regression tester: Fix some bugs in simple-timing-mp-ruby.py.

Information
Submitter:	Marc Orr
Repository:	gem5
Branch:	default
Bugs:
Depends On:
Reviewers
Groups:	Default
People:

Description

The simple-timing-mp-ruby.py configuration file did not get updated to work with some recent updates. It is used in the m5threads regression. Currently the m5threads regression actually skips this test. I left the skip file in there because m5threads does not seem to work with ruby for the sparc ISA with > 2 CPUs. That said, it does work with x86, which I plan to add a regression for.

Changeset 8985:94ac17de30e1
---------------------------
regression tester: Fix some bugs in simple-timing-mp-ruby.py.

Testing Done

Description:

	+	The simple-timing-mp-ruby.py configuration file did not get updated to work with some recent updates. It is used in the m5threads regression. Currently the m5threads regression actually skips this test. I left the skip file in there because m5threads does not seem to work with ruby for the sparc ISA with > 2 CPUs. That said, it does work with x86, which I plan to add a regression for.
	+
		Changeset 8985:94ac17de30e1

		regression tester: Fix some bugs in simple-timing-mp-ruby.py.

Ship It!

Have you tested this? It seems this is not correct. Try to mimic ruby_fs.py.

Marc Orr May 18, 2012, 3:10 a.m. (May 18, 2012, 3:10 a.m.)

Hrmmm... It seemed to work ok. Let me be specific about the test I did:
1. SPARC atomics 2 CPUs: pass
2. SPARC threads 2 CPUs: pass
3. SPARC atomics 4 CPUs: HANGS
4. SPARC threads 4 CPUs: HANGS
3. X86 atomics 4 CPUs: pass
4. X86 threads 4 CPUS: pass

I sort of tried to follow o3-timing-mp-ruby.py test configuration, except that I let the ruby_create_system function call hook up the memory and system ports.

Initially I did try to use the ruby portion of se.py as a guide (it looks similar to ruby_fs.py I think), but the sparc interrupt controller doesn't have master/slave ports, so bad things happened.

Can you be more specific about your concerns with this implementation?

Marc Orr May 18, 2012, 3:33 a.m. (May 18, 2012, 3:33 a.m.)

Some how I posted the wrong comment. Specifically, the testing I intended to post is:
1. SPARC simple_atomics: HANGS
2. SPARC simple___threads: pass
3. X86 simple_atomics: pass
4. X86 simple___threads: pass

Nilay Vaish May 18, 2012, 8 a.m. (May 18, 2012, 8 a.m.)

Now I see why connectAllPorts() is working correctly with Ruby as well.
I suggest changing se.py and ruby_fs.py to use connectAllPorts() instead
of individually connecting the ports.

Marc Orr May 20, 2012, 10:49 a.m. (May 20, 2012, 10:49 a.m.)
```
Your suggestion is orthogonal to this patch - right?
```
Andreas Hansson June 10, 2012, 5:28 p.m. (June 10, 2012, 5:28 p.m.)
```
The patch is pushed. Could you close this Marc?
```

You have a pending review.

Review Board 2.0.15

This change has been marked as submitted.

Screenshots

Files

Description:

Changeset 8985:94ac17de30e1

Status: Closed (submitted)