RFC: Page size on PPC/PPC64 builders

[Tom Lane]

Jesse Keating <jkeating at redhat.com> writes:
> On Mon, 2008-03-03 at 13:54 -0500, Bill Nottingham wrote:
>> OK, then. Why would we explicitly make people build under a
>> combination that isn't
>> in Fedora (that they can't reproduce in Fedora) that can cause their
>> build to fail
>> (for whatever reason)? Do we appreciate being cruel?

> Because our builders can't always be running the kernel of the day.  For
> better or worse, our builders run RHEL5 as the host OS, and use mock for
> the chroots.  Except for very few occasions, the running kernel doesn't
> really play into the build.  The places where it does seem to be pretty
> fragile IMHO and need work anyway.

The bugs that are exposed by this problem are real bugs --- for example,
the PPC64 64KB page problem is just pushing the envelope a bit further
on problems that likely exist already on IA64 (16KB pages, not to
mention the dual-stack issue).  I don't believe anyone can honestly
think that it's okay for userland code to be dependent on 4KB hardware
page size.

The answer is not to try to avoid the bugs, but to provide the
infrastructure that developers will need to identify and fix the bugs.

			regards, tom lane