MINIX3: A Reliable and Secure Operating System
Andrew S. Tanenbaum
and a team of students and programmers who actually did all the work
Vrije Universiteit Amsterdam, The Netherlands
1 GOAL OF OUR WORK: BUILD A RELIABLE OS
Tanenbaum’s definition of a reliable OS:
“An operating system is said to be reliable when a typical user has never experienced even a single failure in his or her lifetime and does not know anybody who has ever experienced a failure.”
In engineering terms, this is probably mean time to failure > 50 years
I don’t think we are there yet
2 THE TELEVISION MODEL
1. You buy the television 2. You plug it in 3. It works perfectly for the next 10 years
3 THE COMPUTER MODEL (WINDOWS EDITION)
1. You buy the computer 2. You plug it in 3. You install service packs 1 through 9f 4. You install 18 new emergency security patches 5. You find and install 7 new device drivers 6. You install antivirus software 7. You install antispyware software 8. You install antihacker software (firewall) 9. You install antispam software 10. You reboot the computer
4 THE COMPUTER MODEL (2)
11. It doesn’t work 12. You call the helpdesk 13. You wait on hold for 30 minutes 14. They tell you to reinstall Windows
5 TYPICAL USER REACTION
The New York Times recently reported that 25% of computer users have gotten so angry at their computer that they physically hit it.
6 IS RELIABILITY SO IMPORTANT?
• Annoying • Lost work • But also think about – Industrial control systems in factories – Power grids – Hospital operating rooms – Banking and e-commerce servers – Emergency phone centers – Control software in cars, airplanes, etc.
7 IS THIS FEASIBLE?
• We won’t find out if we don’t try • Dutch Royal Academy gave me €2 million to try • European Union gave me €2.5 million to give it a shot • So, we’re trying
8 IS RELIABILITY ACHIEVABLE AT ALL?
• Systems can survive hardware failures! – RAIDs can survive failed disks – ECC memory can survive parity errors in memory – TCP/IP can survive lost packets – CD-ROM drives can correct many simultaneous errors • We need to be able to survive software failures, too
9 A NEED TO RETHINK OPERATING SYSTEMS
• Operating systems research need to be refocused – We have nearly infinite hardware on PC-class machines – Plenty of CPU cycles, RAM, bandwidth – Current software has tons of (useless) features – Consequently, the software is slow, bloated, and buggy
• To achieve the TV model, future OSes, must be – Small – Simple – Modular – Reliable – Secure – Self-healing 10 BRIEF HISTORY OF OUR WORK
• (1976) John Lions wrote a book on UNIX V6 • (1979) AT&T released V7 and forbade books on it L • (1985) I started to write a UNIX-like OS from scratch • (1987) MINIX 1 + book for teaching OS classes released • (1997) MINIX 2 (POSIX) & 2nd edition of book released • (2000) MINIX 2 license changed to BSD • (2004) MINIX 3: start of work making a reliable OS • (2006) 3rd edition of book • (2008) European grant • (2010) Focus moved towards embedded systems • (2013) MINIX 3.3.0 moves to NetBSD “compatibility”
11 THREE EDITIONS OF THE BOOK
1 2 3
12 INTELLIGENT DESIGN AS APPLIED TO OPERATING SYSTEMS
• Microkernel (15,000 LoC vs. > 15 million for Linux) – Bugs per 1000 LoC: Most S/W (1-10) – MINIX 3 at least 15 kernel bugs; Linux has > 15,000 – Drivers have 3-7x more bugs than rest of kernel – About 70% of the code is drivers • Highly modular • OS runs as multiple user-mode server processes
13 STEP 1: ISOLATE COMPONENTS
• Move all loadable modules out of the kernel – includes all device drivers and file systems • Run each module as a separate process with POLA (Principle Of Least Authority)
