Return to BSD News archive
Newsgroups: comp.unix.bsd Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!haven.umd.edu!uunet!acd4!iedv7!smm From: smm@iedv7.acd.com ( Steve McCoole ) Subject: BSDI BSD/386 Ported Software List 02/02 Message-ID: <1993Apr8.015905.22562@acd4.acd.com> Followup-To: comp.unix.bsd Keywords: BSD, PC, Available Software, Patches and Enhancements Sender: news@acd4.acd.com (USENET News System) Organization: Applied Computing Devices, Inc., Terre Haute IN References: <1993Apr8.015726.22485@acd4.acd.com> Date: Thu, 8 Apr 1993 01:59:05 GMT Lines: 1032 ======================================================================== Subject: Section 3. Patches or enhancments for BSDI systems. Date: Mon Apr 05 19:00:29 EST 1993 ======================================================================== [ Admin Note: Most (All?) of the patches and extensions listed here are available on BSDI.COM in one of the download directories or in one of the mailing list archives. Some of the early listings are probably only available from one of the mailing list archives. Notes for beta level stuff have been deleted. -SMM ] ----- (GAMMA) A version of the Logitech busmouse driver has been created by Erik and is available. erik@eab.retix.com (Erik Forsberg) ----- Patches for gcc2.2.1 compilation errors. block@gmf.eds.com (Jon Block) ----- Patches for gnu linker bug that shows up with g++. donn@bsdi.com (Donn Seeley) ----- Patches for pbmplus compilation on bsdi. John T Kohl <jtkohl@kolvir.elcr.ca.us> ----- Patches to allow you to build libg++-2.2. You may have configure problems using /bin/sh, I used bash and did a 'configure bsdi'. block@gmf.eds.com (Jon Block) ------ Filehandles patch for perl. Required for perl man package to work. block@gmf.eds.com (Jon Block) ------ Patches to allow the SoundBlaster driver posted to alt.sources work under BSDI. John T Kohl <jtkohl@kolvir.elcr.ca.us> (GAMMA) Patches to John's patches to allow the SoundBlaster driver to work under GAMMA release. Steve McCoole <smm@acd4.acd.com> ------ Another set of patches for gcc-2.2.2. Roland Wilcher <raw@ushiva.ncoast.org> ------ Patches to tar so that it does not require the minus in front of options. John T Kohl <jtkohl@kolvir.elcr.ca.us> ------ Posting of SLIP scripts and experiences in setting it up. "Lee A. Butler" <butler@brl.mil> ------- (GAMMA) Fixes to beta and gamma db(3) bugs. Robert Willis <rwillis@bbn.com> ------- Patches to make elm mailer compile by Brian Litzinger. Sent by Marc. Marc G. Frank <mgfrank@avernus.com> ------- Gcc 2.2.2 fix for recursive code generation bug. Taavi Talvik <taavi@abc.postimees.ee> ------- (GAMMA) Fix to boot stopping while updating the LEDs on the keyboard. This fix was for beta and gamma. Tom Markson <tom@usblues.rt.com> ------- (GAMMA) Fix to wangtek tape driver to allow it to read more than one archive off of a tape. Patch is for beta and gamma. Tom Markson <tom@usblues.rt.com> ------- (GAMMA) Patch to correctly set paths.el for emacs in beta and gamma. composer@beyond.dreams.org (Jeff Kellem) ------- (GAMMA) Fix for getloadavg in libkvm for not closing file descriptors. This fixes the bug mentioned in the gamma release notes and takes care of the sendmail problem. composer@beyond.dreams.org (Jeff Kellem) ------- (GAMMA) Fix for monop that makes the random number generator work properly. Tom Markson <tom@usblues.rt.com> ------- (GAMMA) Filters and setup to support postscript printers under gamma. erik@eab.retix.com (Erik Forsberg) ------- (GAMMA) Fix for sort dumping core on "sort +3". composer@beyond.dreams.org (Jeff Kellem) ------- (GAMMA) Fix for uucp problem of SEGV when a "from" argument is not specified. Tom Markson <tom@usblues.rt.com> ------- (GAMMA) Fix to the new uucp from UUNET to recognize entries in the CONFIG file. Tom Markson <tom@usblues.rt.com> ------- (GAMMA) Another fix to the new uucp from UUNET. Eric Ziegast <ziegast@uunet.uu.net> ------- (GAMMA) Fixes to xload to make it work under BSD/386 gamma (0.9.1). composer@beyond.dreams.org (Jeff Kellem) ------- Agetty a replacement for getty on machines running SysV, SunOS or BSDI BSD/386. It supports hardware handshaking, bidirectional line support and other stuff. brian@apt.bungi.com (Brian Litzinger) ------- Perl scripts for starting up slip automatically. Curtis Villamizar <curtis@ans.net> ------- Su program, sudo, that supports the -c option. brian@apt.bungi.com (Brian Litzinger) ------- Attached is a uuencoded tar.Z of vgaset, a program for interactively sizing screens. It can be run from another terminal and is thus usable even if the display is not readable. It outputs a mode line suitable for inclusion in Xconfig. Documentation is included. uunet!lemis!grog (Greg Lehey) ------- Yagetty. Yet Another getty program with handling for dial-ins and dial-outs on the same port. uunet!lemis!grog (Greg Lehey) ------- (GAMMA) Recently I purchased a Longshine LCS-8880 el-cheapo 4-port serial card from Altex Electronics for about $75 plus shipping. The card has a single DB37 on the back panel with a cable that fans out to four male DB25 connectors, has socketted 16450-style UARTs (easy to swap out for 16550's if needed), and in general looks like pretty solid hardware. It