Subject: Snapshot Report - July 8th tar_files
To: None <current-users@NetBSD.ORG>
From: Alistair G. Crooks <>
List: current-users
Date: 07/11/1995 07:50:32
Management summary:
Agassi lost, after playing the best tennis I've seen for years.
Good ladies final, too.

For the rest of us:
[8th July 1995 tar_files]

1.  If you have options ISO in your kernel config file, you'll still
be unable to compile a kernel.  So the following diff will help you to
get a working kernel (it's a remnant of the LIST -> CIRCLEQ changes
that took place some time ago, and seems to be a file that got missed)

*** /usr/src/sys/netiso/tuba_usrreq.c   1995/07/10 09:46:08     1.1
--- /usr/src/sys/netiso/tuba_usrreq.c   1995/07/10 10:47:41
*** 136,143 ****
                } else {
                        inp = sotoinpcb(so);
!                       LIST_REMOVE(inp, inp_list);
!                       LIST_INSERT_HEAD(&tuba_inpcb.inpt_list, inp, inp_list);
                        inp->inp_table = &tuba_inpcb;
                        tp = intotcpcb(inp);
                        if (tp == 0)
--- 136,143 ----
                } else {
                        inp = sotoinpcb(so);
!                       CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, inp, inp_queue);
!                       CIRCLEQ_INSERT_HEAD(&tuba_inpcb.inpt_queue, inp, inp_queue);
                        inp->inp_table = &tuba_inpcb;
                        tp = intotcpcb(inp);
                        if (tp == 0)

[I don't even know if this is the right change, although I suspect it
is, but it makes a bootable kernel for me.  I'll send-pr this if ever
I get the time - agc].

2.  com.c 1.61 lets me run X again with a serial mouse.  Thanks to
Torsten Duwe for finding the bug.

3.  xlint has a small bootstrapping problem.  The best method is to cd
to usr.bin/xlint and flail around for a few minutes, typing "make" and
"make install" at random.  The more boring method is:

mkdir /usr/libdata
mkdir /usr/libdata/lint
(cd usr.bin/xlint ; make ; make install ; make ; make install)

4.  It's time we started a supported-devices list again, and not just
for the i386 port either, so please mail me if you have weirdo,
odd-ball devices and the driver that supports them (or even run of the
mill stuff, too):

To start:

Device			Driver		NetBSD		Contact/Notes
Adaptec 1542CF		aha1542		i386 1.0A	delete bt from GENERIC
Linksys NE2000 clone 	if_ed.c		i386 1.0A	bog standard
IBM 650MB M-O SCSI-2	standard sd	i386 1.0A
Spider Tarantula 64 VLB	XF86_S3		i386 1.0A	250K xstones

5.  I had a small hiccough with the ed driver when I rebooted on the
July 8th kernel - I was getting the following message:

ed2: length does not match next packet pointer
ed2: len 4040 nlen ff40 start 4c first 00 curr 68 next 00 stop 80
ed2: NIC memory corrupt - invalid packet length 65344

although everything seems to be working fine.

6.  Snugness wiped off face instantly

I've been encouraging people for ages to check their SCSI termination
whenever they get problems with SCSI hardware - indeed, I've been so
eager to do that, that I haven't checked it properly on one of the
machines around here, a portable that I lug around with me that's got
two SCSI-2 disks and a CD-ROM attached.  I've just found out that the
disk on the end of the SCSI ribbon cable wasn't the disk that was
terminated.  I'd attributed the flakiness of this machine to cheapo

So check your SCSI termination, folks.

7.  Notable changes this week in -current

> svr4 emulation
Fix alignment of svr4_getdents, so that it works for solaris2.4/sparc
svr4_machdep.c: Fix {get,set}context and sendsig.
locore.s: add signal trampoline code for svr4. There is a lot of code
          duplication there. Maybe we should share the code?
SVR4 status now:
        /bin/sh:        works
        /bin/ls:        works
        /bin/sleep:     works
        /bin/csh:       gives unknown software trap 0xa7
        /bin/ldd:       works
> sys/arch/sparc/sparc/conf.c
Add the svr4 network cloning devices. Now telnet(1) works too. There are
still problems with X-windows programs.
> src/etc/etc.svr4
SVR4_MAKEDEV:   Script to make devices in /emul/svr4/dev
nsswitch.conf:  Name service configuration switch file for /emul/svr4/etc
netconfig:      Network configuration file for /emul/svr4/etc
> xlint
lint(1) implementation, by Jochen Pohl.  named 'xlint' for a similar
reason to why 'install' is named 'xinstall'.
> share/zoneinfo/datfiles
updated to tzdata95d
> share/termcap/
Install termcap.src directly instead of processing it with reorder.
The terminfo-derived termcap database we are now using has been pre-
reordered, with the most popular terminal entries first.  Besides it
makes very little difference, since most NetBSD application will be
using the db-version of the termcap database.
Removed reorder script.
Removed tck script, we don't have the showtc program (script?) that it
> sys/net/
Latest version of PPP stuff, with packet compression and other
improvements.  The PPP kernel code is now split into if_ppp.c,
containing generic PPP support, and ppp_tty.c, which specifically
supports PPP on async tty devices (as a line discipline).  This is
so that other devices can be supported without making them look
like ttys.
New version of chat from ppp-2.2 distribution.  The locking code is
gone, because it didn't belong here (chat doesn't open the serial
device), and there is a new "report" feature to report lines beginning
with specified strings (e.g. "CONNECT").
> sys/dev/isa/com.c
Fix a condition where ttywait() wouldn't be woken up.  Also, short-circuit
the case of no pending input in compoll().
> various
Make each disk and tape driver define its own read and write functions.
Deprecate rawread() and rawwrite() completely.  Remove d_strategy from
cdevsw to force the abstraction barrier.
> sys/compat/svr4
Added sockioctls... Only getifnum supported for now.
Added SVR4_EMT_TAGOVF, needed by new machdep translations.
Add a few more sockio ioctls... Now sunxdoom is running!
> sys/arch/`arch`/include/types.h
add <sys/cdefs.h> inclusions.  namsspace-protect physadr, label_t
def'ns against _POSIX_SOURCE and _ANSI_SOURCE.
> sys/sys/audioio.h
Add typedef for audio_prinfo_t (PR#1167), Change AUMODE_* to flags, and add new strings for mixer devs
> sys/dev/audio.c
Make mode a flag set. Full-duplex fixes from John Kohl.
> sys/dev/audio_if.h
Common splaudio and ISA_IPL_AUDIO
> sys/dev/isa/cs4231*.h
add support for cs4231
> sys/dev/isa/aha1542.c,bt1742a.c
insert missing call to splx() when {aha,bt}_send_mbo() returns null.
> gnu/usr.bin/gdb/gdb/kcorelow.c
on i386, use VM_MAXUSER_ADDRESS for kernel_start, since the real
kernel start is much different that the start of KVM.  (VM_MAXUSER_ADDRESS
isn't correct, but seems to mostly work...)  (XXX -- there should be a 
better way.)
if curproc == null, use address of proc0; code doesn't know how to deal
with null curproc.  (XXX: (1) should make it deal, or (2) should use address
of last proc run, but there's no real way to do that.)
> sys/vm/vm_meter.c
fix bug in load average calculation that prevented processes which were
uninterruptibly sleeping for a short period from being counted.

8. Administrivia

I'll be very short of time (due to work commitments) for the next few
weeks, so don't be surprised if the next few reports are delayed.

Enjoy your week,
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]