Received: from SOUTH-STATION-ANNEX.MIT.EDU by po7.MIT.EDU (5.61/4.7) id AA03284; Thu, 28 Dec 95 07:16:02 EST Received: from SENATOR-BEDFELLOW.MIT.EDU by MIT.EDU with SMTP id AA02550; Thu, 28 Dec 95 07:15:56 EST Received: (from root@localhost) by senator-bedfellow.MIT.EDU (8.6.12/2.3JIK) id HAA27685 for Linux-Development-System-Dist@senator-bedfellow.mit.edu; Thu, 28 Dec 1995 07:13:36 -0500 Message-Id: <199512281213.HAA27685@senator-bedfellow.MIT.EDU> From: Digestifier To: Linux-Development-System@senator-bedfellow.MIT.EDU Reply-To: Linux-Development-System@senator-bedfellow.MIT.EDU Date: Thu, 28 Dec 95 07:13:33 EST Subject: Linux-Development-System Digest #160 Linux-Development-System Digest #160, Volume #2 Thu, 28 Dec 95 07:13:33 EST Contents: Re: Linux on Suns (Miguel de Icaza) Re: Thanks kernel developers! (Miguel de Icaza) Latest version (Javier Fernandez Marquina) Re: Latest version (Javier Fernandez Marquina) libelf? (Wen-Chun Ni) C++ mangling and dlsym() (Wen-Chun Ni) Re: RAID disk (Nat) Re: C++ mangling and dlsym() (Drew Eckhardt) Re: Can I get fine timing/sleep in a kernel module? (Robert Brockway) Re: Upgraded memory ... how do i benchmark ... (Robert Brockway) Re: Elf Sluggish - interactive perf measure wanted. (Robert Brockway) Build of 1.3.51 fails on ppp.c (none) Re: Problems with Motif 2.0 ELF and Slackware 3.0 (none) Re: Elf Sluggish - interactive perf measure wanted. (John S. Dyson) Linux 1.2.13 and n_tty.c ("James L. Henrickson") Problem with 1.3.51 (Aurel Balmosan) Is there a GCC HOWTO in the works? (Daniel Reish) Stable Filesystem only when boot from floppy (Winfried Truemper) ---------------------------------------------------------------------------- From: miguel@luthien.nuclecu.unam.mx (Miguel de Icaza) Subject: Re: Linux on Suns Date: 27 Dec 1995 18:06:23 -0600 > *Does* linux actually run on a SPARC chip? I know the code is in arch/, > but I didn't know that it was actually functional at this stage. Yes, it does. It's still missing some thing (a cute distribution, and some bug fixes), but it runs most SunOS programs (netscape, mosaic, emacs and others), has a native Libc/SPARC and will soon be working on the 4m machines. Miguel. -- miguel@roxanne.nuclecu.unam.mx The Midnight Commander: http://stekt.oulu.fi/~jtklehto/mc/ ------------------------------ From: miguel@luthien.nuclecu.unam.mx (Miguel de Icaza) Crossposted-To: comp.os.linux.misc Subject: Re: Thanks kernel developers! Date: 27 Dec 1995 18:12:17 -0600 > I just grabbed a copy of the 1.3.50 kernel, and was very pleased with the > new feature where it lets you get information about almosy any given > question that is asked when you configure the kernel. FOr example, I Not only that, but before 1.3.40 (more or less), I could just run X, a couple of terminals and a gdb cross debugger to a SPARC and that was about it. Trying to do anything when X and the debugger were running was very slow, all this on a 66Mhz Pentium with 16 mb of ram. Since the kswap patches went in (plus other enhancements as well), yesterday, when I was running X, I switched to text mode to run the DOS emulator and play Dark Forces. When I was waiting for the game to load (it is located an nfs partition), I telneted from my other linux machine (Linux/SPARC) to my PC and found that I had netscape and the giant cross debugger running :-) and things were running smooth. Quite impressive. Miguel. -- miguel@roxanne.nuclecu.unam.mx The Midnight Commander: http://stekt.oulu.fi/~jtklehto/mc/ ------------------------------ From: jfm@iluso.ci.uv.es (Javier Fernandez Marquina) Subject: Latest version Date: 27 Dec 1995 21:42:44 GMT ------------------------------ From: jfm@uv.es (Javier Fernandez Marquina) Subject: Re: Latest version Date: Wed, 27 Dec 1995 22:48:48 +1000 In article <4bsekk$18co@power.ci.uv.es> jfm@iluso.ci.uv.es (Javier Fernandez Marquina) writes: >From: jfm@iluso.ci.uv.es (Javier Fernandez Marquina) >Subject: Latest version >Date: 27 Dec 1995 21:42:44 GMT Does someone know which is the last version of Linux? Is it worth it installing 1.3.