Enter your search terms Submit search form
	Web pdp-10.trailing-edge.com

	Debugging facilities in Dec-10 Simula-67 compiler (test version)
	================================================================

I. Test character handling:
   =======================
	Whenever the special symbol  occurs in the source file,
	it will cause the next three characters (2 letters and 1 digit)
	to be scanned, and the resulting identification is then matched
	against three tables:

	ZLSTT1 - a match will cause a debug switch in reg x1mask to be
	         set on. (used e.g. to start trace actions)
	ZLSTT2 - a match will cause a switch to be set off
		 (e.g. stop tracing)
	ZLSTT3 - a match will invoke a routine.
		 (used e.g. to initiate a dump)

	If no match occurs, a debug symbol followed by the identification
	found, will be output to file IC1, in order to be handled
	later by pass 2.

	So far, the following identifications have been
	implemented:

	Ident	Action

pass 1:	LS1	start successive output of symbols found by scanner
	LS0	setoff LS1
	SR1	start successive output of entry/exit of
		syntactical routines
	SR0	setoff SR1
	SD1	start successive output of block stack entries 5 and 6,
		for every execution of routines SDBEG and SDEND
	SD0	setoff SD1
	O11	start successive output of symbols output to IC1
	O10	setoff O11
	IC1	init dump of IC1 (to be printed when pass 1 is completed)
	DF1	init dump of DF1 (to be printed when pass 1 is completed)


pass 2:	M20	stop trace program
	M21	start trace program
	M22	abort with dump
	M23	set SM2DB3
	CG0	setoff CG1
	CG1	dump expression tree before compilation
	O21	output assembly list
	O22	trace all fixups

II. Debug File handling:
    ===================
	In order to facilitate edited test output, a special debug
	file is constructed, into which file any submodule can place
	debug information records for later output.
	Every record in the debug file starts with a control half word,
	indicating the meaning of the record. This half word is divided
	into six fields, each one comprising of 3 bits of information.
	case 1:	first field is zero.
		meaning of second field:	1 = new page
						2 = new line
						3 = define new identifier,
						    in this case the control
						    half word is followed by
						    a half word containing
						    an index (1024-4095) and
						    by 4 half words containing
						    the identifier in sixbit

	case 2:	first field nonzero.
			every subsequent field indicates how the following
			half words of data should be output.
			codes:	0 = end of record
				1 = convert field to octal
				2 = convert field to decimal
				3 = print a symbol or an identifier.
				    half word contains corr index
				4 = print text.
				    the next 2 half words contains
				    char in sixbit
				5 = print text.
				    half word contains index to text table
				6 = print text.
				    half word contains three sixbit char
				7 = print text.
				    the next 2 half words contains
				    5 characters in ASCII (left adjusted)

	Placing records in the debug file is accomplished by calls
	=================================
	to one or more of the subroutines O1DB1-O1DB7, where
	the last digit indicates how many half words the record
	consists of. On call to such a routine, the record
	is to be placed in the system stack.

	ex. record consists of control half word and 4 data half words:

 
		    		/                   /  return address   /
				/ data half word 4  /     not used      /
				/ data half word 2  / data half word 3  /
		     stacktop:  / control half word / data half word 1  /


	Generated output:
	================
		Each record written, starts in a position, which
		is the next multiple of 20. If remaining space of
		present line is not sufficient, the record is written at
		the beginning of next line. One blank separates the fields
		from each other.
	ex.
	debug file:	/520000/000000/000010/
			/020000/
			/630000/SR /000067/
			/030000/002000/SKR/#P /   /   /
			/630000/LS0/002000/
			/020000/
			/500000/000002/

	output:
			DEBUG OUTPUT SIMULA-67 VERSION: 8
		SR  BPROG           LS0 SKR#P
		START PASS2

General record dump macro:
=========================
	Define Record Dump (DRD)

	Format:	DRD(R,<F1,F2,...,Fk>)
	------

		R = record name
		Fi= field name (previously defined by a call to
				Define Field (DF) - macro)

	    DRD expands a subroutine named .R and it also
		defines a macro R, which when later called, will
		produce a symbolic dump of all the Fi-fields that were
		specified in the call to the DRD-macro.
		The dump will be octal and if the bit length, L, of a
		field is not a multiple of 3, then ENTIER(L/3)+1 digits
		will appear in the dump.
		The fields will be dumped in the order of appearance
		in the DRD-call.

		The address of the record (first word) is supposed
		to be in reg X1 at execution.

	    DRD may not be called inside a PROC/EPROC body
		(but the expanded routine may of course).

ex. of use:
----------

DRD	DCZHE,<ZDETYP,ZHETYP,ZHESOL,ZHEDLV,ZHEEBL,ZHELEN,ZHEBNM,ZHEFIX,ZLINK>
.
.
.
PROC
	.
	.
	.
	DCZHE
	.
	.
	.
EPROC
.
.
.


Output on execution:
-------------------

=====
DCZHE AT ADDRESS 400982
=====

ZDETYP=1
ZHETYP=2
ZHESOL=003
ZHEDLV=777776
ZHEEBL=02
ZHELEN=0132
ZHEBNM=002
ZHEFIX=0000
ZLINK=000118