Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 42 views

    A Computational Model For SAT-based Verification of Hardware-Dependent Low-Level Embedded System Software

    Uploaded by

    nguyenthanhdt5
    The document presents a computational model for SAT-based formal verification of hardware-dependent low-level embedded system software. The model generates a program netlist (PN) that represents the program flow explicitly through a combinational circuit of instruction cells, enabling SAT-based reasoning. The PN model has several advantages over prior approaches: it represents program computation implicitly through instruction cell concatenation; simplifies individual instructions through constant propagation; and uses active signals to capture global reachability, improving scalability for verification.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    A Computational Model For SAT-based Verification of Hardware-Dependent Low-Level Embedded System Software

    Uploaded by

    nguyenthanhdt5
    42 views25 pages
    The document presents a computational model for SAT-based formal verification of hardware-dependent low-level embedded system software. The model generates a program netlist (PN) that represents the program flow explicitly through a combinational circuit of instruction cells, enabling SAT-based reasoning. The PN model has several advantages over prior approaches: it represents program computation implicitly through instruction cell concatenation; simplifies individual instructions through constant propagation; and uses active signals to capture global reachability, improving scalability for verification.

    Original Description:

    abc

    Original Title

    8D-2

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document presents a computational model for SAT-based formal verification of hardware-dependent low-level embedded system software. The model generates a program netlist (PN) that represents the program flow explicitly through a combinational circuit of instruction cells, enabling SAT-based reasoning. The PN model has several advantages over prior approaches: it represents program computation implicitly through instruction cell concatenation; simplifies individual instructions through constant propagation; and uses active signals to capture global reachability, improving scalability for verification.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    42 views25 pages

    A Computational Model For SAT-based Verification of Hardware-Dependent Low-Level Embedded System Software

    Uploaded by

    nguyenthanhdt5
    The document presents a computational model for SAT-based formal verification of hardware-dependent low-level embedded system software. The model generates a program netlist (PN) that represents the program flow explicitly through a combinational circuit of instruction cells, enabling SAT-based reasoning. The PN model has several advantages over prior approaches: it represents program computation implicitly through instruction cell concatenation; simplifies individual instructions through constant propagation; and uses active signals to capture global reachability, improving scalability for verification.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 25

    A Computational Model for

    SAT-based Verification
    of
    Hardware-Dependent Low-Level
    Embedded System Software
    Bernard Schmidt, Carlos Villarraga, Jrg Bormann,

    Dominik Stoffel, Markus Wedler, Wolfgang Kunz

    Yokohama, 1/25/2013

    Content

    Motivation
    Related Works
    Model Generation
    Basis: Abstract HW/SW Model
    Flow
    Advantages of Model
    Experiment
    Conclusion / Future Work

    Motivation

    Embedded System

    Goal

    Close interaction
    between HW and SW
    Examples: drivers,
    communication
    structures

