Received: from PACIFIC-CARRIER-ANNEX.MIT.EDU by po7.MIT.EDU (5.61/4.7) id AA10911; Tue, 9 Jan 96 08:11:05 EST Received: from SENATOR-BEDFELLOW.MIT.EDU by MIT.EDU with SMTP id AA29011; Tue, 9 Jan 96 08:11:08 EST Received: (from root@localhost) by senator-bedfellow.MIT.EDU (8.6.12/2.3JIK) id HAA29993 for Linux-Development-System-Dist@senator-bedfellow.mit.edu; Tue, 9 Jan 1996 07:13:53 -0500 Message-Id: <199601091213.HAA29993@senator-bedfellow.MIT.EDU> From: Digestifier To: Linux-Development-System@senator-bedfellow.MIT.EDU Reply-To: Linux-Development-System@senator-bedfellow.MIT.EDU Date: Tue, 9 Jan 96 07:13:47 EST Subject: Linux-Development-System Digest #210 Linux-Development-System Digest #210, Volume #2 Tue, 9 Jan 96 07:13:47 EST Contents: Sound Blaster Quad CD ROM Help (Barry Young) Re: Linux user's comments (since 0.98! :-) on FreeBSD 2.1 (J Wunsch) Re: clone(), pthreads, tasking GNAT? (Sean McNeil) compatibility: libc 5.2.18 and fileutils 3.12? (Joe Buck) Re: Elf Sluggish - interactive perf measure wanted. ("H.J. Lu") Re: "ramdisk=4096" but I get a 16Mb ramdrive? (Paul Gortmaker) Re: JAVA is hype. ** Http filesystems ** is the holy grail. (B. Scott Ferguson) Re: Switching to 1.3 kernel (Jerzy Kedra) Re: Progs to measure HD performance? (Louis J. LaBash Jr.) Re: Elf Sluggish - interactive perf measure wanted. (Arnt Gulbrandsen) SCSI_MO/WORM_jukebox support (Bakki Kudva) Re: Win95 filesystem support? (Jan Wielemaker) Re: Linux on Suns (Boris Tobotras) Re: JAVA is hype. ** Http filesystems ** is the holy grail. (Michael Hoegeman) Bug in libc 5.0.9 !! (Thomas Leitner) ---------------------------------------------------------------------------- From: Barry Young Subject: Sound Blaster Quad CD ROM Help Date: 9 Jan 1996 05:03:13 GMT I have the Plug and Play Linux by Yggdrasil. I have a Sound Blaster 16 CD. The installation searches all memory addresses but it does not detect my CD-ROM. Can any one help? Thanks, Barry ------------------------------ From: j@uriah.heep.sax.de (J Wunsch) Crossposted-To: comp.unix.bsd.freebsd.misc Subject: Re: Linux user's comments (since 0.98! :-) on FreeBSD 2.1 Date: 8 Jan 1996 23:18:03 GMT Reply-To: joerg_wunsch@uriah.heep.sax.de (Joerg Wunsch) jrmt@uk.gdscorp.com (Jon Thackray) writes: > - How can I turn off synchronous updates on meta-data under FFS? You can mount it with -o async. Anyway, this didn't do much until recently. After the greate ext2fs integration, John Dyson found out a few spots where it's worth the while to make things asynchronous. Anyway, you'll only notice most of the speedup during things like rm -rf's or ln -s orgys (lndir comes to mind). For the daily work, the speed improvement isn't that noticeable. (I'm running my /tmp and my /usr/release mounted async.) > I'd like my disk to be fast and quiet again! Also, what's the status > of the LFS? An incomplete left-over from Berkeley? Left-over, with some people interested in wiping the dust off it. > - Loadable modules don't appear to work out of the box. Compiling the > kernel without MSDOS support, and then saying mount_msdos gives > '(vfs)cannot find msdos_module'. Sounds about as broken as Linux at > the moment ;-) Interesting. It works for me all day. I usually have at least cd9660 and procfs vfs-loaded. > - Is there any documentation on ext2fs within FreeBSD? > How stable is it? (guess I'll have to join the mailing list > @freebsd.org) It's only in FreeBSD-current, since it's still experimental. No clues about its stability, but since John has been releasing it to the public, i assume it's not too bad. > - Can I map alt+left and right cursor keys for switching virtual > consoles, as well as alt+f1, alt+f2, etc? No real idea. > - Linux compatibility compile option? Can I run Linux a.out