Ulrich Drepper <drepper <at> gmail.com> writes:
> I guess that this is really a difficult way to solve. I wouldn't want
> to pay for something which is hardly ever really used.
>
> But there are programs out there which would like to use the inode
> uniqueness. Therefore the next best thing to do is perhaps to return
> a flag in the getdents information (in d_type, perhaps) to indicate
> that this is a mount point and/or that there are multiple ways to
> access the file in question. Then programs which can use the inode
> information can be watching for this flag and enter the slow path only
> if it's set.
An alternative to a flag in d_type might be setting d_ino to a sentinel value
(there is plenty of existing code that refuses to use a readdir entry with d_ino
of 0, so it would have to be something else; but maybe -1 would work). But I
definitely like your idea of making it obvious to the application (whether by
d_ino or d_type) when it is necessary to use lstat to validate the inode number,
and promising that if the flag is not set then d_ino is correct.
--
Eric Blake
On Thu, 3 Sep 2009 14:50:52 +0000 (UTC)
Eric Blake <[email protected]> wrote:
> Ulrich Drepper <drepper <at> gmail.com> writes:
>
> > I guess that this is really a difficult way to solve. I wouldn't want
> > to pay for something which is hardly ever really used.
> >
> > But there are programs out there which would like to use the inode
> > uniqueness. Therefore the next best thing to do is perhaps to return
> > a flag in the getdents information (in d_type, perhaps) to indicate
> > that this is a mount point and/or that there are multiple ways to
> > access the file in question. Then programs which can use the inode
> > information can be watching for this flag and enter the slow path only
> > if it's set.
>
> An alternative to a flag in d_type might be setting d_ino to a sentinel value
> (there is plenty of existing code that refuses to use a readdir entry with d_ino
> of 0, so it would have to be something else; but maybe -1 would work). But I
> definitely like your idea of making it obvious to the application (whether by
> d_ino or d_type) when it is necessary to use lstat to validate the inode number,
> and promising that if the flag is not set then d_ino is correct.
>
The only problem with those schemes is that they assume that you're
using getdents. If you just lstat something then it's not clear.
It seems like a better scheme might be to do something that makes this
clear when using an lstat call too. Maybe we could coopt a bit in the
st_mode?
S_IFXDEV 020000
That's outside of the S_IFMT mask, so it shouldn't break S_IS* checks,
right?
--
Jeff Layton <[email protected]>