[Unionfs] RT patch - again

[Erez Zadok]

In message <200711270926.23750.paul at gccs.co.za>, Paul Hewlett writes:

> Erez
> 
> What would solve this is if the CONFIG variables in the kernel are always 
> defined but to either 0 (disabled) or to 1 (enabled) and alas that is not 
> going to happen.

Yes: we have three mutually-exclusive cases to distinguish:

1. RT patches not applied
2. RT patches applied and CONFIG_PREEMPT_RT=y
2. RT patches applied and CONFIG_PREEMPT_RT=n

You can't distinguish among three cases with only one yes/no bit.  You need
two bits at least.

The easiest solution, honestly, is if the RT maintainers rename that
.__nrpages field back to .nrpages -- then a simple ifdef will work for me.
Do you think renaming the field back is something they'd agree to?

BTW, I understand why they did renamed it: to catch every person who uses
that field w/o the wrapper.  Makes sense while developing code, as I use
that trick too.  But after that, it's ok to rename the field back to its
original name.

If there's an RT ML you can point me to, I'll be happy to post this request
to them.

Thanks,
Erez.