* ??? Or should I wait for 1.4.* ??? Thank you. ------------------------------ From: wcn@doorknob.cs.brown.edu (Wen-Chun Ni) Subject: libelf? Date: 28 Dec 1995 00:10:53 GMT I have to compile a program that needs the calls from libelf.h in order to run. The include file is there while the library is not shipped with libc 5.2.16 and 5.2.18. I can't find any information from the ELF-HOWTO. Can someone with the knowledge tell me where I can find the source of the library? Thanx very much! Wen-Chun -- ================================ Linux: the revenge of communism. ------------------------------ From: wcn@doorknob.cs.brown.edu (Wen-Chun Ni) Subject: C++ mangling and dlsym() Date: 28 Dec 1995 00:17:51 GMT Since dlsym() is a purely C call, we have no way of getting something like MyClass::draw() directly from a shared library. I've found that, in libg++, the calls like cplus_mangle() can be used. Before I probe into it, can anybody tell me whether this is a *portable* way of doing things like mangling. Or is it a GNU-only call? For example, if I want to port my program to Solaris and compile it using Sun CC, is it possible to use the GNU call to do it? Any information or advice is appreciated! Wen-Chun -- ================================ Linux: the revenge of communism. ------------------------------ From: Nat.Makarevitch@nataa.frmug.fr.net (Nat) Subject: Re: RAID disk Date: 27 Dec 1995 01:45:31 +0100 >>>>> "VB" == Valery Brasseur writes: VB> Is there any drivers for RAID disk under Linux ? Is this for a special VB> hardware or can it be done in software? Title: Multiple Device driver Version: v0.32 Entered-date: 28NOV95 Description: Device driver used to group several block devices as a single drive, whatever the interface type. Support for RAID-0 (disk striping). Also supports linear concatenation of devices. Keywords: device driver hard disk RAID striping Author: zyngier@amertume.ufr-info-p7.ibp.fr (Marc ZYNGIER) Maintained-by: zyngier@amertume.ufr-info-p7.ibp.fr (Marc ZYNGIER) Primary-site: sweet-smoke.ufr-info-p7.ibp.fr public/Linux 32k md032.tar.gz Alternate-site: ftp.mcc.ac.uk pub/linux/ALPHA/md linux.nrao.edu pub/linux/packages/MD-driver Platform: Linux 1.3.45 Copying-policy: GPL -- Linux ------------------------------ From: drew@qualcomm.com (Drew Eckhardt) Subject: Re: C++ mangling and dlsym() Date: 27 Dec 1995 17:36:41 -0700 In article <4bsnnf$mlf@cocoa.brown.edu>, Wen-Chun Ni wrote: >Since dlsym() is a purely C call, we have no way of getting something like > MyClass::draw() >directly from a shared library. I've found that, in libg++, the calls >like cplus_mangle() can be used. Before I probe into it, can anybody >tell me whether this is a *portable* way of doing things like mangling. No. ARM/cfront mangles one way; GNU mangles another way; the AIX compiler mangles a third, etc. >Or is it a GNU-only call? Yes. >For example, if I want to port my program >to Solaris and compile it using Sun CC, is it possible to use the >GNU call to do it? No. You probably want to implement a name mangling abstraction class; under Linux, use the libg++ stuff. Elsewhere, you may need to setup a pipe to a demangler program and parse the output. -- Four boxes : soap, ballot, jury, ammo. | Work: drew@Qualcomm.COM Use in that order. | Play: drew@PoohSticks.ORG ------------------------------ From: robert@slip.cc.uq.oz.au (Robert Brockway) Subject: Re: Can I get fine timing/sleep in a kernel module? Date: 28 Dec 1995 06:16:02 GMT Reply-To: ec531667@student.uq.edu.au : >Well in order to make Linux conform to the POSIX real time spec, a new : >system call nanosleep has been implemented recent 1.3 kernels. : >Yep - it sleeps nano seconds. Impressed yet :-) : That must be