fixed drainCount
Review Request #1092 - Created March 11, 2012 and updated
| Information | |
|---|---|
| Andrew Lukefahr | |
| gem5 | |
| default | |
| Reviewers | |
| Default | |
Changeset 8896:6f0169698281 --------------------------- fixed drainCount
Allows multiple switchouts of O3CPU using the attached config file. Both CPU's maintain separate (overlapping) seqNums, see ./src/cpu/o3/cpu.cc:1239 for further detail...
Files
Issue Summary
6
2
4
0
| Description | From | Last Updated | Status |
|---|---|---|---|
| == true is unnecessary and makes code less readable. | Tony Gutierrez | May 16, 2012, 3:12 a.m. | Open |
| I know NULL is used extensively in gem5 but, should we prefer 0? | Tony Gutierrez | May 16, 2012, 3:12 a.m. | Open |
Review request changed
Updated (March 11, 2012, 8:55 a.m.)
Description: |
|
||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Testing Done: |
|
||||||||||||
Diff: |
Revision 2 (+29 -4) |
Ship It!
Posted (March 14, 2012, 3:55 a.m.)
Hi Andrew, This is great! How extensively have you managed to test this change? does it still work with standard switching? Does it work switching between timing and o3? I've made a few small comments for the patch that i'd like to see changed. Also, it would be really great if we could get this functionality into the normal config scripts. I would really like to see a regression test that uses this functionality because without it, it will probably get broken accidentally. Thanks again, Ali
-
src/cpu/base.cc (Diff revision 2) -
extraneous blank line
-
src/cpu/o3/commit_impl.hh (Diff revision 2) -
just remove old code please
-
src/cpu/o3/cpu.cc (Diff revision 2) -
what is the right way then?
-
src/cpu/o3/lsq_unit.hh (Diff revision 2) -
space between = please
Review request changed
Updated (March 14, 2012, 4:20 a.m.)
Summary: |
|
|||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Description: |
|
|||||||||||||||
Diff: |
Revision 3 (+25 -4) |
Review request changed
Updated (March 14, 2012, 4:20 a.m.)
Summary: |
|
||||
|---|---|---|---|---|---|
Diff: |
Revision 4 (+25 -4) |
Ship It!
Review request changed
Updated (March 14, 2012, 4:28 a.m.)
Description: |
|
||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Diff: |
Revision 5 (+24 -4) |
Posted (March 14, 2012, 4:30 a.m.)
sorry, accidentally renamed the patch, and found a small bug
Posted (March 14, 2012, 4:43 a.m.)
These changes are reasonable, however, I made similar changes when trying to address this issue months ago. It would work for various switching frequencies but would fail for others. Please ensure that this works for the regressions for various switching frequencies. Also, make sure it works while switching back and forth between atomic <-> o3 at various switching frequencies. I know timing will not work with switching since it has its own issues with draining/switching out.
Posted (March 15, 2012, 12:57 a.m.)
Lets treat this patch as a work-in-progress until I can get around to some regression testing. :)
Review request changed
Updated (March 15, 2012, 7:09 a.m.)
Summary: |
|
|||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Description: |
|
|||||||||||||||
Diff: |
Revision 6 (+72 -12) |
Review request changed
Updated (March 15, 2012, 7:11 a.m.)
Summary: |
|
|||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Description: |
|
|||||||||||||||
Diff: |
Revision 7 (+73 -13) |
Review request changed
Updated (March 15, 2012, 7:16 a.m.)
Description: |
|
||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Diff: |
Revision 8 (+72 -13) |
Review request changed
Updated (March 15, 2012, 11:04 p.m.)
Summary: |
|
|---|
Review request changed
Updated (March 16, 2012, 12:38 a.m.)
Summary: |
|
||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Description: |
|
||||||||||||
Diff: |
Revision 9 (+99 -20) |
Review request changed
Updated (March 16, 2012, 12:38 a.m.)
Summary: |
|
|---|
Review request changed
Updated (March 16, 2012, 12:40 a.m.)
Summary: |
|
||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Description: |
|
||||||||||||
Diff: |
Revision 10 (+98 -19) |
Posted (May 2, 2012, 5:53 a.m.)
anyone have any issues with this or should I commit it?
Posted (May 16, 2012, 3:12 a.m.)
I've recently tried using this patch to get switchout functionality working, as others on the mailing list have pointed out there are still many assertions failures as well as failures related to the specific ISA being used.
-
src/cpu/o3/commit_impl.hh (Diff revision 10) -
== true is unnecessary and makes code less readable.
-
src/cpu/o3/fetch_impl.hh (Diff revision 10) -
I know NULL is used extensively in gem5 but, should we prefer 0?