supports two address ranges, 0x1a0-0x1bf, or 0x2a0-0x2bf in the 'enhanced' mode where all four ports share one interrupt line, which can be 2/9,3,4,5,6,7. bdale@gag.com (Bdale Garbee) ------- (GAMMA) We seem to get a lot of requests to look up ASC/ASCQ codes... I'll append a patch to the scsicmd sources below which will let you extract an English translation of these values like this: % scsicmd -a 0x14/0 recorded entity not found % Scsicmd appears in gamma; if you don't have gamma and you want scsicmd, ask Jeff for a copy. donn@bsdi.com (Donn Seeley) ------- (GAMMA) Patch for problems with floating point instructions in rundos envirionment. karels@bsdi.com (Mike Karels) ------- (GAMMA) I can't live without my 8 port serial card, so I quickly hacked gamma 0.9 com.c to support it with my generic multiport modifications. While testing the bidirectional line usage, I discovered a bug, at least in my opinion, in the DCD detection (fix included). This version isn't as robust as the latest stuff I had for beta 0.3.3, but it works. Also, something is interfering with the bidirectional line usage. (Sorry about posting this to both gamma and beta lists, but I wasn't sure if a gamma list exists) You have to add something along the lines of the following to your config file: # COM ports # APT HSS08 multi-port serial card device com0 at isa? port 0x100 device com1 at isa? port 0x108 device com2 at isa? port 0x110 device com3 at isa? port 0x118 device com4 at isa? port 0x120 device com5 at isa? port 0x128 device com6 at isa? port 0x130 device com7 at isa? port 0x138 The HSS08 shares one IRQ for all the ports, so I get warnings about conflicting IRQs from the autoconfigurator during boot, but it works. I also added: stty -f /dev/com0 clocal stty -f /dev/com1 clocal ... to my /etc/rc.local, so things would stop blocking unexpectedly. Brian Litzinger <brian@apt.bungi.com> ------- (GAMMA) The recently completed support for the Panasonic cdrom audio is now available by anonymous ftp from bsdi.com:/sw/gamma/progs/libcdrom.tar.Z. This tar archive contains the sources and objects for the library, plus a binary for cdctl(1). Besides the Panasonic drive, it includes support for the Toshiba 3201B SCSI, the Mitsumi ISA, and any SCSI-2 drive. Pace Willisson pace@bsdi.com ------- (GAMMA) An updated MH is available at bsdi.com in the /bsdi.sw/gamma/progs directory. It is the LBL-UCB MH 6.5.1 version. Tony Sanders <sanders@austin.bsdi.com> ------- Here is the diff to add automatic file decompression to more, which means that if you compress your man pages, that more will automatically decompress them as they are displayed to you by "man". File: bsdi.com:compressed_more.patch. webb@napa.telebit.com (Bill Webb) ------- Here is a portion of my fdisk program which prints out lines for /etc/mtoolsrc written in C++. This file is in File: bsdi.com:fdisk.mtoolsrc.prog. Paul Borman <prb@cray.com> ------- File and instructions on how to compile and get gcc 2.3.2 going under BSDI. This file is under /bsdi.sw/sw/gnu/gcc2.3.2.compile on File: bsdi.com:gcc2.3.2.compile Bert Driehuis <DRIEHUIS@rulfsw.leidenuniv.nl> ------- Instructions on how to install BSDI on a notebook computer that did not have an ethernet interface. This file is stored as File: bsdi.com:webb.install.notebook.noether. webb@napa.telebit.com (Bill Webb) ------- (GAMMA) Shell script to build symboic link trees for compiling off of CD-ROM. File: bsdi.com:cdlink and cdlink.patch01. Paul Borman <prb@random.cray.com>. ------- (GAMMA) The following are my corrections to 4 math functions missing from the gamma release. I have tested erf() and erfc(). I have not tested drand48 and srand48. File: File: bsdi.com:gamma.math.funcs.patch. bts!bill@uunet.uu.net (Bill Hatch) ------- (GAMMA3) The new sed in gamma-3 fails both to compile and execute a substitute expression like 's/\\/\\\\/g' which is supposed to replace each back-slash with two back-slashes. File: bsdi.com:gamma3.sed.patch root@eab.retix.com (Erik Forsberg) ------- (GAMMA) Someone reported a problem with slip where they would get host unreachable errors even though the routing entries looked fine. I found a bug in slip that would cause this error if the line was not using hardware carrier detect, but instead used the clocal flag. Slip was checking carrier detect without checking clocal as well. The fix is appended. File: bsdi.com:host.unreachable.slip.patch karels@bsdi.com (Mike Karels) ------ (GAMMA) Man page describing the ISO9660 CD-ROM filesystem. File: bsdi.com:iso9660.man pace@bsdi.com (Pace Willisson) ------- (GAMMA) I find it quite annoying that stock ps reports the name of /dev/com1 to be co, the same as for /dev/console. The following 4 additional lines to print.c in the ps directory fixes that: Files: bsdi.com:ps.com.report.patch* Paul Borman <prb@random.cray.com> ------- (GAMMA) The distributed new sed is unable to compile regular expressions like this one (occuring frequently in GNU configure scripts, found this one from the gdb 4.7 configure) s/^[-a-z_]*=// The problem is with the first character in the [] character set. Traditional sed (BSD 4.3) allows a single - to be treated as a ordinary character. As sed uses the regex stuff in libcompat, I fixed