github.com/DoctorWkt.unix-jun72
1
0
Fork 0
mirror of https://github.com/DoctorWkt/unix-jun72.git synced 2026-01-27 04:32:48 +00:00
Code Issues Releases Wiki Activity

- move various notes files into notes directory.

Browse Source
This commit is contained in:
tim.newsham
2008-05-04 03:43:56 +00:00
parent fbeb9842db
commit e6223da295
4 changed files with 0 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
notes/TODO.txt Normal file
View File

@@ -0,0 +1,29 @@
- can we get fpsym?
- I don't see any mount utility in the s2 bits, we'll need one.
- init does a mount, so we might not need one.
- we should get the /usr stuff onto the rk03 disk and get it mounted
- we should see how hard it is to get the serial ttys working.
It sounds like getting a hack working in simh might not be too
hard, but to get it integrated back in they probably want something
robust. Perhaps having 1ed around will give someone the motivation
to get dc11 support in simh?
- I thought "as" wouldn't work, but it seems to be trying to run
it. If possible we should investigate what is necessary to get
the system to build its own kernel and its own tape.
- we should investigate the original boot process and see how
much of it would be possible to restore and determine if its
worth the effort.
- there's a bunch of file fragments in the 1972_stuff that might
prove useful. It sounds like Warren has started to think about
how to triage these. This might be a bit of effort.
- modify kernel to run 2nd ed 0407 bins.
- bootstrap Early_C_Compilers compiler to build itself as a
2nd ed binary.
- at some point we should probably put together some kind of package
for others who want to run the system in simh. I'm not really
sure what this should look like and perhaps its a bit early to talk
about it... Maybe two sets - a set of rf0 and rk0 images with
a kernel that can run instantly, and a set with a cold kernel,
a normal kernel and a tape image?

142
notes/machine.txt Normal file
View File

@@ -0,0 +1,142 @@
Collection of notes on the hardware this should run on.
------------------
device layout per u0 src:
address simh dev
rk03/rk11 177400 disk RK
dc11 174000 tty? (not supp?)
tc11/tu56 177340 dec tape DTn (not showing up in simh?)
rf11/rs11 177460 fixed head disk RF
kw11-l 177546 clock CLK
pc11 177550 paper tape PTR/PTP
asr-33 177560 tty? TTI, TTO
------------------
man IV tap0 ... tap7 :
"all eight files exist even though at present there are only two physical
drives."
man IV tty
claims that tty is treated differently than tty0..tty5 and implies
that it is not used in multi-user mode. This may pose a problem
if DC-11 is not supported.. we may only be able to run in single user
mode.
"By appropriate console switch settings, it is possible to cause UNIX
to come up as a single-user system with I/O on this device."
man IV tty0 ... tty5
says there are six tty devices. u0 srcs sets ntty to 8+1 (8 dc-11 plus
one console?)
man V core
"It is exactly 8192+64 bytes long" 8k of user core.
this agrees with the definition of core in u0 as 4096 words long.
This also lines up with the statement that there was 24k ram:
u0 sets orig to 0, and core to 16k and ecore to 24k (core+8k).
------------------
set rk3 enabled
att rk3 rk3.dsk
set rf enabled
set cpu 11/20
set cpu 32K ? there was really just 24k in the real machine
-----------------------
Disks:
rf0 - 1024 blocks, always mounted, has root and swap.
last 64 blocks (32kbyte) allocated to swap by u0.s
u0.s can setup rudimentary fs on this if built with "cold=1"
rk0 - 4871 blocks
u0.s sets up rf0 as:
- superblock layout
- free storage map contains 128 bytes
- inode map contains 64 bytes
- allocate top 64 blocks "to unix" (960..1023)
- allocate 17 blocks per process (nproc) (688..960 - 16 procs)
- free blocks 687..34
- zero out blocks 33..1 for use as inodes
(and 2nd half of superblock which is unused?)
- for each special inode i-1 to i-40 initialize it
- mode 100017 (allocated, read, write, other read, othe write)
- num links = 1
- uid = 1 (yup, not root)
- fill in timestamp info
- write out premade i-nodes on i-41 .. i-47
with premade directory data.
i-41 root (with dev, bin, etc, usr, tmp)
i-42 dev (with tty, ppt, mem, rf0, rk0, tap0 .. tap7, tty0..tty7, lpr, tty8)
i-43 bin (empty)
i-44 etc (with init)
i-45 usr (empty)
i-46 tmp (empty)
i-47 init (binary, included in u0.s)
special version that opens /dev/tap0 and reads files from it,
creates them, chmods them and chowns them
special init:
- seek to 65
again:
- read 512 bytes
- use first word as size, if zero, break out of loop
- call create with name at offset 4
- call chmod with byte at offset 2
- call chown with byte at offset 3
- read 512 bytes from tape
- write min(remaining len, 512) to disk
- decrement remaining len and loop if more left
- close file
- loop to again
tape image format:
for each file:
first block:
size : word, zero for end
mode : byte
(setuid, exec, user read, user write, other read, other write)
uid : byte
name : 8 bytes
followed by ((size+511) / 512) blocks of data.
EOF is marked by zero length file.
-----------------------
cpu
probably 11/20 with 24kb (or should that be 24k words = 48k?)
could be 11/45 though.
pdp-11 releases:
1970: 11/20, 11/15
1972: 11/40, 11/45, 11/05, 11/10
1973: 11/35
"In 1971, when this manual was written, we had move off the original PDP-7
to the PDP-11/20; it had 24KB of core memory, and no memory management
hardware at all. "
[...]
"By this time we knew about the upcoming PDP-11/45, and had visited
Digital in Maynard to talk about it; in particular, we had the specs
for the floating-point instructions it supported. So the system
described here included a simulator for the instructions (fptrap(III))."
- http://www.cs.bell-labs.com/who/dmr/1stEdman.html
(describing nov 3, 1971 manuals)
-------------------------------------
section VII pg 5 has "boot procedures" section:
http://cm.bell-labs.com/cm/cs/who/dmr/man71.pdf
------------------------------------
http://en.wikipedia.org/wiki/PDP-11#PDP-11_instruction_repertoire
Has a reference on all the AS instructions in the PDP-11. This may come in
handy while fixing the OCR text.

