From: "Andrew McLaren" Newsgroups: alt.folklore.computers Subject: Correct usage of "Image" ??? Lines: 26 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.00.2919.6700 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6700 Message-ID: Date: Thu, 10 Feb 2000 20:55:12 +1100 NNTP-Posting-Host: 139.134.110.180 X-Trace: newsfeeds.bigpond.com 950176167 139.134.110.180 (Thu, 10 Feb 2000 20:49:27 EST) NNTP-Posting-Date: Thu, 10 Feb 2000 20:49:27 EST Organization: Telstra BigPond Internet Services (http://www.bigpond.com) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.datacomm.ch!newsmaster-01.vbs.at!newsfeed03.univie.ac.at!newsrouter.chello.at!newsfeed01.sul.t-online.de!t-online.de!newsfeed.tli.de!news.maxwell.syr.edu!intgwpad.nntp.telstra.net!newsfeeds.bigpond.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49586 Brothers, Over the years I've heard the term "image" used to refer to both the "image" of a program on disk, or a "image" of the executable file in memory. For example "the process loader reads the executable image into memory" ... ie, it reads a file. Or, "there was only a single image executing" ... ie, there was only one instance of the program sitting in memory. Can anyone shed light on the origin, etymology, and correct usage of the term "image". BTW refering to an "image" is a wonderful way to intimidate non-technical people with a bit of vague obfuscation; but I just *know* it must have had a useful meaning once. Possibly from the early days, before relocatable code became commonplace, so that a program file really *was* an "image" of the array of opcodes, sitting in memory? So, what is an image really? On disk, or in memory?? Cheers Andrew ###### From: Peter N. M. Hansteen Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: 10 Feb 2000 11:33:40 +0100 Organization: Datadokumentasjon A/S Lines: 17 Message-ID: References: NNTP-Posting-Host: gw.datadok.no Original-Sender: peter@datadok.no X-Newsreader: Gnus v5.7/Emacs 20.5 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!news.algonet.se!algonet!uninett.no!torget.bgnett.no!delilah.datadok.no!nobody Xref: chonsp.franklin.ch alt.folklore.computers:49557 "Andrew McLaren" writes: > So, what is an image really? On disk, or in memory?? Both. Unless the code is modified in some way[1], the in-memory image would be the same as on the disk. The binary image would in any case be immediately executable, without intermediary steps (interpretation). [1] OK, variables will get altered, but then you keep your code separate from your data, don't you? -- SPECIAL OFFER! I proofread unsolicited commercial email sent to this address at a rate of US $500.00 per incident! Include billing address in your message and save US $500.00 per hour off ordinary address resolution and tracking charge! ###### From: "Mike Duffy" Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: Thu, 10 Feb 2000 10:03:20 -0500 Lines: 39 Message-ID: <87ujuo$qji$1@bob.news.rcn.net> References: X-Trace: o0sCwUcaWqU+5XGeMSpxfgfTeEmhqetizuMr05ahzr0= X-Complaints-To: abuse@rcn.com NNTP-Posting-Date: 10 Feb 2000 15:02:48 GMT X-Newsreader: Microsoft Outlook Express 4.72.3110.5 X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.datacomm.ch!newscore.gigabell.net!news.gigabell.net!newsfeed.nacamar.de!newsfeed.icl.net!netnews.com!feed1.news.rcn.net!rcn!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49563 Andrew McLaren wrote in message ... >Brothers, > >Over the years I've heard the term "image" used to refer to both the "image" >of a program on disk, or a "image" of the executable file in memory. > > >For example "the process loader reads the executable image into memory" ... >ie, it reads a file. Or, "there was only a single image executing" ... ie, >there was only one instance of the program sitting in memory. > Some systems do not have a "program loader" per se. For example, VMS maps a virtual address range and assigns the image file as the backing store for page faults. In effect, the image file on disk (at least the read-only portions of it) become an instant "page file", ready to resolve page faults. Pages of the image come into memory in just this way - as the result of page faults. (once resident, read-write portions must be faulted out to process-specific paging locations). My assumption has always been that the image file on disk is an exact "image" of what it looks like in memory. So I usually refer to the file as an image, and the program in memory as a "program", even though some parts of it might not exist in memory at a given time. Of course, that's just an opinion rather than historical evidence, based mainly on the behavior of the system I use most eften. -Mike Duffy Process Software Corp. ###### From: michael.wojcik@merant.com (Michael Wojcik) Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: 10 Feb 2000 15:19:43 GMT Organization: MERANT Inc. Lines: 51 Message-ID: <87ukuf01lho@news2.newsguy.com> References: Reply-To: michael.wojcik@merant.com NNTP-Posting-Host: p-948.newsdawg.com X-Newsreader: xrn 9.00 Originator: mww@lorelei-n Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!newshunter!cosy.sbg.ac.at!news-feed.inet.tele.dk!bofh.vszbr.cz!news.maxwell.syr.edu!logbridge.uoregon.edu!pln-w!spln!extra.newsguy.com!newsp.newsguy.com!mww Xref: chonsp.franklin.ch alt.folklore.computers:49574 In article , "Andrew McLaren" writes: > Brothers, Some sisters may wish to comment as well (BAH?). > Over the years I've heard the term "image" used to refer to both the "image" > of a program on disk, or a "image" of the executable file in memory. > > For example "the process loader reads the executable image into memory" ... > ie, it reads a file. Or, "there was only a single image executing" ... ie, > there was only one instance of the program sitting in memory. > > Can anyone shed light on the origin, etymology, and correct usage of the > term "image". I don't know the etymological specifics, but using "image" to refer to a program's machine-readable form isn't an usual use of the term. "Correct usage" is a chimera; natural language use is determined by the speech community, particularly for a language like English which has many dialects, regionalisms, and jargons and lacks a recognized standardizing body. An "image" is simply a static, non-discursive representation of something - a photograph, painting, sculpture, etc. Images in the ordinary artistic sense of the term are intended for human consumption, but it's easy to see how the idea could be applied to representations of programs in machine-readable form. So "image" on disk is as reasonable as "image" in memory. Both are representations of the program (which could also be considered as a sequence of instructions, or a systematic application of techniques, or a filter for data, etc.). Treating the program as an image emphasizes its nature in that representation as recognizable to the computer (its "audience", so to speak) and static. The program in memory may not actually be static - if we consider its dynamic data as part of the image, or account for cases like self-modifying code - but when we speak of a program image in memory we're treating it as a static entity for purpose of discussion. -- Michael Wojcik michael.wojcik@merant.com AAI Development, MERANT (block capitals are a company mandate) Department of English, Miami University Pocket #9: A complete "artificial glen" with rocks, and artificial moon, and forester's station. Excellent for achieving the effect of the sublime without going out-of-doors. -- Joe Green ###### From: Howard S Shubs Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: Thu, 10 Feb 2000 12:15:38 -0600 Organization: MindSpring Enterprises Lines: 18 Message-ID: References: <87ujuo$qji$1@bob.news.rcn.net> NNTP-Posting-Host: c7.ae.37.77 X-Server-Date: 10 Feb 2000 18:15:43 GMT User-Agent: MT-NewsWatcher/3.0 (PPC) X-Face: "S"r{U%bs].&Ud}Pc~~~0a]M:t5l>>EN\1Faw10M9NK1Xq59wo7-"s0S+[{etQorO /Nf-Ci"i9v'MT!R8)J]N[4|2&x1r^Iq&{SB"6dknr0=+6UFb.>+{zMn_1=rw&/V+"d@* ZS5\LoW_ Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!newsfeed00.sul.t-online.de!newsfeed01.sul.t-online.de!t-online.de!newsfeed.tli.de!news.maxwell.syr.edu!news.mindspring.net!newsfeed.mindspring.net!firehose.mindspring.com!hshubs Xref: chonsp.franklin.ch alt.folklore.computers:49588 In article <87ujuo$qji$1@bob.news.rcn.net>, "Mike Duffy" wrote: >Some systems do not have a "program loader" per se. For example, >VMS maps a virtual address range and assigns the image file as the >backing store for page faults. In effect, the image file on disk (at least >the >read-only portions of it) become an instant "page file", ready to resolve >page faults. Pages of the image come into memory in just this way - as >the result of page faults. (once resident, read-write portions must be >faulted >out to process-specific paging locations). This is called "image activation" on VMS. -- Howard S Shubs hshubs@mindspring.com hshubs@bix.com The Denim Adept Which is better, Maryann or pickled Ginger? SPAM: uce@ftc.gov postmaster@[127.0.0.1] abuse@[127.0.0.1] ###### From: "Lee Courtney" Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: Thu, 10 Feb 2000 10:31:23 -0800 Organization: Posted via Supernews, http://www.supernews.com Lines: 21 Message-ID: References: X-Complaints-To: newsabuse@supernews.com X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.00.2314.1300 X-Mimeole: Produced By Microsoft MimeOLE V5.00.2314.1300 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!newsfeed.stanford.edu!remarQ70!remarQ.com!supernews.com!rQdQ!corp.supernews.