cpu: Fix incorrect speculative branch predictor behavior
Review Request #2298 - Created June 12, 2014 and submitted
| Information | |
|---|---|
| Ali Saidi | |
| gem5 | |
| default | |
| Reviewers | |
| Default | |
Changeset 10253:992be57d8ca3 --------------------------- cpu: Fix incorrect speculative branch predictor behavior When a branch mispredicted gem5 would squash all history after and including the mispredicted branch. However, the mispredicted branch is still speculative and its history is required to rollback state if another, older, branch mispredicts. This leads to things like RAS corruption.
Posted (June 21, 2014, 10:10 a.m.)
-
src/cpu/pred/2bit_local.hh (Diff revision 1) -
should have a blank line here IMO. Also this comment is redundnant wrt comment on original (overridden) declaration on BPredUnit. Do we have a policy on this? Nice to have comments in both places, but the redundancy could lead to divergence if one is updated and the other not. Already have redundant typos...
-
src/cpu/pred/2bit_local.hh (Diff revision 1) -
and only one blank line here
-
src/cpu/pred/bpred_unit.hh (Diff revision 1) -
blank line here. also typo 'iwth'
-
src/cpu/pred/bpred_unit.hh (Diff revision 1) -
typo 'Whether'
-
src/cpu/pred/tournament.hh (Diff revision 1) -
don't delete blank line here
-
src/cpu/pred/tournament.hh (Diff revision 1) -
YA redundant comment w/typo
-
src/cpu/pred/tournament.hh (Diff revision 1) -
add blank line here