269
notes/notes.txt Normal file
View File

@@ -0,0 +1,269 @@
Original Source:
http://www.bitsavers.org/pdf/bellLabs/unix/PreliminaryUnixImplementationDocument_Jun72.pdf
------
I started with the pdf, using Preview in mac I selected regions and
copied and created new doc, saved it as jpeg and then used
the online OCR program: http://asv.aso.ecei.tohoku.ac.jp/tesseract/
I then manually edited each file.
The files are named according to the page index in the pdf and the
section and page given in the page footer. Page headers and footers
were not captured.
------------------
e00-01 to e00-10 (10) - u0 -
assigned: tim newsham, complete
reviewed:
e01-01 to e01-10 (10) - u1 -
assigned: Hellwig Geisse
reviewed:
(reviewed e01-01 to e01-06 - tim newsham)
e02-01 to e02-10 (10) - u2 -
assigned: Brantley Coile, complete
reviewed: tim newsham
e03-01 to e03-04 (4) - u3 -
assigned: Tim Bradshaw,
(Brad Parker, complete)
reviewed:
e04-01 to e04-07 (7) - u4 -
assigned: Alex Garbutt,
(Brad Parker, complete)
reviewed:
e05-01 to e05-06 (6) - u5 -
assigned: Ralph Logan, complete
reviewed: johan beiser
e06-01 to e06-06 (6) - u6 -
assigned: johan beisser, complete
reviewed:
e07-01 to e07-08 (8) - u7 -
assigned: Naoki Hamada, complete
reviewed:
e08-01 to e08-09 (9) - u8 -
assigned: Christian David,
(Brad Parker, complete)
reviewed:
e09-01 to e09-08 (8) - u9 -
assigned: tim newsham, complete
reviewed: Ralph Logan
e10-01 to e10-02 (2) - ux -
assigned: tim newsham, complete
reviewed:
e11-01 to e11-07 (7) - sh
we have a binary for the sh already I think. this may be optional
assigned: johan beisser, complete
reviewed:
e12-01 to e12-04 (4) - init
we have a binary for init already
assigned:
reviewed:
------------------
e00-01
- I used spaces to line up the comment columns, not tabs. I
picked the column distance arbitrarily.
- the bottom section looks like it could be tabbed but my tabstops
didnt line up with the values given so I used spaces
- (Brad Parker) removed commented out 3 lines at "4;4". The comment
appears to be written in and is wrong (I think it might just be a
stroke of the pen rather than a correction)
e00-02
- I used tabs for the start of each asm command but not for the
separation between mnemonic and arguments since these didnt line up.
with my tab stops.
e00-03
- left "dis addresses" typo in per original
e00-04
- This instruction doesnt make sense: "dec r0 / r0=33.,...,1 "
r0 is used as the jsr link register. However, r1 has the
value of 34 prior to this. Changed both occurances of "r0" to
"r1" in this code.
e00-05
- cant tell if the "O" in quotes after 1: is a O or a 0.
- left mispelled "mormally" in comment
- left "file/etc/init" without space.
e00-06
- left mis-spacing on 5: move r5,r1.
e00-10
- left "/ect/init" typo
- changed this to "/etc/init". There is no "/ect" directory
on the rf0 drive at the point the cold init is run and
the cold init program doesnt make any new directories, so
this is likely just a typo in the original printout.
------
e01-06
- "r2, p pid-2(r1)" has written correction "p.ppid". Using written
correction.
- "mov $sysret1,(sp)" has written correction for "-(sp)". Using
--------------------
e02-05
- label "1:" is overstruck with "/" in the listing. leaving the
label in place. Should be reviewed for semantic meaning.
- the struck label is referenced below as "1b" right before
"nig:" on e02-06. I believe removing the label would cause
an extra comparison to occur but not change he semantics of
the loop.
e02-09
- extra label "sysquit:" handwritten after "sysintr:". Leaving
the extra label out.
- adding it back in. this label is referenced in the syscall
table (u1.s) but not defined elsewhere.
e02-10
- comment says "cdev has device =" it probably means "#" and could
have just been a weak print of "#", but leaving as "=".
--------------------
e03-*
- I used spaces to line up columns, not tabs.
--------------------
e04-*
- I used spaces to line up columns, not tabs.
e04-06
- comment seems to have "not equal" character in it (equals + slash);
since my keyboard won't do that, I substituted "<>"
--------------------
I noticed that "inode" and "i-node" were used interchangably, which could possibly
indicate two different people commenting the code. I left each as the original.
e05-01
- line 9 has a crossed out term, so i used the correction "r2"
e05-02
- line 48 comment says "panic / found no tres storage" - left it as "tres"
