From: Digestifier To: Linux-Development@senator-bedfellow.mit.edu Reply-To: Linux-Development@senator-bedfellow.mit.edu Date: Tue, 1 Mar 94 21:13:14 EST Subject: Linux-Development Digest #513 Linux-Development Digest #513, Volume #1 Tue, 1 Mar 94 21:13:14 EST Contents: Sendto: Invalid Arguemnet? [NETWORK] (Jay D. Allen) Re: undefined symbols in modules (Matthias Urlichs) Re: accessing BIOS (Robert Sanders) X11R6? (Corey Brenner) Re: ISDN card comments wanted (Alan Cox) Re: LINUX FOR SUN (Ron Smits) Re: Linux and X WordPerfect (Brandon S. Allbery) Re: Why not put cluster diffs in nominal kernel before 1.0? (Matthias Urlichs) Re: ipcs utility, /proc page allocation (Matthias Urlichs) Re: Help! GCC errors (Dean Junk) Re: Linux and X WordPerfect (Ken Clark) Re: Help! GCC errors (Rob Janssen) Re: undefined symbols in modules (Erik Troan) ISDN support for Linux -> status ? (Bernd Hentig) COFF / iABI support with linux 0.99p15 ?!? (Bernd Hentig) ---------------------------------------------------------------------------- From: jay@caeser.geog.pdx.edu (Jay D. Allen) Subject: Sendto: Invalid Arguemnet? [NETWORK] Date: 1 Mar 1994 18:26:22 GMT A week ago I saw an article asking about multiple entrys in the listed routing table.. I just upgraded to libc2.4.21, and pl15. Now I also get this "problem". Although it does not apear to have any ill sideeffects. but.. The binary version of "routed" does not work anymore. When I run a strace on it, it appears to fail on the "sendto:" system call; Sendto: Invalid arguement So, if this is answered somewhere, just point me in the right direction. -- ======================================================================= * Jay D. Allen - METNET at Portland State University * * Department of Geography AND The Center For Science Education * * jay@caeser.geog.pdx.edu * ======================================================================= ------------------------------ From: urlichs@smurf.noris.de (Matthias Urlichs) Subject: Re: undefined symbols in modules Date: 1 Mar 1994 10:52:06 +0100 In comp.os.linux.development, article <1994Feb26.222416.9740@unlv.edu>, ftlofaro@unlv.edu (Frank Lofaro) writes: > > How about someone changing the modules code to not depend on a static > ksyms list? If I while having a system running, want to use a module to > load code that depends on functions or variables not provided for in ksyms, > I have to recompile after adding it to ksyms and reboot before I can use > the module. This defeats the purpose of loadable modules. > You gotta get the list of symbols from _somewhere_. Since it doesn't make sense to list all kernel symbols in the kernel itself (too big) and you shouldn't depend on an external file like /vmlinuz (compressed, you don't know that that's the kernel that got loaded), you need a small list of exported symbold compiled into the kernel. -- Standing on your head makes a smile of a frown, but the rest of your face is also upside down. -- Matthias Urlichs \ XLink-POP Nürnberg | EMail: urlichs@smurf.noris.de Schleiermacherstraße 12 \ Unix+Linux+Mac | Phone: ...please use email. 90491 Nürnberg (Germany) \ Consulting+Networking+Programming+etc'ing 42 Click here. ------------------------------ From: gt8134b@prism.gatech.EDU (Robert Sanders) Subject: Re: accessing BIOS Date: 28 Feb 94 22:59:29 GMT whitney@galileo.Meakins.McGill.CA writes: >: >So, the question remains : How do I get from protected mode >: >to real mode and back ? >I am not intrested in the v86 feature for my experiment - I want >to know how to switch to real mode. If you switch to real mode, you'll kill Linux, plain and simple. Real mode uses a different interrupt vector scheme, so all the Linux interrupt and trap handlers will be useless. I didn't see the original post, but if you really want to go into real mode -- and keep in mind that it's a one-way trip until a reboot -- you can call iopl with an argument of 3. This lets outer ring programs use I/O services. Your program must be running as root to do this. You will then want to set up low memory (0-640k PHYSICAL memory) for the transition to real mode. The easiest way to do this is to mmap() 4k-640k to somewhere in your memory (I believe that the zero page is not available from Linux). You'll need access to the zero page, however (PHYSICAL addresses 0-4095), as you'll need to set up some interrupt handlers. I believe that Linux leaves the bootup interrupt vectors alone, so you can probably