This is an archive of the discontinued LLVM Phabricator instance.

LiveVariables should not clobber MachineOperand::IsDead, ::IsKill on reserved physical registers
ClosedPublic

Authored by npjdesres on Nov 20 2015, 8:45 AM.

Details

Summary

The problem

tl;dr- pass LiveVariables forgets to restore the MO::IsDead bit for reserved, physical registers.

The LiveVariables pass analyzes MachineInstrs to set the IsKill and IsDead bit on each MachineOperand. It first clears the IsKill and IsDead bits on every machine operand, then recomputes liveness, and finally sets the IsKill/IsDead bits appropriately.

Although the LiveVariables pass clears these bits for *all* MachineOperands, LiveVariables only computes liveness on virtual registers and non-reserved physical registers; it does *not* compute liveness on reserved physical registers, such as condition-code "flag" registers or the stack pointer. Consequently, *no* reserved, physical registers are ever marked as dead or killed. This is conservative, but it pessimizes subsequent passes.

Consequences of that problem

tl;dr- scheduling suffers.

In particular, LiveVariables preceeds the machine scheduler. The machine scheduler builds a dependence graph, including edges representing data, anti, and output dependencies among instructions which use physical registers (see ScheduleDAGInstrs::addPhysRegDeps). In order to avoid an overly-constrained scheduling graph, that routine attempts to skip output dependencies if the physical register is dead.

However, since LiveVariables has cleared the MachineOperand::IsDead bit for reserved, physical registers, the routine ScheduleDAGInstrs::addPhysRegDeps adds these output dependences even though they are not necessary.

In concrete terms, if there is a sequence of instructions where each defines a flag register (e.g., some ADD instructions set the OVERFLOW flag), then MISched considers them locked in order and is unwilling to schedule them in any other order. This sometimes forces the scheduler to emit very-far-from-optimal schedules.

Solution

Modify the LiveVariables pass so that it does not clear the MachineOperand::IsDead or MachineOperand::IsKill bits when the MachineOperand represents a def/use of a physical, reserved register. This criterion directly corresponds to the condition later in the same routine that guard HandlePhysRegUse and handlePhysRegDef.

I tested this patch on today's latest (git 72bf3d33aa1b7e50bca13eecf576011fcc327d51) and it passed regressions:

[100%] Running the LLVM regression tests
Testing Time: 388.74s
  Expected Passes    : 14991
  Expected Failures  : 132
  Unsupported Tests  : 111
[100%] Built target check-llvm
Scanning dependencies of target check
[100%] Built target check

In my out-of-tree target, I see significant improvement to the scheduling DAG (tall and narrow becomes short and broad).

Other references

Diff Detail

Repository
rL LLVM

Event Timeline

npjdesres updated this revision to Diff 40781.Nov 20 2015, 8:45 AM
npjdesres retitled this revision from to LiveVariables should not clobber MachineOperand::IsDead, ::IsKill on reserved physical registers.
npjdesres updated this object.
npjdesres added a subscriber: llvm-commits.
asb added a subscriber: asb.Nov 21 2015, 8:35 AM
MatzeB accepted this revision.Nov 23 2015, 10:17 AM
MatzeB added a reviewer: MatzeB.
MatzeB added a subscriber: MatzeB.

LGTM, but before committing please check the changes with clang-format (which is a nice way of saying: The lines are longer than 80 chars and the spaces before the closing braces seem strange).

This revision is now accepted and ready to land.Nov 23 2015, 10:17 AM
npjdesres updated this revision to Diff 41039.Nov 24 2015, 6:45 AM
npjdesres edited edge metadata.

Thanks for looking it over. Andy Trick agrees on the llvm-dev list.

I've uploaded a new patch after passing it though clang-format.

I do not have commit access.

This revision was automatically updated to reflect the committed changes.