CheckerCPU: Make some basic regression tests for CheckerCPU
Review Request #1033 - Created Feb. 7, 2012 and submitted
| Information | |
|---|---|
| Geoffrey Blake | |
| gem5 | |
| default | |
| Reviewers | |
| Default | |
CheckerCPU: Make some basic regression tests for CheckerCPU Dual purposes the se/00.hello and fs/10.linux-boot regression tests as tests for the CheckerCPU for the ARM ISA only.
Compiles. The se/00.hello an fs/10.linux-boot tests run to completion sucessfully.
Issue Summary
3
3
0
0
| Description | From | Last Updated | Status |
|---|---|---|---|
| I'm not sure you should do things this way. I think it would be cleaner to create a new configuration ... | Gabe Black | Feb. 8, 2012, 6:08 p.m. | Open |
| My comment above applies here as well. | Gabe Black | Feb. 8, 2012, 6:08 p.m. | Open |
| The checker for the simple CPU doesn't seem to make a lot of sense. In that case it should just ... | Gabe Black | Feb. 8, 2012, 6:08 p.m. | Open |
Posted (Feb. 8, 2012, 6:08 p.m.)
-
tests/long/fs/10.linux-boot/test.py (Diff revision 1) -
I'm not sure you should do things this way. I think it would be cleaner to create a new configuration like realview-o3-checker.py which is the same but turns on the checker.
-
tests/quick/se/00.hello/test.py (Diff revision 1) -
My comment above applies here as well.
-
tests/quick/se/00.hello/test.py (Diff revision 1) -
The checker for the simple CPU doesn't seem to make a lot of sense. In that case it should just have a NULL checker.
Review request changed
Updated (Feb. 10, 2012, 7:26 a.m.)
Posted (Feb. 10, 2012, 7:31 a.m.)
Implemented request to make checker cpu regressions their own tests. I included the config.ini and stats.txt files, but have left the creation of the simout, simerr and system.terminal outputs for someone with access to Zizzer so they do not fail the diff when the regression is performed.
Posted (Feb. 11, 2012, 4:47 a.m.)
Without picking through it all in too much detail it looks ok to me. You should feel free to add all the reference files. The regressions aren't just diffed, generally they'll only fail if the stats change. Even then, ones that reflect host performance (number of host seconds, host instruction rate, amount of host memory consumed) are ignored.
Review request changed
Updated (Feb. 13, 2012, 3 a.m.)
Diff: |
Revision 3 (+3728 -1)
|
|---|
Posted (Feb. 13, 2012, 3:01 a.m.)
Added remaining files (simout, simerr, system.terminal) to regressions for the checker cpu on ARM only.
Review request changed
Updated (Feb. 17, 2012, 2:08 a.m.)
Diff: |
Revision 4 (+3685 -1)
|
|---|
Posted (Feb. 17, 2012, 2:09 a.m.)
Re-based to account for Andreas's recent changes to ports.
Review request changed
Updated (Feb. 27, 2012, 7:52 a.m.)
Diff: |
Revision 5 (+3685 -1)
|
|---|
Posted (Feb. 27, 2012, 7:54 a.m.)
Rebasing to keep up with recent changes.
Ship It!
It would be much nicer if we could factor out the common parts in all these regressions.
Review request changed
Updated (Feb. 29, 2012, 12:57 a.m.)
Diff: |
Revision 6 (+3712 -1)
|
|---|
Posted (Feb. 29, 2012, 1:01 a.m.)
Updated Checker regression test config file copyright headers appropriately.
Ship It!