    Formal verification (FV) of combined HW/SW behavior

    Objective of this work

    Computational model and algorithms for


    FV of hardware-dependent, low-level software

    Related Works

    [2] R. Jhala and R. Majumdar, Software model checking, ACM Comput. Surv., vol. 41, pp. 21:121:54, October 2009.
    [3] T. Ball, A. Podelski, and S. K. Rajamani, Boolean and Cartesian Abstraction for Model Checking C Programs, in Proceedings of the
    7th International Conference on Tools and Algorithms for the Construction and Analysis of Systems, ser. TACAS 2001. London, UK:
    Springer-Verlag, 2001, pp. 268283.
    [4] D. Beyer, T. A. Henzinger, R. Jhala, and R. Majumdar, The software model checker Blast: Applications to Software Engineering, Int.
    J. Softw. Tools Technol. Transf., vol. 9, pp. 505525, October 2007.
    [5] P. Godefroid, Model
    Checking for Programming Languages using VeriSoft, in Proceedings of the 24th ACM SIGPLAN-SIGACT
    Sympo-sium on Principles of Programming Languages, ser. POPL 97. New York, NY, USA: ACM, 1997, pp. 174186.
    [6] K. Havelund and Thomas Pressburger, Model Checking Java Programs Using Java PathFinder, International Journal on Software
    Tools for Technology Transfer STTT, vol. 2, no. 4, pp. 366381, 2000.
    [7] G. J. Holzmann, The model checker spin, IEEE Transactions On Software Engineering, 1997.
    [8] D. Babic and A. J. Hu, Calysto: scalable and precise extended static checking, in Proceedings of the 30th International Conference on
    Software Engineering,
    ser. ICSE 08. New York, NY, USA: ACM, 2008, pp. 211220.
    [9] F.Ivancic, Z. Yang, M. Ganai, A. Gupta, and P. Ashar, Efficient SATbased bounded model checking for software verification, in Proc.
    In- ternational Symposium on Leveraging Applications of Formal Methods, 2004.
    [10] E. Clarke, D. Kroening, and K. Yorav, Behavioral consistency of C and Verilog programs using Bounded Model Checking, in
    Proceedings of the 40th annual Design Automation Conference, ser. DAC 03. New York, NY, USA: ACM, 2003, pp. 368371.
    [11] D. W. Currie, A. J. Hu, and S. Rajan, Automatic formal verification of DSP software, in Proceedings of the 37th Annual Design
    Automation Conference, ser. DAC 00. New York, NY, USA: ACM, 2000, pp. 130135.
    [12] B. Schlich, Model checking of software for microcontrollers, ACM Trans. Embed. Comput. Syst., vol. 9, no. 4, pp. 36:136:27, Apr.
    2010.
    [13] C. S. Psreanu and W. Visser, A survey of new trends in symbolic execution for software testing and analysis, Int. J. Softw. Tools
    Technol. Transf., vol. 11, no. 4, pp. 339353, Oct. 2009.
    [14] T. Arons, E. Elster, S. Ozer, J. Shalev, and E. Singerman, Efficient symbolic simulation of low level software, in Design, Automation
    and Test in Europe, 2008. DATE 08, march 2008, pp. 825 830.
    [15] M. D. Nguyen, M. Wedler, D. Stoffel, and W. Kunz, Formal Hardware/Software Co-Verification by Interval Property Checking with
    Abstraction, in Proceedings of the 48th Design Automation Conference, ser. DAC 11. New York, NY, USA: ACM, 2011, pp. 510515.
    [16] D. Groe, U. Khne, and R. Drechsler, HW/SW co-verification of embedded systems using bounded model checking, in GLSVLSI
    06: Proceedings of the 16th ACM Great Lakes symposium on VLSI, 2006, pp. 4348.
    [17] A. Biere, A. Cimatti, E. M. Clarke, M. Fujita, and Y. Zhu, Symbolic model checking using SAT procedures instead of BDDs, in Proc.
    International Design Automation Conference (DAC), June 1999, pp. 317320

    Large research body


    on SW verification

    Our focus:
    New computational model for
    hardware-dependent SW

    State of the Art: Bounded Model Checking (BMC)


    E.g.,
    M. D. Nguyen, M. Wedler, D. Stoffel, and W. Kunz, Formal Hardware/Software CoVerification by Interval Property Checking with Abstraction DAC 2011.
    D. Groe, U. Khne, and R. Drechsler, HW/SW co-verification of embedded systems
    using bounded model checking in GLSVLSI 2006.

    Basic approach:
    Unroll CPU + software in memory

    program flow (instruction sequencing) as given by the


    program is represented only implicitly (by PC + memory)

    program computation also is represented only implicitly


    (by hardware implementation of CPU)

    High computational complexity for realistic designs


    5

    State of the Art: Symbolic Execution


    E.g.,
    C. S. Psreanu and W. Visser, A survey of new trends in symbolic execution for
    software testing and analysis Int. J. Softw. Tools Technol. Transf., 2009.
    T. Arons, E. Elster, S. Ozer, J. Shalev, and E. Singerman, Efficient symbolic simulation
    of low level software DATE 2008

    Basic approach

    Enumerate program execution paths and check conditions for


    verification along these paths by specialized verification algorithms

    Symbolic formulas explicitly represent all possible input scenarios


    (along path)

    program flow is enumerated explicitly (by path enumeration)

    program computation is represented explicitly (through symbolic


    formulas)

    high complexity of symbolic formulas


    6

    State of the Art: CBMC and related work


    E.g.,
    E. Clarke, D. Kroening, and K. Yorav, Behavioral consistency of C and Verilog programs
    using Bounded Model Checking DAC 2003

    Basic approach

    Build a SAT formula for the program computation based on


    CFG unrolling

    SAT formula contains information about control flow only


    implicitly in terms of functional dependencies between
    variables of the formula

    HW-independent paradigm (high-level programming language)

    State of the Art: CBMC and related work


    E.g.,
    E. Clarke, D. Kroening, and K. Yorav, Behavioral consistency of C and Verilog programs
    using Bounded Model Checking DAC 2003

    Extension of CBMC for HW-dependent software is difficult!


    The computational model (SAT formula):

    lacks information about temporal relationships between


    statements/instructions as is needed when relating program
    behavior to HW periphery

    cannot be integrated with environment hardware model

    is not compositional
    8

    Our Approach
    New hardware-dependent computational model

    Program computation should be represented implicitly


    compact model

    Program flow should be represented explicitly


    facilitates SAT reasoning (since global information about
    global execution paths is explicitly available)

    Representation of the temporal dependencies between


    instructions and I/O
    enables HW-dependent verification
    enables compositionality of model
    9

    Basis: Abstract HW/SW Model

    Input Ports
    Program

    Instruction
    Logic

    State
    (AS, memory
    variables)

    w
    Program Control

    Program
    Counter

    jump

    Data Path

    destination

    Output Ports

    Program

    10

    Basis: Abstract HW/SW Model


    Instruction Logic:
    Input Ports

    ADD
    Program

    PS

    Output
    Ports
    Output
    Ports

    MUL

    PS

    MOV

    ...

