xref: /original-bsd/lib/libc/vax/stdio.old/doprnt.c (revision 76571ef5)
1 /* @(#)doprnt.c	4.4 (Berkeley) 11/25/81 */
2 	# C library -- conversions
3 
4 .globl	__doprnt
5 .globl	__flsbuf
6 
7 #define vbit 1
8 #define flags r10
9 #define ndfnd 0
10 #define prec 1
11 #define zfill 2
12 #define minsgn 3
13 #define plssgn 4
14 #define numsgn 5
15 #define caps 6
16 #define blank 7
17 #define gflag 8
18 #define dpflag 9
19 #define width r9
20 #define ndigit r8
21 #define llafx r7
22 #define lrafx r6
23 #define fdesc -4(fp)
24 #define exp -8(fp)
25 #define sexp -12(fp)
26 #define nchar -16(fp)
27 #define sign -17(fp)
28 	.set ch.zer,'0			# cpp doesn't like single appostrophes
29 
30 	.align 2
31 strtab:		# translate table for detecting null and percent
32 	.byte	0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15
33 	.byte	16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31
34 	.byte	' ,'!,'",'#,'$, 0,'&,'','(,'),'*,'+,',,'-,'.,'/
35 	.byte	'0,'1,'2,'3,'4,'5,'6,'7,'8,'9,':,';,'<,'=,'>,'?
36 	.byte	'@,'A,'B,'C,'D,'E,'F,'G,'H,'I,'J,'K,'L,'M,'N,'O
37 	.byte	'P,'Q,'R,'S,'T,'U,'V,'W,'X,'Y,'Z,'[,'\,'],'^,'_
38 	.byte	'`,'a,'b,'c,'d,'e,'f,'g,'h,'i,'j,'k,'l,'m,'n,'o
39 	.byte	'p,'q,'r,'s,'t,'u,'v,'w,'x,'y,'z,'{,'|,'},'~,127
40 	.byte	128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143
41 	.byte	144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159
42 	.byte	160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175
43 	.byte	176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191
44 	.byte	192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207
45 	.byte	208,209,210,211,212,213,214,215,216,217,218,219,220,221,222,223
46 	.byte	224,225,226,227,228,229,230,231,232,233,234,235,236,237,238,239
47 	.byte	240,241,242,243,244,245,246,247,248,249,250,251,252,253,254,255
48 
49 	.align	1
50 __doprnt:
51 	.word	0xfc0			# uses r11-r6
52 	jbr	doit
53 
54 strfoo:
55 	clrl r4					# fix interrupt race
56 	jbr strok				# and try again
57 strmore:
58 	movzbl (r1)+,r2			# one char
59 	tstb strtab[r2]			# translate
60 	jeql stresc2			# bad guy in disguise (outbuf is full)
61 strout2:		# enter here to force out r2; r0,r1 must be set
62 	pushr $3				# save input descriptor
63 	pushl fdesc				# FILE
64 	pushl r2				# the char
65 	calls $2,__flsbuf		# please empty the buffer and handle 1 char
66 	tstl r0					# successful?
67 	jgeq strm1				# yes
68 	jbcs $31,nchar,strm1	# turn on sign bit of nchar to signify error
69 strm1:
70 	incl nchar				# count the char
71 	popr $3					# get input descriptor back
72 strout:			# enter via bsb with (r0,r1)=input descriptor
73 	movab strtab,r3			# table address
74 	movq *fdesc,r4			# output descriptor
75 	jbs $31,r4,strfoo		# negative count is a no no
76 strok:
77 	addl2 r0,nchar			# we intend to move this many chars
78 /******* Start bogus movtuc workaround  *****/
79 	clrl r2
80 	tstl    r0
81 	bleq    movdon
82 movlp:
83 	tstl    r4
84 	bleq    movdon
85 	movzbl  (r1)+,r3
86 	tstb    strtab[r3]
87 	bneq    1f
88 	mnegl   $1,r2
89 	decl    r1
90 	brb     movdon
91 1:
92 	movb    r3,(r5)+
93 	decl    r4
94 	sobgtr  r0,movlp
95   /******* End bogus movtuc workaround ***
96 	movtuc r0,(r1),$0,(r3),r4,(r5)
97 	movpsl r2                       /*  squirrel away condition codes */
98   /******* End equally bogus movtuc ****/
99 movdon: movq r4,*fdesc                  /*  update output descriptor */
100 	subl2 r0,nchar			# some chars not moved
101 	jbs $vbit,r2,stresc		# terminated by escape?
