Commit Graph

174 Commits

Author SHA1 Message Date
Dehao Chen ea62ae9844 Using branch probability to guide critical edge splitting.
Summary:
The original heuristic to break critical edge during machine sink is relatively conservertive: when there is only one instruction sinkable to the critical edge, it is likely that the machine sink pass will not break the critical edge. This leads to many speculative instructions executed at runtime. However, with profile info, we could model the splitting benefits: if the critical edge has 50% taken rate, it would always be beneficial to split the critical edge to avoid the speculated runtime instructions. This patch uses profile to guide critical edge splitting in machine sink pass.

The performance impact on speccpu2006 on Intel sandybridge machines:

spec/2006/fp/C++/444.namd                  25.3  +0.26%
spec/2006/fp/C++/447.dealII               45.96  -0.10%
spec/2006/fp/C++/450.soplex               41.97  +1.49%
spec/2006/fp/C++/453.povray               36.83  -0.96%
spec/2006/fp/C/433.milc                   23.81  +0.32%
spec/2006/fp/C/470.lbm                    41.17  +0.34%
spec/2006/fp/C/482.sphinx3                48.13  +0.69%
spec/2006/int/C++/471.omnetpp             22.45  +3.25%
spec/2006/int/C++/473.astar               21.35  -2.06%
spec/2006/int/C++/483.xalancbmk           36.02  -2.39%
spec/2006/int/C/400.perlbench              33.7  -0.17%
spec/2006/int/C/401.bzip2                  22.9  +0.52%
spec/2006/int/C/403.gcc                   32.42  -0.54%
spec/2006/int/C/429.mcf                   39.59  +0.19%
spec/2006/int/C/445.gobmk                 26.98  -0.00%
spec/2006/int/C/456.hmmer                 24.52  -0.18%
spec/2006/int/C/458.sjeng                 28.26  +0.02%
spec/2006/int/C/462.libquantum            55.44  +3.74%
spec/2006/int/C/464.h264ref               46.67  -0.39%

geometric mean                                   +0.20%

Manually checked 473 and 471 to verify the diff is in the noise range.

Reviewers: rengolin, davidxl

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D24818

llvm-svn: 284541
2016-10-18 21:36:11 +00:00
Eugene Zelenko 1804a77b2a Fix some Clang-tidy modernize-use-using and Include What You Use warnings; other minor fixes.
Differential revision: https://reviews.llvm.org/D23861

llvm-svn: 279695
2016-08-25 00:45:04 +00:00
Jacques Pienaar 71c30a14b7 Rename AnalyzeBranch* to analyzeBranch*.
Summary: NFC. Rename AnalyzeBranch/AnalyzeBranchPredicate to analyzeBranch/analyzeBranchPredicate to follow LLVM coding style and be consistent with TargetInstrInfo's analyzeCompare and analyzeSelect.

Reviewers: tstellarAMD, mcrosier

Subscribers: mcrosier, jholewinski, jfb, arsenm, dschuff, jyknight, dsanders, nemanjai

Differential Revision: https://reviews.llvm.org/D22409

llvm-svn: 275564
2016-07-15 14:41:04 +00:00
Duncan P. N. Exon Smith cb38ffa74d CodeGen: Use MachineInstr& in MachineSink, NFC
Use MachineInstr& instead of MachineInstr* in MachineSinker to help
avoid implicit conversions from iterator to pointer.

llvm-svn: 274303
2016-07-01 00:11:48 +00:00
Duncan P. N. Exon Smith 9cfc75c214 CodeGen: Use MachineInstr& in TargetInstrInfo, NFC
This is mostly a mechanical change to make TargetInstrInfo API take
MachineInstr& (instead of MachineInstr* or MachineBasicBlock::iterator)
when the argument is expected to be a valid MachineInstr.  This is a
general API improvement.

Although it would be possible to do this one function at a time, that
would demand a quadratic amount of churn since many of these functions
call each other.  Instead I've done everything as a block and just
updated what was necessary.

This is mostly mechanical fixes: adding and removing `*` and `&`
operators.  The only non-mechanical change is to split
ARMBaseInstrInfo::getOperandLatencyImpl out from
ARMBaseInstrInfo::getOperandLatency.  Previously, the latter took a
`MachineInstr*` which it updated to the instruction bundle leader; now,
the latter calls the former either with the same `MachineInstr&` or the
bundle leader.

As a side effect, this removes a bunch of MachineInstr* to
MachineBasicBlock::iterator implicit conversions, a necessary step
toward fixing PR26753.

Note: I updated WebAssembly, Lanai, and AVR (despite being
off-by-default) since it turned out to be easy.  I couldn't run tests
for AVR since llc doesn't link with it turned on.

llvm-svn: 274189
2016-06-30 00:01:54 +00:00
Andrew Kaylor aa641a5171 Re-commit optimization bisect support (r267022) without new pass manager support.
The original commit was reverted because of a buildbot problem with LazyCallGraph::SCC handling (not related to the OptBisect handling).

Differential Revision: http://reviews.llvm.org/D19172

llvm-svn: 267231
2016-04-22 22:06:11 +00:00
Vedant Kumar 6013f45f92 Revert "Initial implementation of optimization bisect support."
This reverts commit r267022, due to an ASan failure:

  http://lab.llvm.org:8080/green/job/clang-stage2-cmake-RgSan_check/1549

llvm-svn: 267115
2016-04-22 06:51:37 +00:00
Quentin Colombet 23341a84ca [MachineBasicBlock] Make the pass argument truly mandatory when
splitting edges.

