ifpsp060/src/fpsp.S

665 # since, I believe, only NORMs and DENORMs can come through here,
997 # underflow exception. Since this is incorrect, we check here to see
1141 #	Two main instruction types can enter here: (1) DENORM or UNNORM	#
1166 # (a Trace stack frame must be created here, too). If an FP exception	#
1221 # exception, then the a7' calculated here is wrong since it doesn't
1318 #	INEX1	: none (packed doesn't go through here)
1395 # here, we insert the correct fsave status value into the fsave frame for the
1452 # here, we count on norm not to alter a0...
1484 # here, we count on norm not to alter a0...
1516 # call here. just figure out what it is...
1547 #	INEX1	: none (packed doesn't travel through here)
1561 # it here. if it was used from supervisor mode, then we have to handle this
1589 # here, we're counting on the top of the stack to be the old place-holders
1672 	short		tbl_fu_out	- tbl_fu_out	# INEX1 won't make it here
1786 # fmove in and out enter here.
1973 # here, we insert the correct fsave status value into the fsave frame for the
2456 # exception. Therefore, the FPU could be disabled when entering here.
2484 # here, we will have:
2767 # do it here.
2945 # F Emulator" exception. So, here we create an 8-word stack frame
2997 # here, we actually create the 8-word frame from the 4-word frame,
3131 # here, we simply see if the operand in the fsave frame needs to be "unskewed".
3134 # cause an operr so we don't need to check for them here.
3152 # operand error exceptions. we do this here before passing control to
3156 # through here. we simply need to test the sign of the src
3161 # exceptions, they won't pass through here since they are caught
3199 	short		foperr_exit  - tbl_operr # packed won't enter here
3203 	short		tbl_operr    - tbl_operr # packed won't enter here
3327 # here, we simply see if the operand in the fsave frame needs to be "unskewed".
3330 # fixed here.
3348 # snan exceptions. we do this here before passing control to
3352 # through here. since packed format operations already were handled by
3353 # fpsp_unsupp(), then we need to do nothing else for them here.
3621 # w/ an exponent value of 0x401e. we convert this to extended precision here.
3638 # to extended before arriving here. Therefore, we need to retain the
3713 # inexact exceptions. we do this here before passing control to
3717 # through here. so can double and single precision.
3719 # exceptions, they won't pass through here since they are caught
3790 # here, we simply see if the operand in the fsave frame needs to be "unskewed".
3960 #	There are two main cases of instructions that may enter here to	#
4296 # the ftrapcc instruction should take a trap. so, here we must create a
4338 # I am assuming here that an "fs<cc>.b -(An)" or "fs<cc>.b (An)+" instruction
4426 # as usual, we have to check for trace mode being on here. since instructions
4427 # modifying the supervisor stack frame don't pass through here, this is a
4454 # the fscc instruction should take a trace trap. so, here we must create a
4975 # fdbcc emulation failed. here, we create an access error frame
5309 # here, the operation may underflow iff the precision is sgl or dbl.
6638 # This catch is added here for the '060 QSP. Originally, the call to
7208 #--entry point for EXP(X), here X is finite, non-zero, and not NaN's
7371 #--entry point for EXPM1(X), here X is finite, non-zero, non-NaN
7597 #--entry point for EXPM1(X), here X is denormalized
7649 # here, we build the result in a tmp location so as not to disturb the input
8428 # if these 2 lines weren't here, the correct answer would be returned
10162 #	- we only return the EXOP here if either underflow or	#
10992 # has been decoded. Flow continues here where we now decode
11355 #	The only exceptions capable of passing through here are		#
11372 # FP_DST here...
11376 # frem,fmod, and fscale would get the dst trashed here. But, for
12047 #	Unnorms don't pass through here.				#
13219 # here. For norms, load the rounding mode/prec, execute a "fint", then	#
13325 # here. For norms, load the rounding mode/prec, execute a "fintrz",	#
13444 #	Unnorms don't pass through here.				#
13842 # 'N' bit for a negative QNAN or SNAN input so we must squelch it here.
15656 # we're on the line here and the distinguising characteristic is whether
18178 # have our own fcalc_ea() routine here. If an access error is flagged	#
18254 # here, we convert it to be just like the others...
18979 # _imem_read_word() increased the extwptr by 2. need to adjust here.
19068 # we clear the value here if it should actually be suppressed.
19386 # need to set immediate data flag here since we'll need to do
19393 # here, the <ea> is stacked correctly. however, we must update the
19405 # mode so we don't even worry about this tricky case here : )
20220 # Only "Unimplemented Data Type" exceptions enter here. The operand
20266 # Only "Unimplemented Data Type" exceptions enter here. The operand
20312 # Only "Unimplemented Data Type" exceptions enter here. The operand
20359 # Only "Unimplemented Data Type" exceptions enter here. The operand
20490 # here, we know that the operand would UNFL if moved out to single prec,
20719 # here, we know that the operand would UNFL if moved out to double prec,
21621 # careful that we don't take an exception here.				#
21874 # subtle step here; or in the g,r,s at the bottom of FTEMP_LO to minimize
22951 # its binary representation here. Else, call decbin() which will	#
23770 	bra.b		A7_str		# finished here
24606 #	Flow jumps here when an FP data fetch call gets an error	#
24686 # here's where we actually create the access error frame from the
24717 # we do that here.