Revision 227823d2074da0c138d2abc0074b2dd281bbf923 authored by Dai Ngo on 23 January 2020, 01:45:39 UTC, committed by Anna Schumaker on 04 February 2020, 15:50:44 UTC
When the directory is large and it's being modified by one client
while another client is doing the 'ls -l' on the same directory then
the cache page invalidation from nfs_force_use_readdirplus causes
the reading client to keep restarting READDIRPLUS from cookie 0
which causes the 'ls -l' to take a very long time to complete,
possibly never completing.

Currently when nfs_force_use_readdirplus is called to switch from
READDIR to READDIRPLUS, it invalidates all the cached pages of the
directory. This cache page invalidation causes the next nfs_readdir
to re-read the directory content from cookie 0.

This patch is to optimise the cache invalidation in
nfs_force_use_readdirplus by only truncating the cached pages from
last page index accessed to the end the file. It also marks the
inode to delay invalidating all the cached page of the directory
until the next initial nfs_readdir of the next 'ls' instance.

Signed-off-by: Dai Ngo <dai.ngo@oracle.com>
Reviewed-by: Trond Myklebust <trond.myklebust@hammerspace.com>
[Anna - Fix conflicts with Trond's readdir patches]
[Anna - Remove redundant call to nfs_zap_mapping()]
[Anna - Replace d_inode(file_dentry(desc->file)) with file_inode(desc->file)]
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
1 parent 93a6ab7
Raw File
msync.c
// SPDX-License-Identifier: GPL-2.0
/*
 *	linux/mm/msync.c
 *
 * Copyright (C) 1994-1999  Linus Torvalds
 */

/*
 * The msync() system call.
 */
#include <linux/fs.h>
#include <linux/mm.h>
#include <linux/mman.h>
#include <linux/file.h>
#include <linux/syscalls.h>
#include <linux/sched.h>

/*
 * MS_SYNC syncs the entire file - including mappings.
 *
 * MS_ASYNC does not start I/O (it used to, up to 2.5.67).
 * Nor does it marks the relevant pages dirty (it used to up to 2.6.17).
 * Now it doesn't do anything, since dirty pages are properly tracked.
 *
 * The application may now run fsync() to
 * write out the dirty pages and wait on the writeout and check the result.
 * Or the application may run fadvise(FADV_DONTNEED) against the fd to start
 * async writeout immediately.
 * So by _not_ starting I/O in MS_ASYNC we provide complete flexibility to
 * applications.
 */
SYSCALL_DEFINE3(msync, unsigned long, start, size_t, len, int, flags)
{
	unsigned long end;
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma;
	int unmapped_error = 0;
	int error = -EINVAL;

	start = untagged_addr(start);

	if (flags & ~(MS_ASYNC | MS_INVALIDATE | MS_SYNC))
		goto out;
	if (offset_in_page(start))
		goto out;
	if ((flags & MS_ASYNC) && (flags & MS_SYNC))
		goto out;
	error = -ENOMEM;
	len = (len + ~PAGE_MASK) & PAGE_MASK;
	end = start + len;
	if (end < start)
		goto out;
	error = 0;
	if (end == start)
		goto out;
	/*
	 * If the interval [start,end) covers some unmapped address ranges,
	 * just ignore them, but return -ENOMEM at the end.
	 */
	down_read(&mm->mmap_sem);
	vma = find_vma(mm, start);
	for (;;) {
		struct file *file;
		loff_t fstart, fend;

		/* Still start < end. */
		error = -ENOMEM;
		if (!vma)
			goto out_unlock;
		/* Here start < vma->vm_end. */
		if (start < vma->vm_start) {
			start = vma->vm_start;
			if (start >= end)
				goto out_unlock;
			unmapped_error = -ENOMEM;
		}
		/* Here vma->vm_start <= start < vma->vm_end. */
		if ((flags & MS_INVALIDATE) &&
				(vma->vm_flags & VM_LOCKED)) {
			error = -EBUSY;
			goto out_unlock;
		}
		file = vma->vm_file;
		fstart = (start - vma->vm_start) +
			 ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
		fend = fstart + (min(end, vma->vm_end) - start) - 1;
		start = vma->vm_end;
		if ((flags & MS_SYNC) && file &&
				(vma->vm_flags & VM_SHARED)) {
			get_file(file);
			up_read(&mm->mmap_sem);
			error = vfs_fsync_range(file, fstart, fend, 1);
			fput(file);
			if (error || start >= end)
				goto out;
			down_read(&mm->mmap_sem);
			vma = find_vma(mm, start);
		} else {
			if (start >= end) {
				error = 0;
				goto out_unlock;
			}
			vma = vma->vm_next;
		}
	}
out_unlock:
	up_read(&mm->mmap_sem);
out:
	return error ? : unmapped_error;
}
back to top