Open MikeOpenHWGroup opened 1 month ago
Yes, probably this should not be bind to RVFI variable instead of a UVM variable
Indeed. AFAIK, there is no phase difference between the clk_i
input to this module and clknrst_if.clk
in the UVM "clk-and-reset" interface. So, maybe it is just as simple as replacing:
always @(posedge clknrst_if.clk) begin
with
always @(posedge clk_i) begin
Will that work?
Bug Description
Hi @mario. In #184 you indicated that compiling RVFI requires inclusion of the cv32e20-dv environment. This sounds odd. Looking at cve2_cs_registers.sv we see:
The reference to
clknrst_if.clk
is problematic as it places a dependency on the testbench upon the RTL. This is bad practise in general. Is there a way to implement the RVFI without that dependance?