RFC: Page size on PPC/PPC64 builders

[David Woodhouse]

On Tue, 2008-03-04 at 17:16 -0600, Josh Boyer wrote:
> On Tue, 04 Mar 2008 21:57:26 +0000
> David Woodhouse <dwmw2 at infradead.org> wrote:
> 
> > 
> > On Tue, 2008-03-04 at 12:47 -0600, Josh Boyer wrote:
> > > But I really think we should be eating our own dog food here.
> > 
> > Your employer and mine also care about people who use 64KiB pages... :)
> 
> I don't speak for my employer in any way shape or form.  My comments
> are my own and do not represent those of my employer.  <Insert random
> other ways of saying I speak for me alone and not my employer>.  Fedora
> is done in my spare time.

Yeah, sorry -- that was poorly phrased. I didn't mean that we should be
pursuing 64KiB pages because we are inappropriately beholden to our
employers -- it was more an observation that some poor bugger at some
point is going to have to deal with the issue anyway.

And since we're already _doing_ it (and mostly, we already _did_ it to
get FC6 out the door), and since it's helping us find real bugs on all
architectures, maybe we should just keep doing it. I don't recall a
single case of a program failing on a 4KiB-page kernel because it was
built on a 64KiB-page kernel, although I can certainly imagine how it
might happen.

Yes, there is the issue of (organised and documented) access to test
machines; we need to deal that with whatever the page size, and the test
machines need to match the builders.

> Besides, my employer is probably more concerned with the fact that
> Fedora 9 Alpha (and likely Beta) doesn't even install on their class of
> hardware because of Anaconda/HAL refusing to see the disks on Power6
> and LPARs (as well as PS3).

Christ, isn't that fixed yet? 

/me looks... er, did we not even assign the corresponding bug to HAL?
Or is there a separate HAL bug other than 431045?

-- 
dwmw2