102 	sobgeq r0,strmore		# no; but out buffer might be full
103 stresc:
104 	rsb
105 stresc2:
106 	incl r0					# fix the length
107 	decl r1					# and the addr
108 	movl $1<vbit,r2			# fake condition codes
109 	rsb
110 
111 errdone:
112 	jbcs $31,nchar,prdone	# set error bit
113 prdone:
114 	movl nchar,r0
115 	ret
116 
117 doit:
118 	movab -256(sp),sp		# work space
119 	movl 4(ap),r11			# addr of format string
120 	movl 12(ap),fdesc		# output FILE ptr
121 	movl 8(ap),ap			# addr of first arg
122 	clrl nchar				# number of chars transferred
123 loop:
124 	movzwl $65535,r0		# pseudo length
125 	movl r11,r1				# fmt addr
126 		# comet sucks.
127 	movq *fdesc,r4
128 	subl3 r1,r5,r2
129 	jlss lp1
130 	cmpl r0,r2
131 	jleq lp1
132 	movl r2,r0
133 lp1:
134 		#
135 	bsbw strout				# copy to output, stop at null or percent
136 	movl r1,r11				# new fmt
137 	jbc $vbit,r2,loop		# if no escape, then very long fmt
138 	tstb (r11)+				# escape; null or percent?
139 	jeql prdone				# null means end of fmt
140 
141 	movl sp,r5			# reset output buffer pointer
142 	clrq r9				# width; flags
143 	clrq r6				# lrafx,llafx
144 longorunsg:				# we can ignore both of these distinctions
145 short:
146 L4a:
147 	movzbl (r11)+,r0		# so capital letters can tail merge
148 L4:	caseb r0,$' ,$'x-' 		# format char
149 L5:
150 	.word space-L5			# space
151 	.word fmtbad-L5			# !
152 	.word fmtbad-L5			# "
153 	.word sharp-L5			# #
154 	.word fmtbad-L5			# $
155 	.word fmtbad-L5			# %
156 	.word fmtbad-L5			# &
157 	.word fmtbad-L5			# '
158 	.word fmtbad-L5			# (
159 	.word fmtbad-L5			# )
160 	.word indir-L5			# *
161 	.word plus-L5			# +
162 	.word fmtbad-L5			# ,
163 	.word minus-L5			# -
164 	.word dot-L5			# .
165 	.word fmtbad-L5			# /
166 	.word gnum0-L5			# 0
167 	.word gnum-L5			# 1
168 	.word gnum-L5			# 2
169 	.word gnum-L5			# 3
170 	.word gnum-L5			# 4
171 	.word gnum-L5			# 5
172 	.word gnum-L5			# 6
173 	.word gnum-L5			# 7
174 	.word gnum-L5			# 8
175 	.word gnum-L5			# 9
176 	.word fmtbad-L5			# :
177 	.word fmtbad-L5			# ;
178 	.word fmtbad-L5			# <
179 	.word fmtbad-L5			# =
180 	.word fmtbad-L5			# >
181 	.word fmtbad-L5			# ?
182 	.word fmtbad-L5			# @
183 	.word fmtbad-L5			# A
184 	.word fmtbad-L5			# B
185 	.word fmtbad-L5			# C
186 	.word decimal-L5		# D
187 	.word capital-L5		# E
188 	.word fmtbad-L5			# F
189 	.word capital-L5		# G
190 	.word fmtbad-L5			# H
191 	.word fmtbad-L5			# I
192 	.word fmtbad-L5			# J
193 	.word fmtbad-L5			# K
194 	.word fmtbad-L5			# L
195 	.word fmtbad-L5			# M
196 	.word fmtbad-L5			# N
197 	.word octal-L5			# O
198 	.word fmtbad-L5			# P
199 	.word fmtbad-L5			# Q
200 	.word fmtbad-L5			# R