MachineBasicBlock::SplitCriticalEdges will crash if a nullptr would have
been passed for the Pass argument. Do not allow that by turning this
argument into a reference.
The alternative would have been to make the Pass a truly optional
argument, but although this is easy to do, I was afraid users using it
like this would not be aware the livness information, dominator tree and
such would silently be broken.

llvm-svn: 267052
2016-04-21 21:01:13 +00:00
Andrew Kaylor f0f279291c Initial implementation of optimization bisect support.
This patch implements a optimization bisect feature, which will allow optimizations to be selectively disabled at compile time in order to track down test failures that are caused by incorrect optimizations.

The bisection is enabled using a new command line option (-opt-bisect-limit).  Individual passes that may be skipped call the OptBisect object (via an LLVMContext) to see if they should be skipped based on the bisect limit.  A finer level of control (disabling individual transformations) can be managed through an addition OptBisect method, but this is not yet used.

The skip checking in this implementation is based on (and replaces) the skipOptnoneFunction check.  Where that check was being called, a new call has been inserted in its place which checks the bisect limit and the optnone attribute.  A new function call has been added for module and SCC passes that behaves in a similar way.

Differential Revision: http://reviews.llvm.org/D19172

llvm-svn: 267022
2016-04-21 17:58:54 +00:00
Fiona Glaser 44a2f7a298 MachineSink: make shouldSink a TII target hook
Some targets may disagree on what they want sunk or not sunk,
so make this a target hook instead of hardcoded.

llvm-svn: 264799
2016-03-29 22:44:57 +00:00
Chad Rosier c27a18f39f [TII] Allow getMemOpBaseRegImmOfs() to accept negative offsets. NFC.
http://reviews.llvm.org/D17967

llvm-svn: 263021
2016-03-09 16:00:35 +00:00
Richard Trieu 7a08381403 Remove uses of builtin comma operator.
Cleanup for upcoming Clang warning -Wcomma.  No functionality change intended.

llvm-svn: 261270
2016-02-18 22:09:30 +00:00
Sanjoy Das 16901a3e20 [MachineSink] Don't break ImplicitNulls
Summary:
This teaches MachineSink to not sink instructions that might break the
implicit null check optimization that runs later.  This should not
affect frontends that do not use implicit null checks.

Reviewers: aadg, reames, hfinkel, atrick

Subscribers: majnemer, llvm-commits

Differential Revision: http://reviews.llvm.org/D14632

llvm-svn: 258254
2016-01-20 00:06:14 +00:00
Owen Anderson d95b08a0a7 Refine the definition of convergent to only disallow the addition of new control dependencies.
This covers the common case of operations that cannot be sunk.
Operations that cannot be hoisted should already be handled properly via
the safe-to-speculate rules and mechanisms.

llvm-svn: 249865
2015-10-09 18:06:13 +00:00
Chandler Carruth 7b560d40bd [PM/AA] Rebuild LLVM's alias analysis infrastructure in a way compatible
with the new pass manager, and no longer relying on analysis groups.

This builds essentially a ground-up new AA infrastructure stack for
LLVM. The core ideas are the same that are used throughout the new pass
manager: type erased polymorphism and direct composition. The design is
as follows:

- FunctionAAResults is a type-erasing alias analysis results aggregation
  interface to walk a single query across a range of results from
  different alias analyses. Currently this is function-specific as we
  always assume that aliasing queries are *within* a function.

- AAResultBase is a CRTP utility providing stub implementations of
  various parts of the alias analysis result concept, notably in several
  cases in terms of other more general parts of the interface. This can
  be used to implement only a narrow part of the interface rather than
  the entire interface. This isn't really ideal, this logic should be
  hoisted into FunctionAAResults as currently it will cause
  a significant amount of redundant work, but it faithfully models the
  behavior of the prior infrastructure.

- All the alias analysis passes are ported to be wrapper passes for the
  legacy PM and new-style analysis passes for the new PM with a shared
  result object. In some cases (most notably CFL), this is an extremely
  naive approach that we should revisit when we can specialize for the
  new pass manager.

- BasicAA has been restructured to reflect that it is much more
  fundamentally a function analysis because it uses dominator trees and
  loop info that need to be constructed for each function.

All of the references to getting alias analysis results have been
updated to use the new aggregation interface. All the preservation and
other pass management code has been updated accordingly.

The way the FunctionAAResultsWrapperPass works is to detect the
available alias analyses when run, and add them to the results object.
This means that we should be able to continue to respect when various
passes are added to the pipeline, for example adding CFL or adding TBAA
passes should just cause their results to be available and to get folded
into this. The exception to this rule is BasicAA which really needs to
be a function pass due to using dominator trees and loop info. As
a consequence, the FunctionAAResultsWrapperPass directly depends on
BasicAA and always includes it in the aggregation.

This has significant implications for preserving analyses. Generally,
most passes shouldn't bother preserving FunctionAAResultsWrapperPass
because rebuilding the results just updates the set of known AA passes.
The exception to this rule are LoopPass instances which need to preserve
all the function analyses that the loop pass manager will end up
needing. This means preserving both BasicAAWrapperPass and the
aggregating FunctionAAResultsWrapperPass.

