linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Bernardo Innocenti <bernie@develer.com>
To: Christoph Hellwig <hch@infradead.org>,
	"David S. Miller" <davem@redhat.com>
Cc: uclinux-dev@uclinux.org, linux-kernel@vger.kernel.org
Subject: Re: [uClinux-dev] Kernel 2.6 size increase - get_current()?
Date: Thu, 24 Jul 2003 00:35:38 +0200	[thread overview]
Message-ID: <200307240035.38502.bernie@develer.com> (raw)
In-Reply-To: <20030723212755.A608@infradead.org>

On Wednesday 23 July 2003 22:27, Christoph Hellwig wrote:

> On Wed, Jul 23, 2003 at 01:22:56PM -0700, David S. Miller wrote:
> > Drivers weren't audited much, and there's a lot of boneheaded
> > stuff in this area.  But these should be mostly identical
> > to what would happen on the 2.4.x side
>
> Please read the original message again - he stated that every single
> module in fs/ got alot bigger - if it gets smaller or at least the
> same size as 2.4 it's clearly a sign of inlines gone mad in the
> filesystem/VM code and we need to look at that.  If not we have to look
> elsewhere.

I have my humbling opinion:

In 2.4.20 (m68knommu):
-------------------------------------------------------------------------
#define current _current_task
-------------------------------------------------------------------------

In 2.6.0-test1 (m68knommu):
-------------------------------------------------------------------------
#define current get_current()
static inline struct task_struct *get_current(void)
{
        return(current_thread_info()->task);
}
static inline struct thread_info *current_thread_info(void)
{
        struct thread_info *ti;
        __asm__(
                "move.l %%sp, %0 \n\t"
                "and.l  %1, %0"
                : "=&d"(ti)
                : "d" (~(THREAD_SIZE-1))
                );
        return ti;
}
-------------------------------------------------------------------------

The latter expands to:

 0:	movel #-8192,%d0
 6:	movel %sp,%d2
 8:	andl %d0,%d2
 a:	moveal %d2,%a1
 c:	moveal %a1@,%a0
 e:	moveal %a0@(92),%a0
12:

It's a sequence of 6 instructions, 18 bytes long, clobbering 4 registers.
The compiler cannot see around it.

"current" is being used very lightly all over the kernel, like in this
code snippet from fs/open.c:

        old_fsuid = current->fsuid;
        old_fsgid = current->fsgid;
        old_cap = current->cap_effective;
        current->fsuid = current->uid;
        current->fsgid = current->gid;
        if (current->uid)
                cap_clear(current->cap_effective);
        else
                current->cap_effective = current->cap_permitted;

This takes 18*11 = 198 bytes just for invoking the 'current'
macro so many times.

Perhaps adding __attribute__((const)) on current_thread_info() and
get_current() would help eliminating some unnecessary accesses.

-- 
  // Bernardo Innocenti - Develer S.r.l., R&D dept.
\X/  http://www.develer.com/

Please don't send Word attachments - http://www.gnu.org/philosophy/no-word-attachments.html



  reply	other threads:[~2003-07-23 22:20 UTC|newest]