14 STEP 2: ISOLATE I/O
• Isolate I/O devices • Limit access to I/O ports • Constrain DMA (needs hardware assistance)
15 STEP 3: ISOLATE COMMUNICATION
• Limit interprocess communication • Restrict kernel calls on a per component basis • Restrict IPC on a ‘need-to-communicate’ basis • Make sure faulty receiver cannot hang sender
16 ARCHITECTURE OF MINIX 3
Process
Shell Make ... User
User mode FS 1 FS 2 Proc. ... Other Servers
Disk TTY Net Print ... Other Drivers
Kernel Microkernel handles interrupts, mode processes, scheduling, IPC
17 USER-MODE DEVICE DRIVERS
• Each driver runs as a user-mode process • No superuser privileges • Protected by the MMU • Do not have access to I/O ports, privileged instrs
18 USER-MODE SERVERS
• Each server runs as a separate process • Some key servers – Virtual file server – Actual file servers – Process manager – Memory manager – Network server – Reincarnation server
19 A SIMPLIFIED EXAMPLE: DOING A READ
User Users 1 User 4 mode FS Servers
Disk 2 3 Drivers
Kernel
File access when the block is in the FS cache
20 FILE SERVER (2)
User Users 1 User 9 mode FS Servers 2 6 7,8 5 Disk Drivers 3 4
Notification Kernel
File access when the block is NOT in the FS cache
21 REINCARNATION SERVER
• Parent of all the drivers and servers • When a driver or server dies, RS collects it • RS checks a table for action to take e.g., restart it • RS also pings drivers and servers frequently
22 DISK DRIVER RECOVERY
RS 1 User Users
User 5 4 FS Servers mode 2
New Disk driver driverX 3. Crash! Drivers
Kernel
System is self healing—this is how we hope to make it reliable
23 KERNEL RELIABILITY/SECURITY
• Fewer LoC means fewer kernel bugs • Small kernel (15,000 LoC) means reduced TCB • NO foreign code (e.g., drivers) in the kernel • Static data structures (no malloc in kernel) • Moving bugs to user space reduces their power
24 IPC RELIABILITY/SECURITY
• Fixed-length messages (no buffer overruns) • Rendezvous system was simple – No lost messages – No buffer management – We had to add asynchronous messages • Interrupts and messages are unified
25 DRIVER RELIABILITY/SECURITY
• Untrusted code: heavily isolated • Bugs, viruses cannot spread to other modules • Cannot touch kernel data structures • Bad pointers crash only one driver; recoverable • Infinite loops detected and driver restarted • Restricted power to do damage (not superuser)
26 OTHER ADVANTAGES OF USER DRIVERS
• Short development cycle • Normal programming model • No down time for crash and reboot • Easy debugging • Good flexibility
27 FAULT INJECTION EXPERIMENT
• We injected 800,000 faults into each of 3 drivers • Done on the binary drivers • Examples, change src addr, dest addr, loop condition • 100 faults were injected on each experiment • Waited 1 sec to see if the driver crashed • If no crash, inject another 100 faults and repeat • The driver crashed in 18,038 trials • The operating system NEVER crashed
28 PORT OF MINIX 3 TO ARM
• Restructured source tree for multiple architectures • Changed booting to support uboot for ARM • Rewrote the low-level code dealing with hardware • Changed code for context switching, paging, etc. • Removed x86 segmentation code • Imported NetBSD ARM headers and libraries • Ported build.sh for cross-toolchain support • Wrote drivers for SD card and other Beagle devices
29 EMBEDDED SYSTEMS
BeagleBone Black 5 cm
9 cm
30 CHARACTERISTICS
Item Beaglebone Black
31 CHARACTERISTICS
Item Beaglebone Black CPU ARM v7
32 CHARACTERISTICS
Item Beaglebone Black CPU ARM v7 Clock 1 GHz
33 CHARACTERISTICS
Item Beaglebone Black CPU ARM v7 Clock 1 GHz RAM 512 MB
34 CHARACTERISTICS
