Google
 

Trailing-Edge - PDP-10 Archives - kludoc - documentation/kl-ucode-release.doc
There are 2 other files named kl-ucode-release.doc in the archive. Click here to see a list.


KL MICROCODE RELEASE DOCUMENT -- V326
September 1983






























Copyright (C)1977,1978,1979,1980,1981,1982,1983,1984
Digital Equipment Corporation, Maynard, Mass.


This software is furnished under a license for use only  on  a  single
computer system and may be copied only 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 except for
use on such system and to one  who  agrees  to  these  license  terms.
Title  to  and  ownership of the software shall at all times remain in
DEC.

The information in this software is subject to change  without  notice
and  should  not  be  construed  as  a commitment by Digital Equipment
Corporation.

DEC assumes no responsibility  for  the  use  or  reliability  of  its
software on equipment which is not supplied by DEC.
KL MICROCODE RELEASE DOCUMENT -- V326                           Page 2


V326RG.DOC -- v326
September 1983









1.0  SUMMARY


     1.  Microcode version 326  is  the  tenth  release  of  the  KL10
         microcode.   Microcode  version  326 appears as KLX.MCB only.
         It supports a model B KL10 for TOPS-20 V5.1 or earlier only.

         This is the fifth release of the KL10 microcode that supports
         an  'Extended  KL10'  described  in  the  Hardware  Reference
         Manual.  Only KLX.RAM supports the 'Extended KL10'.  (KLX.MCB
         is  derived  from  KLX.RAM  and  does  support  the 'Extended
         KL10'.)

         The KL10 microcode  is  different  for  TOPS-10  and  TOPS-20
         systems.   The  KL10  microcode  is different for model A and
         model B  processors.   (What  engineering  calls  a  model  B
         processor  is  sometimes  known  as  a  KL10-PV.) The TOPS-10
         microcode for model A processors is called U.RAM, for model B
         processors  it  is UB.RAM.  The TOPS-20 microcode for model A
         processors is called KLL.RAM, for model B  processors  it  is
         called  KLX.RAM.   The  PDP-11  front  end requires different
         formats of the microcode files.  These translated files  have
         different names.  TOPS-10 model A:  UA.MCB.  TOPS-10 model B:
         UB.MCB.   TOPS-20  model  A:   KLA.MCB.   TOPS-20  model   B:
         KLX.MCB.

     2.  The KL10 microcode is  not  dependent  on  any  monitor,  but
         because  of  hardware  differences  between  the  various -10
         processors, a monitor must  be  built  specifically  for  the
         KL10.   KL10  microcode  version 326 has been tested with the
         TOPS-20 release 5.1 monitor.

See LCREV for information regarding microcode requirements for various
hardware revision levels.



2.0  KNOWN DEFICIENCIES


     When executing a CVTBDO using a One Word Global Byte Pointer  the
fill is taken from E0 rather than E0+1.
KL MICROCODE RELEASE DOCUMENT -- V326                           Page 3


3.0  FIXED PROBLEMS

The following problems have been fixed:

326  Change VMA restoration in INC2WD and CNV2WD (See edits 320 and 307)
     to use RSTR VMA_MQ in order to keep the global sense of the
     reference. This was causing ILDBs of OWGs in shadow memory to
     save the incremented byte pointer in the ACs instead of memory.
325  Add VMA/LOAD to local indexed EA computation for EXTEND E1 to make
     it read the section number from VMA instead of PCS (!) if the index
     is section local.
324  Force the XADDR conditional to use RPW type references for DPB and
     IDPB if the SMP conditional is on, even if one word globals are not
     active.
323  Add missing constraint near NOT.WR, accidentally broken by 322.
322  Generate the A(cessible) bit in a page fail word caused by a read
     violation if the page is otherwise accessible and if no CST is present.
     This could be fixed for the CST present case as well, but has been
     deferred since we are tight on space and no one seems to need it
     anyway.
321  Prevent statistics microcode from losing traps by forcing NICOND
     dispatch 11 to ignore the statistics and take the trap.
320  Restore the VMA again in INC2WD (broken by 307), since the state
     register bits may have changed in the interim.  This was causing
     PXCT to do surprising things (mostly bad).
317  Originally, this was an attempt to uncount multiply counted op
     codes which resulted from interrupts during long instructions.
     That project has been shelved for now.  Instead, the second
     NICOND dispatch during op code counting has had its final constraint
     fixed.
316  Make counting only version compatible with time and counting by making
     counting only version use TRX2 and TRX3, removing physical contiguity
     requirement.
315  Op code counting lives again!  The setup code activated by DATAO PI
     was attempting to write the TRX registers with data fresh from memory,
     resulting in parity checks when it was used (see edit 73, for example).
     Juggle code to overlap next address calculation with parity wait.
314  Add CST.WRITE conditional to facilitate assembly of microcode
     without the CST writable bit (see edit 303).
313  Put TIME/3T on XFERW, as the assembler was getting the wrong
     value with both AR_MEM and ARX_MEM macros present.
312  Fix definition of BYTE RPW to include a write test.  This was
     causing the SMP version of DPB to hang when memory was readable
     but not writable.
311  Make all IOP function 7 style of references look in the cache.
310  Improve the fix in 307 to save the computed E0+1 in FILL during
     OWGBP conversion and to restore the VMA from there when done.
     Also, make sure that the VMA is initialized to PC for all cases
     when doing effective address calculations for two word globals
     in string instructions.  307 was not enough to clean up the
     CMPSx fill problem, since VMA HELD was never loaded.
     Force EXT2WD to prereference AC4 and AC5 so that glitch discovered
     for second edit 210 will not be activated.
307  Restore VMA from MQ at end of CNV2WD (and remove it from INC2WD,
     saving a word in the process).  This was causing CMPSx to load
KL MICROCODE RELEASE DOCUMENT -- V326                           Page 4


     a random fill word and MOVSLJ to store to a random place when the
     source length was zero if one word globals were in use.
     Force page fail code to look for ARX as well as AR parity errors
     (now possible with BYTE RPW implemented).
     Make sign extension of E1 go to right place in EXTEND decoding of
     offset instructions (broken in 301).
306  Add University of Essex code to statistics (TRACKS) code to make
     it work with address break enabled.
305  Fix CST write bit logic to not test bit 18 when reading.
304  Switch byte read interlock from LDB to DPB (broken in 303).
303  Implement bit 18 of a CST entry as a write enable bit in addition
     to all the other write enable functions.
     Knock one cycle out of byte deposit where the byte is being
     deposited into the high order byte of a word.
     Implement the SMP conditional for extended addressing by
     replicating all the byte effective address calculation code for
     DPB.  This is unfortunate, but necessary due to the huge dispatch
     table that ends this subroutine.
302  Move XFERW out of EIS (which no longer absolutely requires it
     in line) into SKPJMP (more in the heart of things).  Also
     juggle comment lines and code layout to reduce the listing
     size a bit and to force some of the .TOC lines into the table
     of contents (even though the code nearby may be suppressed).
301  Fix ADJBP so that instructions which occur at the last word on
     a page do not cause a page failure of some random type (one cycle
     too many between I FETCH and NICOND).
     Fix effective address calculation for EXTEND so that only offset
     instructions (and not GSNGL, for example) will have E1 sign
     smeared.
     Implement XJRST.  Also force JSP and JSR to do full 30 bit
     effective address calculations.
300  ADD LOAD OF VMA FROM PC IN PUTDST TO GET THE SECTION ADDRESS
     CORRECT ON THE STRING INSTRUCTIONS.
277  Add EA CALC table for SMP configurations of extended addressing
     for TOPS-10.  (TOPS-20 paging)
276  Force global EA CALC for EXTEND instructions in PUTDST.



4.0  RELATED DOCUMENTATION


The microcode is implicitly documented in the System Reference Manual,
in  that  it  is  an  implementation  of  a  PDP-10.   The  only other
documentation is in the listing and prints of the KL10 processor.
KL MICROCODE RELEASE DOCUMENT -- V326                           Page 5


5.0  FUNCTION OF THE DIFFERENT MICROCODE VERSIONS.


U.RAM is the microprogram which directs  the  operation  of  the  KL10
model  A  hardware  to  emulate the behavior of a PDP-10 for a TOPS-10
system.  UB.RAM is the microprogram which directs the operation of the
KL10  model  B  hardware (KL10-PV) to emulate the behavior of a PDP-10
for a TOPS-10 system.  UB.RAM should look exactly like U.RAM to PDP-10
software.



KLL.RAM is the microprogram which directs the operation  of  the  KL10
model  A  hardware  to  emulate the behavior of a PDP-10 for a TOPS-20
system.  KLX.RAM is the microprogram which directs  the  operation  of
the  KL10  model  B hardware to emulate the behavior of a PDP-10 for a
TOPS-20 system.  KLX.RAM implements the 'Extended KL10'  explained  in
the Hardware Reference Manual.



[End of UCODE.RND]