↩ 010. tme booting SunOS 3 on Sun-2 in 2023 ↩
Mon, 19 Jun 2023 23:53:58 +0200, Solaris on Sun-4c on Wed, 21 Jun 2023 23:56:00 +0200
As usual, some program has seen disproportionate development on SunOS.
This was (relatively-)easily-handled for hostid.
Text processing is impossible to guess from foreign-arch binaries, however, especially since all manuals consist primarily of lies;
fmt being part of the Mail package doesn't help.
The Sun-2 emulator is, fittingly, tme.
This is great, since there's not a dozen of half-broken emulators.
This is bad, because 0.8 was published in 2010 and targets exclusively NetBSD through 2007-era autohell.
This is good, since it seems to be the porterbox for NetBSD/sun?.
This is sub-optimal, because the only working guest is NetBSD.
$ curl https://people.csail.mit.edu/fredette/tme/tme-0.8.tar.gz | tar -xz
$ cd tme-0.8
$ ./configure
$ find -name Makefile -exec sed -i s/-Werror// {} + # the author's "reasonably fast machine" is a 500MHz Pentium III
$ find -name Makefile -exec sed -i 's/^all-local:/& $(pkglib_LTLIBRARIES)/' {} + # unclear if this has ever worked or if the author had simply only ever ran single-job makes
$ make -j25
If you have libltdl-dev installed, you'll get
module.c:93:3: error: use of undeclared identifier 'lt_preloaded_symbols'
93 | LTDL_SET_PRELOADED_SYMBOLS();
| ^
/usr/include/ltdl.h:109:23: note: expanded from macro 'LTDL_SET_PRELOADED_SYMBOLS'
109 | lt_dlpreload_default(lt_preloaded_symbols)
| ^
because for some reason the system – instead of the embedded – libltdl is used.
This is remedied by just commenting it out with
$ ed libtme/module.c
93s:^://:
Usually, this ought to be enough; however, in this state it only boots NetBSD; thankfully,
other
users
have come up with patchsets that fix tme enough to boot SunOS 3.
SunOS 2 doesn't bootstrap (and even if installed from SunOS 3, it still
just but errors).
The only SunOS 3 that worked for me was 3.5; earlier installation tapes were broken to varying extents.
Additionally, during SunOS 3 installation,
the emulator will SIGABRT in the tape driver,
since WRITE MARKS is unimplemented.
This is trivially fixed by turning it into a no-op instead.
$ curl http://www.heeltoe.com/download/sun2/diffs-20111125 | patch -p1 patching file bus/multibus/sun-mie.c patching file generic/bus.c patching file generic/bus-el.c patching file host/posix/posix-memory.c patching file ic/i825x6.c patching file ic/m68k/m68010.c patching file ic/m68k/m68k-execute.c patching file ic/m68k/m68k-impl.h patching file ic/m68k/m68k-insns-auto.c patching file ic/m68k/m68k-misc.c patching file libtme/module.c patching file machine/sun/sun-bwtwo.c patching file machine/sun2/sun2-mainbus.c patching file machine/sun2/sun2-mmu.c patching file machine/sun3/sun3-mainbus.c patching file machine/sun4/sun4-mainbus.c patching file scsi/emulexmt02.c patching file tmesh/tmesh.c patching file tmesh/tmesh-cmds.c patching file tmesh/tmesh-input.y $ ed scsi/scsi-tape.c 38495 585 abort(); s:^://: w 38497 q $ make -j25
With this, the official Sun-2 instructions may continue. Well. It's said that when configuring with --disable-shared, plugins will be linked in statically: this doesn't hold (and I needed to manually edit the final linker run).
It's also said that to run tme from the build directory, you can just export LTDL_LIBRARY_PATH=$PWD;
this almost works except it doesn't (presumably because bookworm libtool lays the output out differently,
and the final binary is still linked to the embedded libltdl(?)).
Thus, on the first run, no system component will be loaded to a choir of ENOENTs.
