llvm-6502/lib/Target/Hexagon
Quentin Colombet 2f7322b348 [ShrinkWrap] Add (a simplified version) of shrink-wrapping.
This patch introduces a new pass that computes the safe point to insert the
prologue and epilogue of the function.
The interest is to find safe points that are cheaper than the entry and exits
blocks.

As an example and to avoid regressions to be introduce, this patch also
implements the required bits to enable the shrink-wrapping pass for AArch64.


** Context **

Currently we insert the prologue and epilogue of the method/function in the
entry and exits blocks. Although this is correct, we can do a better job when
those are not immediately required and insert them at less frequently executed
places.
The job of the shrink-wrapping pass is to identify such places.


** Motivating example **

Let us consider the following function that perform a call only in one branch of
a if:
define i32 @f(i32 %a, i32 %b)  {
 %tmp = alloca i32, align 4
 %tmp2 = icmp slt i32 %a, %b
 br i1 %tmp2, label %true, label %false

true:
 store i32 %a, i32* %tmp, align 4
 %tmp4 = call i32 @doSomething(i32 0, i32* %tmp)
 br label %false

false:
 %tmp.0 = phi i32 [ %tmp4, %true ], [ %a, %0 ]
 ret i32 %tmp.0
}

On AArch64 this code generates (removing the cfi directives to ease
readabilities):
_f:                                     ; @f
; BB#0:
  stp x29, x30, [sp, #-16]!
  mov  x29, sp
  sub sp, sp, #16             ; =16
  cmp  w0, w1
  b.ge  LBB0_2
; BB#1:                                 ; %true
  stur  w0, [x29, #-4]
  sub x1, x29, #4             ; =4
  mov  w0, wzr
  bl  _doSomething
LBB0_2:                                 ; %false
  mov  sp, x29
  ldp x29, x30, [sp], #16
  ret

With shrink-wrapping we could generate:
_f:                                     ; @f
; BB#0:
  cmp  w0, w1
  b.ge  LBB0_2
; BB#1:                                 ; %true
  stp x29, x30, [sp, #-16]!
  mov  x29, sp
  sub sp, sp, #16             ; =16
  stur  w0, [x29, #-4]
  sub x1, x29, #4             ; =4
  mov  w0, wzr
  bl  _doSomething
  add sp, x29, #16            ; =16
  ldp x29, x30, [sp], #16
LBB0_2:                                 ; %false
  ret

Therefore, we would pay the overhead of setting up/destroying the frame only if
we actually do the call.


** Proposed Solution **

This patch introduces a new machine pass that perform the shrink-wrapping
analysis (See the comments at the beginning of ShrinkWrap.cpp for more details).
It then stores the safe save and restore point into the MachineFrameInfo
attached to the MachineFunction.
This information is then used by the PrologEpilogInserter (PEI) to place the
related code at the right place. This pass runs right before the PEI.

Unlike the original paper of Chow from PLDI’88, this implementation of
shrink-wrapping does not use expensive data-flow analysis and does not need hack
to properly avoid frequently executed point. Instead, it relies on dominance and
loop properties.

The pass is off by default and each target can opt-in by setting the
EnableShrinkWrap boolean to true in their derived class of TargetPassConfig.
This setting can also be overwritten on the command line by using
-enable-shrink-wrap.

Before you try out the pass for your target, make sure you properly fix your
emitProlog/emitEpilog/adjustForXXX method to cope with basic blocks that are not
necessarily the entry block.


** Design Decisions **

1. ShrinkWrap is its own pass right now. It could frankly be merged into PEI but
for debugging and clarity I thought it was best to have its own file.
2. Right now, we only support one save point and one restore point. At some
point we can expand this to several save point and restore point, the impacted
component would then be:
- The pass itself: New algorithm needed.
- MachineFrameInfo: Hold a list or set of Save/Restore point instead of one
  pointer.
- PEI: Should loop over the save point and restore point.
Anyhow, at least for this first iteration, I do not believe this is interesting
to support the complex cases. We should revisit that when we motivating
examples.

Differential Revision: http://reviews.llvm.org/D9210

<rdar://problem/3201744>


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@236507 91177308-0d34-0410-b5e6-96231b3b80d8
2015-05-05 17:38:16 +00:00
..
Disassembler [Hexagon] Moving remaining methods off of HexagonMCInst in to HexagonMCInstrInfo and eliminating HexagonMCInst class. 2015-02-19 21:10:50 +00:00
MCTargetDesc Fix -Wmicrosoft warning by making enum unsigned 2015-05-04 18:21:35 +00:00
TargetInfo
CMakeLists.txt Expand MUX instructions early on Hexagon 2015-03-31 13:35:12 +00:00
Hexagon.h Hexagon: Remove pass that does nothing at all 2015-03-10 15:06:38 +00:00
Hexagon.td Make the Hexagon ISelDAGToDAG pass set the subtarget dynamically 2015-03-21 03:12:59 +00:00
HexagonAsmPrinter.cpp [AsmPrinter] Make AsmPrinter's OutStreamer member a unique_ptr. 2015-04-24 19:11:51 +00:00
HexagonAsmPrinter.h Define a runOnMachineFunction for the Hexagon AsmPrinter and 2015-02-03 06:40:22 +00:00
HexagonCallingConv.td
HexagonCFGOptimizer.cpp Grab TargetInstrInfo off of the MachineFunction and remove 2015-02-02 18:46:27 +00:00
HexagonCopyToCombine.cpp [Hexagon] Fix compiler warnings in release build 2015-04-23 20:26:21 +00:00
HexagonExpandCondsets.cpp [Hexagon] Avoid an unused variable warning when assertions are off 2015-03-31 19:43:47 +00:00
HexagonExpandPredSpillCode.cpp [Hexagon] Intrinsics for circular and bit-reversed loads and stores 2015-03-18 16:23:44 +00:00
HexagonFixupHwLoops.cpp [Hexagon] Use constant extenders to fix up hardware loops 2015-04-27 14:16:43 +00:00
HexagonFrameLowering.cpp [ShrinkWrap] Add (a simplified version) of shrink-wrapping. 2015-05-05 17:38:16 +00:00
HexagonFrameLowering.h [ShrinkWrap] Add (a simplified version) of shrink-wrapping. 2015-05-05 17:38:16 +00:00
HexagonHardwareLoops.cpp Remove the remaining uses of abs64 and nuke it. 2015-03-09 20:20:16 +00:00
HexagonInstrFormats.td [Hexagon] Separating InstHexagon from OpcodeHexagon. 2015-03-10 20:56:22 +00:00
HexagonInstrFormatsV4.td [Hexagon] Separating InstHexagon from OpcodeHexagon. 2015-03-10 20:56:22 +00:00
HexagonInstrInfo.cpp [Hexagon] Use constant extenders to fix up hardware loops 2015-04-27 14:16:43 +00:00
HexagonInstrInfo.h [Hexagon] Use constant extenders to fix up hardware loops 2015-04-27 14:16:43 +00:00
HexagonInstrInfo.td Reapply r235977 "[DebugInfo] Add debug locations to constant SD nodes" 2015-04-28 14:05:47 +00:00
HexagonInstrInfoV3.td [Hexagon] Removing v2-4 flags. V4 is the minimum supported version. 2015-02-09 21:07:35 +00:00
HexagonInstrInfoV4.td Reapply r235977 "[DebugInfo] Add debug locations to constant SD nodes" 2015-04-28 14:05:47 +00:00
HexagonInstrInfoV5.td Eliminate constant-extender profitability checks from Hexagon isel 2015-03-12 00:19:59 +00:00
HexagonInstrInfoVector.td [Hexagon] Add support for vector instructions 2015-03-19 16:33:08 +00:00
HexagonIntrinsics.td [Hexagon] Intrinsics for circular and bit-reversed loads and stores 2015-03-18 16:23:44 +00:00
