Date: Sat, 12 Jun 1999 21:55:45 -0700 (PDT) From: Matthew Dillon <dillon@apollo.backplane.com> To: The Hermit Hacker <scrappy@hub.org> Cc: freebsd-current@FreeBSD.ORG, freebsd-stable@FreeBSD.ORG Subject: Re: MMAP() in STABLE/CURRENT ... Message-ID: <199906130455.VAA65569@apollo.backplane.com> References: <Pine.BSF.4.05.9906091422000.49155-100000@thelab.hub.org>
next in thread | previous in thread | raw e-mail | index | archive | help
:Over the past week and a bit, the INN -CURRENT source tree had a major
:upheaval of code, in order to fix the major problem where reader speed
:sucked. What used to take >1min to load up a newsgruop now generally
:takes mere seconds (no exaggeration, try it)...
:
:The problem is that the new code makes *heavy* use of MMAP() ...
:
:When the code was put in place, I immediately upgraded my server to give
:the code a good test (am one of the developers over there) in a production
:environment, and ever since then, my system locks up around 24hrs uptime,
:requiring a cold boot.
:
:I cannot get into DDB, since I'm over a serial console, so debugging this
:has proved to be next to impossible :(
I'll bet that it is another mmap/write deadlock.
:I run CURRENT at home and STABLE in production, but the production server
:is the only one that I believe has enough load on her to be able to
:trigger it...
:
:Is this a problem that is fixed in CURRENT, but hasn't migrated down into
:STABLE? Is there any way of adding some sort of debug code to STABLE such
:that if the race condition is trigger, it will at least pop up a message
:onto the console letting me know its happened so that I know its MMAP()
:related?
:
:Thanks...
:Marc G. Fournier ICQ#7615664 IRC Nick: Scrappy
If it is what I think it is, it is *NOT* yet fixed in -current.
-Matt
Matthew Dillon
<dillon@backplane.com>
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199906130455.VAA65569>