|
vmstat -a command to see the active memory and inactive memory, but what do they mean it?
$ Vmstat-a
procs ----------- memory ------------- swap ------- io ----- system -------- cpu -----
r b swpd free inact active si so bi bo in cs us sy id wa st
1013809631956013724081757848002323109900
Their meaning in manpage only gave a simple explanation, did not explain in detail:
inact: the amount of inactive memory (-a option).
active: the amount of active memory (-a option).
Here we try to understand exactly what it means. Vmstat by reading the source code (vmstat.c and proc / sysinfo.c) that, vmstat command is taken directly from the / proc / meminfo data:
$ Grep-i act / proc / meminfo
Active: 1767928 kB
Inactive: 1373760 kB
The / proc / meminfo data is generated in the following core functions:
fs / proc / meminfo.c:
==================
0023staticint meminfo_proc_show (struct seq_file * m, void * v)
0024 {
...
0032unsignedlong pages [NR_LRU_LISTS];
...
0051for (lru = LRU_BASE; lru
0052 pages [lru] = global_page_state (NR_LRU_BASE + lru);
...
0095 "Active:% 8lu kB \ n"
0096 "Inactive:% 8lu kB \ n"
0097 "Active (anon):% 8lu kB \ n"
0098 "Inactive (anon):% 8lu kB \ n"
0099 "Active (file):% 8lu kB \ n"
0100 "Inactive (file):% 8lu kB \ n"
...
0148 K (pages [LRU_ACTIVE_ANON] + pages [LRU_ACTIVE_FILE]),
0149 K (pages [LRU_INACTIVE_ANON] + pages [LRU_INACTIVE_FILE]),
0150 K (pages [LRU_ACTIVE_ANON]),
0151 K (pages [LRU_INACTIVE_ANON]),
0152 K (pages [LRU_ACTIVE_FILE]),
0153 K (pages [LRU_INACTIVE_FILE]),
This code means that all statistical LRU list, which is equal to ACTIVE_ANON Active Memory and ACTIVE_FILE sum, Inactive Memory and INACTIVE_FILE INACTIVE_ANON equal sum.
LRU list is the Linux kernel memory pages to retrieve data structure algorithms (Page Frame Reclaiming Algorithm) used, LRU is an acronym for the Least Recently Used. The core idea of this algorithm is: Recycling page should be the least recently used.
To achieve this goal, the ideal situation is that each page has an age term for recording the last time the page is accessed, but x86 CPU hardware does not support this feature, x86 CPU can do when you visit the page settings a flag Access bit, time can not be recorded.
So Linux kernel uses a compromise approach: It uses the LRU list list to just visited pages on the column header, the closer the end of the column is the longer unvisited pages, so although it can not access the recording time , but the use of the page in the relative positions of the LRU list can easily find the oldest page.
Linux kernel designed two LRU list: active list and inactive list, just visited pages into the active list, not a long time visited pages into the inactive list, this page recovered from the inactive list becomes simple. Kswapd kernel thread periodically the active list in qualifying moved to inactive list in the page, the transfer is done by refill_inactive_zone () completed. This code means that all statistical LRU list, which is equal to ACTIVE_ANON Active Memory and ACTIVE_FILE sum, Inactive Memory and INACTIVE_FILE INACTIVE_ANON equal sum.
LRU_list
vmstat see active / inactive memory on the active list are inactive list and the memory size. If the inactive list is large, indicating that the page can be recovered if necessary a lot; and if the inactive list is small, indicating that the page can be recovered much.
Active / inactive memory for user processes in terms of memory occupied by the kernel memory occupied (including slab) is not among them.
As seen in the source code ACTIVE_ANON and ACTIVE_FILE, respectively anonymous pages and mapped pages. User process is divided into two pages of memory: memory file associated with (such as program files, data files corresponding to the memory pages) and has nothing to do with the file memory (such as process stack, the application memory using malloc), the former is called file pages or mapped pages, the latter known as anonymous pages. In the event of paging file pages (page-in or page-out) when a read or write from its corresponding file; anonymous pages when paging occurs, the exchange zone is read / write operations. |
|
|
|