All of lore.kernel.org
 help / color / mirror / Atom feed
From: Christopher Yeoh <cyeoh@au1.ibm.com>
To: Bryan Donlan <bdonlan@gmail.com>
Cc: Avi Kivity <avi@redhat.com>,
	linux-kernel@vger.kernel.org,
	Linux Memory Management List <linux-mm@kvack.org>,
	Ingo Molnar <mingo@elte.hu>
Subject: Re: [RFC][PATCH] Cross Memory Attach
Date: Thu, 16 Sep 2010 10:48:19 +0930	[thread overview]
Message-ID: <20100916104819.36d10acb@lilo> (raw)
In-Reply-To: <AANLkTi=rmUUPCm212Sju-wW==5cT4eqqU+FEP_hX-Z_y@mail.gmail.com>

On Wed, 15 Sep 2010 23:46:09 +0900
Bryan Donlan <bdonlan@gmail.com> wrote:

> On Wed, Sep 15, 2010 at 19:58, Avi Kivity <avi@redhat.com> wrote:
> 
> > Instead of those two syscalls, how about a vmfd(pid_t pid, ulong
> > start, ulong len) system call which returns an file descriptor that
> > represents a portion of the process address space.  You can then
> > use preadv() and pwritev() to copy memory, and
> > io_submit(IO_CMD_PREADV) and io_submit(IO_CMD_PWRITEV) for
> > asynchronous variants (especially useful with a dma engine, since
> > that adds latency).
> >
> > With some care (and use of mmu_notifiers) you can even mmap() your
> > vmfd and access remote process memory directly.
> 
> Rather than introducing a new vmfd() API for this, why not just add
> implementations for these more efficient operations to the existing
> /proc/$pid/mem interface?

Perhaps I'm misunderstanding something here, but
accessing /proc/$pid/mem requires ptracing the target process.
We can't really have all these MPI processes ptraceing each other
just to send/receive a message....

Regards,

Chris
-- 
cyeoh@au.ibm.com

WARNING: multiple messages have this Message-ID (diff)
From: Christopher Yeoh <cyeoh@au1.ibm.com>
To: Bryan Donlan <bdonlan@gmail.com>
Cc: Avi Kivity <avi@redhat.com>,
	linux-kernel@vger.kernel.org,
	Linux Memory Management List <linux-mm@kvack.org>,
	Ingo Molnar <mingo@elte.hu>
Subject: Re: [RFC][PATCH] Cross Memory Attach
Date: Thu, 16 Sep 2010 10:48:19 +0930	[thread overview]
Message-ID: <20100916104819.36d10acb@lilo> (raw)
In-Reply-To: <AANLkTi=rmUUPCm212Sju-wW==5cT4eqqU+FEP_hX-Z_y@mail.gmail.com>

On Wed, 15 Sep 2010 23:46:09 +0900
Bryan Donlan <bdonlan@gmail.com> wrote:

> On Wed, Sep 15, 2010 at 19:58, Avi Kivity <avi@redhat.com> wrote:
> 
> > Instead of those two syscalls, how about a vmfd(pid_t pid, ulong
> > start, ulong len) system call which returns an file descriptor that
> > represents a portion of the process address space.  You can then
> > use preadv() and pwritev() to copy memory, and
> > io_submit(IO_CMD_PREADV) and io_submit(IO_CMD_PWRITEV) for
> > asynchronous variants (especially useful with a dma engine, since
> > that adds latency).
> >
> > With some care (and use of mmu_notifiers) you can even mmap() your
> > vmfd and access remote process memory directly.
> 
> Rather than introducing a new vmfd() API for this, why not just add
> implementations for these more efficient operations to the existing
> /proc/$pid/mem interface?

Perhaps I'm misunderstanding something here, but
accessing /proc/$pid/mem requires ptracing the target process.
We can't really have all these MPI processes ptraceing each other
just to send/receive a message....

