typical engineering
development flow
 Feedback loops need to exist
– Interference research may drive
system or software requirements
– Changes in system functional
requirements may drive new
interference channels or alter
existing channels
Research Processor Design
Assess Non-Determinism &
Mitigation Techniques
Create CAST-32A Plan
Define Configuration
Analyze Interference Channels &
Mitigate Impacts
Perform Partitioning & Shared
Memory Analyses
Capture CAST-32A
Accomplishments in HAS/SAS
Functional
System
Requirements
Develop
System
Verify System
19 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
MULTI-CORE CERTIFICATION APPROACH—
DETERMINISM ANALYSIS
 Determinism analysis addresses many CAST-32A concerns
 The determinism analysis is split into five activities:
– Configuration analysis
– Interference channel analysis
– Partition analysis
– Shared memory analysis
– Errata analysis
 Configuration analysis documents SoC settings
 Interference channel analysis quantifies the behavior of the SoC
 Partition/share memory analysis ensures that the software
architecture results in robust behavior
 Errata analysis demonstrates maturity of the platform and addresses
reported problems
20 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
RESULTS AND SUMMARY
 Analysis of SoC processors is extremely complex
 Interference in multi-core platforms is real and impactful, but it can
be measured, characterized, and mitigated
– For the chosen SoC, the behavior and impacts are consistent with
expectations
– Behavior is extremely dependent on SoC configuration and usage
– It is also dependent on application resource usage
 Final configuration rests with the integrator, but the approach
taken can minimize those responsibilities
– Analysis tools and integration guides make that job even easier
 Certification is progressing with SOI4 submittal later this year and
TSO early next year
21 © 2018 WIND RIVER. ALL RIGHTS RESERVED.
FURTHER INFORMATION
 Rockwell Collins and Wind River resources:
https://resources.windriver.com/wind-river-rockwell-collins
 Continue the conversation in our next web seminar:
– Certifying Avionics COTS Hardware & Software to DAL A
http://bit.ly/certify-cots
© 2018 WIND RIVER. ALL RIGHTS RESERVED.
Q&A
© 2018 WIND RIVER. ALL RIGHTS RESERVED.
BACKUP SLIDES
Audience Q & A
Harold G. Tiedeman, Jr.,
Technical Fellow,
Rockwell Collins
Paul J. Parkinson,
Principal Systems Architect,
Wind River
Thanks for joining us
Event archive available at:
http://ecast.opensystemsmedia.com/
E-mail us at: jgilmore@opensystemsmedia.com

More Related Content

PPT
Cisco catalyst3750presspresentation
PPTX
How to Leverage Open Architectures for Existing Systems
PDF
Nsx t reference design guide 3-0
PDF
MT135_Simplifying web-scale systems management with the Dell PowerEdge Embedd...
PPTX
RTI Support for FACE TSS
PDF
OCP Telco Engineering Workshop at BCE2017
PPTX
Make IT Simple, Make Business Agile
PDF
Introducing Cisco HyperFlex Systems: The Next Generation in Complete Hypercon...
Cisco catalyst3750presspresentation
How to Leverage Open Architectures for Existing Systems
Nsx t reference design guide 3-0
MT135_Simplifying web-scale systems management with the Dell PowerEdge Embedd...
RTI Support for FACE TSS
OCP Telco Engineering Workshop at BCE2017
Make IT Simple, Make Business Agile
Introducing Cisco HyperFlex Systems: The Next Generation in Complete Hypercon...

What's hot (19)

PDF
Accelerating Virtual Machine Access with the Storage Performance Development ...
PPT
Weaving the Future - Enable Networks to Be More Agile for Services
PPTX
TechWiseTV Workshop: Application Hosting on Catalyst 9000 Series Switches
PDF
Intel NFVi Enabling Kit Demo/Lab
PDF
Marvell QLogic 2600 Series 16Gb Gen 5 FC HBAs Double Performance and Flexibility
PDF
What are latest new features that DPDK brings into 2018?
PPTX
VIPRION 2400 and vCMP
PDF
2011-11-03 Intelligence Community Cloud Users Group
PDF
Running Kubernetes on OpenStack
PDF
TechWiseTV Workshop: Cisco UCS C4200
PDF
IBM System Networking Portfolio Update, June 2014
PPT
EMC for V Mware Overview
PDF
Enabling new protocol processing with DPDK using Dynamic Device Personalization
PDF
Data Center Storge Architecture comparison EMC VMAX vs HUAWEI 18000 series
PDF
Security and Virtualization in the Data Center
PDF
NFV features in kubernetes
PDF
ONS 2018 LA - Intel Tutorial: Cloud Native to NFV - Alon Bernstein, Cisco & K...
