[lkml]   [2002]   [Sep]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [BK-PATCH-2.5] Introduce new VFS inode cache lookup function
On Tue, Sep 03, 2002 at 04:31:32PM +0100, Anton Altaparmakov wrote:
> On Tue, 3 Sep 2002, Jan Harkes wrote:
> > On Tue, Sep 03, 2002 at 06:20:44AM +0100, Anton Altaparmakov wrote:
> > > 2) It will return inodes that are I_FREEING or I_CLEAR. I will have to
> > > test for these in NTFS and then iput() to wash my hands clean of such
> > > garbage. And if I am not mistaken, the iput() actually will BUG().
> >
> > If that is the case iget is broken. Perhaps it should test for these
> > states in find_inode (and find_inode_fast) and never return them. Are
> > those types of inodes still on the inode hash?
> That is of course the big question. As I said I haven't looked at this
> very closely...

<handwaving> There is no problem here </handwaving>

invalidate_list, prune_icache, generic_delete_inode, and
generic_forget_inode all move the inode off of the inode hash before
setting the I_FREEING flag. So inodes in I_FREEING state can not be
found by find_inode.

Clear inode bugs when the inode is not I_FREEING, it then clears out the
inode and sets I_CLEAR. As a result, inodes in I_CLEAR state can not be
found by find_inode either.

So you can safely remove that test in ilookup.

> > > 4) If anything, as Christoph Hellwig suggested to me on #kernel,
> > > iget{,5}_locked() should be reimplemented in terms of my ilookup()
> > > implementation and not vice versa. (-:
> >
> > Well, considering that this function (modulo the I_FREEING|I_CLEAR test
> > is identical to the first 10 lines in iget5_locked, this could call that
> > function. Ofcourse iget_locked is using the 'fast' version of find_inode.
> Yes, indeed, and yes. I recall that Al wanted the _fast() versions to
> remain separate. This can be done easily enough here, too. For example
> ilookup() and ilookup5() (to remain consistent with iget5() or if you
> have a better idea for a name...).

I have no problem with the name. And coda_fid_to_inode could use
ilookup5 as well, as currently uses iget with the 'set_fail' trick.


To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
More majordomo info at
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:28    [W:0.052 / U:2.780 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site