Now, when preserving an alias analysis, you do so by directly preserving
that analysis. This is only necessary for non-immutable-pass-provided
alias analyses though, and there are only three of interest: BasicAA,
GlobalsAA (formerly GlobalsModRef), and SCEVAA. Usually BasicAA is
preserved when needed because it (like DominatorTree and LoopInfo) is
marked as a CFG-only pass. I've expanded GlobalsAA into the preserved
set everywhere we previously were preserving all of AliasAnalysis, and
I've added SCEVAA in the intersection of that with where we preserve
SCEV itself.

One significant challenge to all of this is that the CGSCC passes were
actually using the alias analysis implementations by taking advantage of
a pretty amazing set of loop holes in the old pass manager's analysis
management code which allowed analysis groups to slide through in many
cases. Moving away from analysis groups makes this problem much more
obvious. To fix it, I've leveraged the flexibility the design of the new
PM components provides to just directly construct the relevant alias
analyses for the relevant functions in the IPO passes that need them.
This is a bit hacky, but should go away with the new pass manager, and
is already in many ways cleaner than the prior state.

Another significant challenge is that various facilities of the old
alias analysis infrastructure just don't fit any more. The most
significant of these is the alias analysis 'counter' pass. That pass
relied on the ability to snoop on AA queries at different points in the
analysis group chain. Instead, I'm planning to build printing
functionality directly into the aggregation layer. I've not included
that in this patch merely to keep it smaller.

Note that all of this needs a nearly complete rewrite of the AA
documentation. I'm planning to do that, but I'd like to make sure the
new design settles, and to flesh out a bit more of what it looks like in
the new pass manager first.

Differential Revision: http://reviews.llvm.org/D12080

llvm-svn: 247167
2015-09-09 17:55:00 +00:00
Reid Kleckner 0e2882345d [WinEH] Add some support for code generating catchpad
We can now run 32-bit programs with empty catch bodies.  The next step
is to change PEI so that we get funclet prologues and epilogues.

llvm-svn: 246235
2015-08-27 23:27:47 +00:00
Arnaud A. de Grandmaison c8a694fd27 [MachineSink] Address post-commit review comments
The successors cache is now a local variable, making it more visible that it
is only valid for the MBB being processed.

llvm-svn: 239807
2015-06-16 08:57:21 +00:00
Arnaud A. de Grandmaison d8673edc2d [MachineSink] Improve runtime performance. NFC.
This patch fixes a compilation time issue, when MachineSink faces PHIs
with a huge number of operands. This can happen for example in goto table
based interpreters, where some basic blocks can have several of those PHIs,
each one with several hundreds operands. MachineSink was spending a
significant time re-building and re-sorting the list of successors of
the current MachineBasicBlock. The computing and sorting of the current
MachineBasicBlock successors is now cached.

llvm-svn: 239720
2015-06-15 09:09:06 +00:00
Owen Anderson 55313d21dc Disable MachineSink on convergent operations, similar to how IR Sink is
restricted.  No test because no in-tree target currently has convergent
MachineInstr's.

llvm-svn: 238763
2015-06-01 17:26:30 +00:00
Matthias Braun 07066cca20 MachineInstr: Remove unused parameter.
llvm-svn: 237726
2015-05-19 21:22:20 +00:00
Matthias Braun 352b89c460 MachineSink: Collect registers before clearing their killflags.
Currently whenever we sink any instruction, we do clearKillFlags for
every use of every use operand for that instruction, apparently there
are a lot of duplication, therefore compile time penalties.

This patch collect all the interested registers first, do clearKillFlags
for it all together at once at the end, so we only need to do
clearKillFlags once for one register, duplication is avoided.

Patch by Lawrence Hu!

Differential Revision: http://reviews.llvm.org/D9719

llvm-svn: 237510
2015-05-16 03:11:07 +00:00
Pete Cooper 85b1c48b20 Clear kill flags on all used registers when sinking instructions.
The test here was sinking the AND here to a lower BB:

	%vreg7<def> = ANDWri %vreg8, 0; GPR32common:%vreg7,%vreg8
	TBNZW %vreg8<kill>, 0, <BB#1>; GPR32common:%vreg8

which meant that vreg8 was read after it was killed.

This commit changes the code from clearing kill flags on the AND to clearing flags on all registers used by the AND.

llvm-svn: 236886
2015-05-08 17:54:32 +00:00
Pete Cooper ff5064a188 80 cols fix since i'm looking at this function anyway. NFC
llvm-svn: 236885
2015-05-08 17:54:29 +00:00
Patrik Hagglund d06de4b954 Use DomTree in MachineSink to sink over diamonds.
According to a previous FIXME comment we now not only look at MBB
successors, but also handle code sinking past them:

  x = computation
  if () {} else {}
  use x

The instruction could be sunk over the whole diamond for the
if/then/else (or loop, etc), allowing it to be sunk into other blocks
after that.

Modified test added in r204522, due to one spill less present.

Minor fixes in comments.

Patch provided by Jonas Paulsson. Reviewed by Hal Finkel.

llvm-svn: 223350
2014-12-04 10:36:42 +00:00
David Blaikie 70573dcd9f Update SetVector to rely on the underlying set's insert to return a pair<iterator, bool>
This is to be consistent with StringSet and ultimately with the standard
library's associative container insert function.

