On Mon, 2007-02-12 at 21:11 -0800, Andrew Morton wrote:
> On Fri, 09 Feb 2007 14:53:37 -0800 Dave Hansen <[EMAIL PROTECTED]> wrote:
>
> > diff -puN
> > fs/file_table.c~14-24-tricky-elevate-write-count-files-are-open-ed
> > fs/file_table.c
> > --- lxc/fs/file_table.c~14-24-tricky-elevate-write-count-files-are-open-ed
> > 2007-02-09 14:26:54.000000000 -0800
> > +++ lxc-dave/fs/file_table.c 2007-02-09 14:26:54.000000000 -0800
> > @@ -209,8 +209,11 @@ void fastcall __fput(struct file *file)
> > if (unlikely(S_ISCHR(inode->i_mode) && inode->i_cdev != NULL))
> > cdev_put(inode->i_cdev);
> > fops_put(file->f_op);
> > - if (file->f_mode & FMODE_WRITE)
> > + if (file->f_mode & FMODE_WRITE) {
> > put_write_access(inode);
> > + if(!special_file(inode->i_mode))
> > + mnt_drop_write(mnt);
> > + }
> > put_pid(file->f_owner.pid);
> > put_user_ns(file->f_owner.user_ns);
> > file_kill(file);
> > diff -puN fs/namei.c~14-24-tricky-elevate-write-count-files-are-open-ed
> > fs/namei.c
> > --- lxc/fs/namei.c~14-24-tricky-elevate-write-count-files-are-open-ed
> > 2007-02-09 14:26:54.000000000 -0800
> > +++ lxc-dave/fs/namei.c 2007-02-09 14:26:54.000000000 -0800
> > @@ -1548,8 +1548,17 @@ int may_open(struct nameidata *nd, int a
> > return -EACCES;
> >
> > flag &= ~O_TRUNC;
> > - } else if (IS_RDONLY(inode) && (flag & FMODE_WRITE))
> > - return -EROFS;
> > + } else if (flag & FMODE_WRITE) {
> > + /*
> > + * effectively: !special_file()
> > + * balanced by __fput()
> > + */
> > + error = mnt_want_write(nd->mnt);
> > + if (error)
> > + return error;
> > + if (IS_RDONLY(inode))
> > + return -EROFS;
> > + }
>
> 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.
> Are you sure that fget_light() and fput_light() don't accidentally bypass this
> new logic?
Pretty sure. My code actually surrounds all of the permission() checks
in the VFS. To even use fget, you had to get a fd at some point, and to
do that you have to go through open, where both the mount and normal
filesystem checks are.
Is there something particular you had in mind?
-- Dave
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