201 	.word fmtbad-L5			# S
202 	.word fmtbad-L5			# T
203 	.word unsigned-L5		# U
204 	.word fmtbad-L5			# V
205 	.word fmtbad-L5			# W
206 	.word hex-L5			# X
207 	.word fmtbad-L5			# Y
208 	.word fmtbad-L5			# Z
209 	.word fmtbad-L5			# [
210 	.word fmtbad-L5			# \
211 	.word fmtbad-L5			# ]
212 	.word fmtbad-L5			# ^
213 	.word fmtbad-L5			# _
214 	.word fmtbad-L5			# `
215 	.word fmtbad-L5			# a
216 	.word fmtbad-L5			# b
217 	.word charac-L5			# c
218 	.word decimal-L5		# d
219 	.word scien-L5			# e
220 	.word float-L5			# f
221 	.word general-L5		# g
222 	.word short-L5			# h
223 	.word fmtbad-L5			# i
224 	.word fmtbad-L5			# j
225 	.word fmtbad-L5			# k
226 	.word longorunsg-L5		# l
227 	.word fmtbad-L5			# m
228 	.word fmtbad-L5			# n
229 	.word octal-L5			# o
230 	.word fmtbad-L5			# p
231 	.word fmtbad-L5			# q
232 	.word fmtbad-L5			# r
233 	.word string-L5			# s
234 	.word fmtbad-L5			# t
235 	.word unsigned-L5		# u
236 	.word fmtbad-L5			# v
237 	.word fmtbad-L5			# w
238 	.word hex-L5			# x
239 fmtbad:
240 	movb r0,(r5)+			# print the unfound character
241 	jeql errdone			# dumb users who end the format with a %
242 	jbr prbuf
243 capital:
244 	bisl2 $1<caps,flags		# note that it was capitalized
245 	xorb2 $'a^'A,r0			# make it small
246 	jbr L4					# and try again
247 
248 string:
249 	movl ndigit,r0
250 	jbs $prec,flags,L20		# max length was specified
251 	mnegl $1,r0			# default max length
252 L20:	movl (ap)+,r2			# addr first byte
253 	locc $0,r0,(r2)			# find the zero at the end
254 	movl r1,r5			# addr last byte +1
255 	movl r2,r1			# addr first byte
256 	jbr prstr
257 
258 htab:	.byte	'0,'1,'2,'3,'4,'5,'6,'7,'8,'9,'a,'b,'c,'d,'e,'f
259 Htab:	.byte	'0,'1,'2,'3,'4,'5,'6,'7,'8,'9,'A,'B,'C,'D,'E,'F
260 
261 octal:
262 	movl $30,r2			# init position
263 	movl $3,r3			# field width
264 	movab htab,llafx	# translate table
265 	jbr L10
266 
267 hex:
268 	movl $28,r2			# init position
269 	movl $4,r3			# field width
270 	movab htab,llafx	# translate table
271 	jbc $caps,flags,L10
272 	movab Htab,llafx
273 L10:	mnegl r3,r6			# increment
274 	clrl r1
275 	addl2 $4,r5			# room for left affix (2) and slop [forced sign?]
276 	movl (ap)+,r0			# fetch arg
277 L11:	extzv r2,r3,r0,r1		# pull out a digit
278 	movb (llafx)[r1],(r5)+		# convert to character
279 L12:	acbl $0,r6,r2,L11		# continue until done
280 	clrq r6				# lrafx, llafx
281 	clrb (r5)			# flag end
282 	skpc $'0,$11,4(sp)		# skip over leading zeroes
283 	jbc $numsgn,flags,prn3	# easy if no left affix
284 	tstl -4(ap)				# original value
285 	jeql prn3			# no affix on 0, for some reason
286 	cmpl r3,$4			# were we doing hex or octal?