Item Beaglebone Black CPU ARM v7 Clock 1 GHz RAM 512 MB Flash 4 GB
35 CHARACTERISTICS
Item Beaglebone Black CPU ARM v7 Clock 1 GHz RAM 512 MB Flash 4 GB Video HDMI/1080p
36 CHARACTERISTICS
Item Beaglebone Black CPU ARM v7 Clock 1 GHz RAM 512 MB Flash 4 GB Video HDMI/1080p GPIO pins 92
37 CHARACTERISTICS
Item Beaglebone Black CPU ARM v7 Clock 1 GHz RAM 512 MB Flash 4 GB Video HDMI/1080p GPIO pins 92 Ethernet 10/100 Mbps
38 CHARACTERISTICS
Item Beaglebone Black CPU ARM v7 Clock 1 GHz RAM 512 MB Flash 4 GB Video HDMI/1080p GPIO pins 92 Ethernet 10/100 Mbps USB 1
39 CHARACTERISTICS
Item Beaglebone Black CPU ARM v7 Clock 1 GHz RAM 512 MB Flash 4 GB Video HDMI/1080p GPIO pins 92 Ethernet 10/100 Mbps USB 1 Open source Yes
40 CHARACTERISTICS
Item Beaglebone Black CPU ARM v7 Clock 1 GHz RAM 512 MB Flash 4 GB Video HDMI/1080p GPIO pins 92 Ethernet 10/100 Mbps USB 1 Open source Yes Price (quantity 1) $45
41 CHARACTERISTICS
Item Beaglebone Black Raspberry Pi B+ CPU ARM v7 Clock 1 GHz RAM 512 MB Flash 4 GB Video HDMI/1080p GPIO pins 92 Ethernet 10/100 Mbps USB 1 Open source Yes Price (quantity 1) $45
42 CHARACTERISTICS
Item Beaglebone Black Raspberry Pi B+ CPU ARM v7 ARM v6 Clock 1 GHz RAM 512 MB Flash 4 GB Video HDMI/1080p GPIO pins 92 Ethernet 10/100 Mbps USB 1 Open source Yes Price (quantity 1) $45
43 CHARACTERISTICS
Item Beaglebone Black Raspberry Pi B+ CPU ARM v7 ARM v6 Clock 1 GHz 700 MHz RAM 512 MB Flash 4 GB Video HDMI/1080p GPIO pins 92 Ethernet 10/100 Mbps USB 1 Open source Yes Price (quantity 1) $45
44 CHARACTERISTICS
Item Beaglebone Black Raspberry Pi B+ CPU ARM v7 ARM v6 Clock 1 GHz 700 MHz RAM 512 MB 512 MB Flash 4 GB Video HDMI/1080p GPIO pins 92 Ethernet 10/100 Mbps USB 1 Open source Yes Price (quantity 1) $45
45 CHARACTERISTICS
Item Beaglebone Black Raspberry Pi B+ CPU ARM v7 ARM v6 Clock 1 GHz 700 MHz RAM 512 MB 512 MB Flash 4 GB None Video HDMI/1080p GPIO pins 92 Ethernet 10/100 Mbps USB 1 Open source Yes Price (quantity 1) $45
46 CHARACTERISTICS
Item Beaglebone Black Raspberry Pi B+ CPU ARM v7 ARM v6 Clock 1 GHz 700 MHz RAM 512 MB 512 MB Flash 4 GB None Video HDMI/1080p HDMI/1080p GPIO pins 92 Ethernet 10/100 Mbps USB 1 Open source Yes Price (quantity 1) $45
47 CHARACTERISTICS
Item Beaglebone Black Raspberry Pi B+ CPU ARM v7 ARM v6 Clock 1 GHz 700 MHz RAM 512 MB 512 MB Flash 4 GB None Video HDMI/1080p HDMI/1080p GPIO pins 92 40 Ethernet 10/100 Mbps USB 1 Open source Yes Price (quantity 1) $45
48 CHARACTERISTICS
Item Beaglebone Black Raspberry Pi B+ CPU ARM v7 ARM v6 Clock 1 GHz 700 MHz RAM 512 MB 512 MB Flash 4 GB None Video HDMI/1080p HDMI/1080p GPIO pins 92 40 Ethernet 10/100 Mbps 10/100 Mbps USB 1 Open source Yes Price (quantity 1) $45
49 CHARACTERISTICS
Item Beaglebone Black Raspberry Pi B+ CPU ARM v7 ARM v6 Clock 1 GHz 700 MHz RAM 512 MB 512 MB Flash 4 GB None Video HDMI/1080p HDMI/1080p GPIO pins 92 40 Ethernet 10/100 Mbps 10/100 Mbps USB 1 4 Open source Yes Price (quantity 1) $45
50 CHARACTERISTICS
Item Beaglebone Black Raspberry Pi B+ CPU ARM v7 ARM v6 Clock 1 GHz 700 MHz RAM 512 MB 512 MB Flash 4 GB None Video HDMI/1080p HDMI/1080p GPIO pins 92 40 Ethernet 10/100 Mbps 10/100 Mbps USB 1 4 Open source Yes No Price (quantity 1) $45
51 CHARACTERISTICS
Item Beaglebone Black Raspberry Pi B+ CPU ARM v7 ARM v6 Clock 1 GHz 700 MHz RAM 512 MB 512 MB Flash 4 GB None Video HDMI/1080p HDMI/1080p GPIO pins 92 40 Ethernet 10/100 Mbps 10/100 Mbps USB 1 4 Open source Yes No Price (quantity 1) $45 $35
52 I ADMIT I WAS WRONG
• On 29 Jan 1992 I posted to comp.os.minix this:
• “Don`t get me wrong, I am not unhappy with LINUX. It will get all the people who want to turn MINIX in BSD UNIX off my back.”
