From: Kirill Korotaev <dev@sw.ru>
To: Hubertus Franke <frankeh@watson.ibm.com>
Cc: Sam Vilain <sam@vilain.net>, Rik van Riel <riel@redhat.com>,
"Eric W. Biederman" <ebiederm@xmission.com>,
Kirill Korotaev <dev@openvz.org>,
Linus Torvalds <torvalds@osdl.org>, Andrew Morton <akpm@osdl.org>,
linux-kernel@vger.kernel.org, clg@fr.ibm.com,
haveblue@us.ibm.com, greg@kroah.com, alan@lxorguk.ukuu.org.uk,
serue@us.ibm.com, arjan@infradead.org, kuznet@ms2.inr.ac.ru,
saw@sawoct.com, devel@openvz.org, Dmitry Mishin <dim@sw.ru>
Subject: Re: [PATCH 1/4] Virtualization/containers: introduction
Date: Wed, 08 Feb 2006 18:36:40 +0300 [thread overview]
Message-ID: <43EA1008.5040502@sw.ru> (raw)
In-Reply-To: <43E8D160.4040803@watson.ibm.com>
> The pid-namespace (pspace) provides an approach of fully separate
> the allocation and maintenance of the pids and treating the <pspace,pid>
> tuple as an entity to uniquely identify a task and vice versa.
> As a result the logic of lookup can be pushed down the find_task_by_pid()
> lookup. There are specific cases where the init_task of a container or
> pspace needs to be checked to ensure that signals/waits and alike are
> properly
> handled across pspace boundaries. I think this is an intuitive and clean
> way.
> It also completely avoids the problem of having to think about all the
> locations
> at the user/kernel boundary where a vpid/pid conversion needs to take
> place.
> It also avoids the problems that logically vpids and pids are different
> types and
> therefore it would have been good to have type checking automatically
> identify
> problem spots.
> On the negative side, it does require to maintain a pidmap per pidspace.
Additional negative sides:
- full isolation can be inconvinient from containers management point of
view. You will need to introduce new modified tools such as top/ps/kill
and many many others. You won't be able to strace/gdb processes from the
host also.
- overhead when virtualization is off, result is not the same.
- additional args everywhere (stack usage, etc.)
> The vpid approach has the drawbacks of having to identify the conversion
> spots
> of all vpid vs. pid semantics. On the otherhand it does take advantage
> of the fact that no virtualization has to take place until a "container"
> has been migrated, thus rendering most of the vpid<->pid calls to be
> noops.
It has some other additional advantages:
- flexible: you can select full isolation or weak is required. I really
believe this is very important.
> The container is just an umbrella object that ties every "virtualized"
> subsystem
> together.
Yep. And containers were what I wanted to start with actually. Not VPIDs.
Kirill
next prev parent reply other threads:[~2006-02-08 15:35 UTC|newest]
Thread overview: 80+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-02-06 21:57 [PATCH 1/4] Virtualization/containers: introduction Kirill Korotaev
2006-02-06 22:12 ` [PATCH 2/4] Virtualization/containers: CONFIG_CONTAINER Kirill Korotaev
2006-02-06 22:17 ` [PATCH 3/4] Virtualization/containers: UID hash Kirill Korotaev
2006-02-06 22:22 ` [PATCH 4/4] Virtualization/containers: uts name Kirill Korotaev
2006-02-06 23:00 ` [PATCH 1/4] Virtualization/containers: introduction Dave Hansen
2006-02-07 12:24 ` Kirill Korotaev
