Transcript 9. Chorus
9. Chorus History of Chorus Chorus started out at the French research institute INRIA in 1980, as a research project in distributed systems. It has since gone through four versions, numbered from 0 through 3. The idea behind Version 0 was to model distributed applications as a collection of actors. Version 1, which lasted from 1982 to 1984, focused on multiprocessor research. Version 2 (1984-1986) was a major rewrite of the system, in C. Version 3 was started in 1987. The version marked the transition from a research system to a commercial product. Goals of Chorus 1. 2. 3. 4. High-performance UNIX emulation. Use on distributed systems. Real-time applications. Integrating object-oriented programming into Chorus. System Structure UNIX subsystem Object-oriented subsystem u1 User addr. space Kernel addr. space s1 s2 k1 Microkernel u2 u3 User process s3 System process k2 Kernel process Management of names, processes, threads, memory, and communication. Six key abstractions Address space Thread Port: holding incoming messages at any instant, each port belongs to one process. Microkernel message Region of address space Microkernel network A capability in Chorus UI of a port Bits 13 Creation site 3 key to distinguish objects 48 Type Epoch number + counter 64 Defined by the subsystem Can indicate site, port, or port group; the other five combinations are reserved for future use. An address space with four mapped regions Region stack Scratched segment Unmapped address Region file Unmapped address Region data Region program Mapped portion of file F Unmapped portion of file F Copy of program’s initial data Read-only segment Kernel structure Responsible for ports and messages Interprocess communication manager (portable) Handles, processes, threads, and Real-time executive scheduling (portable) Virtual memory (portable) Supervisor (machine dependent) Caches traps, interrupts, and exceptions Manages paging (low-level part of the paging system) Machine-dependent Portion of the virtua Memory manager The UNIX Subsystem Since Chorus is now a commercial product, it must be compatible with UNIX. Chorus accomplishes this goal by providing a standard subsystem, called MiX, that is compatible with System V. MiX is also compatible with UNIX in other ways. For example, the file system is compatible, so Chorus can read a UNIX disk. Furthermore, the Chorus device drivers are interface compatible with the UNIX ones, so if UNIX device drivers exist for a device machine, they can be ported to Chorus with relatively littler work. The Object-Oriented Subsytem It consists of three layers. The bottom layer does object management in a generic way and is effectively a microkernel for objectoriented systems. The middle layer provides a general runtime system. The top layer is the language runtime system. This subsystem, called COOL. Process Management in Chorus A process in Chorus is a collection of active and passive elements that work together to perform some computation. The active elements are the threads. The passive elements are an address space (containing some regions) and a collection of ports (for sending and receiving messages). Three kinds of processes Type User Trust Privilege Untrusted Unpriviledged Mode Space User User User System Trusted Unpriviledged User Kernel Privileged Kernel Kernel Trusted Privilege refers to the ability to execute I/O and other protected instructions. Trust means that the process is allowed to call the kernel directly. Threads Every thread has its own private context (i.e., stack, program counter, and registers). A thread is tied to the process in which it was created, and cannot be moved to another process. Chorus threads are known to the kernel and scheduled by the kernel, so creating and destroying them requires making kernel calls. An advantage of having kernel threads is that when one thread blocks waiting for some event (e.g., a message arrival), the kernel can schedule other threads. Another advantage is the ability to run different threads on different CPUs when a multiprocessor is available. The disadvantage of kernel threads is the extra overhead required to manage them. Threads communicate with one another by sending and receiving messages. Chorus distinguishes the following states, but they are not mutually exclusive: 1. ACTIVE – The thread is logically able to run. 2. SUSPENDED – The thread has been intentionally suspended. 3. STOPPED – The thread’s process has been suspended. 4. WAITING – The thread is waiting for some event to happen. Two synchronization mechanisms Traditional semaphore, with operations UP and DOWN. mutex, which is essentially a semaphore whose values are restricted to 0 and 1. Mutexes are used only for mutual exclusion. Scheduling CPU scheduling is done using priorities on a perthread basis. Each process has a priority and each thread has a relative priority within its process. The absolute priority of a thread is the sum of its process’ priority and its own relative priority. The kernel keeps track of the priority of each thread in ACTIVE state and runs the one with the highest absolute priority. On a multiprocessor with k CPUs, the k highest-priority threads are run. High priority A These threads are not timesliced These threads are timesliced Low priority B C D C D D C Traps, Exceptions, and Interrupts 1. 