• I Apologize. Now I do want to turn MINIX into BSD. It just took me 20 years to realize this.
53 MINIX 3 MEETS BSD
+ =
BSD Daemon is copyright 1988 by Marshall Kirk McKusick and is used with permission. 54 OR MAYBE
55 WHY BSD?
• MINIX 3 didn’t have enough application software • BSD is a proven, portable, quality product • BSD has better code quality than Linux • Pkgsrc handles packages better than what we had • Thousands of excellent packages available • Active community • License compatibility • Why NetBSD? • Mostly due to its emphasis on portability
56 NETBSD FEATURES IN MINIX 3.3.0
• Clang/LLVM compiler • NetBSD build system • ELF file format • Source code tree modeled on NetBSD • Headers and libraries are from NetBSD • X11 • Pkgsrc works and builds 5040 NetBSD packages • Nevertheless, it is built on MINIX 3 kernel & servers
57 NETBSD FEATURES MISSING IN MINIX 3.3.0
• Kernel threads (we do have userland pthreads) • Some system calls: – All _LWP*, MSG*, SEM* calls – CLONE – Some GET, IOCTL calls – KQUEUE, KTRACE – VFORK – Job control – Some other minor calls • Nevertheless, we can build over 5000 packages
58 KYUA TESTS
Conclusion: 2139 out of 2651 passed (81%)
59 SYSTEM ARCHITECTURE
Users User- Land Clang Pkgsrc (libc) Pkg 1 Pkg n … (NetBSD) Servers VFS FS MM Rein carnat … OS (MINIX) Drivers Disk Net TTY… USB …
Microkernel (this is the only part running in kernel mode)
60 MINIX 3 ON THE THREE BEAGLE BOARDS
61 YOUR ROLE
• MINIX 3 is an open-source project • I hope some of you will join and help us • Things to do – Add crucial missing system calls – Port more packages (Java, a browser, etc.) – Write the missing drivers for Beagle series – Get it running on Raspberry Pi & other platforms – Port Rump – Port required libraries and then port a GUI
62 MINIX 3 IN A NUTSHELL
• Microkernel reimplementation of NetBSD • Fully open source with BSD license • Highly compatible with NetBSD • Supports both LLVM and gcc • Uses NetBSD pkgsrc • Over 5000 packages build • Go get it at www.minix3.org and try it
63 POSITIONING OF MINIX
• Show that multiserver systems are reliable • Demonstrate that drivers belong in user mode • High-reliability and fault-tolerant applications • $50 single-chip, small-RAM laptops for 3rd world • Embedded systems
64 FUTURE FEATURE: LIVE UPDATE
• Software is updated to: – Fix bugs – Improve performance – Add new features • Goal is to update OS to a new version w/o reboot • Running processes must NOT be restarted • New version of OS may have new data structures • Lots of state in there: open files, timers,etc.
65 EXAMPLE OF HOW WOULD THIS WORK
User A Apache running A Apache still running
Kernel FreeBSD 10.1 FreeBSD 10.2
• Replace the OS while user processes are running • Very difficult to do with BSD, Linux, Windows, etc.