287 	jneq L12a			# octal
288 	movb $'x,r0
289 	jbc $caps,flags,L12b
290 	movb $'X,r0
291 L12b:	movb r0,-(r1)
292 	movl $2,llafx		# leading 0x for hex is an affix
293 L12a:	movb $'0,-(r1)	# leading zero for octal is a digit, not an affix
294 	jbr prn3			# omit sign (plus, blank) massaging
295 
296 unsigned:
297 lunsigned:
298 	bicl2 $1<plssgn|1<blank,flags	# omit sign (plus, blank) massaging
299 	extzv $1,$31,(ap),r0		# right shift logical 1 bit
300 	cvtlp r0,$10,(sp)		# convert [n/2] to packed
301 	movp $10,(sp),8(sp)		# copy packed
302 	addp4 $10,8(sp),$10,(sp)	# 2*[n/2] in packed, at (sp)
303 	blbc (ap)+,L14			# n was even
304 	addp4 $1,pone,$10,(sp)		# n was odd
305 	jbr L14
306 
307 patdec:					# editpc pattern for decimal printing
308 	.byte 0xAA			# eo$float 10
309 	.byte 0x01			# eo$end_float
310 	.byte 0				# eo$end
311 
312 decimal:
313 	cvtlp (ap)+,$10,(sp)		# 10 digits max
314 	jgeq L14
315 	incl llafx			# minus sign is a left affix
316 L14:	editpc $10,(sp),patdec,8(sp)	# ascii at 8(sp); r5=end+1
317 	skpc $' ,$11,8(sp)		# skip leading blanks; r1=first
318 
319 prnum:			# r1=addr first byte, r5=addr last byte +1, llafx=size of signs
320 				# -1(r1) vacant, for forced sign
321 	tstl llafx
322 	jneq prn3			# already some left affix, dont fuss
323 	jbc $plssgn,flags,prn2
324 	movb $'+,-(r1)		# needs a plus sign
325 	jbr prn4
326 prn2:	jbc $blank,flags,prn3
327 	movb $' ,-(r1)		# needs a blank sign
328 prn4:	incl llafx
329 prn3:	jbs $prec,flags,prn1
330 	movl $1,ndigit		# default precision is 1
331 prn1:	subl3 r1,r5,lrafx	# raw width
332 	subl2 llafx,lrafx	# number of digits
333 	subl2 lrafx,ndigit	# number of leading zeroes needed
334 	jleq prstr			# none
335 	addl2 llafx,r1		# where current digits start
336 	pushl r1			# movcx gobbles registers
337 		# check bounds on users who say %.300d
338 	movab 32(r5)[ndigit],r2
339 	subl2 fp,r2
340 	jlss prn5
341 	subl2 r2,ndigit
342 prn5:
343 		#
344 	movc3 lrafx,(r1),(r1)[ndigit]	# make room in middle
345 	movc5 $0,(r1),$ch.zer,ndigit,*(sp)	# '0 fill
346 	subl3 llafx,(sp)+,r1	# first byte addr
347 	addl3 lrafx,r3,r5	# last byte addr +1
348 
349 prstr:			# r1=addr first byte; r5=addr last byte +1
350 				# width=minimum width; llafx=len. left affix
351 				# ndigit=<avail>
352 	subl3 r1,r5,ndigit		# raw width
353 	subl3 ndigit,width,r0	# pad length
354 	jleq padlno				# in particular, no left padding
355 	jbs $minsgn,flags,padlno
356 			# extension for %0 flag causing left zero padding to field width
357 	jbs $zfill,flags,padlz
358 			# this bsbb needed even if %0 flag extension is removed
359 	bsbb padb				# blank pad on left
360 	jbr padnlz
361 padlz:
362 	movl llafx,r0
363 	jleq padnlx				# left zero pad requires left affix first
364 	subl2 r0,ndigit			# part of total length will be transferred
365 	subl2 r0,width			# and will account for part of minimum width
366 	bsbw strout				# left affix
367 padnlx:
368 	subl3 ndigit,width,r0	# pad length
369 	bsbb padz				# zero pad on left
370 padnlz:
371 			# end of extension for left zero padding
372 padlno:			# remaining: root, possible right padding
373 	subl2 ndigit,width		# root reduces minimum width
374 	movl ndigit,r0			# root length
375 p1:	bsbw strout				# transfer to output buffer
376 p3:	jbc $vbit,r2,padnpct	# percent sign (or null byte via %c) ?
377 	decl r0					# yes; adjust count
378 	movzbl (r1)+,r2			# fetch byte
379 	movq *fdesc,r4			# output buffer descriptor
380 	sobgeq r4,p2			# room at the out [inn] ?
381 	bsbw strout2			# no; force it, then try rest
382 	jbr p3					# here we go 'round the mullberry bush, ...
383 p2:	movb r2,(r5)+			# hand-deposit the percent or null
384 	incl nchar				# count it
385 	movq r4,*fdesc			# store output descriptor
386 	jbr p1					# what an expensive hiccup!