This lead to updating SmallSet::insert to return pair<iterator, bool>,
and then to update SmallPtrSet::insert to return pair<iterator, bool>,
and then to update all the existing users of those functions...

llvm-svn: 222334
2014-11-19 07:49:26 +00:00
Jingyue Wu 2954280f6a [MachineSink] Use the real post dominator tree
Summary:
Fixes a FIXME in MachineSinking. Instead of using the simple heuristics in
isPostDominatedBy, use the real MachinePostDominatorTree and MachineLoopInfo.
The old heuristics caused instructions to sink unnecessarily, and might create
register pressure.

This is the second try of the fix. The first one (D4814) caused a performance
regression due to failing to sink instructions out of loops (PR21115). This
patch fixes PR21115 by sinking an instruction from a deeper loop to a shallower
one regardless of whether the target block post-dominates the source.

Thanks Alexey Volkov for reporting PR21115! 

Test Plan:
Added a NVPTX codegen test to verify that our change prevents the backend from
over-sinking. It also shows the unnecessary register pressure caused by
over-sinking.

Added an X86 test to verify we can sink instructions out of loops regardless of
the dominance relationship. This test is reduced from Alexey's test in PR21115.

Updated an affected test in X86.

Also ran SPEC CINT2006 and llvm-test-suite for compilation time and runtime
performance. Results are attached separately in the review thread.

Reviewers: Jiangning, resistor, hfinkel

Reviewed By: hfinkel

Subscribers: hfinkel, bruno, volkalexey, llvm-commits, meheff, eliben, jholewinski

Differential Revision: http://reviews.llvm.org/D5633

llvm-svn: 219773
2014-10-15 03:27:43 +00:00
Eric Christopher eb9e87f6e3 Access subtarget specific variables off of the MachineFunction's
cached subtarget and not the TargetMachine.

llvm-svn: 219668
2014-10-14 07:00:33 +00:00
Jingyue Wu fd47fb9976 Revert r216862 due to a performance regression
Reported by Alexey Volkov in PR21115

llvm-svn: 218771
2014-10-01 15:22:13 +00:00
Bruno Cardoso Lopes d04f7596e7 [MachineSink+PGO] Teach MachineSink to use BlockFrequencyInfo
Machine Sink uses loop depth information to select between successors BBs to
sink machine instructions into, where BBs within smaller loop depths are
preferable.  This patch adds support for choosing between successors by using
profile information from BlockFrequencyInfo instead, whenever the information
is available.

Tested it under SPEC2006 train (average of 30 runs for each program); ~1.5%
execution speedup in average on x86-64 darwin.

<rdar://problem/18021659>

llvm-svn: 218472
2014-09-25 23:14:26 +00:00
Patrik Hagglund 57d315b7c1 [MachineSinking] Conservatively clear kill flags after coalescing.
This solves the problem of having a kill flag inside a loop
with a definition of the register prior to the loop:

%vreg368<def> ...

Inside loop:

        %vreg520<def> = COPY %vreg368
        %vreg568<def,tied1> = add %vreg341<tied0>, %vreg520<kill>

=> was coalesced into =>

        %vreg568<def,tied1> = add %vreg341<tied0>, %vreg368<kill>

MachineVerifier then complained:
*** Bad machine code: Virtual register killed in block, but needed live out. ***

The kill flag for %vreg368 is incorrect, and is cleared by this patch.

This is similar to the clearing done at the end of
MachineSinking::SinkInstruction().

Patch provided by Jonas Paulsson.

Reviewed by Quentin Colombet and Juergen Ributzka.

llvm-svn: 217427
2014-09-09 07:47:00 +00:00
Juergen Ributzka 4bea494569 Revert r216803 "[MachineSinking] Clear kill flag of all operands at all their uses."
This reverts commit r216803, because it might have broken the buildbot.
The issue is tracked in PR20842.

llvm-svn: 217120
2014-09-04 02:07:36 +00:00
Jingyue Wu 5208cc5dbe [MachineSink] Use the real post dominator tree
Summary:
Fixes a FIXME in MachineSinking. Instead of using the simple heuristics
in isPostDominatedBy, use the real MachinePostDominatorTree. The old
heuristics caused instructions to sink unnecessarily, and might create
register pressure.

Test Plan:
Added a NVPTX codegen test to verify that our change is in effect. It also
shows the unnecessary register pressure caused by over-sinking. Updated
affected tests in AArch64 and X86.

Reviewers: eliben, meheff, Jiangning

Reviewed By: Jiangning

Subscribers: jholewinski, aemerson, mcrosier, llvm-commits

Differential Revision: http://reviews.llvm.org/D4814

llvm-svn: 216862
2014-09-01 03:47:25 +00:00
Juergen Ributzka 00d78221ab [MachineSinking] Clear kill flag of all operands at all their uses.
When sinking an instruction it might be moved past the original last use of one
of its operands. This last use has the kill flag set and the verifier will
obviously complain about this.

Before Machine Sinking (AArch64):
%vreg3<def> = ASRVXr %vreg1, %vreg2<kill>
%XZR<def> = SUBSXrs %vreg4, %vreg1<kill>, 160, %NZCV<imp-def>
...

After Machine Sinking:
%XZR<def> = SUBSXrs %vreg4, %vreg1<kill>, 160, %NZCV<imp-def>
...
%vreg3<def> = ASRVXr %vreg1, %vreg2<kill>

This fix clears all the kill flags in all instruction that use the same operands
as the instruction that is being sunk.

