



or1200 test case
by sudhanshubillore on Mar 18, 2013 |
sudhanshubillore
Posts: 16 Joined: Aug 10, 2012 Last seen: Apr 18, 2023 |
||
Hi all,
I am working on OR1200 processor. I modified current cache architecture of or1200 to 4-way set associative cache architecture. I modified both IC & DC cache. Now I am trying to run all existing test cases by disabling IC & DC just to ensure while modifying IC & DC I didn't disturb other modules. I found "or1200-dsxinsn" test case is not getting over, when I am terminating it through ctrl+c then message like "unspecified location accessed" is coming. Please suggest how to debug it? How to debug such an exception related problems? Thank You. regards, Sudhanshu Billore |
RE: or1200 test case
by stekern on Mar 19, 2013 |
stekern
Posts: 84 Joined: Apr 28, 2009 Last seen: Nov 10, 2016 |
||
You can generate a VCD by running the test with:
make rtl-test TEST=or1200-dsxinsn VCD=1 |
RE: or1200 test case
by sudhanshubillore on Mar 20, 2013 |
sudhanshubillore
Posts: 16 Joined: Aug 10, 2012 Last seen: Apr 18, 2023 |
||
I generated VCD file and I am able to see waveforms too. There I found "if_pc" signal is suddenly changed from 0x130 to 0x1000. I am not sure is this the correct signal I am looking at. It seems it is an exception, even if I know it is an exception then how to resolve it? How to go to exact root of the problem?
|
RE: or1200 test case
by stekern on Mar 20, 2013 |
stekern
Posts: 84 Joined: Apr 28, 2009 Last seen: Nov 10, 2016 |
||
Your problem has nothing to do with that, it's simply an oversight in the
or1200-dsxinsn test when caches are disabled, this http://pastie.org/6638818 patch fixes it. |
RE: or1200 test case
by sudhanshubillore on Mar 28, 2013 |
sudhanshubillore
Posts: 16 Joined: Aug 10, 2012 Last seen: Apr 18, 2023 |
||
Hi,
Here is the combination which I use & corresponding results. I hope with table problem is very much clear. First of all I am not able to understand whats the need of that patch? What its purpose? Please help me out.. Thank you. --------------------------------------------------------------------------------------- combination dsxinsn testcase ticksyscall testcase --------------------------------------------------------------------------------------- 1.DC & IC(disable) test is ok infinite loop (with patch) 2.DC & IC(disable) infinite loop infinite loop (without patch) 3.DC & IC(enable) ERROR 1 test is ok (with patch) (checking simulation results....) 4.DC & IC(enable) ERROR 1 test is ok (without patch) (checking simulation results....) |
RE: or1200 test case
by stekern on Mar 30, 2013 |
stekern
Posts: 84 Joined: Apr 28, 2009 Last seen: Nov 10, 2016 |
||
The purpose of the patch is to fix the test, which was broken.
The patch have now been committed to SVN too. I'm not sure I understand what you mean with ERROR and 1 test is ok in: --- 3.DC & IC(enable) ERROR 1 test is ok (with patch) (checking simulation results....) 4.DC & IC(enable) ERROR 1 test is ok (without patch) (checking simulation results....) --- is that with your cache changes applied or without? if so, they're obviously not working correctly (yet) ;) If not, I don't know, the test passes here with caches enabled before and after that patch. |



