Re: seq_file and exporting dynamically allocated data

[viro@parcelfarce.linux.theplanet.co.uk]

On Sat, Nov 15, 2003 at 06:33:10PM +0100, Harald Welte wrote:
> On Sat, Nov 15, 2003 at 05:18:44PM +0000, viro@parcelfarce.linux.theplanet.co.uk wrote:
> > On Sat, Nov 15, 2003 at 10:38:33AM +0100, Harald Welte wrote:
> > > that doesn't help.  As I am aware, the seq_file structure is only
> > > allocated in the seq_open() call.  How does seq_open() know which
> > > private data (i.e. hash table) to associate with struct file?
> > 
> > Why should seq_open() know that?  Its caller does and it can set the damn
> > thing to whatever it wants.
> 
> So who is the caller? it's the ->open() member of struct
> file_operations.  and struct file_operations doesn't have some private
> member where I could hide my pointer before saving it to
> seq_file.private in seq_open().

If arguments of ->open() were not enough to find your data, how the hell would
current code manage to find it?

You've got inode; you've got (if that's on procfs) proc_dir_entry - from
inode; you've got dentry (from struct file *).  If that's not enough to
find your data, what is?

Which files do you have in mind?