skip this, but if not, you'll need to modify the zero page. You can find the address of the page directory from the CR3 register (you'll have to use inline assembly to get this). You will then find some unused page in one of the page tables and map the zero page there R/W. Set up any interrupt handlers you need to, insert whatever real-mode code you plan on running at the proper address within your mmap()ed /dev/mem space, twiddle the magic bits in CR0 (PE and PG) and you're in real mode. Note that the except sequence you should execute to change PE and PG is a bit complicated. For example, once you change PE, you'll be running at the physical equivalent of whataver linear address is in the PC, so it's a bit tricky. Also, I don't kno if you get one instruction (i.e. a jmp) before the PG change takes effect, or if you immediately drop into real mode. And that's just a skeleton. Now, are you absolutely *sure* vm86 mode won't do twhat you want? It's much easier. Pardon the short lines, but I'm working on a 100 column screen and am not quite sure where the 80 column mark is. -- _g, '96 --->>>>>>>>>> gt8134b@prism.gatech.edu <<<<<<<<<--- CompSci ,g_ W@@@W__ |-\ ^ | disclaimer: <---> "Bow before ZOD!" __W@@@W W@@@@**~~~' ro|- archie@cory.EECS.Berkeley.EDU (Archie Cobbs) writes: >A company I've worked for is interested in the possibility >of developing an ISDN card for the PC... I suggested that >writing a simple (?) device driver for Linux would be the >quickest/easiset way to test and play with it :-) I have to be careful here as I'm working with a real PC isdn card. I'm still trying to convince the powers that be to let me do a Linux version but we are a bit busy.... > >Since it's still just an idea, the hardware folks were wondering >if people out there more experienced than me with the interplay >between hardware and device drivers had any specific suggestions >about how the hardware could be designed to make things easier, >especially in the context of an O/S like Linux. 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 unlike DOS it can't afford to poll all the time. Alan iiitac@pyr.swan.ac.uk ------------------------------ From: ron@draconia.hacktic.nl (Ron Smits) Subject: Re: LINUX FOR SUN Date: 1 Mar 1994 15:32:52 GMT Stuart Tener (stuart@jane.sas.upenn.edu) wrote: : LINUX FOR SUN (IPC, IPX, Classic, LX, etc.) : Does it exist? NO : where do I get it? NOWHERE.GONE.COM : what will it cost? Alot of blood sweat and tears : reply via email: stuart@jane.sas.upenn.edu : (310)-358-0202 : thanks -- Ron Smits ron@draconia.hacktic.nl Ron.Smits@Netherlands.NCR.COM /*-( My opinions are my opinions, My boss's opinions are his opinions )-*/ /*-( They might not be the same -*/ ------------------------------ From: bsa@kf8nh.wariat.org (Brandon S. Allbery) Subject: Re: Linux and X WordPerfect Date: Tue, 1 Mar 1994 17:15:45 GMT In article <2ktvff$cbc@bradley.bradley.edu>, guru@camelot.bradley.edu (Jerry Whelan) says: +--------------- | Other than the size difference, is there any technical reason why | Linux shouldn't just adopt the ELF format as the native binary format? +------------->8 No technical reason, but a logistical one: the GNU utilities don't handle ELF fully yet, so we'd be left without a linker and maybe an assembler. ++Brandon -- Brandon S. Allbery kf8nh@kf8nh.ampr.org bsa@kf8nh.wariat.org "MSDOS didn't get as bad as it is overnight -- it took over ten years of careful development." ---dmeggins@aix1.uottawa.ca ------------------------------ From: urlichs@smurf.noris.de (Matthias Urlichs) Subject: Re: Why not put cluster diffs in nominal kernel before 1.0? Date: 1 Mar 1994 09:41:52 +0100 In comp.os.linux.development, article , pclink@qus102.qld.tne.oz.au (Rick) writes: > > While cluster-08a on pl15a and pl15h boost the iozone performance on my > FD885+767Mb Seagate, fsck runs 3-4 times slower on a 75% full 200Mb > ext2 partition. Anybody else noticed this? > Yes. It's a bug in refill_freelist; it doesn't check whether enough buffers already are on the list. Since refill_freelist is called from read_block unconditionally, this results in a slowdown or a hangup. The fix is rather simple. --- linux.current/fs/buffer.c Thu Feb 24 19:08:48 1994 +++ linux.work/fs/buffer.c Sun Feb 27 10:49:57 1994 @@ -110,8 +112,8 @@ every 5 seconds. If you change the frequency of update, the time constant will also change. */ /* These are the min and max parameter values that we will allow