it to allow this syntax in /usr/src/lib/libcompat/regcomp.c See the case that was removed by #ifdef notdef This problem also exists in the libc regexp version. But as I do not know what POSIX has to say about this syntax, I haven't bothered to fix it there too. File: bsdi.com:sed.dash.patch root@eab.retix.com (Erik Forsberg) ------- (GAMMA) Tcpdump patches: Patch 1 This is a real bug. If you have NE[12]000 Ethernet card or 3Com EtherLink III(ef?) or EtherLink Plus(ep?) card, you should better to apply this patch. Patch2 Yesterday, I posted my fix to here. Today, I found a much better way to fix this problem. If you already applied my previous fix, please restore files(both net/bpf.c and i386/isa/if_ne.c) and then apply the following fix. File: bsdi.com:patches/tcpdump.patch[1-2] Yoshitaka Tokugawa <toku@dit.co.jp> ------- (GAMMA) Another script to make a set of symlinks to compile off of a CD-ROM filesystem. File: bsdi.com:web.makelinks.script. webb@napa.telebit.com (Bill Webb) ------ (GAMMA) If you are running version 8.9 of Ingres on BSD/386, you may have noticed some problems with the locking deamon. The biggest problem, which I fixed, is that the ingreslock program is using older semantics for select. They used long's with select instead of fd_set's so select gets horribly confused. I assume this is a relic of the days when the second parameter to select was just an int. Anyway, I removed the bit fiddling and changed all of the manipulations to their correct FD_XXX counterparts. Plus, I initialized a table that seemed to need it (maybe it didn't, but it doesn't to hurt!) If you have picked up ingres89.tar.Z from bsdi.com or applied the Ingres.diff file sent there, just: cd /usr/ingres/source/support (or your source area/support) patch -p <this_file I have found a couple of other changes from the Lynix port. I'll be posting those very soon: File: bsdi.com:ingress.lock.driver.patches Tom Markson <tom@usblues.rt.com> ------- (GAMMA) Below is a shar archive with the necessary changes for BSDI. They has been tested on gamma 1 and gamma 3. I do not know if they will work on the beta release. To configure and compile smail, you will need to install a current version of bash as /bin/sh. The shipped version of /bin/sh (ash) will not work, and neither will older versions of bash. Also, at least on gamma 1, you will need to create the file mkdriv.perl by running mkdriv.awk through a2p (awk to perl). THe makefile shipped with smail3-28 will then know to do the right thing. File: bsdi.com:smail-3.28.patches bjorn@sysadmin.com (Bjorn Satdeva) ------- (GAMMA) Patches to compile libg++ version 2.3 under BSDI. File: bsdi.com:libg++-2.3.patches Bert Driehuis <DRIEHUIS@rulfsw.leidenuniv.nl> ------- (GAMMA) Third, I wrote a new getopt for perl. I was hoping a few people here might like to test it before I post it to comp.lang.perl, to sniff out any bugs I may have missed. What distinguishes it from the getopts provided with perl is that you can specify the variable the command-line option affects. Additionally, you can have simple flags, options that take single arguments, or options that take multiple arguments. File: bsdi.com:frank.perl.getopt mgfrank@avernus.com (Marc G. Frank) ------- (GAMMA) Patches to make ELM 2.4pl17 mailer compile under BSDI gamma. File: bsdi.com:elm2.4.pl17.patch nerd@percival.rain.com (Michael Galassi) ------- (GAMMA) These are changes to GDB 4.7 in order to support it for BSD/386 from BSDI. Unfortunately, because BSDI uses a new a.out format (magic number 0314) it is kind of difficult to merge it into the 386BSD support without pain. So these changes adds the configuration option 'i386-unknown-bsdi'. NOTE: Does not support kernel debugging! File: bsdi.com:gdb4.7.patch erik@eab.retix.com (Erik Forsberg) ------- (GAMMA) Here are my patches for gamma4. I didn't have to hack on xmix, just recompiled it from scratch (using gcc-2, not the gcc-1.xx which comes with the system, though). File: bsdi.com:sblast1.5.gamma.patches John T Kohl <jtkohl@kolvir.elcr.ca.us> ------- (GAMMA) I got tired of the autorepeating Caps/Num/Scroll Lock keys and tried to do something about it. The console keyboard handling is fixed easy enough, but the X server uses the raw scancodes from the keyboard and needs to be fixed as well. Since I use the ATI Graphics Ultra X server I can't fix this myself. A crude patch to disable autorepeating "Lock" keys in the console driver is included later in this mail. File: bsdi.com:gamma4.autorepeat.key.patch Hans Nasten <nasten@everyware.se> ------- (GAMMA) Several folks have run into problems with the new bidirectional modem feature; there are a few things that slipped through the cracks when that was integrated into gamma.4. We have a package of fixes that should make things work correctly with getty, tip, uucp and kermit, courtesy of Pace Willison. I have included the source patches in this message as a shar file; a complete package including pre-compiled binaries is available via anonymous FTP from BSDI.COM, bidir.tar.Z. If anyone needs binaries and cannot get them via FTP, contact polk@BSDI.COM for a uuencoded copy by mail. File: bsdi.com:bidir.tar.Z karels@bsdi.com (Mike Karels) ------- (GAMMA) Here's my pet peeve. When