387 padnpct:
388 	movl width,r0	# size of pad
389 	jleq loop
390 	bsbb padb
391 	jbr loop
392 
393 padz:
394 	movb $'0,r2
395 	jbr pad
396 padb:
397 	movb $' ,r2
398 pad:
399 	subl2 r0,width			# pad width decreases minimum width
400 	pushl r1				# save non-pad addr
401 	movl r0,llafx			# remember width of pad
402 	subl2 r0,sp				# allocate
403 	movc5 $0,(r0),r2,llafx,(sp)	# create pad string
404 	movl llafx,r0			# length
405 	movl sp,r1				# addr
406 	bsbw strout
407 	addl2 llafx,sp			# deallocate
408 	movl (sp)+,r1			# recover non-pad addr
409 	rsb
410 
411 pone:	.byte	0x1C			# packed 1
412 
413 charac:
414 	movl (ap)+,r0		# word containing the char
415 	movb r0,(r5)+		# one byte, that's all
416 
417 prbuf:
418 	movl sp,r1			# addr first byte
419 	jbr prstr
420 
421 space:	bisl2 $1<blank,flags		# constant width e fmt, no plus sign
422 	jbr L4a
423 sharp:	bisl2 $1<numsgn,flags		# 'self identifying', please
424 	jbr L4a
425 plus:	bisl2 $1<plssgn,flags		# always print sign for floats
426 	jbr L4a
427 minus:	bisl2 $1<minsgn,flags		# left justification, please
428 	jbr L4a
429 gnum0:	jbs $ndfnd,flags,gnum
430 	jbs $prec,flags,gnump		# ignore when reading precision
431 	bisl2 $1<zfill,flags		# leading zero fill, please
432 gnum:	jbs $prec,flags,gnump
433 	moval (width)[width],width	# width *= 5;
434 	movaw -ch.zer(r0)[width],width	# width = 2*witdh + r0 - '0';
435 	jbr gnumd
436 gnump:	moval (ndigit)[ndigit],ndigit	# ndigit *= 5;
437 	movaw -ch.zer(r0)[ndigit],ndigit # ndigit = 2*ndigit + r0 - '0';
438 gnumd:	bisl2 $1<ndfnd,flags		# digit seen
439 	jbr L4a
440 dot:	clrl ndigit			# start on the precision
441 	bisl2 $1<prec,flags
442 	bicl2 $1<ndfnd,flags
443 	jbr L4a
444 indir:
445 	jbs $prec,flags,in1
446 	movl (ap)+,width		# width specified by parameter
447 	jgeq gnumd
448 	xorl2 $1<minsgn,flags		# parameterized left adjustment
449 	mnegl width,width
450 	jbr gnumd
451 in1:
452 	movl (ap)+,ndigit		# precision specified by paratmeter
453 	jgeq gnumd
454 	mnegl ndigit,ndigit
455 	jbr gnumd
456 
457 float:
458 	jbs $prec,flags,float1
459 	movl $6,ndigit			# default # digits to right of decpt.
460 float1:	bsbw fltcvt
461 	addl3 exp,ndigit,r7
462 	movl r7,r6			# for later "underflow" checking
463 	bgeq fxplrd
464 	clrl r7				# poor programmer planning
465 fxplrd:	cmpl r7,$31			# expressible in packed decimal?
466 	bleq fnarro			# yes
467 	movl $31,r7
468 fnarro:	subl3 $17,r7,r0			# where to round
469 	ashp r0,$17,(sp),$5,r7,16(sp)	# do it
470 	bvc fnovfl
471 		# band-aid for microcode error (spurious overflow)
472 	#	clrl r0				# assume even length result
473 	#	jlbc r7,fleven			# right
474 	#	movl $4,r0			# odd length result
475 	#fleven:	cmpv r0,$4,16(sp),$0		# top digit zero iff true overflow
476 	#	bneq fnovfl
477 		# end band-aid
478 	aobleq $0,r6,fnovfl		# if "underflow" then jump
479 	movl r7,r0
480 	incl exp
481 	incl r7
482 	ashp r0,$1,pone,$0,r7,16(sp)
483 	ashl $-1,r7,r0			# displ to last byte
484 	bisb2 sign,16(sp)[r0]		# insert sign
485 fnovfl:
486 	movab 16(sp),r1		# packed source
487 	movl r7,r6		# packed length
488 	pushab prnum	# goto prnum after fall-through call to fedit
489 
490 
491 	# enter via bsb
492 	#	r1=addr of packed source
493 	#	   16(r1) used to unpack source
494 	#	   48(r1) used to construct pattern to unpack source
495 	#	   48(r1) used to hold result
496 	#	r6=length of packed source (destroyed)
497 	#	exp=# digits to left of decimal point (destroyed)
498 	#	ndigit=# digits to right of decimal point (destroyed)
499 	#	sign=1 if negative, 0 otherwise
500 	# stack will be used for work space for pattern and unpacked source
501 	# exits with
502 	#	r1=addr of punctuated result
503 	#	r5=addr of last byte +1
504 	#	llafx=1 if minus sign inserted, 0 otherwise
505 fedit:
506 	pushab 48(r1)			# save result addr
507 	movab 48(r1),r3			# pattern addr
508 	movb $0x03,(r3)+		# eo$set_signif
509 	movc5 $0,(r1),$0x91,r6,(r3)	# eo$move 1
510 	clrb (r3)				# eo$end
511 	editpc r6,(r1),48(r1),16(r1)	# unpack 'em all
512 	subl3 r6,r5,r1			# addr unpacked source
513 	movl (sp),r3			# punctuated output placed here
514 	clrl llafx
515 	jlbc sign,f1
516 	movb $'-,(r3)+		# negative
517 	incl llafx
518 f1:	movl exp,r0
519 	jgtr f2
520 	movb $'0,(r3)+		# must have digit before decimal point
521 	jbr f3
522 f2:	cmpl r0,r6			# limit on packed length
523 	jleq f4
524 	movl r6,r0
525 f4:	subl2 r0,r6			# eat some digits
526 	subl2 r0,exp		# from the exponent
527 	movc3 r0,(r1),(r3)	# (most of the) digits to left of decimal point
528 	movl exp,r0			# need any more?
