Another update that fixes Al's latest comments:
- the per-sb options are merged into the fs-specific options
- mount source is shown as a separate field (because commas are not
escaped within it)
- group id checking before allocation/freeing: this one doesn't
actually make a difference for the moment (I hope) but Al's variant
is cleaner
- simplify ID allocation and freeing
- dominator renaming
- function header comment updating
Updated the git tree:
git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs.git mountinfo
I suspect it doesn't work for -mm, because it's not based on -linus
directly, but on the mountinfo-base branch in the vfs-2.6.git tree.
Andrew, I'm happy that you applied the patches, but if it's the
slightest inconvenience to deal with them, just leave it. It'll get
there through the vfs-2.6 tree anyway.
Thanks,
Miklos
--
Miklos Szeredi <[email protected]> wrote:
> Another update that fixes Al's latest comments:
>
> - the per-sb options are merged into the fs-specific options
> - mount source is shown as a separate field (because commas are not
> escaped within it)
> - group id checking before allocation/freeing: this one doesn't
> actually make a difference for the moment (I hope) but Al's variant
> is cleaner
> - simplify ID allocation and freeing
> - dominator renaming
> - function header comment updating
Is it robust against mounting /dev/$'abc def\x0dghi'
to ~/mnt/`perl -e 'print "a" x 4050'`?
Besides that, I don't like having 'none' as the mount source. Instead, I
like to e.g. mount proc /proc -t proc, this will put the important
information into the first column of the mount(1) output, right where I
start reading.
> > Another update that fixes Al's latest comments:
> >
> > - the per-sb options are merged into the fs-specific options
> > - mount source is shown as a separate field (because commas are not
> > escaped within it)
> > - group id checking before allocation/freeing: this one doesn't
> > actually make a difference for the moment (I hope) but Al's variant
> > is cleaner
> > - simplify ID allocation and freeing
> > - dominator renaming
> > - function header comment updating
>
> Is it robust against mounting /dev/$'abc def\x0dghi'
> to ~/mnt/`perl -e 'print "a" x 4050'`?
It is.
> Besides that, I don't like having 'none' as the mount source. Instead, I
> like to e.g. mount proc /proc -t proc, this will put the important
> information into the first column of the mount(1) output, right where I
> start reading.
So do that, this patch doesn't force you to put 'none' in there.
Miklos