This fixes rdar://problem/18180996.

llvm-svn: 216803
2014-08-29 23:48:03 +00:00
Quentin Colombet 5cded89d12 [MachineSink] Improve the compile time by preserving the dominance information
as long as possible.

** Context **

Each time the dominance information is modified, the dominator tree analysis
switches in a slow query mode. After a few queries without any modification on
the dominator tree, it performs an expensive update of its internal structure to
provide fast queries again.

** Problem **

Prior to this patch, the MachineSink pass was splitting the critical edges on
demand while relying heavy on the dominator tree information. In some cases,
this leads to pathological behavior where:
- We end up in the slow query mode right after splitting an edge.
- We update the dominance information.
- We break the dominance information again, thus ending up in the slow query
  mode and so on.

** Proposed Solution **

To mitigate this effect, this patch postpones all the splitting of the edges at
the end of each iteration of the main loop.
The benefits are:
- The dominance information is valid for the life time of an iteration.
- This simplifies the code as we do not have to special treat instructions that
  are sunk on critical edges. Indeed, the related block will be available
  through the next iteration.

The downside is that when edges splitting is required, this incurs an additional
iteration of the main loop compared to the previous scheme.

** Performance **

Thanks to this patch, the motivating example compiles in 6+ minutes instead of
10+ minutes. No test case added as the motivating example as nothing special but
being huge!

I have measured only noise for both the compile time and the runtime on the llvm
test-suite + SPECs with Os and O3.

Note: The current implementation of MachineBasicBlock::SplitCriticalEdge also
uses the dominance information and therefore, hits this problem. A subsequent
patch will address that.

<rdar://problem/17894619>

llvm-svn: 215410
2014-08-11 23:52:01 +00:00
Eric Christopher d913448b38 Remove the TargetMachine forwards for TargetSubtargetInfo based
information and update all callers. No functional change.

llvm-svn: 214781
2014-08-04 21:25:23 +00:00
Jiangning Liu c3053129b9 Add TargetInstrInfo interface isAsCheapAsAMove.
llvm-svn: 214158
2014-07-29 01:55:19 +00:00
Chandler Carruth 1b9dde087e [Modules] Remove potential ODR violations by sinking the DEBUG_TYPE
define below all header includes in the lib/CodeGen/... tree. While the
current modules implementation doesn't check for this kind of ODR
violation yet, it is likely to grow support for it in the future. It
also removes one layer of macro pollution across all the included
headers.

Other sub-trees will follow.

llvm-svn: 206837
2014-04-22 02:02:50 +00:00
Craig Topper c0196b1b40 [C++11] More 'nullptr' conversion. In some cases just using a boolean check instead of comparing to nullptr.
llvm-svn: 206142
2014-04-14 00:51:57 +00:00
Paul Robinson 7c99ec5b99 Disable each MachineFunctionPass for 'optnone' functions, unless that
pass normally runs at optimization level None, or is part of the
register allocation pipeline.

llvm-svn: 205228
2014-03-31 17:43:35 +00:00
Owen Anderson b36376efcb Switch a number of loops in lib/CodeGen over to range-based for-loops, now that
the MachineRegisterInfo iterators are compatible with it.

llvm-svn: 204075
2014-03-17 19:36:09 +00:00
Owen Anderson 16c6bf49b7 Phase 2 of the great MachineRegisterInfo cleanup. This time, we're changing
operator* on the by-operand iterators to return a MachineOperand& rather than
a MachineInstr&.  At this point they almost behave like normal iterators!

Again, this requires making some existing loops more verbose, but should pave
the way for the big range-based for-loop cleanups in the future.

llvm-svn: 203865
2014-03-13 23:12:04 +00:00
Craig Topper 4584cd54e3 [C++11] Add 'override' keyword to virtual methods that override their base class.
llvm-svn: 203220
2014-03-07 09:26:03 +00:00
Benjamin Kramer 3a377bce4e Now that we have C++11, turn simple functors into lambdas and remove a ton of boilerplate.
No intended functionality change.

llvm-svn: 202588
2014-03-01 11:47:00 +00:00
Will Dietz 5cb7f4e3f2 MachineSink: Fix and tweak critical-edge breaking heuristic.
Per original comment, the intention of this loop
is to go ahead and break the critical edge
(in order to sink this instruction) if there's
reason to believe doing so might "unblock" the
sinking of additional instructions that define
registers used by this one.  The idea is that if
we have a few instructions to sink "together"
breaking the edge might be worthwhile.

This commit makes a few small changes
to help better realize this goal:

First, modify the loop to ignore registers
defined by this instruction.  We don't
sink definitions of physical registers,
and sinking an SSA definition isn't
going to unblock an upstream instruction.

Second, ignore uses of physical registers.
Instructions that define physical registers are
rejected for sinking, and so moving this one
won't enable moving any defining instructions.
As an added bonus, while virtual register
use-def chains are generally small due
to SSA goodness, iteration over the uses
and definitions (used by hasOneNonDBGUse)
for physical registers like EFLAGS
can be rather expensive in practice.
(This is the original reason for looking at this)

Finally, to keep things simple continue
to only consider this trick for registers that
have a single use (via hasOneNonDBGUse),
but to avoid spuriously breaking critical edges
only do so if the definition resides
in the same MBB and therefore this one directly
blocks it from being sunk as well.
If sinking them together is meant to be,
let the iterative nature of this pass
sink the definition into this block first.