529 	jleq f3
530 	movc5 $0,(r1),$'0,r0,(r3)	# '0 fill
531 f3:	movl ndigit,r0		# # digits to right of decimal point
532 	jgtr f5
533 	jbs $numsgn,flags,f5	# no decimal point unless forced
534 	jbcs $dpflag,flags,f6	# no decimal point
535 f5:	movb $'.,(r3)+		# the decimal point
536 f6:	mnegl exp,r0		# "leading" zeroes to right of decimal point
537 	jleq f9
538 	cmpl r0,ndigit		# cant exceed this many
539 	jleq fa
540 	movl ndigit,r0
541 fa:	subl2 r0,ndigit
542 	movc5 $0,(r1),$'0,r0,(r3)
543 f9:	movl ndigit,r0
544 	cmpl r0,r6			# limit on packed length
545 	jleq f7
546 	movl r6,r0
547 f7:	subl2 r0,ndigit		# eat some digits from the fraction
548 	movc3 r0,(r1),(r3)	# (most of the) digits to right of decimal point
549 	movl ndigit,r0			# need any more?
550 	jleq f8
551 		# check bounds on users who say %.300f
552 	movab 32(r3)[r0],r2
553 	subl2 fp,r2
554 	jlss fb
555 	subl2 r2,r0			# truncate, willy-nilly
556 	movl r0,ndigit		# and no more digits later, either
557 fb:
558 		#
559 	subl2 r0,ndigit		# eat some digits from the fraction
560 	movc5 $0,(r1),$'0,r0,(r3)	# '0 fill
561 f8:	movl r3,r5			# addr last byte +1
562 	popr $1<1			# [movl (sp)+,r1] addr first byte
563 	rsb
564 
565 patexp:	.byte	0x03			# eo$set_signif
566 	.byte	0x44,'e			# eo$insert 'e
567 	.byte	0x42,'+			# eo$load_plus '+
568 	.byte	0x04			# eo$store_sign
569 	.byte	0x92			# eo$move 2
570 	.byte	0			# eo$end
571 
572 scien:
573 	incl ndigit
574 	jbs $prec,flags,L23
575 	movl $7,ndigit
576 L23:	bsbw fltcvt			# get packed digits
577 	movl ndigit,r7
578 	cmpl r7,$31				# expressible in packed decimal?
579 	jleq snarro				# yes
580 	movl $31,r7
581 snarro:	subl3 $17,r7,r0		# rounding position
582 	ashp r0,$17,(sp),$5,r7,16(sp) # shift and round
583 	bvc snovfl
584 		# band-aid for microcode error (spurious overflow)
585 	#	clrl r0				# assume even length result
586 	#	jlbc ndigit,sceven		# right
587 	#	movl $4,r0			# odd length result
588 	#sceven:	cmpv r0,$4,16(sp),$0		# top digit zero iff true overflow
589 	#	bneq snovfl
590 		# end band-aid
591 	incl exp			# rounding overflowed to 100...