pressing ctrl-alt-del in the normal kernel it reboots (gasp)! I mean are we running MS-DOG here or what? :-) Here's my patch so it at least asks. Wasn't this the default before? Was there a reason for its removal? File: bsdi.com:gamma4.ctrl.alt.del.no.reboot.patch gray@antaire.com (Gray Watson) ------- (GAMMA) Here is a patch for the diskdefect command that fixes the above problem (diskdefect -a corruped the bad-block table if no new bad sectors were found). File: bsdi.com:gamma4.diskdefect.patch Tony Sanders <sanders@bsdi.com> ------- (GAMMA) The changes in the latest SCSI tape driver (gamma-4) that increased the size of all buffers used for a scsi mode sense command upto 128 was not such a good idea. It completely broke tape support for EISA 174x host adapters. The problem is that first a mode sense command is used to fill the buffer, then something is modified and the same buffer is then passed to the mode select function. But the length passed is sizeof buffer (always 128) instead of the ACTUAL length of the buffer. This causes a data underrun error on the 1740 host adapter that I use. File: bsdi.com:gamma4.esia.174x.tape.patch erik@eab.retix.com (Erik Forsberg) ------- (GAMMA) Aside from the fact that a number of programs will have to be recompiled, is there a reason why the following patch should not be applied? This allows /etc/fstab to have '#' comments and blank lines. File: bsdi.com:gamma4.fstab.comment.patch gray@antaire.com (Gray Watson) ------- (GAMMA) Here is an official patch for fstat.c in case anyone needs to use it. File: bsdi.com:gamma4.fstat.patch Tony Sanders <sanders@bsdi.com> ------- (GAMMA) Join as shipped with gamma/4 will SEGV if you try to use it. THe code was correct in gamma/3, but the fix must have slipped through the cracks. This fix will make join function as advertised: File: bsdi.com:gamma4.join.patch Tom Markson <tom@usblues.rt.com> ------- (GAMMA) We have had problems finding a spot for the ne2000. The problem in gamma4 is that it is blocked by the lp0/pe0 device on the standard LPT1 address of 0x378 (ne0 is 360-37f). If the parallel port is present, ne0 will not configure. The problem is that every reasonable address conflicts with some standard device/address. Here are the main current choices: 300 conflicts with wt0 320 conflicts with aha0 340 conflicts with mcd0 360 conflicts with lp0/pe0 The plan for production is to add an mcd0 configuration at 334, allowing ne0 to work at 340. I also plan to include an ne0 configuration at 320, as lots of people are using that (with no Adaptec 154x, of course). I have included diffs for the GENERIC config file in /usr/src/sys/i386/conf for anyone who would like to test this. Note that the ne0 config at 320 is not particularly safe if an Adaptec is present without the patches I posted for if_ne.c last week. (It doesn't appear that neprobe would hit the Adaptec registers, but I am not certain.) File: bsdi.com:gamma4.ne2000.generic.kernel.conf.patch karels@bsdi.com (Mike Karels) ------- (GAMMA) Here is the fix for NE-2000 clones that fail to configure with gamma4. The new code has been tested on the clones that failed and on Novell NE-1000 and NE-2000 cards. Let me know if anyone has cards that still fail. File: bsdi.com:gamma4.ne2000.patch karels@bsdi.com (Mike Karels) ------- (GAMMA) Here is a fix for at least part of the problem causing various crashes when using rundos and other programs that use the floppy drives. File: bsdi.com:gamma4.rundos.floppy.panic.patch karels@bsdi.com (Mike Karel) ------- (GAMMA) This is a patch for the problem with scsi tape drives that cannot be opened, printing the error message "st0: unsupported density selection". An earlier fix was posted that was incompatible with some Exabyte drives. This fix should work with all drives. Please let me know if there are still any problems after applying this patch to /usr/src/sys/scsi/st.c. File: bsdi.com:gamma4.scsi.tape.exebyte.patch karels@bsdi.com (Mike Karels) ------- (GAMMA) As distributed, the gamma.4 slattach program doesn't exit when it's slip line is droped. The problem is that the process waiting for HUP's has not issued a TIOCSCTTY to make it's controlling tty the slip line; hence no HUP's are delivered when the carrier detect drops. The following patch to /usr/src/sbin/slattach/slattach.c solves this problem: File: bsdi.com:gamma4.slattach.patch Brad Huntting <huntting@glarp.com> ------- (GAMMA) I gave a possible fix for slip hangs to a couple of people, but haven't gotten any feedback on it. I would be interested to know if it solves the problem you had. To the best of my knowledge, the problem is actually in hardware, and it happens only with some com boards. The com port sometimes loses a transmit-complete interrupt, and exhaustive testing indicated that software never got the interrupt and it was not pending. However, ISA-bus interrupts are not exactly robust. File: bsdi.com:gamma4.slip.hang.patch karels@bsdi.com (Mike Karels) ------- (GAMMA) Several people have had a problem that resulted in a kernel protection fault in the low-level handler for irq 7. In several cases this happened when booting from a floppy. After staring at the instruction that causes the fault for a while, I