PDF
SDN in the Enterprise
PDF
Red Hat NFV solution overview
Accelerating Virtual Machine Access with the Storage Performance Development ...
Weaving the Future - Enable Networks to Be More Agile for Services
TechWiseTV Workshop: Application Hosting on Catalyst 9000 Series Switches
Intel NFVi Enabling Kit Demo/Lab
Marvell QLogic 2600 Series 16Gb Gen 5 FC HBAs Double Performance and Flexibility
What are latest new features that DPDK brings into 2018?
VIPRION 2400 and vCMP
2011-11-03 Intelligence Community Cloud Users Group
Running Kubernetes on OpenStack
TechWiseTV Workshop: Cisco UCS C4200
IBM System Networking Portfolio Update, June 2014
EMC for V Mware Overview
Enabling new protocol processing with DPDK using Dynamic Device Personalization
Data Center Storge Architecture comparison EMC VMAX vs HUAWEI 18000 series
Security and Virtualization in the Data Center
NFV features in kubernetes
ONS 2018 LA - Intel Tutorial: Cloud Native to NFV - Alon Bernstein, Cisco & K...
SDN in the Enterprise
Red Hat NFV solution overview
Ad

Similar to Plan with confidence: Route to a successful Do178c multicore certification (20)

PPTX
Enhancement of ARINC 653 for Multi-core Hardware.pptx
PDF
Accelerating EDA workloads on Azure – Best Practice and benchmark on Intel EM...
PDF
Procesamiento multinúcleo óptimo para aplicaciones críticas de seguridad
PPTX
Meet the New VxWorks (Sep.2019)
PPTX
Power8 sales exam prep
PDF
Cots moves to multicore: Wind River
PPTX
IBM Power Systems E850C and S824
PPT
Wind River Overview
PDF
Superior Cloud Economics with Power Systems
PPTX
Learn About the FACE Standard for Avionics Software and a Ready-to-Go COTS Pl...
PDF
Migrating Mission-Critical Workloads to Intel Architecture
PPTX
IBM Power Systems Update
PPT
Power 7 Overview
PDF
Multicore 101: Migrating Embedded Apps to Multicore with Linux
PDF
Track A-Shmuel Panijel, Windriver
PDF
OpenPOWER Acceleration of HPCC Systems
PPTX
IBM Power Systems - enabling cloud solutions
PPT
STG101 Power Product_PRESENTATION VERSION 1.ppt
PPT
Wind River Medical Devices
PPTX
Wind River For Medical
Enhancement of ARINC 653 for Multi-core Hardware.pptx
Accelerating EDA workloads on Azure – Best Practice and benchmark on Intel EM...
Procesamiento multinúcleo óptimo para aplicaciones críticas de seguridad
Meet the New VxWorks (Sep.2019)
Power8 sales exam prep
Cots moves to multicore: Wind River
IBM Power Systems E850C and S824
Wind River Overview
Superior Cloud Economics with Power Systems
Learn About the FACE Standard for Avionics Software and a Ready-to-Go COTS Pl...
Migrating Mission-Critical Workloads to Intel Architecture
IBM Power Systems Update
Power 7 Overview
Multicore 101: Migrating Embedded Apps to Multicore with Linux
Track A-Shmuel Panijel, Windriver
OpenPOWER Acceleration of HPCC Systems
IBM Power Systems - enabling cloud solutions
STG101 Power Product_PRESENTATION VERSION 1.ppt
Wind River Medical Devices
Wind River For Medical
Ad

More from ICTperspectives (20)

PDF
UNI EN 12464-1-Illuminazione dei Luoghi di Lavoro
PDF
Mastering Modern C++: C++11, C++14, C++17, C++20, C++23
PDF
Modello-MAD-interpello2024-scuola-PDFeditabile-v1
PDF
User guide of VectorCast 2024 C/C++ for safety critical applications
PDF
User guide of VectorCast 2024 ADA testing tool for safety critical software
PDF
Comparison-SiemensTIAPortal-RockwellStudio5000
PDF
Levelised Cost of Hydrogen (LCOH) Calculator Manual
PDF
Guide on the use of Artificial Intelligence-based tools by lawyers and law fi...
