LinuxSir.cn,穿越时空的Linuxsir!

 找回密码
 注册
搜索
热搜: shell linux mysql
查看: 871|回复: 5

问两个2.6内核的基础性问题---请高手帮忙回答-

[复制链接]
发表于 2005-7-11 11:41:46 | 显示全部楼层 |阅读模式
有没有哪位对2.6内核里的各进程含义熟悉的?比如pdflush,migration。还有我想找找2.6内核编译时make时前面标志的含义在哪能找到?比如CHK,UPD,HOSTCC
发表于 2005-7-11 15:21:39 | 显示全部楼层
pdflush:
  1.When free memory shrinks below a specified threshold, the kernel must write dirty data back to disk to free memory.
  2.When dirty data grows older than a specific threshold, sufficiently old data is written back to disk, to ensure that dirty data does not remain dirty indefinitely.

migration:
  It is a highprio kernel thread that performs thread migration by bumping thread off CPU then 'pushing' onto another runqueue. It's used by syscall sched_setaffinity and "Load Balancer"
回复 支持 反对

使用道具 举报

发表于 2005-7-12 17:43:09 | 显示全部楼层
CHK HOSTCC 应该在makefile中可以找到。HOSTCC其实就是你主机上的c编译器
回复 支持 反对

使用道具 举报

 楼主| 发表于 2005-7-13 09:59:22 | 显示全部楼层
谢谢二位---     |-events/0
     |-events/1
     |-events/2
     |-events/3
     |-khelper
     |-kirqd
     |-klogd
     |-kseriod
     |-ksoftirqd/0
     |-ksoftirqd/1
     |-ksoftirqd/2
     |-ksoftirqd/3
     |-kswapd0
     |-kthread-+-aio/0
     |         |-aio/1
     |         |-aio/2
     |         |-aio/3
     |         |-kacpid
     |         |-kblockd/0
     |         |-kblockd/1
     |         |-kblockd/2
     |         |-kblockd/3
     |         |-2*[pdflush]
     |         |-reiserfs/0
     |         |-reiserfs/1
     |         |-reiserfs/2
     |         `-reiserfs/3
     |-migration/0
     |-migration/1
     |-migration/2
     |-migration/3
     |-scsi_eh_0
     |-scsi_eh_1
还有这些进程的含义呢?到哪可以找到?
回复 支持 反对

使用道具 举报

发表于 2005-7-13 13:23:44 | 显示全部楼层
events/n is used by "work queues" (a bottom Half mechanism in kernel)。queued works are executed in events/n kernel thread.

ksoftirqd/n is used by "Softirq" (another bottom Half mechanism in kernel)。It's used to solve the problem that a Softirq continally raise itself again. if the new Softirq is handled immediately,nothing but Softirq and interrupt handlers run when system under intense load. If the new Softirq is not handled until next time kernel handles pedning Softirqs, it will stay there for a long time on an idle system.
Some sort of compromise is needed.The solution is to not immediately process reraised Softirq. Instead, kernel wake up ksoftirqd kernel thread to handle pending Softirqs.it solves the problem on an idle system described above(Ksoftirqd kernel thread run with lowest priority)
回复 支持 反对

使用道具 举报

发表于 2005-7-13 14:22:44 | 显示全部楼层
K开头的是内核里的函数吧    d结尾的都是守护进程喽
回复 支持 反对

使用道具 举报

您需要登录后才可以回帖 登录 | 注册

本版积分规则

快速回复 返回顶部 返回列表