Giter Club home page Giter Club logo

statethreads's Introduction

state threads library

state threads library(以下简称st)是一个基于setjmp/longjmp实现的C语言版用户线程库或协程库(user level thread)即:
调度单位减小到函数,上下文切换不需要内核参与,不存在系统调用。上下文切换开销降到最低,系统调用降到最低,没有锁竞争,没有
信号处理。保留了程序对请求的线性处理逻辑,提高了程序的开发效率,可扩展性和可维护性。

WELCOME!

The State Threads Library is a small application library which provides a foundation for writing fast and highly scalable Internet applications (such as web servers, proxy servers, mail transfer agents, and so on, really any network-data-driven application) on UNIX-like platforms. It combines the simplicity of the multithreaded programming paradigm, in which one thread supports each simultaneous connection, with the performance and scalability of an event-driven state machine architecture. In other words, this library offers a threading API for structuring an Internet application as a state machine. For more details, please see the library documentation in the "docs" directory or on-line at

http://state-threads.sourceforge.net/docs/

The State Threads Project is an open source project for maintaining and enhancing the State Threads Library. For more information about this project, please see

http://state-threads.sourceforge.net/

BUILDING

To build the library by hand, use the GNU make utility. Run the make command (e.g., `gmake') with no arguments to display all supported targets.

To build more or less automatically, first set the CONFIG_GUESS_PATH variable in either osguess.sh or your environment then run "make default" which guesses your OS and builds. Requires the "config.guess" utility from GNU autoconf (not included with ST). You can use one from a larger "main" software project or just use any config.guess available on your system. You can also get it directly from GNU: ftp://ftp.gnu.org/gnu/autoconf/

To build rpms (RedHat Linux 6.2 or later, Linux/Mandrake, Solaris with gnome, etc.): download the latest st-x.y.tar.gz # rpm -ta st-x.y.tar.gz The .rpms will land in /usr/src/RPMS/. Install them with: # rpm -i libst*.rpm Requires GNU automake and rpm 3.0.3 or later.

Debian users: If you run potato, please upgrade to woody. If you run woody, "apt-get install libst-dev" will get you v1.3. If you run testing/unstable, you will get the newest available version. If you must have the newest libst in woody, you may follow these not-recommended instructions: 1. Add "deb-src unstable main" to your /etc/apt/sources.list 2. apt-get update 3. apt-get source st 4. cd st-1.4 (or whatever version you got) 5. debuild 6. dpkg -i ../*.deb

If your application uses autoconf to search for dependencies and you want to search for a given version of libst, you can simply add PKG_CHECK_MODULES(MYAPP, st >= 1.3 mumble >= 0.2.23) to your configure.ac/in. This will define @MYAPP_LIBS@ and @MYAPP_CFLAGS@ which you may then use in your Makefile.am/in files to link against mumble and st.

LICENSE

The State Threads library is a derivative of the Netscape Portable Runtime library (NSPR). All source code in this directory is distributed under the terms of the Mozilla Public License (MPL) version 1.1 or the GNU General Public License (GPL) version 2 or later. For more information about these licenses please see http://www.mozilla.org/MPL/ and http://www.gnu.org/copyleft/.

All source code in the "examples" directory is distributed under the BSD style license.

PLATFORMS

Please see the "docs/notes.html" file for the list of currently supported platforms.

DEBUGGER SUPPORT

It's almost impossible to print SP and PC in a portable way. The only way to see thread's stack platform-independently is to actually jump to the saved context. That's what the _st_iterate_threads() function does. Do the following to iterate over all threads:

  • set the _st_iterate_threads_flag to 1 in debugger
  • set breakpoint at the _st_show_thread_stack() function (which does nothing)
  • call the _st_iterate_threads() function which jumps to the next thread
  • at each break you can explore thread's stack
  • continue
  • when iteration is complete, you return to the original point (you can see thread id and a message as arguments of the _st_show_thread_stack() function).