Regards,

Chris
-- 
cyeoh@au.ibm.com

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

  parent reply	other threads:[~2010-09-16  1:18 UTC|newest]

Thread overview: 62+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-09-15  1:18 [RFC][PATCH] Cross Memory Attach Christopher Yeoh
2010-09-15  8:02 ` Ingo Molnar
2010-09-15  8:02   ` Ingo Molnar
2010-09-15  8:16   ` Ingo Molnar
2010-09-15  8:16     ` Ingo Molnar
2010-09-15 13:23     ` Christopher Yeoh
2010-09-15 13:23       ` Christopher Yeoh
2010-09-15 13:20   ` Christopher Yeoh
2010-09-15 13:20     ` Christopher Yeoh
2010-09-15 10:58 ` Avi Kivity
2010-09-15 10:58   ` Avi Kivity
2010-09-15 13:51   ` Ingo Molnar
2010-09-15 13:51     ` Ingo Molnar
2010-09-15 16:10     ` Avi Kivity
2010-09-15 16:10       ` Avi Kivity
2010-09-15 14:42   ` Christopher Yeoh
2010-09-15 14:42     ` Christopher Yeoh
2010-09-15 14:52     ` Linus Torvalds
2010-09-15 14:52       ` Linus Torvalds
2010-09-15 15:44       ` Robin Holt
2010-09-15 15:44         ` Robin Holt
2010-09-16  6:32     ` Brice Goglin
2010-09-16  6:32       ` Brice Goglin
2010-09-16  9:15       ` Brice Goglin
2010-09-16  9:15         ` Brice Goglin
2010-09-16 14:00         ` Christopher Yeoh
2010-09-16 14:00           ` Christopher Yeoh
2010-09-15 14:46   ` Bryan Donlan
2010-09-15 14:46     ` Bryan Donlan
2010-09-15 16:13     ` Avi Kivity
2010-09-15 16:13       ` Avi Kivity
2010-09-15 19:35       ` Eric W. Biederman
2010-09-15 19:35         ` Eric W. Biederman
2010-09-16  1:18     ` Christopher Yeoh [this message]
2010-09-16  1:18       ` Christopher Yeoh
2010-09-16  9:26       ` Avi Kivity
2010-09-16  9:26         ` Avi Kivity
2010-11-02  3:37         ` Christopher Yeoh
2010-11-02  3:37           ` Christopher Yeoh
2010-11-02 11:10           ` Avi Kivity
2010-11-02 11:10             ` Avi Kivity
2010-09-16  1:58     ` KOSAKI Motohiro
2010-09-16  1:58       ` KOSAKI Motohiro
2010-09-16  8:08       ` Ingo Molnar
2010-09-16  8:08         ` Ingo Molnar
2010-09-15 15:11 ` Linus Torvalds
2010-09-15 15:14   ` Linus Torvalds
2010-09-16  2:25     ` Christopher Yeoh
2010-09-16 16:27   ` Peter Zijlstra
2010-09-16 16:54     ` Linus Torvalds
2010-09-16 17:13       ` Peter Zijlstra
2010-09-16 17:34         ` Linus Torvalds
2010-09-16 17:47           ` Peter Zijlstra
2010-09-16 17:54             ` Linus Torvalds
2010-09-16 18:00               ` Linus Torvalds
2010-09-19  4:44                 ` Yuhong Bao
2010-09-19 19:20               ` Yuhong Bao
2010-09-19 21:48                 ` Russell King - ARM Linux
2010-09-19 22:47                   ` Yuhong Bao
2010-09-19  4:55           ` Yuhong Bao
2010-09-15 16:07 ` Valdis.Kletnieks
2010-09-16  2:17   ` Christopher Yeoh

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=20100916104819.36d10acb@lilo \
    --to=cyeoh@au1.ibm.com \
    --cc=avi@redhat.com \
    --cc=bdonlan@gmail.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=mingo@elte.hu \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.