finally realized that I got bit by segmentation; the failing instruction tries to modify kernel data space without having switched to a kernel data selector. The following patch to /usr/src/sys/i386/isa/icu.h should fix the problem. File: bsdi.com:gamma4.stray.interrupt.panic.patch karels@bsdi.com (Mike Karels) ------- (GAMMA) Here are the diffs to make Taylor 1.03 [ftp.uu.net:/networking/uucp/taylor/taylor-uucp-1.03.tar.Z] work with the serial device locking scheme under Gamma.4. File: bsdi.com:gamma4.taylor.uucp.patch Alan B Clegg <abc@concert.net> ------- (GAMMA) The "hayes" entry for tip didn't seem to like my telebit too much, so I took the telebit dialer from tip/slip (available from gatekeeper.dec.com, I think) and added it to BSD/386 tip. The shar file below contains: 1. Diffs to the Makefile and acutab.c to add the telebit 2. The telebit dial program. Files: bsdi.com:gamma4.telebit.tip.patch[1-2] tom@usblues.rt.com ------- (GAMMA) Here are two fixes to gamma/4 (gee, am I the first?): 1. tex doesn't work as shipped. cmb10.tfm is missing from /usr/contrib/lib/tex/fonts/cmfonts. I built it by going into /usr/src/contrib/tex/fonts/cmfonts and doing the following: /usr/contrib/bin/mf '\mode:=localfont;batchmode;mag:=magstep(0);input cmb10' 2. getpar.c is corrupt in /usr/src/games/trek. Here is the replacement: File: bsdi.com:gamma4.tex.trek.patch Tom Markson <tom@usblues.rt.com> ------- (GAMMA) Several people have had problems with crashes when modems produced character input as they were being opened. Oddly enough, this problem did not show up before gamma4, although the bug seems to have been present since the alpha release. Someone proposed two different fixes earlier; I decided to use this one, as it does not require changes to existing tty hardware drivers or conventions. This patch should be applied to the file /usr/src/sys/kern/tty.c. File: bsdi.com:gamma4.tty.crash.patch karels@bsdi.com (Mike Karels) ------- (GAMMA) I've always mourned the days of a simple 1 LOGFILE uucp system. The more /var/log/uucp/uucico/hostsite stuff is big time overkill on my system with 1 or 2 uucp sites. Of course, if I were running SCO, I'd have to live with it. But, since we all bought BSD/386 so we'd have the source, changes are possible. File: bsdi.com:gamma4.uucp.1logfile.patch tom@usblues.rt.com ------- (GAMMA) Several people have had problems with the gamma4 driver for wt tape; either the device did not configure, or the drive would hang on the first use. Both problems have been traced to the same bug. Here is the fix for /usr/src/i386/isa/wt.c. The patch also adds some debugging code, and just in case, forces the irq to the default if the drive does not interrupt when probed (which should not be necessary now, but we will probably include in the next release). File: bsdi.com:gamma4.wt.tape.patch karels@bsdi.com (Mike Karels) ------- There was a message on comp.unix.bsd a while back with a DOS program that determined the real disk parameters on IDE disks. It works on at least some ESDI and probably most other modern controllers as well. It prints the CMOS geometry, asks the drive for its current geometry and prints it, then resets the controller and asks again for the native geometry. It then forces DOS to be rebooted, as the controller is not in synch with the software. We may incorporate something like this into disksetup in the future, but it is a bit tricky (not all controllers necessarily support this). When I ran it, I discovered that one of my drives was using a mapped geometry, although I hadn't realized it. The files aren't large, so I will include them here. The shar contains the README file, the sources, and a uuencoded DOS executable. This will not work under rundos. File: bsdi.com:disk.geometry.dos.prog karels@bsdi.com (Mike Karels) ------- (GAMMA 3) [One of the most prolific kernel fixers I've seen on the list is Yuval Yarom <yval@cs.huji.ac.il>. Several kernel patches against gamma 3 have been sent to the list. I will try to list each patch and the Subject line of the message with the name that the file will have on the archive. Thanks Yuval! SMM] gamma3.boottrap.patch: Subject: (Gamma 3) traps on boot. gamma3.execve.lock.patch: Subject: (Gamma 3) deadlock in execve gamma3.kernel.size.patch: Subject: (Gamma 3) solution to kernel size problem gamma3.memleak.deadlock.patch: Subject: (Gamma 3) Kernel memory leak in exec. gamma3.mmap.range.patch: Subject: (Gamma 3) Wrong range check in mmap. gamma3.msync.patch: Subject: (Gamma 3) panic in msync gamma3.munmap.patch: Subject: (Gamma 3) Wrong range check in munmap. gamma3.physio.patch: Subject: (Gamma 3) Wrong address validation in physio. gamma3.stack.alloc.patch: Subject: (Gamma 3) Stack allocation problem gamma3.vmmap.anon.patch: Subject: (Gamma 3) Wrong caching of anonymous memory regions gamma3.pagermap.wait.patch: Subject: (Gamma 3) Deadlock in page daemon gamma3.security.patch: Subject: Fix for a security hole in Gamma 3. gamma3.sigsuspend.patch: Subject: (Gamma 3) Sigsuspend(2) problem. gamma3.socket.patch: Subject: (Gamma 3) Several bugs in receive from socket. gamma3.unp_discard.patch: Subject: (Gamma 3) Wrong access rights handling in unix domain sockets. ------- (GAMMA 