Trailing-Edge
-
PDP-10 Archives
-
BB-M081Y-SM
-
monitor-sources/cidll.mac
There are 19 other files named cidll.mac in the archive. Click here to see a list.
; Edit= 9029 to CIDLL.MAC on 8-Nov-88 by LOMARTIRE
;Merge Production changes to BUG text
; Edit= 8875 to CIDLL.MAC on 10-Aug-88 by RASPUZZI
;Update BUG. documentation again.
; UPD ID= 8490, RIP:<7.MONITOR>CIDLL.MAC.5, 9-Feb-88 12:17:52 by GSCOTT
;TCO 7.1218 - Update copyright notice.
; UPD ID= 8407, RIP:<7.MONITOR>CIDLL.MAC.4, 4-Feb-88 10:27:45 by GSCOTT
;TCO 7.1210 - Set CINRWP to be not normally dumpable.
; UPD ID= 70, RIP:<7.MONITOR>CIDLL.MAC.3, 12-Aug-87 14:47:02 by MCCOLLUM
;TCO 7.1037 - Fix up calls to SCAMPI, which is now in section XCDSEC.
; *** Edit 7211 to CIDLL.MAC by GRANT on 18-Dec-85
; Add CI counters for WATCH
; UPD ID= 2283, SNARK:<6.1.MONITOR>CIDLL.MAC.42, 26-Jun-85 14:34:51 by NICHOLS
;TCO 6.1.2296 - Round up receive buffer size to account for up to 7 bytes
;of padding.
; UPD ID= 2056, SNARK:<6.1.MONITOR>CIDLL.MAC.41, 3-Jun-85 14:21:34 by MCCOLLUM
;TCO 6.1.1406 - Update copyright notice.
; UPD ID= 1974, SNARK:<6.1.MONITOR>CIDLL.MAC.40, 14-May-85 11:29:31 by GLINDELL
;More documentation around BEGSTR's for 6.1 monitor tables manual
; UPD ID= 1868, SNARK:<6.1.MONITOR>CIDLL.MAC.39, 4-May-85 10:54:00 by MCCOLLUM
;TCO 6.1.1238 - Fix more BUG. documentation
; UPD ID= 1810, SNARK:<6.1.MONITOR>CIDLL.MAC.38, 24-Apr-85 14:28:22 by MCCOLLUM
;TCO 6.1.1238 - Fix more BUG. documentation
; UPD ID= 1766, SNARK:<6.1.MONITOR>CIDLL.MAC.37, 19-Apr-85 17:04:34 by GLINDELL
; TCO 6.1.1332 - return NF.NDP instead of NF.FNS on ZERO LINE COUNTERS
; UPD ID= 1451, SNARK:<6.1.MONITOR>CIDLL.MAC.36, 1-Feb-85 14:19:32 by LOMARTIRE
;Add new symbols to describe the priority used by SC.SDG
; UPD ID= 1344, SNARK:<6.1.MONITOR>CIDLL.MAC.35, 16-Jan-85 11:26:30 by GLINDELL
;Remove TSTS6 tests
; UPD ID= 1154, SNARK:<6.1.MONITOR>CIDLL.MAC.34, 5-Dec-84 17:23:02 by GLINDELL
;Remove unnecessary intra-section transfers
; UPD ID= 1101, SNARK:<6.1.MONITOR>CIDLL.MAC.33, 19-Nov-84 15:44:05 by GLINDELL
;Fix typo
; UPD ID= 1087, SNARK:<6.1.MONITOR>CIDLL.MAC.32, 16-Nov-84 17:29:05 by GLINDELL
;Change line parameter 2848 to 2500, and circuit counter 2500 to 1064
; UPD ID= 1071, SNARK:<6.1.MONITOR>CIDLL.MAC.31, 14-Nov-84 13:32:10 by GLINDELL
;Fix IFIW's
;Add line parameter "buffer size" and circuit counter "data overrun"
; UPD ID= 1061, SNARK:<6.1.MONITOR>CIDLL.MAC.30, 13-Nov-84 09:07:06 by LOMARTIRE
;Use new sending priority symbols in SC.SDG calls.
; UPD ID= 969, SNARK:<6.1.MONITOR>CIDLL.MAC.29, 6-Nov-84 16:08:13 by GLINDELL
;Remove FTT1, insert comments, take out length check
; UPD ID= 864, SNARK:<6.1.MONITOR>CIDLL.MAC.28, 10-Oct-84 17:56:18 by GLINDELL
;Move CIDLL to extended section
; UPD ID= 813, SLICE:<6.1.MONITOR>CIDLL.MAC.24, 20-Sep-84 16:46:01 by GLINDELL
;Forgot to set PBCFLG for .SSPBC callback.. Caused CINBCD bugchk's.
; UPD ID= 770, SNARK:<6.1.MONITOR>CIDLL.MAC.23, 31-Aug-84 12:16:55 by PALMIERI
;Change to new counter macro format
; UPD ID= 738, SNARK:<6.1.MONITOR>CIDLL.MAC.22, 11-Aug-84 11:45:47 by GLINDELL
;Add additional data, and fix comments for all BUG.'s in CIDLL
; UPD ID= 730, SNARK:<6.1.MONITOR>CIDLL.MAC.21, 8-Aug-84 09:34:20 by GLINDELL
;.SSPBC callback has changed: T2 may contain -1 (no CID)
; UPD ID= 702, SNARK:<6.1.MONITOR>CIDLL.MAC.20, 23-Jul-84 17:54:50 by HALPIN
;Add a RET statement to the literal CALL in the PARAMETER macro for Line
;parameter 0 (STATE).
; UPD ID= 698, SNARK:<6.1.MONITOR>CIDLL.MAC.19, 20-Jul-84 13:37:43 by GLINDELL
;Add network management parameters STATE and RECEIVE BUFFERS
;Fix AC bug in network management dispatch
; UPD ID= 692, SNARK:<6.1.MONITOR>CIDLL.MAC.18, 18-Jul-84 09:47:51 by GLINDELL
;Forgot to change .SSMNT to .SSDMA in function code range check
; UPD ID= 691, SNARK:<6.1.MONITOR>CIDLL.MAC.17, 18-Jul-84 09:10:44 by GLINDELL
;More SCAMPI changes: .SSDDG, ACs in .SSDGR, remove .SSCBD, .SSNWO, .SSMNT
; UPD ID= 682, SNARK:<6.1.MONITOR>CIDLL.MAC.16, 13-Jul-84 17:57:46 by GLINDELL
;SCAMPI will no more put back a 'rejected listener' in listen state
; UPD ID= 669, SNARK:<6.1.MONITOR>CIDLL.MAC.15, 6-Jul-84 18:06:40 by PALMIERI
;Turn off FTT1 - for multinode CI testing (requires newer DNADLL)
; UPD ID= 667, SNARK:<6.1.MONITOR>CIDLL.MAC.14, 5-Jul-84 15:47:12 by PALMIERI
;Reorder DLL NTMAN calls to agree with NTMAN
; UPD ID= 654, SNARK:<6.1.MONITOR>CIDLL.MAC.13, 2-Jul-84 16:48:51 by GLINDELL
;Remove PBSBI since SBI # is now equal to node number
;DNAOPN/DNACLS can be called at both jsys and scheduler level
;Correct AC usage in PARAMETER macro
;Test for KLPX9 as well as for SCSISB on SC.CON fail return
;
; UPD ID= 617, SNARK:<6.1.MONITOR>CIDLL.MAC.12, 11-Jun-84 13:42:43 by GLINDELL
;Use RNMXER instead of RETBAD for network management error codes
; UPD ID= 610, SNARK:<6.1.MONITOR>CIDLL.MAC.11, 8-Jun-84 17:27:00 by GLINDELL
;Remove network management entry point, calls come through CINDSP
;
; UPD ID= 591, SNARK:<6.1.MONITOR>CIDLL.MAC.10, 1-Jun-84 18:06:09 by GLINDELL
;Step 1 in allowing DNAOPN/DNACLS to be called at scheduler level
;
; UPD ID= 571, SNARK:<6.1.MONITOR>CIDLL.MAC.9, 28-May-84 15:58:01 by GLINDELL
;New network management entry point
;Remove use of SC.SBI in light of future SCAMPI changes
;
; UPD ID= 523, SNARK:<6.1.MONITOR>CIDLL.MAC.8, 14-May-84 17:29:57 by GLINDELL
;Change reject codes in accordance with new SCAPAR
;
; UPD ID= 506, SNARK:<6.1.MONITOR>CIDLL.MAC.7, 7-May-84 13:07:18 by GLINDELL
;Remove usage of FREE3, make CID be P2 instead
;Remove CIOFF/CION in transmit code
;
; UPD ID= 486, SNARK:<6.1.MONITOR>CIDLL.MAC.6, 30-Apr-84 13:23:10 by GLINDELL
;Change feature tests FTT2 and FTT7 to use 'new' DNADLL
;
; UPD ID= 456, SNARK:<6.1.MONITOR>CIDLL.MAC.5, 25-Apr-84 18:22:04 by GLINDELL
;RONCO is now node 11, change DRPTAB accordingly
;
; UPD ID= 445, SNARK:<6.1.MONITOR>CIDLL.MAC.4, 19-Apr-84 14:12:15 by GLINDELL
;Move NTPARM, NTCTRS and COPMSD (DNCMSG) to D36COM
;
; UPD ID= 441, SNARK:<6.1.MONITOR>CIDLL.MAC.3, 17-Apr-84 11:29:33 by GLINDELL
;Network management interface coded
;Move BEGSTRs CB and PB back because of multiple definition (SCPAR)
;
; UPD ID= 434, SNARK:<6.1.MONITOR>CIDLL.MAC.2, 8-Apr-84 18:34:53 by GLINDELL
;DNADLL and SCA interface coded.
;
; UPD ID= 366, SNARK:<6.1.MONITOR>CIDLL.MAC.1, 1-Mar-84 16:24:11 by GLINDELL
; COPYRIGHT (c) DIGITAL EQUIPMENT CORPORATION 1985, 1988.
; ALL RIGHTS RESERVED.
;
; THIS SOFTWARE IS FURNISHED UNDER A LICENSE AND MAY BE USED AND COPIED
; ONLY IN ACCORDANCE WITH THE TERMS OF SUCH LICENSE AND WITH THE
; INCLUSION OF THE ABOVE COPYRIGHT NOTICE. THIS SOFTWARE OR ANY OTHER
; COPIES THEREOF MAY NOT BE PROVIDED OR OTHERWISE MADE AVAILABLE TO ANY
; OTHER PERSON. NO TITLE TO AND OWNERSHIP OF THE SOFTWARE IS HEREBY
; TRANSFERRED.
;
; THE INFORMATION IN THIS SOFTWARE IS SUBJECT TO CHANGE WITHOUT NOTICE
; AND SHOULD NOT BE CONSTRUED AS A COMMITMENT BY DIGITAL EQUIPMENT
; CORPORATION.
;
; DIGITAL ASSUMES NO RESPONSIBILITY FOR THE USE OR RELIABILITY OF ITS
; SOFTWARE ON EQUIPMENT THAT IS NOT SUPPLIED BY DIGITAL.
;CIDLL - DECnet-36 CI Data Link Layer
SUBTTL Gunnar Lindell 1984
SEARCH D36PAR,SCAPAR
;The functional and design specification for this module is found in
; R61SPC:CI-DECNET-SPEC.MEM, Gunnar Lindell, 13-Nov-1984
IFN FTOPS20 <
SEARCH PROLOG
TTITLE CIDLL,,< - CI Data Link Layer for DECnet-36>
XRESCD
>
IFN FTOPS10 <
SEARCH F,S
TITLE CIDLL - DECnet-36 CI Data Link Layer
$RELOC
>
D36SYM ;Set up D36 specific parameters
RESCD
Subttl Table of Contents
; Table of Contents for CIDLL
;
; Section Page
;
;
; 1. Definitions
; 1.1 External and Internals . . . . . . . . . . . . 5
; 1.2 Local symbols and macros . . . . . . . . . . . 6
; 1.3 Protocol parameters . . . . . . . . . . . . . 7
; 1.4 DECnet/CI
; 1.4.1 Controller block . . . . . . . . . . . . 8
; 1.4.2 Port block . . . . . . . . . . . . . . . 9
; 1.5 Port connection states and macros . . . . . . 10
; 1.6 Network management
; 1.6.1 Parameters . . . . . . . . . . . . . . . 11
; 1.6.2 Counters . . . . . . . . . . . . . . . . 12
; 1.7 Local storage . . . . . . . . . . . . . . . . 13
; 2. DNADLL General calls
; 2.1 CININI - Module initialization . . . . . . . . 14
; 2.2 CINSEC - Once-a-second routine . . . . . . . . 15
; 2.3 CINTIC - Once-a-tick service . . . . . . . . . 16
; 3. Dispatches
; 3.1 DECnet-36 calls . . . . . . . . . . . . . . . 17
; 3.2 SCA callbacks . . . . . . . . . . . . . . . . 19
; 4. Connection management
; 4.1 DNADLL
; 4.1.1 Open data link layer port . . . . . . . 20
; 4.1.2 Close data link layer port . . . . . . . 21
; 4.2 SCA callback
; 4.2.1 "Port broke connection" . . . . . . . . 22
; 4.2.2 "Node came online" . . . . . . . . . . . 23
; 4.2.3 "Connection to listen" . . . . . . . . . 24
; 4.2.4 "Connection response available" . . . . 25
; 4.2.5 "Ok to send data" . . . . . . . . . . . 26
; 4.2.6 "Remote initiated disconnect" . . . . . 27
; 4.3 Subroutines
; 4.3.1 DOLIS . . . . . . . . . . . . . . . . . 28
; 4.3.2 ACTIVAte . . . . . . . . . . . . . . . . 29
; 4.3.3 DOCONN . . . . . . . . . . . . . . . . . 30
; 4.3.4 LOCDIS . . . . . . . . . . . . . . . . . 31
; 4.3.5 REMDIS . . . . . . . . . . . . . . . . . 32
; 4.3.6 CIDDIS . . . . . . . . . . . . . . . . . 33
; 4.3.7 CANBUF . . . . . . . . . . . . . . . . . 34
; 4.3.8 SCAUCB . . . . . . . . . . . . . . . . . 35
; 5. Data transmission
; 5.1 DNADLL
; 5.1.1 Transmit message . . . . . . . . . . . . 36
; 5.2 SCA callback
; 5.2.1 "Datagram received" . . . . . . . . . . 37
; 5.2.2 "Message/datagram send complete" . . . . 38
; 5.2.3 "Dropped datagram" . . . . . . . . . . . 39
Subttl Table of Contents (page 2)
; Table of Contents for CIDLL
;
; Section Page
;
;
; 6. Network management
; 6.1 Read parameter . . . . . . . . . . . . . . . . 40
; 6.2 Set parameter . . . . . . . . . . . . . . . . 41
; 6.3 Clear parameter . . . . . . . . . . . . . . . 42
; 6.4 Show and zero counters . . . . . . . . . . . . 43
SUBTTL Definitions -- External and Internals
;CIDLL has the following entry points:
INTERN CININI ;Initialization
INTERN CINSEC ;Once-a-second routine
; INTERN CINTIC ;Once-a-tick service - not used
INTERN CINDSP ;DECnet-36 function dispatch
;CIDLL makes the following external references:
EXTERN DNGMSG ;Get a DECnet-36 message block
EXTERN DNSKBY ;Skip bytes in MSD
EXTERN DNCMSD ;Copy MSD chain to contiguous block
EXTERN NTPARM ;Network management parameters
EXTERN NTCTRS ;Network management counters
SUBTTL Definitions -- Local symbols and macros
;CIDLL specific AC definitions: CB, PB and CID
DEFAC(CID,P2)
DEFAC(CB,FREE1)
DEFAC(PB,FREE2)
;Dispatch macros
DEFINE $SDSP <...DSP=.>
DEFINE $DSP(A,B,C) <
IFN <A - <.-...DSP>>,
<IF1 <PRINTX Dispatch out of range: 'A 'B>>
B
>
DEFINE $EDSP <PURGE ...DSP>
SUBTTL Definitions -- Protocol parameters
XRESCD
;DNANAM is the DECnet SCA process name
DNANAM: ASCI8(<DECNET$PHASE_III>)
;.CNDRP is the DECnet/CI protocol version
XP .CNDRP,1
;DNADAT is the DECnet/CI connection data
DNADAT: BYTE (8) .CNDRP,0,0,0
REPEAT 3,<BYTE (8) 0,0,0,0>
;Define DECnet/CI reject codes
XP CNRJNL,.CMNRV+1 ;Reject, not listening now
XP CNRJWP,CNRJNL+1 ;Reject, bad protocol version
;Define DECnet/CI disconnect codes
XP CNDSWS,1 ;Wrong state
XP CNDSLD,2 ;Local disconnect
XP CNDSRD,3 ;Remote disconnect
;NDGBUF is the number of receive buffers to queue for each connection
XP NDGBUF,4 ;Heuristic value
ASSUME NDGBUF,GT,0 ;Make sure reasonable value
;POLTIM is the number of seconds between connect attempts
XP POLTIM,^D30 ;Heuristic value
ASSUME POLTIM,GT,0 ;Make sure reasonable
SUBTTL Definitions -- DECnet/CI -- Controller block
;This structure defines one CI controller for DECnet use. Since TOPS-20
;only supports one CI20, only one CB block is needed. That block is
;allocated in static resident space with name CBBLK.
BEGSTR CB ;DECnet/CI controller block
FIELD CPU,3 ;Cpu #
FIELD CI,3 ;Ci interface #
FIELD KLP,1 ;CI ("KLIPA") interface present
HWORD LPT ;Local port #
WORD CID ;Connect ID of current listener
WORD PRT,^D16 ;Pointers to port blocks, indexed by port #
ENDSTR
SUBTTL Definitions -- DECnet/CI -- Port block
;This structure defines one CI port for DECnet use. The port block is
;allocated when a user says "SET CIRCUIT CI-0-0.x STATE ON" to NCP. It
;is alloacted from resident extended free space, and is pointed to by
;a table in CBBLK (see BEGSTR CB).
;
;The port block is never deallocated.
BEGSTR PB
WORD CB ;Pointer to DECnet/CI controller block
WORD CID ;Connect ID
WORD DID ;DNADLL ID for this circuit
HWORD STA ;Connect state for port
HWORD PNR ;Port # of this port
HWORD DRE ;Disconnect reason
WORD BYR ;# of bytes received
WORD BYS ;# of bytes sent
WORD DGR ;# of datagrams received
WORD DGS ;# of datagrams sent
WORD DDG ;# of datagrams dropped by DECnet
WORD SDG ;# of datagrams dropped by SCA
WORD DOR ;# of data overruns
ENDSTR
SUBTTL Definitions -- Port connection states and macros
;The Port connection states are:
XP .PSOFF,0 ;OFF
XP .PSLIS,1 ;LISTEN
XP .PSACC,2 ;ACCEPT
XP .PSCDOWN,3 ;CONNECT, remote believed down
XP .PSCUP,4 ;CONNECT, remote believed up
XP .PSCSE,5 ;CONNECT sent
XP .PSRUN,6 ;RUN
;The GETSTATE macro loads the port connect state into the AC argument
DEFINE GETSTATE (A) <
LOAD A,PBSTA,(PB)
>
;The CHKSTATE macro skips if the port connect state is equal to the
; state given as argument
DEFINE CHKSTATE (A) <
LOAD CX,PBSTA,(PB)
CAIE CX,.PS'A
>
;The NEWSTATE macro sets the port connect state to a new value
DEFINE NEWSTATE (A) <
MOVX CX,.PS'A
STOR CX,PBSTA,(PB)
>
SUBTTL Definitions -- Network management -- Parameters
;Generate the parameter table
XSWAPCD
LINTAB:
PARAMETER(^D0,PANST!PANCL,,,,JFCL,<CALL [ LOAD T2,CBKLP,+CBBLK
TRC T2,1
RET ]>,JFCL,<STATE>)
PARAMETER(^D1110,PANST!PANCL,,,,JFCL,<SETZ T2,>,JFCL,<CONTROLLER NORMAL>)
PARAMETER(^D1105,PANST!PANCL,,,,JFCL,<MOVX T2,NDGBUF>,JFCL,<RECEIVE BUFFERS>)
PARAMETER(^D1112,PANST!PANCL,,,,JFCL,<MOVX T2,^D7>,JFCL,<PROTOCOL CI>)
PARAMETER(^D2500,PANST!PANCL,,,,JFCL,<MOVE T2,RCVBSZ>,JFCL,<Buffer size>)
NRLPAR==.-LINTAB
CIRTAB:
PARAMETER(^D1112,PANST!PANCL,,,,JFCL,<MOVX T2,^D7>,JFCL,<TYPE CI>)
NRCPAR==.-CIRTAB
XRESCD
SUBTTL Definitions -- Network management -- Counters
;Generate the counter table
XRESCD
COUTAB:
COUNTER(^D1000,^D32,<LOAD T1,PBBYR,(PB)>,<SETZRO PBBYR,(PB)>,,<# of bytes read>)
COUNTER(^D1001,^D32,<LOAD T1,PBBYS,(PB)>,<SETZRO PBBYS,(PB)>,,<# of bytes sent>)
COUNTER(^D1010,^D32,<LOAD T1,PBDGR,(PB)>,<SETZRO PBDGR,(PB)>,,<# of datagrams read>)
COUNTER(^D1011,^D32,<LOAD T1,PBDGS,(PB)>,<SETZRO PBDGS,(PB)>,,<# of datagrams sent>)
COUNTER(^D1020,^D8,<CALL [ LOAD T1,PBDDG,(PB)
OPSTR <ADD T1,>,PBSDG,(PB)
RET
]>,<CALL [ SETZRO PBDDG,(PB)
SETZRO PBSDG,(PB)
RET
]>,,<# of dropped datagrams>)
COUNTER(^D1064,^D8,<LOAD T1,PBDOR,(PB)>,<SETZRO PBDOR,(PB)>,,<Data overrun>)
NRCOUN==.-COUTAB
SUBTTL Definitions -- Local storage
;TOPS-20 can have only one CB block, and it is statically defined here.
; In a possible future TOPS-10 implementation, the CININI should loop
; over all possible CI interfaces and build a CB block for each.
;
;Also, TOPS-20 relies on there being only 1 CB block when locating the
; CB block associated with a connection. On TOPS-10, the "XMOVEI CB,CBBLK"
; has to be replaced with something else.
;The CB block
RESVR ;Into zeroed resident storage
CBBLK: BLOCK CB.LEN
;DNACBK has the DNA callback address as supplied in the CININI call
DNACBK: BLOCK 1
;RCVBSZ has the maximum DECnet receive buffer size
RCVBSZ: BLOCK 1
;CONTIM is the connect timer, decremented by CINSEC
RESDT
CONTIM: EXP -1 ;No timer set initially
XRESCD
SUBTTL DNADLL General calls -- CININI - Module initialization
;CININI - initializes CIDLL.
;Call: T1/ line ID (BEGSTR LI)
; T2/ callback address
; T3/ maximum DECnet receive buffer size
; CALL CININI
; +1 on error with error code in T1
; +2 on success
;
;Context: Process
;CININI will initialize the CBBLK, and post a "promiscuous listen".
; Note that since CBBLK is static and zeroed at boot time, only the
; non-zero fields are filled in
XSWAPCD
CININI: SAVEAC <CB,P1>
MOVEM T2,DNACBK ;Save DNADLL callback address
ADDI T3,7 ; round up to account for padding
MOVEM T3,RCVBSZ ; and DECnet receive buffer size
CALL STCBPB ;Set CB from device ID (LI word) supplied in T1
RETBAD (CNNCI%) ; -propagate error return
;Note: CPU and CI # are always zero for TOPS-20
CALL SC.PRT ;Get local port # from SCA
RETBAD (CNNCI%) ; -error, "No CI interface present"
SETONE CBKLP,(CB) ;Set bit in CB "KLIPA present"
STOR T1,CBLPT,(CB) ; and save local port number
;Since there is a KLIPA on the system, get the poll timer going
MOVX T1,POLTIM ;Get poll interval
MOVEM T1,CONTIM ; and store in timer
;Post a "promiscuous listen"
CALL DOLIS
RETBAD (CNNRE%) ; On error, return "No resources"
;Enable "node came online" interrupts
BLCAL. (<XENT SC.SOA>,<[XADDR. SCACBK]>) ;[7.1037]
RETBAD (CNNRE%) ; On error, return "No resources"
RETSKP
XRESCD
SUBTTL DNADLL General calls -- CINSEC - Once-a-second routine
;CINSEC - called once a second
;Call: CALL CINSEC
; +1 return always
;
;Context: Scheduler
;
;Retries connect attempts
XRESCD
CINSEC: SKIPL CONTIM ;Any timer set?
SOSLE CONTIM ; -yes, see if time for cycle?
RET ; -no, return
;It is time to do a retry on all ports in CUP state.
SAVEAC <CB,PB,P1> ;Save ACs
MOVX T1,POLTIM ;Get new timer going
MOVEM T1,CONTIM ; ...
;Now loop over all CBs. It is of course easy on TOPS-20 since there is
; only one: this has to be reworked on the 10.
XMOVEI CB,CBBLK ;Load CB address
;Now loop over all ports on this CB
SETZ P1, ;P1 := port number (begin at 0)
DO.
CAILE P1,<C%SBLL-1> ;Done all ports yet?
RET ; -yes, just return
XMOVEI T1,CB.PRT(CB) ;Get address to port blocks
ADD T1,P1 ;Add in port number
SKIPN PB,(T1) ;Is there a port block?
IFSKP. ; -yes, check connection state
CHKSTATE CUP ; Is it CONNECT, and we believe remote is UP?
ANSKP. ; -yes, then
CALL DOCONN ; call DOCONN
ENDIF.
AOJA P1,TOP. ;Count up port and loop back
ENDDO.
;Will never get here
SUBTTL DNADLL General calls -- CINTIC - Once-a-tick service
;CIDLL does not use the once-a-tick service that DNADLL provides
; RESCD
;CINTIC:RET
SUBTTL Dispatches -- DECnet-36 calls
;CINDSP - called to perform a DECnet-36 function
;Call: T1/ function code
; T2/ CIDLL ID
; T3/ data
; T4/ additional data, if needed
; CALL CINDSP
; +1 on error with error code in T1
; +2 on success
;
;Context: Jsys, scheduler
ASSUME DF.OPN,EQ,0
;Define flags to determine what predispatch processing that should be done
OPNFLG==:1B1 ;Device ID is LI word, and PB may not exist
PBFLG==:1B2 ;Device ID is PB address
NTMFLG==:1B3 ;Network management call
XRESCD
CINDSP:
IFN FTDEBUG < ;Range check while debugging
CAIL T1,DF.OPN
CAILE T1,DF.SZC
BUG.(CHK,CINDOR,CIDLL,SOFT,<DNA dispatch out of range>,<<T1,FUNCOD>>,<
Cause: ROUTER called CIDLL with a function code that is out of range.
Data: FUNCOD - The bad function code
>,RTN)
>
SAVEAC <T5,T6,P1,MS,CB,PB,MB> ;Save ACs
DMOVE T5,T2 ;Save T2-T3
MOVE MB,DNADSP(T1) ;Get dispatch word and flags
;Check PBFLG - if set, then the PB address was passed
TXZN MB,PBFLG ;Is PB flag set?
IFSKP. ; -yes,
MOVE PB,T5 ; T2 contains PB
LOAD CB,PBCB,(PB) ; and PB points to CB
ENDIF.
;Check OPNFLG - if set, then this is a call to open a new port, and a PB
; may or may not exist
TXZN MB,OPNFLG ;Should we get CB and PB from LI word, but
; allow PB to be non-existing?
IFSKP. ; -yes,
SETZ PB, ; Clear PB to flag non-existent port block
MOVE T1,T5 ; Get LI word
CALL STCBPB ; Set CB and PB from LI word
CAIN T1,CNPNA% ; Is the error code "port block not allocated"
SKIPA ; -skip return or the "right" error code, skip
RET ; the error return
ENDIF.
;Check NTMFLG - if set, then this is a network management call
TXZN MB,NTMFLG ;Is NTMFLG set (network management request)
IFSKP. ; -yes,
MOVE P1,T5 ; move NF address to a safer place
LOAD T1,NFEID,(P1) ; get LI word from NF block
CALL STCBPB ; and set up CB (and PB) from block
RNMXER (NF.URC) ; -error "Unrecognized component"
ENDIF.
;Dispatch to processing routine
DMOVE T2,T5 ;Restore T2-T3
JRST (MB) ; and dispatch
;DNA function dispatch table
DNADSP: $SDSP
$DSP(DF.OPN,<IFIW!OPNFLG!<DNAOPN&777777>>,<Open data link layer port>)
$DSP(DF.CLS,<IFIW!PBFLG!<DNACLS&777777>>,<Close data link layer port>)
$DSP(DF.XMT,<IFIW!PBFLG!<DNAXMT&777777>>,<Transmit DECnet message>)
$DSP(DF.SET,<IFIW!NTMFLG!<NTMSET&777777>>,<Set parameter>)
$DSP(DF.CLR,<IFIW!NTMFLG!<NTMCLR&777777>>,<Clear parameter>)
$DSP(DF.RED,<IFIW!NTMFLG!<NTMRED&777777>>,<Read parameter>)
$DSP(DF.SHC,<IFIW!NTMFLG!<NTMCOU&777777>>,<Show counters>)
$DSP(DF.SZC,<IFIW!NTMFLG!<NTMSZC&777777>>,<Show and zero counters>)
$EDSP
;STCBPB sets up CB and PB from a LI word.
;
;Call: T1/ LI word
; CALL STCBPB
; +1 error return, status code in T1
; +2 success return, CB and PB set up
;
;Context: Jsys or scheduler
;
;Note: If the only thing wrong is that the PB is not yet allocated and
; built, CB is set to contain the CB address, and the status code is set
; to CNPNA% (Port block not allocated)
;
;Note: this routine will reject a specification of the local port
STCBPB: SAVEAC <P1>
MOVE P1,T1 ;Put LI word in a safer place
IFN FTDEBUG <
LOAD T1,LIDEV,+P1 ;Get device type
CAIE T1,LD.CIP ; and verify that it is indeed CI port
IFNSK.
BUG.(CHK,CINNCI,CIDLL,SOFT,<Device type not CI>,<<P1,LI>>,<
Cause: The device type supplied to CIDLL was not a CI. ROUTER and NTMAN
should only call CIDLL for CI lines and circuits.
Data: LI - The rejected line ID
>)
RETBAD (CNNSP%) ; and return error after bugcheck
ENDIF.
>
;Now setup CB
LOAD T1,LIKON,+P1 ;Get controller number
LOAD T2,LIUNI,+P1 ; and unit number
SKIPN T1 ;Both should be zero on TOPS-20
SKIPE T2 ; ...
RETBAD (CNNSP%) ; -return error if not
XMOVEI CB,CBBLK ;Ok, get address of CB block
;Is it line or circuit?
TMNE LILXC,+P1 ;Line?
RETSKP ; -yes, return, all done
;Circuit, proceed and get drop number
OPSTR <SKIPL T1,>,LIDRP,+P1 ;Get drop number
CAILE T1,<C%SBLL - 1> ;Between 0 and 15?
RETBAD (CNNSP%) ; -no, error return
OPSTR <CAMN T1,>,CBLPT,(CB) ;Is this local port?
RETBAD (CNLPT%) ; -yes, thats illegal
XMOVEI T2,CB.PRT(CB) ;Get address of port block
ADD T2,T1 ;Add in port number
SKIPN PB,(T2) ;Load PB and test if zero
RETBAD (CNPNA%) ; -not there, return "Port block not allocated"
RETSKP
SUBTTL Dispatches -- SCA callbacks
;SCACBK - all SCA callbacks come through here
;Call: T1/ function code
; T2/ CID (in general).
; T3-4/ function specific data
;
;Context: Scheduler or SCA interrupt
;Define flags to determine what predispatch processing that needs to be done
CIDFLG==:1B1 ;Get CB and PB from CID
LISFLG==:1B2 ;Get CB from CID
NODFLG==:1B3 ;Get CB and PB from node #
BUGFLG==:1B4 ;Generate BUGCHK (unexpected callback)
NOPFLG==:1B5 ;Ignore, i.e. just return
PBCFLG==:1B6 ;Port broke connection call
ASSUME .SSDGR,EQ,0
XRESCD ;Run in extended resident code
SCACBK: SAVEAC <T5,T6,MB,CB,PB,CID> ;Save Acs
;MB is used for the dispatch address and flags
MOVE MB,SCADSP(T1) ;Dispatch table entry to MB
MOVE CID,T2 ;CID to T2
DMOVE T5,T3 ;Move T3-4 to T5-6 to save them for a while
IFN FTDEBUG < ;Range check only while debugging
CAIL T1,.SSDGR ;Range check function code
CAILE T1,.SSDMA
SKIPA ;Go bugchk
>
;Check BUGFLG - if set, then we should never see this callback so BUGCHK
TXZE MB,BUGFLG ;Is BUGCHK flag set?
BUG. (CHK,CINBSC,CIDLL,SOFT,<Unexpected SCA callback>,<<T1,SCAFUN>,<CID,SCACID>>,<
Cause: SCAMPI issued a callback to CIDLL with an unexpected callback.
Data: SCAFUN - The bad function code
SCACID - The connect ID that SCAMPI supplied on the call
>,RTN) ;Return after bugcheck
;Check NOPFLG - if set, then we can ignore this callback
TXZE MB,NOPFLG ;Ignore flag set?
RET ; then just return
;Check PBCFLG - if set, then it is a "port broke connection" callback. This
; callback is special because it may have a negative "CID", indicating an
; "unsolicited" node offline callback.
TXZN MB,PBCFLG ;Is it "port broke connection"?
IFSKP. ; -yes,
JUMPL CID,RTN ; If CID is -1 then it is an unsolicited
; pbc, and we are not interested
TXO MB,CIDFLG ; T2 is not -1 =) T2 contains CID. Set CIDFLG.
ENDIF.
;Check CIDFLG and LISFLG - if any set, then load the CB block address
TXNN MB,LISFLG!CIDFLG ;CIDFLG or LISFLG set?
IFSKP. ; -yes, start with getting CB
IFN FTDEBUG <
OPSTR <SKIPE>,SID,CID ; Verify that user field in CID is zero
CALLRET BADCID ; -no, bugcheck and return
>
XMOVEI CB,CBBLK ; -yes, load CB address
ENDIF.
;Check LISFLG - if set, then this is a listener call and we may not have
; a PB set up for the port in question. Therefore we do not load a PB here.
TXZN MB,LISFLG ;Listener flag set?
IFSKP. ; -yes, verify listener CID in CB
OPSTR <CAME CID,>,CBCID,(CB) ;Equal?
CALLRET BADCID ; -no, bugcheck and return
ENDIF.
;Check CIDFLG - if set, then the CID is valid and we should have a PB
TXZN MB,CIDFLG ; Is CID flag set?
IFSKP. ; -yes, setup and verify PB also
MOVE T1,CID ; Get port number
CALL <XENT SC.NOD> ;[7.1037] from CID into T2
XMOVEI T1,CB.PRT(CB) ; Address of port table
ADD T1,T2 ; Add offset into table
SKIPN PB,(T1) ; Load PB and check if zero
CALLRET BADCID ; -yes, bugcheck and return
OPSTR <CAME CID,>,PBCID,(PB) ;Verify CID in PB
CALLRET BADCID ; -bad, bugcheck and return
ENDIF.
;Check NODFLG - if set, then this is a node online call and we may not have
; a CID.
TXZN MB,NODFLG ;Node flag set?
IFSKP. ; -yes, means node # is in CID
XMOVEI CB,CBBLK ; Load CB address
XMOVEI T1,CB.PRT(CB) ; Get address of port table
ADD T1,CID ; Add port # as offset
MOVE PB,(T1) ; And load PB (possibly zero)
ENDIF.
;Predispatch processing done, retrieve T3-4 and dispatch
DMOVE T3,T5 ;Restore T3-4
JRST (MB) ; and dispatch off MB
;BADCID generates a bugcheck "Bad CID" and returns
BADCID: BUG.(CHK,CINBCD,CIDLL,SOFT,<Bad CID>,<<CID,CID>,<MB,DSPTCH>>,<
Cause: SCAMPI supplied a bad connect ID on a callback to CIDLL.
CIDLL does not have any connection open with the particular
connect ID. This bugcheck may occur if one system crashes
while a DECnet/CI message is outstanding.
Data: CID - The bad connect ID
DSPTCH - The function dispatch word
>,RTN)
;SCA callback dispatch table
SCADSP: $SDSP
$DSP(.SSDGR,<IFIW!CIDFLG!<SCADGR&777777>>,<Datagram received>)
$DSP(.SSMGR,<BUGFLG>,<Message received>)
$DSP(.SSPBC,<IFIW!PBCFLG!<SCAPBC&777777>>,<Port broke connection>)
$DSP(.SSCTL,<IFIW!LISFLG!<SCACTL&777777>>,<Connection to listen>)
$DSP(.SSCRA,<IFIW!CIDFLG!<SCACRA&777777>>,<Connect resp. available>)
$DSP(.SSMSC,<IFIW!CIDFLG!<SCAMSC&777777>>,<Send complete>)
$DSP(.SSDDG,<IFIW!CIDFLG!<SCADDG&777777>>,<Dropped datagram>)
$DSP(.SSLCL,<NOPFLG>,<Little credit left>)
$DSP(.SSNCO,<IFIW!NODFLG!<SCANCO&777777>>,<Node came online>)
$DSP(.SSOSD,<IFIW!CIDFLG!<SCAOSD&777777>>,<Ok to send data>)
$DSP(.SSRID,<IFIW!CIDFLG!<SCARID&777777>>,<Remote disconnected>)
$DSP(.SSCIA,<NOPFLG>,<Credit is available>)
$DSP(.SSDMA,<BUGFLG>,<Named buffer operation complete>)
$EDSP
SUBTTL Connection management -- DNADLL -- Open data link layer port
;DNAOPN opens a data link layer port.
;
;Call: CB and PB set up: PB is zero if port block is not yet allocated
; T2/ port ID as begstr LI
; T3/ DNADLL ID for this port
; CALL DNAOPN
; +1 error return, status code in T1
; +2 success return
;
;Context: Process or scheduler
;
;Ac usage: CB, PB and MB may be clobbered.
ASSUME LI.LEN,EQ,1 ;Make sure LI BEGSTR fits into one word
XRESCD
DNAOPN: STKVAR <LIID,DLLID>
MOVEM T2,LIID ;Save port ID
MOVEM T3,DLLID ;Save DNADLL ID
NOSKED ;Protect
SKIPN PB ;Does PB exist?
IFSKP.
OKSKED ; Unprotect
CHKSTATE OFF
RETBAD (CNPAO%,) ; Return status "Port already open"
ELSE.
;Come here to build a port block
MOVX T1,PB.LEN ;Allocate port block
CALL DNGWDZ ;Get zeroed data
RETBAD (CNNRE%) ; -error "No resources"
MOVE PB,T1 ;Make PB point to allocated block
STOR CB,PBCB,(PB) ;Store CB address in PB
MOVE T1,DLLID ;Get DNADLL ID
STOR T1,PBDID,(PB) ; and store in port block
NEWSTATE OFF ;Initial state is OFF
MOVE T1,LIID ;Get LI word
LOAD T1,LIDRP,+T1 ; and extract drop number
STOR T1,PBPNR,(PB) ; which is port number
;Link PB into CB. When that is done, the port is in operation
XMOVEI T2,CB.PRT(CB) ;Get address of port table
ADD T2,T1 ; and offset to the port number
MOVEM PB,(T2) ; and store pointer to PB in CB
OKSKED ;Ok to unprotect
ENDIF.
;Activate the port, port state =) LISTEN or CONNECT
CALL ACTIVA
MOVE T1,PB ;Return port handle ("PB") in T1
RETSKP
ENDSV.
SUBTTL Connection management -- DNADLL -- Close data link layer port
;DNACLS closes a data link layer port
;
;Call: PB and CB are set up
; CALL DNACLS
; +1 error return, status code in T1
; +2 success return
;
;Context: Jsys or scheduler level
;
;Ac usage: CB, PB and MB may be clobbered
;Interlocking: DNACLS has to interlock between getting the old state and
; disconnecting the connection if there was one. If CIOFF was not done, DNACLS
; could assume that the status was, say, LISTEN and then set the state to
; OFF. In the window between reading LISTEN and setting state to OFF there
; could be a SCA interrupt establishing a connection.
XRESCD ;Must be resident since CIOFF is done
DNACLS: CIOFF ;Implies NOSKED too..
GETSTATE T1 ;Get current state
CALL @DNACLT(T1) ;Call transition routine
NEWSTATE OFF ;New state is OFF in all cases
CION ;No harm can be done now
RETSKP
;DNACLT is state transition vector for DNADLL Close event
DNACLT: $SDSP
$DSP(.PSOFF,<IFIW <R&777777>>,<Current state is OFF>)
$DSP(.PSLIS,<IFIW <R&777777>>,<Current state is LISTEN>)
$DSP(.PSACC,<IFIW <LOCDIS&777777>>,<Current state is ACCEPT>)
$DSP(.PSCDO,<IFIW <R&777777>>,<Current state is CDOWN>)
$DSP(.PSCUP,<IFIW <R&777777>>,<Current state is CUP>)
$DSP(.PSCSE,<IFIW <LOCDIS&777777>>,<Current state is CSENT>)
$DSP(.PSRUN,<IFIW <LOCDIS&777777>>,<Current state is RUN>)
$EDSP
SUBTTL Connection management -- SCA callback -- "Port broke connection"
;"Port broke connection" callback indicate that the connection to the remote
; node has been broken for some reason.
;
;At the callback the CID is unusable, and the connection is broken. Buffers
; outstanding for it will be cleaned when the SCA reaper runs at a future
; time.
;
;CIDLL shall activate the connection if needed, and report "port state change"
; to DNADLL if the current state was RUN.
;
;Call: CB,PB and CID set up
; CALL SCAPBC
;
;Context: Scheduler or SCA interrupt
;
;Note: there is no interlock problem since even if the callback occurred at
; scheduler level, SCA will not interrupt for the same CID (since its broken)
XRESCD
SCAPBC: GETSTATE T1 ;Get current state
JRST @SCAPBT(T1) ; and go to transition routine
;SCAPBT is the state transition vector for SCA callback "port broke connection"
SCAPBT: $SDSP
$DSP(.PSOFF,<IFIW <SCAUCB&777777>>,<Current state is OFF>)
$DSP(.PSLIS,<IFIW <SCAUCB&777777>>,<Current state is LISTEN>)
$DSP(.PSACC,<IFIW <PBCACC&777777>>,<Current state is ACCEPT>)
$DSP(.PSCDO,<IFIW <SCAUCB&777777>>,<Current state is CDOWN>)
$DSP(.PSCUP,<IFIW <PBCCUP&777777>>,<Current state is CUP>)
$DSP(.PSCSE,<IFIW <PBCCSE&777777>>,<Current state is CSENT>)
$DSP(.PSRUN,<IFIW <PBCRUN&777777>>,<Current state is RUN>)
$EDSP
;PBCACC handles ACCEPT state
;
;Move state to LISTEN
PBCACC: NEWSTATE LISTEN ;CID is disconnected, need not do any more
RET
;PBCCUP handles CUP state
;PBCCSE handles CSENT state
;
;Move state to CDOWN
PBCCUP:
PBCCSE: NEWSTATE CDOWN ;Remote is down
RET
;PBCRUN handles RUN state
;
;Signal port state change (to down) to DNADLL, and activate port
PBCRUN: MOVX T1,DI.LSC ;Port state change
LOAD T2,PBDID,(PB) ;DNADLL id
MOVX T3,LS.OFF ;Port OFF
CALL @DNACBK ;Call DNADLL at callback address
CALLRET ACTIVA ;Activate this port and return
SUBTTL Connection management -- SCA callback -- "Node came online"
;SCANCO is called on SCA callback "node came online"
;
;Call: CB, PB setup (PB possibly zero)
; CALL SCANCO
; +1 return always
;
;Context: SCA interrupt
XRESCD
SCANCO: SKIPN PB ;Is there a port block?
RET ; -no, just return
GETSTATE T1 ;Get state into T1
JRST @SCANCT(T1) ; and go to transition routine
;SCANCT is the state transition vector for "node came online"
SCANCT: $SDSP
$DSP(.PSOFF,<IFIW <R&777777>>,<Current state is OFF>)
$DSP(.PSLIS,<IFIW <R&777777>>,<Current state is LISTEN>)
$DSP(.PSACC,<IFIW <SCAUCB&777777>>,<Current state is ACCEPT>)
$DSP(.PSCDO,<IFIW <NCOCDO&777777>>,<Current state is CDOWN>)
$DSP(.PSCUP,<IFIW <R&777777>>,<Current state is CUP>)
$DSP(.PSCSE,<IFIW <NCOCSE&777777>>,<Current state is CSENT>)
$DSP(.PSRUN,<IFIW <SCAUCB&777777>>,<Current state is RUN>)
$EDSP
;NCOCSE handles CSENT state
;NCOCDO handles CDOWN state
;
;Move state to CUP, bugcheck if in CSENT state
NCOCSE: CALL SCAUCB ;Bugcheck if in CSENT state
NCOCDO: NEWSTATE CUP
RET
SUBTTL Connection management -- SCA callback -- "Connection to listen"
;SCACTL is called on SCA callback "connection to listen"
;
;Call: CB, PB and CID set up
; T3/ Address of connection data from remote system
; CALL SCACTL
; +1 return always
;
;Context: SCA interrupt
XRESCD
SCACTL: STKVAR <REMDAT>
MOVEM T3,REMDAT ;Save address of remote connect data
CALL DOLIS ;Post a new promiscuous listener
JFCL ; ...
;Now get the port number for the remote end
MOVE T1,CID ;Load CID into T1 to
CALL <XENT SC.NOD> ;[7.1037] get port number into T2
;See if we are listening on this port: do we have a PB?
XMOVEI T1,CB.PRT(CB) ;Get address of port blocks
ADD T2,T1 ; add port number offset
SKIPN PB,(T2) ;Load PB and check for zero
CALLRET REJNL ; -no, no PB, reject and return
;Yes, we have a PB: are we listening?
;Note: the following section of code replaces the state transition vector
CHKSTATE LISTEN
CALLRET REJNL ; and do reject
;The only thing that can stop us now is bad connection data
MOVE T2,REMDAT ;Get address of remote connection data
LDB T1,[POINT 8,(T2),7] ; and 1st byte of connection data
CAIN T1,.CNDRP ;Check if protocol type is .CNDRP "Multidrop"
IFSKP. ; -no,
NEWSTATE OFF ; Set state to OFF
CALL REJWP ; Reject connection
MOVX T2,.CNDRP ; Get our protocol version
BUG.(INF,CINRWP,CIDLL,SOFT,<Remote supplied wrong protocol version>,<<T1,REMVER>,<T2,LOCVER>>,<
Cause: The remote port is not running the same protocol version of DECnet/CI.
This may happen if you are running DECnet/CI between two different
versions of TOPS-20, or if you are running DECnet/CI between TOPS-20
and a VMS system. DECnet/CI between TOPS-20 and VMS is not
supported.
Action: If this is caused by a VMS system, then remove the system from
your cluster or disable the CI dircuit to that system. If it is
caused by a TOPS-20 system, then you should make both systems
run the same version of the software or disable the CI circuit
between them.
Data: REMVER - The remote ends protocol version
LOCVER - The local TOPS-20 systems protocol version
>,RTN,<DB%NND>) ;[7.1210]
ENDIF.
;All ok, accept the connection
BLCAL. (<XENT SC.ACC>,<CID,[0],[0],[NDGBUF]>) ;[7.1037]Accept
IFSKP. ;If succeded,
STOR CID,PBCID,(PB)
NEWSTATE ACCEPT ; and new state is ACCEPT
ELSE. ;Accept failed
;Note: if the ACCEPT fails nothing bad can happen. State stays LISTEN,
; and no CID has been moved into the PB to confuse things
BUG.(CHK,CINACF,CIDLL,SOFT,<Accept failed>,<<T1,ERRCOD>>,<
Cause: CIDLL decided to accept an incoming connection, but the accept
call to SCA failed.
Data: ERRCOD - The error code returned by SCAMPI
>)
ENDIF.
RET ;Return
ENDSV.
;REJECT a connection with code CNRJNL "no listener"
;
REJNL: MOVX T1,CNRJNL ;Load rejection code
JRST REJECT ; and go reject
;REJECT a connection with code CNRJWP "wrong protocol version"
;
REJWP: MOVX T1,CNRJWP ;Load rejection code
; JRST REJECT ; and go reject
REJECT: BLCAL. (<XENT SC.REJ>,<CID,T1>) ;[7.1037]
JFCL ;Ignore failure to reject
RET
SUBTTL Connection management -- SCA callback -- "Connection response available"
;SCACRA is called on SCA callback "Connection response available". It is
; expected to happen while in CSENT state, and informs CIDLL whether the
; connection was accepted (go to RUN) or if it was rejected (go to OFF)
;
;Call: CB, PB and CID set up.
; T3/ -1 if accepted, 0 if rejected
; T4/ reject code, or pointer to connection data
; CALL SCACRA
; +1 return always
;
;Context: SCA interrupt
XRESCD
SCACRA: STKVAR <REMSTS,REJCOD>
MOVEM T3,REMSTS ;Save status and rejection code
MOVEM T4,REJCOD
;The following section of code implements the state transitions for SCACRA:
; if the current state is not CONNECT, then bugchk and do no more. Else process
; the connect response
CHKSTATE CSENT
BUG.(CHK,CINUEC,CIDLL,SOFT,<Unexpected connect response>,<<CID,SCACID>>,<
Cause: CIDLL received a callback from SCA stating that a connection response
was available. CIDLL was not expecting any such callback for
the port.
Data: SCACID - The connect ID
>,RTN) ; -return after the BUGCHK
;Now, if state is CSENT, then check if it was rejected. If so, go to state
; OFF and BUGINF, else go to state RUN and signal "port state change"
SKIPN REMSTS ;Rejected or accepted
IFSKP. ; -accepted!
NEWSTATE RUN ; New state is RUN
MOVX T1,DI.LSC ; Flag port state change to DNADLL
LOAD T2,PBDID,(PB) ; DNADLL id
MOVX T3,LS.ON ; New port state is ON
CALLRET @DNACBK ; Call DNADLL and return
ELSE. ; -rejected
BLCAL. (<XENT SC.DIS>,<CID,[CNDSWS]>) ;[7.1037]Disconnect "wrong state"
JFCL ; -ignore failure to disconnect
MOVE T1,REJCOD ; Get rejection code
CAIN T1,CNRJWP ; Is it wrong protocol version?
IFSKP. ; -no,
NEWSTATE CUP ; just move to CUP
ELSE. ; -yes,
NEWSTATE OFF ; new state is OFF
MOVX T1,.CNDRP ; Get our protocol version
BUG.(INF,CINRRL,CIDLL,SOFT,<Remote rejected our protocol version>,<<T1,LOCVER>>,<
Cause: The remote port does not understand the protocol version of DECnet/CI.
The other system may be running another version of TOPS-20, or the
remote system may be running VMS. DECnet/CI between TOPS-20
and VMS is not supported.
Action: If this is caused by a VMS system, then remove the system from
your cluster or disable the CI dircuit to that system. If it is
caused by a TOPS-20 system, then you should make both systems
run the same version of the software or disable the CI circuit
between them.
Data: LOCVER - TOPS-20 DECnet/CI protocol version
>)
ENDIF.
ENDIF.
RET
ENDSV.
SUBTTL Connection management -- SCA callback -- "Ok to send data"
;SCAOSD is called on SCA callback "Ok to send data". It is expected to
; occur while in ACCEPT state, and is an indication that the connection
; is established, i.e.
;
;Call: CB, PB and CID set up
; CALL SCAOSD
; +1 return always
;
;Context: SCA interrupt
XRESCD
SCAOSD: GETSTATE T1 ;Get current state
JRST @SCAOST(T1) ; and go to transition routine
;SCAOST is the state transition vector for SCA callback "Ok to send data"
SCAOST: $SDSP
$DSP(.PSOFF,<IFIW <SCAUCB&777777>>,<Current state is OFF>)
$DSP(.PSLIS,<IFIW <SCAUCB&777777>>,<Current state is LISTEN>)
$DSP(.PSACC,<IFIW <OSDACC&777777>>,<Current state is ACCEPT>)
$DSP(.PSCDO,<IFIW <SCAUCB&777777>>,<Current state is CDOWN>)
$DSP(.PSCUP,<IFIW <SCAUCB&777777>>,<Current state is CUP>)
$DSP(.PSCSE,<IFIW <SCAUCB&777777>>,<Current state is CSENT>)
$DSP(.PSRUN,<IFIW <R&777777>>,<Current state is RUN>)
$EDSP
;OSDACC is called in ACCEPT state
OSDACC: NEWSTATE RUN ;Hurrah!
MOVX T1,DI.LSC ;Tell DNADLL also "port state change"
LOAD T2,PBDID,(PB) ;DNADLL id
MOVX T3,LS.ON ;New state is ON
CALLRET @DNACBK ;Call at supplied address and return
SUBTTL Connection management -- SCA callback -- "Remote initiated disconnect"
;SCARID is called on SCA callback "remote disconnect". CIDLL will disconnect
; the port if running, notify DNADLL if so, and activate the port again.
;
;Call: CB, PB and CID set up
; CALL SCARID
; +1 return always
;
;Context: SCA interrupt
SCARID: GETSTATE T1 ;Get current state
JRST @SCARIT(T1) ; and go to transition routine
;SCARIT is the transition vector for SCA callback "Remote initiated disconnect"
SCARIT: $SDSP
$DSP(.PSOFF,<IFIW <SCAUCB&777777>>,<Current state is OFF>)
$DSP(.PSLIS,<IFIW <SCAUCB&777777>>,<Current state is LISTEN>)
$DSP(.PSACC,<IFIW <RIDACC&777777>>,<Current state is ACCEPT>)
$DSP(.PSCDO,<IFIW <SCAUCB&777777>>,<Current state is CDOWN>)
$DSP(.PSCUP,<IFIW <SCAUCB&777777>>,<Current state is CUP>)
$DSP(.PSCSE,<IFIW <RIDCSE&777777>>,<Current state is CSENT>)
$DSP(.PSRUN,<IFIW <RIDRUN&777777>>,<Current state is RUN>)
$EDSP
;RIDACC is called in ACCEPT state
RIDACC: CALL REMDIS ;Return receive buffers and disconnect
NEWSTATE LISTEN ; and new state is LISTEN
RET
;RIDCSE is called in CSENT state
RIDCSE: CALL REMDIS ;Return receive buffers and disconnect
NEWSTATE CUP ; and new state is CUP
RET
;RIDRUN is called in RUN state
RIDRUN: CALL REMDIS ;Return receive buffers and disconnect
MOVX T1,DI.LSC ;Signal port down to DNADLL
LOAD T2,PBDID,(PB) ;DNADLL id
MOVX T3,LS.OFF ;Down
CALL @DNACBK ;Call at supplied address
CALLRET ACTIVA ; and activate and return
SUBTTL Connection management -- Subroutines -- DOLIS
;DOLIS posts a promiscuous listen
;
;Call: CB/ points to CB block
; CALL DOLIS
; +1 error return, SCA error code in T1
; +2 success return, CID for listen inserted into CBCID
;
;Context: Jsys or SCA interrupt
XRESCD
DOLIS: BLCAL. (<XENT SC.LIS>,<[XADDR. DNANAM],[XADDR. DNANAM],[-1],[XADDR. SCACBK],[0],[0],[0]>) ;[7.1037]
BUG.(CHK,CINLIE,CIDLL,SOFT,<Listen failed>,<<T1,ERRCOD>>,<
Cause: CIDLL asked SCA for a "promiscuous listen" but the call failed.
As a consequence, the system does not accept any future incoming
DECnet/CI connections.
Data: ERRCOD - The error code returned from SCAMPI
>,RTN) ;Return after bugcheck
STOR T1,CBCID,(CB) ;Store new listener CID
RETSKP
SUBTTL Connection management -- Subroutines -- ACTIVAte
;ACTIVAte decides whether a port should go to LISTEN or CONNECT state
;
;Call: CB and PB setup
; CALL ACTIVA
; +1 always
;
;Context: Jsys or SCA interrupt
;
;Connection algorithm says that if local port # is less than remote, then
; our side LISTENs, else connects.
XRESCD
ACTIVA: LOAD T1,CBLPT,(CB) ;Get local port #
OPSTR <CAML T1,>,PBPNR,(PB) ; Less than remote port # ?
IFSKP. ; -yes, do LISTEN
NEWSTATE LISTEN ; So move state to LISTEN
ELSE.
OPSTR <CAMG T1,>,PBPNR,(PB) ;Local greater than remote port # ?
IFSKP.
NEWSTATE CUP ; -yes, move state to CUP
CALL DOCONN ; and do a connect
ELSE. ; Local port = remote port, we're confused
NEWSTATE OFF ; Go to OFF state
BUG.(CHK,CINLER,CIDLL,SOFT,<Local port # equal to remote>,,<
Cause: The ACTIVATE routine noticed that we tried to connect to ourselves.
CIDLL should have detected this before.
>,RTN)
ENDIF.
ENDIF.
RET
SUBTTL Connection management -- Subroutines -- DOCONN
;DOCONN does an active connect
;
;Call: CB and PB setup
; CALL DOCONN
; +1 always
;
;Context: Jsys or scheduler
;
XRESCD
DOCONN:
;Must interlock here since SCA may interrupt between BLCAL. and the next
; instruction that stores the CID.......
CIOFF
;Make sure that -
; STATE is indeed CUP. This will protect against connect responses
; coming in between the call of this routine and the actual connect call
; to SCA
CHKSTATE CUP ;Make sure we are in CUP state
IFSKP. ; -yes,
BLCAL. (<XENT SC.CON>,<[XADDR. DNANAM],[XADDR. DNANAM],<PBPNR,(PB)>,[0],[0],[XADDR. SCACBK],[0],[XADDR. DNADAT],[0],[NDGBUF]>) ;[7.1037]
IFSKP. ; Success return
STOR T1,PBCID,(PB) ; Store CID since connect now going
NEWSTATE CSENT ; and move state to connect sent
ELSE. ; Fail return
CAIE T1,SCSISB ; Is there a system block
CAIN T1,KLPX9 ; and is VC open?
IFNSK. ; -yes, move state to CDOWN
NEWSTATE CDOWN
ENDIF.
ENDIF.
ENDIF.
CION ;Allow interrupts
RET
SUBTTL Connection management -- Subroutines -- LOCDIS
;LOCDIS - disconnect a connection
;
;Called at many state transitions when local node decides to disconnect
;
;Will cancel SCA receive buffers and disconnect the connection
LOCDIS: CALL CANBUF ;Cancel outstanding receive buffers
MOVX T1,CNDSLD ;Load disconnect code "Local disconnect"
CALLRET CIDDIS ;Go disconnect
SUBTTL Connection management -- Subroutines -- REMDIS
;REMDIS - disconnect a connection
;
;Called at many state transitions when remote node disconnects
;
;Will cancel SCA receive buffers and disconnect the connection
REMDIS: CALL CANBUF ;Cancel outstanding receive buffers
MOVX T1,CNDSRD ;Load disconnect code "Remote disconnect"
CALLRET CIDDIS ;Go disconnect
SUBTTL Connection management -- Subroutines -- CIDDIS
;CIDDIS - disconnect a disconnection
;
;T1/ disconnect reason code
CIDDIS: STOR T1,PBDRE,(PB) ;Store disconnect reason code in PB
BLCAL. (<XENT SC.DIS>,<<PBCID,(PB)>,T1>) ;[7.1037]
JFCL ;Ignore error return
RET
SUBTTL Connection management -- Subroutines -- CANBUF
;CANBUF - cancel receive buffers
CANBUF: BLCAL. (<XENT SC.CRD>,<<PBCID,(PB)>,[NDGBUF]>) ;[7.1037]
JFCL ;Ignore error return
RET
SUBTTL Connection management -- Subroutines -- SCAUCB
;SCAUCB issues a callback "Unexpected callback function"
SCAUCB: BUG.(CHK,CINUCB,CIDLL,SOFT,<Unexpected SCA callback>,<<T1,STATE>,<CID,SCACID>,<MB,ROUADR>>,<
Cause: SCA issued a callback to DECnet/CI that was not expected in the
current state of the connection.
Data: STATE - The DECnet/CI connection state
SCACID - The connect ID for the connection in question
ROUADR - The address of the SCA callback processing routine
>,RTN) ;Return after bugcheck
SUBTTL Data transmission -- DNADLL -- Transmit message
;DNAXMT transmits a DECnet message
;
;Call: CB and PB are set up
; T3/ MB address
; CALL DNAXMT
; +1 error return, status code in T1
; +2 success return
;
;Context: Process or scheduler. **Note: one call to DNAXMT may interrupt
; another one**
;
;AC usage: CB, PB, MB, MS, T1-T6, P1 may be clobbered
XRESCD
DNAXMT: MOVE MB,T3 ;Get MB address
;Check that connect state is RUN. However, no problems if this changes until
; actual send time
CHKSTATE RUN ;Is state RUN?
RETBAD(CNPWS%) ; -error return "Port in wrong state"
;Allocate a datagram buffer
MOVX T1,1 ;Need one buffer
CALL <XENT SC.ALD> ;[7.1037]Try to get it
RETBAD(CNNRE%) ; -error "resource error"
MOVE P1,T1 ;Save buffer address in P1
CAIE T2,1 ;Make sure we got only one buffer
BUG.(CHK,CINWNB,CIDLL,SOFT,<Wrong number of buffers>,<<T2,COUNT>>,<
Cause: CIDLL asked SCAMPI to allocate 1 buffer, but received more than
one. The extra buffers are now lost.
Data: COUNT - The number of buffers SCAMPI allocated and returned
>)
;Copy the MSD chain to the buffer
XMOVEI T1,.MHUDA(P1) ;Address of user area in SCA buffer
MOVX T2,<C%BYTD-C%OVHD> ;Size of maximum user buffer in bytes
CALL DNCMSD ;Go copy....
IFNSK. ; -error return means DECnet message longer
MOVE T1,P1 ; than SCA buffer, return the buffer
CALL <XENT SC.RLD> ;[7.1037] to SCA
RETBAD(CNIMS%) ; and return "illegal message size"
ENDIF.
;On success return from DNCMSD, T1 has a bp to destination, T2 # of pad
; bytes to apply and T3 length of message (except pad bytes) in bytes
SKIPN T2 ;Any padding needed?
IFSKP.
ADD T3,T2 ;Add up length of message
MOVE T1,[POINT 8,.MHUDA(P1)] ;Make our own bp indexed by P1
TRO T2,200 ;Make a ROUTER pad byte with count
IDPB T2,T1 ;Put it in front
ENDIF.
;Verify that state is RUN and send the datagram
;
;If port is closed between the CHKSTATE and the SC.SDG doesnt really matter,
; since we either will get an error return from SC.SDG and then deallocate
; the transmit buffer, or well get a message send complete callback.
CHKSTATE RUN ;Is it still RUN?
IFSKP. ;Yes, send datagram
OPSTRM <ADDM T3,>,PBBYS,(PB) ;Update # of bytes sent
OPSTRM <AOS>,PBDGS,(PB) ; and # of datagrams sent
BLCAL.(<XENT SC.SDG>,<<PBCID,(PB)>,[F.RTB],T3,P1,[CIDPRI],[0]>) ;[7.1037]
IFSKP. ;Transmit succeeded
AOS DNASND ;[7211] COUNT THE SEND
MOVX T4,CNSUC% ; so load success code
ELSE. ;Error on send
MOVX T4,CNXMF% ; and load error code "transmit failed"
ENDIF.
ELSE.
MOVX T4,CNPWS% ; -if not RUN, then "port in wrong state"
ENDIF.
;Between here and when we callback DNADLL the port may be closed by SCA,
; but DECnet can handle that
CAIE T4,CNSUC% ;Was it success?
IFSKP. ;-yes, callback DNADLL
MOVX T1,DI.ODN ; Function is "output done"
LOAD T2,PBDID,(PB) ; DNADLL ID
MOVE T3,MB ; Message block in T3
; Status is already in T4
CALL @DNACBK ; Call back
RETSKP ; and skip return to indicate success
ELSE. ;-no, release buffer and give error return
MOVE T1,T4 ; Get status into T1 for RET
EXCH T1,P1 ; Now save error code while getting buffer adr
CALL <XENT SC.RLD> ;[7.1037] Release the SCA buffer
MOVE T1,P1 ; Retrieve the error code
RET ; and return +1 to indicate error
ENDIF.
;Will never get here
SUBTTL Data transmission -- SCA callback -- "Datagram received"
;SCADGR is called on SCA callbacks "Datagram received"
;
;Call: CB, PB and CID are set up.
; T3/ Address of datagram buffer
; T4/ Flags (! address of routine to return buffer - not used)
; Flags are: F.SPM set =) high density mode,
; clear =) industry compatiblw mode
; Packet length (including packet header) (in bytes if
; industry compatible mode) is in offset .MHPKL in the buffer
; CALL SCADGR
; +1 return always
;
;Context: SCA interrupt
;
;Ac usage: CB, PB, CID, MB, MS, T6. P1 = SCA buffer adr
; T1-T6 temporary
XRESCD
SCADGR: SAVEAC <P1,MS> ;P1 contains SCA buffer adr
AOS DNARCV ;[7211] COUNT THE RECEPTION
MOVE P1,T3 ;Buffer address
CALL SCADG1 ;Call coroutine to do the job
CALLRET RDGBUF ; on +1 return, return buffer and return
CALL RDGBUF ;+2 return: return buffer and ..
MOVX T1,DI.INC ; give "input complete" callback to DNADLL
LOAD T2,PBDID,(PB) ; Get DNADLL ID
MOVE T3,MB ; MB address
MOVX T4,CNSUC% ; Success
CALLRET @DNACBK ;Callback and return
;SCADG1 - worker routine for SCADGR
; T4/ flags
; returns +1 if message was dropped or some other error
; +2 with message to be passed to higher layer
SCADG1: STKVAR <BYTCNT> ;# of DECnet bytes in datagram
CHKSTATE RUN ;Verify that state is 'RUN'
BUG.(CHK,CINUDR,CIDLL,SOFT,<Unexpected datagram receive>,<<CID,SCACID>>,<
Cause: CIDLL received a datagram with the connection state not being RUN.
Data: SCACID - The connect ID
>,RTN) ;Return after bugcheck
TXNE T4,F.SPM ;Verify that it is industry compatible
BUG.(CHK,CINNIC,CIDLL,SOFT,<Received illegal packet format>,<<T4,FLAGS>>,<
Cause: CIDLL received a DECnet datagram, but the packet mode was not
industry compatible.
Data: FLAGS - The flags and mode of the packet
>,RTN) ;Return after bugcheck
;Get length of data portion of packet into BYTCNT
MOVE T1,.MHPKL(P1) ;Total packet length
SUBI T1,C%OVHD ; and subtract overhead bytes
MOVEM T1,BYTCNT ; and store result
;Verify that the message will fit into a DECnet buffer
CAMG T1,RCVBSZ ;Too big?
IFSKP. ; -yes, 'data overrun'
OPSTRM <AOS>,PBDOR,(PB) ; Update count
RET ; and return
ENDIF.
;Allocate message block and input buffer
CALL DNGMSG ; and ask DECnet-36 for MB and buffer
JRST SCADRP ; -couldnt, drop buffer
MOVE MB,T1 ;Message block address
XMOVEI MS,UD.MSD(MB) ; and load MSD address
;Update counters for received message
MOVE T1,BYTCNT ;Get # of bytes in message
OPSTRM <ADDM T1,>,PBBYR,(PB) ;# of bytes received
OPSTRM <AOS>,PBDGR,(PB) ; and # of datagrams received
;All set, copy message into DECnet buffer
STOR T1,MDBYT,(MS) ;Store # of bytes in MSD
MOVEI T1,3(T1) ;Make ..
LSH T1,-2 ; .. # of words to copy
XMOVEI T2,.MHUDA(P1) ;Source address
LOAD T3,MDALA,(MS) ; and destination address from MSD
EXTEND T1,[XBLT] ;Here we go...
;See if we should strip off (VMS) padding of -1 bytes..
; Loop over SCA buffer and check how many -1 bytes there are
; T1/ # of -1 bytes, T2/ bp, T3/ scratch
SETZ T1, ;No bytes to start with
MOVE T2,[POINT 8,.MHUDA(P1)] ;Make bp
DO.
ILDB T3,T2 ;Get next byte
CAIE T3,377 ;Is it -1?
EXIT. ; -no, exit
AOJA T1,TOP. ;-yes, count one more and loop to next
ENDDO.
CALL DNSKBY ;Skip that many bytes (count in T1)
RET ; -ran out of bytes, dont return a message
;Return success indicating there is a message that should go to DNADLL
RETSKP
ENDSV.
;SCADRP is called when a received message is dropped. Update the counters
; and return
SCADRP: OPSTRM <AOS>,PBDDG,(PB) ;Update # of datagrams lost because of DECnet
RET
;RDGBUF - return datagram buffer to SCA receive queues
RDGBUF: BLCAL.(<XENT SC.RDG>,<CID,[0],P1>) ;[7.1037]Recycle buffer to free queue
BUG.(CHK,CINFRB,CIDLL,SOFT,<Failed to recycle buffer>,<<CID,SCACID>,<P1,BUFADR>>,<
Cause: CIDLL has received a datagram, and failed to return the buffer to
the SCA receive queue. One less buffer is now on the port
datagram receive queue.
Data: SCACID - The connect ID
BUFADR - The address of the buffer that could not be posted
>)
RET ;Return
SUBTTL Data transmission -- SCA callback -- "Message/datagram send complete"
;SCAMSC is called on SCA callbacks "Datagram send complete"
;
;Call: CB, PB and CID are set up
; T3/ address of buffer
; T4/ length of buffer
; CALL SCAMSC
; +1 always
;
;Context: SCA interrupt
;
;Ac usage: CB, PB, CID, MB, T1-T6
XRESCD
SCAMSC: MOVE T1,T3 ;Load address of buffer
CALLRET <XENT SC.RLD> ;[7.1037] and return it and return
SUBTTL Data transmission -- SCA callback -- "Dropped datagram"
;SCADDG is called on SCA callbacks "Datagram dropped"
;
;Call: CB, PB and CID are set up
; CALL SCADDG
; +1 always
;
;Context: SCA interrupt
XRESCD
SCADDG: OPSTRM <AOS>,PBSDG,(PB) ;Update # of datagrams lost beacuse of SCA
RET
SUBTTL Network management -- Read parameter
;NTMRED - called to read a network management parameter
;Call: P1/ address of NF block
; CALL NTMRED
; +1 on error with code in T1
; +2 on success
;
;Context: Jsys
XSWAPCD
NTMRED:
;Line or circuit?
LOAD T1,NFEID,(P1) ;Get entity ID
TMNN LILXC,+T1 ;Line or circuit?
IFSKP. ; -line
MOVX T1,<XADDR. LINTAB> ; then load address to line parameter table
MOVEI T2,NRLPAR ; and length
ELSE. ; -circuit
MOVX T1,<XADDR. CIRTAB> ; then load address to circuit param table
MOVEI T2,NRCPAR ; and length
ENDIF.
MOVX T3,NF.RED ;Load function code
CALLRET NTPARM ; and call common function to do work
SUBTTL Network management -- Set parameter
;NTMSET - called to set a network management parameter
;Call: P1/ address of NF block
; CALL NTMSET
; +1 on error with code in T1
; +2 on success
;
;Context: Jsys
;CIDLL does not support the setting of any parameters, so it will always
; return NF.OPF ("operation failure").
XSWAPCD
NTMSET: RNMXER (NF.OPF)
SUBTTL Network management -- Clear parameter
;NTMCLR - called to clear a network management parameter
;Call: P1/ address of NF block
; CALL NTMCLR
; +1 on error with code in T1
; +2 on success
;
;Context: Jsys
;CIDLL does not support the setting of any parameters, so it will always
; return NF.OPF ("operation failure").
XSWAPCD
NTMCLR: RNMXER (NF.OPF)
SUBTTL Network management -- Show and zero counters
;NTMCOU - called to show counters
;Call: P1/ address of NF block
; CALL NTMCOU
; +1 on error with code in T1
; +2 on success
;
;Context: Jsys
;NTMSZC - called to show and zero counters
;Call: P1/ address of NF block
; CALL NTMSZC
; +1 on error with code in T1
; +2 on success
;
;Context: Jsys
XSWAPCD
NTMCOU: SKIPA T3,[NF.COU] ;Load function code and skip
NTMSZC: MOVX T3,NF.SZC ;Load function code
LOAD T1,NFEID,(P1) ;Get entity ID
TMNE LILXC,+T1 ;Is it line?
RNMXER (NF.NDP) ; -yes, return "no data present"
;Call NTCTRS to do the real work
MOVX T1,<XADDR. COUTAB> ;Address of counter table
MOVEI T2,NRCOUN ;Length of counter table
CALLRET NTCTRS ;Read the counters and return
XRESCD
TNXEND
END