Difference between revisions of "Kcmp trees"

From CRIU
Jump to navigation Jump to search
Line 1: Line 1:
FIXME
+
== Overview ==
  
- write about type of trees
+
Usually we dump not just a single process but a set of them, where every process may be sharing some resources with other processes.
- use of kcmp syscall
+
Thus we need somehow to distinguish which resources are shared and which are not.
- internal representation
 
  
 +
For this sake [http://man7.org/linux/man-pages/man2/kcmp.2.html kcmp] system call has been introduced to Linux kernel.
 +
It takes two processes and compare a resource asked, returning result similar to well known
 +
[http://man7.org/linux/man-pages/man3/strcmp.3.html strcmp] call. This allows CRIU to track resources with a sorting algorithm.
 +
 +
=== Kcmp-trees ===
 +
 +
CRIU gather files, filesystems, vitrual memory descriptors, signal handlers and file descriptors associated with a process
 +
each into Kcmp-tree. Thus at moment we are carrying five Kcmp-trees. Each declared with <code>DECLARE_KCMP_TREE</code> helper.
 +
For example
 +
 +
<pre>DECLARE_KCMP_TREE(vm_tree, KCMP_VM);</pre>
 +
 +
Each tree internally represented as [http://en.wikipedia.org/wiki/Red%E2%80%93black_tree red-black] tree.
 +
 +
=== Requesting new ID ===
 +
 +
When CRIU gathers process resources it check if a resource is already sitting inside of a tree calling
 +
<code>kid_generate_gen()</code> helper. If a resource is not in a tree - it pushed into a tree
 +
and a caller obtains new abstract ID which may be used inside CRIU images, otherwise the helper
 +
returns zero notifying that this kind of resource already known to CRIU and has been handled earlier.
 +
 +
This feature is quite important to eliminate duplication of entries inside CRIU dump images, because
 +
two processes might share a lot of resources and dumping them multiple times would cause very serious
 +
performance issue.
  
 
[[Category: Under the hood]]
 
[[Category: Under the hood]]

Revision as of 21:34, 16 December 2014

Overview

Usually we dump not just a single process but a set of them, where every process may be sharing some resources with other processes. Thus we need somehow to distinguish which resources are shared and which are not.

For this sake kcmp system call has been introduced to Linux kernel. It takes two processes and compare a resource asked, returning result similar to well known strcmp call. This allows CRIU to track resources with a sorting algorithm.

Kcmp-trees

CRIU gather files, filesystems, vitrual memory descriptors, signal handlers and file descriptors associated with a process each into Kcmp-tree. Thus at moment we are carrying five Kcmp-trees. Each declared with DECLARE_KCMP_TREE helper. For example

DECLARE_KCMP_TREE(vm_tree, KCMP_VM);

Each tree internally represented as red-black tree.

Requesting new ID

When CRIU gathers process resources it check if a resource is already sitting inside of a tree calling kid_generate_gen() helper. If a resource is not in a tree - it pushed into a tree and a caller obtains new abstract ID which may be used inside CRIU images, otherwise the helper returns zero notifying that this kind of resource already known to CRIU and has been handled earlier.

This feature is quite important to eliminate duplication of entries inside CRIU dump images, because two processes might share a lot of resources and dumping them multiple times would cause very serious performance issue.