binaries, > and how? I've not got many ELF binaries yet, so I'm not too worried > about those. I think so, but you have to resort to some documentation. I don't need it, and i don't have some Linux binaries handy. -- cheers, J"org joerg_wunsch@uriah.heep.sax.de -- http://www.sax.de/~joerg/ -- NIC: JW11-RIPE Never trust an operating system you don't have sources for. ;-) ------------------------------ From: sean@mcneil.com (Sean McNeil) Crossposted-To: comp.lang.ada Subject: Re: clone(), pthreads, tasking GNAT? Date: 08 Jan 1996 22:15:04 -0800 In article <4crkak$911@fg70.rz.uni-karlsruhe.de> ig25@fg70.rz.uni-karlsruhe.de (Thomas Koenig) writes: > Is clone() stable enough for real work yet? I don't know, but I'd be interested in any information on this. > Is anybody working on a Linux pthreads implementation based on it? There was a Project entry awhile back that made mention of a threaded kernel and I beleive clone() was created for this purpose. I haven't heard anything else on this. There a two pthread packages that I know of at this time: Chris Provenzano's pthread package. The source is included in libc-5.2.18. mtp@big.aa.net (Michael T. Peterson) PCthreads(tm). I will not support this pthread package unless the full source becomes freely available. > Is anybody working on a tasking GNAT version which includes this > pthread version? Not that I am aware of. I am officially part of ACT now and I am pretty much in charge of the Linux port. I have been working with Chris Provenzano's pthread package, but will listen to any and all arguments regarding the best pthread package to support. Note that the source must be freely available as there may be problems and/or enhancements that need to be made. If you have concerns and/or requests regarding the Linux port of GNAT, please send email to report@gnat.com. ------------------------------ From: jbuck@synopsys.com (Joe Buck) Subject: compatibility: libc 5.2.18 and fileutils 3.12? Date: 9 Jan 1996 06:17:35 GMT Hi, I've attempted to upgrade libc to version 5.2.18 so that I can run the JDK, but it seems to break /bin/ls and /bin/pwd from GNU fileutils 3.12 and GNU sh-utils 3.12 respectively. These are the versions included with Red Hat 2.1. Typing "ls" prints nothing, though "ls *" works as expected. pwd prints pwd: cannot get current directory: Bad address But many things do work. Could these be related to the problem HJ described with "make" where there's a problem reading directories? (I'm puzzled why I haven't seen this mentioned). Do I need a newer set of fileutils/binutils? -- -- Joe Buck (not speaking for Synopsys, Inc) ------------------------------ From: "H.J. Lu" Subject: Re: Elf Sluggish - interactive perf measure wanted. Date: 8 Jan 1996 23:40:15 GMT I guess I have to say something here. First of all, I just moved to CA from NY. I still haven't got my computer yet and I don't what UPS will do to my computer :-(. I know the ELF binaries linked with shared library are slower than the a.out binaries linked with DLL. I have made some measurements for compiling libc. You can find them in one of the libc ChangeLogs. The ELF binary is about 5% slower than the a.out one. But it is far to easy to maintain and build a shared library under ELF. Also dynamic loading and shared C++ library are also very simple to build/use in ELF. You can find an ELF paper at ftp://sunsite.unc.edu/pub/linux/GCC/elf.ps.gz. BTW, when you measure the ELF performance, you should keep most of ELF shared libraries in memory all the time to save the loading time. You can do this by using a login shell linked with the shared ELF C library. Otherwise, you may wind up loading ELF libc.so every time when you run an ELF binary. H.J. ------------------------------ From: gpg109@leonard.anu.edu.au (Paul Gortmaker) Subject: Re: "ramdisk=4096" but I get a 16Mb ramdrive? Date: 9 Jan 1996 19:09:42 +1100 vefatica@syr.edu (Vincent Fatica) writes: >I have always used "ramdisk=4096" in lilo.conf, and would get a 4Mb ramdisk. >But now (1.3.52), I'm getting a 16Mb ramdisk (I've put well over 4Mb of >stuff on it). I tried changing the lilo.conf entry to "ramdisk=1024" and I >still get a 16Mb ramdisk (as reported by "df"). I don't get it. If it's >working correctly, I'd like someone to explain just how it works now. # cd /usr/src/linux # less Documentation/ramdisk.txt Paul. ------------------------------ Crossposted-To: comp.lang.java,comp.os.linux.advocacy From: fergie@netcom.com (B. Scott Ferguson) Subject: Re: JAVA is hype. ** Http filesystems ** is the holy grail. Date: Tue, 9 Jan 1996 04:19:19 GMT Jonathan Payne (jpayne@starwave.com) wrote: : > >So why did dumb X-terminals fail to find a market? : > : > Ask NCD :) : What does Java and Internet terminals have to do with X terminals? : They have very little in common. If I understand correctly, the basic idea for both X terminals and Internet terminals is: 1) They're not general purpose machines. 2) They're cheaper. (At least they're supposed to be cheaper.) If X-terminals failed in the unix market, what market will the Internet terminals find? I have a hard time believing that someone who won't pay < $1,000 for a bottom-of-the-line PC will pay $500 for an internet terminal that won't accept CD-ROMs, won't word-process, won't play DOOM, and requires better than a 14k modem, a $20/month ISP and ties up the phone for hours at a time. But I'm willing to be convinced. -- Scott Ferguson fergie@netcom.com ------------------------------ From: jxa@galaxy.uci.agh.edu.pl (Jerzy Kedra) Subject: Re: Switching to 1.3 kernel Date: 9 Jan 1996 07:50:55 GMT Derek Wollenstein (rickwol@chelsea.ios.com) wrote: > Could anyone tell me what the most recent, stable, 1.3.xx is available. > I'd like to switch to 1.3 but I don't want to find that everything > breaks with the new kernel. Thanks. > -Derek Wollenstein 1.3.48 seems to be ok. -- -- Jerzy Kedra, AGH Krakow, Poland -- PGP Mail Welcome ------------------------------ From: louis@lcjones.aclib.siue.edu (Louis J. LaBash Jr.) Crossposted-To: comp.os.linux.hardware Subject: Re: Progs to measure HD performance? Date: 8 Jan 1996 18:32:44 GMT On Sun, 07 Jan 1996 00:33:07 -0500, Rick Niles wrote: |I can't remember the names of any of the linux progs |to measure HD performance. I've searched the HOWTOs, the sunsite |archive and I can't find them. The crazy thing is I already have |many of them installed! Can someone remind me of the some the |ways to measure HD performance. (i.e. program names.) Try Drew's site, bonnie: lynx ftp://ftp.poohsticks.org/pub/poohsticks/ total 1669 -rw-rw-r-- 1 3494 3494 1393569 Sep 21 03:59 18usc.txt.gz -rw-rw-r-- 1 3494 3494 113141 Sep 21 03:43 SCSI-HOWTO -rw-r--r-- 1 3494 3494 10378 Nov 15 21:39 bonnie.c.gz -rw-rw-r-- 1 3494 3494 5103 Aug 24 16:06 diffs -rw-r--r-- 1 3494 3494 27028 Jun 13 20:08 linux-conference.tgz -rw-r--r-- 1 3494 3494 48275 Nov 15 20:22 pcibios.ps.gz drwx--x--x 3 3494 3494 512 Sep 25 17:26 private -rw-rw-r-- 1 3494 3494 82406 Nov 9 02:47 snapshot.tgz BTW: there are others: hdparm lets you fool with them. Hope this is of some utility. -- Louis-ljl-{LLaBash@eniac.ac.siue.edu | lou@minuet.siue.edu} ------------------------------ From: Arnt Gulbrandsen Subject: Re: Elf Sluggish - interactive perf measure wanted. Date: 09 Jan 1996 07:33:32 +0100 stock@cptpc157.stm.tudelft.nl (Robert Stockmann) > Maybe its time to kick out mr. H.J. Lu ?? > > I assume that the current libc is totally fucked up with respect > to speed. 