1 / 25

Shlomi Dolev and Reuven Yagel,

S. Self-stabilizing. Operating System. O S. SSS 06. Self-Stabilizing Device Drivers. Shlomi Dolev and Reuven Yagel, Computer Science Department Ben-Gurion University of the Negev, Beer-Sheva, Israel {dolev|yagel}@cs.bgu.ac.il. 11/06. Motivation for S elf-stabilizing OS.

bryanl
Download Presentation

Shlomi Dolev and Reuven Yagel,

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. S Self-stabilizing Operating System O S SSS06 Self-Stabilizing Device Drivers Shlomi Dolev and Reuven Yagel, Computer Science Department Ben-Gurion University of the Negev, Beer-Sheva, Israel {dolev|yagel}@cs.bgu.ac.il 11/06

  2. Motivation forSelf-stabilizing OS • Coping with transient faults(e.g. soft-error (SEU) in a remote sensor results unpredictable system states) • Growing interest in self-* / autonomic computing systems • Self-stabilizing algorithms/programs assume hardware and operating system are also stabilizing • Pentium HALTING problem: “… if the ESP or SP register is 1 when the PUSH instruction is executed, the processor shuts down…”

  3. Proposed solution • To build according to the well defined and understood paradigm of self-stabilization (traditionally used in distributed systems) • Thereby achieving: trustworthiness, dependability, self-healing, automatic recovery, adaptive systems, …

  4. OS Dependability • Past examples: • Dijkstra, “THE” Multiprogramming System ’68 (Layered Approach) • Denning, Lampson ~’70 (Protection) • KeyKOS ’85 (Capabilities) • EROS ’92 (Checkpoints) • Micro-kernel ~‘90, Exo-kernel ’94 (Minimal TCB) • Recent: • IBM: K42 (Autonomic Computing), SUN: Solaris 10, Predictive Self-Healing • MSR: Singularity (managed code, protection, verification) • JHU: The Coyotos Secure Operating System (Capabilities, type safety) • HW e.g. Intel : Itanium 2, Machine Check Arch.

  5. Goal: Autonomic Computer • Following any sequence of transient faults, the (operating) system converges • Using self stabilization: • A system can be started in an arbitrary state and converge to a desired behavior • Using fair composition to run hardware + OS layaers • BGU: Self-stabilizing systems, tools & paradigms • Microprocessor [DH04] • Operating System [DY04, DY05, DY06] • Compiler [DH05] • Framework: autonomic recoverer [BDK03, BD06] • Middleware: File System [DK02], Group Comm. [DS01]

  6. SOS - Directions • Black-box • Take existing (Desktop/Real-time) OS • Add stabilization layer • Detailed formal specification needed • Carefully tailoring a tiny kernel • Processor scheduling [SAACS04] • Memory management [SOSP05, SSS05] • Device drivers [SSS06]

  7. A problem has been detected and Windows has been shut down to prevent damage to your computer. PFN_LIST_CORRUPT If this is the first time you've seen this error screen, restart your computer. If this screen appears again, follow these steps: Check to make sure any new hardware or software is properly installed. If this is a new installation, ask your hardware or software manufacturer for any Windows updates you might need. If problems continue, disable or remove any newly installed hardware or software. Disable BIOS memory options such as caching or shadowing. If you need to use Safe Mode to remove or disable components, restart your computer, press F8 to select Advanced Startup Options, and then select Safe Mode. Technical information: *** STOP: 0x0000004e (0x00000099, 0x00000000, 0x00000000, 0x00000000) Beginning dump of physical memory Physical memory dump complete. Contact your system administrator or technical support group for further assistance. /* Third Edition UNIX, early 1973 */ panic(s) char *s; { prproc(); update(); printf("panic: %s\n", s); for(;;) idle(); }