very recent kernels, I don't see it in 1.3.50 It turns out to be mips specific at the moment. -Robert --Robert Brockway, email: ec531667@student.uq.edu.au WWW: http://student.uq.edu.au/~ec531667 where did my .sig go? ------------------------------ From: robert@slip.cc.uq.oz.au (Robert Brockway) Crossposted-To: comp.os.linux.setup,comp.os.linux.hardware Subject: Re: Upgraded memory ... how do i benchmark ... Date: 28 Dec 1995 06:26:22 GMT Reply-To: ec531667@student.uq.edu.au Gary Heston (gary@scisysgw.sci.com) wrote: : In article <4bia10$sap@bcarh8ab.bnr.ca> hadi@bnr.ca (Jamal Hadi Salim) writes: : >I have just upgraded my PCs mem from 8 to 16 M. : [ ... ] : >I have a 486-dx4100 (50.08 bogomips) and i was able to compile the : >kernel in 16 minutes flat. unfortunately i never kept track when i had : >8 meg. What i have kept records of is the result of Mark Lord's : >hdparm on my system. And i see no difference with what i have at the : >moment : So pull the two SIMMs you added back out and do a kernel compile. : hdparm isn't going to show any significant difference because you : didn't make changes to your hard drive subsystem; as long as there's : enough memory for buffers, hdparm should be happy. Setting the MEM="8Mb" kernel parameter from lilo or loadlin would be less painful :-) Better check the exact syntax, that is from memory. Followups set to c.o.l.hardware. -Robert --Robert Brockway, email: ec531667@student.uq.edu.au WWW: http://student.uq.edu.au/~ec531667 Computers: Can't live with them, can't play Doom without them. ------------------------------ From: robert@slip.cc.uq.oz.au (Robert Brockway) Subject: Re: Elf Sluggish - interactive perf measure wanted. Date: 28 Dec 1995 06:36:19 GMT Reply-To: ec531667@student.uq.edu.au Edmund Ronald (eronald@cnam.fr) wrote: : Questions: : 1. am i the only one who thinks the latest ELF offerings are slower than the a.out's ? Due to the way elf works, it is inherently slower at starting a program than a.out. Once a program is running, there should be no speed difference. Everytime i try to explain why the change to elf is being made i always leave something, so RTFM the ELF-howto for an explanation. : 2. is this the fault of top-heavy releases or of the kernel/libs as rumored ? Well, i think Slackware could do with a bit of an overhaul. It has some persistent bugs, that haven't been fixed from 2.0 to 3.0. : 3. Do we have an interactive performance measure tool which can check things like : interactive lag - that time you wait after doing 'ls' or 'top' and the system : - the new ELF system - just sits there and thinks about things before putting : up a screen - is there such a tool? Well there r alot of bechmarks out there :-) try some. : I'm afraid Linux is moving from "running lite without overbyte' into the mainstream : 128Meg + RAM world, as the developers get free alphas... In other words, I'm afraid : that now the developers have *REAL* machines, pentiums and such, the lightning-fast : character of Linux ios disappearing. well Linus got one. Linux us actually becomming faster, and more efficient (Except for the change to ELF of course). : Edmund. btw: Watch your line length. The standard for Usenet is 78 chars. Cheers, -Robert --Robert Brockway, email: ec531667@student.uq.edu.au WWW: http://student.uq.edu.au/~ec531667 My sig is still gone. Damn. where did it go? ------------------------------ From: none Subject: Build of 1.3.51 fails on ppp.c Date: Thu, 28 Dec 1995 17:42:07 +1100 When building the more recent kernels (1.3.46 - 1.3.51) with all of SLIP, CSLIP, PPP turned on, the compilation of drivers/net/ppp.c fails. I recall reading a post about this problem for 1.3.48, but it still hasn't been fixed in 1.3.51. make[2]: Entering directory `/usr/src/linux/drivers/net' gcc -D__KERNEL__ -I/usr/src/linux/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -pipe -m486 -c -o ppp.o ppp.c In file included from ppp.c:107: slhc.h:128: field `cs_tcp' has incomplete type make[2]: *** [ppp.o] Error 1 make[2]: Leaving directory `/usr/src/linux/drivers/net' make[1]: *** [sub_dirs] Error 2 make[1]: Leaving directory `/usr/src/linux/drivers' make: *** [linuxsubdirs] Error 2 The type cs_tcp is incomplete since the type 'struct tcphdr' is undefined at that point. 