592 	subl3 $1,r7,r0
593 	ashp r0,$1,pone,$0,r7,16(sp)
594 	ashl $-1,r7,r0		# displ to last byte
595 	bisb2 sign,16(sp)[r0]		# insert sign
596 snovfl:
597 	jbs $gflag,flags,gfmt		# %g format
598 	movab 16(sp),r1
599 	bsbb eedit
600 eexp:
601 	movl r1,r6		# save fwa from destruction by cvtlp
602 	subl3 $1,sexp,r0	# 1P exponent
603 	cvtlp r0,$2,(sp)	# packed
604 	editpc $2,(sp),patexp,(r5)
605 	movl r6,r1		# fwa
606 	jbc $caps,flags,prnum
607 	xorb2 $'e^'E,-4(r5)
608 	jbr prnum
609 
610 eedit:
611 	movl r7,r6		# packed length
612 	decl ndigit		# 1 digit before decimal point
613 	movl exp,sexp	# save from destruction
614 	movl $1,exp		# and pretend
615 	jbr fedit
616 
617 gfmt:
618 	addl3 $3,exp,r0		# exp is 1 more than e
619 	jlss gfmte		# (e+1)+3<0, e+4<=-1, e<=-5
620 	subl2 $3,r0		# exp [==(e+1)]
621 	cmpl r0,ndigit
622 	jgtr gfmte		# e+1>n, e>=n
623 gfmtf:
624 	movl r7,r6
625 	subl2 r0,ndigit		# n-e-1
626 	movab 16(sp),r1
627 	bsbw fedit
628 g1:	jbs $numsgn,flags,g2
629 	jbs $dpflag,flags,g2	# dont strip if no decimal point
630 g3:	cmpb -(r5),$'0		# strip trailing zeroes
631 	jeql g3
632 	cmpb (r5),$'.		# and trailing decimal point
633 	jeql g2
634 	incl r5
635 g2:	jbc $gflag,flags,eexp
636 	jbr prnum
637 gfmte:
638 	movab 16(sp),r1		# packed source
639 	bsbw eedit
640 	jbsc $gflag,flags,g1	# gflag now means "use %f" [hence no exponent]
641 
642 general:
643 	jbs $prec,flags,gn1
644 	movl $6,ndigit		# default precision is 6 significant digits
645 gn1:	tstl ndigit		# cannot allow precision of 0
646 	jgtr gn2
647 	movl $1,ndigit		# change 0 to 1, willy-nilly
648 gn2:	jbcs $gflag,flags,L23
649 	jbr L23			# safety net
650 
651 	# convert double-floating at (ap) to 17-digit packed at (sp),
652 	# set 'sign' and 'exp', advance ap.
653 fltcvt:
654 	clrb sign
655 	movd (ap)+,r5
656 	jeql fzero
657 	bgtr fpos
658 	mnegd r5,r5
659 	incb sign
660 fpos:
661 	extzv $7,$8,r5,r2		# exponent of 2
662 	movab -0200(r2),r2		# unbias
663 	mull2 $59,r2			# 59/196: 3rd convergent continued frac of log10(2)
664 	jlss eneg
665 	movab 196(r2),r2
666 eneg:
667 	movab -98(r2),r2
668 	divl2 $196,r2
669 	bsbw expten
670 	cmpd r0,r5
671 	bgtr ceil
672 	incl r2
673 ceil:	movl r2,exp
674 	mnegl r2,r2
675 	cmpl r2,$29			# 10^(29+9) is all we can handle
676 	bleq getman
677 	muld2 ten16,r5
678 	subl2 $16,r2
679 getman:	addl2 $9,r2			# -ceil(log10(x)) + 9
680 	jsb expten
681 	emodd r0,r4,r5,r0,r5		# (r0+r4)*r5; r0=int, r5=frac
682 fz1:	cvtlp r0,$9,16(sp)		# leading 9 digits
683 	ashp $8,$9,16(sp),$0,$17,4(sp)	# as top 9 of 17
684 	emodd ten8,$0,r5,r0,r5
685 	cvtlp r0,$8,16(sp)		# trailing 8 digits
686 		# if precision >= 17, must round here
687 	movl ndigit,r7			# so figure out what precision is
688 	pushab scien
689 	cmpl (sp)+,(sp)
690 	jleq gm1			# who called us?