You can call _st_iterate_threads() in three ways:

  • Insert it into your source code at the point you want to go over threads.
  • Just run application and this function will be called at the first context switch.
  • Call it directly from the debugger at any point.

This works with gdb and dbx.

Example using gdb:

(gdb) set _st_iterate_threads_flag = 1 (gdb) b _st_show_thread_stack ... (gdb) call _st_iterate_threads() ... (gdb) bt ... (gdb) c ... (gdb) bt ... (gdb) c ... and so on...

_st_iterate_threads_flag will be set to 0 automatically after iteration is over or you can set it to 0 at any time to stop iteration.

Sometimes gdb complains about SIGSEGV when you call a function directly at gdb command-line. It can be ignored -- just call the same function right away again, it works just fine. For example:

(gdb) set _st_iterate_threads_flag = 1 (gdb) b _st_show_thread_stack Breakpoint 1 at 0x809bbbb: file sched.c, line 856. (gdb) call _st_iterate_threads() Program received signal SIGSEGV, Segmentation fault. .... (gdb) # just call the function again: (gdb) call _st_iterate_threads() Breakpoint 1, _st_show_thread_stack (thread=0x4017aee4, messg=0x80ae7a2 "Iteration started") at sched.c:856 856 } ....

You can use simple gdb command-line scripting to display all threads and their stack traces at once:

(gdb) while _st_iterate_threads_flag

bt c end ....

Another script to stop at the thread with the specific thread id (e.g., 0x40252ee4):

(gdb) # set the flag again: (gdb) set _st_iterate_threads_flag = 1 (gdb) call _st_iterate_threads() Breakpoint 1, _st_show_thread_stack (thread=0x4017aee4, messg=0x80ae7a2 "Iteration started") at sched.c:856 856 } .... (gdb) while thread != 0x40252ee4

c end .... .... Breakpoint 1, _st_show_thread_stack (thread=0x40252ee4, messg=0x0) at sched.c:856 856 } (gdb) bt .... (gdb) # don't want to continue iteration, unset the flag: (gdb) set _st_iterate_threads_flag = 0 (gdb) c Continuing. Breakpoint 1, _st_show_thread_stack (thread=0x0, messg=0x80ae78e "Iteration completed") at sched.c:856 856 } (gdb) c Continuing. (gdb) return Make selected stack frame return now? (y or n) y #0 0x4011254e in __select () from /lib/libc.so.6 (gdb) detach

CHANGE LOG

Changes from 1.8 to 1.9.

o Support 32-bit and 64-bit Intel Macs.

o Added ST_VERSION string, and ST_VERSION_MAJOR and ST_VERSION_MINOR [bug 1796801].

o Fixed some compiler warnings, based on a patch from Brian Wellington [bug 1932741].

Changes from 1.7 to 1.8.

o Added support for kqueue and epoll on platforms that support them. Added ability to choose the event notification system at program startup.

o Long-overdue public definitions of ST_UTIME_NO_TIMEOUT (-1ULL) and ST_UTIME_NO_WAIT (0) [bug 1514436].

o Documentation patch for st_utime() [bug 1514484].

o Documentation patch for st_timecache_set() [bug 1514486].

o Documentation patch for st_netfd_serialize_accept() [bug 1514494].

o Added st_writev_resid() [rfe 1538344].

o Added st_readv_resid() [rfe 1538768] and, for symmetry, st_readv().

Changes from 1.6 to 1.7.

o Support glibc 2.4, which breaks programs that manipulate jump buffers. Replaced Linux IA64 special cases with new md.S that covers all Linux.

Changes from 1.5.2 to 1.6.

none

Changes from 1.5.1 to 1.5.2.

o Alfred Perlstein's context switch callback feature.

o Claus Assmann's st_recvmsg/st_sendmsg wrappers.

o Extra stack padding for platforms that need it.

o Ron Arts's timeout clarifications in the reference manual.

o Raymond Bero and Anton Berezin's AMD64 FreeBSD port.

o Claus Assmann's AMD64 SunOS 5.10 port.

o Claus Assmann's AMD64 OpenBSD port.

o Michael Abd-El-Malek's Mac OS X port.

o Michael Abd-El-Malek's stack printing patch.

