https://github.com/torvalds/linux
Revision e8a03feb54ca7f1768bbdc2b491f9ef654e6d01d authored by Rik van Riel on 14 April 2010, 21:59:28 UTC, committed by Linus Torvalds on 19 April 2010, 23:28:20 UTC
The recent anon_vma fixes cause many anonymous pages to end up
in the parent process anon_vma, even when the page is exclusively
owned by the current process.

Adding exclusively owned anonymous pages to the top anon_vma
reduces rmap scanning overhead, especially in workloads with
forking servers.

This patch adds a parameter to __page_set_anon_rmap that can
be used to indicate whether or not the added page is exclusively
owned by the current process.

Pages added through page_add_new_anon_rmap are exclusively
owned by the current process, and can be added to the top
anon_vma.

Pages added through page_add_anon_rmap can be either shared
or exclusively owned, so we do the conservative thing and
add it to the oldest anon_vma.

A next step would be to add the exclusive parameter to
page_add_anon_rmap, to be used from functions where we do
know for sure whether a page is exclusively owned.

Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Johannes Weiner <hannes@cmpxchg.org>
Lightly-tested-by: Borislav Petkov <bp@alien8.de>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
[ Edited to look nicer  - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1 parent 9b030e2
Raw File
Tip revision: e8a03feb54ca7f1768bbdc2b491f9ef654e6d01d authored by Rik van Riel on 14 April 2010, 21:59:28 UTC
rmap: add exclusively owned pages to the newest anon_vma
Tip revision: e8a03fe
nfsctl.c
/*
 *	fs/nfsctl.c
 *
 *	This should eventually move to userland.
 *
 */
#include <linux/types.h>
#include <linux/file.h>
#include <linux/fs.h>
#include <linux/nfsd/syscall.h>
#include <linux/cred.h>
#include <linux/sched.h>
#include <linux/linkage.h>
#include <linux/namei.h>
#include <linux/mount.h>
#include <linux/syscalls.h>
#include <asm/uaccess.h>

/*
 * open a file on nfsd fs
 */

static struct file *do_open(char *name, int flags)
{
	struct nameidata nd;
	struct vfsmount *mnt;
	int error;

	mnt = do_kern_mount("nfsd", 0, "nfsd", NULL);
	if (IS_ERR(mnt))
		return (struct file *)mnt;

	error = vfs_path_lookup(mnt->mnt_root, mnt, name, 0, &nd);
	mntput(mnt);	/* drop do_kern_mount reference */
	if (error)
		return ERR_PTR(error);

	if (flags == O_RDWR)
		error = may_open(&nd.path, MAY_READ|MAY_WRITE, flags);
	else
		error = may_open(&nd.path, MAY_WRITE, flags);

	if (!error)
		return dentry_open(nd.path.dentry, nd.path.mnt, flags,
				   current_cred());

	path_put(&nd.path);
	return ERR_PTR(error);
}

static struct {
	char *name; int wsize; int rsize;
} map[] = {
	[NFSCTL_SVC] = {
		.name	= ".svc",
		.wsize	= sizeof(struct nfsctl_svc)
	},
	[NFSCTL_ADDCLIENT] = {
		.name	= ".add",
		.wsize	= sizeof(struct nfsctl_client)
	},
	[NFSCTL_DELCLIENT] = {
		.name	= ".del",
		.wsize	= sizeof(struct nfsctl_client)
	},
	[NFSCTL_EXPORT] = {
		.name	= ".export",
		.wsize	= sizeof(struct nfsctl_export)
	},
	[NFSCTL_UNEXPORT] = {
		.name	= ".unexport",
		.wsize	= sizeof(struct nfsctl_export)
	},
	[NFSCTL_GETFD] = {
		.name	= ".getfd",
		.wsize	= sizeof(struct nfsctl_fdparm),
		.rsize	= NFS_FHSIZE
	},
	[NFSCTL_GETFS] = {
		.name	= ".getfs",
		.wsize	= sizeof(struct nfsctl_fsparm),
		.rsize	= sizeof(struct knfsd_fh)
	},
};

SYSCALL_DEFINE3(nfsservctl, int, cmd, struct nfsctl_arg __user *, arg,
		void __user *, res)
{
	struct file *file;
	void __user *p = &arg->u;
	int version;
	int err;

	if (copy_from_user(&version, &arg->ca_version, sizeof(int)))
		return -EFAULT;

	if (version != NFSCTL_VERSION)
		return -EINVAL;

	if (cmd < 0 || cmd >= ARRAY_SIZE(map) || !map[cmd].name)
		return -EINVAL;

	file = do_open(map[cmd].name, map[cmd].rsize ? O_RDWR : O_WRONLY);	
	if (IS_ERR(file))
		return PTR_ERR(file);
	err = file->f_op->write(file, p, map[cmd].wsize, &file->f_pos);
	if (err >= 0 && map[cmd].rsize)
		err = file->f_op->read(file, res, map[cmd].rsize, &file->f_pos);
	if (err >= 0)
		err = 0;
	fput(file);
	return err;
}
back to top