On Fri, 25 Feb 2011 13:21:50 +0300
Pavel Shilovsky <piastry@xxxxxxxxxxx> wrote:
> Use invalidate_inode_pages2 that don't leave pages even if shrink_page_list()
> has a temp ref on them. It prevents a data coherency problem on exclusive
> oplocks opens.
>
> Signed-off-by: Pavel Shilovsky <piastry@xxxxxxxxxxx>
> ---
> fs/cifs/inode.c | 16 +++++++++++-----
> 1 files changed, 11 insertions(+), 5 deletions(-)
>
> diff --git a/fs/cifs/inode.c b/fs/cifs/inode.c
> index 589f3e3..0011e95 100644
> --- a/fs/cifs/inode.c
> +++ b/fs/cifs/inode.c
> @@ -1687,12 +1687,18 @@ cifs_invalidate_mapping(struct inode *inode)
>
> cifs_i->invalid_mapping = false;
>
> - /* write back any cached data */
> - if (inode->i_mapping && inode->i_mapping->nrpages != 0) {
> - rc = filemap_write_and_wait(inode->i_mapping);
> - mapping_set_error(inode->i_mapping, rc);
> + if (inode->i_mapping) {
> + /* write back any cached data */
> + if (inode->i_mapping->nrpages != 0) {
> + rc = filemap_write_and_wait(inode->i_mapping);
> + mapping_set_error(inode->i_mapping, rc);
> + }
> + rc = invalidate_inode_pages2(inode->i_mapping);
> + if (rc)
> + cERROR(1, "%s: could not invalidate inode %p", __func__,
> + inode);
> }
> - invalidate_remote_inode(inode);
> +
> cifs_fscache_reset_inode_cookie(inode);
> }
>
I think using invalidate_inode_pages2 is the right thing to do. I'm not
so keen however on simply popping a printk when that fails. The user is
going to see that and say "huh?"
I think we need to consider allowing EBUSY bubble up to userspace
appropriately. Otherwise we still risk data coherency problems, right?
Perhaps cifs_invalidate_mapping should be changed to an int return and
the callers could return errors from it?
--
Jeff Layton <jlayton@xxxxxxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe linux-cifs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