To save you some round-trips between
strace -oss tmesh/tmesh MY-SUN2,
grep 'tme_.*so' ss | grep ENOENT, and
find $PWD -name whatever.so.0,
I found this sufficient for SUN2-MULTIBUS:
$ ln -s $PWD/bus/multibus/.libs/tme_bus_multibus.so.0 $PWD/generic/.libs/tme_generic.so.0 $PWD/host/gtk/.libs/tme_host_gtk.so.0 $PWD/host/posix/.libs/tme_host_posix.so.0 $PWD/ic/.libs/tme_ic_am9513.so.0 $PWD/ic/.libs/tme_ic_mm58167.so.0 $PWD/ic/.libs/tme_ic_z8530.so.0 $PWD/ic/m68k/.libs/tme_ic_m68k.so.0 $PWD/machine/sun2/.libs/tme_machine_sun2.so.0 $PWD/scsi/.libs/tme_scsi.so.0 $PWD/serial/.libs/tme_serial_kb.so.0 tme
With this, the official Sun-2 instructions may continue:
$ curl -O https://people.csail.mit.edu/fredette/tme/sun2-multi-rev-R.bin
$ cp $(find -name SUN2-MULTIBUS) MY-SUN2
+ cp ./machine/sun2/SUN2-MULTIBUS MY-SUN2
$ perl -w $(find -name tme-sun-idprom) 2/120 8:0:20:0:0:0 > my-sun2-idprom.bin # s/0:0:0/whatever/ if you have other sunmicro hosts
+ perl -w ./machine/sun/tme-sun-idprom 2/120 8:0:20:0:0:0
$ find -name sun-keyboards.txt
./machine/sun/sun-keyboards.txt
$ cp ./machine/sun/sun-keyboards.txt ./machine/sun/my-sun-macros.txt .
$ truncate -s 40G my-sun2-disk.img # fun: the emulator SIGBUSses if the guest reads past the end here!
Thankfully, customisation points for SunOS enjoyers and auto-boot are outlined explicitly:
$ grep -nC3 SunOS MY-SUN2 82-## a SCSI disk: 83-## 84:# if you're using SunOS, change the disk type from tme-scsi-1 to 85-# acb4000, to get an Adaptec ACB-4000 compatible disk: 86-# 87-sd0 at scsibus0: tme/scsi/disk id 0 type tme-scsi-1 -- 90-## a SCSI tape: 91-## 92:# if you're using SunOS, remove the vendor EMULEX and product "MT-02 QIC" 93-# to get a 100% Emulex MT-02 compatible tape: 94-# 95-st0 at scsibus0: tme/scsi/tape id 4 type emulex-mt02 vendor EMULEX product "MT-02 QIC" $ ed MY-SUN2 3579 87s/tme-scsi-1/acb4000/p sd0 at scsibus0: tme/scsi/disk id 0 type acb4000 95s/ ve.*//p st0 at scsibus0: tme/scsi/tape id 4 type emulex-mt02 $s/# //p command mainbus0 power up w 3540 q
By default, the emulator boots with the GTK display.
This is fine, it works the most; however, for the purposes of transcribing this tutorialisation I took advantage of the
serial I/O capability
— which can be toggled by inverting the comments on both mentions of "GTK" in MY-SUN2 —
to use my preferred xterm, which I can copy from, as the head.
Naturally, it only supports BSD PTYs, which Debian hasn't configured with on any arch since 2007. Thankfully, all /dev/ptmx support needs is
--- host/posix/posix-serial.c 2007-08-24 02:57:01.000000000 +0200
+++ host/posix/posix-serial.c 2023-06-19 01:26:46.891606875 +0200
@@ -34,4 +34,5 @@
*/
+#define _GNU_SOURCE
#include <tme/common.h>
_TME_RCSID("$Id: posix-serial.c,v 1.11 2007/08/24 00:57:01 fredette Exp $");
@@ -43,4 +44,5 @@
#include <fcntl.h>
#include <stdio.h>
+#include <signal.h>
#include <termios.h>
#include <sys/types.h>
@@ -342,4 +344,6 @@
if (rc < 0) {
/* XXX diagnostic */
+ if (errno == EIO)
+ raise(SIGHUP);
continue;
}
@@ -842,4 +846,5 @@
const char *filename_in;
const char *filename_out;
+ char *vt;
int fd_in, fd_out;
int usage;
@@ -943,4 +948,9 @@
return (errno);
}
+ if (vt = ptsname(fd_in)) {
+ printf("opened %s\n", vt);
+ grantpt(fd_in);
+ unlockpt(fd_in);
+ }
}
if (fd_out < 0) {
I've personally added
if (!fork())
_exit(execlp("st", "st", "-l", vt, (char *)0));
after the unlockpt(), yielding an upgraded-over-stock UX of an xterm showing up on boot and closing the xterm killing the emulator
(set device to /dev/ptmx).
Strictly, this ought to be st -l vt -- -parenb 9600 clocal to match the author's remote(5) line, but none of these do anything on a PTY.