4) The ash bug which almost everyone seems to report is the one that makes ash require a semicolon before 'do' in a 'for' loop if there is no 'in' clause. Without the 'in' clause, there is no ambiguity about whether the 'do' must be a keyword, so this seems to be reasonable, if inconsistent, behavior. Here are some examples: File: bsdi.com:gamma4.ash.for.patch donn@bsdi.com (Donn Seeley) ------- (GAMMA 4) At least two of my audio CD's induce an infinite loop in mitsumi_probe(). Release: Gamma.4 (0.9.4) Repeat-By: Insert the disk "Claude Bolling Suite for Flute & Jazz Piano Trio, No. 2) (CBS records, catalog MK 42318). Run cdctl. Watch it spend eternity frobbing with the disk. Fix: I don't know what this test in the code is for, but commenting it out gets rid of the problem. File: bsdi.com:gamma4.cdrom.infinite.loop.patch John T Kohl <jtkohl@kolvir.blrc.ma.us> ------- (GAMMA 4) Description: The floppy controller is returning EINVAL instead of ENOSPC when EOF is reached making multi-volume tar archives impossible. Release: 0.9.4 Repeat-By: tar -cMvf /dev/fd0 something-larger-than-the-floppy Fix: The below *seems* to work. I've not tested it much and don't have time. File: bsdi.com:gamma4.fd.ENOSPC.patch gray@antaire.com (Gray Watson) ------- (GAMMA 4) I have the following problem with low density floppies: I cannot access 360 or 720 KB floppies in my 1.2 MB 5.25 drive and 720 KB floppies in my 1.44 MB 3.5 drive (I use the floppycontroler from the adaptec 1542b). Low density access works, if I use a typed device node, but not when I want to use automatic density detection. The autodetection works for HD-floppies, but fails for others. So the driver uses the first entry of a drive for low density floppies (the HD entry) and the floppy access fails. I dont know why the autodetection fails, but the following workaround works for 720 KB in 1.44 drive and 360 KB in 1.2 drive. If you comment out the line, which sets the first entry as default, the driver uses the last entry as default, if autodetection fails. This is 360 KB for a 1.2 drive and 720 Kb for a 1.44 drive. This Workaround does not work for 720 KB in a 1.2 MB 5.25 drive, because the detection fails and the 360 KB entry is used as default. HD floppy access works also with this code, because the autodetection does not fail for HD-floppies. File: bsdi.com: gamma4.floppy.lowdensity.workaround "Christoph Koppe (Inf4 - hiwi)" <chkoppe@immd4.informatik.uni-erlangen.de> ------- (GAMMA 4) ftime(3) uses data returned by gettimeofday(2) to fill out the timeb structure it is passed, this is a problem because per gettimeofday's man page it does not fill out the timezone part. The attached ftime.c does a call to localtime to get the added information. Just replace /usr/src/lib/libcompat/ftime.c with what folows. File: bsdi.com: gamma4.ftime.patch nerd@percival.rain.com ------- (GAMMA 4) This m4 input: define(msg, printf $1) msg(("fmt", arg)); results in: printf ("fmt" arg); Notice the dropped comma between "fmt" and arg. This is the patch for /usr/src/usr.bin/m4/main.c. This patch is going to be dropped into the master source tree. Anyone that uses m4 might want to try this to be sure it doesn't break anything else. I've run our m4 tests and everything works ok. File: bsdi.com: gamma4.m4.patch sanders@bsdi.com ------- (GAMMA 4) Here are the changes for the console driver to support multiple console screens, done by Vadim Antonov. As reported earlier, this driver allows multiple character-mode console sessions (default 8, can be changed with "options NPCSCREENS=12", etc). X and rundos can still be used, but they take over the entire console while running and it is not possible to switch to another screen until they exit (or rundos is suspended). Output to any of the virtual screens while X or rundos is using the display will be discarded. A shar file below includes the patches for the gamma4 console driver, /usr/src/sys/i386/isa/pccons.c. File: bsdi.com: gamma4.multiconsole.patch karels@bsdi.com (Mike Karels) ------- (GAMMA 4) > > The termcap entries provided with the new console driver 'break' less, vi, > > emacs, and probably lots of other stuff too. > The new termcap entry for ibmpc3 is incorrect. Correct escape > sequence for Insert Line should be al=\E[L. Oops. You are right. Anyone who installed/installs the new driver should use this corrected termcap. I'm not sure how that happened, but obviously it happened after I had done most of my testing. File: bsdi.com: gamma4.pc3termcap.patch karels@bsdi.com (Mike Karels) ------- (GAMMA 4) > At the end of the script I get this message > stty: stdout appears redirected, but stdin is the control descriptor > Is this normal? There is a little bug in stty that causes this message to be printed even when the "-f" flag is used. It's harmless but if it annoys you here's the patch: File: bsdi.com: gamma4.stty.message.patch huntting@misc.glarp.com ------- (GAMMA 4) Ain't this insidious? This fix seems to make xview work reasonably well (enough so that I can now work on the CDROM bugs in workman rather than xview bugs in workman...watch this space for more details!). Description: syscall(2) doesn't work right. It leaves $sp off by a word, which can lead to disaster