PDF
Nuclear Power Economics and Structuring 2024
PDF
BPMN tutorial by Draw Libre Office
PDF
Dynamical systems simulation in Python for science and engineering
PDF
Dlgs 165/2001 e dlgs 75/2017-Ordinamento del lavoro nelle PA e successive int...
PDF
A Framework of Knowledge, Skills and Attitudes Conductive to High Performanc...
PDF
Progettazione ed installazione impianti d'antenna
PDF
Report 2030 Digital Decade
PDF
Artificial Intelligence The Future of Humankind
PDF
Intro to embedded systems programming
PDF
Implementing-Value-Based-Healthcare-In-Europe-EIT-Summit-2019
PDF
Codice-ATECO-ISTAT-62.0
PDF
Auriga-DiProgess MAX2-Misuratore-di-campo-TV-Fibra-Ottica
UNI EN 12464-1-Illuminazione dei Luoghi di Lavoro
Mastering Modern C++: C++11, C++14, C++17, C++20, C++23
Modello-MAD-interpello2024-scuola-PDFeditabile-v1
User guide of VectorCast 2024 C/C++ for safety critical applications
User guide of VectorCast 2024 ADA testing tool for safety critical software
Comparison-SiemensTIAPortal-RockwellStudio5000
Levelised Cost of Hydrogen (LCOH) Calculator Manual
Guide on the use of Artificial Intelligence-based tools by lawyers and law fi...
Nuclear Power Economics and Structuring 2024
BPMN tutorial by Draw Libre Office
Dynamical systems simulation in Python for science and engineering
Dlgs 165/2001 e dlgs 75/2017-Ordinamento del lavoro nelle PA e successive int...
A Framework of Knowledge, Skills and Attitudes Conductive to High Performanc...
Progettazione ed installazione impianti d'antenna
Report 2030 Digital Decade
Artificial Intelligence The Future of Humankind
Intro to embedded systems programming
Implementing-Value-Based-Healthcare-In-Europe-EIT-Summit-2019
Codice-ATECO-ISTAT-62.0
Auriga-DiProgess MAX2-Misuratore-di-campo-TV-Fibra-Ottica

Recently uploaded (20)

PPTX
KTU 2019 -S7-MCN 401 MODULE 2-VINAY.pptx
PDF
PPT on Performance Review to get promotions
PPT
CRASH COURSE IN ALTERNATIVE PLUMBING CLASS
PPTX
CH1 Production IntroductoryConcepts.pptx
PPTX
Engineering Ethics, Safety and Environment [Autosaved] (1).pptx
PPTX
UNIT-1 - COAL BASED THERMAL POWER PLANTS
PDF
composite construction of structures.pdf
PPTX
web development for engineering and engineering
PDF
Evaluating the Democratization of the Turkish Armed Forces from a Normative P...
PDF
keyrequirementskkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk
PDF
Digital Logic Computer Design lecture notes
PPTX
Infosys Presentation by1.Riyan Bagwan 2.Samadhan Naiknavare 3.Gaurav Shinde 4...
PDF
BMEC211 - INTRODUCTION TO MECHATRONICS-1.pdf
PPTX
OOP with Java - Java Introduction (Basics)
PDF
Mitigating Risks through Effective Management for Enhancing Organizational Pe...
PPTX
bas. eng. economics group 4 presentation 1.pptx
PDF
Embodied AI: Ushering in the Next Era of Intelligent Systems
PPTX
Internet of Things (IOT) - A guide to understanding
PDF
Model Code of Practice - Construction Work - 21102022 .pdf
PDF
Enhancing Cyber Defense Against Zero-Day Attacks using Ensemble Neural Networks
KTU 2019 -S7-MCN 401 MODULE 2-VINAY.pptx
PPT on Performance Review to get promotions
CRASH COURSE IN ALTERNATIVE PLUMBING CLASS
CH1 Production IntroductoryConcepts.pptx
Engineering Ethics, Safety and Environment [Autosaved] (1).pptx
UNIT-1 - COAL BASED THERMAL POWER PLANTS
composite construction of structures.pdf
web development for engineering and engineering
Evaluating the Democratization of the Turkish Armed Forces from a Normative P...
keyrequirementskkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk
Digital Logic Computer Design lecture notes
Infosys Presentation by1.Riyan Bagwan 2.Samadhan Naiknavare 3.Gaurav Shinde 4...
