Return to BSD News archive
Path: sserve!newshost.anu.edu.au!munnari.oz.au!spool.mu.edu!howland.reston.ans.net!cs.utexas.edu!swrinde!sgiblab!cs.uoregon.edu!news.uoregon.edu!usenet.coe.montana.edu!bsd.coe.montana.edu!nate From: nate@bsd.coe.montana.edu (Nate Williams) Newsgroups: comp.os.386bsd.bugs Subject: Re: Why are no processes swapped in FreeBSD-1.0? Date: 12 Feb 1994 17:54:59 GMT Organization: Montana State University - Bozeman MT Lines: 67 Message-ID: <2jj55j$gqv@pdq.coe.montana.edu> References: <1994Feb9.003953.7932@sserve.cc.adfa.oz.au> NNTP-Posting-Host: bsd.coe.montana.edu In article <1994Feb9.003953.7932@sserve.cc.adfa.oz.au>, Warren Toomey <wkt@cserve.cs.adfa.oz.au> wrote: >Looking at the output of ps -aux under FreeBSD-1.0, it seems that NO processes >are ever swapped out to disk (i.e there are no IW processes marked). However, >under 386bsd-0.1+patchkits, this does happen. I've ensured that the system has >had enough time and inclination to swap out some processes (e.g processes >idle for >30 seconds, lots of memory active in other processes). From what I understand given the original VM system (I'm not expert, but one of our VM experts is not on Internet and the other one is too busy) that 386BSD did not swap like typical unix systems. Maybe 386BSD did not print out the correct statistics? (See notes below) >Can somebody tell me if this is a problem in FreeBSD-1.0, or if ps itself >is not properly reporting process status. If it's any consolation, FreeBSD 1.1 does indeed swap. .... nate 83 0.0 0.0 52 0 v0 IW+ 5:49PM 0:00.28 xinit root 76 0.0 0.0 24 0 v1 IWs+ 5:48PM 0:00.08 /usr/libexec/ge nate 75 0.0 0.0 412 0 v0 IWs 5:48PM 0:00.96 -tcsh (tcsh) Here's a bit out of some notes John posted to the mailing lists. ... As you might know, the version of the MACH VM system that is being used by most of the BSD variants is not really complete and has some problems. The inadequacies included: .. 3) Non-implementation of swapping. 4) Paging in/out of ONLY single pages. .. 9) The system will hang on swap space full. But FreeBSD now has.. 4) True swapping support, tied in with RSS limits and UPAGE paging. The UPAGES have been moved into process space, and they are paged when a process is swapped out. (UPAGES MUST be wired when a process is swapped in.) 5) The page daemons have been essentially redone, and vm_fault has had significant additions and improvements to support CLUSTER paging. .. 16) The system can relieve its problem of running out of swap space, eliminating an impending hang. --------------------------- As you can, I only posted a few of the problems and solutions that the new VM system attempts to solve since those are the problems spoken about above, but there are many more. >P.S Why doesn't iostat show disk activity :-) ??? Because no-one has (yet) integrated the counters into the disk drivers. (There's a project for someone to do. :-) Nate -- nate@bsd.coe.montana.edu | FreeBSD core member and all around tech nate@cs.montana.edu | weenie. work #: (406) 994-4836 | Graduating May '94 with a BS in EE home #: (406) 586-0579 | - looking for work in CS/EE field.