'struct tcphdr' is defined in include/linux/tcp.h. By adding an include line for in ppp.c, the build succeeds. The subsequent kernel works as expected for SLIP, CSLIP, PPP. For my configuration, only ppp.c fails to compile. Here is my /usr/src/linux/.config if it helps: begin 664 .config.gz M'XL("-TVXC```RYC;VYF:6<`A5;?;Z,X$'[O7V&I#[OW<-NT>QO=G;0/CC&) M%<"N#6FZ+Q8-;H.:0`2DJ_[W-]"F,8GI2952YOMF/#.>'[Z\N$1XWY3;M,E7 MZ6;SBIY,8:JT,1G:IL\&KH;Q&1=FTM(MW0D3CGZ\V?ZFOQZ&#*`@#XO%#I!*[ M(E;QC]&'6V)CTAJ\,09E5?YBJOJJ.^I*4NR%]%N>&?185FAM-KNKO'@L45D@ MYM$W=FV[?SAUYCD./8"@:N7Q*,0Q%LR)$&=H`(`&_>2D6+*81R[=I7>6WL@T MO\OJ.2^>4+EK\K*P]P$-BFH+PT#AA6#@?86^T=OH3VN;S#TNO)H'1C M1K`Z83J.`C$F)+8/ZKG!!!-.T+*GH:XF7+G2U?J"U6U")>ZNQE8/)4]B>I:L MKWG3RNKTT:"F1%!!+_#/VD`=[8NFW*_6)OO#KHPV(]!;,1EPM(4EEFXTXEJT MW1[&B?:8(AQBN7 M_!BZGH!%\[,$*J*8:P!T\H'2?.]]2^UX"!0U(U3947A)&-Z[FC]@PDH`?&G" M0R&I4M2J4"'$84(0N+55N=U5IJ[+JNZF`<`HK0S,`ICF#_M\TZ"T?I]L];=^ M2,0U=-^\^VP2ZZ1>5@U6.`S'#@^B"1AZ76H7$20EP1.C_*7^'/+@7B68* M][R]N1Z-G.8\*HC;2WHGY]^=R$R`>Y"11`W#`\B)&Q\>WXQ&(]MA-=?3Z_%@ M`BC$-P@*3N;4/7UB]W5A2:*^QEM!>']*R/#[ED%?6["MI:[HX1]0S,Q)F%]O5R_U3JMTE5C*NO!<;R*>Q+`JA@P"$/2DPLG)B2+XH$6GR0JY(ER-ZGX M!`O5Y[KP/EP&GU-N&1G=Z('7!!:AG4SXU&P:<4DUF(.)DR@!X\.MJ#VN:80G M@?7D:<5$)%#*`3TU#-L-FOT>WBDXGYQNEW!?9<288#/ <3568#5JGJ_9]8U\EU)_/@E[8_P%I"A;VK0L``,>G ` end ------------------------------ From: none Subject: Re: Problems with Motif 2.0 ELF and Slackware 3.0 Date: Thu, 28 Dec 1995 17:53:27 +1100 > I upgraded to Slackware 3.0 ELF with gcc 2.7.0. When I try > to compile code that contains Motif functions, make says > thatatthere are undefined references to all the Motif functions. I found that after upgrading I needed to insert -lXext in my link line: gcc file.c -L/usr/X11R6/lib -lMrm -lXm -lXext -lXt -lSM -lICE -lX11 To see whether the linker is finding any libraries and what it is doing with them, run the linker with --verbose, when compiling. I find a convenient way to do this is first run gcc -v which outputs the command gcc uses when invoking ld. Copy this command and insert the --verbose flag at the start of the arg list. ------------------------------ From: root@dyson.iquest.net (John S. Dyson) Subject: Re: Elf Sluggish - interactive perf measure wanted. Date: Thu, 28 Dec 1995 00:18:11 GMT In article <4bqhh8$l8d@sheckley.cnam.fr>, Edmund Ronald wrote: >Hi folks - > > I have been running an old Suse (Slackware 2.1 look-alike) a.out with kernel >1.2.5 on my P5/90/16 EIDE system, and have now installed a partition to try the >latest ELF releases on the InfoMagic CD set. > > RedHat 2.1 and Slackware 3.0 both seem SLOOOOOOOOW > >I like the RedHat, (kernel 2.13) really - it works out of the box. I removed > lotsof the superfluous demons from the inittab (nfsd, httpd etc) . It's > still SLOW, although top shows it eats less RAM now, and I don't go into > swapping. Note that FreeBSD has been working this problem all along -- my guess is that the problem that you are seeing is a result of the shared lib implementation. Eventually, FreeBSD using the SunOS style shared libs ended up with approx the same performance as Linux using the old SVR3-style shared libs -- but it took alot of work. In fact FreeBSD really screams when not using shared libs. Now, it is my guess that are just seeing growing pains of Linux using a better shared lib scheme like FreeBSDs. I suspect that the Linux developers will eventually work out the bugs in their implementation (on FreeBSD, slow performance is considered a bug.) It took significant