From: Anton Blanchard <anton@samba.org>
To: Nick Piggin <npiggin@suse.de>
Cc: Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
linux-fsdevel@vger.kernel.org,
Ravikiran G Thirumalai <kiran@scalex86.org>,
Peter Zijlstra <peterz@infradead.org>,
Linus Torvalds <torvalds@linux-foundation.org>,
Jens Axboe <axboe@kernel.dk>
Subject: Re: Latest vfs scalability patch
Date: Thu, 15 Oct 2009 22:23:29 +1100 [thread overview]
Message-ID: <20091015112329.GH4808@kryten> (raw)
In-Reply-To: <20091015105332.GB3127@wotan.suse.de>
[-- Attachment #1: Type: text/plain, Size: 744 bytes --]
Hi Nick,
> I wonder what other good performance tests you can add to your test
> framework? creat/unlink is another easy one. And for each case, putting
> threads in their own cwd versus a common cwd are the variants.
I did try the two combinations of creat/unlink but haven't had a chance to
digest the profiles yet. I've attached them (taken at 64 cores, ie worst
case :)
In both cases performance was significantly better than mainline.
> BTW. for these cases in your tests it will be nice if you can run on
> ramfs because that will isolate purely the vfs. Perhaps also include
> other filesystems as you get time, but I think ramfs is the most
> useful for us to start with.
Good point. I'll add that into the setup scripts.
Anton
[-- Attachment #2: open_unlink.profile --]
[-- Type: text/plain, Size: 10567 bytes --]
# Samples: 82617
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
99.16% unlink1_process [kernel] [k] ._spin_lock
|
|--99.98%-- ._spin_lock
| |
| |--49.80%-- .path_get
| | (nil)
| | |
| | |--71.62%-- .path_init
| | | |
| | | |--51.37%-- .do_path_lookup
| | | | .user_path_parent
| | | | .do_unlinkat
| | | | syscall_exit
| | | | 0xfff8e90bd4c
| | | | .testcase
| | | | .affinitize
| | | | .new_task
| | | | .main
| | | | 0xfff8e84f33c
| | | | 0xfff8e84f55c
| | | | (nil)
| | | |
| | | --48.63%-- .do_filp_open
| | | .do_sys_open
| | | syscall_exit
| | | 0xfff8e909c7c
| | | .testcase
| | | .affinitize
| | | .new_task
| | | .main
| | | 0xfff8e84f33c
| | | 0xfff8e84f55c
| | | (nil)
| | |
| | --28.38%-- .path_walk
| | |
| | |--50.49%-- .do_filp_open
| | | .do_sys_open
| | | syscall_exit
| | | 0xfff8e909c7c
| | | .testcase
| | | .affinitize
| | | .new_task
| | | .main
| | | 0xfff8e84f33c
| | | 0xfff8e84f55c
| | | (nil)
| | |
| | --49.51%-- .do_path_lookup
| | .user_path_parent
| | .do_unlinkat
| | syscall_exit
| | 0xfff8e90bd4c
| | .testcase
| | .affinitize
| | .new_task
| | .main
| | 0xfff8e84f33c
| | 0xfff8e84f55c
| | (nil)
| |
| |--49.58%-- .dput
| | (nil)
| | |
| | |--64.21%-- .path_put
| | | |
| | | |--48.48%-- .path_walk
| | | | |
| | | | |--51.00%-- .do_path_lookup
| | | | | .user_path_parent
| | | | | .do_unlinkat
| | | | | syscall_exit
| | | | | 0xfff8e90bd4c
| | | | | .testcase
| | | | | .affinitize
| | | | | .new_task
| | | | | .main
| | | | | 0xfff8e84f33c
| | | | | 0xfff8e84f55c
| | | | | (nil)
| | | | |
| | | | --49.00%-- .do_filp_open
| | | | .do_sys_open
| | | | syscall_exit
| | | | 0xfff8e909c7c
| | | | .testcase
| | | | .affinitize
| | | | .new_task
| | | | .main
| | | | 0xfff8e84f33c
| | | | 0xfff8e84f55c
| | | | (nil)
| | | |
| | | |--25.79%-- .do_filp_open
| | | | .do_sys_open
| | | | syscall_exit
| | | | 0xfff8e909c7c
| | | | .testcase
| | | | .affinitize
| | | | .new_task
| | | | .main
| | | | 0xfff8e84f33c
| | | | 0xfff8e84f55c
| | | | (nil)
| | | |
| | | |--25.72%-- .do_path_lookup
| | | | .user_path_parent
| | | | .do_unlinkat
| | | | syscall_exit
| | | | 0xfff8e90bd4c
| | | | .testcase
| | | | .affinitize
| | | | .new_task
| | | | .main
| | | | 0xfff8e84f33c
| | | | 0xfff8e84f55c
| | | | (nil)
| | | --0.00%-- [...]
