All of lore.kernel.org
 help / color / mirror / Atom feed
From: Zoltan Menyhart <Zoltan.Menyhart@bull.net>
To: Andi Kleen <ak@suse.de>
Cc: Jes Sorensen <jes@sgi.com>, Tony Luck <tony.luck@intel.com>,
	discuss@x86-64.org, linux-kernel@vger.kernel.org,
	libc-alpha@sourceware.org, vojtech@suse.cz,
	linux-ia64@vger.kernel.org
Subject: Re: FOR REVIEW: New x86-64 vsyscall vgetcpu()
Date: Fri, 16 Jun 2006 15:31:12 +0000	[thread overview]
Message-ID: <4492CEC0.2080102@bull.net> (raw)
In-Reply-To: <200606161656.40930.ak@suse.de>

Andi Kleen wrote:

> That is not how user space TLS works. It usually has a base a register.

Can you please give me a real life (simplified) example?

> This means it cannot be cache colored (because you would need a static
> offset) and you couldn't share task_structs on a page.

I do not see the problem. Can you explain please?
E.g. the scheduler pulls a task instead of the current one. The CPU
will see "current->thread_info.cpu"-s of all the tasks at the same
offset anyway.

> Also you would make task_struct part of the userland ABI which
> seems like a very very bad idea to me. It means we couldn't change
> it anymore.

We can make some wrapper, e.g.:

	user_per_cpu_var(name, offset)

"vgetcpu()" would also be added to the ABI which we couldn't change
easily either.

Thanks,

Zoltan

WARNING: multiple messages have this Message-ID (diff)
From: Zoltan Menyhart <Zoltan.Menyhart@bull.net>
To: Andi Kleen <ak@suse.de>
Cc: Jes Sorensen <jes@sgi.com>, Tony Luck <tony.luck@intel.com>,
	discuss@x86-64.org, linux-kernel@vger.kernel.org,
	libc-alpha@sourceware.org, vojtech@suse.cz,
	linux-ia64@vger.kernel.org
Subject: Re: FOR REVIEW: New x86-64 vsyscall vgetcpu()
Date: Fri, 16 Jun 2006 17:31:12 +0200	[thread overview]
Message-ID: <4492CEC0.2080102@bull.net> (raw)
In-Reply-To: <200606161656.40930.ak@suse.de>

Andi Kleen wrote:

> That is not how user space TLS works. It usually has a base a register.

Can you please give me a real life (simplified) example?

> This means it cannot be cache colored (because you would need a static
> offset) and you couldn't share task_structs on a page.

I do not see the problem. Can you explain please?
E.g. the scheduler pulls a task instead of the current one. The CPU
will see "current->thread_info.cpu"-s of all the tasks at the same
offset anyway.

> Also you would make task_struct part of the userland ABI which
> seems like a very very bad idea to me. It means we couldn't change
> it anymore.

We can make some wrapper, e.g.:

	user_per_cpu_var(name, offset)

"vgetcpu()" would also be added to the ABI which we couldn't change
easily either.

Thanks,

Zoltan

  reply	other threads:[~2006-06-16 15:31 UTC|newest]

Thread overview: 69+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-06-14  7:42 FOR REVIEW: New x86-64 vsyscall vgetcpu() Andi Kleen
2006-06-14 10:47 ` Alan Cox
2006-06-14 14:54 ` Steve Munroe
2006-06-15 23:17   ` Benjamin Herrenschmidt
     [not found] ` <449029DB.7030505@redhat.com>
     [not found]   ` <200606141752.02361.ak@suse.de>
2006-06-14 16:30     ` Ulrich Drepper
2006-06-14 17:34       ` [discuss] " Andi Kleen
2006-06-15 18:44 ` Tony Luck
2006-06-16  6:22   ` Andi Kleen
2006-06-16  7:23     ` Gerd Hoffmann
2006-06-16  7:37       ` Andi Kleen
2006-06-16  9:48     ` Jes Sorensen
2006-06-16  9:48       ` Jes Sorensen
2006-06-16 10:09       ` Andi Kleen
2006-06-16 10:09         ` Andi Kleen
2006-06-16 11:02         ` Jes Sorensen
2006-06-16 11:02           ` Jes Sorensen
2006-06-16 11:17           ` Andi Kleen
2006-06-16 11:17             ` Andi Kleen
2006-06-16 11:58             ` Jes Sorensen
2006-06-16 11:58               ` Jes Sorensen
2006-06-16 12:36               ` Zoltan Menyhart
2006-06-16 12:36                 ` Zoltan Menyhart
2006-06-16 12:41                 ` Jes Sorensen
2006-06-16 12:41                   ` Jes Sorensen
2006-06-16 12:48                   ` Zoltan Menyhart
2006-06-16 12:48                     ` Zoltan Menyhart
2006-06-16 21:04                     ` Chase Venters
2006-06-16 21:04                       ` Chase Venters
2006-06-16 14:56                 ` Andi Kleen
2006-06-16 14:56                   ` Andi Kleen
2006-06-16 15:31                   ` Zoltan Menyhart [this message]
2006-06-16 15:31                     ` Zoltan Menyhart
2006-06-16 15:37                     ` Andi Kleen
2006-06-16 15:37                       ` Andi Kleen
2006-06-16 15:58                       ` Jakub Jelinek
2006-06-16 15:58                         ` Jakub Jelinek
2006-06-16 16:24                         ` Andi Kleen
2006-06-16 16:24                           ` Andi Kleen
2006-06-16 16:33                           ` Jakub Jelinek
2006-06-16 16:33                             ` Jakub Jelinek
2006-06-16 21:12                     ` Chase Venters
2006-06-16 21:12                       ` Chase Venters
2006-06-16 15:36                   ` Brent Casavant
2006-06-16 15:36                     ` Brent Casavant
2006-06-16 15:40                     ` Andi Kleen
2006-06-16 15:40                       ` Andi Kleen
2006-06-16 21:15                       ` Chase Venters
2006-06-16 21:15                         ` Chase Venters
2006-06-16 21:19                       ` Chase Venters
2006-06-16 21:19                         ` Chase Venters
2006-06-16 23:40                         ` Brent Casavant
2006-06-16 23:40                           ` Brent Casavant
2006-06-17  6:58                           ` Andi Kleen
2006-06-17  6:58                             ` Andi Kleen
2006-06-17  6:55                         ` [discuss] " Andi Kleen
2006-06-17  6:55                           ` Andi Kleen
2006-06-19  8:42                           ` Zoltan Menyhart
2006-06-19  8:42                             ` Zoltan Menyhart
2006-06-19  8:54                             ` Andi Kleen
2006-06-19  8:54                               ` Andi Kleen
2006-06-16 14:54               ` Andi Kleen
2006-06-16 14:54                 ` Andi Kleen
2006-06-20  8:28                 ` Jes Sorensen
2006-06-20  8:28                   ` Jes Sorensen
2006-06-19  0:15 ` Paul Jackson
2006-06-19  8:21   ` Andi Kleen
2006-06-19 10:09     ` Paul Jackson
2006-06-21  1:18     ` Paul Jackson
2006-06-21  1:21       ` Paul Jackson

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=4492CEC0.2080102@bull.net \
    --to=zoltan.menyhart@bull.net \
    --cc=ak@suse.de \
    --cc=discuss@x86-64.org \
    --cc=jes@sgi.com \
    --cc=libc-alpha@sourceware.org \
    --cc=linux-ia64@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=tony.luck@intel.com \
    --cc=vojtech@suse.cz \
    /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.