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49581 Andrew McLaren wrote in message news:HQvo4.10458$VJ1.19399@newsfeeds.bigpond.com... > Brothers, > > Over the years I've heard the term "image" used to refer to both the "image" > of a program on disk, or a "image" of the executable file in memory. Of course there's also IMAGE, aka TurboIMAGE, the network DBMS bundled with the HP3000 business server. Originally released in mid-70s and still in use today at 1000's of sites. Lee Courtney Monterey Software Group Inc. 1350 Pear Avenue, Suite J Mountain View, California 94043-1302 (650) 964-7052 voice, (650) 964-6735 fax, (408) 237-1705 beeper www.editcorp.com/Businesses/MontereySoftware From: "Lee Courtney" Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: Thu, 10 Feb 2000 10:31:23 -0800 Organization: Posted via Supernews, http://www.supernews.com Lines: 21 Message-ID: References: X-Complaints-To: newsabuse@supernews.com X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.00.2314.1300 X-Mimeole: Produced By Microsoft MimeOLE V5.00.2314.1300 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!newsfeed.stanford.edu!remarQ70!remarQ.com!supernews.com!rQdQ!corp.supernews.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49581 Andrew McLaren wrote in message news:HQvo4.10458$VJ1.19399@newsfeeds.bigpond.com... > Brothers, > > Over the years I've heard the term "image" used to refer to both the "image" > of a program on disk, or a "image" of the executable file in memory. Of course there's also IMAGE, aka TurboIMAGE, the network DBMS bundled with the HP3000 business server. Originally released in mid-70s and still in use today at 1000's of sites. Lee Courtney Monterey Software Group Inc. 1350 Pear Avenue, Suite J Mountain View, California 94043-1302 (650) 964-7052 voice, (650) 964-6735 fax, (408) 237-1705 beeper www.editcorp.com/Businesses/MontereySoftware ###### From: viro@weyl.math.psu.edu (Alexander Viro) Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: 10 Feb 2000 13:50:40 -0500 Organization: -ENOENT Lines: 20 Message-ID: <87v1a0$uj@weyl.math.psu.edu> References: <87ujuo$qji$1@bob.news.rcn.net> NNTP-Posting-Host: weyl.math.psu.edu Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!newshunter!cosy.sbg.ac.at!newsfeed.Austria.EU.net!nmaster.kpnqwest.net!npeer.kpnqwest.net!EU.net!news.maxwell.syr.edu!news.cis.ohio-state.edu!news.ems.psu.edu!news3.cac.psu.edu!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49594 In article <87ujuo$qji$1@bob.news.rcn.net>, Mike Duffy wrote: >Some systems do not have a "program loader" per se. For example, >VMS maps a virtual address range and assigns the image file as the >backing store for page faults. In effect, the image file on disk (at least >the >read-only portions of it) become an instant "page file", ready to resolve >page faults. Pages of the image come into memory in just this way - as >the result of page faults. (once resident, read-write portions must be >faulted >out to process-specific paging locations). ...ahem. s/Some systems/Almost everything on boxen with paging MMU/. Demand-paging is not VMS-specific (I would be really amazed if VMS was the first who did it - I'ld bet on MULTICS or TENEX, depending on the timing. Or on the early BBN stuff). -- "You're one of those condescending Unix computer users!" "Here's a nickel, kid. Get yourself a better computer" - Dilbert. ###### From: Alexandre Pechtchanski Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Organization: Rockefeller University Hospital (GCRC), New York Message-ID: References: X-Newsreader: Forte Agent 1.7/32.534 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 30 Date: Thu, 10 Feb 2000 14:00:01 -0500 NNTP-Posting-Host: 129.85.24.56 X-Trace: rockyd.rockefeller.edu 950209241 129.85.24.56 (Thu, 10 Feb 2000 14:00:41 EST) NNTP-Posting-Date: Thu, 10 Feb 2000 14:00:41 EST Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!newspeer.te.net!news.indigo.ie!news-out.cwix.com!newsfeed.cwix.com!news.maxwell.syr.edu!newsfeed.nyu.edu!rockyd.rockefeller.edu!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49598 On Thu, 10 Feb 2000 20:55:12 +1100, "Andrew McLaren" wrote: >Brothers, > >Over the years I've heard the term "image" used to refer to both the "image" >of a program on disk, or a "image" of the executable file in memory. > >For example "the process loader reads the executable image into memory" ... >ie, it reads a file. Or, "there was only a single image executing" ... ie, >there was only one instance of the program sitting in memory. > >Can anyone shed light on the origin, etymology, and correct usage of the >term "image". > >BTW refering to an "image" is a wonderful way to intimidate non-technical >people with a bit of vague obfuscation; but I just *know* it must have had a >useful meaning once. Possibly from the early days, before relocatable code >became commonplace, so that a program file really *was* an "image" of the >array of opcodes, sitting in memory? > >So, what is an image really? On disk, or in memory?? Was not there a time long ago [in Galaxy far away... sorry, wrong group ;-)] when the final stage of the program development was an executable code in the core _the image of which_ then got dumped to external media (paper tape?) ? This would explain usage of "image" quite nicely. [ When replying, remove *'s from address ] Alexandre Pechtchanski, Systems Manager, RUH, NY ###### From: brucehoult@pobox.com (Bruce Hoult) Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: Fri, 11 Feb 2000 15:00:01 +1300 Organization: The Internet Group Ltd Lines: 19 Message-ID: References: <87ujuo$qji$1@bob.news.rcn.net> NNTP-Posting-Host: macinnat.static.star.net.nz X-Newsreader: MT-NewsWatcher 2.4.4 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news-fra1.dfn.de!unlisys!news.snafu.de!diablo.theplanet.net!newsfeed.icl.net!newsfeed.icl.net!ihug.co.nz!brucehoult Xref: chonsp.franklin.ch alt.folklore.computers:49639 In article <87ujuo$qji$1@bob.news.rcn.net>, "Mike Duffy" wrote: > Some systems do not have a "program loader" per se. For example, > VMS maps a virtual address range and assigns the image file as the > backing store for page faults. In effect, the image file on disk (at least > the > read-only portions of it) become an instant "page file", ready to resolve > page faults. Pages of the image come into memory in just this way - as > the result of page faults. (once resident, read-write portions must be > faulted > out to process-specific paging locations). That's true of lots of systems these days -- even MacOS (for PowerPC programs). Win32 does it. Linux does it. It would be more intersting to know if there is any current OS that *doesn't*. -- Bruce ###### From: "Mike Duffy" Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: Fri, 11 Feb 2000 09:22:50 -0500 Lines: 42 Message-ID: <8815uo$rde$1@bob.news.rcn.net> References: <87ujuo$qji$1@bob.news.rcn.net> <87v1a0$uj@weyl.math.psu.edu> X-Trace: 6Voh1FmfkK/eLCP4AB24h8RJ9jIA6ieG/6JG4O/tbas= X-Complaints-To: abuse@rcn.com NNTP-Posting-Date: 11 Feb 2000 14:22:16 GMT X-Newsreader: Microsoft Outlook Express 4.72.3110.5 X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49616 Alexander Viro wrote in message <87v1a0$uj@weyl.math.psu.edu>... >In article <87ujuo$qji$1@bob.news.rcn.net>, >Mike Duffy wrote: >>Some systems do not have a "program loader" per se. For example, >>VMS maps a virtual address range and assigns the image file as the >>backing store for page faults. In effect, the image file on disk (at least >>the >>read-only portions of it) become an instant "page file", ready to resolve >>page faults. Pages of the image come into memory in just this way - as >>the result of page faults. (once resident, read-write portions must be >>faulted >>out to process-specific paging locations). > >...ahem. s/Some systems/Almost everything on boxen with paging MMU/. >Demand-paging is not VMS-specific (I would be really amazed if VMS was >the first who did it - I'ld bet on MULTICS or TENEX, depending on the >timing. Or on the early BBN stuff). > Certainly. I didn't mean to imply that this was VMS-specific, nor that there were "few" systems which acted this way. Just an example based on what I most often use. One of the design decisions for VMS (circa 1976?) was "what kind" of page replacement to use (whether to fault against all processes, or against yourself, etc). They studied several systems already using various forms, so no, they certainly were not first. I was limiting my comment to how an image gets into memory, and I suspect there may be (whether modern or historic) a few virtual-memory, paging systems around which copy the entire image into virtual memory before beginning execution, whether using common or separate backing store. Perhaps someone could post an example if they know of one? -Mike ###### Sender: lynn@LYNNLT Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? References: <87ujuo$qji$1@bob.news.rcn.net> <87v1a0$uj@weyl.math.psu.edu> <8815uo$rde$1@bob.news.rcn.net> Reply-To: Anne & Lynn Wheeler MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii From: Anne & Lynn Wheeler Message-ID: Organization: Wheeler&Wheeler Lines: 22 User-Agent: Gnus/5.07008 (Pterodactyl Gnus v0.80) Emacs/20.3 Date: Fri, 11 Feb 2000 14:42:08 GMT NNTP-Posting-Host: 209.63.29.109 X-Complaints-To: support@adcomsys.net X-Trace: news-west.eli.net 950280128 209.63.29.109 (Fri, 11 Feb 2000 07:42:08 MST) NNTP-Posting-Date: Fri, 11 Feb 2000 07:42:08 MST Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!hermes.visi.com!news-out.visi.com!uunet!ffx.uu.net!news-west.eli.net!