no idea whether it is supposed to be "tree", "free", or "tres"
- reviewed, looks clearly "free", fixed.
e05-04
- some of the latter parts of the code have comments "/ ?", left as is
e05-05
- looks like some of the comments contain a "does not equal symbol" manually
created by typing =<backspace>/. I symbolized it with "=/".
--------------------
e06:
- Used tabs between sections of code instead of spaces.
For example:
bpn rtty0 / comment
This may need some review later. I'll go over it again this afternoon.
--------------------
I used spaces in all occasions because I did not fully understand
where to put tabs. We can convert them easily by a small program
if the rule is specified, so I leave them there.
e07-01
- left "cmp r0,$4" as is. There is a write in correction
for "r1" for "r0".
- reviewed this, r1 has the character at this point and $4 is
control-D. The write-in looks correct.
- left "to" as is after ttych: There is a write in correction
for "from" instead of "to".
e07-04
- left "asr r0" as is. There is a write in correction
for "r2" for "r0".
- changed it to "r2". The comments clearly say "r2". The
value comes from a call to "get" right before the call to
this code ("put"), a page earlier inside the getc function.
"r2" was previously multipled by 2 in get, so this divide
by 2 makese sense to undo that.
e07-07
- printout has: "mov $240.*$ps" but this is invalid syntax and
there are other occurances where the "." is a comma. I'm
changing this to be valid assembly -timnewsham.
e07-08
- There is a handwritten addendum "super block" after the line
"jsr r0, ppoke" which clarifies the comment.
--------------------
e08-*
- I used spaces to line up columns, not tabs.
e08-02
- the bread function is missing a branch when copying data from
kernel memory (r5) to user memory (r2). I added a branch:
1: / r5 points to beginning of data in I/O buffer, r2 points to beginning
/ of users data
movb (r5)+,(r2)+ / move data from the I/O buffer
dec r3 / to the user's area in core starting at u.base
bne 1b
^^^^^^^^^^^
tst u.count / done
beq 1f / yes, return
without this it only copies one byte to userland.
--------------------
e09-01
- I'm using tabs between opcode and operands now
- left "bic $!77,r1" although written comment says "? !177"
- changing to "$!177". This is masking off the high bit of
the received ascii character. shortly after this it compares
with $177 (delete) so clearly its not supposed to mask off
both hight bits ($!77). Also later in the other e09 pages
similar operations are performed.
e09-02
- left "movb tty+38r0),r1" as is. There's a write in
correction for "(" instead of "8". This concerns me as it
means there are probably some errors in this listing as
the paren is most likely the right character here.
- fixed this to allow assembly to work.
e09-03
- hard to tell due to hand written correction, but I believe
the line before "rts r0" reads "mov (sp)+,r0"
- this makes sense since r0 is pushed onto the stack at the
start of the function.
- comment "ASC11" (one instead of I) left as is.
e09-04
- text comment says a label was lost! leaving it out for now.
- adding it back. This looks like a switch table starting
at the missing label (or actually one instruction before
the label, weird that they wrote it this way).
- correction "if count >=0" obscures original text in comment.
no choice but to use new text.
- "ASC11" sic
e09-05
- 5th to last "byte" line 2nd column is uneadable. pattern suggests
the value "000"
--------------------
e10-01
e10-02
- I used tabs for the first tab stop and spaced out to the next step.
- Per discussions with Brantley Coile it appears that the origianl
source probably used tabs, and that the printout has a tab stop
after 8 characters and another after 7 characters. The spaces
between opcode and operands are probably due to a tab, as can be
seen in later assembly code we have in machine readable form.
-------------------
e11-01
- I'm using a whole bunch of guesswork on some of these. Keystrokes
are highly faded, and have to be guessed. The OCR utterly failed.
e11-02
- Line 15 has manual corrections, making it difficult to read what
was added or subtracted.
e11-03
e11-04
- First lines are horribly mangled. Guessed.
- Line 7 has a typo, preserved
- Line 8 has what resembles a "|", not sure if this is a screwup or
not. Preserved, and assuming it's a "|".
- line 21/22: I have no idea.

