Subject: Linux-Development Digest #518 From: Digestifier To: Linux-Development@senator-bedfellow.MIT.EDU Reply-To: Linux-Development@senator-bedfellow.MIT.EDU Date: Fri, 4 Mar 94 01:13:05 EST Linux-Development Digest #518, Volume #1 Fri, 4 Mar 94 01:13:05 EST Contents: Re: Help! GCC errors (Dean Junk) Help: inode information, a correction (Manish Gupta) Re: Specialix driver (Johannes Stille) Re: Is there a driver for BusLogic 445 VLB (not aha1540). (Mr Ivan Alastair Beveridge) Re: Multi-Serial Cards? (Bill Mitchell) Re: Amiga FileSystem, Anyone? (Eyvind Bernhardsen) Re: Why not put cluster diffs in nominal kernel before 1.0? (Ferny) Re: ROMmable Linux? (Donald Jeff Dionne) Re: Amiga FileSystem, Anyone? (Eyvind Bernhardsen) Re: LINUX FOR SUN (Alan Braggins) Re: Amiga FileSystem, Anyone? (Alan Braggins) Re: Please stop about the .sig (dan@oea.hacktic.nl) Re: effectiveness of cache ram? (Kai Henningsen) Re: Is there support for HPFS? (Thomas Quinot) Re: ISDN card comments wanted (Donald J. Becker) HPIII/IVsi Network Printer Drivers (Keith Smith) Re: Specialix driver (Keith Smith) Re: effectiveness of cache ram? (Keith Smith) ---------------------------------------------------------------------------- From: us292121@bulldog.mmm.com (Dean Junk) Subject: Re: Help! GCC errors Date: 3 Mar 1994 15:57:00 GMT Christopher L Seawood (mgrcls@manager) wrote: : Dean Junk (us292121@bulldog.mmm.com) wrote: : : : I am having the following problem compiling xmix: : : : /usr/lib/libgcc.sa(__libc.o): Definition of symbol __NEEDS_SHRLIB_libc_4 (multiply defined) : : /usr/lib/libc.sa(__libc.o): Definition of symbol __NEEDS_SHRLIB_libc_4 (multiply defined) : : make: *** [xmix] Error 1 : : : : Do you have any ideas? I have everything else working great but this! : : Take this how you wish.... Read the release notes. It specifically says to 'rm -f /usr/lib/libgcc.*' : Take this as you wish ... piss off! I can't beleive the attitude of some of the people on this newsgroup. I was upgrading my kernel just for tape support and ended upgrading just about everything on my system. So when you say 'Read the release notes', I just want to puke because I have read about 30 novels of release notes in the last week! And, I can't beleive people are wasting bandwidth by following up to this post after I have already posted that I KNOW THE ANSWER! Besides, 30 people before you have already posted the same response! So, gee whiz, I missed a line in the release notes. I am sorry. But for gods sake, stop following up on this post! -- Dean Junk "An ounce of perception, a pound of obscure" Internet (dpjunk@mmm.com) --RUSH ------------------------------ From: manish@ms.uky.edu (Manish Gupta) Subject: Help: inode information, a correction Date: 3 Mar 1994 17:39:15 -0500 Hello, I guess I had put my question incorrectly earlier. I will try again. I am writing a daemon process to write some data into a file on the disk. Since the application is time critical, I don't want toi make any system call, `read`, `write`, `open` etc. The easiest thing would be directly call the function which is called by filesystem's `read` call handler [struct file_operations.read()...]. To do this I my daemon needs to have a pointer to the actual inode structure for the file. I hope this time I am much more clear than before. Now, can anybody help me? Daemon process does have root privileges and everything I have said, meant or would like know is in the context of Linux OS. thanks again for your time and replies. sincerely, - manish -- I call them as I see them. If I can't see them, I make them up. -- Biff Barf ------------------------------ From: johannes@titan.westfalen.de (Johannes Stille) Subject: Re: Specialix driver Date: Thu, 3 Mar 1994 13:04:54 GMT In article <1994Feb24.021506.25055@rpp386> jfh@rpp386.cactus.org (John F. Haugh II) writes: >In article <1994Feb22.173853.19781@super.org> becker@super.org (Donald J. Becker) writes: >>If your downloaded code implements a defined, public interface that's used >>by several operating systems, it probably doesn't fall under the GPL. >> >>A Linux device driver, on the other hand, is an integral and >>inseparable(1) part of the Linux kernel. By the terms of the GPL it must be >>released in source form. > >Only if it is distributed integrally and inseparably. The GPL cannot >cover something which isn't contaminated by GPL'd code. If you include >no GPL'd code in what you distribute, you needn't GPL what you are >putting out. My guess is