The merge command is initiated with the following call:
covered merge [options] existing_CDD CDD_to_merge+
The existing_CDD specifier relates to any existing CDD. If the -o option is not specified (see below for details), this file will be replaced with the merged CDD created by the merge. This is important to remember since merging is an irreversible operation (apart from generating a brand-new CDD from the design and dumpfile using the score command).
The CDD_to_merge specifier relates to any existing CDD that is not the existing_CDD. Though it is not illegal to merge the same CDD to itself, the result is the same as the original CDD. Note that all files treated as CDD_to_merge are guaranteed not to be modified by the merge command. These files are only read by Covered and then merged into the information contained in the existing_CDD.
Table 10.1. Options to merge Command
|-d directory||Directory to search for CDD files to include. This option is used in conjunction with the -ext option which specifies the file extension to use for determining which files in the directory are CDD files.|
Specifies how to handle exclusion reason resolution. If two or more CDD files being merged have exclusion reasons specified for the same coverage point, the exclusion reason needs to be resolved (unless it is the same string value). If this option is not specified and a conflict is found, Covered will interactively request input for each exclusion as to how to handle it. If this option is specified, it tells Covered how to handle all exclusion reason conflicts. The values are as follows:
|-ext extension||Used in conjunction with the -d option. If no -ext options are specified on the command-line, the default value of '.cdd' is used. Note that a period (.) should be specified.|
|-f filename||Name of file containing additional arguments to parse. You may specify this option more than once on a command-line.|
|-h||Outputs usage information for the merge command.|
|-m message||Allows the user to specify information about this CDD file which can be most anything. All messages found in merged CDDs will not be included in the merged CDD file.|
|-o filename||Name of CDD to write merged coverage information to (if not specified, the existing_CDD is overwritten with merged CDD).|
In previous versions of Covered, merging could only be accomplished when the top-most scored modules of each CDD were the same the module. Typically, the user would score the entire DUT in each diagnostic run, creating a set of CDD files that could be merged together to see the overall coverage of the diagnostic runs. While this capability still exists, Covered has recently been enhanced to allow more complex merging which provides some new capabilities.
Covered's merging algorithms have now been enhanced to allow merging CDDs that are generated for different parts of the same DUT. That is to say that each instance or sub-tree in the design can be scored individually and can be merged together. This ability allows for the DUT to be split up into smaller parts, scored in parallel and then merged back together to create a single CDD containing the scored results for the entire DUT. This parallel scoring capability can now be used to speed up scoring of large designs by taking advantage of multiple CPUs or CPU cores.
The following is an example of a DUT (top-most module of the DUT is block A) and the generic covered calls to score and create a single CDD file containing coverage for the entire design.
The following commands could be used to score the DUT comprised of blocks A-G in parallel (each score command below could occur in parallel with all other score commands).
covered score -t A -i MAIN.A -e B -e C -o A.cdd ... covered score -t B -i MAIN.A.B -e D -e E -o B.cdd ... covered score -t C -i MAIN.A.C -e F -e G -o C.cdd ... covered score -t D -i MAIN.A.C.D -o D.cdd ... covered score -t E -i MAIN.A.C.E -o E.cdd ... covered score -t F -i MAIN.A.B.F -o F.cdd ... covered score -t G -i MAIN.A.B.G -o G.cdd ... covered merge -o total.cdd A.cdd B.cdd C.cdd D.cdd E.cdd F.cdd G.cdd
In addition to allowing for parallel scoring, this merge method allows users to gain coverage results for smaller parts of the design and merge them into a single CDD containing coverage results for the entire design. This can be advantageous for those that are focused on increasing coverage for a specific portion of the design. Scoring can be done on that small part and the coverage gained can be added into an existing CDD containing coverage information for the entire design.
Using Figure 10.1, “Example of Verilog hierarchy to be scored in parallel” this type of merging would be performed as such:
covered score -t A -i MAIN.A total.cdd // The initial coverage file for the entire design covered score -t C -i MAIN.A.C -o CDE.cdd ... // Coverage obtained for modules C, D and E only covered merge total.cdd CDE.cdd
Along with the capability to merge different portions of the same large design, it is also possible to merge CDDs that were generated from different testbenches but have modules that are in common with each other. This allows different teams to work together to enhance coverage for shared design blocks. It is important to note that when this type of merging is being performed that the top-most covered module in the CDDs labeled CDD_to_merge exist exactly once in the CDD labeled existing_CDD for instance coverage information to be reported correctly. If the module is instantiated more than once in the existing_CDD, the CDD will still be merged but only the module coverage information will be correct.