to be assigned */ -static int bdflush_min[N_PARAM] = { 0, 10, 5, 25, 0, 100, 100, 1, 1}; +static int bdflush_min[N_PARAM] = { 0, 10, 16, 25, 0, 100, 100, 1, 1}; static int bdflush_max[N_PARAM] = {100,5000, 2000, 2000,100, 60000, 60000, 2047, 5}; /* @@ -600,6 +602,9 @@ int buffers[NR_LIST]; int i; int needed; + int isize = BUFSIZE_INDEX(size); + + if(nr_free[isize] > bdf_prm.b_un.nrefill * 2) return; /* If there are too many dirty buffers, we wake up the update process now so as to ensure that there are still clean buffers available -- Even the night must fail When light sleeps in the eyes, When dark becomes dark on dark And into darkness dies. Soon the eye dissolves, Perplexed by the teasing night, Into a stillness of the heart, A fable of fallen light. -- Matthias Urlichs \ XLink-POP Nürnberg | EMail: urlichs@smurf.noris.de Schleiermacherstraße 12 \ Unix+Linux+Mac | Phone: ...please use email. 90491 Nürnberg (Germany) \ Consulting+Networking+Programming+etc'ing 42 Click here. ------------------------------ From: urlichs@smurf.noris.de (Matthias Urlichs) Subject: Re: ipcs utility, /proc page allocation Date: 1 Mar 1994 10:01:47 +0100 In comp.os.linux.development, article <2kimjb$obb@openwx.networx.com>, bonn@badger.networx.com (David Bonn) writes: > > (2) Check for overrun and fail if we overrun. This is likely to be > tedious to detect. I'd rather just make sure it won't happen. What's tedious about it? int info_ipcs(char *arg, int len) { int totallen = 0; int buflen; char buf[100]; int error = verify_area(VERIFY_WRITE,arg,len); if(error) return error; while(whatever) { buflen = sprintf(arg,"whatever\n",what,ev,er); if(buflen < len) return -ENOSPC; /* or whatever */ memcpy_tofs(buf,arg,buflen); totallen += buflen; arg += buflen; len -= buflen; } return totallen; } No malloced buffers, artificial limitations, or similar nonsense. -- Comitas comitatum, omnia comitas. -- Professor Charles P. Issawi -- Matthias Urlichs \ XLink-POP Nürnberg | EMail: urlichs@smurf.noris.de Schleiermacherstraße 12 \ Unix+Linux+Mac | Phone: ...please use email. 90491 Nürnberg (Germany) \ Consulting+Networking+Programming+etc'ing 42 Click here. ------------------------------ From: us292121@bulldog.mmm.com (Dean Junk) Subject: Re: Help! GCC errors Date: 1 Mar 1994 17:07:19 GMT Mitchum DSouza (m.dsouza@mrc-apu.cam.ac.uk) wrote: : Dean Junk: : : | I just upgraded to .99.15 kernel from .99.14 kernel along with the : | following libraries: : | : | ld.so.1.4.3 : | libc.so.4.5.21 : | libm.so.4.5.21 : | : | and the following tar archives: : | : | image-4.5.21 : | inc-4.5.21 : | : | 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! : | : | System setup: : | : | 386DX-40 AMD : | 16MB Memory : | 2-250 WD IDE HD's : | Mitsumi CD-ROM : | SB Multi-CD 16 card : | Conner 250MB Tape Backup : | 3.5" Floppy : | 5.25" Floppy : | 14400 Pract Per modem : | Linux .99.15 from sunsite.unc.edu : | gcc 2.4.5 : | xfree 2.0 : : Do one of the following. : : 1) Read the library release notes TO THE LETTER - EVERY SINGLE SENTENCE. : 2) Read the GCC-FAQ before asking GCC related queries. : : Mitch The solution is to delete the libgcc.sa library since it is a carry over from a previous release. Thanks to all who responded by e-mail! -- Dean Junk "An ounce of perception, a pound of obscure" Internet (dpjunk@mmm.com) --RUSH ------------------------------ From: ken@darwin.mbb.sfu.ca (Ken Clark) Subject: Re: Linux and X WordPerfect Date: Tue, 1 Mar 1994 22:24:49 GMT In article eric@tantalus.nrl.navy.mil (Eric Youngdale) writes: >In article <2ktvff$cbc@bradley.bradley.edu> guru@camelot.bradley.edu (Jerry Whelan) writes: >> Other than the size difference, is there any technical reason why >>Linux shouldn't just adopt the ELF format as the native binary format? >>(Debugging C++?) > > Yes. The GNU as and ld do not support ELF shared libraries. No one is >working on this right now, so it is not at all clear when this will become >available. I thought that the folks at Cygnus were working on this. Wasn't there the implication in the last release of gld that ELF would be readly RSN, or was that just a wishfull dream on my part? I thought gas allready supported ELF sort of. Right now this seems to be what everyone is holding their breath for (Linux, FreeBSD, BSDI BSD/386). I think that a working ELF gld would also allow a completely free development system (libs and all) for Unixware and friends, no? Eric, you mention that finishing