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49642 misc. on page replacement ... circa 1968 ... of course there was atlas on others before. http://www.garlic.com/~lynn/93.html#0 http://www.garlic.com/~lynn/93.html#4 http://www.garlic.com/~lynn/93.html#5 http://www.garlic.com/~lynn/94.html#1 http://www.garlic.com/~lynn/96.html#0a http://www.garlic.com/~lynn/98.html#54 http://www.garlic.com/~lynn/99.html#18 this was on cp/67 which was port from cp/40 circa 65 or 66. some of information on virtual memory is history paper at: http://pucc.princeton.edu/~melinda/ -- -- Anne & Lynn Wheeler | lynn@garlic.com, finger for pgp key http://www.garlic.com/~lynn/ ###### From: jmfbah@aol.com (JMFBAH) Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Lines: 46 NNTP-Posting-Host: ladder06.news.aol.com X-Admin: news@aol.com Date: 11 Feb 2000 15:12:41 GMT References: Organization: AOL http://www.aol.com Message-ID: <20000211101241.23575.00002605@ng-fu1.aol.com> Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.cwix.com!cyclone-east.rr.com!news.rr.com!news-east.rr.com!portc05.blue.aol.com!audrey05.news.aol.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49631 From: Howard S Shubs Date: Thu, Feb 10, 2000 13:15 EST Message-id: >>In article <87ujuo$qji$1@bob.news.rcn.net>, "Mike Duffy" >> >>wrote: >>Some systems do not have a "program loader" per se. For example, >>VMS maps a virtual address range and assigns the image file as the >>backing store for page faults. In effect, the image file on disk (at least >>the >>read-only portions of it) become an instant "page file", ready to resolve >>page faults. Pages of the image come into memory in just this way - as >>the result of page faults. (once resident, read-write >> portions must be faulted >>out to process-specific paging locations). >This is called "image activation" on VMS. Barf. In the olden days, there used to be a LOAD command. This took the relocatables that had been generated by the COMPILE command, did some magic, and created an image that was in core. This image could then be SAVEd to disk as an executable. This executable could then be STARTed (if the image was still pure in core) or RUN if the image had to be retrieved from a storage media such as disk, DECtape, magtape, etc. To confuse things even further there were various file attributes that stored the way the file was written; or, in other words, the format of the data in the file. There were a few flavors of image, ASCII, and I can't remember any more. Damn....my monitor just sparked. Now note that this nomenclature is purely DEC-defined. Other manufacturers had other standards. /BAH ###### From: jmfbah@aol.com (JMFBAH) Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Lines: 59 NNTP-Posting-Host: ladder06.news.aol.com X-Admin: news@aol.com Date: 11 Feb 2000 15:22:12 GMT References: <8815uo$rde$1@bob.news.rcn.net> Organization: AOL http://www.aol.com Message-ID: <20000211102212.23575.00002607@ng-fu1.aol.com> Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.cwix.com!cyclone-east.rr.com!news.rr.com!news-east.rr.com!portc05.blue.aol.com!audrey05.news.aol.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49632 From: "Mike Duffy" Date: Fri, Feb 11, 2000 09:22 EST Message-id: <8815uo$rde$1@bob.news.rcn.net> >Alexander Viro wrote in message <87v1a0$uj@weyl.math.psu.edu>... >>In article <87ujuo$qji$1@bob.news.rcn.net>, >>Mike Duffy wrote: >>>Some systems do not have a "program loader" per se. For example, >>>VMS maps a virtual address range and assigns the image file as the >>>backing store for page faults. In effect, the image file on disk (at >>>least >>>the >>>read-only portions of it) become an instant "page file", ready to resolve >>>page faults. Pages of the image come into memory in just this way - as >>>the result of page faults. (once resident, read-write portions must be >>>faulted >>>out to process-specific paging locations). >> >>...ahem. s/Some systems/Almost everything on boxen with paging MMU/. >>Demand-paging is not VMS-specific (I would be really amazed if VMS was >>the first who did it - I'ld bet on MULTICS or TENEX, depending on the >>timing. Or on the early BBN stuff). >> >I was limiting my comment to how an image gets into memory, >and I suspect there may be (whether modern or historic) a few >virtual-memory, paging systems around which copy the entire image >into virtual memory before beginning execution, whether using >common or separate backing store. First of all, it depends what the hardware architecture is. Second of all, it depends on how the program is segmented. On TOPS10 a trick that was used so that very large programs could be executed (before extended addressing was implemented) was to allow multiple segments. Now, the key to this was to organize one's code so that a rountine in segment one isn't called by the Nth segment because it won't be there. Another trick to save memory space was sharable high segments. That allowed the monitor to have only one copy of the segment in core even though more than one user was using it. Note that, in this case, a sharable _writable_ high segment was frowned upon [very wry emoticon here]. >Perhaps someone could post an example if they know of one? It's not clear if you're talking hardware or software. You'ld almost have to also know the evolution of a OS and its hardware to get any idea of "what was done". /BAH ###### From: jmfbah@aol.com (JMFBAH) Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Lines: 18 NNTP-Posting-Host: ladder06.news.aol.com X-Admin: news@aol.com Date: 11 Feb 2000 15:26:48 GMT References: <87ukuf01lho@news2.newsguy.com> Organization: AOL http://www.aol.com Message-ID: <20000211102648.23575.00002609@ng-fu1.aol.com> Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.cwix.com!cyclone-east.rr.com!news.rr.com!news-east.rr.com!portc05.blue.aol.com!audrey05.news.aol.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49635 From: michael.wojcik@merant.com (Michael Wojcik) Date: Thu, Feb 10, 2000 10:19 EST Message-id: <87ukuf01lho@news2.newsguy.com> >In article , "Andrew >McLaren" writes: >> Brothers, >Some sisters may wish to comment as well (BAH?). I tried, even though I had to do all my prefixing by hand. I'm not sure that what I wrote is what the guy was asking for. DEC has their own standards w.r.t. definitions of words (this was pre-VMS, of course ). /BAH ###### From: Howard S Shubs Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: Fri, 11 Feb 2000 10:32:31 -0600 Organization: Denim Software Lines: 16 Message-ID: References: <20000211101241.23575.00002605@ng-fu1.aol.com> NNTP-Posting-Host: c7.ae.35.5c X-Server-Date: 11 Feb 2000 16:32:32 GMT User-Agent: MT-NewsWatcher/3.0 (PPC) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!howland.erols.net!newsfeed.mindspring.net.MISMATCH!news.mindspring.net!firehose.mindspring.com!hshubs Xref: chonsp.franklin.ch alt.folklore.computers:49654 In article <20000211101241.23575.00002605@ng-fu1.aol.com>, jmfbah@aol.com (JMFBAH) wrote: >In the olden days, there used to be a LOAD command. This >took the relocatables that had been generated by the COMPILE >command, did some magic, and created an image that was in >core. This image could then be SAVEd to disk as an >executable. This executable could then be STARTed (if the >image was still pure in core) or RUN if the image had to >be retrieved from a storage media such as disk, DECtape, >magtape, etc. That must have been previous to VMS v3. I've never seen those. -- Howard S Shubs, the Denim Adept ###### From: Tim Shoppa Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: Fri, 11 Feb 2000 12:54:40 -0400 Organization: Trailing Edge Technology Lines: 20 Message-ID: <38A40690.50CF13CE@trailing-edge.com> References: <20000211101241.23575.00002605@ng-fu1.aol.com> NNTP-Posting-Host: timaxp.trailing-edge.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Trace: ffx2nh5.news.uu.net 950291681 16983 63.73.218.130 (11 Feb 2000 17:54:41 GMT) X-Complaints-To: news@ffx2nh5.news.uu.net NNTP-Posting-Date: 11 Feb 2000 17:54:41 GMT X-Mailer: Mozilla 3.03Gold (X11; I; OpenVMS V7.0 DEC 3000 Model 300L) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newspeer.monmouth.com!uunet!ffx.uu.net!ffx2nh5!