BMEC211 - INTRODUCTION TO MECHATRONICS-1.pdf
OOP with Java - Java Introduction (Basics)
Mitigating Risks through Effective Management for Enhancing Organizational Pe...
bas. eng. economics group 4 presentation 1.pptx
Embodied AI: Ushering in the Next Era of Intelligent Systems
Internet of Things (IOT) - A guide to understanding
Model Code of Practice - Construction Work - 21102022 .pdf
Enhancing Cyber Defense Against Zero-Day Attacks using Ensemble Neural Networks

Plan with confidence: Route to a successful Do178c multicore certification

  • 1. Harold G. Tiedeman, Jr., Technical Fellow, Rockwell Collins Paul J. Parkinson, Principal Systems Architect, Wind River Plan With Confidence: Route to a Successful DO-178C Multi-Core Certification Moderator: John McHale, OpenSystems Media Speakers:
  • 2. Agenda  Housekeeping  Presentation  Questions and Answers  Wrap-up
  • 3. © 2018 WIND RIVER. ALL RIGHTS RESERVED. PLAN WITH CONFIDENCE: ROUTE TO A SUCCESSFUL DO-178C MULTI-CORE CERTIFICATION Harold G. Tiedeman, Jr. Technical Fellow, Rockwell Collins Paul Parkinson Principal Systems Architect, Wind River Web Seminar, September 27, 2018
  • 4. 4 © 2018 WIND RIVER. ALL RIGHTS RESERVED.  Current state of the art in civil certified processors – Single core, complex system-on-chip – Single core active designs  Partitioned operating systems – Support multi-DAL software – Mature and in service, both civil and military systems  Next logical step: full multi-core with all cores active – Realize performance and SWaP benefits – Must prove safety and deterministic behavior  Never before accomplished INTRODUCTION
  • 5. 5 © 2018 WIND RIVER. ALL RIGHTS RESERVED. WHAT IS UNIQUE ABOUT MULTI-CORE PROCESSORS?  Provide increased performance potential  Challenges for application isolation and determinism  Variation in suitability of designs for use in avionics  Shared resources provide multi- core interference potential  Analysis of behaviour and worst- case execution timing (WCET) is more complex  Processor design information may not be available for hardware analysis Notional Multi-core Processor Architecture with Shared L2 Cache Core #1 L1 I-Cache L1 D-Cache Core #2 L1 I-Cache L1 D-Cache L2 Cache Coherency Module
  • 6. 6 © 2018 WIND RIVER. ALL RIGHTS RESERVED. RELATIONSHIPS AND INDUSTRY INVOLVEMENT  Engage OEMs, civil, and military certification authorities – Start early to help them understand issues – Agree how to address each one  Use proprietary relationships – Operating system (OS) vendors – System-on-chip (SoC) vendors  Aligned with the EASA multi-core certification review item (CRI) and FAA certification authorities software team CAST-32A position paper  Engage internal subject matter experts to make the leap to multi-core systems  Use of certification authorities and certification liaisons to help guide efforts
  • 7. 7 © 2018 WIND RIVER. ALL RIGHTS RESERVED. SELECTING THE RIGHT MULTI-CORE PROCESSOR: MULTICORE FOR AVIONICS (MCFA)  Rockwell Collins MCFA involvement – Founding member and executive board member – Multiple individuals engaged from both Commercial and Government Systems businesses  Processor assessments conducted by Rockwell Collins – Reviews conducted at supplier’s site – Based on a set of questions and artifacts established by MCFA – Provides insight into vendor’s processes for development, production, and quality – Freescale/NXP QorIQ P-Series assessment completed summer 2011 – Freescale/NXP QorIQ T-Series assessment completed summer 2016  FSL became part of NXP Dec 2015 NXP public presentation on MCFA https://community.nxp.com/docs/DOC-331635
  • 8. 8 © 2018 WIND RIVER. ALL RIGHTS RESERVED. SELECTING THE RIGHT MULTI-CORE PROCESSOR: PROCESSING PLATFORM DEFINITION  Multi-core common processing resource (MCPR) – Common reusable logical and schematic design – Can be easily laid out repeatedly as necessary for various form factors  Includes – Multi-core processor – System memory – DAL A hardware monitor FPGA and software monitors – Ethernet networking – Basic I/O (serial and discretes) – Non-volatile storage (NAND, NOR, nvROM)  4x existing product line performance