2006-02-07 3:34 ` Eric W. Biederman
2006-02-07 3:40 ` Rik van Riel
2006-02-07 6:30 ` Sam Vilain
2006-02-07 11:51 ` Kirill Korotaev
2006-02-07 14:31 ` Eric W. Biederman
2006-02-07 15:42 ` Eric W. Biederman
2006-02-07 16:18 ` Kirill Korotaev
2006-02-07 17:20 ` Eric W. Biederman
2006-02-07 22:43 ` Sam Vilain
2006-02-07 16:57 ` Hubertus Franke
2006-02-07 20:19 ` Serge E. Hallyn
2006-02-07 20:46 ` Hubertus Franke
2006-02-07 22:00 ` Eric W. Biederman
2006-02-07 22:19 ` Hubertus Franke
2006-02-07 22:06 ` The issues for agreeing on a virtualization/namespaces implementation Eric W. Biederman
2006-02-07 23:35 ` Hubertus Franke
2006-02-08 0:43 ` Alexey Kuznetsov
2006-02-08 2:49 ` Eric W. Biederman
2006-02-08 3:36 ` Serge E. Hallyn
2006-02-08 3:52 ` Eric W. Biederman
2006-02-08 4:37 ` Herbert Poetzl
2006-02-08 4:46 ` Eric W. Biederman
2006-02-08 19:24 ` Stephen Hemminger
2006-02-08 5:23 ` Eric W. Biederman
2006-02-08 14:40 ` Hubertus Franke
2006-02-08 15:17 ` Serge E. Hallyn
2006-02-08 15:35 ` Kirill Korotaev
2006-02-08 15:57 ` Hubertus Franke
2006-02-08 19:02 ` Herbert Poetzl
2006-02-08 16:48 ` Eric W. Biederman
2006-02-08 17:46 ` Eric W. Biederman
2006-02-08 18:03 ` Serge E. Hallyn
2006-02-08 18:31 ` Hubertus Franke
2006-02-08 20:21 ` Dave Hansen
2006-02-08 21:22 ` Serge E. Hallyn
2006-02-08 22:28 ` Eric W. Biederman
2006-02-20 12:11 ` Kirill Korotaev
2006-02-20 12:41 ` Herbert Poetzl
2006-02-20 14:26 ` Kirill Korotaev
2006-02-20 15:16 ` Herbert Poetzl
2006-02-08 4:56 ` Herbert Poetzl
2006-02-08 14:38 ` Serge E. Hallyn
2006-02-08 14:51 ` Hubertus Franke
2006-02-09 4:45 ` Kyle Moffett
2006-02-09 5:41 ` Eric W. Biederman
2006-02-09 22:25 ` Eric W. Biederman
2006-02-07 22:58 ` [PATCH 1/4] Virtualization/containers: introduction Sam Vilain
2006-02-07 23:18 ` Hubertus Franke
2006-02-08 5:03 ` Eric W. Biederman
2006-02-08 14:13 ` Hubertus Franke
2006-02-08 15:44 ` Kirill Korotaev
2006-02-08 16:39 ` Eric W. Biederman
2006-02-08 2:08 ` Kevin Fox
2006-02-08 1:16 ` Sam Vilain
2006-02-08 4:21 ` Paul Jackson
2006-02-08 15:36 ` Kirill Korotaev [this message]
2006-02-08 17:16 ` Eric W. Biederman
2006-02-08 20:43 ` Dave Hansen
2006-02-08 21:04 ` Eric W. Biederman
2006-02-07 12:14 ` Kirill Korotaev
2006-02-07 14:06 ` Eric W. Biederman
2006-02-07 14:52 ` Rik van Riel
2006-02-07 15:13 ` Eric W. Biederman
2006-02-09 0:24 ` Eric W. Biederman
2006-02-09 2:18 ` Jeff Dike
2006-02-09 3:16 ` Eric W. Biederman
2006-02-09 14:28 ` Kirill Korotaev
2006-02-09 15:40 ` Jeff Dike
2006-02-09 15:49 ` Kirill Korotaev
2006-02-09 17:50 ` Jeff Dike
2006-02-09 16:38 ` Hubertus Franke
2006-02-09 17:48 ` Jeff Dike
2006-02-09 22:09 ` Sam Vilain
2006-02-09 21:56 ` Eric W. Biederman
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=43EA1008.5040502@sw.ru \
--to=dev@sw.ru \
--cc=akpm@osdl.org \
--cc=alan@lxorguk.ukuu.org.uk \
--cc=arjan@infradead.org \
--cc=clg@fr.ibm.com \
--cc=dev@openvz.org \
--cc=devel@openvz.org \
--cc=dim@sw.ru \
--cc=ebiederm@xmission.com \
--cc=frankeh@watson.ibm.com \
--cc=greg@kroah.com \
--cc=haveblue@us.ibm.com \
--cc=kuznet@ms2.inr.ac.ru \
--cc=linux-kernel@vger.kernel.org \
--cc=riel@redhat.com \
--cc=sam@vilain.net \
--cc=saw@sawoct.com \
--cc=serue@us.ibm.com \
--cc=torvalds@osdl.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox