github.com/PDP-10.its
1
0
Fork 0
mirror of https://github.com/PDP-10/its.git synced 2026-03-05 11:04:02 +00:00
Code Issues Releases Wiki Activity

Added support for FTP client and server.

Browse Source
This commit is contained in:
Eric Swenson
2016-11-24 09:10:52 -08:00
committed by Lars Brinkhoff
parent eb3aec175b
commit a94bfe9cf7
12 changed files with 11418 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
Makefile
View File

@@ -1,6 +1,6 @@
EMULATOR ?= simh
SRC = system syseng sysen1 sysen2 sysnet kshack dragon channa midas _teco_ emacs rms klh syshst sra mrc
SRC = system syseng sysen1 sysen2 sysnet kshack dragon channa midas _teco_ emacs rms klh syshst sra mrc ksc
DOC = info _info_ sysdoc kshack _teco_ emacs emacs1
MINSYS = _ sys sys2 sys3 device emacs _teco_ sysbin
RAM = bin/boot/ram.262

5
README.md
View File

@@ -62,6 +62,8 @@ from scratch.
- TELSER, Telnet/Supdup server
- TELNET, Telnet client
- SUPDUP, Supdup client
- FTPS, FTP Server
- FTPU, FTP Client
6. A brand new host table is built from the host table source and
installed into SYSBIN; HOSTS3 > using H3MAKE.
@@ -76,7 +78,8 @@ host table tools, and binary host table are installed.
Currently, basic TCP network support is in the build, in addition to
both a TELNET/SUPDUP server, and both TELNET and SUPDUP clients.
Other network services will appear in a subsequent release.
Additionally, both an FTP server and client are included. Other network
services will appear in a subsequent release.
The KLH10 dskdmp.ini file has an IP address (192.168.1.100) and gateway IP
address (192.168.0.45) configured for the ITS system. The IP address

20
build/build.tcl
View File

@@ -320,6 +320,7 @@ respond "*" ":midas sysbin;telnet_sysnet;telnet\r"
expect ":KILL"
respond "*" ":link sys;ts telnet,sysbin;telnet bin\r"
respond "*" ":link sys;ts tn,sys;ts telnet\r"
# supdup port (95) uses telser
respond "*" ":link device;tcp syn137,sysbin;telser bin\r"
@@ -333,6 +334,25 @@ type ":vk\r"
respond "*" ":link sys1;ts supdup,sysbin;supdup bin\r"
respond "*" ":link syseng;fsdefs 999999,system;fsdefs >\r"
# these two links are expected by sysnet; ftps > and are present
# in the PI distribution
respond "*" ":link ksc;nuuos 999999,klh;nuuos >\r"
respond "*" ":link ksc;macros 999999,klh;macros >\r"
respond "*" ":link ksc;out 999999,klh; out >\r"
respond "*" ":midas sysbin;ftps_sysnet;ftps \r"
expect ":KILL"
respond "*" ":link device;tcp syn025,sysbin;ftps bin\r"
respond "*" ":link device;tcp syn031,sysbin;ftps bin\r"
respond "*" ":midas sysbin;ftpu_sysnet;ftpu \r"
expect ":KILL"
respond "*" ":link sys;ts ftp,sysbin;ftpu bin\r"
respond "*" ":link kshack;good ram,.;ram ram\r"
respond "*" ":link kshack;ddt bin,.;@ ddt\r"
respond "*" $emulator_escape

221
doc/_info_/ftp.info Executable file
View File

@@ -0,0 +1,221 @@
Description of FTP commands, as or 2/5/77
Some Useful FTP Commands:
CONNECT <host> Connects to a foreign host. Use the name of the host
or its (octal) numeric address. Any pre-existing
connection will be flushed.
FTP prompts with "$$" when it is not connected to
a foreign host, and with "$" when a connection exists.
<host> Is equivalent to CONN <host>.
HOSTS Lists the names of hosts which FTP knows
LOGIN <name> Logs in at foreign host (if needed by that host)
If you get an error message implying that you are not
logged in, this is where to start. Sometimes FTP
will figure out by itself that logging in is what you
have to do; then it will ask for for a user name,
send it, and go ahead with the operation you had
already requested.
TEN Mode command for pdp10 36 bit binary files.
(May also be used for pdp10 ascii files if you don't
mind extraneous characters at the end of the file)
NOTE: This is the DEFAULT mode when communicating with
another PDP10. You can specify it explicitly, also,
with any computer (although some may not accept it).
It is an abbreviation for TYPE I and BYTE 36
TEXT Mode command for Network ASCII data transmission.
This mode is the DEFAULT when communicating with a
machine that is not a PDP10. It is an abbreviation
for TYPE A and BYTE 8.
GET or RETRIEVE Transfers a file from foreign host to here.
Asks for local name (end with CR) and then foreign name.
PUT or STORE Transfers a file from here to foreign host.
or SEND Asks for local name (end with CR) and then foreign name.
APPEND Appends a local file onto a file at the foreign host.
Arguments are local file (not changed by the operation)
followed by the foreign file (which will be created if
it does not exist), as for PUT.
TRAN <host1><file1>=<host2><file2>
Transfers <file2> at <host2> to <file1> at <host1>.
Two foreign hosts can't be used, though; one of
<host1> and <host2> must be the local host's name.
Abbreviations are allowed: You can leave out the
name of the local host (omitting the altmode as well),
as in TRAN FOO BAR=MCFOO BAR or TRAN MCFOO BAR=X Y,
and if the two filespecs are identical either one may
be omitted, as in TRAN FOO BAR=MC which is equivalent
to TRAN FOO BAR=MCFOO BAR.
NOTE: "_" is equivalent to "=". To include a "=" or
"_" or a ^Q in a filename in a TRAN command, precede
it with a ^Q (which will be gobbled up as a quoter).
^Q's followed by other characters (not _, = or ^Q)
will NOT be absorbed by TRAN (so you can use them to
quote at other levels, with hosts that understand it).
QUIT Closes connection to foreign host, and then exit from FTP
^C Is the same as QUIT - no CR is needed.
Any command may be terminated with a ^C instead of a
CR, in which case that command will be followed by
a QUIT (unless it encounters an error).
DISC Closes connection to foreign host. You may then connect
to another host, or QUIT, or ^Z...
HELP or ? Types a list of commands, with brief descritions.
Less useful commands:
DELETE <file> Deletes the specified file at the foreign host.
RENAME <oldname>
<newname>
Renames the file <oldname> at the foreign host to be
called <newname>. Separate the two names with a CR.
DIRECTORY <dir> or
LISTF <dir> Types out a listing of the foreign host's directory <dir>.
Some hosts may also allow you to specify a filename,
in which case info on that file will be given.
LISTB <dir> Types out a brief listing of a foreign directory.
It is like LISTF except that onlty the filenames are
typed - no other info on the files.
LISTL <dir> Types out a listing of the directory <dir> at
the local host.
CWD <dir> Sets the foreign default directory to <dir>
DEFAULTS <filenames>
Sets the local filename defaults to <filenames>.
Note: the local defaults are sticky; each local
filename's names default to the last filenames
specified for a local file.
Rarely Useful Commands:
STATUS <arg> Asks for a status message from the foreign host.
The meaning of <arg> is site-dependent. ITS sites
don't implement the command at all.
QUOTE <command> Sends <command> directly to the foreign host over
the TELNET connection. Used if the foreign host
implements some simple command that FTP doesn't know
about itself. It won't work for commands that do
anything hairy (such as use the data connection at
all) since FTP doesn't expect it do do that.
SOAK Waits for and describes one reply from the server.
This is never necessary with any of the normal
FTP commands, since they know when to wait for
replies.
DEBUG Toggles the switch which says whether to print expected
replies from the server (unexpected replies such as
error messages and notices to the user are always
printed). The DEBUG switch also causes reply codes
to be given with all replies, not just error messages.
PASS <password> Sends a password to a foreign host. Note that after
a LOGIN command, if a password is needed, FTP will
usually be able to figure that out; it will ask you
for the password and send it automatically, so you
need not explicitly give a PASSWORD command.
ACCT <account> Sends an account number to the foreign host. Note
that if after a LOGIN command an account number is
needed, FTP will usually figure that out and just
ask you for it. So you usually won't need to give
an explicit ACCT command.
BYE Is like QUIT, but it notifies the foreign host
with a BYE command before closing the connection.
Is this ever useful? ^C is easier.
ALLOCATE <size> Sends the foreign host the estimated size of a file
you are about to send it (measured in bytes). Most
hosts don't need ALLOCATE commands since they can
allocate file storage dynamically.
TYPE <type> Says what transfer type to use. Of the many types
specified in the FTP protocol, only A (ASCII) and
I (Image) are implemented on ITS. You usually
would rather use the TEXT or TEN commands, since
they specify the two most useful combinations of
TYPE and BYTE.
BYTE <size> Says what byte size to use for transfers. The only
byte sizes implemented on ITS for FTP'ing are 8,
32 and 36. In any case, you should usually use the
TEXT and TEN commands for mode-setting.
Giving FTP commands from DDT:
If you do :FTP <command> and <command> contains an = or a _,
FTP automatically does a TRAN command with <command>
as arguments. After that transfer completes, FTP
kills itself. For example, :FTP FOO BAR=MC will
transfer FOO BAR from MC to the local host, leaving
you back in DDT.
If the command string from DDT does not contain an = or a _,
it is simply executed as an FTP command. FTP will then read
more commands from the TTY, since one FTP command (unless it
is TRAN) is not usually enough for a whole transaction.
Filename Defaulting:
Local filenames default "stickily", so that each file's device,
sname, and first and second filenames all default to the last
such names specified for a local file.
Giving a null argument for the local file will cause the default
file to be used.
When you specify only one filename, it is the first filename,
and ">" is used as the second.
The initial default directory is your working directory.
If you wish to include a space, colon, or semicolon in a local
filename (or in a foreign filename if the foreign site is an ITS),
precede it with a ^Q. The same technique is used for including
a = or _ or ^Q anywhere in a pathname, in the TRAN command.
Type-in Conventions:
You can use Rubout to delete one character, ^D, ^G or ^U to
cancel an entire line, and ^L to redisplay the current line
as typed so far. Commands with no arguments should be terminated
with a CR. When giving a command with arguments, you can
separate the command name from the first argument with either
a space or a CR (if you use CR, FTP will tell you what sort
of argument is expected). When a command requires several
arguments, the arguments are separated with CRs.
You can use a ^C instead of CR to terminate a line. That
causes FTP to return to DDT, killing itself, after the current
command is finished (as long as it does not get an error).
If you plan to disown a FTP while it is transfering, you should
end the last command with a ^C so that the FTP will go away
when it is finished; otherwise it might hang around for hours
until someone notices it and kills it.
You can use ^N instead of CR to terminate a line. This is not
especially useful when you are typing on the TTY, but it is
useful when passing commands from DDT, since it allows you to
put several lines' worth of input into the one line that DDT lets
you send.

342
doc/info/ftp.14 Executable file
View File

@@ -0,0 +1,342 @@
-*-Text-*-

File: FTP Node: TOP Up: (DIR) Next: BASIC
Description of FTP commands, as of 2/5/77
FTP is a utility program for transferring files over the ARPA network.
The name of the program comes from the name of the protocol used
(File Transfer Protocol).
* Menu:
* BASIC:: The basic FTP commands.
* TYPE IN:: FTP's typein conventions.
* LESS:: Some less useful commands.
* DEFAULTS:: How FTP does file name defaulting.
* RARE:: The rarely used commands.
* JCL:: Giving FTP commands from DDT.
* XFILE:: Command files.
* INFERIOR:: Programs can use FTP as a subroutine.

File: FTP, Node: BASIC, Previous: TOP, Up: TOP, Next: TYPE IN
The Most Useful FTP Commands:
Note that unambiguous abbreviations may be used for command and
host names. See the Next node for how to edit your input.
CONNECT <host> Connects to a foreign host. Use the name of the host
or its (octal) numeric address. A pre-existing
connection will be flushed.
FTP prompts with "$$" when it is not connected to
a foreign host, and with "$" when a connection exists.
<host> Is equivalent to CONN <host>.
HOSTS Lists the names of hosts which FTP knows
LOGIN <name> Logs in at foreign host (if needed by that host)
If you get an error message implying that you are not
logged in, this is where to start. Sometimes FTP
will figure out by itself that logging in is what you
have to do; then it will ask for for a user name,
send it, and go ahead with the operation you had
already requested.
TEN Mode command for pdp10 36 bit binary files.
(May also be used for pdp10 ascii files if you don't
mind extraneous characters at the end of the file)
NOTE: This is the DEFAULT mode when communicating with
another PDP10. You can specify it explicitly, also,
with any computer (although some may not accept it).
It is an abbreviation for TYPE I and BYTE 36
TEXT Mode command for Network ASCII data transmission.
This mode is the DEFAULT when communicating with a
machine that is not a PDP10. It is an abbreviation
for TYPE A and BYTE 8.
GET or RETRIEVE Transfers a file from foreign host to here.
Asks for local name (end with CR) and then foreign name.
PUT or STORE Transfers a file from here to foreign host.
or SEND Asks for local name (end with CR) and then foreign name.
APPEND Appends a local file onto a file at the foreign host.
Arguments are local file (not changed by the operation)
followed by the foreign file (which will be created if
it does not exist), as for PUT.
TRANSFER <host1><file1>=<host2><file2>
Transfers <file2> at <host2> to <file1> at <host1>.
Two foreign hosts can't be used, though; one of
<host1> and <host2> must be the local host's name.
Abbreviations are allowed: You can leave out the
name of the local host (omitting the altmode as well),
as in TRAN FOO BAR=MCFOO BAR or TRAN MCFOO BAR=X Y,
and if the two filespecs are identical either one may
be omitted, as in TRAN FOO BAR=MC which is equivalent
to TRAN FOO BAR=MCFOO BAR.
NOTE: "_" is equivalent to "=". To include a "=" or
"_" or a ^Q in a filename in a TRAN command, precede
it with a ^Q (which will be gobbled up as a quote).
^Q's followed by other characters (not _, = or ^Q)
will NOT be absorbed by TRAN (so you can use them to
quote at other levels, with hosts that understand it).
QUIT Closes connection to foreign host, and then exit from FTP
^C Is the same as QUIT - no CR is needed.
Any command may be terminated with a ^C instead of a
CR, in which case that command will be followed by
a QUIT (unless it encounters an error).
DISC Closes connection to foreign host. You may then connect
to another host, or QUIT, or ^Z...
HELP or ? Types a list of commands, with brief descriptions.

File: FTP, Node: TYPE IN, Up: TOP, Previous: BASIC, Next: DEFAULTS
Type-in Conventions:
You can use Rubout to delete one character, ^D, ^G or ^U to
cancel an entire line, and ^L to redisplay the current line
as typed so far. Commands with no arguments should be terminated
with a CR. When giving a command with arguments, you can
separate the command name from the first argument with either
a space or a CR (if you use CR, FTP will tell you what sort
of argument is expected). When a command requires several
arguments, the arguments are separated with CRs. Command names may be
abbreviated as long as the abbreviations are unambiguous.
You can use a ^C instead of CR to terminate a line. That
causes FTP to return to DDT, killing itself, after the current
command is finished (as long as it does not get an error).
If you plan to disown a FTP while it is transferring, you should
end the last command with a ^C so that the FTP will go away
when it is finished; otherwise it might hang around for an hour
until ITS kills it.
You can use ^N instead of CR to terminate a line. This is not
especially useful when you are typing on the TTY, but it is
useful when passing commands from DDT, since it allows you to
put several lines' worth of input into the one line that DDT lets
you send.

File: FTP, Node: DEFAULTS, Up: TOP, Previous: TYPE IN, Next: LESS
Filename Defaulting:
Local filenames default "stickily", so that each file's device,
sname, and first and second filenames all default to the last
such names specified for a local file.
Giving a null argument for the local file will cause the default
file to be used.
When you specify only one filename, it is the first filename,
and ">" is used as the second.
The initial default directory is your working directory.
If you wish to include a space, colon, or semicolon in a local
filename (or in a foreign filename if the foreign site is an ITS),
precede it with a ^Q. The same technique is used for including
a = or _ or ^Q anywhere in a pathname, in the TRAN command.

File: FTP, Node: LESS, Previous: DEFAULTS, Up: TOP, Next: RARE
Less useful commands:
DELETE <file> Deletes the specified file at the foreign host.
RENAME <oldname>
<newname>
Renames the file <oldname> at the foreign host to be
called <newname>. Separate the two names with a CR.
DIRECTORY <dir> or
LISTF <dir> Types out a listing of the foreign host's directory <dir>.
Some hosts may also allow you to specify a filename,
in which case info on that file will be given.
LISTB <dir> Types out a brief listing of a foreign directory.
It is like LISTF except that only the filenames are
typed - no other info on the files.
LISTL <dir> Types out a listing of the directory <dir> at
the local host.
CWD <dir> Sets the foreign default directory to <dir>
DEFAULTS <filenames>
Sets the local filename defaults to <filenames>.
Note: the local defaults are sticky; each local
filename's names default to the last filenames
specified for a local file.

File: FTP, Node: RARE, Previous: LESS, Up: TOP, Next: JCL
Rarely Useful Commands:
STATUS <arg> Asks for a status message from the foreign host.
The meaning of <arg> is site-dependent. ITS sites
don't implement the command at all.
QUOTE <command> Sends <command> directly to the foreign host over
the TELNET connection. Used if the foreign host
implements some simple command that FTP doesn't know
about itself. It won't work for commands that do
anything hairy (such as use the data connection at
all) since FTP doesn't expect it to do that.
SOAK Waits for and describes one reply from the server.
This is never necessary with any of the normal
FTP commands, since they know when to wait for
replies.
DEBUG Toggles the switch which says whether to print expected
replies from the server (unexpected replies such as
error messages and notices to the user are always
printed). The DEBUG switch also causes reply codes
to be given with all replies, not just error messages.
PASS <password> Sends a password to a foreign host. Note that after
a LOGIN command, if a password is needed, FTP will
usually be able to figure that out; it will ask you
for the password and send it automatically, so you
need not explicitly give a PASSWORD command.
ACCT <account> Sends an account number to the foreign host. Note
that if after a LOGIN command an account number is
needed, FTP will usually figure that out and just
ask you for it. So you usually won't need to give
an explicit ACCT command.
ALLOCATE <size> Sends the foreign host the estimated size of a file
you are about to send it (measured in bytes). Most
hosts don't need ALLOCATE commands since they can
allocate file storage dynamically.
TYPE <type> Says what transfer type to use. Of the many types
specified in the FTP protocol, only A (ASCII) and
I (Image) are implemented on ITS. You usually
would rather use the TEXT or TEN commands, since
they specify the two most useful combinations of
TYPE and BYTE.
BYTE <size> Says what byte size to use for transfers. The only
byte sizes implemented on ITS for FTP'ing are 8,
32 and 36. In any case, you should usually use the
TEXT and TEN commands for mode-setting.
ICPSOCKET <number>
Causes FTP to try to connect to servers using socket
<number> rather than socket 3, the standard FTP server
listen socket. Useful for testing experimental
versions of the server, not actually installed yet.
SILENT Tells FTP not to type out on the terminal at all.
This remains in effect until FTP tries to read input
from the terminal. It does not affect printing in a
script file.
PROCEED Tells FTP to continue running without the terminal.
FTP accomplishes this by valretting a ^P command. A
SILENT is done automatically. This command is illegal
when commands are being read from the terminal, unless
TTY input was actually translated to a file.
DISOWN Tells FTP to continue running disowned. Just like
PROCEED except that the FTP is disowned as well as
^P'd.
VALRET Returns to FTP's superior with a .BREAK 16,100000
(which does not kill the FTP). The FTP can be
proceeded, or it can be given new JCL and restarted.

File: FTP, Node: JCL, Up: TOP, Previous: RARE, Next: XFILE
Giving FTP commands from DDT:
If you do :FTP <command> and <command> contains an = or a _,
FTP automatically does a TRAN command with <command>
as arguments. After that transfer completes, FTP
kills itself. For example, :FTP FOO BAR=MC will
transfer FOO BAR from MC to the local host, leaving
you back in DDT.
If the command string from DDT does not contain an = or a _,
it is simply executed as an FTP command. FTP will then read
more commands from the TTY, since one FTP command (unless it
is TRAN) is not usually enough for a whole transaction.

File: FTP, Node: XFILE, Up: TOP, Previous: JCL, Next: INFERIOR
Command Files:
XFILE <file> Tells FTP to open <file> and read commands from it.
Commands in an execute file look just like commands
typed on the TTY. Linefeeds are ignored, so you ought
to use CRLF between lines. Any error (such as a
failure to establish a connection or a failure to find
a file) causes reading from the command file to be
discontinued.
SCRIPT <file> Tells FTP to open <file> and print init everything
that appears on the TTY. If there is no command file
(commands are still being read from the TTY) then
output continues to go to the TTY as well as the
script file. If there is a command file and a script
file, then nothing is output to the TTY. If you wish
to have an FTP run without your attention, you need to
have both a script file and a command file.
ESCRIPT Tells FTP to stop using a script file. If there is a
script file open, it is closed. This isn't necessary
if you are doing a QUIT or ^C.
One reasonable way to operate is to give FTP a SCRIPT command and then
an XFILE command from the TTY, and then ^P the FTP. If the command
file does not do a QUIT then when it is finished you will be told
that "FTP wants the TTY". Alternatively, the command file might
contain a PROCEED command which would do the ^P automatically.

File: FTP, Node: INFERIOR, Up: Top, Previous: XFILE
Use of FTP as a Subroutine
You can invoke FTP as an inferior to transfer some files as long as
you provide it with a source of commands and a place to list its
commands and replies. The command source can be provided either by
translating TTY input to a file, or by giving an XFILE command as JCL.
A file for listing can be provided either by translating TTY output to
a file or by including a SCRIPT command in the JCL. If you are not
interested in looking at the output (you trust that the foreign hosts
will be available, etc) then you can use NUL: as an output sink. The
SILENT command may be helpful in suppressing inconvenient typeout that
might occur before you have both a command file and a script file set
up. Either way, you do not need to give FTP actual ownership of the
TTY.
If you wish to include all the commands in the JCL string, there is
no need for a command file. There will still be a need for a script
file, however. Multiple commands in JCL are separated either with
CRLF or with ^N. Note, however, that commands that came from the JCL
will not be echoed to the script file, while commands from an XFILE or
a translated TTY will be echoed.
If you did not translate TTY input (you gave a XFILE or put all the
commands in the JCL) then on any error FTP will execute a .BREAK 16,
which specifies that type-ahead is to be discarded. If you are not
watching the FTP replies, this can be used as an indication that there
was recently an error.
If you wish to use a single FTP job for several operations, not all
at once, the VALRET command is useful. It makes FTP stop and return
control to its superior with a .BREAK 16,100000, without altering any
of its state variables. The VALRET command should go in the command
stream at the end of the commands for a single run. After the FTP
returns, it can be proceeded (to continue with commands from the same
source) or it can be given some fresh JCL and restarted from its
normal starting address.

1356
src/klh/nuuos.205 Executable file
View File

File diff suppressed because it is too large Load Diff

2709
src/klh/out.250 Executable file
View File

File diff suppressed because it is too large Load Diff

80
src/ksc/nfnpar.7 Executable file
View File

@@ -0,0 +1,80 @@
if1 ifndef %%scan,.insrt ksc;nscan >
if2 ife %%scan,.insrt ksc;nscan >
fnpard:
fnpart=fnpard+1
.begin fnpar ; filename parser. addr of string in 1
; returns dev/sname/fn1/fn2 in accs 1,2,3,4
tdca b,b ;FNPARD entry point = DDT type
seto b, ;FNPART entry point = TECO type
movem b,fmode' ;set type
movem a,finstr' ;save addr of input string
setzm fdev' ;clear accumulated vars
setz b, ;fdir
setzb c,d ;fn1,fn2
fnpr20: push p,[brkchr']
push p,[fstr]
push p,[fbrktb]
push p,finstr
pushj p,scanrt
move a,brkchr
cain a,40
jrst [pushj p,fcvstr ? pushj p,ffnput ? jrst fnpr20]
cain a,";
jrst [pushj p,fcvstr ? move b,a ? jrst fnpr20]
cain a,":
jrst [pushj p,fcvstr ? movem a,fdev ? jrst fnpr20]
pushj p,fcvstr ;handle EOF or foreign char
pushj p,ffnput
move a,fdev
popj p,
ffnput: jumpe a,[popj p,] ;nothing to deposit
skipn fmode ;DDT or TECO style?
jrst [jumpe c,[move c,a ? popj p,] ;DDT
move d,a ? popj p,]
jumpe d,[move d,a ? popj p,] ;TECO
move c,d
move d,a
popj p,
;break table
fbrktb: cain a,^Q
jrst scbrqt
cain a,":
jrst scbrsk
cain a,";
jrst scbrsk
cain a,40
jrst scbrsk
cain a,^M
jrst scbrsk
cain a,^J
jrst scbrsk
jrst (b)
.scalar fstr(2)
;convert string (ptr in fstr) to sixbit wd in a
fcvstr: pushae p,[b,c,d]
setz a,
move c,[440600,,a]
hrrz d,fstr ;get cnt
fcvst5: sojl d,fcvst9
ildb b,fstr+1
cain b,^Q ; if quoter,
jrst fcvst5 ; don't deposit it.
cail b,"a
caile b,"z
caia ;not lowercase
subi b,40 ;cvt to uppercase
subi b,40 ;cvt to 6bit
idpb b,c ;deposit
tlne c,770000
jrst fcvst5 ;if sixbit word not exhausted, get more chars
fcvst9: popae p,[d,c,b]
popj p,
.end fnpar

117
src/ksc/nscan.8 Executable file
View File

@@ -0,0 +1,117 @@
IF1 %%SCAN==0 ;so files .insrt'ing can do right thing; should
IF2 %%SCAN==1 ;have their .INSRT as follows:
; if1 ifndef %%scan, .insrt ksc;nscan >
; if2 ife %%scan, .insrt ksc;nscan >
ifndef pushae,[
DEFINE PUSHAE AC,LIST
IRP LOC,,[LIST]
PUSH AC,LOC
TERMIN
TERMIN
DEFINE POPAE AC,LIST
IRP LOC,,[LIST]
POP AC,LOC
TERMIN
TERMIN
]
DEFINE SCAN ?INPUT,BRKTBL,RESULT,BRKCHR
PUSH P,[BRKCHR]
PUSH P,[RESULT]
PUSH P,[BRKTBL]
PUSH P,[INPUT]
PUSHJ P,SCANRT
TERMIN
%BRKAD==1 ;break char flag for appending to result
SCBRAD: TLO A,%BRKAD
JRST 2(B) ;return to right place in SCANRT
%BRKSK==2 ;break char flag for skip of break char
SCBRSK: TLO A,%BRKSK
JRST 2(B)
%BRKLV==4 ;break char flag for leaving on front of input
SCBRLV: TLO A,%BRKLV
JRST 2(B)
%BRKQT==10 ;break char flag for gobbling following char
SCBRQT: TLO A,%BRKQT
JRST 2(B)
SCANRT:
.begin scnblk
define arg1
(p)-5!termin
define arg2
arg1-1!termin
define arg3
arg2-1!termin
define arg4
arg3-1!termin
PUSHAE P,[A,B,C,D]
MOVE D,ARG1 ;get addr of input string
MOVE C,1(D) ;get BP
HRRZ D,(D) ;get char cnt
JUMPE D,[MOVE A,[-1,,3] ;none left? Then EOS, no result string.
MOVEM A,@ARG4 ;store -1,,3 as "break char"
SETZM @ARG3 ;and clear 1st descriptor wd of result string (count)
JRST SCNR90] ;and return.
SCNR20: ILDB A,C ;get char from input string
JSP B,@ARG2 ;execute break table.
SOJG D,SCNR20 ;no break, continue
JRST SCNR70 ;again none left, EOS.
;break char found! decipher flag(s) and return
TLNE A,%BRKQT ; quote next
JRST SCNR45
TLNE A,%BRKSK
JRST SCNR50 ;skip break char.
TLNE A,%BRKAD
JRST SCNR55 ;go append break char to result
TLNE A,%BRKLV
JRST SCNR60 ;leave break char on front of input.
JRST SCNR20 ;no flags found?! continue.
; quote next char
SCNR45: SOJLE D,SCNR70 ; jump if none left.
ILDB A,C ; get it, ignore it.
SOJG D,SCNR20 ; get next.
JRST SCNR70
;skip break char - goes on neither result nor input.
SCNR50: SOS @ARG1 ;D is subtracted from this for result cnt, so ensure brkchar
;isn't included within result.
;append break char to result string
SCNR55: SUBI D,1 ;D becomes new input cnt, so ensure brkchr isn't included.
JRST SCNR80 ;and return thusly.
;retain break char on front of input string
SCNR60: ADD C,[70000,,0] ;must decrement bp to point back at brkchar.
CAIG C,
SUB C,[430000,,1] ;(reset if go off wd edge)
JRST SCNR80
;EOS found and non-null result string.
SCNR70: MOVE A,[-1,,3] ;"break char".
MOVEM A,@ARG4
JRST .+2
;return result and update input as determined by count in D.
;if control came directly here, result string would contain everything up to
;break char, and input string cnt would include brk char but bp would point
;immediately after it.
SCNR80: HRRZM A,@ARG4 ;store break char
MOVE A,ARG1 ;get addr of input string again.
MOVE B,(A) ;get 1st wd of descriptor before clobbering.
HRRM D,(A) ;store updated input string count.
SUB B,D ;get descriptor with proper result string cnt
EXCH C,1(A) ;exchange bp's to update input and get bp for result.
MOVE A,ARG3 ;get addr of result string now
MOVEM B,(A) ;store 1st descriptor wd
MOVEM C,1(A) ;and 2nd (bp).
SCNR90: POPAE P,[D,C,B,A]
SUB P,[5,,5] ;flush args and return addr
JRST @5(P) ;and return.
.end scnblk

