Commit Graph

2838 Commits

Author SHA1 Message Date
Bill Wendling 4822a7ac8a For this pattern, ROTR is the cheaper option.
llvm-svn: 55576
2008-08-31 01:04:56 +00:00
Bill Wendling fc72416447 - Fix comment so that it describes how the code really works:
// fold (or (shl x, (*ext y)), (srl x, (*ext (sub 32, y)))) ->
   //   (rotl x, y)
   // fold (or (shl x, (*ext y)), (srl x, (*ext (sub 32, y)))) ->
   //   (rotr x, (sub 32, y))

Example: (x == 0xDEADBEEF and y == 4)

    (x << 4) | (x >> 28)
 => 0xEADBEEF0 | 0x0000000D
 => 0xEADBEEFD

    (rotl x, 4)
 => 0xEADBEEFD

    (rotr x, 28)
 => 0xEADBEEFD

- Fix comment and code for second version. It wasn't using the rot* propertly.

   // fold (or (shl x, (*ext (sub 32, y))), (srl x, (*ext r))) -> 
   //   (rotr x, y)
   // fold (or (shl x, (*ext (sub 32, y))), (srl x, (*ext r))) ->
   //   (rotl x, (sub 32, y))

    (x << 28) | (x >> 4)
 => 0xD0000000 | 0x0DEADBEE
 => 0xDDEADBEE

    (rotl x, 4)
 => 0xEADBEEFD

    (rotr x, 28)
 => (0xEADBEEFD)

llvm-svn: 55575
2008-08-31 00:37:27 +00:00
Gabor Greif 66ccf603a9 typo
llvm-svn: 55574
2008-08-30 22:16:05 +00:00
Gabor Greif e12264bf41 fix some 80-col violations
llvm-svn: 55571
2008-08-30 19:29:20 +00:00
Evan Cheng cfb7f3abdf Transform (x << (y&31)) -> (x << y). This takes advantage of the fact x86 shift instructions 2nd operand (shift count) is limited to 0 to 31 (or 63 in the x86-64 case).
llvm-svn: 55558
2008-08-30 02:03:58 +00:00
Owen Anderson 6f0c51d9da Fix an issue where a use might be selected before a def, and then we didn't respect the pre-chosen vreg
assignment when selecting the def.  This is the naive solution to the problem: insert a copy to the pre-chosen
vreg.  Other solutions might be preferable, such as:
  1) Passing the dest reg into FastEmit_.  However, this would require the higher level code to know about reg classes, which they don't currently.
  2) Selecting blocks in reverse postorder.  This has some compile time cost for computing the order, and we'd need to measure its impact.

llvm-svn: 55555
2008-08-30 00:38:46 +00:00
Evan Cheng 894be333f1 Fix 80 col. violations.
llvm-svn: 55551
2008-08-29 23:20:46 +00:00
Evan Cheng 5e7658c2e4 Back out 55498. It broken Apple style bootstrapping.
llvm-svn: 55549
2008-08-29 22:21:44 +00:00
Dan Gohman d58f3e36d0 Add a target callback for FastISel.
llvm-svn: 55512
2008-08-28 23:21:34 +00:00
Gabor Greif f304a7aa4d erect abstraction boundaries for accessing SDValue members, rename Val -> Node to reflect semantics
llvm-svn: 55504
2008-08-28 21:40:38 +00:00
Dan Gohman c45733f194 Implement null and undef values for FastISel.
llvm-svn: 55500
2008-08-28 21:19:07 +00:00
Dan Gohman f27e33baa7 Optimize DAGCombiner's worklist processing. Previously it started
its work by putting all nodes in the worklist, requiring a big
dynamic allocation. Now, DAGCombiner just iterates over the AllNodes
list and maintains a worklist for nodes that are newly created or
need to be revisited. This allows the worklist to stay small in most
cases, so it can be a SmallVector.

This has the side effect of making DAGCombine not miss a folding
opportunity in alloca-align-rounding.ll.

llvm-svn: 55498
2008-08-28 21:01:56 +00:00
Dan Gohman 17da671922 Move CaseBlock, JumpTable, and BitTestBlock to be members of
SelectionDAGLowering instead of being in an anonymous namespace.
This fixes warnings about SelectionDAGLowering having fields
using anonymous namespaces.

llvm-svn: 55497
2008-08-28 20:38:18 +00:00
Dan Gohman 360c57f683 Fix a FastISel bug where the instructions from lowering the arguments
were being emitted after the first instructions of the entry block.

llvm-svn: 55496
2008-08-28 20:28:56 +00:00
Rafael Espindola 6c8a99a778 Reduce the size of the Parts vector.
llvm-svn: 55483
2008-08-28 18:29:58 +00:00
Owen Anderson d8a82b75e2 Hook up support for fast-isel of trunc instructions, using the newly working support for EXTRACT_SUBREG.
llvm-svn: 55482
2008-08-28 18:26:01 +00:00
Owen Anderson 9cd1a5e530 FastEmitInst_extractsubreg doesn't need to be passed the register class. It can get it from MachineRegisterInfo instead.
llvm-svn: 55476
2008-08-28 17:47:37 +00:00
Rafael Espindola 029c1c8460 Correctly resize the Parts array.
llvm-svn: 55471
2008-08-28 14:24:45 +00:00
Dale Johannesen 41be0d4445 Split the ATOMIC NodeType's to include the size, e.g.
ATOMIC_LOAD_ADD_{8,16,32,64} instead of ATOMIC_LOAD_ADD.
Increased the Hardcoded Constant OpActionsCapacity to match.
Large but boring; no functional change.