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49624 Howard S Shubs wrote: > > In article <20000211101241.23575.00002605@ng-fu1.aol.com>, > jmfbah@aol.com (JMFBAH) wrote: > > >In the olden days, there used to be a LOAD command. This > >took the relocatables that had been generated by the COMPILE > >command, did some magic, and created an image that was in > >core. This image could then be SAVEd to disk as an > >executable. This executable could then be STARTed (if the > >image was still pure in core) or RUN if the image had to > >be retrieved from a storage media such as disk, DECtape, > >magtape, etc. > > That must have been previous to VMS v3. I've never seen those. I believe BAH is talking about the "monitor" type OS's, i.e. TOPS-10, RT-11, OS/8, etc. Tim. ###### Message-ID: <38A45D26.808610D6@earthlink.net> From: jchausler X-Mailer: Mozilla 4.04 [en] (Win95; I) MIME-Version: 1.0 Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? References: <20000211101241.23575.00002605@ng-fu1.aol.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 35 NNTP-Posting-Host: 63.15.101.119 X-Complaints-To: abuse@earthlink.net X-Trace: newsread2.prod.itd.earthlink.net 950296394 63.15.101.119 (Fri, 11 Feb 2000 11:13:14 PST) NNTP-Posting-Date: Fri, 11 Feb 2000 11:13:14 PST Organization: EarthLink Network, Inc. X-ELN-Date: Fri Feb 11 11:13:14 2000 Date: Fri, 11 Feb 2000 19:13:14 GMT Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news-fra1.dfn.de!fu-berlin.de!newsfeed.berkeley.edu!newsfeed1.earthlink.net!nntp.earthlink.net!newsread2.prod.itd.earthlink.net.POSTED!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49606 JMFBAH wrote: > In the olden days, there used to be a LOAD command. This > took the relocatables that had been generated by the COMPILE > command, did some magic, and created an image that was in > core. This image could then be SAVEd to disk as an > executable. This executable could then be STARTed (if the > image was still pure in core) or RUN if the image had to > be retrieved from a storage media such as disk, DECtape, > magtape, etc. > > Now note that this nomenclature is purely DEC-defined. > Other manufacturers had other standards. > > /BAH This was certainly similar to Data General. The compilers/assemblerwould normally generate "relocatable binaries" (The filename extension was .rb) The relocatable loader would then either load them into memory to execute under one of DG's OS's (RDOS in this case) or make an "absolute binary" (.ab) file which, assuming it was a "stand-alone" program (we usually programmed to the "bare metal" in those days) could be loaded directly into memory and executed. These .ab's were typically called "executable images" and when punched to paper tape could be directly read in and executed using the machines program load facility or a manually entered binary loader. This was true whether the image was a small "toy" program or a complete multiprocessing system. Regards, Chris AN GETTO$;DUMP;RUN,ALGOL,TAPE $$ ###### From: "Mike Yankus" Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: Fri, 11 Feb 2000 13:56:57 -0600 Organization: Sterling Software, Inc. Lines: 27 Message-ID: <881phv$f5b$1@reuters.plano.sterling.com> References: NNTP-Posting-Host: pc-504.itd.sterling.com X-Trace: reuters.plano.sterling.com 950299007 15531 192.245.39.124 (11 Feb 2000 19:56:47 GMT) X-Complaints-To: usenet@reuters.plano.sterling.com NNTP-Posting-Date: 11 Feb 2000 19:56:47 GMT X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.00.2615.200 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2615.200 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!hermes.visi.com!news-out.visi.com!uunet!ffx.uu.net!hype.plano.sterling.com!news.plano.sterling.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49651 I can give you one example of an "image" from RSX11D. A sysgen built an "core image" file on disk - and I do mean image! Kernel, device drivers, file system, and any other program that you wanted to be running at immediately after booting were contained therein. This includes non-privleged applications. This file was up to 124K (16 bit) words in size (enough to cover 18 bit addresses). It was stored contiguously on the disk and read in all in one fell swoop. The bootstrap code reloaded the registers (which were saved on the kernel stack and called functions in each device driver which reloaded their volatile registers. A knowledgable programmer could patch the "core image" file to effect changes upon next reboot. The file represented the image of physical memory (minus the external page, i.e. the memory mapped hardware registers). BTW this is a PDP11 system - 16 bit addressing, 18 bit physical memory in the earlier models, and 22 bit later. You could not have anything running in the upper part of memory at bootstrap. It ran in 18 bit mode until it determined that it was booted on a 22 bit system. Mike Yankus ###### Newsgroups: alt.folklore.computers From: ehrice@his.com (Edward Rice) Subject: Re: Correct usage of "Image" ??? Message-ID: Organization: NDS Date: Fri, 11 Feb 2000 21:58:55 -0500 References: <20000211101241.23575.00002605@ng-fu1.aol.com> NNTP-Posting-Host: pm8-229.his.com X-Trace: 11 Feb 2000 21:58:15 -0500, pm8-229.his.com Lines: 113 X-Authenticated-User: ehrice Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!howland.erols.net!newsfeed.skycache.com!news1.his.com!user Xref: chonsp.franklin.ch alt.folklore.computers:49647 In article <20000211101241.23575.00002605@ng-fu1.aol.com>, jmfbah@aol.com (JMFBAH) wrote: > Barf. Thanks, Barbara, I knew someone would give me a conversational opening here. This mini-primer is for people who don't know GCOS, and/or don't know old "job-stream"-oriented systems. It's pretty typical of how IBM and other makers did the compilations and the loading, although I happened to find my GCOS control cards pamphlet first so this is all GCOS examples. It's a little more complex than CDC handled, but looks very much the same in the simpler cases. I believe it's virtually identical to UNIVAC, except that the syntax and all the names are different. These ALL go back to IBSYS, the first true IBM operating system, and oddly enough the GCOS implementation is the closest to IBM's IBSYS that I know of, much closer to it than IBM JCL. Okay. Grab your hat and take a deep breath... System input -- the control card stream, sometimes including source decks with the appropriate compiler invocation cards, and sometimes including some object decks as well, and often including some data cards, came in a file called "I*". GCOS used a standard set of 2-character identifiers for the system files, all using an asterisk as one character -- any other valid file name was permitted to the user. The system read I* and split it out (IIRC) into a formatted version of the same file, called *I. *I was invisible, utterly, and I'm not sure I remember it correctly. I'll ignore it. Each compiler was invoked with a file called S*, the source input to that compiler. If the user's input decks (or files, if doing remote job entry, or tape input) called for a Fortran compile and a GMAP assembly run, then Fortran would be handed one S* file of its source, and when it was finished GMAP would get a different S* with just its source on it. Assuming you asked for this to be done (didn't suppress it, as it was the default), the binary object deck images were fed to a common file known as C*. Punched output was sent to P*, as was any printed output (listings), and these would be de-spooled later when the job finished. Once compilations were done, a $ EXECUTE card announced that it was time to do linking functions and in most cases, run the job. The loader knew about reading card decks of binary data in from C*, although you could fiddle how that worked with loader control cards. The basic goal of the linker (known in GCOS circles as the loader) was to collect C* binaries, acquire necessary library routines, and build another file called B*, which was a larger binary file also deck-oriented, which could be loaded into memory and executed. Along the way, libraries had to be harvested, and this is where L* and *L, the primary and secondary system libraries, came into play. You could override at least *L (and possible L*, I forget) by opening files with those filecodes, thereby using your own version of a library rather than the default. This was not N-level, it was 2-level, so some sites had to play games with moving some code to L* so that people who used their own *L would still get most of the system routines made available. The card decks that you sent out to C* earlier would also go to O* (I think, or maybe this happened during loader operations), so the loader read O* while pulling this all together. Control cards to the loader, which the system set aside earlier on, were passed to the loader on file R*. Okay. All of this led up to building what was basically a non-privileged-mode core image, interspersed with the necessary I/O control words to make it all load into the right place. This was placed on **, which was a slave-bootable file. GCOS made slave programs look much like system programs, with their own transfer vector and communications region up front -- the ** file, once the loader had set it up, included the