ELF for gld would be a daunting task (which terrifies me, since there seems to be no limit to Eric's talents at writing Linux kernel code). What exactly needs to be done to gld and the GNU BFD to get this working? I thought that BFD was invented exactly to make such a job "easy" (obviously not :-). - Ken ------------------------------ From: rob@pe1chl.ampr.org (Rob Janssen) Subject: Re: Help! GCC errors Date: Tue, 1 Mar 1994 17:00:00 GMT Reply-To: pe1chl@rabo.nl In <2kuitm$or1@dawn.mmm.com> us292121@bulldog.mmm.com (Dean Junk) writes: >I just upgraded to .99.15 kernel from .99.14 kernel along with the >following libraries: > ld.so.1.4.3 > libc.so.4.5.21 > libm.so.4.5.21 >and the following tar archives: > image-4.5.21 > inc-4.5.21 >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! >System setup: > 386DX-40 AMD > 16MB Memory > 2-250 WD IDE HD's > Mitsumi CD-ROM > SB Multi-CD 16 card > Conner 250MB Tape Backup > 3.5" Floppy > 5.25" Floppy > 14400 Pract Per modem > Linux .99.15 from sunsite.unc.edu > gcc 2.4.5 > xfree 2.0 User setup: Did not read the instructions Rob -- ========================================================================= | Rob Janssen | AMPRnet: rob@pe1chl.ampr.org | | e-mail: pe1chl@rabo.nl | AX.25 BBS: PE1CHL@PI8UTR.#UTR.NLD.EU | ========================================================================= ------------------------------ From: ewt@sunSITE.unc.edu (Erik Troan) Subject: Re: undefined symbols in modules Date: 1 Mar 1994 18:54:03 GMT In article <2kv386$mud@smurf.noris.de>, Matthias Urlichs wrote: >> >You gotta get the list of symbols from _somewhere_. Since it doesn't make >sense to list all kernel symbols in the kernel itself (too big) and you >shouldn't depend on an external file like /vmlinuz (compressed, you don't >know that that's the kernel that got loaded), you need a small list of >exported symbold compiled into the kernel. > This problem will probably solve itself. As more people write modules, Linus will get a better idea what symbols need to be made available to modules. I don't think it will take that long for the list of symbols to become pretty complete. Erik -- ======================================================================== "Could I bend yer ear fer a tick?" ewt@sunsite.unc.edu = Erik Troan sasewt@unx.sas.com - Strictly Ballroom ------------------------------ From: behe@inasys.uucp (Bernd Hentig) Subject: ISDN support for Linux -> status ? Date: 1 Mar 1994 10:48:32 +0100 Hi, we're currently using a DOS router (KA9Q/PCROUTE/PAPI) to communicate to the outside TCP/IP world and it works great. However, since BSD386 began supporting ISDN drivers since Version 1.0, I wonder whether this has already been done for Linux. I remember there was some noise about this item months ago, so if someone could mail me the current status of this project or give me some hints on the appropriate mailing list ? Thanx Bernd -- Bernd Hentig | Email behe%inasys.de@Germany.Eu.net inasys GmbH | Phone Voice (Germany) (0)228 5205 451 FAX ~ 100 D-53121 Bonn | Germany | ======== INTERNET - Gateway to Infinity ======== ------------------------------ From: behe@inasys.uucp (Bernd Hentig) Subject: COFF / iABI support with linux 0.99p15 ?!? Date: 1 Mar 1994 10:43:23 +0100 Hi, I just installed 0.99p15 on top of SLS 1.04 (yes, I know the libs are quite old and lilo doesn't cooperate with AMI BIOS/DOS in this version). To my great surprise I found that there finally was support for the COFF a.out included in the kernel, but the iABI to the system calls was not supported yet. I tried to start some programs ftped from an SCO machine nearby and all I got was "iSBC not supported yet" or some message like this. Is there already a beta (or even alpha) version of iABI available ? I didn't find anything in the ftp archives about it (not even docu), so could someone please inform me about the status of the iABI / COFF project ? Please mail me, I don't want to waste too much bandwidth in this group as it's already big enough. Bernd P.S. I'm quite willing to spend some time testing iABI; however I can't promise I'll find the time to add any codes or bug fixes. -- Bernd Hentig | Email behe%inasys.de@Germany.Eu.net inasys GmbH | Phone Voice (Germany) (0)228 5205 451 FAX ~ 100 D-53121 Bonn | Germany | ======== INTERNET - Gateway to Infinity ======== ------------------------------ ** 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 ******************************