that without GPL'ing the headers (that is, >if the /usr/include files are LGPL'd) you couldn't prevent this from >occuring. And you would have a really hard time if the person had >their own "clean" versions of the system header files. > [...] > >It also doesn't mean the GPL covers them. Despite claims to the >contrary, the GPL cannot infest any code which is not a derivative >work. There isn't a legal basis for it to do so, "inseperable" or >not. [...] I'm afraid you (that is most or all of you, not only John F. Haugh II) are overlooking an important aspect: You only consider the definition of "derivative", but forget the definition of "work". I doubt that a driver that needs to be linked (even at runtime) into the Linux kernel is itself a "work". The driver can't be produced except as a part of a Linux kernel + driver combination. Don't tell me you can write such a driver without testing by compiling it, linking it into the kernel, and running it as part of a modified Linux kernel. And the driver can't be used except as a part of a Linux kernel + driver combination. So the "work" is the modified kernel, and the driver is just a part of this "work", so it is affected by the copyright of the whole "work", even if you distribute only this part. The driver author first creates the complete "work" (modified kernel), and it is in this case covered by the GPL, because it is obviously derived from the GPL'd Linux kernel (obvious because it contains Linus' code). Taking one .o file from this "work" won't change the legal status of this part. This is where a defined, public interface makes a difference. With a public interface, you _use_ the original "work" instead of _modifying_ it, so the code using the original "work" can itself be another "work". BTW, I think your definition of "derive" is too narrow. IMHO there are other ways of "deriving" besides actually copying source code. It e.g. can be enough if you only take over major principles or ideas (as with Apple and the idea of the trash can on your desktop). (This doesn't mean that I like the general current legal status of software.) All this doesn't depend on the wording of the GPL and of RMS' (or Linus' or whoever's) interpretation, but only on the copyright laws. And of course a final decision can only be made in court. (Let's hope this won't be necessary.) Johannes ------------------------------ From: zceed04@ucl.ac.uk (Mr Ivan Alastair Beveridge) Subject: Re: Is there a driver for BusLogic 445 VLB (not aha1540). Date: Thu, 3 Mar 1994 16:08:36 GMT In article <2l4nqc$o3d@sun2.ruf.uni-freiburg.de> pit@lupo.kis.uni-freiburg.de (Peter Suetterlin) writes: >Joseph P DeCello III (decello@discovery.uucp) wrote: > >> Is there a driver available or in development that FULLY supports >> the capabilities of this card. It's work well with AHA1540 emulation >> but I imagine, I won't get the performance of a BusLogic VLB driver. >> Thanks, >> Joe > >Yes, there is. It's on tsx-11, in the BETA directory. I'm using it since >about 3 months now and am very content. It is indeed faster than >in Adaptec mode. > Hmm - I can't find it - the only one that I have seen woz under ALPHA - where exactly is it?? Ivan ------------------------------ From: mitchell@mdd.comm.mot.com (Bill Mitchell) Subject: Re: Multi-Serial Cards? Date: 3 Mar 1994 07:58:53 -0800 Reply-To: mitchell@mdd.comm.mot.com (Bill Mitchell) in comp.os.linux.development, marauder@lod.amaranth.com (marauder) said: >I am looking for som input on which multi-serial cards are available and >compatible for use with Linux. I am basically looking for a 4-port card that >would support dialup modems. Are there any people who have a setup like this >in place? -- any input would be greatly appteciated. > >td Shameless plug follows: Electronics & Computer Surplus City 1490 W. Artesia Blvd Gardenia, CA 90248 1-800-543-0540 1-310-217-0950 (fax) 1-310-217-1922 (BBS on-line Inventory) I just bought an "AST Four-Port/XN" board from them. It has four socketed 16450's. They have it priced at $29.95 (yes, $29.95), including a four-headed octopus (quadopus?) cable. The cable alone probably cost that much originally. It was a snap to install and configure, and came up fine on my linux system. The only problem I have is that setserial insists that the 4th port has an 8250 uart - and I haven't pursued that. -- mitchell@mdd.comm.mot.com (Bill Mitchell) ------------------------------ From: eyvind@lise.unit.no (Eyvind Bernhardsen) Subject: Re: Amiga