work on both the dynamic run-time linker and the VM system. John Dyson dyson@freebsd.org ------------------------------ From: "James L. Henrickson" Subject: Linux 1.2.13 and n_tty.c Date: Thu, 28 Dec 1995 04:57:04 -0500 Hi, While comparing the 1.2.13 source with 1.3.50, I saw that some changes were made to linux/drivers/char/n_tty.c. In a few places, there is now a check for 0 before decrementing a variable. Were these changes made to fix a bug, or are they there simply because it is safe programming practice (like the additional NULL pointer checks)? Jim Henrickson jim@borg.com ------------------------------ From: aurel@xylo.owl.de (Aurel Balmosan) Subject: Problem with 1.3.51 Date: Thu, 28 Dec 1995 09:57:46 GMT Hello, I use the kernel 1.3.51 and I have found a problem: When I link a program the first time (no exec-file is present) the program works fine. (In this case procps-0.97 ps) After I have recompiled a .o file (ps.o) and linked it the program (ps) crashed in __main. But when I delete ps and link ps completely the program runs fine. I have tried this several times and depending on the change in the source the program crashs on different locations. Is it possible, that the new buffering has a problem with file updates? Or is this a problem of the process cash (if there is one) which rereads not all necessary pages of the program? Also I have found out that this problem don't depent on the flags past to the C-Compiler. I am using following tools: gcc-2.7.2 (i486-linuxaout) binutils-2.5.2.6 (orginal binaries from sunsite) libc-4.7.5 (orginal binaries from sunsite) linux-1.3.51 (none ELF compiled) All programs running are a.out If someone can verify this problem please notice me. Otherwise I have to check for unstable memory or compilation problems of the gcc-2.7.2 Thanks in advance. -- ================================================================ Aurel Balmosan aurel@xylo.owl.de ------------------------------ From: dreish@umich.edu (Daniel Reish) Subject: Is there a GCC HOWTO in the works? Date: 28 Dec 1995 11:28:45 GMT Is anyone working on a GCC HOWTO? I know there is a GCC FAQ, but I'm looking for a guide to keeping a Linux compiling environment working and up-to-date (without waiting for the Red Hat or equivalent package to be released). If such a thing doesn't exist, please mail me everything you know about the subject, and I'll try to put it together in a useful form and submit it as a HOWTO. (I'm still having extensive problems with GCC 2.7.2 &al, and I like to think I'm not too much dumber than the average Linux user, so there must be some demand for such a thing. On the other hand, maybe I am.) -- Dan ------------------------------ From: truemper@ElFi.MI.Uni-Koeln.DE (Winfried Truemper) Subject: Stable Filesystem only when boot from floppy Date: 28 Dec 1995 11:53:41 GMT I have a strange problem with a P/I-P55TP4XE Motherboard: If I boot Linux via lilo, it will lock up after 0-5 minutes with errors from the ext2-filesystem ("ext2-find-....") resulting in massive filesystem-damage. If I boot Linux via a boot-floppy, the system works fine (disk-transfers are beyond 10GB in the past 24 h). The kernels are exactly the same. I tried 1.3.37 and 1.4.49. My setup is: P/I-P55TP4XE Motherboard with 80MB RAM (60ns), 512MB Cache (15ns) Adaptec 2940 SCSI-Controller with three SCSI-disk and one SCSI-CDROM attached: - IBM/DPES-31080 (1GB, internal, end of the scsi-chain) - SEAGATE/ST410800N (10GB, external, end of the scsi-chain) - FUJITSU/M2684S-512 (512MB, internal) - NEC/CD-ROM DRIVE:84 (internal) Miro20SD (S3) Graphics Card (PCI) 3com509 Ethernet-Card (ISA) The same setup with a PCI/II 55SP4 (?) Motherboard worked flawlessly with or without boot-disk (but it is much slower ...). Comments? Winfried ------------------------------ ** 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 ******************************