879
src/ksc/pagser.49 Executable file
View File

@@ -0,0 +1,879 @@
; -*- Mode: MIDAS -*-
.BEGIN PAGSER ; New page/core manager.
IFNDEF $$PDBG,$$PDBG==0 ; 1 - include debugging printout routines
IFNDEF $$PPGS,$$PPGS==0 ; 1 - include GETPAG
comment |
PAGSER is a storage allocation management package. It will
allocate, expand, and release areas of memory in units of either pages
or words. All the core which PAGSER knows about is divided into
"blocks" which are either "active" (in use by the program) or
"unactive" (free, available for allocation requests). At
initialization, PAGSER will control a single large "unactive" block;
all future allocate requests will be satisfied from this primal block.
Because PAGSER's data structures are themselves allocated
dynamically, there is no space problem for them. Because they are
localized and distinct from the areas they refer to, page faults
are minimized and allocation can be done with respect to the theoretical
address space only, without implying that anything actually exists
within a block. This is useful for programs which need to dynamically
hack their page map.
In PAGSER's data structures, each block (active or not) is
described by a single node, which is $PNSIZ words long and
has the following format:
$PNPTR 0: <C-pred>,,<C-succ> ; Corelist pointers
$PNSPP 1: <S-pred>,,<S-succ> ; Spacelist pointers (0 if active)
$PNFLG 2: <ID> ,,<unused> ; Consistency check.
$PNBLK 3: <# wds>,,<addr> ; Size & location of block.
There are three lists composed of these nodes: the Corelist, Spacelist,
and Freelist.
The FREELIST is simply a list of all unused nodes, which describe
no blocks at all. When a node is required (e.g. when a new block is
created, and needs a new node to describe it) one is procured from
this list. Freelist nodes use the RH of the first word to point
to the next free node; zero implies no more.
The CORELIST holds ALL nodes which represent a block, in an order
corresponding EXACTLY to their sequence in memory. <C-pred> points at
the node for the block preceding this one in core, and <C-succ>
similarly points at the node for the block immediately succeeding this
one in core. The Corelist is used to determine the state of blocks
preceding and succeeding a particular block, and always completely
describes the state of the entire section of memory entrusted to
PAGSER.
The SPACELIST is an efficiency hack. The <S-pred> and <S-succ>
pointers are used to string together all UNACTIVE blocks, so that when
an allocation request is made, no time is wasted checking active
blocks. An "active" block is one for which the Spacelist pointer word
($PNSPP) is zero. The current algorithm for constructing and searching
the Spacelist favors the most recently freed blocks.
About $PNBLK, note that <# wds>,,<addr> specify the location and
length of the core block that the node represents. There need not
necessarily be any actual core there, especially for the LAST block
on the Corelist (<C-succ> zero). The code tries to keep unmapped
all pages in the last block, if it is unactive.
$PNFLG is primarily used for verifying that a "node address",
as furnished to some PAGSER routines by the user program,
actually does refer to an active PAGSER node. This helps to quickly
catch any bugs that give invalid arguments to the core routines,
which otherwise might be extremely hard to uncover. It isn't
foolproof, but is much better than nothing.
|
; Indices into a node.
OFFSET -.
$PNPTR:: 0 ; <pred>,,<succ> for Corelist
$PNSPP:: 0 ; <pred>,,<succ> for Spacelist
$PNBLK:: 0 ; <# wds>,,<addr>
$PNFLG:: 0 ; <ID chk>,,<flags> ; RH not really used now.
; $PNPGS:: 0 ; <# pages included>
$PNSIZ:: ; Size of a node.
OFFSET 0
$PVLID==:<(SIXBIT /PSR/)> ; ID value used in LH of $PNFLG.
; Note only used for active nodes.
; Other random parameters
IFNDEF $PMAKN,$PMAKN==:40 ; # nodes to create each time need more.
IFNDEF $PTRIV,$PTRIV==:10 ; # words OK to waste when satisfying request.
; Page size parameters
IFNDEF PG$BTS,{ IFN OS%TNX, PG$BTS==:9.
IFN OS%ITS, PG$BTS==:10.
}
PG$SIZ==:1_PG$BTS
PG$MSK==:PG$SIZ-1
;;; PSRERR is an error macro which can be skipped over.
;;; The user may define this for us, otherwise it is a .VALUE.
;;; PSRERR is called with the PAGSER error code.
IFNDEF PSRERR,[
DEFINE PSRERR CODE
.VALUE [CODE]
TERMIN
];IFNDEF PSRERR
;;; Preliminary list of error codes.
ERBAD==0 ;Internal error
ERARG==1 ;Bad argument
EROOM==2 ;Not enough room
;;; User may also supply the similar macro CORLUZ,
;;; which is called when a CORBLK fails.
subttl Initialization - PSRINI
.SCALAR SPCLST ; Ptr to Spacelist
.SCALAR CORLST ; Ptr to Corelist
.SCALAR FRELST ; Ptr to Freelist, 0 when none left.
.SCALAR FRESAV ; Ptr to saved node, used when getting block for more.
.SCALAR FFPAGE ; # of first free page
.VECTOR INILST($PNSIZ) ; Initial list node, set up pointing at free core.
.VECTOR INIFSV($PNSIZ) ; Initial scratch node for FRESAV to point at.
; PSRINI - Initialize PAGSER routines
; U1 - -<# pages>,,<1st page #> defines area available for hacking.
.M"PSRINI:
.M"CORINI:
PUSH P,U1
JUMPGE U1,.+3 ; If already positive skip convert hack.
TLC U1,-1
ADD U1,[1,,] ; Make count positive
TRNE U1,-1 ; Make sure page # isn't 0.
TLNN U1,-1 ; Likewise make sure # pages isn't 0.
PSRERR ERARG ; Ugh, zero component!
HRRZM U1,FFPAGE
LSH U1,PG$BTS ; Get in terms of words
MOVEM U1,$PNBLK+INILST ; Initialize count of first, unactive, node.
MOVSI U1,SPCLST-$PNSPP
MOVEM U1,$PNSPP+INILST ; and initialize spacelist pointer
MOVSI U1,CORLST-$PNPTR
MOVEM U1,$PNPTR+INILST ; and corelist pointer
MOVEI U1,INILST ; Initialize
MOVEM U1,SPCLST ; Spacelist, and
MOVEM U1,CORLST ; Corelist, and
SETZM FRELST ; Freelist.
MOVEI U1,INIFSV
MOVEM U1,FRESAV ; Point to extra node...
SETZM INIFSV ; Clear it out.
MOVE U1,[INIFSV,,INIFSV+1]
BLT U1,INIFSV+$PNSIZ-1
POP P,U1
POPJ P,
;;; PSRP - See if enough space available.
;;; U1 - # words needed
;;; Skips if enough space is available, does not skip if not enough.