Update tests to accomodate this change,
add new testcase where sinking avoids pipeline stalls.

llvm-svn: 192608
2013-10-14 16:57:17 +00:00
Craig Topper b94011fd28 Use SmallVectorImpl& instead of SmallVector to avoid repeating small vector size.
llvm-svn: 186274
2013-07-14 04:42:23 +00:00
Craig Topper e1c1d363a5 Use SmallVectorImpl instead of SmallVector for iterators and references to avoid specifying the vector size unnecessarily.
llvm-svn: 185512
2013-07-03 05:11:49 +00:00
Chandler Carruth ed0881b2a6 Use the new script to sort the includes of every file under lib.
Sooooo many of these had incorrect or strange main module includes.
I have manually inspected all of these, and fixed the main module
include to be the nearest plausible thing I could find. If you own or
care about any of these source files, I encourage you to take some time
and check that these edits were sensible. I can't have broken anything
(I strictly added headers, and reordered them, never removed), but they
may not be the headers you'd really like to identify as containing the
API being implemented.

Many forward declarations and missing includes were added to a header
files to allow them to parse cleanly when included first. The main
module rule does in fact have its merits. =]

llvm-svn: 169131
2012-12-03 16:50:05 +00:00
Jakob Stoklund Olesen 244beb42ce Remove unused BitVectors from getAllocatableSet().
llvm-svn: 165999
2012-10-16 00:05:06 +00:00
Manman Ren f288d2f120 MachineSink: Sort the successors before trying to find SuccToSinkTo.
Use stable_sort instead of sort. Follow-up to r161062.

rdar://11980766

llvm-svn: 161075
2012-07-31 20:45:38 +00:00
Manman Ren 8c549b586c MachineSink: Sort the successors before trying to find SuccToSinkTo.
One motivating example is to sink an instruction from a basic block which has
two successors: one outside the loop, the other inside the loop. We should try
to sink the instruction outside the loop.

rdar://11980766

llvm-svn: 161062
2012-07-31 18:10:39 +00:00
Andrew Trick 1fa5bcbe2a Codegen pass definition cleanup. No functionality.
Moving toward a uniform style of pass definition to allow easier target configuration.
Globally declare Pass ID.
Globally declare pass initializer.
Use INITIALIZE_PASS consistently.
Add a call to the initializer from CodeGen.cpp.
Remove redundant "createPass" functions and "getPassName" methods.

While cleaning up declarations, cleaned up comments (sorry for large diff).

llvm-svn: 150100
2012-02-08 21:23:13 +00:00
Andrew Trick 9e761997d8 whitespace
llvm-svn: 150094
2012-02-08 21:22:43 +00:00
Jakob Stoklund Olesen 86ae07f049 Extract method for detecting constant unallocatable physregs.
It is safe to move uses of such registers.

llvm-svn: 148259
2012-01-16 22:34:08 +00:00
Devang Patel c268688643 Do not sink instruction, if it is not profitable.
On ARM, peephole optimization for ABS creates a trivial cfg triangle which tempts machine sink to sink instructions in code which is really straight line code. Sometimes this sinking may alter register allocator input such that use and def of a reg is divided by a branch in between, which may result in extra spills. Now mahine sink avoids sinking if final sink destination is post dominator.

Radar 10266272.

llvm-svn: 146604
2011-12-14 23:20:38 +00:00
Devang Patel 706574a994 Fix comment.
llvm-svn: 146226
2011-12-09 01:25:04 +00:00
Devang Patel 2f9a0e1b86 Update stale comment.
llvm-svn: 146220
2011-12-09 01:18:48 +00:00
Devang Patel 202cf2f6fc Revert r146184. I am seeing performance regression cause by this patch in one test case.
llvm-svn: 146205
2011-12-08 23:52:00 +00:00
Devang Patel b94c9a47e9 Refactor. No intentional functionality change.
llvm-svn: 146187
2011-12-08 21:48:01 +00:00
Devang Patel 1a3c1697f9 Filter "sink to" candidate blocks sooner. This avoids unnecessary computation to determine whether the block dominates all uses or not.
llvm-svn: 146184
2011-12-08 21:33:23 +00:00
Evan Cheng 7f8e563a69 Add bundle aware API for querying instruction properties and switch the code
generator to it. For non-bundle instructions, these behave exactly the same
as the MC layer API.

For properties like mayLoad / mayStore, look into the bundle and if any of the
bundled instructions has the property it would return true.
For properties like isPredicable, only return true if *all* of the bundled
instructions have the property.
For properties like canFoldAsLoad, isCompare, conservatively return false for
bundles.

llvm-svn: 146026
2011-12-07 07:15:52 +00:00
Devang Patel 9de7a7db26 While sinking machine instructions, sink matching DBG_VALUEs also otherwise live debug variable pass will drop DBG_VALUEs on the floor.
llvm-svn: 139208
2011-09-07 00:07:58 +00:00
Evan Cheng fe917efc8b Fix a couple of places where changes are made but not tracked.
llvm-svn: 129287
2011-04-11 18:47:20 +00:00
Owen Anderson 6c18d1aac0 Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which
must be called in the pass's constructor.  This function uses static dependency declarations to recursively initialize
the pass's dependencies.

Clients that only create passes through the createFooPass() APIs will require no changes.  Clients that want to use the
CommandLine options for passes will need to manually call the appropriate initialization functions in PassInitialization.h
before parsing commandline arguments.