    (PC) a

    ROT

    Data Path

    w
    LOAD
    Program Control

    (icode) w

    jump

    Input Ports

    State
    (AS, memory
    variables)

    Instruction
    Logic

    destination

    BEQ
    Program
    Counter

    w
    Program

    J
    11

    Instruction Cell
    program state

    Abstract model for a CPU


    instruction

    Hardware-dependent

    Describes the modification of the


    program state

    Can be formally verified against


    RTL implementation of CPU

    new
    program state

    12

    Program State
    program state

    Registers

    Control signals in the CPU

    Program data in RAM

    Active signal (program flow


    variable)

    new
    program state

    13

    Active Signal

    Additional 1-bit signal in program state

    Signal is asserted if corresponding program state


    belongs to active execution path

    Additional logic in instruction cells to handle active


    signal:

    Datapath and load/store instructions:


    propagate active flag from input state to output state
    (no extra logic required)
    Branch instructions: propagate active flag depending on
    branch condition from input state to branch target output
    state. All other output states are set inactive.
    14

    Transform CFG to Execution Graph (EXG)


    CFG

    1st step

    2nd step

    Final EXG

    always
    inactive?

    Execution graph:

    Models all possible execution paths


    Optimized by pruning and merging
    Not unique

    always
    inactive?

    always
    inactive!
    15

    Verifying if a branch is active

    Branch instruction

    Propagates active signal to exactly


    one of the branches

    program state

    Generated property:

    Claim: active signal is never


    asserted
    Property holds: branch never taken
    Property does not hold: branch is
    taken in some executions

    taken

    not
    taken

    16

    Program Netlist (PN)

    PN generation

    Replace every EXG node by an


    instruction cell
    HW-dependent description of
    program computation
    Insert merge cells

    Merge cell

    Merges two paths into one


    Path is selected by incoming
    active signal
    17

    Advantages of PN Model
    Replacing EXG nodes by instruction cells

    PN is a combinational circuit

    CFG-based unrolling

    Supports SAT-based FV

    Explicitly represented program flow

    Concatenation of instruction cells

    Implicitly represented program computation


    18

    Advantages of PN Model (2)


    Explicit program flow:
    program paths are considered separately

    Simplification of program computation

    CPU logic is simplified to single instruction cell


    Machine instruction parameters are constant:
    individual instruction cell is further simplified by
    Boolean constraint propagation
    Simplification takes place during PN generation
    (not during property verification)
    19

    Advantages of PN Model (3)

    Active flags:

    Small set of signals captures global reachability


    information of entire program
    Valuation on active signals selects possible execution
    paths or sets of execution paths
    Helps SAT solver to take into account reachability without
    detailed reasoning on all intermediate instructions

    Example:
    a

    Property:
    A: = true
    C: if b.active
    then b=a

    20

    Experiment Software LIN Master (1)

    Industrial automotive design

    1309 lines of C code

    Send/receive message frames via UART

    Interrupt-driven

    Compiled for SuperH2 based 32-bit CPU, RISC


    architecture, 5-stage pipeline

    21

    Experiment Software LIN Master (2)

    Model generation:

    Interrupt Service Routine (ISR), main program


    (scheduler) and initialization generated automatically
    Composed model: assembled manually

    Program
    component
    LIN-Init
    LIN-Scheduler
    LIN-ISR

    #instructions
    CFG
    PN
    225
    385
    85
    790

    84
    1138

    CPU mem.
    (s) (MB)
    1.32
    36
    0.13
    11.00

    27
    102

    22

    Experiment Software LIN Master (3)


    Property verification:

    Composed model: 24001 instructions, 62592 primary inputs


    Verify correct generation of LIN frame
    (header, data, checksum and control signal for UART)
    Commercial HW property checker (OneSpin 360MV)
    Property
    RX frame 4or8 data bytes incl. checksum
    RX frame 4or8 data bytes wrong checksum
    TX frame 4or8 data bytes incl. Checksum
    Wrong PID or not matching ID (8bytes)

    CPU(s)

    MEM(MB)

    17
    28
    15
    14

    1641
    1545
    1584
    1566

    Conclusion

    Program netlist (PN)

    New hardware-dependent model for low-level software


    verification combining the advantages of HW-style
    bounded model checking and SW-style symbolic
    execution
    Automatic PN generation successful for industrial
    automotive software

    Future Work

    Equivalence checking
    Integrate PN with hardware for FV of firmware

    24

    Thank you!

    Questions?

    bschmidt@eit.uni-kl.de

    You might also like