  • 9. 9 © 2018 WIND RIVER. ALL RIGHTS RESERVED. SELECTING THE RIGHT MULTI-CORE PROCESSOR: PROCESSOR SELECTION  Freescale Quad-Core T2080 QorIQ SoC has been chosen for the MCPR – Process can be iterative as deep dives occur – Switch from previous plan to use the NXP P-series  T2080 provides significant processing advantage – E6500 Core improves integer and floating point processing (2x improvement) – About 5.5x DMIPS performance over baseline processor  Higher-performance internal SoC interconnect fabric significantly reduces interference  Additional I/O controllers provide better mitigation options to reduce interference  Performance enhancements, i.e., AltiVec™ and dual threading at no added power NXP QorIQ® P4080 Processor (Image: NXP) NXP QorIQ® T2080 Processor (Image: NXP) Peripheral Access Management Unit0 128 KB Backside L2 Cache Power Architecture e500mc Core 32 KB D-Cache 32 KB I-Cache 1024 KB CoreNet Platform Cache 1024 KB CoreNet Platform Cache 64-bit DDR2/3 Memory Controller with ECC 64-bit DDR2/3 Memory Controller with ECC CoreNet Coherency Fabric PAMU PAMU PAMU PAMU Security Fuse Processor Security Monitor 2x USB 2.0 w/ULPI eLBC Power Management SD/MMC 2x DUART 1x 12C SPI, GPIO SEC 4.0 PME 2.0 Queue Mgr. Buffer Mgr. Frame Manager Parse, Classify, Distribute 10 G 1 G 1 G 1 G 1 G Frame Manager Parse, Classify, Distribute 10 G 1 G 1 G 1 G 1 G Real-Time Debug 2x DMA PCIe PCIe PCIe sRIO sRIO Watchpoint Cross Trigger Perf. Monitor Trace Aurora 18 Lanes, 5 GHz SerDes Core Complex (CPU, L2 and Frontside CoreNet Platform Cache) P4080 and P4081 Only Networking Elements P4080 and P4040 Only Accelerators and Memory Control Basic Peripherals and Interconnect Peripheral Access Management Unit 2 MB Banked L2 521 KB Platform Cache 64-bit DDR2/3 Memory Controller Coherency Fabric PAMU PAMU PAMU Security Fuse Processor Security Monitor IFC Power Management SDXC/eMMC 2x DUART 4x 12C SPI, GPIO 2x USB 2.0 + PHY DCE 1.0 PME 2.1 Security 5.2 (XoR, CRC) RMan Frame Manager Parse, Classify, Distribute hIGIG/+ 4x 1 / 2.5 / 10 G DCB 2x 1 / 2.5 G Real-Time Debug Watchpoint Cross Trigger Perf. Monitor Trace Aurora Core Complex (CPU, L2, L3 Cache) Accelerators and Memory Control Basic Peripherals and Interconnect Networking Elements Pre-fetch 32 KB D-Cache 32 KB I-Cache 32 KB D-Cache 32 KB I-Cache 32 KB D-Cache 32 KB I-Cache 32 KB D-Cache 32 KB I-Cache PowerTM E6500 T1 T2 PowerTM E6500 T1 T2 PowerTM E6500 T1 T2 PowerTM E6500 T1 T2 Peripheral Access Management Unit Queue Mgr. Buffer Mgr. 8ch DMA PCIe PCIe PCIe PCIe sRIO sRIO 8ch DMA 8ch DMA SATA2.0 SATA2.0 8-Lane 1- GHz SerDes 8-Lane 1- GHz SerDes
  • 10. 10 © 2018 WIND RIVER. ALL RIGHTS RESERVED. REQUIREMENTS FOR A SAFETY-CRITICAL OPERATING PLATFORM  Open architecture reduces system integration costs  Preserves investment in existing applications Multiple Guest OS  Consolidates onto a single platform, reducing size, weight, and power (SWaP) requirements Consolidation  Enables flexible configuration using efficient hardware virtualization support Virtualization Support  Enables hosting of portable POSIX®, ARINC 653, and FACE™ applications Open Standards Flexible Business Model  Supports product line approach and multiple customers without additional licensing overhead Safety Certification  Low perceived DO-178C certification risk  Supports incremental certification