when it's called repeatedly (as it is in XView3.0). Release: Gamma.4 (0.9.4) Repeat-By: Write a program that has a local variable on the stack; initialize it, then call syscall() repeatedly. Note that the value eventually gets trashed when the stack works its way up over the variable. File: bsdi.com: gamma4.syscall.patch John T Kohl <jtkohl@kolvir.blrc.ma.us> ------- (GAMMA 4) Seems xman has a problem. It appeared to have a problem with the MANPATH, but in reality it was with the "cat" subdirectories. In looking at the source, it appears that man.c requires MAN to be set to the subdirectory prefix which in our case is "cat". However, vendor.h sets MAN to "man". File: bsdi.com: gamma4.xman.patch ektron1!wutt@uunet.uu.net (William Wuttke) ------- If any of you have had to SLIP into a site with a terminal server you know how difficult it can be to get the suplied startslip program to function with anything but the "standard" bsd login sequence. The enclosed shar file has a perl script which can be "easily" modified to deal with a wide range of send-expect login sequences. It does not have a chit-chat config file, you must edit the script itself. I've tried to make it easy to modify, but it's still perl (the worlds most pathalogically unreadable computer language)... The present config is used to dial in to the Annex where I work. I sugest placing the "slipup" script in /usr/local/bin, and the slipup.conf in /etc (if you place slipup.conf elsewhere, be shure to change the pathname in the "slipup" script that reads it). File: bsdi.com: huntting.slipup.perl Brad Huntting <huntting@glarp.com> ------- (GAMMA 4) tmpnam() will return a filename in the form of "/var/tmp//tmp..." because it assumes P_tmpdir (from stdio.h) does not end in a slash (which it does). For those who care here is a "graceful" patch -- or you can remove the extra / from tmpnam.c. :-) File: bsdi.com: tmpnam.patch gray@antaire.com (Gray Watson) ------- (GAMMA 4) Fix to the w_char redefinition problem in Xlib.h. File: bsdi.com : Xlib.h.patch.Z rd@pixie.aii.com (Robert D. Thrush) ------- (GAMMA 4) I'd like to ask you to test the new version of VGA driver on your VGA cards -- it does some trickery to save/restore fonts and it is possible that some cards don't like that. The changes are supposed to fix the problem with restoring screen on exit from X on Trident and some other VGA cards. I also added code for loading user font. File: bsdi.com : antovov.vga.driver.Z avg@uunet.uu.net (Vadim Antonov) ------- (GAMMA 4) I have had enough requests for the new 3C509 driver that I will use multicast to reply. Here are the patches for /usr/src/sys/i386/isa/if_ef.c. They fix three problems: there was a possible hang if there were excessive collisions, some machines could not configure more than 1 card, and some machines could not configure the 3C509 on a reboot without a power-up. I have forgotten whether the 0.9.4.1 patches include this, but I think the answer is "no". A couple of folks asked about performance; I don't have performance numbers (or a 3C509!), but several reports put it substantially ahead of the other cards we support. At least one person asked about the card for which we just added support. It is the Intel EtherExpress 16. File: bsdi.com : ethernet.3c509.patch.Z karels@bsdi.com (Mike Karels) ------- (GAMMA 4) > I have a question about the return value from fflush. According to the > manpage, fflush will fail if the stream is not open for writing. > However, it fails when the stream is opened for reading and writing with > mode "w+". I thought I'd sent in this patch -- I'm pretty sure I did, but would need to search through my archives to double check. Just in case, I'll CC this to problems@bsdi.com, also. For those that want to fix it now, I include a patch to /usr/src/lib/libc/stdio/fflush.c below my signature. File: bsdi.com : fflush.patch.Z composer@beyond.dreams.org (Jeff Kellem) ------- (GAMMA 4) Check the 'speed' with stty. If it's 38,400 or higher, more gets confused. If you're using xterm, the simple answer is to drop the speed down to (say) 9600, since it makes no difference to anything else. Otherwise you need to patch more... File : bsdi.com : more.highspeed.patch donn@bsdi.com (Donn Seeley) ------- (GAMMA 4) Followup to msgs(1) problem (bug #001975) The fopen(fname, "a+") didn't actually work either like I thought it had. Soo...opening for read/write ("r+"), and then opening for write ("w") if that doesn't work (if the file doesn't exist) does the trick. I tested it running msgs(1) a few times, and removing my .msgsrc, and repeating the whole process a few times. I can't get it to fail anymore, and here is my resultant patch: File: bsdi.com : msgs.patch.Z From: tod@vision.intel.com (Tod Oace) ------- (GAMMA 4) Fix to allow boot without a keyboard In srt0.s, a call is made to kbd_reset. This call hangs unless the following patch is applied to /sys/i386/stand/kbd.c. The patch will put an upper bound on how long the code waits for the reset command to complete. File: bsdi.com : no_kbd_boot.patch.Z From: Robert Hagens <hagens@reston.ans.net> ------- (GAMMA 4) Another problem with sed. The gamma-4 sed failed the following sed script found in the brand new GROFF 1.07 release (psstrip.sed) #!