使用库函数也有系统调用的开销,为什么不直接使用系统调用呢?
来源:互联网 发布:数控车床怎么编程视频 编辑:程序博客网 时间:2024/05/18 22:40
《C专家编程》中提到,库函数调用通常比行内展开的代码慢,因为它需要付出函数调用的开销。但系统调用比库的数调用还要慢很多,因为它需要把上下文环境切换到内核模式然后再切换回来。
库函数对文件的操作实际上是通过系统调用来实现的。例如 C库函数 fwrite() 就是通过 write() 系统调用来实现的。
但是使用库函数也有系统调用的开销,为什么不直接使用系统调用呢?
这是因为,读写文件通常是大量的数据(这种大量是相对于底层驱动的系统调用所实现的数据操作单位而言)。这时,使用库函数就可以大大减少系统调用的次数。
这一结果又源于缓冲区技术,在用户空间和内核空间对文件操作都使用了缓冲区。
例如:用 fwrite() 写文件,都是先将内容写到用户空间缓冲区,当用户空间缓冲区满或者写结束时才将内核缓冲区内容写到文件对应的硬件媒介。
/****fwrite.c - read from a stream** Copyright (c) 1989-1997, Microsoft Corporation. All rights reserved.**Purpose:* Write to the specified stream from the user's buffer.********************************************************************************/#include <cruntime.h>#include <stdio.h>#include <mtdll.h>#include <io.h>#include <string.h>#include <file2.h>#include <internal.h>/****size_t fwrite(void *buffer, size_t size, size_t count, FILE *stream) -* write to the specified stream from the specified buffer.**Purpose:* Write 'count' items of size 'size' to the specified stream from* the specified buffer. Return when 'count' items have been written* or no more items can be written to the stream.**Entry:* buffer - pointer to user's buffer* size - size of the item to write* count - number of items to write* stream - stream to write to**Exit:* Returns the number of (whole) items that were written to the stream.* This may be less than 'count' if an error or eof occurred. In this* case, ferror() or feof() should be used to distinguish between the* two conditions.**Notes:* fwrite will attempt to buffer the stream (side effect of the _flsbuf* call) if necessary.** No more than 0xFFFE bytes may be written out at a time by a call to* write(). Further, write() does not handle huge buffers. Therefore,* in large data models, the write request is broken down into chunks* that do not violate these considerations. Each of these chunks is* processed much like an fwrite() call in a small data model (by a* call to _nfwrite()).** This code depends on _iob[] being a near array.** MTHREAD/DLL - Handled in just two layers since it is small data* model. The outer layer, fwrite(), handles stream locking/unlocking* and calls _fwrite_lk() to do the work. _fwrite_lk() is the same as* the single-thread, small data model version of fwrite().********************************************************************************/#ifdef _MT/* define locking/unlocking version */size_t __cdecl fwrite ( const void *buffer, size_t size, size_t count, FILE *stream ){ size_t retval; _lock_str(stream); /* lock stream */ retval = _fwrite_lk(buffer, size, count, stream); /* do the read */ _unlock_str(stream); /* unlock stream */ return retval;}#endif /* _MT *//* define the normal version */#ifdef _MTsize_t __cdecl _fwrite_lk (#else /* _MT */size_t __cdecl fwrite (#endif /* _MT */ const void *buffer, size_t size, size_t num, FILE *stream ){ const char *data; /* point to where data comes from next */ unsigned total; /* total bytes to write */ unsigned count; /* num bytes left to write */ unsigned bufsize; /* size of stream buffer */ unsigned nbytes; /* number of bytes to write now */ unsigned nwritten; /* number of bytes written */ int c; /* a temp char */ /* initialize local vars */ data = buffer; count = total = size * num; if (0 == count) return 0; if (anybuf(stream)) /* already has buffer, use its size */ bufsize = stream->_bufsiz; else#if defined (_M_M68K) || defined (_M_MPPC) /* assume will get BUFSIZ buffer */ bufsize = BUFSIZ;#else /* defined (_M_M68K) || defined (_M_MPPC) */ /* assume will get _INTERNAL_BUFSIZ buffer */ bufsize = _INTERNAL_BUFSIZ;#endif /* defined (_M_M68K) || defined (_M_MPPC) */ /* here is the main loop -- we go through here until we're done */ while (count != 0) { /* if the buffer is big and has room, copy data to buffer */ if (bigbuf(stream) && stream->_cnt != 0) { /* how much do we want? */ nbytes = (count < (unsigned)stream->_cnt) ? count : stream->_cnt; memcpy(stream->_ptr, data, nbytes); /* update stream and amt of data written */ count -= nbytes; stream->_cnt -= nbytes; stream->_ptr += nbytes; data += nbytes; } else if (count >= bufsize) { /* If we have more than bufsize chars to write, write data by calling write with an integral number of bufsiz blocks. If we reach here and we have a big buffer, it must be full so _flush it. */ if (bigbuf(stream)) { if (_flush(stream)) { /* error, stream flags set -- we're out of here */ return (total - count) / size; } } /* calc chars to read -- (count/bufsize) * bufsize */ nbytes = ( bufsize ? (count - count % bufsize) : count ); nwritten = _write(_fileno(stream), data, nbytes); if (nwritten == (unsigned)EOF) { /* error -- out of here */ stream->_flag |= _IOERR; return (total - count) / size; } /* update count and data to reflect write */ count -= nwritten; data += nwritten; if (nwritten < nbytes) { /* error -- out of here */ stream->_flag |= _IOERR; return (total - count) / size; } } else { /* buffer full and not enough chars to do direct write, so do a _flsbuf. */ c = *data; /* _flsbuf write one char, this is it */ if (_flsbuf(c, stream) == EOF) { /* error or eof, stream flags set by _flsbuf */ return (total - count) / size; } /* _flsbuf wrote a char -- update count */ ++data; --count; /* update buffer size */ bufsize = stream->_bufsiz > 0 ? stream->_bufsiz : 1; } } /* we finished successfully, so just return num */ return num;}
阅读全文
1 0
- 使用库函数也有系统调用的开销,为什么不直接使用系统调用呢?
- 直接使用系统调用
- 测量系统调用开销
- linux使用系统调用和C库函数调用
- 绕过 libc 直接使用系统调用
- 系统调用之时间开销
- 库函数调用和系统调用的区别
- 库函数调用和系统调用的区别
- 系统调用与库函数调用的区别
- 系统调用和库函数调用的区别
- 库函数调用和系统调用的区别
- 库函数调用和系统调用的区别
- 系统调用与库函数调用的区别
- 库函数调用和系统调用的区别
- 库函数调用和系统调用的区别
- 系统调用与库函数调用的区别
- 系统调用和库函数调用的区别
- 系统调用与库函数调用的区别
- linux创建桌面快捷启动图标
- 输出单向链表中倒数第k个结点
- AutoEventWireup 属性的作用(转载)
- kotlin配置android基本工程
- Python学习笔记-清除变量
- 使用库函数也有系统调用的开销,为什么不直接使用系统调用呢?
- 算法(四):贪心算法
- cgicc thttpd经常用的调试命令
- Python几种并发实现方案的性能比较
- android webview增强版,对原生webview的一些解决方案
- 系统开关机时oracle自启动和关闭脚本(for aix和linux)
- MySQL查询不区分大小写问题
- 访问网页后HTML,CSS,JavaScript文件的加载顺序
- 《Linux C编程实战》中一些基础C语言知识 && 一些笔试面试题总结