1/1
Testbench bug fixes for Icarus windows
by Unknown on Nov 13, 2006 |
Not available! | ||
I noticed the following issues when running the testbench with Icarus
verilog on cygwin/windows. Refer to file tb_ethernet.v, version 1.34 (the latest). 1. Line 2442, I had to change ETH_MIIRX_DATA to ETH_MIITX_DATA, since we are trying to read back and verify a value written to ETH_MIITX_DATA. 2. In lines 4259 - 4261 I had to change the declarations to unsigned: reg [31:0] i_data; // was integer i_data; reg [31:0] i_length; // was integer i_length; reg [31:0] tmp_len; // was integer tmp_len; because expressions like `MEMORY_BASE + i_length[1:0] were causing sign extension if i_length was, for example, 3. This was causing reads and writes to memory locations just below MEMORY_BASE. With those changes, things are MUCH better, but subsequent tests still fail like this: Heading: MAC HALF DUPLEX FLOW TEST *************************************************************************************** *************************************************************************************** ************************************************************************************* At time: 487283647 Test: TEST 0: DEFER, COLL. AND LATE COLL. WHILE TRANSMITTING AND RECEIVING FRM. AT 4 TX/RX BD ( 10Mbps ) *FAILED* because Transmit should NOT start more than 1 CLK after CarrierSense and causing Collision Does anyone have the fixes to make the testbench run on Icarus windows under cygwin? Any information on how to fix this last problem would be greatly appreciated. Does anyone still maintain the testbench? Thanks, Chuck Heller |
1/1