I have tested this with all standard configurations of clang and llvm-gcc on Darwin.  It is possible that there are problems
with the static dependencies that will only be visible with non-standard options.  If you encounter any crash in pass
registration/creation, please send the testcase to me directly.

llvm-svn: 116820
2010-10-19 17:21:58 +00:00
Owen Anderson 8ac477ffb5 Begin adding static dependence information to passes, which will allow us to
perform initialization without static constructors AND without explicit initialization
by the client.  For the moment, passes are required to initialize both their
(potential) dependencies and any passes they preserve.  I hope to be able to relax
the latter requirement in the future.

llvm-svn: 116334
2010-10-12 19:48:12 +00:00
Owen Anderson df7a4f2515 Now with fewer extraneous semicolons!
llvm-svn: 115996
2010-10-07 22:25:06 +00:00
Evan Cheng d4b31a7630 Don't sink insert_subreg, subreg_to_reg, reg_sequence. They are meant to be
close to their sources to facilitate coalescing.

llvm-svn: 114631
2010-09-23 06:53:00 +00:00
Evan Cheng f3e9a48584 Enable machine sinking critical edge splitting. e.g.
define double @foo(double %x, double %y, i1 %c) nounwind {
  %a = fdiv double %x, 3.2
  %z = select i1 %c, double %a, double %y
  ret double %z
}

Was:
_foo:
        divsd   LCPI0_0(%rip), %xmm0
        testb   $1, %dil
        jne     LBB0_2
        movaps  %xmm1, %xmm0
LBB0_2:
        ret

Now:
_foo:
        testb   $1, %dil
        je      LBB0_2
        divsd   LCPI0_0(%rip), %xmm0
        ret
LBB0_2:
        movaps  %xmm1, %xmm0
        ret

This avoids the divsd when early exit is taken.
rdar://8454886

llvm-svn: 114372
2010-09-20 22:52:00 +00:00
Evan Cheng 2031b768ba Avoid splitting critical edge twice for a set of PHI uses.
llvm-svn: 114338
2010-09-20 19:12:55 +00:00
Evan Cheng b339f3da0c Fix code that break critical edges for PHI uses. Watch out for multiple PHIs in different blocks.
llvm-svn: 114270
2010-09-18 06:42:17 +00:00
Evan Cheng e53ab6dffc Teach machine sink to
1) Do forward copy propagation. This makes it easier to estimate the cost of the
   instruction being sunk.
2) Break critical edges on demand, including cases where the value is used by
   PHI nodes.
Critical edge splitting is not yet enabled by default.

llvm-svn: 114227
2010-09-17 22:28:18 +00:00
Evan Cheng e5af930156 Update debug logs.
llvm-svn: 111575
2010-08-19 23:33:02 +00:00
Evan Cheng 361b9be7c6 It's possible to sink a def if its local uses are PHI's.
llvm-svn: 111537
2010-08-19 18:33:29 +00:00
Evan Cheng 681d0c25f9 Remove disabled assertion.
llvm-svn: 111531
2010-08-19 17:33:48 +00:00
Evan Cheng ae9939c839 Teach machine-sink to break critical edges when appropriate. Work in progress.
llvm-svn: 111530
2010-08-19 17:33:11 +00:00
Evan Cheng 25b6068b8f If any def of a machine-sink candidate has local uses, it's obviously not safe to sink it to a successor block. This bug has been hidden because a later check for critical-edge disable these illegal optimizations. This patch should significantly reduce the amount of time spent on checking dominator information for obviously unsafe sinking.
llvm-svn: 111450
2010-08-18 23:09:25 +00:00
Owen Anderson a7aed18624 Reapply r110396, with fixes to appease the Linux buildbot gods.
llvm-svn: 110460
2010-08-06 18:33:48 +00:00
Owen Anderson bda59bd247 Revert r110396 to fix buildbots.
llvm-svn: 110410
2010-08-06 00:23:35 +00:00
Owen Anderson 755aceb5d0 Don't use PassInfo* as a type identifier for passes. Instead, use the address of the static
ID member as the sole unique type identifier.  Clean up APIs related to this change.

llvm-svn: 110396
2010-08-05 23:42:04 +00:00
Owen Anderson a57b97e7e7 Fix batch of converting RegisterPass<> to INTIALIZE_PASS().
llvm-svn: 109045
2010-07-21 22:09:45 +00:00
Bill Wendling e41e40f689 - Reapply r106066 now that the bzip2 build regression has been fixed.
- 2010-06-25-CoalescerSubRegDefDead.ll is the testcase for r106878.

llvm-svn: 106880
2010-06-25 20:48:10 +00:00
Daniel Dunbar ef5a4383ad Revert r106066, "Create a more targeted fix for not sinking instructions into a range where it"... it causes bzip2 to be miscompiled by Clang.
Conflicts:

	lib/CodeGen/MachineSink.cpp

llvm-svn: 106614
2010-06-23 00:48:25 +00:00
Bill Wendling d71bd63600 Improve comment to include that the use of a preg is also verboten in this situation.
llvm-svn: 106119
2010-06-16 18:01:31 +00:00
Bill Wendling 8c0cf0994d Create a more targeted fix for not sinking instructions into a range where it
will conflict with another live range. The place which creates this scenerio is
the code in X86 that lowers a select instruction by splitting the MBBs. This
eliminates the need to check from the bottom up in an MBB for live pregs.

llvm-svn: 106066
2010-06-15 23:46:31 +00:00
Jim Grosbach 01edd68225 Cleanup 80-column and trim trailing whitespace
llvm-svn: 105435
2010-06-03 23:49:57 +00:00
Bill Wendling f82aea634c Machine sink could potentially sink instructions into a block where the physical
registers it defines then interfere with an existing preg live range.