$ curl https://nabijaczleweli.xyz/content/assets/blogn_t/010.1-tme-0.8-ptmx | patch -p0 $ make -j25
Now
$ LTDL_LIBRARY_PATH=$PWD tmesh/tmesh MY-SUN2 [/display0.0]: cannot generate keysym 'Meta_R' directly, or through a macro: No such file or directory # ... [/display0.0]: cannot generate keysym 'R1' directly, or through a macro: No such file or directory lt-tmesh> ls mainbus0: tme/machine/sun2 multibus my-sun2-idprom.bin cpu0 at mainbus0: tme/ic/m68010 obio0 at mainbus0 obio: tme/generic/bus size 8MB obmem0 at mainbus0 obmem: tme/generic/bus size 16MB ram0 at obmem0 addr 0x0: tme/host/posix/memory ram 4MB rom0 at obmem0 addr 0xef0000: tme/host/posix/memory rom sun2-multi-rev-R.bin rom0 at obmem0 addr 0xef8000 clock0 at obio0 addr 0x2800: tme/machine/sun2/clock tod0 at obio0 addr 0x3800: tme/machine/sun2/tod zs0 at obio0 addr 0x2000 ipl 6: tme/machine/sun2/zs mbio0 at mainbus0 mbio: tme/generic/bus size 8MB mbmem0 addr 0x00000 at mainbus0 mbmem: tme/generic/bus size 8MB sc0 at mbmem0 addr 0x80000 ipl 2: tme/bus/multibus/sun-sc scsibus0 at sc0: tme/scsi/bus bwtwo0 at obmem0 addr 0x700000 ipl 4: tme/machine/sun2/bwtwo type multibus zs1 at obmem0 addr 0x780000 ipl 6: tme/machine/sun2/zs kbd0 at zs1 channel A: tme/serial/keyboard type sun-type-2 macros my-sun-macros.txt map sun-keyboards.txt rate 20 ms0 at zs1 channel B: tme/serial/mouse type mousesystems-5 display0 at bwtwo0: tme/host/gtk/display display0 at kbd0 display0 at ms0 sd0 at scsibus0: tme/scsi/disk id 0 type acb4000 disk0 at sd0: tme/host/posix/disk file my-sun2-disk.img st0 at scsibus0: tme/scsi/tape id 4 type emulex-mt02 tape0 at st0: tme/host/posix/tape
or
$ LTDL_LIBRARY_PATH=$PWD tmesh/tmesh MY-SUN2 opened /dev/pts/0 lt-tmesh> ls mainbus0: tme/machine/sun2 multibus my-sun2-idprom.bin cpu0 at mainbus0: tme/ic/m68010 obio0 at mainbus0 obio: tme/generic/bus size 8MB obmem0 at mainbus0 obmem: tme/generic/bus size 16MB ram0 at obmem0 addr 0x0: tme/host/posix/memory ram 4MB rom0 at obmem0 addr 0xef0000: tme/host/posix/memory rom sun2-multi-rev-R.bin rom0 at obmem0 addr 0xef8000 clock0 at obio0 addr 0x2800: tme/machine/sun2/clock tod0 at obio0 addr 0x3800: tme/machine/sun2/tod zs0 at obio0 addr 0x2000 ipl 6: tme/machine/sun2/zs mbio0 at mainbus0 mbio: tme/generic/bus size 8MB mbmem0 addr 0x00000 at mainbus0 mbmem: tme/generic/bus size 8MB sc0 at mbmem0 addr 0x80000 ipl 2: tme/bus/multibus/sun-sc scsibus0 at sc0: tme/scsi/bus console0 at zs0 channel A: tme/host/posix/serial device /dev/ptmx break-carats sd0 at scsibus0: tme/scsi/disk id 0 type acb4000 disk0 at sd0: tme/host/posix/disk file my-sun2-disk.img st0 at scsibus0: tme/scsi/tape id 4 type emulex-mt02 tape0 at st0: tme/host/posix/tape
should hold.
If you get any ENOENTs that aren't due to keysyms, you're missing some plugins that need to be linked to tme/.
Since 1 is an exercise in writing programs for your programs in your programs, the tmesh process is, per nom, a shell;
unlike dectapes everyone's familiar with, tapes here consist of indexed partitions;
thus, running command tape0 load part0 part1 … inserts a tape consisting of parts into st0,
command tape0 unload does the obverse, and loading twice in a row yields a sensible error.
Conversely, ^Ding segfaults the emulator.
