Linux系统资源控制

每一个进程都有自己的一组资源限制,在Linux系统中我们可以通过这2个API来取得和设置资源。


#include <sys/resource.h>
int getrlimit(int resource, struct rlimit *rlim);
int setrlimit(int resource, const struct rlimit *rlim);

getrlimit用来取得,setrlimit用来设置。这二个参数都需要一个要控制的资源,比如控制CPU、内存、文件描述符个数等等的控制,作为第一个参数传入,第二个参数是一个rlimit的结构体地址(指针),他的结构定义放在头文件/usr/include/bits/resource.h中


struct rlimit
{
         rlim_t rlim_cur;       
         rlim_t rlim_max;
};


结构体中,rlim_cur是要取得或设置的资源软限制的值,rlim_max是硬限制。
这两个值的设置有一个小的约束:
1) 任何进程可以将软限制改为小于或等于硬限制
2) 任何进程都可以将硬限制降低,但普通用户降低了就无法提高,该值必须等于或大于软限制
3) 只有超级用户可以提高硬限制
一个无限的限制由常量RLIM_INFINITY指定(The value RLIM_INFINITY denotes no limit on a resource)

RLIMIT_AS
                   The maximum size of the process virtual memory (address
                   space) in bytes.  This limit affects calls to brk(2), mmap(2)
                   and mremap(2), which fail with the error ENOMEM upon exceeding
                   this limit. Also automatic stack expansion will fail (and gen-
                   erate a SIGSEGV that kills the process when no alternate stack
                   has been made available). Since the value is a long,  on
                   machines with a 32-bit long either this limit is at most 2 GiB,
                   or this resource is unlimited.


RLIMIT_CORE
                   Maximum size of core file. When 0 no core dump files are cre-
                   ated.  When nonzero, larger dumps are truncated to this size.
     设定最大的core文件,当值为0时将禁止core文件非0时将设定产生的

     最大core文件大小为设定的值


RLIMIT_CPU
                   CPU time limit in seconds. When the process reaches the soft
                   limit, it is sent a SIGXCPU signal. The default action for
                   this signal  is to terminate the process.However, the signal
                   can be caught, and the handler can return control to the main
                   program.If the process continues to consume CPU time, it will
                   be sent SIGXCPU once per second until the hard limit is
                   reached, at which time it is sent SIGKILL.(This latter point
                   describes Linux 2.2 and 2.4 behaviour. Implementations vary in
                   how they treat processes which continue to consume CPU time
                   after reaching the soft limit. Portable applications that need
                   to catch this signal should perform an orderly termination upon
                   first receipt of SIGXCPU.)
     CPU时间的最大量值(秒),当超过此软限制时向该进程发送SIGXCPU

     或SIGKILL信号


RLIMIT_DATA

                   The maximum size of the process data segment (initialized
                   data,  uninitialized data, and heap).  This limit affects calls
                   to brk() and sbrk(), which fail with the error ENOMEM upon
                   encountering the soft limit of this resource.
     数据段的最大字节长度


RLIMIT_FSIZE
                   The maximum size of files that the process may create.
                   Attempts to extend a file beyond this limit result in delivery
                   of a SIGXFSZ signal. By default, this signal terminates a pro-
                   cess, but a process can catch this signal instead, in which
                   case the relevant system call (e.g., write(), truncate()) fails
                   with the error EFBIG.
     可以创建的文件的最大字节长度,当超过此软限制时向进程发送SIGXFSZ


RLIMIT_MEMLOCK

                   The maximum number of bytes of virtual memory that may be
                   locked into RAM using mlock() and mlockall().


RLIMIT_NOFILE
                   Specifies a value one greater than the maximum file descriptor
                   number that can be opened by this process. Attempts (open(),
                   pipe(), dup(), etc.) to exceed this limit yield the error
                   EMFILE.
     每个进程能够打开的最多文件数。更改此限制将影响到sysconf函数在

     参数_SC_CHILD_MAX中的返回值

 

RLIMIT_NPROC
                   The maximum number of processes that can be created  for the
                   real user ID of the calling process. Upon encountering this
                   limit, fork() fails with the error EAGAIN.
     每个实际用户ID所拥有的最大子进程数,更改此限制将影响到sysconf
     函数在参数_SC_CHILD_MAX中返回的值


RLIMIT_RSS

                   Specifies the limit (in pages) of the process resident set
                   (the number of virtual pages resident in RAM). This limit only
                   has effect in Linux 2.4 onwatrds, and there only affects calls
                   to madvise() specifying MADVISE_WILLNEED.
     最大驻内存集字节长度(RSS)如果物理存储器供不应求则内核将从进程

     处取回超过RSS的部份


RLIMIT_STACK

                   The maximum size of the process stack, in bytes. Upon reaching
                   this limit, a SIGSEGV signal is generated. To handle this sig-
                   nal, a process must employ an alternate signal stack (sigalt-
                   stack(2)).
     栈的最大长度


RLIMIT_VMEM 可映照地址空间的最大字节长茺,这影响到mmap函数


这些限制影响到调用进程并由子进程继承! 可以在SHELL中预设这些值ulimit命令设置
#include <stdlib.h>
#include <stdio.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>
#include <sys/resource.h>
int main(void)
{
     struct rlimit r;
     if(getrlimit(RLIMIT_NOFILE,&r)<0)
     {
         fprintf(stderr,"getrlimit error\n");
         exit(1);
     }
     printf("RLIMIT_NOFILE cur:%d\n",r.rlim_cur);
     printf("RLIMIT_NOFILE max:%d\n",r.rlim_max);

    
     r.rlim_cur=100;
     r.rlim_max=200;

     if (setrlimit(RLIMIT_NOFILE,&r)<0)
     {
         fprintf(stderr,"setrlimit error\n");
         exit(1);
     }

    
     if(getrlimit(RLIMIT_NOFILE,&r)<0)
     {
         fprintf(stderr,"getrlimit error\n");
         exit(1);
     }
     printf("RLIMIT_NOFILE cur:%d\n",r.rlim_cur);
     printf("RLIMIT_NOFILE max:%d\n",r.rlim_max);
     return 0;
}

:!gcc test.c                                                                                                                                                    

:!./a.out
RLIMIT_NOFILE cur:1024
RLIMIT_NOFILE max:1024
RLIMIT_NOFILE cur:100
RLIMIT_NOFILE max:200

posted @ 2013-02-21 19:10  only_eVonne  阅读(617)  评论(0编辑  收藏  举报