linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Michel Lespinasse <walken@google.com>
To: riel@redhat.com, peterz@infradead.org, vrajesh@umich.edu,
	daniel.santos@pobox.com, aarcange@redhat.com,
	dwmw2@infradead.org, akpm@linux-foundation.org
Cc: linux-mm@kvack.org, linux-kernel@vger.kernel.org,
	torvalds@linux-foundation.org
Subject: Re: [PATCH 5/5] rbtree: move augmented rbtree functionality to rbtree_augmented.h
Date: Tue, 7 Aug 2012 18:19:12 -0700	[thread overview]
Message-ID: <CANN689EcSPkawZMQC-L-odANez+T0mVg_w4v6iOLL5WHcKACfw@mail.gmail.com> (raw)
In-Reply-To: <1344324343-3817-6-git-send-email-walken@google.com>

On Tue, Aug 7, 2012 at 12:25 AM, Michel Lespinasse <walken@google.com> wrote:
> Provide rb_insert_augmented() and rb_erase_augmented through
> a new rbtree_augmented.h include file. rb_erase_augmented() is defined
> there as an __always_inline function, in order to allow inlining of
> augmented rbtree callbacks into it. Since this generates a relatively
> large function, each augmented rbtree users should make sure to
> have a single call site.

I should probably add this to show how the code generation works out
in practice (with a gcc 4.6 based compiler)

Before this change:

       text    data     bss     dec     hex filename
       3426       0       0    3426     d62 lib/rbtree.o

    0000000000000af0 g     F .text  000000000000001e rb_last
    0000000000000b80 g     F .text  0000000000000047 rb_next
    0000000000000000 g     F .text  0000000000000165 rb_insert_color
    0000000000000bd0 g     F .text  0000000000000047 rb_prev
    00000000000004e0 g     F .text  00000000000001cd __rb_insert_augmented
    0000000000000170 g     F .text  000000000000036f rb_erase
    00000000000006b0 g     F .text  0000000000000416 rb_erase_augmented
    0000000000000ad0 g     F .text  000000000000001e rb_first
    0000000000000b10 g     F .text  000000000000006e rb_replace_node

       text    data     bss     dec     hex filename
        540       0       0     540     21c lib/interval_tree.o

    0000000000000000 l     F .text  0000000000000054
interval_tree_augment_propagate
    0000000000000060 l     F .text  000000000000000e interval_tree_augment_copy
    0000000000000070 l     F .text  000000000000003e
interval_tree_augment_rotate
    00000000000000b0 g     F .text  0000000000000065 interval_tree_insert
    0000000000000120 g     F .text  0000000000000012 interval_tree_remove
    0000000000000140 g     F .text  000000000000004c interval_tree_iter_first
    0000000000000190 g     F .text  0000000000000074 interval_tree_iter_next

After this change:

       text    data     bss     dec     hex filename
       2976       0       0    2976     ba0 lib/rbtree.o

    0000000000000000 g     F .text  000000000000025c __rb_erase_color
    0000000000000930 g     F .text  000000000000001e rb_last
    00000000000009c0 g     F .text  0000000000000047 rb_next
    0000000000000260 g     F .text  0000000000000165 rb_insert_color
    0000000000000a10 g     F .text  0000000000000047 rb_prev
    0000000000000740 g     F .text  00000000000001cd __rb_insert_augmented
    00000000000003d0 g     F .text  000000000000036f rb_erase
    0000000000000910 g     F .text  000000000000001e rb_first
    0000000000000950 g     F .text  000000000000006e rb_replace_node

       text    data     bss     dec     hex filename
        900       0       0     900     384 lib/interval_tree.o

    0000000000000000 l     F .text  000000000000003e
interval_tree_augment_rotate
    0000000000000040 g     F .text  0000000000000065 interval_tree_insert
    00000000000000b0 g     F .text  000000000000020b interval_tree_remove
    00000000000002c0 g     F .text  000000000000004c interval_tree_iter_first
    0000000000000310 g     F .text  0000000000000074 interval_tree_iter_next

So the code size effect is that the library code for augmented erase
shrinks by 450 bytes, and each augmented rb_erase user grows by (well,
this will very between users, but I think interval tree is
representative of a typical user) ~360 bytes. The rotate callback is
generated as a static function so it can be passed by pointer to the
library rebalancing routines; the copy and propagate functions are
inlined into interval_tree_remove (and the compiler is able to
determine that there are no remaining non-inlined calls, so it doesn't
generate an additional static definition).

-- 
Michel "Walken" Lespinasse
A program is never fully debugged until the last user dies.

--
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>

  reply	other threads:[~2012-08-08  1:19 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-08-07  7:25 [PATCH 0/5] rbtree based interval tree as a prio_tree replacement Michel Lespinasse
2012-08-07  7:25 ` [PATCH 1/5] rbtree: add prio tree and interval tree tests Michel Lespinasse
2012-08-07  7:25 ` [PATCH 2/5] mm: replace vma prio_tree with an interval tree Michel Lespinasse
2012-08-14 12:11   ` Hillf Danton
2012-08-07  7:25 ` [PATCH 3/5] kmemleak: use rbtree instead of prio tree Michel Lespinasse
2012-08-08 17:07   ` Michel Lespinasse
2012-08-09  8:31     ` Catalin Marinas
2012-08-15 16:36       ` Catalin Marinas
2012-08-15 20:53         ` Michel Lespinasse
2012-08-16 15:06           ` Catalin Marinas
2012-08-07  7:25 ` [PATCH 4/5] prio_tree: remove Michel Lespinasse
2012-08-07  7:25 ` [PATCH 5/5] rbtree: move augmented rbtree functionality to rbtree_augmented.h Michel Lespinasse
2012-08-08  1:19   ` Michel Lespinasse [this message]
2012-08-13  8:20 ` [PATCH 0/5] rbtree based interval tree as a prio_tree replacement Peter Zijlstra
2012-08-13 10:37   ` Michel Lespinasse
2012-08-30 21:34 ` Andrew Morton
2012-08-30 21:43   ` Rik van Riel
2012-08-30 22:33   ` Michel Lespinasse

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=CANN689EcSPkawZMQC-L-odANez+T0mVg_w4v6iOLL5WHcKACfw@mail.gmail.com \
    --to=walken@google.com \
    --cc=aarcange@redhat.com \
    --cc=akpm@linux-foundation.org \
    --cc=daniel.santos@pobox.com \
    --cc=dwmw2@infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=peterz@infradead.org \
    --cc=riel@redhat.com \
    --cc=torvalds@linux-foundation.org \
    --cc=vrajesh@umich.edu \
    /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