FileSystem, Anyone? Date: 04 Mar 1994 02:08:44 GMT In article <2l5epv$qm7@bmerha64.bnr.ca> Hamish.Macdonald@bnr.ca (Hamish Macdonald) writes: David> Ok, anybody have minix-fs or ext2fs for the Amiga? Someone has worked on a minix filesystem for AmigaDOS. It's got a few problems, but I've been able to read my (linux/68k) minix partitions from AmigaDOS. I'm not sure if the author has continued work on it. It doesn't allow writes, and I'm not sure how device files are handled. It DOES allow writes. Check out the latest version, it should be on Aminet (ftp.wustl.edu is a good place to start). -- //| Eyvind Bernhardsen - finger for PGP key - eyvind@lise.unit.no \X/-| "MS Word is an ugly, clanking, God-awful mess of a program." -DNA ------------------------------ From: FernyM@pc64.maths.bris.ac.uk (Ferny) Subject: Re: Why not put cluster diffs in nominal kernel before 1.0? Reply-To: Mark.Fernyhough@bristol.ac.uk Date: Thu, 3 Mar 1994 16:45:31 GMT Peter Mutsaers (muts@compi.hobby.nl) wrote: : I don't want to miss the cluster diffs, even for IDE disks. This : afternoon I compared 15d with cluster-08a with 15h (without cluster : diffs since 08a cannot patch 15h :-( ) and 15d with cluster-08a did : 800kb/s read and write on an ext2 filesystem on an IDE disk, 15h did : only 550kb/s. Measured with iozone. Can anyone tell me if the cluster diffs (08a) will patch 15j or will i have to wait for a new version of cluster? Thanks in advance, Mark -- ============================================================================== | Ferny (Mark Fernyhough) | Email: Mark.Fernyhough@bristol.ac.uk | | Dept Maths, Uni of Bristol | or root@pc64.maths.bris.ac.uk | | Bristol, Uk. | Tel: (0272) 303319 | ============================================================================== ------------------------------ From: jeff@ee.ryerson.ca (Donald Jeff Dionne) Subject: Re: ROMmable Linux? Date: 2 Mar 1994 23:12:08 GMT Uri Blumenthal (uri@watson.ibm.com) wrote: : Hi, : Some time ago there was a discussion here about : how possible it was to make embedded Linux... : Could somebody please e-mail me, what's the : chance/status of such project? Is it doable : now? Well, I thought it would be a good thing to start, and asked if there was any takers. I got one reply... be that person wanted to see MCA working properly first. : Thanks! : -- : Regards, : Uri. uri@watson.ibm.com scifi!angmar!uri : ------------ : Jeff#EE.Ryerson.CA ------------------------------ From: eyvind@lise.unit.no (Eyvind Bernhardsen) Subject: Re: Amiga FileSystem, Anyone? Date: 04 Mar 1994 03:15:36 GMT In article eyvind@lise.unit.no (Eyvind Bernhardsen) writes: [Yup, me following up myself] David> Ok, anybody have minix-fs or ext2fs for the Amiga? Someone has worked on a minix filesystem for AmigaDOS. It's got a few problems, but I've been able to read my (linux/68k) minix partitions from AmigaDOS. I'm not sure if the author has continued work on it. It doesn't allow writes, and I'm not sure how device files are handled. It DOES allow writes. Check out the latest version, it should be on Aminet (ftp.wustl.edu is a good place to start). OK, OK, so I may have been wrong--I don't use the Minix file system, and was going by what someone told me. (I've been informed that he was probably thinking about BFFS). Sorry for any confusion this caused. -- //| Eyvind Bernhardsen - finger for PGP key - eyvind@lise.unit.no \X/-| "MS Word is an ugly, clanking, God-awful mess of a program." -DNA ------------------------------ From: armb@setanta.demon.co.uk (Alan Braggins) Subject: Re: LINUX FOR SUN Date: Thu, 3 Mar 1994 12:53:41 GMT In article <2l1tuc$8nt@ifi.uio.no> gunnarr@ifi.uio.no (Gunnar Rxnning) writes: > Probably, but there are an ongoing port to the 680x0 architechture But the machines listed ("IPC, IPX, Classic, LX, etc.") are not 680x0 Suns. -- Alan Braggins armb@setanta.demon.co.uk abraggins@cix.compulink.co.uk "Any technology distinguishable from magic is insufficiently advanced" ------------------------------ From: armb@setanta.demon.co.uk (Alan Braggins) Subject: Re: Amiga FileSystem, Anyone? Date: Thu, 3 Mar 1994 13:02:25 GMT In article dholland@husc9.harvard.edu (David Holland) writes: > > I doubt an Amiga file system would be easier to write under Linux > > than MS-DOS, > That's complete nonsense. MS-DOS has no hooks for adding alternate > file systems. Linux does. Given that the reason Amiga disks can not be read under MS-DOS is hardware related, this is irrelevent. Obviously