/bin/sed -f # Strip a PostScript prologue of unnecessary comments and white space. /^%[%!]/b s/^[ ][ ]*// s/[ ][ ]*$// s/%.*// /^$/d s/[ ]*\([][}/{]\)[ ]*/\1/g The last line fails to compile, sed complaining about unbalanced []'s. This is due to compile_delimited not really knowing the traditionally rules about [] expressions (namely that inside a [s] expression, the first character of s is allowed to be ] and that \ has no special meaning. The following patch corrects this. I have a vague nasty feeling though that we should run sed through a complete set of tests after this change. I have done a few tests, but nothing VERY extensive. File: bsdi.com : sed.patch.Z From: root@eab.retix.com (Erik Forsberg) ------ (GAMMA 4) I found a problem in sort with a large file where it hits the end of the sort buffer, and because of the way trecheaders are allocated, bufend could be less than opos, so the memmove gets a negative length. Context diffs follow: File: bsdi.com : sort.large.file.patch.Z From: Kent.Peacock@eng.sun.com (Kent Peacock [CONTRACTOR]) ------- (GAMMA 4) Haven't seen mention of this, so i hope i'm not replicating something obvious. All that's needed to get tcsh 6.03 up and running on BSD/386 is one small change - BSD/386 defines BSD4_4 but doesn't have the "quad_t" type (yet;-)). [ I believe that tcsh is back in the normal distribution for 1.0 SMM ] File: bsdi.com : tcsh-6.03.patch.Z From: sms@wlv.iipo.gtegsc.com (Steven M. Schultz) ------- (GAMMA 4) Patch for "panic: vm_map_entry_create: out of map entries" Several folks have run into this panic recently. The direct cause of the panic is severe fragmentation of an internal kernel map, which we haven't been able to reproduce when we try to cause it. One way for it to happen is for the buffer cache to fill with lots of small files; a news feed or NFS mounts with small rsize/wsize have both been seen to do this under some circumstances. Here is a change for /usr/src/sys/vm/vm_map.c that alleviates the problem. If you have not seen this panic, you probably don't need the fix. File: bsdi.com : vm_map_entry_create.panic.patch From: karels@bsdi.com (Mike Karels) ------- ======================================================================== Subject: Section 4. BSDI Archive Sites Date: Mon Apr 05 19:00:29 EST 1993 ======================================================================== [ These are the sites that are currently willing to archive stuff for the BSDI system. They are numbered and described below. Please contact the person listed in the description with problems, not me. I won't be able to help you. ] 1. nic.funet.fi Please, if possible, ftp all binaries/sources/patches of BSD/386 to nic.funet.fi (128.214.6.100). There is plenty of disk space available and I'm willing to try to keep things in order there. The directory in nic is ~pub/unix/386ix/bsd386/incoming . I just got added to beta list, after being reading the idle alpha list for a while, so please tell me about the other archive sites containing bsd/386 material! We have had this directory in nic since we got the alpha release, but the only material I have had there is some xview3 patches. Send me a note at <antsu@nic.funet.fi> , if you upload something to nic, describing the bits you sent. Happy hacking, Antti-Pekka 2. ftp.unm.edu In addition, on the machine ftp.unm.edu is available via anon ftp. It has a directory called pub/BSDI-386 (to hopefully cut some of the confusion with 386BSD) and at the moment contains very little. Any patches relating to BSD/386 that I receive (via email or other means) will be place there. At the moment, I only have ~60 megs of space to deal with so I can't deal with too many full sources, but we are working on a 2 gig ftp server and when that becomes a reality, I don't think space will be a problem. Any patches posted to the beta list from this point on will appear there, if you have posted a patch to the list in the past, please mail it to me directly. Steve Freed <sfreed@ariel.unm.edu> 3. bsdi.com An archive is now available for programs, patches, information, and ports for BSD/386 systems. The archive is at bsdi.com [153.39.128.11] in the directories under "/bsdi.sw/sw". Please report problems with the site to mccoole@bsdi.com or smm@acd4.acd.com. Sorry it took me so long to get things going but I think that you will like it. The archive includes the old postings from the beta and gamma mailing lists and the new bsdi-users mailing list. Patches are on-line for the gamma release of BSD/386. The beta patches will be removed in a reorganization that will occur very soon. Also available are whole ported programs or patches to some well known packages to make them run under BSD/386. If you have software or patches that you would like to contribute to the archive, please drop me a message. Steve 4. ftp.uni-erlangen.de ftp.uni-erlangen.de:/pub/BSD/BSDI contains a mirror of the relevant/interesting files on bsdi.com. Dirk 5. ftp.dit.co.jp For the Japanese BSDI users, ftp.dit.co.jp:/pub/bsd386 contains a mirror of bsdi.com:/sw. -- toku -- Yoshitaka Tokugawa Internet: toku@dit.co.jp dit Co., Ltd. Yamada Bldg. 4F, 2-11-10 Iidabashi Phone +813-3239-7240 Chiyoda-ku, Tokyo, 102 Japan