The hard(-to-bear) part (other people's C) is largely over, and can be followed by the hard(-to-bear) part of unpacking 1. Western Electric and UCB licensees may obtain the SunOS 3.5/Sun-2 tape set from WinWorld, extract it somewhere, and ease their lives by running ln -s $somewhere/'Sun UNIX 4.2 Release 3.5 (Tape) [Sun-2]'/tape? .. All SunOS 3 tapes, contrary to their predecessors, have a TOC as the second partition:
$ cat tape1/02 # Inter-tape distances enlarged to show texture
TABLE OF CONTENTS
MC68010 tape 1
1 1 tape boot block
1 2 toc
1 3 copyright
1 4 diag
1 5 copy
1 6 mini-root
1 7 root
1 8 pub
1 9 client image
1 10 copyright
2 1 copyright
2 2 toc
2 3 Sys
2 4 Networking tools and programs
2 5 Debugging tools
2 6 usr
2 7 copyright
3 1 copyright
3 2 toc
3 3 SunView User Programs
3 4 SunView Programmers Files
3 5 SunView and Demo Program source
3 6 Text Processing tools
3 7 Setup tools
3 8 Standalone Diagnostics
3 9 Fortran Compiler & Libraries
3 10 User Level Diagnostics
3 11 SunCore & CGI Libraries
3 12 Pascal Interpreter & Compiler
3 13 Profiled Libraries
3 14 uucp programs
3 15 copyright
4 1 copyright
4 2 toc
4 3 System V programs and libraries
4 4 Manual Pages
4 5 Demonstration Programs
4 6 Games
4 7 SunPro
4 8 copyright
5 1 copyright
5 2 toc
5 3 Versatec Printer Software
5 4 copyright
# This is, thus, not relevant, but is nevertheless humourous
$ paste <(ls -l tape?/* | cut -c35-) <(file tape?/* | cut -d: -f2-)
30720 Mar 28 1994 tape1/01 data
1024 Mar 28 1994 tape1/02 GTA in-game text (GXT), version 3, used in GTA VC/LS/VCS
512 Mar 28 1994 tape1/03 data
98816 Mar 28 1994 tape1/04 a.out SunOS mc68010 executable
41472 Mar 28 1994 tape1/05 a.out SunOS mc68010 executable not stripped
4718592 Mar 28 1994 tape1/06 Unix Fast File system [v1] (big-endian), last mounted on /mnt, last written at Wed Dec 2 18:02:04 1987, clean flag 0, number of blocks 4608, number of data blocks 4303, number of cylinder groups 3, block size 8192, fragment size 1024, minimum percentage of free blocks 0, rotational delay 4ms, disk rotational speed 60rps, TIME optimization
1843200 Mar 28 1994 tape1/07 tar archive
4198400 Mar 28 1994 tape1/08 tar archive
1740800 Mar 28 1994 tape1/09 tar archive
512 Mar 28 1994 tape1/10 data
$ echo tape1/*; echo tape2/*; echo tape3/*; echo tape4/*; echo tape5/* # and this is useful for later
tape1/01 tape1/02 tape1/03 tape1/04 tape1/05 tape1/06 tape1/07 tape1/08 tape1/09 tape1/10
tape2/01 tape2/02 tape2/03 tape2/04 tape2/05 tape2/06 tape2/07
tape3/01 tape3/02 tape3/03 tape3/04 tape3/05 tape3/06 tape3/07 tape3/08 tape3/09 tape3/10 tape3/11 tape3/12 tape3/13 tape3/14 tape3/15
tape4/01 tape4/02 tape4/03 tape4/04 tape4/05 tape4/06 tape4/07 tape4/08
tape5/01 tape5/02 tape5/03 tape5/04
Therefore
lt-tmesh> command tape0 load tape1/01 tape1/02 tape1/03 tape1/04 tape1/05 tape1/06 tape1/07 tape1/08 tape1/09 tape1/10
and on the console:
Self Test completed successfully. Sun Workstation, Model Sun-2/120 or Sun-2/170, Sun-2 keyboard ROM Rev R, 4MB memory installed Serial #83661, Ethernet address 8:0:20:0:0:0 Probing Multibus: sd Auto-boot in progress... Boot: sd(0,0,0)vmunix Exception 2C at 006000 >b st() # equivalent to st(0,0,0) Boot: st(0,0,0) Boot: st(0,0,3) # "1 4 diag" Size: 58636+33256+29072 bytes Version 1.3 86/07/25 Disk Initialization and Diagnosis When asked if you are sure, respond with 'y' or 'Y' specify controller: 0 - Interphase SMD-2180 1 - Xylogics 440 (prom set 926) 2 - Xylogics 450/451 3 - Adaptec ACB 4000 - SCSI/ST506 4 - Emulex MD21 - SCSI/ESDI which one? 4 # even though the configuration includes theacb4000; the emulex appears to be functionally equivalent but it has bigger disks available Specify controller address on the mainbus (in hex): 80000 #disk0 at sd0 at scsibus0 at sc0 at mbmem0 addr 0x80000Device address: 80000 Which target? 