Thread overview: 64+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2003-07-23 18:46 Kernel 2.6 size increase Bernardo Innocenti
2003-07-23 19:14 ` Richard B. Johnson
2003-07-23 20:07   ` David S. Miller
2003-07-23 21:47     ` Randy.Dunlap
2003-07-23 19:32 ` [uClinux-dev] " Christoph Hellwig
2003-07-23 20:11   ` David S. Miller
2003-07-23 20:15     ` Christoph Hellwig
2003-07-23 20:22       ` David S. Miller
2003-07-23 20:27         ` Christoph Hellwig
2003-07-23 22:35           ` Bernardo Innocenti [this message]
2003-07-23 22:37             ` [uClinux-dev] Kernel 2.6 size increase - get_current()? Alan Cox
2003-07-23 23:00               ` Bernardo Innocenti
2003-07-24  5:06                 ` David McCullough
2003-07-24 11:28                   ` Alan Cox
2003-07-24 12:04                     ` David McCullough
2003-07-24 14:48                       ` Alan Cox
2003-07-25 18:25                       ` bill davidsen
2003-07-24 15:30                     ` Hollis Blanchard
2003-07-24 19:37                       ` Alan Cox
2003-07-24 19:51                         ` Hollis Blanchard
2003-07-24 21:20                           ` J.A. Magallon
2003-07-25  4:22                             ` Otto Solares
2003-07-25 14:38                               ` Hollis Blanchard
2003-07-28  3:19                                 ` Miles Bader
2003-07-28  8:14                                   ` Ihar "Philips" Filipau
2003-07-28  8:58                                     ` Miles Bader
2003-07-28  9:03                                       ` Ihar "Philips" Filipau
2003-07-23 21:57   ` [uClinux-dev] Kernel 2.6 size increase Bernardo Innocenti
2003-07-23 22:07     ` Bernardo Innocenti
2003-07-23 22:27       ` Willy Tarreau
2003-07-23 22:34         ` Alan Cox
2003-07-24 20:27         ` Bernardo Innocenti
2003-07-25 15:46           ` Christoph Hellwig
2003-07-25 23:55             ` [PATCH] Make I/O schedulers optional (Was: Re: Kernel 2.6 size increase) Bernardo Innocenti
2003-07-26  8:17               ` Andrew Morton
2003-07-26 12:40                 ` Bernardo Innocenti
2003-07-26 14:07                   ` Jens Axboe
2003-07-26 23:42                     ` Bernardo Innocenti
2003-07-26 23:41                       ` Jens Axboe
2003-07-28 17:13             ` [uClinux-dev] Kernel 2.6 size increase Nicolas Pitre
2003-07-28 23:02               ` Bernardo Innocenti
2003-07-29  2:36                 ` Miles Bader
2003-08-08 13:25                 ` [uClinux-dev] " David Woodhouse
2003-08-08 14:37                   ` Bernardo Innocenti
2003-08-08 14:43                     ` David Woodhouse
2003-07-25 18:16           ` bill davidsen
2003-07-29 22:29           ` Tom Rini
2003-07-29 22:48             ` Alan Cox
2003-07-29 23:06               ` Tom Rini
2003-07-30  2:07                 ` Miles Bader
2003-07-30 15:33                   ` Tom Rini
2003-07-31  1:49                     ` Miles Bader
2003-07-31  4:17                       ` Tom Rini
2003-07-31  5:03                         ` Miles Bader
2003-07-31 15:24                           ` Tom Rini
2003-07-30  2:49             ` [uClinux-dev] " Bernardo Innocenti
2003-07-30 15:35               ` Tom Rini
2003-07-30 16:45                 ` [uClinux-dev] Kernel 2.6 size increase (PATCH) Bernardo Innocenti
     [not found] <cwQJ.3BO.29@gated-at.bofh.it>
     [not found] ` <cypH.5dM.35@gated-at.bofh.it>
     [not found]   ` <cyza.5lN.13@gated-at.bofh.it>
     [not found]     ` <cArg.74D.11@gated-at.bofh.it>
2003-07-24  8:13       ` [uClinux-dev] Kernel 2.6 size increase - get_current()? Ihar "Philips" Filipau
2003-07-25  7:25         ` Denis Vlasenko
2003-07-25 18:36         ` bill davidsen
2003-07-24  8:27 Ihar "Philips" Filipau
2003-07-24 11:50 ` David McCullough
     [not found] <d2nx.4QV.15@gated-at.bofh.it>
     [not found] ` <dbTZ.5Z5.19@gated-at.bofh.it>
2003-07-25 15:37   ` Ihar "Philips" Filipau

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=200307240035.38502.bernie@develer.com \
    --to=bernie@develer.com \
    --cc=davem@redhat.com \
    --cc=hch@infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=uclinux-dev@uclinux.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;
as well as URLs for NNTP newsgroup(s).