control words to set or skip those areas as appropriate, and to get all the executable into memory. Except, if you were doing system overlaying of code, all the overlays lived on your ** file. As your job ran, when it was time to load in overlay it would read a word or two from the ** file, use that as I/O control orders, and the appropriate overlay would be read in to the right place. The ** file was as close to a true core image file as GCOS had, and as I've mentioned it had I/O controls and other small overhead in there. This was a great way to run a production program really FAST, and it was useful to be able to capture this file. For this purpose, the H* file could be allocated, and if it was (in user-filespace, not in system-filespace where the ** was located), then the loader would copy ** to H* just before diving into the ** file for execution. With a NOGO directive on the EXECUTE card, you could build an H* file without actually executing a program. Of course. (I can't find the NOGO or NGO option in my BJ69 rev 2 -- either it's undocumented or I've trivially confused where it goes.) This represents roughly a third of the standard filecodes in the system: others were assigned to, for example, compressed source deck input and output, intermediate COBOL compiler files, Integrated Data Store temporaries, system editors, console typewriters, etc. And while it sounds complex, most people didn't know about all the details. If you set yourself up with... $ SNUMB (a 5-character system number (job identifier) goes here) $ IDENT (accounting information goes here) $ FORTRAN (your Fortran source deck) $ EXECUTE $ ENDJOB then all that stuff would happen automagically, and your Fortran job would compile, appropriate libraries would be searched, your printout would end up on paper, and standard installation limits would be applied to your compilation and to your execution. Way cool, high speed, exceptionally easy to work with. Cf. IBM's JCL, which was less powerful and much, much uglier to deal with. ###### From: "Andrew McLaren" Newsgroups: alt.folklore.computers References: <87ukuf01lho@news2.newsguy.com> <20000211102648.23575.00002609@ng-fu1.aol.com> Subject: Re: Correct usage of "Image" ??? Lines: 25 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.00.2919.6700 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6700 Message-ID: Date: Sat, 12 Feb 2000 15:52:56 +1100 NNTP-Posting-Host: 139.134.107.13 X-Trace: newsfeeds.bigpond.com 950330826 139.134.107.13 (Sat, 12 Feb 2000 15:47:06 EST) NNTP-Posting-Date: Sat, 12 Feb 2000 15:47:06 EST Organization: Telstra BigPond Internet Services (http://www.bigpond.com) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newscore.univie.ac.at!uninett.no!newsfeed.nacamar.de!europa.netcrusader.net!205.231.236.10!newspeer.monmouth.com!intgwpad.nntp.telstra.net!newsfeeds.bigpond.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49643 > From: michael.wojcik@merant.com (Michael Wojcik) > > >In article , "Andrew >McLaren" > writes: > >> Brothers, > > >Some sisters may wish to comment as well (BAH?). > Absolutely! Brothers *and* sisters. I was conscious of the somewhat sexist overtones in my use of "Brothers" and yet "Siblings" sounds affected, and "Brothers and sisters" sounds more like a revival meeting, than a collegiate salutation. My sincere apologies if anyone was offended or felt excluded. Barbara Stephenson, for example, was the first woman engineer to work at Digital (started c. 1960?) Cheers, Andrew ###### From: jmfbah@aol.com (JMFBAH) Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Lines: 36 NNTP-Posting-Host: ladder06.news.aol.com X-Admin: news@aol.com Date: 12 Feb 2000 14:47:47 GMT References: Organization: AOL http://www.aol.com Message-ID: <20000212094747.26290.00003057@ng-cn1.aol.com> Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!remarQ-easT!remarQ.com!supernews.com!portc05.blue.aol.com!audrey05.news.aol.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49673 From: "Andrew McLaren" Date: Fri, Feb 11, 2000 23:52 EST Message-id: >> From: michael.wojcik@merant.com (Michael Wojcik) >> >> >In article , "Andrew >>McLaren" <> writes: >>> >> Brothers, >> >> >Some sisters may wish to comment as well (BAH?). >> >Absolutely! Brothers *and* sisters. I was conscious of the somewhat sexist >overtones in my use of "Brothers" and yet "Siblings" sounds affected, and >"Brothers and sisters" sounds more like a revival meeting, than a collegiate >salutation. My sincere apologies if anyone was offended or felt excluded. Piffle. I didn't even notice the failure to cover all bases when I read your post. I didn't feel excluded nor was I offended. :-) If I spent my time noticing such things, I would have never got any work done. >Barbara Stephenson, for example, was the first woman engineer to work at >Digital (started c. 1960?) I don't know of her and don't remember hearing her name. Could she be the woman in the PDP-10 project team picture? The only one I know by sight is Alan Kotok. I don't think I know the rest of them. Anybody remember Mary Paine? /BAH ###### From: jmfbah@aol.com (JMFBAH) Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Lines: 32 NNTP-Posting-Host: ladder06.news.aol.com X-Admin: news@aol.com Date: 12 Feb 2000 14:52:17 GMT References: Organization: AOL http://www.aol.com Message-ID: <20000212095217.26290.00003059@ng-cn1.aol.com> Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.cwix.com!cyclone-east.rr.com!news.rr.com!news-east.rr.com!portc05.blue.aol.com!audrey05.news.aol.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49672 From: Howard S Shubs Date: Fri, Feb 11, 2000 11:32 EST Message-id: >In article <20000211101241.23575.00002605@ng-fu1.aol.com>, >jmfbah@aol.com (JMFBAH) wrote: >>In the olden days, there used to be a LOAD command. This >>took the relocatables that had been generated by the COMPILE >>command, did some magic, and created an image that was in >>core. This image could then be SAVEd to disk as an >>executable. This executable could then be STARTed (if the >>image was still pure in core) or RUN if the image had to >>be retrieved from a storage media such as disk, DECtape, >>magtape, etc. >That must have been previous to VMS v3. I've never seen those. Shhh...[emoticon looking around to see if anybody is listening, then whispers...] VMS didn't follow convention if there was a hint that the source was TOPS10. VMS was brain damaged (I considered it senile) as of its first ship. And for those of you who like VMS and think it's a great operating system, you should extrapolate the sentiment and image what a TOPS10 or TOPS20 operating system was like. There, now, have I showed by bias :-)) /BAH ###### From: jmfbah@aol.com (JMFBAH) Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Lines: 18 NNTP-Posting-Host: ladder06.news.aol.com X-Admin: news@aol.com Date: 12 Feb 2000 14:55:16 GMT References: Organization: AOL http://www.aol.com Message-ID: <20000212095516.26290.00003060@ng-cn1.aol.com> Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!europa.netcrusader.net!24.30.200.2!cyclone-east.rr.com!news.rr.com!news-east.rr.com!portc05.blue.aol.com!audrey05.news.aol.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49671 From: ehrice@his.com (Edward Rice) Date: Fri, Feb 11, 2000 21:58 EST Message-id: >In article <20000211101241.23575.00002605@ng-fu1.aol.com>, >jmfbah@aol.com (JMFBAH) wrote: > > Barf. >Thanks, Barbara, I knew someone would give me a conversational opening >here. Happy to oblige. /BAH ###### From: jmfbah@aol.com (JMFBAH) Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Lines: 34 NNTP-Posting-Host: ladder06.news.aol.com X-Admin: news@aol.com Date: 12 Feb 2000 15:01:16 GMT References: <38A45D26.808610D6@earthlink.net> Organization: AOL http://www.aol.com Message-ID: <20000212100116.26290.00003062@ng-cn1.aol.com> Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.cwix.com!cyclone-east.rr.com!news.rr.com!news-east.rr.com!portc05.blue.aol.com!audrey05.news.aol.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49669 From: jchausler Date: Fri, Feb 11, 2000 14:13 EST Message-id: <38A45D26.808610D6@earthlink.net> >JMFBAH wrote: > >> Now note that this nomenclature is purely DEC-defined. >> Other manufacturers had other standards. >This was certainly similar to Data General. The compilers/assemblerwould >normally generate "relocatable binaries" (The filename extension >was .rb) I will use the word foo to indicate a filename. Our default was foo.REL. > The relocatable loader would then either load them into memory >to execute under one of DG's OS's (RDOS in this case) or make an >"absolute binary" (.ab) file which, assuming it was a "stand-alone" program >(we usually programmed to the "bare metal" in those days) could be loaded >directly into memory and executed. These .ab's were typically called >"executable images" and when punched to paper tape could be directly >read in and executed using the machines program load facility or a manually >entered binary loader. This was true whether the image was a small "toy" >program or a complete multiprocessing system. Our extension default name for these was foo.EXE. In olden days there did exist foo.SAV, foo.SHR, foo.HGH. (Then we got sane :-) and spec'ed out foo.EXE.) /BAH ###### Message-ID: <38A756F5.C16E5FED@silcom.com> From: Lars Poulsen X-Mailer: Mozilla 4.61 [en] (Win98; I) X-Accept-Language: da,en MIME-Version: 1.0 Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? References: <20000211101241.23575.00002605@ng-fu1.aol.