.M"PSRP: ;Bash UUO ACs with abandon!
SKIPN U2,SPCLST ;Get spacelist.
JRST PSRP9 ; If missing take failure return.
PSRP2: HLRZ U4,$PNBLK(U2) ;Get core spec.
CAILE U1,(U4) ;Loop until we find one big enough.
JRST [ HRRZ U2,$PNSPP(U2)
JUMPN U2,PSRP2
JRST PSRP9 ] ; If no more left, failure return.
AOS (P) ;On success, skip return.
PSRP9: RET
subttl PSRGET - Get a block
; PSRGET - Get core.
; U1 - # words needed
; Returns
; U1 - actual # words available in block
; U2 - <addr of block>,,<node addr> ; COREXP - 0,,<addr of blk>
.M"CORGET:
PUSHJ P,PSRGET
HLRZ U2,U2 ; Return block addr only.
POPJ P,
.M"PSRGET:
PUSH P,U3
PUSH P,U4
SKIPN U2,SPCLST
PSRERR ERBAD ; No spacelist???
PSRGT2: HLRZ U4,$PNBLK(U2) ; Get core spec
CAIG U1,(U4)
JRST PSRGT4 ; Found block big enough...
HRRZ U2,$PNSPP(U2) ; Nope, get next
JUMPN U2,PSRGT2
PSRERR EROOM ; Failed to find big enough block?!
; Found block big enuf. See if want to split or swallow whole.
PSRGT4: CAILE U4,$PTRIV(U1) ; Excess greater than some trivial value?
JRST [ PUSHJ P,PSRADJ ; Yes, split block...
JRST PSRGT8]
; Here when block needn't be split up. Must remove from spacelist.
MOVEI U3,(U2)
IFN OS%ITS,{ ; If on ITS,
MOVE U4,$PNPTR(U3) ; and gobbling very last block,
TRNN U4,-1 ; we're bypassing PSRADJ
PUSHJ P,PSRCOR ; so must make sure core exists for it.
}
PUSHJ P,PSRACT ; Remove from spacelist.
; All done with split or whatever, U1 still has size and U2
; address of node for block we want.
PSRGT8: HLRZ U1,$PNBLK(U2) ; Update U1 to actual size
HRL U2,$PNBLK(U2) ; and U2 to <block addr>,,<node addr>.
POP P,U4
POP P,U3
POPJ P,
subttl GETPAG - Get pages.
ifn $$ppgs,{
; GETPGN - Variant of GETPAG that returns negative rather than positive
; count in LH, AOBJN style.
.M"PAGGTN:PUSHJ P,PSRPGT
TLC U1,-1
ADD U1,[1,,]
POPJ P,
; GETPAG - Get block of pages.
; U1 - # of pages desired
; Returns
; U1 - <# pages>,,<1st page #>
.M"GETPAG:
.M"PSRPGT:
PUSH P,U2
PUSH P,U3
PUSH P,U4
LSH U1,PG$BTS ; Get # words
SKIPN U2,SPCLST
PSRERR ERBAD ; No spacelist??
PSRPG2: HLRZ U4,$PNBLK(U2)
CAIG U1,(U4)
JRST PSRPG4
PSRPG3: HRRZ U2,$PNPTR(U2)
JUMPN U2,PSRPG2
PSRERR EROOM ; Couldn't find enough pages!!
PSRPG4: MOVE U3,$PNBLK(U2) ; Get address core block starts at
ANDI U3,PG$MSK ; Find # words past page boundary it starts at
CAIE U3,
SUBI U3,PG$SIZ ; Get -<# words to next boundary>
ADD U4,U3 ; Thus get # wds in block past first boundary
CAIGE U4,(U1) ; Compare (U4 may be neg), have enuf words?
JRST PSRPG3 ; Nope, keep searching.
MOVMS U3
CAILE U3,$PTRIV
JRST [ PUSH P,U1
PUSH P,U4
MOVM U1,U3
PUSHJ P,PSRADJ ; Split off low end.
EXCH U2,U3 ; Get low-end node in U3, and
PUSHJ P,PSRMKU ; put it on spacelist.
POP P,U4
POP P,U1
JRST .+1]
CAILE U4,$PTRIV(U1) ; Need to split things up at high end?
JRST [ PUSHJ P,PSRADJ ; Get block of right size.
JRST PSRPG8]
;IFN OS%ITS,{ ; If on ITS,
; MOVE U4,$PNPTR(U3) ; and gobbling very last block,
; TRNN U4,-1 ; we're bypassing PSRADJ
; PUSHJ P,PSRCOR ; so must make sure core exists for it.
;}
PSRPG8: TRNE U1,PG$MSK
PSRERR ERARG ; Just a check...
LSH U1,<18.-PG$BTS> ; Before shifting # pages out into LH.
HRRZ U3,$PNBLK(U2)
ADDI U3,PG$SIZ-1
LSH U3,-PG$BTS
HRRI U1,(U3) ; Now have <# pages>,,<page # starts at>
POP P,U4
POP P,U3
POP P,U2
POPJ P,
} ; end ifn $$ppgs
subttl PSREXP - Expand block
; PSREXP - Expand core block.
; U1 - # wds to increase by
; U2 - node of block to increase ; COREXP - addr of block
; Returns
; U1 - new # wds total in block
; U2 - <new addr of block>,,<new node> ; COREXP - 0,,<new addr of blk>
comment |
There are 4 cases possible, listed in order of optimality:
1) Enough room in unactive successor.
Merge and perhaps split.
2) Enough room in unactive predecessor.
Must copy up and perhaps split.
3) Enough room if both predecessor and successor combined.
Must copy up, then merge with following, then perhaps split.
4) No free blocks on either side, or not enough room.
Must find completely new block and copy into it. Old
block must be released.
|
.M"COREXP:
PUSHJ P,PSREQV ; Chg blk addr to node addr for compatibility
PUSHJ P,PSREXP
HLRZ U2,U2 ; Return block addr only.
POPJ P,
.M"PSREXP:
PUSH P,U3
PUSH P,U4
; Verify that node is kosher.
HLRZ U3,$PNFLG(U2) ; Get ID-check value from where it should be
CAIE U3,$PVLID ; and make sure it's what it should be...
PSRERR ERBAD ; Yuckity yuck!
; See if enough room in successor.
MOVE U3,$PNPTR(U2) ; Get ptr to successor
TRNE U3,-1 ; Not there?
SKIPN $PNSPP(U3) ; or Active?
JRST [HLRZS U3 ; Yes, can't hack succ, check pred.
CAIN U3,CORLST-$PNPTR
JRST PSRXP7 ; Not there?
SKIPN $PNSPP(U3)
JRST PSRXP7 ; Can't hack pred either, need new block.
HLRZ U4,$PNBLK(U3)
SUBI U4,(U1)
JUMPL U4,PSRXP7 ; If not enough words, also can't hack.
JRST PSRXP2] ; Can, go hack predecessor.
HLRZ U4,$PNBLK(U3) ; Check successor - get # words in it.
SUBI U4,(U1) ; Find diff.
JUMPGE U4,[HLRZ U4,$PNBLK(U2)
ADDI U1,(U4)
JRST PSRXP4] ; Jump if successful.
HLRZS U3 ; Fooey, but see if predecessor can save day.
CAIE U3,CORLST-$PNPTR ; No predecessor?
SKIPN $PNSPP(U3) ; or not available?
JRST PSRXP7 ; Ugh, must find whole new block.
HLRZ U3,$PNBLK(U3) ; Get # words
ADDI U4,(U3)
JUMPL U4,PSRXP7 ; Again, jump if must get new block.
; Here, can win by combining pred & succ blocks.
PSRXP2: HLRZ U3,$PNPTR(U2) ; Get ptr to pred.
PUSHJ P,PSRACT ; Make U3 block active.
EXCH U2,U3
PUSHJ P,PSRCPY ; Copy into U2 from U3.
HLLZ U4,$PNBLK(U3) ; Get count of old block
ADDM U4,$PNBLK(U2) ; Add into pred block.
HLRZS U4 ; While still have count,
ADDI U1,(U4) ; Find total # words required
PUSHJ P,PSRFRX ; Put U3 block on Freelist.
HLRZ U4,$PNBLK(U2) ; Find how much room we have now.
CAIG U1,(U4)
JRST PSRXP6 ; Aha, got enough.
; Here, can win by adding successor block.
PSRXP4: HRRZ U3,$PNPTR(U2) ; Get succ
JUMPE U3,PSRXP7
PUSHJ P,PSRMRG ; Merge successor in with U2 block.
; Here have block all together, see if want to split some off.
PSRXP6: HLRZ U4,$PNBLK(U2) ; Find total # words we have
CAIGE U4,(U1)
PSRERR EROOM ; somehow didn't get enough???
CAIG U4,$PTRIV(U1) ; More than a trivial amount?
JRST PSRXP8 ; No, return.
PUSHJ P,PSRADJ ; Split them up...
PUSHJ P,PSRMKU ; Put split-off block in U3 on spacelist.
JRST PSRXP8
; Here, must find a whole new block.
PSRXP7: HLRZ U4,$PNBLK(U2)
ADDI U1,(U4) ; Find # words total we need.
MOVE U3,U2 ; Save U2
PUSHJ P,PSRGET ; Get a block of right size (node ptr in U2)
PUSHJ P,PSRCPY ; Copy U3 into U2.
MOVEI U1,(U3)
PUSHJ P,PSRREL ; Free up old core.
PSRXP8: HLRZ U1,$PNBLK(U2) ; Return U1 - <# words>
HRL U2,$PNBLK(U2) ; Return U2 - <block addr>,,<node addr>
POP P,U4
POP P,U3
POPJ P,
subttl PSRREL - Release Block
; PSRREL - Release core block.
; U1 - addr of node to release. (CORREL - addr of block)
.M"CORREL:
PUSH P,U2
MOVEI U2,(U1)
PUSHJ P,PSREQV ; Temp hack - change block addr to node addr
JRST PSRRL1
.M"PSRREL:
PUSH P,U2
MOVEI U2,(U1)
PSRRL1: PUSH P,U3
PUSH P,U4
; Verify that node is kosher.
HLRZ U3,$PNFLG(U2) ; Get ID-check value from where it should be
CAIE U3,$PVLID ; and make sure it's what it should be...
PSRERR ERBAD ; Yuckity yuck!
HLRZ U3,$PNPTR(U2) ; Get <pred>
CAIE U3,CORLST-$PNPTR ; Special check before taking look at "node".
SKIPN $PNSPP(U3) ; Is predecessor unactive?
JRST PSRRL2 ; No, or not real node, don't merge.
EXCH U2,U3
PUSHJ P,PSRMRG ; Merge if so
PSRRL2: HRRZ U3,$PNPTR(U2) ; Get <succ>
JUMPE U3,PSRRL4
SKIPE $PNSPP(U3) ; Is successor unactive?
PUSHJ P,PSRMRG
SKIPN $PNSPP(U2) ; Is block itself on spacelist by now?
PUSHJ P,PSRMK2 ; Make it unactive, put on spacelist.
; Could check here for freeing up pages.
PSRRL4:
PSRRL8: PUSHJ P,PSRGC
POP P,U4
POP P,U3
POP P,U2
POPJ P,
; PSREQV - Temporary hack which converts a block address to a node address,
; so as to maintain compatibility with CORSER. In future the node
; address will be returned to caller of CORGET so as to eliminate this
; searching overhead.
; Takes U2 - block addr
; Returns U2 - node addr for the block
PSREQV: PUSH P,U3 ? PUSH P,U4
MOVEI U3,CORLST-$PNPTR
JRST PSREQ4
PSREQ2: HRRZ U4,$PNBLK(U3)
CAIN U2,(U4)
JRST PSREQ6
PSREQ4: HRRZ U3,$PNPTR(U3)
JUMPN U3,PSREQ2
PSRERR ERBAD ; No block corresponds to this address!!
PSREQ6: MOVEI U2,(U3)
POP P,U4 ? POP P,U3
POPJ P,
subttl Primitives - OS dependent page hacking
; PSRGC - Frees up core from U2 block insofar as possible.
; U2 - block just freed
; Clobbers U3,U4
PSRGC: MOVE U4,$PNPTR(U2)
TRNE U4,-1
POPJ P,
MOVE U4,$PNBLK(U2) ; Get <# wds>,,<addr>
MOVEI U3,PG$SIZ-1(U4)
LSH U3,-PG$BTS ; Find 1st page # freed.
MOVE U4,FFPAGE
SUBI U4,(U3)
JUMPLE U4,PSRGC9
IFN OS%ITS,{
IMUL U4,[-1,,]
HRRI U4,(U3)
SYSCAL CORBLK,[MOVEI 0
MOVEI %JSELF
U4]
PSRERR ERARG ; Should only happen if bad args.
}
IFN OS%TNX,{
PUSH P,1 ? PUSH P,2 ? PUSH P,3
SETO 1,
MOVEI 2,(U3)
HRLI 2,.FHSLF
MOVEI 3,(U4)
TLO 3,(PM%CNT)
PMAP
ERJMP .+1
POP P,3 ? POP P,2 ? POP P,1
}
MOVEM U3,FFPAGE
PSRGC9: POPJ P,
; PSRADJ - Splits up block of core into 2 blocks.
; This is the ONLY PLACE where new nodes/blocks are added to
; the corelist!!
; U1 - # words to put in active new block
; U2 - ptr to block to split up (unactive, on spacelist)
; Returns
; U2 - ptr to active new block
; Clobbers U3,U4
PSRADJ: MOVEI U3,(U2) ; Put big unactive node in U3
PUSHJ P,PSRGTN ; Get a new node in U2
HLRZ U4,$PNBLK(U3)
SUBI U4,(U1)
CAIG U4, ; Make sure block is big enough for split...
PSRERR ERARG ; Ugh?? Caller messed up.
HRLM U4,$PNBLK(U3) ; Store reduced size for big unactive block
HRLM U1,$PNBLK(U2) ; and set desired size for new block.
MOVE U4,$PNBLK(U3) ; Get addr of big block,
HRRM U4,$PNBLK(U2) ; and store it as addr for new block
ADDI U4,(U1) ; and find new addr of big unactive block.
HRRM U4,$PNBLK(U3) ; Store away, all block-size specs done.
; Insert new block into corelist, and make it active!
MOVE U4,$PNPTR(U3) ; Get <pred>,,<succ> for old node
HRRI U4,(U3) ; <succ> for new node is old one.
MOVEM U4,$PNPTR(U2) ; Fix up new node...
HLRZS U4 ; Now get <pred>,
HRRM U2,$PNPTR(U4) ; So as to zap predecessor to point at new.
HRLM U2,$PNPTR(U3) ; And zap successor also.
MOVEI U4,$PVLID ; Final touch to active node is ID...
HRLM U4,$PNFLG(U2) ; set it.
IFN OS%ITS,{
MOVE U4,$PNPTR(U3)
TRNN U4,-1
PJRST PSRCOR
}
POPJ P,
IFN OS%ITS,{
IFDEF CORLUZ,.SCALAR CBKERR
PSRCOR: HLRZ U4,$PNBLK(U2) ; Get length of block
ADD U4,$PNBLK(U2) ; Find first unused addr
MOVEI U4,PG$SIZ-1(U4) ; Clear LH and adjust for
LSH U4,-PG$BTS ; finding 1st unused page #.
CAMG U4,FFPAGE
POPJ P,
PUSH P,U4
SUB U4,FFPAGE
IMUL U4,[-1,,]
HRR U4,FFPAGE
IFDEF CORLUZ,[
SETZM CBKERR
SYSCAL CORBLK,[ MOVEI %CBNDR+%CBNDW
MOVEI %JSELF
U4
MOVEI %JSNEW
%CLERR,,CBKERR]
CORLUZ
]
.ELSE,[
SYSCAL CORBLK,[ MOVEI %CBNDR+%CBNDW
MOVEI %JSELF
U4
MOVEI %JSNEW]
PSRERR EROOM
]
POP P,FFPAGE
POPJ P,
};OS%ITS
subttl Primitives - Merging, Freeing, Copying
; PSRMKU - Make unactive, put node on spacelist.
; U3 - ptr to node to put on spacelist.
; Clobbers U4
PSRMK2: MOVEI U3,(U2)
PSRMKU: MOVE U4,SPCLST
HRLI U4,SPCLST-$PNSPP
MOVEM U4,$PNSPP(U3)
TRNE U4,-1 ; If successor exists,
HRLM U3,$PNSPP(U4) ; zap it.
HRRZM U3,SPCLST ; and zap predecessor.
HRRZS $PNFLG(U3) ; Also zap ID check.
POPJ P,
; PSRMRG - Merge two adjacent blocks
; U2 - ptr to node, 1st block (only count is changed)
; U3 - ptr to node, 2nd block (can be on Spacelist or Corelist)
; Clobbers U3,U4
PSRMRG: HRRZ U3,$PNPTR(U2) ; Get addr of succ
CAIN U3, ; Make sure actually have a successor.
PSRERR ERARG ; Caller screwed up.
HLLZ U4,$PNBLK(U3) ; Get # words in U3 node, and use to
ADDM U4,$PNBLK(U2) ; increment count of current block.
PUSHJ P,PSRFRE ; Free U3 node.
POPJ P,
; PSRACT - Make U3 block active (already on corelist, just take off
; the spacelist)
; Clobbers U4
PSRACT: MOVEI U4,$PVLID ; Insert ID check
HRLM U4,$PNFLG(U3) ; into node for future refs by user prog.
PSRACX: MOVE U4,$PNSPP(U3)
TRNE U4,-1 ; Test for successor
HLLM U4,$PNSPP(U4) ; Zap successor if there's one.
MOVSS U4
HLRM U4,$PNSPP(U4) ; Zap predecessor (always)
SETZM $PNSPP(U3)
POPJ P,
; PSRFRE - Puts U3 block on Freelist. Takes it off Spacelist/Corelist as nec.
; PSRFRX - Ditto but assumes block isn't on Spacelist.
; PSRFR - Ditto but assumes block isn't on Corelist or Spacelist.
; All clobber U4.
PSRFRE: SKIPE $PNSPP(U3) ; On spacelist?
PUSHJ P,PSRACX ; Take it off, so it's only on corelist.
PSRFRX: MOVE U4,$PNPTR(U3) ; Get <pred>,,<succ>
TRNE U4,-1
HLLM U4,$PNPTR(U4) ; Zap successor if one.
MOVSS U4
HLRM U4,$PNPTR(U4) ; Zap predecessor (always)
PSRFR: MOVE U4,FRELST
MOVEM U4,$PNPTR(U3)
HRRZM U3,FRELST
POPJ P,
; PSRCPY - Copy block from U3 to U2. Overlap is OK as long as
; U2 < U3.
; Clobbers U4
PSRCPY: PUSH P,U2
MOVE U2,$PNBLK(U2) ; Get <dest> address
HRL U2,$PNBLK(U3) ; Get <source> address
HLRZ U4,$PNBLK(U3) ; Get # words to copy
ADDI U4,-1(U2) ; Find last word to copy into
BLT U2,(U4)
POP P,U2
POPJ P,
subttl Primitives - PSRGTN - Get free node.
; PSRGTN - Get a node.
; Returns addr in U2.
PSRGTN: SKIPN U2,FRELST ; Get pointer to current free list
JRST PSRGN3 ; None left? Make more.
HRL U2,(U2) ; Get ptr to next free node
HLRZM U2,FRELST ; and store it away.
POPJ P,
; Must make new nodes. Requires getting new block recursively,
; with special hack.
PSRGN3: MOVE U2,FRESAV ; Make a freelist pointing to special
MOVEM U2,FRELST ; node saved just for this occasion.
PUSH P,U1 ; Because PSRGET will need one...
MOVEI U1,$PNSIZ*$PMAKN ; Get this many words for new block
PUSHJ P,PSRGET ; Call recursively.
HLRZ U2,U2 ; Get blk addr by itself
PUSHJ P,PSRMKF ; Make a freelist out of block.
HRRZ U1,(U2) ; Get ptr to second node on list,
MOVEM U1,FRELST ; and make that the new freelist.
SETZM (U2) ; While first node on list becomes the
HRRZM U2,FRESAV ; saved node.
POP P,U1
JRST PSRGTN
; PSRMKF - Makes freelist out of piece of core.
; U1 - # words
; U2 - address of block
; Clobbers U1. U2 (same) will point to 1st node on freelist.
PSRMKF: CAIGE U1,$PNSIZ*2 ; Must be able to make at least two.
PSRERR ERARG ; Caller screwed up.
PUSH P,U2
PUSH P,U1 ; First must zap whole block...
ADDI U1,(U2) ; addr to last wd + 1
HRLI U2,(U2)
ADDI U2,1 ; addr,,addr+1
SETZM -1(U2)
BLT U2,-1(U1) ; Clear all of block.
POP P,U1 ; restore size.
IDIVI U1,$PNSIZ ; Find # possible nodes.
MOVE U2,(P) ; Restore addr of block
MOVNI U1,(U1)
HRLI U2,(U1) ; Have -<# nodes>,,<addr>
PSRMF2: MOVEI U1,$PNSIZ(U2) ; Get ptr to next node
MOVEM U1,(U2) ; Store in current node
ADDI U2,$PNSIZ-1
AOBJN U2,PSRMF2
SETZM -$PNSIZ(U2) ; Clear out last node to terminate list.
POP P,U2
POPJ P,
subttl Debugging aid - DEBCOR
; Debugging printout routine. Requires MACROS/NUUOS/OUT.
IFN $$PDBG,{
DBC==17
DEBRET: JSR DEBRST
POPJ P,
DEBCOR: JSR DEBSAV
SKIPN A,CORLST
JRST [ FWRITE DBC,[[ No Corelist!!!
]]
JRST DEBRET]
FWRITE DBC,[[Core list:]]
DEBCR2: FWRITE DBC,[[
Node ],RHV,A,[ Blk ],RHV,$PNBLK(A),[/ ],LHV,$PNBLK(A),[ wds, ]]
HRRZ B,$PNBLK(A)
ANDI B,PG$MSK
CAIE B,
SUBI B,PG$SIZ
HLRZ C,$PNBLK(A)
ADDI B,(C)
ASH B,-PG$BTS
CAIGE B,
SETZ B,
FWRITE DBC,[N8,B,[ pgs ]]
SKIPN $PNSPP(A)
JRST [ FWRITE DBC,[[ACTIVE]]
JRST DEBCR4]
FWRITE DBC,[[FREE, Spacelist ptrs: ],LHV,$PNSPP(A),[,,],RHV,$PNSPP(A)]
DEBCR4: FWRITE DBC,[[
]]
; Now verify ID, if active.
SKIPN $PNSPP(A) ; Active?
JRST [ HLRZ B,$PNFLG(A) ; Yes, get ID.
CAIN B,$PVLID ; See if valid.
JRST .+1 ; Yup.
FWRITE DBC,[[ Bad ID! Should be ],RHV,[[$PVLID]],[, is ],RHV,B,[
]]
JRST .+1]
; Now verify that next node points back to this node.
HRRZ C,$PNPTR(A) ; Get addr of successor
JUMPE C,DEBC60 ; No successor? Done.
HLRZ B,$PNPTR(C) ; Get successor's pred
CAIN A,(B) ; Should point back to current node.
JRST DEBC40 ; Yep, no sweat here.
; Ugh, list failure!!
FWRITE DBC,[[ Succ node has bad pred!
This node: ],RHV,A,[/ ],LHV,$PNPTR(A),[,,],RHV,$PNPTR(A),[
Next node: ],RHV,C,[/ ],LHV,$PNPTR(C),[,,],RHV,$PNPTR(C),[
]]
; Probably screwed from here on, but continue anyway.
; Now verify that next node's block is successor to current block.
DEBC40: HRRZ D,$PNBLK(A) ; Get block addr
HLRZ C,$PNBLK(A) ; and # wds
ADDI D,(C) ; to find next addr after this block.
HRRZ B,$PNPTR(A) ; Get ptr to successor node
JUMPE B,DEBC60 ; If no successor, done with corelist.
HRRZ C,$PNBLK(B) ; Get address next block starts at.
CAIN C,(D) ; Compare...
JRST DEBC50 ; Equal, all's well, so process next.
; Ugh, next block isn't contiguous to this one!
FWRITE DBC,[[ Next block not contiguous!!
Should start at ],RHV,D,[, but succ node claims ],RHV,C,[
]]
DEBC50: HRRZ B,$PNPTR(A)
JUMPE B,DEBC60
MOVEI A,(B)
JRST DEBCR2
; Corelist done, now write out Spacelist.
DEBC60: FWRITE DBC,[[End of Corelist, last managed addr is ]]
HRRZ C,$PNBLK(A) ; Get 1st inaccessible addr
HLRZ D,$PNBLK(A) ; according to last node on spacelist
ADDI C,-1(D) ; via <blk addr>+<# wds>-1
FWRITE DBC,[N8,C,[
First free page is ],N8,FFPAGE,[ at ]]
MOVE B,FFPAGE
IMULI B,PG$SIZ
FWRITE DBC,[N8,B,[
-----------------------------
Spacelist:]]
SKIPN A,SPCLST
JRST [ FWRITE DBC,[[ 0 ??]]
JRST DEBRET]
MOVEI A,SPCLST-$PNSPP
DEBC70: HLRZ B,$PNSPP(A)
SETZ C,
CAIE B,
HRRZ C,$PNSPP(B)
CAIE A,(C)
JRST [ FWRITE DBC,[N8,B]
JRST .+1]
CAIN A,(C)
JRST [ FWRITE DBC,[[*]]
JRST .+1]
FWRITE DBC,[[.]]
HRRZ A,$PNSPP(A)
JUMPE A,[FWRITE DBC,[[0]]
JRST DEBC90]
DEBC75: FWRITE DBC,[[* -> ],N8,A,[:]]
JRST DEBC70
DEBC90: FWRITE DBC,[[
Done.
]]
JRST DEBRET
; Save/restore support for PAGSER debug stuff. Requires MACROS/NUUOS/OUT.
LVAR DEBSAV: 0 ? JRST DEBSV0 ; jump to pure
LVAR DEBACS: BLOCK 20
DEBSV0: PUSH P,U40
IFE $$UCAL,PUSH P,UUORPC
MOVEM 17,DEBACS+17
MOVEI 17,DEBACS
BLT 17,DEBACS+16 ; Save ACs
MOVE 17,DEBACS+17
IFN OS%ITS,[
.OPEN DBC,[.UAO,,'TTY]
.VALUE
OUT(DBC,OPEN(UC$IOT))
]
IFN OS%TNX, OUT(DBC,OPEN(UC$IOT,[.PRIOU]))
JRST @DEBSAV
LVAR DEBRST: 0 ? JRST DEBRS0 ; jump to pure
DEBRS0:
IFN OS%ITS,.CLOSE DBC,
MOVSI 17,DEBACS
BLT 17,17
IFE $$UCAL,POP P,UUORPC
POP P,U40
JRST @DEBRST
} ;end ifn $$pdbg
.END PAGSER

2873
src/sysnet/ftps.336 Executable file
View File

File diff suppressed because it is too large Load Diff

2816
src/sysnet/ftpu.161 Executable file
View File

File diff suppressed because it is too large Load Diff