Re: [PATCH 1/2][RESEND] Default iosched fixes (was: Fall back io scheduler for 2.6.15?)

[Nate Diller <nate.diller@gmail.com>]

On 1/21/06, Jens Axboe <axboe@suse.de> wrote:
> On Fri, Jan 20 2006, Nate Diller wrote:
> > My previous default iosched patch did a poor job dealing with the
> > 'elevator=' boot-time option.  The old behavior falls back to the
> > compiled-in default if the requested one is not registered at boot
> > time.  This patch dynamically evaluates which default
> > to use, and emits a suitable error message when the requested scheduler
> > is not available.  It also does the 'as' -> 'anticipatory' conversion
> > before elevator registration, which along with a modified registration
> > function, allows it to correctly indicate which default scheduler is
> > in use.
>
> I'm a little confused by your description - what problem does this patch
> actually solve? We already fall back to the default, and we already do
> the "as" conversion. It does seem to cleanup the code, just curious
> since your description seems to promise a little more than what it
> actually adds.

It makes the ' (default)' printk that happens at elevator registration
time behave (more) correctly.  My original patch rather ignored that
segment of code.  The current behavior is to only print ' (default)'
when one was specified at boot time, and not if 'as' was requested
either, since it doesn't understand the 'as -> anticipatory'
conversion.  Now, it will display correctly the one selected at
compile-time, if none was specified at boot, and when the boot-time
option was 'as'.

It also handles modular defaults better; although they cannot be
specified in kconfig, a default requested at boot-time will now still
work, even if it's a module.  When the boot-time requested scheduler
is not loaded, it will fall back to the compiled-in default; when it
is, it gets used.

NATE