0 # for a second disk, duplicate both lines mentioningsd0, rename tosd1,disk1,id 1and use 1 here; this ends up being/dev/sd2Which unit? 0 Specify drive: 0 - Micropolis 1355 1 - Toshiba MK 156F 2 - Micropolis 1558 3 - Other which one? 2 # this is the largest disk available; the 1355 is the second-largest, and the 1558 is not available on earlier systems ncyl 1218 acyl 2 nhead 15 nsect 35 interleave 1 reset the SCSI bus... diag> label label this disk... OK to use logical partition map 'Micropolis 1558'? y Are you sure you want to write? y verify label id: <Micropolis 1558 cyl 1218 alt 2 hd 15 sec 35> Partition a: starting cyl=0, # blocks=16275 Partition b: starting cyl=31, # blocks=32550 Partition c: starting cyl=0, # blocks=639450 Partition g: starting cyl=93, # blocks=590625 diag> q Boot: st(,,4) # "1 5 copy" Size: 23088+5248+113452 bytes Standalone Copy From: st(0,0,5) # "1 6 mini-root" To: sd(0,0,1) # sd0b, conventionally swap Copy completed - 4720640 bytes Boot: sd(0,0,1)vmunix -as # -a to prompt for /; -s for single-user Size: 481704+76100+64288 bytes Sun UNIX 4.2 Release 3.5 (GENERIC) #1: Tue Dec 1 16:56:07 PST 1987 Copyright (c) 1986 by Sun Microsystems, Inc. mem = 4096K (0x400000) avail mem = 3280896 Ethernet address = 8:0:20:0:0:0 sc0 at mbmem 80000 pri 2 sd0 at sc0 slave 0 sd0: <Micropolis 1558 cyl 1218 alt 2 hd 15 sec 35> sd1 at sc0 slave 1 st0 at sc0 slave 32 sd2 at sc0 slave 8 st1 at sc0 slave 40 zs0 at obio 2000 pri 3 pi0 at obio 1800 tod0 at obio 3800 root device? sd0* # the*enables cooperative swap-on-root (instead of normally dumping to sd0b, which instantly destroys the filesystem there); this doesn't appear to be documented using 100 buffers containing 366592 bytes of main memory WARNING: TOD clock not initialized -- CHECK AND RESET THE DATE! Singleuser boot -- fsck not done # stty erase \^H # (^is special in the bourne shell) if using the graphical console; the keyboard mapping appears to match actual keyboards in backspace being ^H; the miniroot has no way of checking whether it's run on /dev/console and always runsstty dec; the installed system does, but also still does # cd dev # MAKEDEV st0 # newfs /dev/rsd0a /dev/rsd0a: 16274 sectors in 31 cylinders of 15 tracks, 35 sectors 8.3Mb in 2 cyl groups (16 c/g, 4.30Mb/g, 1920 i/g) super-block backups (for fsck -b#) at: 32, 8480, # newfs /dev/rsd0g /dev/rsd0g: 590624 sectors in 1125 cylinders of 15 tracks, 35 sectors 302.4Mb in 71 cyl groups (16 c/g, 4.30Mb/g, 1984 i/g) super-block backups (for fsck -b#) at: 32, 8480, 16928, 25376, 33824, 42272, 50720, 59168, 67616, 76064, 84512, 92960, 101408, 109856, 118304, 126752, 134432, 142880, 151328, 159776, 168224, 176672, 185120, 193568, 202016, 210464, 218912, 227360, 235808, 244256, 252704, 261152, 268832, 277280, 285728, 294176, 302624, 311072, 319520, 327968, 336416, 344864, 353312, 361760, 370208, 378656, 387104, 395552, 403232, 411680, 420128, 428576, 437024, 445472, 453920, 462368, 470816, 479264, 487712, 496160, 504608, 513056, 521504, 529952, 537632, 546080, 554528, 562976, 571424, 579872, 588320, # mount sd0a /a # cd /a # mm() { mt -f /dev/nrst0 $*; } # unp() { mm rew; mm fsf $i; tar xvpf /dev/nrst0; } # I'm pretty sure the bourne shell on this system doesn't have scoped variables, because usingmm fsf $*or$1here always runsmm fsf rew# set -x # for i in 6 7 8; do unp; done # "1 7 root", "1 8 pub", "1 9 client image" + unp + mm rew + mt -f /dev/nrst0 rew + mm fsf 6 + mt -f /dev/nrst0 fsf 6 + tar xvpf /dev/nrst0 x ./etc/nd.local, 270 bytes, 1 tape blocks x ./etc/dkinfo, 24576 bytes, 48 tape blocks x ./etc/nd, 49152 bytes, 96 tape blocks # ... x ./pub/bin/sh, 65536 bytes, 128 tape blocks x ./.cshrc, 15 bytes, 1 tape blocks x ./.login, 84 bytes, 1 tape blocks x ./.profile, 70 bytes, 1 tape blocks x ./.rhosts, 0 bytes, 0 tape blocks x ./boot, 35020 bytes, 69 tape blocks # cd dev # MAKEDEV std ttys pty0 win0 sd0 st0 # mount sd0g /a/usr # cd ../etc # cat > fstab /dev/sd0a / 4.2 rw,noquota 1 1 /dev/sd0g /usr 4.2 rw,noquota 1 2 ^D # mv ypbind ypbind_ # this disables ypbind(8), cf. rc.local lt-tmesh> command tape0 unload lt-tmesh> command tape0 load tape2/01 tape2/02 tape2/03 tape2/04 tape2/05 tape2/06 tape2/07 # cd ../usr # for i in 2 3 4 5; do unp; done # "2 3 Sys", "2 4 Networking tools and programs", "2 5 Debugging tools", "2 6 usr" + unp + mm rew + mt -f /dev/nrst0 rew + mm fsf 2 + mt -f /dev/nrst0 fsf 2 + tar xvpf /dev/nrst0 x ./sys/OBJ/addupc.o, 354 bytes, 1 tape blocks x ./sys/OBJ/af.o, 526 bytes, 2 tape blocks x ./sys/OBJ/ar.h, 14 bytes, 1 tape blocks # ... x ./sccs/helpdir/un, 559 bytes, 2 tape blocks x ./sccs/helpdir/ut, 2070 bytes, 5 tape blocks x ./sccs/helpdir/va, 101 bytes, 1 tape blocks x ./sccs/helpdir/vc, 2417 bytes, 5 tape blocks lt-tmesh> command tape0 unload lt-tmesh> command tape0 load tape3/01 tape3/02 tape3/03 tape3/04 tape3/05 tape3/06 tape3/07 tape3/08 tape3/09 tape3/10 tape3/11 tape3/12 tape3/13 tape3/14 tape3/15 # for i in 2 3 4 5 6 7 8 9 10 11 12 13; do unp; done # "3 3 SunView User Programs", "3 4 SunView Programmers Files", "3 5 SunView and Demo Program source", "3 6 Text Processing tools", "3 7 Setup tools", "3 8 Standalone Diagnostics", "3 9 Fortran Compiler & Libraries", "3 10 User Level Diagnostics", "3 11 SunCore & CGI Libraries", "3 12 Pascal Interpreter & Compiler", "3 13 Profiled Libraries", "3 14 uucp programs" + unp + mm rew + mt -f /dev/nrst0 rew + mm fsf 2 + mt -f /dev/nrst0 fsf 2 + tar xvpf /dev/nrst0 x ./bin/adjacentscreens symbolic link to othertools x ./bin/align_equals symbolic link to suntools # ... x ./spool/uucp/LOGFILE, 0 bytes, 0 tape blocks x ./spool/uucp/SYSLOG, 0 bytes, 0 tape blocks lt-tmesh> command tape0 unload lt-tmesh> command tape0 load tape4/01 tape4/02 tape4/03 tape4/04 tape4/05 tape4/06 tape4/07 tape4/08 # for i in 2 3 4 5 6; do unp; done # "4 3 System V programs and libraries", "4 4 Manual Pages", "4 5 Demonstration Programs", "4 6 Games", "4 7 SunPro" # ... + unp + mm rew + mt -f /dev/nrst0 rew + mm fsf 6 + mt -f /dev/nrst0 fsf 6 + tar xvpf /dev/nrst0 x ./sunpro/install_sunpro, 472 bytes, 1 tape blocks x ./sunpro/compile, 73728 bytes, 144 tape blocks x ./sunpro/cpp, 49152 bytes, 96 tape blocks x ./sunpro/ld, 57344 bytes, 112 tape blocks x ./sunpro/m4, 40960 bytes, 80 tape blocks x ./sunpro/make, 98304 bytes, 192 tape blocks x ./sunpro/make.1, 39910 bytes, 78 tape blocks x ./sunpro/default.mk, 3351 bytes, 7 tape blocks x ./sunpro/filemerge, 638976 bytes, 1248 tape blocks x ./sunpro/filemerge.1, 7165 bytes, 14 tape blocks # cd / # umount /a/usr # umount /a
and kill the emulator: the miniroot doesn't have halt, not that it potentially getting corrupted matters.