  • 11. 11 © 2018 WIND RIVER. ALL RIGHTS RESERVED. SELECTING THE RIGHT OS PLATFORM FOR MULTI-CORE CERTIFICATION  Wind River safety-critical track record: – 2000: VxWorks Cert for federated applications – 2002: VxWorks 653 for IMA applications  Integrated modular avionics – Enables many federated systems to be hosted on a common computing platform – Reduces SWaP  VxWorks 653 – Supports ARINC 653, POSIX®, VxWorks, and Ada applications – Provides COTS DO-178 certification packages – First OS to achieve conformance certification for FACE™ OSS Safety Base Profile VxWorks and VxWorks 653 cumulative design wins (federated and IMA) Cumulative Federated Cumulative IMA Projects
  • 12. 12 © 2018 WIND RIVER. ALL RIGHTS RESERVED. SELECTING THE RIGHT OS PLATFORM FOR MULTI-CORE CERTIFICATION  High-level design goals for VxWorks 653 Multi-core Edition • COTS RTCA DO-178C DAL A certification evidence • Support of multiple DALs on multiple cores • Perform static configuration and enforcement as per ARINC 653 • Perform fault isolation and containment (health monitors) • Enable IMA role-based development and delivery as per RTCA DO-297 • Robust partitioning of application and OS environments for ease of updates and reduced certification burden
  • 13. 13 © 2018 WIND RIVER. ALL RIGHTS RESERVED. VxWorks 653 Application Executive Multi-Core Hardware Hardware Virtualization Support Board Support Architecture Support Avionics Bus (MIL-STD-1553B, ARINC 429, ARINC 664, SAE AS6802…) XML Configuration Data Core 1Core 0 Core 2 Core 3 Flight Management Application DAL B VxWorks Partition OS Flight Display Application DAL A ARINC 653 Guest OS Sensor Intelligence Application DAL C VxWorks Partition OS New Applications DAL E Linux Guest OS Legacy Applications DAL E Legacy Guest OS VxWorks 653 Multi-Core Edition system architecture Virtual Machines Virtualized partition Guest Operating Systems and applications Enables consolidation of applications onto common platform Reduces program migration and lifecycle costs compared to multiple LRUs Type-1 Hypervisor-based module operating system Provides lightweight supervision of multiple cores Utilizes full hardware virtualization assistance Implements ARINC 653 health management Independent Build Link and Load (IBLL) Supports role-based development as per DO-297 XML system configuration and DO-178C qualified XML-to-binary compiler Minimises impact of change Significantly reduces cost of incremental certification SELECTING THE RIGHT OS PLATFORM FOR MULTI-CORE CERTIFICATION
  • 14. 14 © 2018 WIND RIVER. ALL RIGHTS RESERVED.14 © 2018 WIND RIVER. ALL RIGHTS RESERVED. COMPLYING WITH FAA CAST-32A MULTI-CORE OBJECTIVES  Initial lack of FAA and EASA formal policy on multi-core processors  Publication of FAA CAST-32A multi-core objectives in 2016  Rockwell Collins and Wind River collaborative approach  Positive feedback on approach from multiple certification agencies
  • 15. 15 © 2018 WIND RIVER. ALL RIGHTS RESERVED. COMPLYING WITH FAA CAST-32A MULTI-CORE OBJECTIVES Objective Why It’s Important MCP_Planning_1: Plans identify MCP software architecture (including dynamic and IMA aspects).  The plans provide the overall system design context for the certifying authority. MCP_Planning_2: Plans provide a high-level description of how MCP shared resources and dynamic features will be used and how the applicant intends to allocate and verify the use of shared resources.  Planning shows that the applicant has thought through critical multi-core issues that could impact the execution of the application software. MCP_Resource_Usage_1: The applicant has determined and documented the MCP configuration settings.  MCPs are extremely complex and likely have configuration settings that could negatively impact system safety. MCP_Resource_Usage_2: The applicant has planned, developed, documented, and verified a means that ensures that in the event of any of the critical configuration settings of the MCP being inadvertently altered, an appropriate means of mitigation is specified.  Configuration settings could be inadvertently modified by software errors or single event upsets in ways that result in undefined behavior if not mitigated.