66 LIVE UPDATE IN MINIX
User A Apache running A Apache still running
User MM Dri- FS MM Dri- FS ver 6.0 ver 7.0
Kernel Microkernel Microkernel
67 HOW DO WE DO THE UPDATE?
• Manager tells some process (e.g. Old-FS) to get ready • Old-FS finishes its work and queues new work • Manager creates New-FS process with new code • LLVM puts tables inside New-FS listing its data objects • New-FS contacts Old-FS and asks for state it needs • The state is transferred one object at a time • When all state is transferred, Third-FS is created • It talks to New-FS and tries to recreate Old-FS • If they agree New-FS becomes FS, else revert to Old-FS • Like translating English to Dutch, then Dutch to English 68 HOW THE UPDATE WORKS
A Apache running
FS Old FS 6.0
Microkernel
69 HOW THE UPDATE WORKS
A Get Apache running ready
FS 6.0
Microkernel
70 HOW THE UPDATE WORKS
A Apache running
FS FS 6.0 7.0
Microkernel
71 HOW THE UPDATE WORKS
A Apache running
I need variable x FS FS G 6.0 7.0
Microkernel
72 HOW THE UPDATE WORKS
A Apache running
FS Here is variable x FS 6.0 7.0
Microkernel
73 HOW THE UPDATE WORKS
A Apache running
FS FS FS 6.0 7.0 ?
Microkernel
74 HOW THE UPDATE WORKS
A Apache running
FS FS I need variable x FS 6.0 7.0 ?
Microkernel
75 HOW THE UPDATE WORKS
A Apache running
FS FS Here is variable x FS 6.0 7.0 ?
Microkernel
76 HOW THE UPDATE WORKS
A Apache running
FS FS FS 6.0 7.0 ?
Are these the same? Microkernel
77 HOW THE UPDATE WORKS
A Apache running
FS 7.0
Microkernel
78 MUCH BETTER THAN KSPLICE
• KSPLICE can handle only small security patches • KSPLICE patches the running process • Over time, crud accumulates in the process • If the update fails, there is no recovery
79 OTHER USES OF LIVE UPDATE
• Enhanced security: – Update the OS at a high rate to foil return-to-libc attacks – Stop any attack that uses knowledge of memory layout – Reduce exposure to information leakage attacks
• Garbage collection in C (!) – Only live data is copied over to the new version – This can “fix” memory leaks (malloc but no free)
80 RESEARCH: FAULT INJECTION
Inject fault?
Original Basic block unmodified with fault basic block injected
This structure is created automatically by the LLVM compiler
81 NEW PROGRAM STRUCTURE
This can be optimized by patching the original binary to get any test without recompilation
Overhead is 8% 82 MINIX 3 LOGO
• Why a raccoon? – Small – Cute – Clever – Agile – Eats bugs – More likely to visit your house than a penguin
83 WEBSITE: www.minix3.org
84 DOCUMENTATION IS IN A WIKI
• Wiki.minix3.org • You can help document the system
85 TRAFFIC TO WWW.MINIX3.ORG
Total visits to the main page since 2004: 3.1 million
Actual downloads since 2007: 650,000 (from the log)
86 MINIX 3 GOOGLE NEWSGROUP
87 CONCLUSION
• Current OSes are bloated and unreliable • MINIX 3 is an attempt at a reliable, secure OS • Kernel is very small (15,000 LoC) • OS runs as a collection of user processes • Each driver is a separate process • Each OS component has restricted privileges • Faulty drivers can be replaced automatically • Live update is possible (not in current release)
88 SURVEY
• Please download MINIX 3 from www.minix3.org • Give it a try • Fill out the survey on the main page • We have had 650,000 downloads but we don’t know who they are or what they are doing • We are trying to build a community
89 THE END
90 WEBSITE: www.minix3.org
91 92 MASTERS DEGREE AT THE VU
• If you are interested in computer systems • Look at our masters in parallel & distributed syst. • Google me • Look at my home page • See video linked there or check out
pdcs.vu.nl
93 DISK PERFORMANCE
94 THE COST OF DRIVER RECOVERY • We killed the Ethernet driver every Δt sec to simulate repeated driver crashes
Driver recovery takes about 360 msec 95 RESEARCH: MULTICORE CHIPS
Multicore chip • Network stack has components • Chips may be heterogeneous TCP IP • Where to put each component?
Ether Kernel • Experiments scaling frequencies • Sometimes slower is faster! Core • Sleep/wakeup is expensive
96 RESEARCH: NEW FILE SYSTEM--LORIS
VFS • Better reliabilty • Better flexibility Naming • Handles heterogeneity better • File rather than block oriented Cache • Uses checksums to detect corruption
Logical Introduces concept of a logical file (1 or more phys files spread or striped over possibly heterogeneous devices) Physical
Driver
97