130 seconds on a ELF system compared to 30 seconds on a a.out > system *is* unacceptable ... Maybe it's time for you to sanity-check your results. If you believe they are representative and correct, I've an old 486 still running a.out which I'd gladly exchange for your P90. Your P90 is at least three times faster, but if a.out is so much faster that you'll still win. --Arnt ------------------------------ From: Bakki Kudva Subject: SCSI_MO/WORM_jukebox support Date: Mon, 08 Jan 1996 22:49:12 -0500 I am interested in using the HP magneto-optical R/W or WORM juke boxes (20Gig for starters) for developing document imaging and HSM applications under Linux. Has anyone used this or similar juke boxes? I have a few general questions on how this might be accomplished. 1. Is it possible to use them with the kernel SCSI code? How about the robot arm movement, media insert/remove commands? I know that they are part of the SCSI 2 spec. 2. In a typical imaging application, is the native file system created on the optical platters? 3. Is the ext2 file system capable of handling the storage of millions of documents, or does this have to be done in conjunction with an rdb and writing raw sectors to the drives. I would appreciate hearing from anyone who has had some experience developing such code. If necessary I would like to undertake the development under the tutelage of a guru! If this is not an appropriate subject for the newsgroup, please email me. TIA, Bakki ------------------------------ From: jan@swisun31 (Jan Wielemaker) Subject: Re: Win95 filesystem support? Date: 9 Jan 1996 09:34:31 GMT leb@toadflax.cs.ucdavis.edu (Tchiu Le) writes: >'You guys know whether or when Linux will support the Windows 95 >filesystem extensions (long file names in particular) ? I've used the driver listed below (version 0.1.6 I think) for a while with Linux 1.3.27. This kernel was a bit unstable for me though (random hangups and repeatable network problems) and I upgraded to 1.3.49, after which this driver fails to compile because the buffer-cache structures have changed. Cheers --- Jan ================================================================ This is an extremely preliminary release of xmsdos, a filesystem that can handle Windows 95 and Windows NT long filenames. Be warned, it may do irrepairable damage to your directories and your disk. It was tested on Linux 1.1.91 and Windows NT 3.5. Where to get: mm-ftp.cs.berkeley.edu:/pub/multimedia/linux/xmsdos/xmsdos-0.1.tgz What works (or is supposed to): * Creating and removing long filenames and directories. * Reading files created by Windows NT and Windows 95. * Writing files that can be read by Windows NT and Windows 95. * Renaming short filenames and directories to other short names. * Renaming long filenames and directories. ================================================================ -- Jan Wielemaker ------------------------------ From: tobotras@horse.jet.msk.su (Boris Tobotras) Subject: Re: Linux on Suns Date: 09 Jan 1996 09:39:51 GMT Miguel, On 08 Jan 1996 17:17:19 -0600, you wrote: MdI> Besides, X is almost running :-). What graphic adapter are you supporting now? -- Best regards, -- Boris. ------------------------------ Crossposted-To: comp.lang.java,comp.os.linux.advocacy From: mh@wx.gtegsc.com (Michael Hoegeman) Subject: Re: JAVA is hype. ** Http filesystems ** is the holy grail. Date: Tue, 9 Jan 1996 10:06:21 GMT B. Scott Ferguson (fergie@netcom.com) wrote: : Brandon J. Van Every (vanevery@blarg.net) wrote: : : You don't understand what Java is trying to do. Java wants to : : _eliminate_ the PC. Clients are diskless dumbed-down boxes with : : modems attached. Servers do all the work. This is a great idea for : : enterprise computing, at least, because its saves a company having to : : pay dozens of