Date: Wed, 19 Sep 2001 11:17:10 -0700 (PDT) From: Matt Dillon <dillon@earth.backplane.com> To: Ian Dowse <iedowse@maths.tcd.ie> Cc: freebsd-stable@freebsd.org, Mike Silbersack <silby@silby.com>, Kirk McKusick <mckusick@mckusick.com>, Grigoriy Orlov <gluk@ptci.ru> Subject: Re: dirpref MFC to -stable: patch set to test Message-ID: <200109191817.f8JIHAa56883@earth.backplane.com> References: <200109191059.aa43270@salmon.maths.tcd.ie>
next in thread | previous in thread | raw e-mail | index | archive | help
:In message <200109190106.f8J16jv50999@earth.backplane.com>, Matt Dillon writes:
:> Ok folks, here is a patchset that MFC's the dirpref code from -current
:> to -stable. I have done some very simple testing, but I need a couple
:> more eyes on the code. I want to make sure that I have MFC'd all
:> necessary rcs pieces.
:
:Looks good! All the relevant revisions appear to be included. There
:should be no problems with switching back and forth between different
:kernels or userlands; the kernel chooses defaults for the new
:superblock parameters if necessary, and the old utilities (fsck,
:newfs) ignore the new fields. I'm running with this patch now, so
:I'll let you know if any problems crop up.
:
:One thing I did notice was that the tunefs code has diverged
:significantly from -current. Specifically, revisions 1.13 and 1.14
:were not MFC'd. I haven't examined at these changes closely to see
:if they are applicable to -stable, but it would be nice if they
:could be MFC'd now to avoid further divergence.
:
:Ian
Yah, I noticed that when I tried to patch it in originally. It's
been rewritten to use getopt(). Normally I'd applaud the change
but the new code is even more messy then the old!
In anycase, I'm not worrying about it. My patch just adds the two
new options and doesn't try to do anything fancy.
-
Looks good, I'm still on schedule to commit it on Friday.
-
I noticed something odd while testing it, though. When doing a
'find /usr/ports -name fubar' on a /usr/ports created w/ dirpref,
I got about 4x the performance as a non-dirpref /usr/ports, but iostat
was showing 600-800 tps with a KB/t of only 1K. In otherwords, the
data was packed more optimally on the disk but FreeBSD was only reading
it in 1K I/O's. I wonder if there is something we can do to improve
that.
-Matt
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200109191817.f8JIHAa56883>