LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
From: Dave Hansen <hansendc@us.ibm.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: linux-kernel@vger.kernel.org, hch@infradead.org
Subject: Re: [PATCH 12/22] elevate write count files are open()ed
Date: Tue, 13 Feb 2007 16:17:37 -0800	[thread overview]
Message-ID: <1171412257.30997.21.camel@localhost.localdomain> (raw)
In-Reply-To: <20070213095817.d2a8a380.akpm@linux-foundation.org>

On Tue, 2007-02-13 at 09:58 -0800, Andrew Morton wrote:
> > On Tue, 13 Feb 2007 08:58:16 -0800 Dave Hansen <hansendc@us.ibm.com> wrote:
> > > yipes.  A new mount-wide spin_lock/unlock for each for-writing open() and close().
> > > Can we have a microbenchmark on this please?
> > 
> > Yeah, I'll schedule some dbench time on a NUMA machine.
> 
> dbench doesn't do open() a lot.  To assess the worst-case we'd need one
> process per cpu camping in an open/close loop.

This is definitely a worst-case scenario.  A 32-way x86_64 NUMA machine
(with a pretty crappy interconnect) with a process-per-cpu all beating
on the same filesystem.  

no patch:
real: 30.111s
user: 0.031s
 sys: 2.685s

r/o bind mount patch:
real: 48.359s
user: 0.146s
 sys: 47.984s

It definitely makes a huge difference in system time, although not a
fatal one.  Christoph, what do you think?  Back to caching the
superblock flag in the mount?

#!/bin/sh
# go.sh
name=`uname -r`
grep -q /mnt/ram /proc/mounts || mount -t ramfs ram /mnt/ram;
make openbench;
nr_cpus=`cat /proc/cpuinfo | grep -c ^processor`
for ((run=0;run<5;run++)); do
        dir=$name.run.$run;
        mkdir -p $dir;
        for ((i=0;i<nr_cpus;i++)); do
                { time taskset -c $i ./openbench $((1<<16))  & } \
                > $dir/openbench.time.$i 2>&1
        done;
        wait
        echo run $run done
done

// openbench.c
#include <stdio.h>

#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>

void main(int argc, char **argv)
{
        pid_t pid = getpid();
        char buf[100];
        int ret;
        int fd;
        int i;
        int loops = atoi(argv[1]);
        sprintf(&buf[0], "/mnt/ram/openbench.%d", pid);

        for (i=0; i< loops; i++) {
                fd = open(&buf[0], O_WRONLY|O_CREAT);
                if (fd < 0) {
                        perror("open error");
                        exit(fd);
                }
                write(fd, "foo");
                close(fd);
        }
        ret = unlink(&buf[0]);
        if (ret)
                perror("unlink error");
}

-- Dave


  reply	other threads:[~2007-02-14  0:17 UTC|newest]

Thread overview: 33+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-02-09 22:53 [PATCH 01/22] filesystem helpers for custom 'struct file's Dave Hansen
2007-02-09 22:53 ` [PATCH 02/22] r/o bind mounts: add vfsmount writer counts Dave Hansen
2007-02-09 23:41   ` Eric Dumazet
2007-02-10  0:10     ` Dave Hansen
2007-02-09 22:53 ` [PATCH 03/22] record when sb_writer_count elevated for inode Dave Hansen
2007-02-09 22:53 ` [PATCH 04/22] elevate writer count for chown and friends Dave Hansen
2007-02-09 22:53 ` [PATCH 05/22] elevate mnt writers for callers of vfs_mkdir() Dave Hansen
2007-02-09 22:53 ` [PATCH 06/22] elevate write count during entire ncp_ioctl() Dave Hansen
2007-02-09 22:53 ` [PATCH 07/22] elevate write count for link and symlink calls Dave Hansen
2007-02-09 22:53 ` [PATCH 08/22] elevate mount count for extended attributes Dave Hansen
2007-02-09 22:53 ` [PATCH 09/22] mount_is_safe(): add comment Dave Hansen
2007-02-09 22:53 ` [PATCH 10/22] unix_find_other() elevate write count for touch_atime() Dave Hansen
2007-02-09 22:53 ` [PATCH 12/22] elevate write count files are open()ed Dave Hansen
2007-02-13  5:11   ` Andrew Morton
2007-02-13 16:58     ` Dave Hansen
2007-02-13 17:58       ` Andrew Morton
2007-02-14  0:17         ` Dave Hansen [this message]
2007-02-09 22:53 ` [PATCH 11/22] elevate write count over calls to vfs_rename() Dave Hansen
2007-02-09 22:53 ` [PATCH 13/22] elevate writer count for do_sys_truncate() Dave Hansen
2007-02-09 22:53 ` [PATCH 14/22] elevate write count for do_utimes() Dave Hansen
2007-02-09 22:53 ` [PATCH 15/22] elevate write count for do_sys_utime() and touch_atime() Dave Hansen
2007-02-09 22:53 ` [PATCH 16/22] sys_mknodat(): elevate write count for vfs_mknod/create() Dave Hansen
2007-02-09 22:53 ` [PATCH 17/22] elevate mnt writers for vfs_unlink() callers Dave Hansen
2007-02-09 22:53 ` [PATCH 18/22] do_rmdir(): elevate write count Dave Hansen
2007-02-09 22:53 ` [PATCH 19/22] elevate writer count for custom struct_file Dave Hansen
2007-02-09 22:53 ` [PATCH 20/22] [PATCH] gfs: check nlink count Dave Hansen
2007-02-09 22:53 ` [PATCH 21/22] honor r/w changes at do_remount() time Dave Hansen
2007-02-09 23:22   ` Andrew Morton
2007-02-10  0:00     ` Dave Hansen
2007-02-10  0:29     ` Anton Altaparmakov
2007-02-10  9:54     ` Jan Engelhardt
2007-02-09 22:53 ` [PATCH 22/22] kill open files traverse on remount ro Dave Hansen
2007-02-09 23:18 ` [PATCH 01/22] filesystem helpers for custom 'struct file's Andrew Morton

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=1171412257.30997.21.camel@localhost.localdomain \
    --to=hansendc@us.ibm.com \
    --cc=akpm@linux-foundation.org \
    --cc=hch@infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --subject='Re: [PATCH 12/22] elevate write count files are open()ed' \
    /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

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).