Re: [PATCH 0/5] Allow more than PAGESIZE data read in configfs

[Chandra Seetharaman <sekharan@us.ibm.com>]

On Mon, 2006-10-16 at 16:09 -0700, Joel Becker wrote:
> On Mon, Oct 16, 2006 at 12:24:35PM -0700, Chandra Seetharaman wrote:
> > On Sat, 2006-10-14 at 13:10 -0700, Joel Becker wrote:
> > > 	Issues of PAGE_SIZE and attribute size aside, the patch posted
> > > was incorrect.  While they used seq_file, they implemented it in a
> > 
> > I do not think it is incorrect. It provides a simplest interface to
> > configfs's clients.
> 
> 	I understand that it provides a simple sed(1)-based change for
> existing clients.  However, it abuses seq_file in exactly the wrong way.
> 
> > > completely inefficent fashion, filling the entire buffer in one ->show()
> > > call rather than letting multiple read(2) calls iterate their list.
> > 
> > So, are you suggesting that we should provide a generic seq_operations
> > interface to configfs's clients ?
> 
> 	Yes, if we provide a vector type in any way, it should be via
> seq_operations or something like it.  If we're going to use seq_file, we
> should use it correctly.
> 

So, you want me to make a patch that uses seq_op instead of seq_file ? 
I am willing to do it.

> Joel
> 
-- 

----------------------------------------------------------------------
    Chandra Seetharaman               | Be careful what you choose....
              - sekharan@us.ibm.com   |      .......you may get it.
----------------------------------------------------------------------