  • 16. 16 © 2018 WIND RIVER. ALL RIGHTS RESERVED. COMPLYING WITH FAA CAST-32A MULTI-CORE OBJECTIVES Objective Why It’s Important MCP_Resource_Usage_3: The applicant has identified the interference channels and has verified the means of mitigation of the interference.  Interference channels are a source of jitter and performance degradation and may have a significant, negative impact on the determinism of the processing system. MCP_Resource_Usage_4: The applicant has identified, allocated, and verified that the available resources of the MCP and of its interconnect are sufficient to meet the demands of the integrated software.  MCP shared resources could be oversubscribed by the full collection of hosted software, resulting in degradation of expected functionality. MCP_Software_1: There is verification that all the hosted software components function correctly and have sufficient time to complete their execution when all the hosted software is executing in the intended final configuration.  Software applications running simultaneously on different cores impact each other’s execution timing and need to be integrated together to understand the impacts to operational behavior.
  • 17. 17 © 2018 WIND RIVER. ALL RIGHTS RESERVED. COMPLYING WITH FAA CAST-32A MULTI-CORE OBJECTIVES Objective Why It’s Important MCP_Software_2: Verification that the data and control coupling is correct during software requirement- based testing.  Data and control coupling across cores is more complex than coupling across partitions on a single-core processor and may result in unintended behavior if not verified to be correct. MCP_Error_Handling_1: Identification of the effects of failures that may occur within the MCP and plan, design, implement, and verify means by which to detect and handle those failures in a fail-safe manner.  The high level of integration of device functions and peripheral interfaces typically found within multi-core SoC designs drives a need for more built-in- test and monitoring for desired behavior. MCP_Accomplishment_Summary_1: The applicant has summarized in their SAS, HAS, or other deliverable documentation how they have met each of the objectives of this document.  Provides a reference for the evidence developed to build assurance that the MCP system design is appropriate for safety-critical use.
  • 18. 18 © 2018 WIND RIVER. ALL RIGHTS RESERVED. MULTI-CORE CERTIFICATION APPROACH— EXECUTION FLOW  Much of the multi-core–specific certification effort executes in parallel with typical engineering development flow  Feedback loops need to exist – Interference research may drive system or software requirements – Changes in system functional requirements may drive new interference channels or alter existing channels Research Processor Design Assess Non-Determinism & Mitigation Techniques Create CAST-32A Plan Define Configuration Analyze Interference Channels & Mitigate Impacts Perform Partitioning & Shared Memory Analyses Capture CAST-32A Accomplishments in HAS/SAS Functional System Requirements Develop System Verify System
  • 19. 19 © 2018 WIND RIVER. ALL RIGHTS RESERVED. MULTI-CORE CERTIFICATION APPROACH— DETERMINISM ANALYSIS  Determinism analysis addresses many CAST-32A concerns  The determinism analysis is split into five activities: – Configuration analysis – Interference channel analysis – Partition analysis – Shared memory analysis – Errata analysis  Configuration analysis documents SoC settings  Interference channel analysis quantifies the behavior of the SoC  Partition/share memory analysis ensures that the software architecture results in robust behavior  Errata analysis demonstrates maturity of the platform and addresses reported problems
  • 20. 20 © 2018 WIND RIVER. ALL RIGHTS RESERVED. RESULTS AND SUMMARY  Analysis of SoC processors is extremely complex  Interference in multi-core platforms is real and impactful, but it can be measured, characterized, and mitigated – For the chosen SoC, the behavior and impacts are consistent with expectations – Behavior is extremely dependent on SoC configuration and usage – It is also dependent on application resource usage  Final configuration rests with the integrator, but the approach taken can minimize those responsibilities – Analysis tools and integration guides make that job even easier  Certification is progressing with SOI4 submittal later this year and TSO early next year
  • 21. 21 © 2018 WIND RIVER. ALL RIGHTS RESERVED. FURTHER INFORMATION  Rockwell Collins and Wind River resources: https://resources.windriver.com/wind-river-rockwell-collins  Continue the conversation in our next web seminar: – Certifying Avionics COTS Hardware & Software to DAL A http://bit.ly/certify-cots
  • 22. © 2018 WIND RIVER. ALL RIGHTS RESERVED. Q&A
  • 23. © 2018 WIND RIVER. ALL RIGHTS RESERVED. BACKUP SLIDES
  • 24. Audience Q & A Harold G. Tiedeman, Jr., Technical Fellow, Rockwell Collins Paul J. Parkinson, Principal Systems Architect, Wind River
  • 25. Thanks for joining us Event archive available at: http://ecast.opensystemsmedia.com/ E-mail us at: jgilmore@opensystemsmedia.com