For instance, if we had something like these machine instructions:

BB#0
  ... = imul ... EFLAGS<imp-def,dead>
  test ..., EFLAGS<imp-def>
  jcc BB#2 EFLAGS<imp-use>

BB#1
  ... ; fallthrough to BB#2

BB#2
  ... ; No code that defines EFLAGS
  jcc ... EFLAGS<imp-use>

Machine sink will come along, see that imul implicitly defines EFLAGS, but
because it's "dead", it assumes that it can move imul into BB#2. But when it
does, imul's "dead" imp-def of EFLAGS is raised from the dead (a zombie) and
messes up the condition code for the jump (and pretty much anything else which
relies upon it being correct).

The solution is to know which pregs are live going into a basic block. However,
that information isn't calculated at this point. Nor does the LiveVariables pass
take into account non-allocatable physical registers. In lieu of this, we do a
*very* conservative pass through the basic block to determine if a preg is live
coming out of it.

llvm-svn: 105387
2010-06-03 07:54:20 +00:00
Bill Wendling 7ee730eb40 Compulsive reformating. No functionalitical changes.
llvm-svn: 105359
2010-06-02 23:04:26 +00:00
Dan Gohman c90f51c00b Teach MachineLICM and MachineSink how to clear kill flags conservatively
when they move instructions.

llvm-svn: 103737
2010-05-13 20:34:42 +00:00
Jakob Stoklund Olesen cdc3df4888 Avoid sinking machine instructions into a loop.
MachineLoopInfo is already available when MachineSinking runs, so the check is
free.

There is no test case because it would require a critical edge into a loop, and
CodeGenPrepare splits those. This check is just to be extra careful.

llvm-svn: 101420
2010-04-15 23:41:02 +00:00
Jakob Stoklund Olesen 20b71e28cc Teach MachineSinking to handle easy critical edges.
Sometimes it is desirable to sink instructions along a critical edge:

x = ...
if (a && b) ...
else use(x);

The 'a && b' condition creates a critical edge to the else block, but we still
want to sink the computation of x into the block. The else block is dominated by
the parent block, so we are not pushing instructions into new code paths.

llvm-svn: 101165
2010-04-13 19:06:14 +00:00
Dan Gohman 918a90a3ca Don't do code sinking on unreachable blocks. It's unprofitable and hazardous.
llvm-svn: 100455
2010-04-05 19:17:22 +00:00
Dale Johannesen 2061c84109 Fix some more places where dbg_value affected codegen.
llvm-svn: 97765
2010-03-05 00:02:59 +00:00
Evan Cheng 62e795ab8c Swap parameters of isSafeToMove and isSafeToReMat for consistency.
llvm-svn: 97578
2010-03-02 19:03:01 +00:00
Chris Lattner b06015aa69 move target-independent opcodes out of TargetInstrInfo
into TargetOpcodes.h.  #include the new TargetOpcodes.h
into MachineInstr.  Add new inline accessors (like isPHI())
to MachineInstr, and start using them throughout the 
codebase.

llvm-svn: 95687
2010-02-09 19:54:29 +00:00
David Greene 4b7aa24842 Change errs() to dbgs().
llvm-svn: 92593
2010-01-05 01:26:00 +00:00
Nick Lewycky 974e12b2d3 Remove includes of Support/Compiler.h that are no longer needed after the
VISIBILITY_HIDDEN removal.

llvm-svn: 85043
2009-10-25 06:57:41 +00:00
Nick Lewycky 02d5f77d26 Remove VISIBILITY_HIDDEN from class/struct found inside anonymous namespaces.
Chris claims we should never have visibility_hidden inside any .cpp file but
that's still not true even after this commit.

llvm-svn: 85042
2009-10-25 06:33:48 +00:00
Dan Gohman b0db9917af Fix a typo in a comment.
llvm-svn: 84504
2009-10-19 14:56:05 +00:00
Dan Gohman 20a327b31f Change a few instance variables to be local variables.
llvm-svn: 84503
2009-10-19 14:52:05 +00:00
Dan Gohman 87b02d5bbc Factor out LiveIntervalAnalysis' code to determine whether an instruction
is trivially rematerializable and integrate it into
TargetInstrInfo::isTriviallyReMaterializable. This way, all places that
need to know whether an instruction is rematerializable will get the
same answer.

This enables the useful parts of the aggressive-remat option by
default -- using AliasAnalysis to determine whether a memory location
is invariant, and removes the questionable parts -- rematting operations
with virtual register inputs that may not be live everywhere.

llvm-svn: 83687
2009-10-09 23:27:56 +00:00
Dan Gohman be8137b0b4 Replace TargetInstrInfo::isInvariantLoad and its target-specific
implementations with a new MachineInstr::isInvariantLoad, which uses
MachineMemOperands and is target-independent. This brings MachineLICM
and other functionality to targets which previously lacked an
isInvariantLoad implementation.

llvm-svn: 83475
2009-10-07 17:38:06 +00:00