HexagonIntrinsicsDerived.td [Hexagon] Deleting a lot of old variants of intrinsics and updating references. 2015-01-28 18:29:11 +00:00
HexagonIntrinsicsV3.td [Hexagon] Converting XTYPE/SHIFT intrinsics. Cleaning out old intrinsic patterns and updating tests. 2015-02-03 20:40:52 +00:00
HexagonIntrinsicsV4.td Eliminate constant-extender profitability checks from Hexagon isel 2015-03-12 00:19:59 +00:00
HexagonIntrinsicsV5.td [Hexagon] Converting XTYPE/SHIFT intrinsics. Cleaning out old intrinsic patterns and updating tests. 2015-02-03 20:40:52 +00:00
HexagonISelDAGToDAG.cpp Reapply r235977 "[DebugInfo] Add debug locations to constant SD nodes" 2015-04-28 14:05:47 +00:00
HexagonISelLowering.cpp Reapply r235977 "[DebugInfo] Add debug locations to constant SD nodes" 2015-04-28 14:05:47 +00:00
HexagonISelLowering.h [hexagon] Remove setHexLibcallName, it leaks memory. 2015-04-25 14:46:46 +00:00
HexagonMachineFunctionInfo.cpp
HexagonMachineFunctionInfo.h [Hexagon] Overhaul of stack object allocation 2015-04-22 16:43:53 +00:00
HexagonMachineScheduler.cpp Move HexagonMachineScheduler to use the subtarget off of the 2015-02-02 22:11:40 +00:00
HexagonMachineScheduler.h Move HexagonMachineScheduler to use the subtarget off of the 2015-02-02 22:11:40 +00:00
HexagonMCInstLower.cpp [Hexagon] Moving remaining methods off of HexagonMCInst in to HexagonMCInstrInfo and eliminating HexagonMCInst class. 2015-02-19 21:10:50 +00:00
HexagonNewValueJump.cpp Re-sort includes with sort-includes.py and insert raw_ostream.h where it's used. 2015-03-23 19:32:43 +00:00
HexagonOperands.td [Hexagon] Use constant extenders to fix up hardware loops 2015-04-27 14:16:43 +00:00
HexagonPeephole.cpp [Hexagon] Eliminating immediate condition set. 2015-03-09 19:57:18 +00:00
HexagonRegisterInfo.cpp [Hexagon] Overhaul of stack object allocation 2015-04-22 16:43:53 +00:00
HexagonRegisterInfo.h [Hexagon] Shrink-wrap stack frame (Hexagon-specific) 2015-04-23 16:05:39 +00:00
HexagonRegisterInfo.td [Hexagon] Adding vector shift instructions and tests. 2015-01-30 21:58:46 +00:00
HexagonRemoveSZExtArgs.cpp [LPM] Stop using the string based preservation API. It is an 2015-01-28 04:57:56 +00:00
HexagonSchedule.td
HexagonScheduleV4.td
HexagonSelectCCInfo.td
HexagonSelectionDAGInfo.cpp
HexagonSelectionDAGInfo.h
HexagonSplitConst32AndConst64.cpp [Hexagon] Use A2_tfrsi for constant pool and jump table addresses 2015-04-22 18:25:53 +00:00
HexagonSubtarget.cpp Remove more superfluous .str() and replace std::string concatenation with Twine. 2015-03-30 15:42:36 +00:00
HexagonSubtarget.h Remove useMachineScheduler and replace it with subtarget options 2015-03-11 22:56:10 +00:00
HexagonTargetMachine.cpp Expand MUX instructions early on Hexagon 2015-03-31 13:35:12 +00:00
HexagonTargetMachine.h Revert r233206 2015-03-25 20:21:16 +00:00
HexagonTargetObjectFile.cpp Compute the ELF SectionKind from the flags. 2015-01-29 17:33:21 +00:00
HexagonTargetObjectFile.h
HexagonVLIWPacketizer.cpp [Hexagon] Treat CFI as solo instructions 2015-04-22 15:47:35 +00:00
LLVMBuild.txt
Makefile