In the pre-RA scheduler, maintain cmp+br proximity.
This is done by pushing physical register definitions close to their
use, which happens to handle flag definitions if they're not glued to
the branch. This seems to be generally a good thing though, so I
didn't need to add a target hook yet.
The primary motivation is to generate code closer to what people
expect and rule out missed opportunity from enabling macro-op
fusion. As a side benefit, we get several 2-5% gains on x86
benchmarks. There is one regression:
SingleSource/Benchmarks/Shootout/lists slows down be -10%. But this is
an independent scheduler bug that will be tracked separately.
See rdar://problem/9283108.
Incidentally, pre-RA scheduling is only half the solution. Fixing the
later passes is tracked by:
<rdar://problem/8932804> [pre-RA-sched] on x86, attempt to schedule CMP/TEST adjacent with condition jump
Fixes:
<rdar://problem/9262453> Scheduler unnecessary break of cmp/jump fusion
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@129508 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp b/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp
index 078533b..b258e6e 100644
--- a/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp
+++ b/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp
@@ -87,6 +87,8 @@
SU->isCommutable = Old->isCommutable;
SU->hasPhysRegDefs = Old->hasPhysRegDefs;
SU->hasPhysRegClobbers = Old->hasPhysRegClobbers;
+ SU->isScheduleHigh = Old->isScheduleHigh;
+ SU->isScheduleLow = Old->isScheduleLow;
SU->SchedulingPref = Old->SchedulingPref;
Old->isCloned = true;
return SU;
@@ -335,6 +337,12 @@
if (!HasGlueUse) break;
}
+ // Schedule zero-latency TokenFactor below any nodes that may increase the
+ // schedule height. Otherwise, ancestors of the TokenFactor may appear to
+ // have false stalls.
+ if (NI->getOpcode() == ISD::TokenFactor)
+ NodeSUnit->isScheduleLow = true;
+
// If there are glue operands involved, N is now the bottom-most node
// of the sequence of nodes that are glued together.
// Update the SUnit.