At this point we have SunOS, and it Just Boots.
The miniroot environment didn't have curses programs, but your fully-unpacked 46MiB system (more if you also unpacked the printer drivers) does;
TERM=sun works for the graphical console; termcap is hell, TERM=ansi
(sx|ansi|any ansi terminal with pessimistic assumptions:\)
doesn't work under my preferred xterm, and I have neither the patience or the will to try others.
Unless you're an ed fiend, I'd recommend just using the graphical console.
(If you won't, you'll need to also fight getty to produce a sane login prompt.)
$ LTDL_LIBRARY_PATH=$PWD tmesh/tmesh MY-SUN2
Self Test completed successfully.
Sun Workstation, Model Sun-2/120 or Sun-2/170, Sun-2 keyboard
ROM Rev R, 4MB memory installed
Serial #83661, Ethernet address 8:0:20:0:0:0
Probing Multibus: sd
Auto-boot in progress...
Boot: sd(0,0,0)vmunix
Load: sd(0,0,0)boot
Boot: sd(0,0,0)vmunix
Size: 481704+76100+64288 bytes
Sun UNIX 4.2 Release 3.5 (GENERIC) #1: Tue Dec 1 16:56:07 PST 1987
Copyright (c) 1986 by Sun Microsystems, Inc.
mem = 4096K (0x400000)
avail mem = 3280896
Ethernet address = 8:0:20:0:0:0
sc0 at mbmem 80000 pri 2
sd0 at sc0 slave 0
sd0: <Micropolis 1558 cyl 1218 alt 2 hd 15 sec 35>
sd1 at sc0 slave 1
st0 at sc0 slave 32
sd2 at sc0 slave 8
st1 at sc0 slave 40
zs0 at obio 2000 pri 3
pi0 at obio 1800
root on sd0
using 100 buffers containing 366592 bytes of main memory
WARNING: no TOD clock -- CHECK AND RESET THE DATE!
/dev/sd0a: 257 files, 5903 used, 1705 free (1 frags, 213 blocks)
/dev/rsd0g: 4749 files, 47151 used, 229399 free (311 frags, 28636 blocks)
Automatic reboot in progress...
Wed Dec 2 10:13:37 PST 1987
starting rpc and net services: portmap router biod.
local daemons: sendmail statd lockd.
preserving editor files
clearing /tmp
standard daemons: update cron printer uucp.
starting network: inet.
Wed Dec 2 10:13:38 PST 1987
�
�o���e�lo�i�:��
�o���e�lo�i�:��
�o���e�lo�i�:�
�
�o���e�lo�i�:�t��t�re�et
Password:
Login incorrect
login: root
Sun UNIX 4.2 Release 3.5 (GENERIC) #1: Tue Dec 1 16:56:07 PST 1987
noname# cat .login
stty dec
set path=($path /etc /usr/etc)
set noglob; eval `tset -Q -s`; unset noglob
noname# ed .login
84
1a
if (`tty` = /dev/console) stty erase ^H
.
w
126
q
noname# cd /etc
noname# ed rc.boot
1439
/host/
hostname=noname
s/noname/sunos3/
w
1439
q
noname# ed rc.local
3079
/nona/
/bin/domainname noname
s/noname/nabijaczleweli.xyz/
w
3091
q
noname# halt
syncing disks... done
Unix Halted
>b
Boot: sd(0,0,0)vmunix
Load: sd(0,0,0)boot
Boot: sd(0,0,0)vmunix
Size: 481704+76100+64288 bytes
Sun UNIX 4.2 Release 3.5 (GENERIC) #1: Tue Dec 1 16:56:07 PST 1987
Copyright (c) 1986 by Sun Microsystems, Inc.
mem = 4096K (0x400000)
avail mem = 3280896
Ethernet address = 8:0:20:0:0:0
# ...
standard daemons: update cron printer uucp.
starting network: inet.
Wed Dec 2 10:16:12 PST 1987
�
���o�3�lo�i�:��
���o�3�lo�i�:��
���o�3�lo�i�:����
Password:
Login incorrect
login: root
Last login: Wed Dec 2 10:14:09 on console
Sun UNIX 4.2 Release 3.5 (GENERIC) #1: Tue Dec 1 16:56:07 PST 1987
sunos3# suntools
starts X (hover over the text that says "Mouse is off" in the bottom left and press something on the keyboard to capture the mouse):
X (and the console in general) achieve single-digit refresh rates on bookworm;
this is likely related to the emulator being in a tight poll() loop with a zero timeout.