98
notes/s2files.txt Normal file
View File

@@ -0,0 +1,98 @@
This is an invetory of the s2 files that are incorporated into the
system as compared to what files should be on a proper first edition
system. Some files are missing, some files are borrowed from 2nd ed
but present as 0405 binaries, and some files are 2nd ed binaries that
the 1st ed kernel cannot run.
/bin
----
: V2 cmd, 0405 binary
ar
as V2 binary
b missing
bas
bcd missing
boot missing
cal
cat
cc V2 binary
chball ? no idea
check
chmod
chown
cmp
cp
date
db
dbppt missing
dc
df
ds V2 binary
dsw
ds V2 binary
dsw
dtf missing
du
echo V2 cmd, 0405 binary
ed
exit V2 cmd, 0405 binary
fc V2 binary
find V2 binary
form
goto V2 cmd, 0405 binary
hup missing
if V2 cmd, 0405 binary
lbppt missing
ld V2 binary
ln
login V2 cmd, 0405 binary
ls
mail
maki V2 binary
mesg
mkdir
mkfs
mount
mv
nm V2 binary
od
pr
rew
rkd missing
rkf missing
rkl missing
rm
rmdir
roff
rmdir
roff
sdate missing
sh
size V2 binary
skip ? no idea
sort
stat
strip V2 binary
stty V2 cmd, 0405 binary
su
sum
tap
tm
tty
type
un V2 binary
wc
who
write
/etc
----
as2 V2 binary
getty V2 cmd, 0405 binary
glob
init
msh
suftab
uids