  8. Assumptions • Every configuration (processor registers/ memory/IO controllers) is possible • (Some) program code is hardwired (in ROM) and is correct [Castro & Liskov, 2000] • Stabilization of other layers • CPU instruction manual (e.g. Pentium) defines a transition function • Restrict access to privileged instructions and registers (e.g. HLT, LIDT) • Segmentation for enforcing isolation

  9. Method • Specify main (additional) requirements for each OS function • Evolve self-stabilizing solutions that follow computer-architecture/OS progress • Detailed proof for self-stabilization of algorithms AND implementation • Using processor instruction set • Don’t rely on existing compilers

  10. Device Driver Challenges • Practically an essential part of any OS • Co-operation with external controllers • Adaptation (connecting interfaces) • Translation (abstracting e.g. speed variation) • (and sometimes adding functionality) • ”a modern Seagate drive contains roughly 400,000 lines of code” [IRON FS] • “modern disk controllers often have many megabytes of memory inside the controller” [Minix]

  11. Unreliable Components • “In Windows XP, for example, device drivers cause 85% of reported failures” [Nooks] • ”Since drivers account for the majority of the code (over 70% in this release)… …the error rate for drivers is almost seven times higher than the error rate for the rest of the kernel” [Engler et. al.]

  12. Where is the Problem? • Device drivers are loaded into kernel and run in privileged mode • No monitoring of operation & resource usage • Usually external & less tested code • Not-stabilizing! • "malicious driver can program a DMA capable device to overwrite any part of memory...We anticipate that future hardware will provide memory protection for DMA transfers." [Singularity]

  13. What Has Been Done? • Isolation: Micro-kernel [Mach], user mode drivers [Minix], Monitoring [Nooks] • Virtualization [Xen, …] • Type safety [Singularity, Coyotos] • Model Checking [SLAM] • ”our goal is to prevent the vast majority of driver-caused crashes with little or no change to existing driver” [Nooks]

  14. MSR Sinuglarity Project • Achieving OS dependability through advances in languages (compilers), tools (verifiers and monitors). • Device drivers isolation through Software Isolated Processes (SIP) and resource useage specification • Recently adding hardware protection:“increased use of hardware error detection and correction techniques, such as error correcting codes (ECC) on memory and data paths, to ensure the assumption of correct execution that underlies all program safety” [MSR-TR-2006-43]

  15. Example: ATA Standard • 1990 ATA-1 a de-facto protocol for storage devices (ANSI 1994) • T13 Technical Committee http://www.t13.org/#Projects • Current draft: ATA-8 • ATA ZOO: AT-Attachment, (E-)IDE, ATAPI, SATA/PATA

  16. ATA Model

  17. ATA Model Host: :Device

  18. ATA Non-stability

  19. Device-Driver Requirements • Error model: state of device driver & controller program (including PC) might get any value • Ping-pong: infinitely I/O commands in which there is update & read of the device buffer • Progress: eventually every (and only) I/O request is executed completely and correctly according to the ATA spec.

  20. Solution 1: Leasing

  21. Solution 1: Leasing • The driver reaches its idle state infinitely often • From any configuration a safe (idle) configuration is eventually reached – device reset • From there ping-pong holds • Eventually progress holds

  22. Solution 2: Consistency Checks • Device driver contains consistency checker • Consistency of driver current state with a snapshot of the controller

  23. Solution 2: Consistency Checks • Eventually ping-pong holds • Eventually progress holds • Combining the 2 solutions: • varying snapshot sizes according to degree of device stabilization • fallback to leases and restarts

  24. Future Work • I/O device drivers • Protocols for co-operation of more than one microprocessor • Detailed driver / General monitoring layer • Gather the different parts

  25. Conclusion • The work shows theoretical and practical ways to achieve the goal of a self-stabilizing OS, particularly the device drivers part • By supplying an infrastructure for practical self-stabilizing systems, robust and dependable systems can be achieved • http://www.cs.bgu.ac.il/~yagel/sos

More Related