Subject: Snapshot report - June 10th tar_files
To: None <current-users@NetBSD.ORG>
From: Alistair G. Crooks <>
List: current-users
Date: 06/14/1995 08:48:01
Snapshot report - June 10th tar_files

Actually, there weren't any tar_files, and so my systems are still on
June 3rd sources.  I think this was due to sun-lamp's move across the
USA.  At the time of writing (Wed Jun 14 08:43:45 PDT 1995) is actually, and so you can still
ftp from there.  CVS commits are taking place on

Other notable events over the last week and a bit are:

1. (Robert V. Baron) has made some mods available,
>         a D-Link DE-600 pocket ether driver from Mach 2.5/CODA
>          (note Mach 2.5 was a project at CMU and CODA is an
>           ongoing activity at CMU involved in file systems and
>           laptop research.)
>         a PCMCIA WaveLan driver from CODA -- not the Klemets Driver
>         an intel pcmcia lan driver using the 82595 chip from CODA
>         changes to pccons.c to exchange the capslock and control
>         changes to pccons.c to let arrow keys and F1,F2,F3 emulate
>          a mouse.
>         a frame buffer device (minor on /dev/mem) for X

2. Jesus M. Gonzalez Barahona ( has ported gnat, the GNU
Ada translator to NetBSD/i386 1.0:

>         Gnat-2.05, an almost complete Ada95 compiler is ready for 
> NetBSD-1.0/i386, available from (in NetBSD-ports/i386
> directory). It has now full Ada tasking support, based on Provenzano's
> threads.

3. (Stefan Grefen) has updated his PCMCIA support:

> I've uploaded a new release of my pcmcia code for NetBSD to 
> and
> I raised the status to Beta test. The SCSI stuff is still alpha.
> This is the last release for NetBSD 1.0 the next one will be for -current.
> ...
> Devices known to work:
>     D-Link DE-650 and DE-650CT ethernet cards.
>     A IBM pcmcia ethernet card (also known as NE 4100 infomover).
>     These cards are NE2000 compatible ones.
>     The Linksys should be a DE-650 (or vice versa, I've seen this card
>       with different  Labels on it).
>     3COM  3c589B (should do 3c589 also) ethernet cards.
>     Megahertz Modem 
>     There is also a pratialy working Adaptec (Trantor) T460 driver.

Note that this is 1.0-based, the next Beta from Stefan will be for -current.

4.  A lovable guy by the name of Alistair Crooks (
has got almost complete support for Postgres95, the O-O database,
rolled back in by the maintainers (it's missing one file, which is
reported to be in the next Beta).  It's also been reported that
Postgres95 will run on NetBSD/sparc as well as the i386 platform, and
BSD/OS 2.0.  Mail him for more info.

5. (Juergen Hannken-Illjes) has implemented
bounce-buffers for the i386 port, specifically for Adaptec 1542
support.  Mail him for the mods (I have them, but haven't had a chance
to try them out yet).

6. (John Kohl) has updated his VM86 support for
the i386 platform.
> I sucked out ALL my changes relative a -current kernel tree, and put
> them into .

7.  I've seen numerous "I'm interested" messages on the java-porting
mailing list from people wishing to port java (Sun's/Gosling's new O-O
language) to various NetBSD platforms - has anyone actually done this? 
Has anyone got the sources?  What platforms are you aiming at?  All
information gratefully received.

8.  I still have local mods here to swap the CAPS_LOCK and Control
keys in pccons.c, compress manual pages (although I seem to have been
overtaken by events here), make rexec compile (in libcompat/4.3),
Kenneth Stailey's mount_portal fix etc. Hopefully, these will make
their way into -current sometime real soon now.

9.  Some people have been reporting problems with the generic kernel
on the i386 platform, which will screw up the display at boot time,
after it's probed for the 4th floppy.  The workaround for this is to
mount the floppy drive without a floppy in it, or any other command
which forces lots of output to the display.  The fix is to replace the
line in your kernel config file which says

fd*	at fdc? drive ?

with a line that says

fd0	at fdc0 drive 0

Of course, this assumes that you only have one floppy drive.

