Date: Mon, 18 Feb 2002 23:09:59 -0800 (PST) From: Matthew Dillon <dillon@apollo.backplane.com> To: Jeroen Ruigrok/asmodai <asmodai@wxs.nl> Cc: Robert Watson <rwatson@FreeBSD.ORG>, Poul-Henning Kamp <phk@FreeBSD.ORG>, cvs-committers@FreeBSD.ORG, cvs-all@FreeBSD.ORG Subject: Re: cvs commit: src/sys/kern kern_time.c Message-ID: <200202190709.g1J79xM58924@apollo.backplane.com> References: <Pine.NEB.3.96L.1020218130420.69361t-100000@fledge.watson.org> <200202181825.g1IIPNe26719@apollo.backplane.com> <20020219065317.GO4470@daemon.ninth-circle.org>
next in thread | previous in thread | raw e-mail | index | archive | help
:[snip rest of eloquent text]
:
:Didn't you take out Alfred's locking for the VM system a while ago?
:Maybe you could kill time on that in the mean time?
:
:--
:Jeroen Ruigrok van der Werven / asmodai / Kita no Mono
It's not possible to test a proper MP solution for the VM system
until we can push Giant down into it from the top down and get a good
piece of the I/O path pulled up from the bottom. The best way to push
Giant down at this stage is from the system calls down and from the
networking and I/O stacks up.
This isn't a matter of perusing the source to find something to do.
just picking a random subsystem and spending time on it does not
result in testable solutions being committed to the tree. This is a
matter of working on familiar subsystems that Giant is already touching
directly and pushing it down further.
ucred, with Julian's commit, is one of the more easily partitionable
subsystems and it covers a good chunk of system calls including parts
of the I/O system, which we will need later. It is what I am focusing on
now and if people had stopped screwing around with it for months in P4
and started doing real commits using safe mechanisms like Giant
wrappers we would be mostly done by now.
From my point of view, 80% of the work that has been done in P4 in
this area has been a complete and utter waste of time, much of which
is totally unsafe for commit because it rips Giant out instead of
instrumenting it in a safe way. All the work that has been done
will have to be repeated as the bits get committed because not a single
sole outside of the P4 developers have been able to test it, not even
in piecemeal. It isn't the coding that takes time here, people, it is
making the stuff available to a wide audience in a manner that allows
easy testing without compromising other development projects.
That means, in a nutshell:
* piecemeal commits to -current
* Instrumented Giant
It's that simple.
-Matt
Matthew Dillon
<dillon@backplane.com>
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200202190709.g1J79xM58924>