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 85 Date: Sun, 13 Feb 2000 17:14:29 -0800 NNTP-Posting-Host: 207.71.222.73 X-Complaints-To: abuse@avtel.com X-Trace: newsfeed.avtel.net 950490339 207.71.222.73 (Sun, 13 Feb 2000 17:05:39 PST) NNTP-Posting-Date: Sun, 13 Feb 2000 17:05:39 PST Organization: None X-Received-Date: Sun, 13 Feb 2000 17:05:39 PST (newsfeed.avtel.net) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!newsfeed.cwix.com!newsfeed.avtel.net!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49732 Edward Rice wrote: > This mini-primer is for people who don't know GCOS, and/or don't > know old "job-stream"-oriented systems. ... > ... Way cool, high speed, exceptionally easy to work with. > Cf. IBM's JCL, which was less powerful and much, much uglier to > deal with. I have worked with many systems in the early 1970s; IBSYS/IBJOB was the first large system I got involved with (on a tape-oriented IBM 7094-II at NEUCC in Copenhagen, Denmark). Shortly after I got access to it, they also acquired an IBM 360/65 with OS/MVT and HASP-II. Your description of GCOS is in many ways similar to my recollection of IBSYS/IBJOB. I must, however, post an exception to your slander of OS/360 JCL. JCL was attempting to solve something that no other system had ever tried to do: To take all the messy details of how to optimize I/O on an extremely large variety of hardware, and externalize it from the program and its libraries into a group of system components called "access methods". The heart of JCL was the "DD" (data definition) statement, where the DCB (device control block) had fields available to override corresponding fields in a macro in the program. Historically, the problem showed up in programs using magnetic tape storage. If your program uses 80-byte records, you might write each record on the tape as a separate block (with a CRC at the end and a block gap between it and the preceding and the next block) but that would be wasteful, since a block gap was about half an inch, and 80 bytes would be one tenth of an inch, so 80% of the tape would be taken up by block gaps, which would both take up most of the reel (indeed, your data might not fit on a single reel that way). So more likely, you would put 10 of these records into an 800 byte data block on the tape. Now you would lose only one thrid of the tape to block gaps. You could even put 50 records in each block, which would reduce the overhead to less than 10%. But now you had another problem: If the tape got damaged, you would not lose a single record, but 50 of them at a time. And you would need to set aside 8KB of memory space for a pair of I/O buffers, which was somewhat extravagant if you only had a 64 KB program partition to begin with. On S/360, this kind of fine-grained tuning was available to files on disk devices as well. Every model of disk drive had different track sizes, and under JCL control you specified how you wanted the tracks allocated to your files to be formatted: //MYFILE DD DISP=(NEW,CATLG),DCB=(BLKSIZE=800,LRECL=80) I am aware of no other system that allowed the user this much control. Everyone else just formatted the whole drive in fixed length sectors. And I have heard that even S/90 has now caved in to that. For naive users running simple jobs, it did not matter much what the job setup was: Everyone got a set of basic control cards and copied them, while inserting their own program deck and data deck in the designated locations. IBSYS: OS/MVT: $JOB NAME //NAME JOB (...),"USER NAME" $IBSYS IBJOB // EXEC FORTGCLG $IBJOB FORTRAN //FORT.SYSIN DD * $FORTRAN DIMENSION ... DIMENSION ... ... ... END END //GO.SYSIN DD * $EXECUTE DATA DATA DATA DATA /* $EOJ // The problem started when users wanted to save the program on permanent storage. Then you had to teach them all the things that were glossed over by the JCL macro language in order to create the illusion of simplicity. For some reason, this "learning hump" was much bigger for OS-JCL than for IBSYS; I think this was because they were making the transition from tape systems to disk systems at the same time. For a truly cumbersome job control language, study ICL 1900 GEORGE from about the same era. -- / Lars Poulsen - http://www.cmc.com/lars - lars@cmc.com 125 South Ontare Road, Santa Barbara, CA 93105 - +1-805-569-5277 ###### Message-ID: <38A7B3A9.249D3D7B@home.com> From: Pete Lamasney X-Mailer: Mozilla 4.61 [en] (WinNT; U) X-Accept-Language: en MIME-Version: 1.0 Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? References: <20000211101241.23575.00002605@ng-fu1.aol.com> <38A756F5.C16E5FED@silcom.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 16 Date: Mon, 14 Feb 2000 07:50:05 GMT NNTP-Posting-Host: 24.0.172.250 X-Complaints-To: abuse@home.net X-Trace: news1.rdc1.sdca.home.com 950514605 24.0.172.250 (Sun, 13 Feb 2000 23:50:05 PST) NNTP-Posting-Date: Sun, 13 Feb 2000 23:50:05 PST Organization: @Home Network Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.grnet.gr!news-feed1.eu.concert.net!newsfeed.cwix.com!newsswitch.lcs.mit.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newshub2.home.com!news.home.com!news1.rdc1.sdca.home.com.POSTED!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:49857 Lars Poulsen wrote: > [...snip...] > components called "access methods". The heart of JCL was the "DD" > (data definition) statement, where the DCB (device control block) > had fields available to override corresponding fields in a macro > in the program. Actually, you could not override something that was coded in the macro in the program, but you could specify a value for something that was omitted. Typical practice was to omit the size of the block, thereby allowing the decision to be postponed until such time as the device characteristics were known, and allowing for new/different devices in the future. Pete ###### Newsgroups: alt.folklore.computers From: ehrice@his.com (Edward Rice) Subject: Re: Correct usage of "Image" ??? Message-ID: Organization: NDS Date: Fri, 18 Feb 2000 01:38:46 -0500 References: <20000211101241.23575.00002605@ng-fu1.aol.com> <38A756F5.C16E5FED@silcom.com> NNTP-Posting-Host: pm9-192.his.com X-Trace: 18 Feb 2000 01:37:48 -0500, pm9-192.his.com Lines: 92 X-Authenticated-User: ehrice Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!uunet!ams.uu.net!newsfeed2.news.nl.uu.net!sun4nl!newsfeed.icl.net!netnews.com!europa.netcrusader.net!208.184.7.66!newsfeed.skycache.com!news1.his.com!user Xref: chonsp.franklin.ch alt.folklore.computers:50082 In article <38A756F5.C16E5FED@silcom.com>, Lars Poulsen wrote: > I have worked with many systems in the early 1970s; IBSYS/IBJOB was > the first large system I got involved with... GECOS/GCOS was a pretty clear descendant of IBSYS. We had tee-shirts made up with "$IB" on the front, since that was the way all IBSYS control cards began ($IBJOB to start a job, $IBFTC to invoke the Fortran compiler, etc.), and I still have mine. Any similarity has a strong reason -- GE was IBM's largest customer for 7094's, with several in each of many operational areas. Later, when GE built the 600-line and Honeywell brought out the 6000's, GE very reasonably upgraded from the 7094 to the 600 to the 6000. IBM provided them only the poor option of emulating the 7094 on a special box that could be hooked up to a 360/50 or 360/65. > Your description of GCOS is in many ways similar to my recollection > of IBSYS/IBJOB. I must, however, post an exception to your slander > of OS/360 JCL. Slander? Slander OS/360 JCL? Not possible. > JCL was attempting to solve something that no other system had > ever tried to do: To take all the messy details of how to optimize > I/O on an extremely large variety of hardware, and externalize it > from the program and its libraries into a group of system > components called "access methods". The heart of JCL was the "DD" > (data definition) statement, where the DCB (device control block) > had fields available to override corresponding fields in a macro > in the program. Yes, and this was good. The solution to the problem already existed in IBSYS, and was better. IBM made it GHASTLY bad. Making users know the physical characteristics of each different device they planned to use, so that they'd know the size and effect of allocating 2C or 12T on a 2301 or a 3350 ... THAT was simply backwards. It was productive to offload that from the application arena to the system arena, but it was the reverse of productive to load it back onto the user's shoulders. > On S/360, this kind of fine-grained tuning was available to files > on disk devices as well. Every model of disk drive had different > track sizes, and under JCL control you specified how you wanted > the tracks allocated to your files to be formatted: > //MYFILE DD DISP=(NEW,CATLG),DCB=(BLKSIZE=800,LRECL=80) The complexity