Answers Database
 
 
 A1.5i (M1.5.25) map: FATAL_ERROR:xvkma:xvkmaslice.c:4041:1.97 - domergeslices got empty slot for frag 
 
 Record #5174
Problem Title: 
A1.5i (M1.5.25) map: FATAL_ERROR:xvkma:xvkmaslice.c:4041:1.97 - domergeslices got empty slot
for frag 
 
 
 
Problem Description: 
Urgency : Standard 
 
General Description: 
Map error out on a LUT component with the following error: 
FATAL_ERROR:xvkma:xvkmaslice.c:4041:1.97 - domergeslices got 
empty slot for frag: comp ix5298 
 
ngdbuild doen't issue any related warning. 
 
Input and output of the  LUT seem to be connected properly to 
LUTs and FDCPE. 
 
 
Solution 1: 
 
There are actually two problems in the design that might 
cause the error: 
 
     	1. User uses FDRS type flops (in this case FDCPE) but 
        	   connects GND to the SET pin, this will trigger a 
        	   special trimming procedure to try to trim the pin 
        	   away, this procedure does not leave the flop in the 
        	   right state for pack/merge. 
 
     	2. When packing or merging a FDRS type flop, we have 
        	   the option of moving the flop from top half (which 
        	   will use up a G-lut for route-through) to the 
        	   bottom half. But there is a bug in the pack/merge 
        	   code that fail to identify all the  variances of 
        	   FDRS configuration (e.g. It fails to update the 
        	   celltform BYMUX if we have 2 FDRS type flops in 
        	   the source slice.) 
 
     	problem 1 can be avoided if the user will not use FDRS 
     	type flop with GND connected SET pin (exchange FDCPE 
     	with FDCE). 
 
     	Problem 2 can only be avoided  by setting the variable 
     	NO_DELAY_BASED_MERGE, to avoid problem-causing merge. 
     	It one case it was also  necessary to set the variable 
     	NO_MERGE_OTHERS. 
 
     	Work Stations: 
     	setenv NO_DELAY_BASED_MERGE 1 
     	setenv NO_MERGE_OTHERS 1 
 
     	PCs: 
     	set NO_DELAY_BASED_MERGE=1  (note: there should be no spaces around "=") 
     	set NO_MERGE_OTHERS=1 
 
 
 
 
 
 
 
 End of Record #5174 - Last Modified: 02/25/99 10:09  |