2. Traps are intentional calls to the kernel or a subsystem to invoke services. Programs cause traps by calling a system call library procedure. The system supports two ways of handling traps. all traps for a particular trap vector go to a single kernel thread that has previously announced its willingness to handle that vector. each trap vector is tied to an array of kernel threads, with the Chorus supervisor using the contents of a certain register to index into the array to pick a thread. Exceptions are unexpected events that are caused by accident, such as the divide-byzero exception, floating-point overflow, or a page fault. Interrupts are caused by asynchronous events, such as clock ticks or the completion of an I/O request. Kernel Calls for Process Management actorCreate Create a new process ActorDelete Remove a process ActorStop Stop a process, put its threads in STOPPED state Restart a process from STOPPED state actoreStart actorPriority Get or set a process’ priority actorExcept Get or set the port used for exception handling threadCreate Create a new thread threadDelete Delete a thread threadSuspend Suspend a thread threadResume Restart a suspended thread threadPriority Get or set a thread’s priority threadLoad Get a thread’s context pointer threadStore Set a thread’s context pointer threadContext Get or set a thread’s execution context mutexInit Initialize a mutex mutexGet Try to acquire a mutex mutexRel Release a mutex semInit Initialize a semaphore semP Do a DOWN on a semaphore semV Do an UP on a semaphore Memory Management in Chorus A region is a contiguous range of virtual address, for example, from 1024 to 6143. A segment is a contiguous collection of bytes named and protected by a capability. Mapping segments onto regions. It is not necessary that a segment be exactly the size of its region. 1. If the segment is larger than the region, only a portion of the segment will be visible in the address space, although which portion is visible can be changed by remapping it. 2. If the segment is smaller than the region, the result of reading an unmapped address is up to the mapper. For example, it can raise an exception, return 0, or extend the segment. Mappers Each mapper controls one or more segments that are mapped onto regions. A segment can be mapped into multiple regions, even in different address spaces at the same time. Segments can be mapped into multiple address space at the same time Process A Segments S1 S2 Process B Distributed Shared Memory Chorus supports paged distributed shard memory in the style of IVY. IT uses a dynamic decentralized algorithm, meaning that different managers keep track of different pages, and the manager for a page change as the page moves around the system. The unit of sharing between multiple machines is the segment. Segments are split up into fragments of one or more pages. At any instant, each fragment is either read-only, and potentially present on multiple machines, or read/write, and present only on one machine. Kernel Calls for Memory Management rgnAllocate Allocate a memory region and set its properties rgnFree Release a previously allocated region rgnInit Allocate a region and fill it from a given segment rgnSetInherit Set the inheritance properties of a region rgnSetPaging Set the paging properties of a region rgnSetProtect Set the protection options of a region rgnStat Get the statistics associated with a region sgRead Read data from a segment sgWrite Write data to a segment sgStat Request information about a page cache sgFlush Request from a mapper to the kernel asking for dirty pages MpCreate Request to create a dummy segment for swapping MpRelease Request asking to release a previously created segment MpPullIn Request asking for one or more pages MpPushOut Request asking mapper to accept one or more pages Communication in Chorus The basic communication paradigm in Chorus is message passing. 64 bytes long header An optional fixed part Identifies the source and destination and contains protection identifiers and flags. Maximum of 64k bytes Optional body Port group 1 Port group 2 network Communication Operations Two kinds of communication operations are provided by Chorus: asynchronous send and RPC. Asynchronous send allows a thread simply to send a message to a port. There is no guarantee that the message arrives and no notification if something goes wrong. RPC uses blocking send and at-most-once semantics. To all 1 2 3 To any 1 2 3 To 1 1 2 3 Not to 1 1 2 3 Kernel calls for communication portCreate Create a port and return its capability portDelete Destroy a port portEnable portDisable Eanble a port so its messages count on a receive from all ports Disable a port portMigrate Move a port to a different process grpAllocate Create a port group grpPortInsert Add a new port to an existing port gro grpPortRemove Delete a port from a port group ipcSend Send a message asynchronously ipcReceive Block until a message arrives ipcGetData Get the current message’s body ipcReply Send a reply to the current message Perform a remote procedure call ipcCall