of distinguishing in fine detail between disks and tapes, to pick a large example, was just silly. The operating system could have -- and should have -- taken on the burden of doing things efficiently, once the user expressed the information requirement of the space. Other systems did this, and much less painfully than IBM tried. I actually did all that stuff, knew my way through the five-foot-shelf of daily-use manuals, but I'm not at all sure I could correctly distinguish between Kept and Cataloged and some other species of file retention. The very idea of files that stayed around but were uncataloged, and could be accessed by any user that knew their name, was an abomination. > I am aware of no other system that allowed the user this much > control. Everyone else just formatted the whole drive in fixed > length sectors. And I have heard that even S/90 has now caved in > to that. Cheap memory. It's now much cheaper to format optimally for the hardware and then buffer in such a way as to use that space productively. When memory was expensive, you had to integrate your storage requirements to the hardware device, to avoid using two full-cylinder buffers per file. > IBSYS: OS/MVT: We used slightly different cards than you listed, on the IBSYS side, but that's a quibble. > For a truly cumbersome job control language, study ICL 1900 > GEORGE from about the same era. Bad things can almost always be made worse. IBM itself had a better control language than OS JCL: the control language from IBSYS. CDC had a pretty similar control language they used for 3000-line and 6000-line machines, and it was much easier to use. GE's was vastly simpler and easier to use, and printed out meaningful error messages to boot. Shall we compare sample error messages from OS/360 and any other computer (possibly excepting the ICL 1900 -- I'm not familiar with that) of its day? GECOS gave nice, semi-English error messages, like "SYSOUT LIMIT EXCEEDED, JOB TERMINATED BY SYSTEM." Compiler error messages were explanatory: "AN APPARENT ARRAY DID NOT APPEAR IN A DIMENSION STATEMENT." IBM, by contrast, gave their users binders full of error codes, with instructions to take "IEW073C" to the appropriate manual and look it up. This was obvious disregard for the user's needs. Slander? Nah. -- Edward "buy you a beer and we can arm-wrestle over this" R ###### Message-ID: <38AEF2E4.82BDDFFA@acm.org> From: "Joel C. Ewing" X-Mailer: Mozilla 4.6 [en] (Win98; U) X-Accept-Language: en MIME-Version: 1.0 Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? References: <20000211101241.23575.00002605@ng-fu1.aol.com> <38A756F5.C16E5FED@silcom.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 73 Date: Sat, 19 Feb 2000 19:46:53 GMT NNTP-Posting-Host: 158.252.167.172 X-Complaints-To: abuse@earthlink.net X-Trace: newsread2.prod.itd.earthlink.net 950989613 158.252.167.172 (Sat, 19 Feb 2000 11:46:53 PST) NNTP-Posting-Date: Sat, 19 Feb 2000 11:46:53 PST Organization: EarthLink Network, Inc. Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!hermes.visi.com!news-out.visi.com!cam-news-hub1.bbnplanet.com!lsanca1-snf1!news.gtei.net!newsfeed2.earthlink.net!newsfeed.earthlink.net!nntp.earthlink.net!newsread2.prod.itd.earthlink.net.POSTED!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:50098 Edward Rice wrote: > > In article <38A756F5.C16E5FED@silcom.com>, > Lars Poulsen wrote: ... > > On S/360, this kind of fine-grained tuning was available to files > > on disk devices as well. Every model of disk drive had different > > track sizes, and under JCL control you specified how you wanted > > the tracks allocated to your files to be formatted: > > //MYFILE DD DISP=(NEW,CATLG),DCB=(BLKSIZE=800,LRECL=80) > > The complexity of distinguishing in fine detail between disks and tapes, to > pick a large example, was just silly. The operating system could have -- > and should have -- taken on the burden of doing things efficiently, once > the user expressed the information requirement of the space. Other systems > did this, and much less painfully than IBM tried. I actually did all that ... While tedious to use and "ugly" it did make cross system compatability possible, which was a much more serious problem then than now, and you still see problems with tape compatability today. Systems which hide all the ugly details of record format and block size from the user make it possible to generate a tape for offsite usage an NOT HAVE A CLUE as to how to accurately describe the tape contents for processing on a different platform. Fortunately MVS JCL provides the flexibility to work around the limitations on data formats imposed by other less-flexible operating system platforms. Now as to the syntax chosen to provide this flexibility, I don't know how to rationally explain that. It must somehow have fallen out of the original code design--surely no one would have actually sat down and designed a user interface so gross as to use reasonable keyword/value pairs in some places, positionally sensitive keywords in others, and meaningless combinations like "OPTCD=Q" in others! > > I am aware of no other system that allowed the user this much > > control. Everyone else just formatted the whole drive in fixed > > length sectors. And I have heard that even S/90 has now caved in > > to that. > The IBM RAMAC and RVA Storage Subsystems used on S/390 undoubtably use fixed-sectoring at the hardware level on their internal 3.5" drives, but they still use that hardware to emulate variable block, Count-Key-Data 3390 drive architecture. From the Operating System's perspective nothing has changed, the DASD is still variable blocksize--and this is crucial to maintaining the upward compatability which has continued to be a very large part of the success of the S/360/370/390 platforms. ... > ... Shall we > compare sample error messages from OS/360 and any other computer (possibly > excepting the ICL 1900 -- I'm not familiar with that) of its day? GECOS > gave nice, semi-English error messages, like "SYSOUT LIMIT EXCEEDED, JOB > TERMINATED BY SYSTEM." Compiler error messages were explanatory: "AN > APPARENT ARRAY DID NOT APPEAR IN A DIMENSION STATEMENT." IBM, by contrast, > gave their users binders full of error codes, with instructions to take > "IEW073C" to the appropriate manual and look it up. This was obvious > disregard for the user's needs. Slander? Nah. > The concept of standardized message ID's was a good one--you could identify the product generating the message and know where to look for clarification of the message and possible corrective action. Using one or more message lines for cryptic return codes, reason codes, etc. when the same amount of space would have sufficed for a clear-text English description of the problem was not. You could perhaps argue now-a-days that message automation might have an easier time dealing with terse return codes & reason codes, but console message automation wasn't an issue when this interface was designed. -- Joel C. Ewing, Fort Smith, AR jcewing@acm.org ###### From: ab528@FreeNet.Carleton.CA (Heinz W. Wiggeshoff) Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: 19 Feb 2000 21:48:58 GMT Organization: The National Capital FreeNet Lines: 45 Message-ID: <88n34a$jr3$1@freenet9.carleton.ca> References: <20000211101241.23575.00002605@ng-fu1.aol.com> <38A756F5.C16E5FED@silcom.com> <38AEF2E4.82BDDFFA@acm.org> Reply-To: ab528@FreeNet.Carleton.CA (Heinz W. Wiggeshoff) NNTP-Posting-Host: freenet10 X-Trace: freenet9.carleton.ca 950996938 20323 134.117.136.30 (19 Feb 2000 21:48:58 GMT) X-Complaints-To: complaints@ncf.ca NNTP-Posting-Date: 19 Feb 2000 21:48:58 GMT X-Given-Sender: ab528@freenet10.carleton.ca (Heinz W. Wiggeshoff) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!newshunter!cosy.sbg.ac.at!news-feed.inet.tele.dk!bofh.vszbr.cz!howland.erols.net!torn!cunews!freenet-news!FreeNet.Carleton.CA!ab528 Xref: chonsp.franklin.ch alt.folklore.computers:50029 "Joel C. Ewing" (jcewing@acm.org) writes: [snip] > Now as to the syntax chosen > to provide this flexibility, I don't know how to rationally explain > that. It must somehow have fallen out of the original code > design--surely no one would have actually sat down and designed a user > interface so gross as to use reasonable keyword/value pairs in some > places, positionally sensitive keywords in others, and meaningless > combinations like "OPTCD=Q" in others! I'm sure that the same crowd that designed the /360 assembler language were tasked with JCL, utility statements, and VM/CMS EXEC implementation - all that && bullshit is a dead-giveaway. Thank god, (oops), Colishaw for REXX, TSO for free-format JCL, etc. > The IBM RAMAC and RVA Storage Subsystems used on S/390 undoubtably use > fixed-sectoring at the hardware level on their internal 3.5" drives, but > they still use that hardware to emulate variable block, Count-Key-Data > 3390 drive architecture. From the Operating System's perspective > nothing has changed, the DASD is still variable blocksize--and this is > crucial to maintaining the upward compatability which has continued to > be a very large part of the success of the S/360/370/390 platforms. VSAM