| | |
| | |--35.78%-- .__link_path_walk
| | | .path_walk
| | | |
| | | |--50.81%-- .do_path_lookup
| | | | .user_path_parent
| | | | .do_unlinkat
| | | | syscall_exit
| | | | 0xfff8e90bd4c
| | | | .testcase
| | | | .affinitize
| | | | .new_task
| | | | .main
| | | | 0xfff8e84f33c
| | | | 0xfff8e84f55c
| | | | (nil)
| | | |
| | | --49.19%-- .do_filp_open
| | | .do_sys_open
| | | syscall_exit
| | | 0xfff8e909c7c
| | | .testcase
| | | .affinitize
| | | .new_task
| | | .main
| | | 0xfff8e84f33c
| | | 0xfff8e84f55c
| | | (nil)
| | --0.01%-- [...]
| --0.62%-- [...]
--0.02%-- [...]
[-- Attachment #3: open_unlink_directories.profile --]
[-- Type: text/plain, Size: 12730 bytes --]
# Samples: 101253
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
^[[31m 99.15%^[[m unlink2_process [kernel] [k] ._spin_lock
|
|^[[31m--99.97%-- ^[[m._spin_lock
| |
| |^[[31m--49.96%-- ^[[m.dput
| | (nil)
| | |
| | |^[[31m--71.64%-- ^[[m.__link_path_walk
| | | .path_walk
| | | |
| | | |^[[31m--50.78%-- ^[[m.do_path_lookup
| | | | .user_path_parent
| | | | .do_unlinkat
| | | | syscall_exit
| | | | 0xfff86a85d4c
| | | | .testcase
| | | | .affinitize
| | | | .new_task
| | | | .main
| | | | 0xfff869c933c
| | | | 0xfff869c955c
| | | | (nil)
| | | |
| | | ^[[31m--49.22%-- ^[[m.do_filp_open
| | | .do_sys_open
| | | syscall_exit
| | | 0xfff86a83c7c
| | | .testcase
| | | .affinitize
| | | .new_task
| | | .main
| | | 0xfff869c933c
| | | 0xfff869c955c
| | | (nil)
| | |
| | |^[[31m--28.35%-- ^[[m.path_put
| | | |
| | | |^[[31m--50.91%-- ^[[m.path_walk
| | | | |
| | | | |^[[31m--53.90%-- ^[[m.do_path_lookup
| | | | | .user_path_parent
| | | | | .do_unlinkat
| | | | | syscall_exit
| | | | | 0xfff86a85d4c
| | | | | .testcase
| | | | | .affinitize
| | | | | .new_task
| | | | | .main
| | | | | 0xfff869c933c
| | | | | 0xfff869c955c
| | | | | (nil)
| | | | |
| | | | ^[[31m--46.10%-- ^[[m.do_filp_open
| | | | .do_sys_open
| | | | syscall_exit
| | | | 0xfff86a83c7c
| | | | .testcase
| | | | .affinitize
| | | | .new_task
| | | | .main
| | | | 0xfff869c933c
| | | | 0xfff869c955c
| | | | (nil)
| | | |
| | | |^[[31m--25.98%-- ^[[m.do_filp_open
| | | | .do_sys_open
| | | | syscall_exit
| | | | 0xfff86a83c7c
| | | | .testcase
| | | | .affinitize
| | | | .new_task
| | | | .main
| | | | 0xfff869c933c
| | | | 0xfff869c955c
| | | | (nil)
| | | |
| | | |^[[31m--23.10%-- ^[[m.do_path_lookup
| | | | .user_path_parent
| | | | .do_unlinkat
| | | | syscall_exit
| | | | 0xfff86a85d4c
| | | | .testcase
| | | | .affinitize
| | | | .new_task
| | | | .main
| | | | 0xfff869c933c
| | | | 0xfff869c955c
| | | | (nil)
| | | --0.01%-- [...]
| | --0.01%-- [...]