691 	addl2 exp,r7			# float; adjust for exponent
692 gm1:	cmpl r7,$17
693 	jlss gm2
694 	cmpd r5,$0d0.5			# must round here; check fraction
695 	jlss gm2
696 	bisb2 $0x10,8+4(sp)		# increment l.s. digit
697 gm2:		# end of "round here" code
698 	addp4 $8,16(sp),$17,4(sp)	# combine leading and trailing
699 	bisb2 sign,12(sp)		# and insert sign
700 	rsb
701 fzero:	clrl r0
702 	movl $1,exp		# 0.000e+00 and 0.000 rather than 0.000e-01 and .000
703 	jbr fz1
704 
705 	.align 2
706 lsb: .long 0x00010000		# lsb in the crazy floating-point format
707 
708 	# return 10^r2 as a double float in r0||r1 and 8 extra bits of precision in r4
709 	# preserve r2, r5||r6
710 expten:
711 	movd $0d1.0,r0			# begin computing 10^exp10
712 	clrl r4				# bit counter
713 	movad ten1,r3			# table address
714 	tstl r2
715 	bgeq e10lp
716 	mnegl r2,r2			# get absolute value
717 	jbss $6,r2,e10lp		# flag as negative
718 e10lp:	jbc r4,r2,el1			# want this power?
719 	muld2 (r3),r0			# yes
720 el1:	addl2 $8,r3			# advance to next power
721 	aobleq $5,r4,e10lp		# through 10^32
722 	jbcc $6,r2,el2			# correct for negative exponent
723 	divd3 r0,$0d1.0,r0		# by taking reciprocal
724 	cmpl $28,r2
725 	jneq enm28
726 	addl2 lsb,r1			# 10**-28 needs lsb incremented
727 enm28:	mnegl r2,r2			# original exponent of 10
728 el2:	addl3 $5*8,r2,r3		# negative bit positions are illegal?
729 	jbc r3,xlsbh-5,eoklsb
730 	subl2 lsb,r1			# lsb was too high
731 eoklsb:
732 	movzbl xprec[r2],r4		# 8 extra bits
733 	rsb
734 
735 	# powers of ten
736 	.align	2
737 ten1:	.word	0x4220,0,0,0
738 ten2:	.word	0x43c8,0,0,0
739 ten4:	.word	0x471c,0x4000,0,0
740 ten8:	.word	0x4dbe,0xbc20,0,0
741 ten16:	.word	0x5b0e,0x1bc9,0xbf04,0
742 ten32:	.word	0x759d,0xc5ad,0xa82b,0x70b6
743 
744 	# whether lsb is too high or not
745 	.byte 1:0,1:0,1:0,1:0,1:1,1:0,1:1,1:0	# -40 thru -33
746 	.byte 1:0,1:1,1:0,1:0,1:0,1:0,1:1,1:0	# -32 thru -25
747 	.byte 1:0,1:0,1:1,1:1,1:1,1:1,1:0,1:0	# -24 thru -17
748 	.byte 1:0,1:1,1:0,1:0,1:1,1:1,1:1,1:1	# -16 thru -9
749 	.byte 1:1,1:1,1:1,1:0,1:0,1:0,1:0,1:1	# -8  thru -1
750 xlsbh:
751 	.byte 1:0,1:0,1:0,1:0,1:0,1:0,1:0,1:0	# 0 thru 7
752 	.byte 1:0,1:0,1:0,1:0,1:0,1:0,1:0,1:0	# 8 thru 15
753 	.byte 1:0,1:0,1:0,1:0,1:0,1:0,1:0,1:0	# 16 thru 23
754 	.byte 1:0,1:1,1:1,1:0,1:1,1:1,1:1,1:1	# 24 thru 31
755 	.byte 1:1,1:1,1:1,1:1,1:1,1:1,1:1    	# 32 thru 38
756 
757 	# bytes of extra precision
758 	.byte           0x56,0x76,0xd3,0x88,0xb5,0x62	# -38 thru -33
759 	.byte 0xba,0xf5,0x32,0x3e,0x0e,0x48,0xdb,0x51	# -32 thru -25
760 	.byte 0x53,0x27,0xb1,0xef,0xeb,0xa5,0x07,0x49	# -24 thru -17
761 	.byte 0x5b,0xd9,0x0f,0x13,0xcd,0xff,0xbf,0x97	# -16 thru -9
762 	.byte 0xfd,0xbc,0xb6,0x23,0x2c,0x3b,0x0a,0xcd	# -8  thru -1
763 xprec:
764 	.byte 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00	# 0  thru 7
765 	.byte 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00	# 8  thru 15
766 	.byte 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00	# 16 thru 23
767 	.byte 0x00,0xa0,0xc8,0x3a,0x84,0xe4,0xdc,0x92	# 24 thru 31
768 	.byte 0x9b,0x00,0xc0,0x58,0xae,0x18,0xef     	# 32 thru 38
769