would be dead without it. > > The concept of standardized message ID's was a good one--you could > identify the product generating the message and know where to look for > clarification of the message and possible corrective action. Using one > or more message lines for cryptic return codes, reason codes, etc. when > the same amount of space would have sufficed for a clear-text English > description of the problem was not. You could perhaps argue now-a-days > that message automation might have an easier time dealing with terse > return codes & reason codes, but console message automation wasn't an > issue when this interface was designed. How much core ( $$$$/kilobyte ) can the designer give over to error messages? Imagine a 256Kb model 50 running MFT, or the largest service bureau machine in 1970 Ottawa running MVT in 2 megabytes? Hence the racks of manuals - if the manual was available, the problem was usually solvable without bugging an SE. Oh ya, dump reading was a required skill until PL/I came along. B-) ###### From: Brian Inglis Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: Sun, 20 Feb 2000 11:17:41 -0700 Organization: Systematic Software Reply-To: Brian.dot.Inglis@SystematicSw.ab.ca Message-ID: References: <87ujuo$qji$1@bob.news.rcn.net> <87v1a0$uj@weyl.math.psu.edu> X-Newsreader: Forte Agent 1.7/32.534 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit NNTP-Posting-Host: 207.148.140.84 X-Original-NNTP-Posting-Host: 207.148.140.84 X-Trace: 20 Feb 2000 11:17:42 -0700, 207.148.140.84 Lines: 29 X-Original-NNTP-Posting-Host: 204.50.1.43 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!logbridge.uoregon.edu!newsfeed.direct.ca!nntp.cadvision.com!news.cadvision.com!207.148.140.84 Xref: chonsp.franklin.ch alt.folklore.computers:50116 On 10 Feb 2000 13:50:40 -0500, viro@weyl.math.psu.edu (Alexander Viro) wrote: >In article <87ujuo$qji$1@bob.news.rcn.net>, >Mike Duffy wrote: >>Some systems do not have a "program loader" per se. For example, >>VMS maps a virtual address range and assigns the image file as the >>backing store for page faults. In effect, the image file on disk (at least >>the >>read-only portions of it) become an instant "page file", ready to resolve >>page faults. Pages of the image come into memory in just this way - as >>the result of page faults. (once resident, read-write portions must be >>faulted >>out to process-specific paging locations). > >...ahem. s/Some systems/Almost everything on boxen with paging MMU/. >Demand-paging is not VMS-specific (I would be really amazed if VMS was >the first who did it - I'ld bet on MULTICS or TENEX, depending on the >timing. Or on the early BBN stuff). ISTR the VAX claim to fame was not providing reference bits, so it was impossible to tell if a page had been accessed, and difficult to implement LRU paging, so VMS didn't. Not sure what BSD VAX did. Anyone know? Thanks. Take care, Brian Inglis Calgary, Alberta, Canada -- Brian_Inglis@CSi.com (Brian dot Inglis at SystematicSw dot ab dot ca) use address above to reply ###### Path: chonsp.franklin.ch!not-for-mail From: Neil Franklin Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: 20 Feb 2000 23:25:00 +0100 Organization: My own Private Self Lines: 63 Message-ID: <6uk8jzqzpf.fsf@chonsp.franklin.ch> References: <87ujuo$qji$1@bob.news.rcn.net> <87v1a0$uj@weyl.math.psu.edu> NNTP-Posting-Host: chonsp.franklin.ch X-Trace: chonsp.franklin.ch 951085500 447 10.0.3.2 (20 Feb 2000 22:25:00 GMT) X-Complaints-To: news@chonsp.franklin.ch NNTP-Posting-Date: 20 Feb 2000 22:25:00 GMT X-Newsreader: Gnus v5.7/Emacs 20.4 Xref: chonsp.franklin.ch alt.folklore.computers:50119 Brian Inglis writes: > On 10 Feb 2000 13:50:40 -0500, viro@weyl.math.psu.edu (Alexander > Viro) wrote: > > >In article <87ujuo$qji$1@bob.news.rcn.net>, > >Mike Duffy wrote: > >>Some systems do not have a "program loader" per se. For example, > >>VMS maps a virtual address range and assigns the image file as the > >>backing store for page faults. In effect, the image file on disk (at least > >>the > >>read-only portions of it) become an instant "page file", ready to resolve > > > >...ahem. s/Some systems/Almost everything on boxen with paging MMU/. Right. Includes even 80[3456]86 such as this one (AMD K6-2) I am writing this on. > ISTR the VAX claim to fame was not providing reference bits, so Yes. > it was impossible to tell if a page had been accessed, and No. See below. > difficult to implement LRU paging, so VMS didn't. AFAIK VMS does implement paging, but with NUR (Not Used Recently) algorithm (like most other OSes actually, at least all 80x86 or M68k based ones I know). > Not sure what BSD VAX did. Anyone know? The same as VMS. It emulated them. At least to the extent needed for NUR. To emulate "accessed" have the "page for reuse" routine first set the "read access" flag to off, if the page is in use it will soon page access fault (so becoming an trap on access flag), then recognize that it is not a real access violation and re-set the flag. Then with a delay regard all pages not yet re-set as not accessed recently -> reusable. To emulate "dirty" set all not yet written to pages to "no write access" and then treat the "write access violation" for actually write-allowed pages as "first write" (so it is a trap on write) and only then set the "write access" flag to on, and then treat it as "dirty" flag (i.e. dirty and not accessed recently are candidates for writing to the pagefile, followed by un-dirtying). Yes, this sucks a bit, but not much. And it does work. And it does make the hardware algorithm independant and simpler. I suppose this would be the "proper" way to do it on a RISC. But even all of them I know have "accessed" and "dirty" flags separate from access rights. -- Neil Franklin, neil@franklin.ch.remove http://neil.franklin.ch/ Nerd, Geek, Hacker, Unix Wizzard, Sysadmin, Roleplayer, Mystic Computer: a toy, speeds work so that you have more time to play ###### From: bjh21@cus.cam.ac.uk (Ben Harris) Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Date: 20 Feb 2000 23:54:00 GMT Organization: University of Cambridge, England Lines: 21 Message-ID: <88puqo$86c$1@pegasus.csx.cam.ac.uk> References: <87v1a0$uj@weyl.math.psu.edu> <6uk8jzqzpf.fsf@chonsp.franklin.ch> NNTP-Posting-Host: ursa.cus.cam.ac.uk Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.grnet.gr!news-feed1.eu.concert.net!newsfeed.gamma.ru!Gamma.RU!news.vas-net.net!server2.netnews.ja.net!pegasus.csx.cam.ac.uk!bjh21 Xref: chonsp.franklin.ch alt.folklore.computers:50169 In article <6uk8jzqzpf.fsf@chonsp.franklin.ch>, Neil Franklin wrote: [Emulated modified/referenced bits] >Yes, this sucks a bit, but not much. And it does work. And it does >make the hardware algorithm independant and simpler. I suppose this >would be the "proper" way to do it on a RISC. But even all of them I >know have "accessed" and "dirty" flags separate from access rights. The MEMC that went with the ARM 2 didn't, but I doubt that will surprise anybody. A slightly scarier thing I realised recently is that it also doesn't have an easy method to find out the address you were accessing to cause a page fault. The OS has to look at the instruction that faulted, and at the processor state at the time, and work out what memory location(s) that instruction was trying to reference. Fun. -- Ben Harris Unix Support, University of Cambridge Computing Service. If I wanted to speak for the University, I'd be in ucam.comp-serv.announce. ###### From: David Wragg Newsgroups: alt.folklore.computers Subject: Re: Correct usage of "Image" ??? Message-ID: References: <87v1a0$uj@weyl.math.psu.edu> <6uk8jzqzpf.fsf@chonsp.franklin.ch> <88puqo$86c$1@pegasus.csx.cam.ac.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii User-Agent: Gnus/5.0802 (Gnus v5.8.2) XEmacs/20.4 (Emerald) Lines: 20 Date: 21 Feb 2000 14:07:40 +0000 NNTP-Posting-Host: 194.119.176.228 X-Complaints-To: news@u-net.net X-Trace: newsr2.u-net.net 951156843 194.119.176.228 (Mon, 21 Feb 2000 18:14:03 GMT) NNTP-Posting-Date: Mon, 21 Feb 2000 18:14:03 GMT Organization: (Posted via) U-NET Internet Ltd. Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!newshunter!cosy.sbg.ac.at!news-feed.inet.tele.dk!bofh.vszbr.cz!newsfeed.tli.de!news-raspail.gip.net!news-lond.gip.net!news.gsl.net!gip.net!peer.news.th.u-net.net!u-net!newsr2.u-net.net!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:50162 bjh21@cus.cam.ac.uk (Ben Harris) writes: > A slightly scarier thing I realised recently is that it also doesn't have > an easy method to find out the address you were accessing to cause a page > fault. The OS has to look at the instruction that faulted, and at the > processor state at the time, and work out what memory location(s) that > instruction was trying to reference. Fun. On RISC machines this isn't so bad, since the load/store instruction formats are few and simple. The Linux ARM2 port manages it in about 150 lines of assembly, though those are prefixed with the comment * This is where I wish that the ARM would tell you which address aborted. (as later ARM generations do). Doing the same kind of thing on an x86 would be an utter nightmare, of course. David Wragg