Changes from 1.5.0 to 1.5.1.

o Andreas Gustafsson's USE_POLL fix.

o Gene's st_set_utime_function() enhancement.

Changes from 1.4 to 1.5.0.

o Andreas Gustafsson's performance patch.

o New extensions: Improved DNS resolver, generic LRU cache, in-process DNS cache, and a program to test the resolver and cache.

o Support for AMD Opteron 64-bit CPUs under Linux.

o Support for SPARC-64 under Solaris.

o Andreas Gustafsson's support for VAX under NetBSD.

o Changed unportable #warning directives in md.h to #error.

Changes from 1.3 to 1.4.

o Andreas Gustafsson's NetBSD port.

o Wesley W. Terpstra's Darwin (MacOS X) port.

o Support for many CPU architectures under Linux and *BSD.

o Renamed private typedefs so they don't conflict with public ones any more.

o common.h now includes public.h for strict prototyping.

o Joshua Levy's recommendation to make st_connect() and st_sendto() accept const struct sockaddr pointers, as the originals do.

o Clarified the documentation regarding blocking vs. non-blocking I/O.

o Cygwin support.

o Created the extensions directory.

o Fixed warnings from ia64asm.S.

Changes from 1.2 to 1.3.

o Added st_read_resid() and st_write_resid() to allow the caller to know how much data was transferred before an error occurred. Updated documentation.

o Updated project link, copyrights, and documentation regarding timeouts. Added comment to st_connect().

o Optimized the _st_add_sleep_q() function in sched.c. Now we walk the sleep queue backward when inserting a thread into it. When you have lots (hundreds) of threads and several timeout values, it takes a while to insert a thread at the appropriate point in the sleep queue. The idea is that often this appropriate point is closer to the end of the queue rather than the beginning. Measurements show performance improves with this change. In any case this change should do no harm.

o Added a hint of when to define USE_POLL and when not to, to the Makefile.

o Added debugging support (files common.h and sched.c). See above.

o Decreased the number of reallocations of _ST_POLLFDS in sched.c. Inspired by Lev Walkin.

o Fixed st_usleep(-1) and st_sleep(-1), and added a warning to the documentation about too-large timeouts.

o Linux/*BSD Alpha port.

o Wesley W. Terpstra modernized the build process:

  • properly build relocatable libraries under bsd and linux
  • use library versioning
  • added rpm spec file
  • added debian/ files See above for build instructions.

Changes from 1.1 to 1.2.

o Added st_randomize_stacks().

o Added a patch contributed by Sascha Schumann.

Changes from 1.0 to 1.1.

o Relicensed under dual MPL-GPL.

o OpenBSD port.

o Compile-time option to use poll() instead of select() for event polling (see Makefile). This is useful if you want to support a large number of open file descriptors (larger than FD_SETSIZE) within a single process.

o Linux IA-64 port. Two issues make IA-64 different from other platforms:

  • Besides the traditional call stack in memory, IA-64 uses the general register stack. Thus each thread needs a backing store for the register stack in addition to the memory stack.

  • Current implementation of setjmp()/longjmp() can not be used for thread context-switching since it assumes that only one register stack exists. Using special assembly functions for context-switching is unavoidable.

o Thread stack capping on IRIX. This allows some profiling tools (such as SpeedShop) to know when to stop unwinding the stack. Without this libexc, used by SpeedShop, traces right off the stack and crashes.

o Miscellaneous documentation additions.

COPYRIGHTS

Portions created by SGI are Copyright (C) 2000 Silicon Graphics, Inc. All Rights Reserved.

statethreads's People

Contributors

tangyibo avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.