I'd attribute this to a GDK incompatibility probably (or the intrusive user-space scheduler), idk what else it could be doing.
The emulator supposedly sleeps when kNetBSD does. This may be true even for kSunOS, but due to the above it sits at a constant 100% CPU always.
Host networking is only available via the BSD /dev/bpf* interface.
That said, host/bsd/bsd-bpf.c is just 1200 lines and most of that is likely able to be duplicated directly.
OTOH it's unclear if the ie device itself would work under SunOS.
# SunOS 4 (Solaris) on Sun-4c
$ curl https://nabijaczleweli.xyz/content/assets/blogn_t/010.7-tme-0.8-sun4 | patch -p0 $ make -j25 $ ln -s $PWD/bus/sbus/.libs/tme_bus_sbus.so.0 $PWD/ic/.libs/tme_ic_am7930.so.0 $PWD/ic/.libs/tme_ic_am7990.so.0 $PWD/ic/.libs/tme_ic_lsi64854.so.0 $PWD/ic/.libs/tme_ic_mk48txx.so.0 $PWD/ic/.libs/tme_ic_ncr53c9x.so.0 $PWD/ic/.libs/tme_ic_nec765.so.0 $PWD/ic/sparc/.libs/tme_ic_sparc.so.0 $PWD/machine/sun4/.libs/tme_machine_sun4.so.0 tme
wherein
--- libtme/log-prf.c 2009-08-30 18:56:42.000000000 +0200
+++ libtme/log-prf.c 2023-06-21 18:23:33.286882822 +0200
@@ -205,7 +205,7 @@
case 's':
prf_value_s = va_arg(prf_args, const char *);
PRF_OUT_ARG_CODE(TME_LOG_ARG_CODE_STRING);
- PRF_OUT_MEM(prf_value_s, strlen(prf_value_s));
+ PRF_OUT_MEM(prf_value_s, prf_value_s ? strlen(prf_value_s) : (size_t)0);
/* enter state zero: */
prf_agg = prf_format;
fixes a printf("%s") implementation (why doess tme ship its own printf body?) to not segfault when passed a null pointer,
and
--- libtme/host/x86/misc-x86.c 2009-11-08 18:21:18.000000000 +0100
+++ libtme/host/x86/misc-x86.c 2023-06-21 21:39:26.322610998 +0200
@@ -49,10 +49,14 @@
unsigned long reg_d;
union tme_value64 value;
+#if 0
asm(" rdtsc \n"
: "=a" (reg_a), "=d" (reg_d));
value.tme_value64_uint32_lo = reg_a;
value.tme_value64_uint32_hi = reg_d;
+#else
+ value.tme_value64_uint = __builtin_ia32_rdtsc();
+#endif
return (value);
}
#define TME_HAVE_MISC_CYCLES
fixes the instruction counter being truncated to 32 bits ⇒ no forward progress when measuring ⇒ tickrate estimation estimating zero ticks per second ⇒
faulting cpu0 with an error on creation.
With this, I could boot a Sun-4c machine as in the upstream instruxions
(the idle-type bit in the config may be adjusted for SunOS and the emulator does actually sleep;
the graphical console still updates at the same rate, and the serial variant doesn't seem to work at all after the initial POST)
with solaris1.1.2.7z from my good friend Juliano
and install by following the script (you want suninstall after rebooting to the miniroot).
I got to the funniest time-zone selection menu in history:
in the third-stage setup (and the hardware clock does work! Poland, at least, appears to be accurate still.
Haven't checked whether "euro-bit-of-the-СССР" matches any part of the РФ).
The UID entry box is broken, and doesn't accept the entered line: do not answer yes to making a new user in sys-setup.
I have found that the system is — regardless of disklabel — very liable to just hang when when quota-checking or fscking on boot, unpacking a bigger preset, or simply cd /usr/etc/install; ls -lling to recover from the above. In many ways I haven't actually gotten the system to boot cleanly (indeed, at all) after the final post-installation reboot, so the time-zone selection menu is all you really get. Or just ^Cing it and hoping it doesn't break before you manage to test what you need, but at that point you might as well just chroot after installing from the miniroot.
Nit-pick? Correction? Improvement? Annoying? Cute? Anything? Don't hesitate to post
or open an issue!