an Amiga disk reader under MS-DOS, were it to exist, wouldn't be an alternate filesystem. > Ok, anybody have minix-fs or ext2fs for the Amiga? I haven't seen them for Amigados. The Amiga Linux port has minix-fs. -- Alan Braggins armb@setanta.demon.co.uk abraggins@cix.compulink.co.uk "Any technology distinguishable from magic is insufficiently advanced" ------------------------------ From: dan@oea.hacktic.nl Subject: Re: Please stop about the .sig Date: Thu, 3 Mar 1994 23:33:08 GMT David Rapchun (rapchun@suicide.sdsu.edu) wrote: : Everyone, please stop sending me mail about that message i posted with the : long .sig. I have corrected the problem as you can see. I just wish that : last post would hurry up and die since i get responses back every day as it ^^^^^^^^^^^^^^^^ You could cancel it, you know. Ask your news admin to do it if you don't know how. -- |< Dan Naas dan@oea.hacktic.nl >| +--------------------------------------+ ------------------------------ Date: 02 Mar 1994 19:09:00 +0100 From: kai@khms.westfalen.de (Kai Henningsen) Subject: Re: effectiveness of cache ram? haw30@eng.amdahl.com wrote on 02.03.94 in <1994Mar2.014033.15596@amdahl.com>: of >> 64K cache. They quoted a report from some independent testing > lab that found 60% of the motherboards they tested ran faster > with external cache disabled (unfortunetly, there were no details > of how they conducted the tests). Designing a good cache subsytem Well, I can only say that I *never, ever* saw a board running faster with external cache disabled. Quite to the contrary, each and every one were a lot smaller. That is, unless the cache rams were defect, of course. Then it doesn't run with cache, period. Kai -- Internet: kh@ms.maus.de, kai@khms.westfalen.de Bang: major_backbone!{ms.maus.de!kh,khms.westfalen.de!kai} ## CrossPoint v2.93 ## ------------------------------ From: thomas@melchior.frmug.fr.net (Thomas Quinot) Subject: Re: Is there support for HPFS? Date: 2 Mar 1994 23:26:56 +0100 Tibor Polgar (tlp00@aimer.spg.amdahl.com) wrote: : Is there support for HPFS as a VFS, similar to the support for FAT? At first i Yup, but read-only at the moment. Thomas. -- ThoThoThoThoTho Totolitoto ! ------------------------------ From: becker@super.org (Donald J. Becker) Subject: Re: ISDN card comments wanted Date: Thu, 3 Mar 1994 05:31:20 GMT In article , Mark Evans wrote: >Alan Cox (iiitac@swan.pyr) wrote: > >: Linux has few 'special' requirements. In general DMA is a nuisance and for >: small transfers on the PC a total loser. Linux does need interrupts because > >What do you mean by 'small'? >i.e. above what size does using DMA become usefull. Basically if you have to program the DMA controller more than once every 4K of transfers, the overhead is too much(1). There are often-winning (2) ways to use DMA for small or random-sized transfers, like ethernet packets. One way is to make them look like big transfers. The i82593 ethernet controller uses two DMA channels, both configured as recirculating ring buffers. Received packets are stored one after another another in the ring, with the host resetting the Rx stop pointer after it removes each packet. The LANCE ethernet controller uses another approach: becoming a bus master. In this mode the LANCE uses the DMA controller to take over the bus, but doesn't use the DMA address generation. (1) Exceptions to this are devices that want data annoyingly infrequently and don't have any buffer space of their own. The floppy controller uses DMA for this reason -- it wants a byte every 4 to 16 usec., which is too short to take an interrupt but too long to just busy loop. (2) I say "often-winning" because even the best DMA device is not always a win. Some processors will flush their cache after every DMA cycle, which slows the processor down much more than if it had done the transfer itself. ____ A few other nice features a device should have: It should be easy to safely probe for a device. The scheme used on the 3c509 is a good solution for the ISA bus. Once a device is found it's nice to be able to read the resources (IRQ, DMA, and shared memory) that the card wants or is already using. A device should never hold the bus for an indefinite time if data isn't ready to be transferred. The NE*000 will hang the machine if you access its dataport and it's not set up to do a data transfer. A more robust solution is to set an 'underrun' error bit and return invalid data. Make it possible to power down the device if it's not being