admins to handle all those broken Windoze PC's on : : everyone's desk. : So why did dumb X-terminals fail to find a market? Well, maybe you don't use any dumb x terminals but there are in fact a sh*tload of x terminals out there. More than you think i suspect. On the whole they're rather unglamourous but most of them "just work" which is why partly why you "never hear" much about them. ============================================================================== Mike Hoegeman email: mh@wx.gtegsc.com tel: (805)497-5560 GTE Weather Systems Group ------------------------------ From: tom@finwds01.tu-graz.ac.at (Thomas Leitner) Subject: Bug in libc 5.0.9 !! Date: 9 Jan 1996 10:18:44 GMT [ Article reposted from comp.os.linux.development.apps ] [ Author was Thomas Leitner ] [ Posted on 9 Jan 1996 10:18:23 GMT ] Hi, We found the following bug in libc 5.0.9 when we ported our app. to Linux: When a file is opened by two processes where one process repeatedly writes a data block with fwrite and the other process reads it with fread, the reader process does *not* get the actual data from the file. It gets the old data from the stream buffer instead which is *wrong*. I tried this on all sorts of Unixes I could get an account on at our campus (OSF/1 3.2, Solaris 2.4, Ultrix 4.4, SunOS 4.1.3_U1) and it works as its supposed to but on Linux (Slackware 3.0) with libc 5.0.9, the same data is read all over again. Note that the fix for this is to set the input stream to unbuffered using "setbuf(fi, NULL);". Comments anyone? Tom P.S: Here's my test program. Start writer process with "lbug" and reader process with "lbug read". /* lbug.c: demonstrates a bug in libc 5.0.9 under Linux */ #define _POSIX_SOURCE 1 #include #include void main(int argc, char *argv[]) { /* if more than one argument given -> READ the file */ if (argc > 1) { int i; char buffer[512]; FILE *fi = fopen("test.dat", "rb"); if (fi == (FILE*) NULL) { perror("fopen(test.dat)"); exit(1); } /* THIS IS NEEDED UNDER LINUX W/ LIBC 5.0.9 TO GET IT GOING */ /*setbuf(fi, NULL);*/ while (1) { if (fseek(fi, 0L, SEEK_SET) != 0) { perror("fseek"); exit(1); } if (fread(buffer, 512, 1, fi) != 1) { perror("fread"); exit(1); } for (i = 0; i < 512; i++) { if (buffer[0] != buffer[i]) printf("BUFFER ERROR\n"); } printf("read buffer %d\n", (int) buffer[0]); sleep(1); } } else { /* no argument given -> WRITE the file */ int i = 0; char buffer[512]; FILE *fo = fopen("test.dat", "wb"); if (fo == (FILE*) NULL) { perror("fopen(test.dat)"); exit(1); } while (1) { i++; i = i % 256; memset(buffer, i, 512); if (fseek(fo, 0L, SEEK_SET) != 0) { perror("fseek"); exit(1); } if (fwrite(buffer, 512, 1, fo) != 1) { perror("fwrite"); exit(1); } printf("wrote buffer %d\n", i); sleep(1); } } } /* E O F */ -- Tom Leitner, tom@finwds01.tu-graz.ac.at |\ _,,,---,,_ Dept. of Communications zzz. /,`.-'`' -. ;-;;,_. Graz University of Technology |,4- ) )-,_. ,\ ( `'-' Inffeldgasse 12 '---''(_/--' `-'\_) A-8010 Graz / Austria / Europe, Phone: +43-316-873-7455, Fax: +43-316-463-697 WWW home page: http://wiis.tu-graz.ac.at/people/tom.html -- Tom Leitner, tom@finwds01.tu-graz.ac.at |\ _,,,---,,_ Dept. of Communications zzz. /,`.-'`' -. ;-;;,_. Graz University of Technology |,4- ) )-,_. ,\ ( `'-' Inffeldgasse 12 '---''(_/--' `-'\_) A-8010 Graz / Austria / Europe, Phone: +43-316-873-7455, Fax: +43-316-463-697 WWW home page: http://wiis.tu-graz.ac.at/people/tom.html ------------------------------ ** 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-System-Request@NEWS-DIGESTS.MIT.EDU You can send mail to the entire list (and comp.os.linux.development.system) via: Internet: Linux-Development-System@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-System Digest ******************************