Re: Fwd: struct inode size reduction.

[Andries Brouwer <aebr@win.tue.nl>]

On Mon, Mar 10, 2003 at 05:25:49PM +0100, Roman Zippel wrote:

> On Mon, 10 Mar 2003, Andries Brouwer wrote:
> 
> > > > -       error = register_chrdev(driver->major, driver->name, &tty_fops);
> > > > +       error = register_chrdev_region(driver->major, driver->minor_start,
> > > > +                                      driver->num, driver->name, &tty_fops);
> > > 
> > > Are that much parameters really needed?
> > 
> > Yes.
> 
> Why? Problems are hardly solved by adding more parameters.
> If going to a larger number space means, that we have to add crappy 
> interfaces, we should rather keep it as it is.
> Why do you need to partition the number space like this? I looked at the 
> users in the last mail for a reason. If we're going to change the 
> interface, it should reflect what we will need in the future.

Maybe I should not react, but let me answer once more.
You do not understand the part about "small steps".

You see a future and ask why I don't jump to the future you
see. The answer is that I take small steps.

Look at the current junk in char_dev.c and cry:
        if (ret && isa_tty_dev(major)) {
                lock_kernel();
                if (need_serial(major,minor)) {
                        /* Force request_module anyway, but what for? */
                        fops_put(ret);
                        ret = NULL;
                }
                unlock_kernel();
        }
Then be happy that it is gone.

You want a different interface? My changes make it easier for you
to get there. Please go ahead.

Andries