| |
| |^[[31m--28.88%-- ^[[m.__follow_mount
| | (nil)
| | .do_lookup
| | .__link_path_walk
| | .path_walk
| | |
| | |^[[31m--50.26%-- ^[[m.do_filp_open
| | | .do_sys_open
| | | syscall_exit
| | | 0xfff86a83c7c
| | | .testcase
| | | .affinitize
| | | .new_task
| | | .main
| | | 0xfff869c933c
| | | 0xfff869c955c
| | | (nil)
| | |
| | ^[[31m--49.74%-- ^[[m.do_path_lookup
| | .user_path_parent
| | .do_unlinkat
| | syscall_exit
| | 0xfff86a85d4c
| | .testcase
| | .affinitize
| | .new_task
| | .main
| | 0xfff869c933c
| | 0xfff869c955c
| | (nil)
| |
| |^[[31m--20.33%-- ^[[m.path_get
| | (nil)
| | |
| | |^[[31m--69.39%-- ^[[m.path_init
| | | |
| | | |^[[31m--50.78%-- ^[[m.do_path_lookup
| | | | .user_path_parent
| | | | .do_unlinkat
| | | | syscall_exit
| | | | 0xfff86a85d4c
| | | | .testcase
| | | | .affinitize
| | | | .new_task
| | | | .main
| | | | 0xfff869c933c
| | | | 0xfff869c955c
| | | | (nil)
| | | |
| | | ^[[31m--49.22%-- ^[[m.do_filp_open
| | | .do_sys_open
| | | syscall_exit
| | | 0xfff86a83c7c
| | | .testcase
| | | .affinitize
| | | .new_task
| | | .main
| | | 0xfff869c933c
| | | 0xfff869c955c
| | | (nil)
| | |
| | ^[[31m--30.61%-- ^[[m.path_walk
| | |
| | |^[[31m--53.29%-- ^[[m.do_filp_open
| | | .do_sys_open
| | | syscall_exit
| | | 0xfff86a83c7c
| | | .testcase
| | | .affinitize
| | | .new_task
| | | .main
| | | 0xfff869c933c
| | | 0xfff869c955c
| | | (nil)
| | |
| | ^[[31m--46.71%-- ^[[m.do_path_lookup
| | .user_path_parent
| | .do_unlinkat
| | syscall_exit
| | 0xfff86a85d4c
| | .testcase
| | .affinitize
| | .new_task
| | .main
| | 0xfff869c933c
| | 0xfff869c955c
| | (nil)
| ^[[32m--0.83%-- ^[[m[...]
--0.03%-- [...]
next prev parent reply other threads:[~2009-10-15 11:26 UTC|newest]
Thread overview: 57+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-10-06 6:49 Latest vfs scalability patch Nick Piggin
2009-10-06 10:14 ` Jens Axboe
2009-10-06 10:26 ` Jens Axboe
2009-10-06 11:10 ` Peter Zijlstra
2009-10-06 12:51 ` Jens Axboe
2009-10-06 12:26 ` Nick Piggin
2009-10-06 12:49 ` Jens Axboe
2009-10-07 8:58 ` [rfc][patch] store-free path walking Nick Piggin
2009-10-07 9:56 ` Jens Axboe
2009-10-07 10:10 ` Nick Piggin
2009-10-12 3:58 ` Nick Piggin
2009-10-12 5:59 ` Nick Piggin
2009-10-12 8:20 ` Jens Axboe
2009-10-12 11:00 ` Jens Axboe
2009-10-13 1:26 ` Christoph Hellwig
2009-10-13 1:52 ` Nick Piggin
2009-10-07 14:56 ` Linus Torvalds
2009-10-07 16:27 ` Linus Torvalds
2009-10-07 16:46 ` Nick Piggin
2009-10-07 19:25 ` Linus Torvalds
2009-10-07 20:34 ` Andi Kleen
2009-10-07 20:51 ` Linus Torvalds
2009-10-07 21:06 ` Andi Kleen
2009-10-07 21:20 ` Linus Torvalds
2009-10-07 21:57 ` Linus Torvalds
2009-10-07 22:22 ` Andi Kleen
2009-10-08 7:39 ` Nick Piggin
2009-10-09 17:53 ` Andi Kleen
2009-10-08 13:12 ` Denys Vlasenko
2009-10-09 7:47 ` Nick Piggin
2009-10-09 17:49 ` Andi Kleen
2009-10-07 16:29 ` Nick Piggin
2009-10-08 12:36 ` Nick Piggin
2009-10-08 12:57 ` Jens Axboe
2009-10-08 13:22 ` Nick Piggin
2009-10-08 13:30 ` Jens Axboe
2009-10-08 18:00 ` Peter Zijlstra
2009-10-09 4:04 ` Nick Piggin
2009-10-09 8:54 ` Jens Axboe
2009-10-09 9:51 ` Jens Axboe
2009-10-09 10:02 ` Nick Piggin
2009-10-09 10:08 ` Jens Axboe
2009-10-09 10:07 ` Nick Piggin
2009-10-09 3:50 ` Nick Piggin
2009-10-09 6:15 ` David Miller
2009-10-09 10:40 ` Nick Piggin
2009-10-09 11:09 ` Jens Axboe
2009-10-09 10:44 ` Nick Piggin
2009-10-09 10:48 ` Jens Axboe
2009-10-09 23:16 ` Paul E. McKenney
2009-10-15 10:08 ` Latest vfs scalability patch Anton Blanchard
2009-10-15 10:39 ` Nick Piggin
2009-10-15 10:46 ` Anton Blanchard
2009-10-15 10:53 ` Nick Piggin
2009-10-15 11:23 ` Anton Blanchard [this message]
2009-10-15 11:41 ` Nick Piggin
2009-10-15 11:48 ` Nick Piggin
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=20091015112329.GH4808@kryten \
--to=anton@samba.org \
--cc=axboe@kernel.dk \
--cc=kiran@scalex86.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=npiggin@suse.de \
--cc=peterz@infradead.org \
--cc=torvalds@linux-foundation.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 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.