used. The ethercard drivers take care to put the devices into low-power mode at boot and close() time. In the case of an ISDN card you should have a just-barely-awake mode as well as a asleep-and-deaf-to-the-world mode. -- Donald Becker becker@super.org IDA Supercomputing Research Center 17100 Science Drive, Bowie MD 20715 301-805-7482 ------------------------------ From: keith@ksmith.com (Keith Smith) Subject: HPIII/IVsi Network Printer Drivers Date: Fri, 04 Mar 94 03:10:29 GMT Anyone know where one could locate a Linux driver for an HPIII/IVsi network printer (TCP/IP)? Does HP publish the source for this thing? I currently am running the machine thru an SCO spooler, but would like to print direct from the linux box. -- Keith Smith keith@ksmith.com 5719 Archer Rd. Digital Designs BBS 1-910-423-4216 Hope Mills, NC 28348-2201 Somewhere in the Styx of North Carolina ... ------------------------------ Crossposted-To: gnu.misc.discuss From: keith@ksmith.com (Keith Smith) Subject: Re: Specialix driver Date: Fri, 04 Mar 94 03:22:14 GMT [ Benifit of Commercial HW/SW for Linux ] In article , Doug DeJulio wrote: >Please explain -- how will *I* benefit? EZ. What kind of machine do you hack linux on? An INTEL arch PC. Why? Because it is the cheapest thing out there. Why? Because the SCADS of SOFTWARE for x86/DOS promoted a market for INTEL clone hardware to run it, which promoted more software, which ... Basically the more shit that sells the cheaper it gets (hardware), And the more software produced, the more "standardized" the programming interface, and stable the binaries are produced. Both are benifical to anyone. You always want to promote growth of almost _any_ kind in the market you are in. I had source for all my AS/400 applications, but I wouldn't wanna ever use the dinosaur again. The API interface is so narrow programming was a breeze, and guaranteeing a program would work was easy. On the cost side however with hard disk capacity running 5 times the PC market, and Memory at over 40 times the PC market... Why? No vendors other than IBM. You want cheap AND reliable you want lots of HW/SW vendors, and the more the merrier. -- Keith Smith keith@ksmith.com 5719 Archer Rd. Digital Designs BBS 1-910-423-4216 Hope Mills, NC 28348-2201 Somewhere in the Styx of North Carolina ... ------------------------------ From: keith@ksmith.com (Keith Smith) Subject: Re: effectiveness of cache ram? Date: Fri, 04 Mar 94 04:02:04 GMT In article <1994Mar2.014033.15596@amdahl.com>, Henry A Worth wrote: >In a recent issue of "Microtimes" (a silicon valley area tabloid and >computer shopping mag), there was an article questioning the value >of >64K cache. They quoted a report from some independent testing >lab that found 60% of the motherboards they tested ran faster >with external cache disabled (unfortunetly, there were no details No way. With _several_ different clone motherboards running SCO Unix and DOS I performed similar tests turning cache's on and off, Pulling Cache RAM chips and the like. In _EVERY_ instance turning on the external cache improved performance DRAMATICALLY. And caches above 64K had a MUCH more significant effect under SCO than DOS (18% improvements vs 5%), and became even more significant when Load Averages went above 1.00 (As high as 50% 64K vs 256K running compute intensive awk script). A 486/33 with no external cache and 8MB RAM is darn near unusable under SCO, (basically with more than one user). Once we passed the 25Mhz mark multi-way memory interleaving doesn't hold a candle to a half-assed cache. Running similar tests under Xenix on 20/25Mhz 386 boxes did not show much of a difference, and in fact some of the cache stuff _was_ slower. Now most everyone is using the same chipsets, so it doesn't much matter. -- Keith Smith keith@ksmith.com 5719 Archer Rd. Digital Designs BBS 1-910-423-4216 Hope Mills, NC 28348-2201 Somewhere in the Styx of North Carolina ... ------------------------------ ** FOR YOUR REFERENCE ** The service address, to which questions about the list itself and requests to be added to or deleted from it should be directed, is: Internet: Linux-Development-Request@NEWS-DIGESTS.MIT.EDU You can send mail to the entire list (and comp.os.linux.development) via: Internet: Linux-Development@NEWS-DIGESTS.MIT.EDU Linux may be obtained via one of these FTP sites: nic.funet.fi pub/OS/Linux tsx-11.mit.edu pub/linux sunsite.unc.edu pub/Linux End of Linux-Development Digest ******************************