10.  As some people have noted, the i386 GENERIC kernel gets its
knickers in a twist if you posses an Adaptec 154?  and no BusLogic
card.  If anyone's really bothered by this, mail me, and I'll see
about creating a kernel for you.

11.  I must admit to getting well pissed off by a certain "gentleman"
on the FreeBSD -ports mailing list.  Apologies to anyone who was
caught in the crossfire about this.  He now has a place in my "prats"
file.  It's a good thing elm has filters.

12.  Notable changes that would have made it into the June 10th tar_files:
(This is my definition of `notable')

> othersrc/sup/{expand.c,scm.c,sup.1,sup.h,supcmeat.c,supcmisc.c,supcparse.c,
Changes to write ascii timestamps in the when files.
> sys/netinet/ip_mroute.c
Eliminate compiler warnings.
> src/sys/dev/isa/com.c
Add an input FIFO, and fix several flow control problems.  Based on code
from Felix A. Croes.
> src/etc/inetd.conf
The tftpd daemon has been changed to use setgid() & setuid() to
explicitly set the user and group IDs to non-priveleged values.  This
was done because the chroot() call used when the secure (-s) option is
used can only be done by the superuser.  This change now requires
tftpd to be executed by root.  So the inetd.conf entry has been
changed to start it that way.  I also added -s /tftpboot arguments, so
people who uncomment the tftpd entry without realizing it's security
ramifications won't open thier whole systems to unauthorized access. 
> src/gnu/usr.bin/ld{ld.c,ld.h,rrs.c,shlib.c} (and others)
Implement compiled-in library search paths: any `-L' argument
encountered while `dynamic link-mode' is on, is recorded in the
executable.  The run-time linker will search these paths after
$LD_LIBRARY_PATH but before the default set and those currently in the
hints file.  The feature is disabled whenever for set[ug]id programs.
> src/include{link.h,dlfcn.h}
Put dlopen() declarations in separate header file for compatibility.
> src/sys/dev/ic
dc21040 driver, from Matt Thomas
> src/lib/csu/{i386 and almost everything else}/Makefile 
Build run-time code for use in static executables, called `scrt0.o'.
> sys/arch/i386/pci/pci_machdep.c 
Use the memory mapping set up by the BIOS.
> src/sys/kern/sys_process.c 
If necessary, wake up the child during attach so we always get a SIGCHLD.
> src/share/mk/ 
make this work whether or not MANZ is defined.  also, clean it up a bit,
remove some extra cruft, and make the way compressed man pages are installed
a bit more sane.  (DON'T blow away uncompressed man pages of the same name,
if they're also there...)
> src/bin/sh/{trap.c,trap.h}
Ignore result of sigaction when setting traps. Traps will succeed even
on SIGKILL or SIGSTOP. This is what other bourne shells do. (suggested
by mycroft)
> src/sys/kern{kern_sig.c,kern_synch.c,sys_process.c}
Fix various signal handling bugs:
* If we got a stopping signal while already stopped with the same signal,
the second signal would sometimes (but not always) be ignored.
* Signals delivered by the debugger always pretended to be stopping
* PT_ATTACH still didn't quite work right.
> src/bin/sh/{eval.c,exec.c,exec.h,input.c}
Changed so that 'PATH=newpath command' works, instead of looking at the
old path. Synced input.c with vangogh.
> src/sys/arch/sun3/stand/netboot
Have netboot!

I'm well aware that most of this is i386 specific.  If there's anyone
out there who's staying -current on other platforms and would like to
contribute some things every week, please get in touch with me.

If there's anything I've missed out from this rambling mail, I'm sure
someone will let me know.  Similarily, if there's anything you think
I should be looking at, please let me know.

Alistair G. Crooks (			   +44 125 234 6377
Amdahl European HQ, Dogmersfield Park, Hartley Wintney, Hants RG27 8TE, UK.
[These are only my opinions, and certainly not those of Amdahl Corporation]