Re: Portmasters alert: was => Re: uvm physseg - put the lid on it.

[Cherry G. Mathew <cherry%zyx.in@localhost>]

>>>>> "Taylor" == Taylor R Campbell <campbell+netbsd-tech-kern%mumble.net@localhost> writes:

    Taylor> Hi, Cherry!
    Taylor> Currently there are a number of builds broken for reasons related to
    Taylor> pmap/uvm changes:

    Taylor> https://releng.netbsd.org/builds/HEAD/201612240730Z/

Hey Taylor,

Sorry about that - acorn26, sh3 and evbppc related builds should now be
fixed.

    Taylor> You added a new kernel option UVM_HOTPLUG, but I assume you must have
    Taylor> made API changes that go beyond the #ifdef UVM_HOTPLUG parts.

    Taylor> Can you write up a summary of the pmap and uvm API changes you made,
    Taylor> from the perspective of a portmaster, or a portmaster's understudy?
    Taylor> What were a pmap's relevant responsibilities before your changes, and
    Taylor> what are they after?

Have you had a chance to look at the man page ?
http://cvsweb.netbsd.org/bsdweb.cgi/src/share/man/man9/uvm_hotplug.9?only_with_tag=MAIN

I'll write a little blog entry on the NetBSD blog in due course of time,
but brief summary for portmasters is:

The vm_physmem array has been replaced, and its equivalent is now
accessed and modified via the uvm_hotplug(9) API.

This is irrespective of options UVM_HOTPLUG

    Taylor> In addition to fixing the build, it would be helpful to be able to
    Taylor> review changes to fix the build more than cursorily via the autobuild
    Taylor> status.

I'm not sure I understand this comment. Are you saying you need more
information about why the builds are failing/what the fixes are ?

-- 
~cherry