Difference between revisions of "Installation"

From CRIU
Jump to navigation Jump to search
(→‎Configure the linux kernel: use code tags; uniform lists)
Line 7: Line 7:
 
# Only pure x86-64 environment is supported, no IA32 emulation allowed.
 
# Only pure x86-64 environment is supported, no IA32 emulation allowed.
 
# There is no way to use cgroups freezer facility yet.
 
# There is no way to use cgroups freezer facility yet.
# No network or IPC checkpoint/restore supported.
+
# No IPC checkpoint/restore supported, and network is under hard development stage.
 
 
== Basic design ==
 
 
 
=== Checkpoint ===
 
 
 
The checkpoint procedure relies heavily on '''/proc''' file system (it's a general place where crtools takes all the information it needs).
 
Which includes
 
 
 
* Files descriptors information (via '''/proc/$pid/fd''' and '''/proc/$pid/fdinfo''').
 
* Pipes parameters.
 
* Memory maps (via '''/proc/$pid/maps''').
 
 
 
The process dumper (lets call it a dumper further) does the following steps during checkpoint stage
 
 
 
# A '''$pid''' of a process group leader is obtained from the command line.
 
# By using this '''$pid''' the dumper walks though '''/proc/$pid/status''' and gathers children '''$pids''' recursively. At the end we will have a process tree.
 
# Then it takes every '''$pid''' from a process tree, sends ''SIGSTOP'' to every process found, and performs the following steps on each '''$pid'''.
 
#* Collects VMA areas by parsing '''/proc/$pid/maps'''.
 
#* Seizes a task via relatively new ptrace interface. Seizing a task means to put it into a special state when the task have no idea if it's being operated by ptrace.
 
#* Core parameters of a task (such as registers and friends) are being dumped via ptrace interface and parsing '''/proc/$pid/stat''' entry.
 
#* The dumper injects a parasite code into a task via ptrace interface. This allows us to dump pages of a task right from within the task's address space.
 
#** An injection procedure is pretty simple - the dumper scans executable VMA areas of a task (which were collected previously) and tests if there a place for <code>syscall</code> call, then (by ptrace as well) it substitutes an original code with <code>syscall</code> instructions and creates a new VMA area inside process address space.
 
#** Finally parasite code get copied into the new VMA and the former code which was modified during parasite bootstrap procedure get restored.
 
#* Then (by using a parasite code) the dumper flushes contents of a task's pages to the file. And pulls out parasite code block completely, since we don't need it anymore.
 
#* Once parasite removed a task get unseized via ptrace call but it remains stopped still.
 
#* The dumper writes out files and pipes parameter and data.
 
# The procedure continues for every '''$pid'''.
 
 
 
=== Restore ===
 
 
 
The restore procedure (aka restorer) proceed in the following steps
 
 
 
# A process tree has been read from a file.
 
# Every process started with saved (i.e. original) '''$pid''' via <code>clone()</code> call.
 
# Files and pipes are restored (by restored it's meant - they are opened and positioned).
 
# A new memory map is created, filled with data the program had at checkpoint time.
 
# Finally the program is kicked to start with rt_sigreturn system call.
 
 
 
<!--
 
=== Kernel requirements ===
 
 
 
Since checkpoint and restore processes require some help from the Linux kernel, the following kernel patches are needed
 
 
 
* procfs-report-eisdir-when-reading-sysctl-dirs-in-proc.patch
 
* proc-fix-races-against-execve-of-proc-pid-fd.patch
 
* proc-fix-races-against-execve-of-proc-pid-fd-fix.patch
 
* proc-force-dcache-drop-on-unauthorized-access.patch
 
* cr-statfs-callback-for-pipefs
 
 
 
These patches are already in -mm tree and rather a preparation patches for the next series.
 
 
 
* fs-proc-switch-to-dentry
 
* cr-proc-map-files-21
 
 
 
These patches introduce '''/proc/$pid/msp_files'''.
 
 
 
* cr-clone-with-pid-support
 
 
 
This one introduce ability to clone process with specified pid.
 
 
 
* cr-proc-add-children
 
 
 
This one introduce "Children" line to '''/proc/$pid/status'''.
 
 
 
* fs-add-do-close
 
* fs-proc-add-tls
 
* fs-proc-add-mm-task-stat
 
 
 
These ones provides missing pieces of process' information which is needed for checkpoint/restore.
 
 
 
* binfmt-elf-for-cr-5
 
 
 
This one provides new Elf file format.
 
-->
 
  
 
== Download crtools ==
 
== Download crtools ==

Revision as of 20:52, 14 January 2012

What CRtools is

CRtools is an utility to checkpoint/restore process tree. Unlike checkpoint/restore implemented completely in kernel space, it tries to achieve the same target operating in user space. Since the tools and overall concept are still under heavy development stage there are some known limitations applied

  1. Only pure x86-64 environment is supported, no IA32 emulation allowed.
  2. There is no way to use cgroups freezer facility yet.
  3. No IPC checkpoint/restore supported, and network is under hard development stage.

Download crtools

The crtools utility itself is hosted at github. Clone this repo to test new functionality.

Also crtools requires some additional patches to be applied on the linux kernel (on top of v3.2-rc6 to be precise).

So clone linux-2.6-crtools.git, checkout crtools branch and compile the kernel.

Configure the linux kernel

Make sure you have the following options turned on

  • General setup -> Checkpoint/restore support (CONFIG_CHECKPOINT_RESTORE)
  • Networking support -> Networking options -> Unix domain sockets -> UNIX: socket monitoring interface (CONFIG_UNIX_DIAG)
  • Processor type and features -> Enable generic object ID infrastructure (CONFIG_GENERIC_OBJECT_ID)

Note you might have to enable

  • General setup -> Configure standard kernel features (expert users) (CONFIG_EXPERT)

option, which depends on

  • General setup -> Embedded system (CONFIG_EMBEDDED)

(welcome to Kconfig reverse chains hell).