This is to support partial-word atomics on ppc; i8 is
not a valid type there, so by the time we get to lowering, the
ATOMIC_LOAD nodes looks the same whether the type was i8 or i32.
The information can be added to the AtomicSDNode, but that is the
largest SDNode; I don't fully understand the SDNode allocation,
but it is sensitive to the largest node size, so increasing
that must be bad.  This is the alternative.

llvm-svn: 55457
2008-08-28 02:44:49 +00:00
Dan Gohman e1a9a780a5 Reorganize the lifetimes of the major objects SelectionDAGISel
works with.

SelectionDAG, FunctionLoweringInfo, and SelectionDAGLowering
objects now get created once per SelectionDAGISel instance, and
can be reused across blocks and across functions. Previously,
they were created and destroyed each time they were needed.

This reorganization simplifies the handling of PHI nodes, and
also SwitchCases, JumpTables, and BitTestBlocks. This
simplification has the side effect of fixing a bug in FastISel
where successor PHI nodes weren't being updated correctly.

This is also a step towards making the transition from FastISel
into and out of SelectionDAG faster, and also making
plain SelectionDAG faster on code with lots of little blocks.

llvm-svn: 55450
2008-08-27 23:52:12 +00:00
Owen Anderson 5f57bc2247 Add a helper method that will be used to support EXTRACT_SUBREG for selecting trunc's in fast-isel.
llvm-svn: 55439
2008-08-27 22:30:02 +00:00
Dan Gohman 61cfa3095d Fix FastISel's bitcast code for the case where getRegForValue fails.
llvm-svn: 55431
2008-08-27 20:41:38 +00:00
Owen Anderson 90609850b2 Use TargetLowering to get the types in fast isel, which handles pointer types correctly for our purposes.
llvm-svn: 55428
2008-08-27 18:58:30 +00:00
Dan Gohman d01789be23 Don't check TLI.getOperationAction. The FastISel way is to
just try to do the action and let the tablegen-generated code
determine if there is target-support for an operation.

llvm-svn: 55427
2008-08-27 18:15:05 +00:00
Dan Gohman b0b5a27438 Add a new FastISel method, getRegForValue, which takes care of
the details of materializing constants and other values into
registers, and make use of it in several places.

llvm-svn: 55426
2008-08-27 18:10:19 +00:00
Dan Gohman f2a6c1579f Add a comment about the current floating-point constant code in FastISel.
llvm-svn: 55425
2008-08-27 18:01:42 +00:00
Dan Gohman 3a3a52de58 Optimize ScheduleDAGRRList's topological sort to use one pass instead
of two, and to not need a scratch std::vector. Also, compute the ordering
immediately in the result array, instead of in another scratch std::vector
that is copied to the result array.

llvm-svn: 55421
2008-08-27 16:29:48 +00:00
Dan Gohman 9cbdedcbcf Optimize ScheduleDAG's ComputeDepths and ComputeHeights to not need
a scratch std::vector.

llvm-svn: 55420
2008-08-27 16:27:25 +00:00
Dan Gohman 5ca269e684 Basic FastISel support for floating-point constants.
llvm-svn: 55401
2008-08-27 01:09:54 +00:00
Owen Anderson 54aff7bb23 Fix handling of inttoptr and ptrtoint when unhandled operands are present.
llvm-svn: 55400
2008-08-27 00:35:37 +00:00
Owen Anderson 140549256f Add support for fast isel of inttoptr and ptrtoint in the cases where truncation is not needed.
llvm-svn: 55399
2008-08-27 00:31:01 +00:00
Owen Anderson ca1711a5b5 Factor out a large amoutn of the cast handling code in fast isel into helper methods.
This simultaneously makes the code simpler and adds support for sext as well.

llvm-svn: 55398
2008-08-26 23:46:32 +00:00
Owen Anderson 343310a715 Add support for fast isel of zext.
llvm-svn: 55396
2008-08-26 23:14:49 +00:00
Gabor Greif abfdf928d8 disallow direct access to SDValue::ResNo, provide a getter instead
llvm-svn: 55394
2008-08-26 22:36:50 +00:00
Owen Anderson 655c1dc63d Add support for fptosi of constants in fast isel.
llvm-svn: 55393
2008-08-26 22:34:28 +00:00
Dan Gohman d56f73f2f2 Optimize SelectionDAG's topological sort to use one pass instead
of two, and to not need a scratch std::vector. Also, use the
SelectionDAG's topological sort in LegalizeDAG instead of having
a separate implementation.

llvm-svn: 55389
2008-08-26 21:42:18 +00:00
Dan Gohman 6fda9208d9 Refactor the bitcast code into its own function.
llvm-svn: 55387
2008-08-26 21:28:54 +00:00
Dan Gohman b5e04bfb18 Make FastISel use the correct argument type when casting GEP indices.
llvm-svn: 55384
2008-08-26 20:57:08 +00:00
Dan Gohman 3bcbbece19 Don't select binary instructions with illegal types.
llvm-svn: 55383
2008-08-26 20:52:40 +00:00
Owen Anderson 3c4dc434ee Add support for fast isel of sitofp, and remove some unnecessary and imprecise legality checks.
llvm-svn: 55381
2008-08-26 20:37:00 +00:00
Owen Anderson e0ac9765b2 Use a combination of copyRegToReg and ISD::BIT_CONVERT when doing fast isel of bitcasts,
allowing it to support the full range of conversions people might ask for in a correct manner.

llvm-svn: 55378
2008-08-26 18:51:24 +00:00
Owen Anderson 27fb3dcbc7 Make TargetInstrInfo::copyRegToReg return a bool indicating whether the copy requested
was inserted or not.  This allows bitcast in fast isel to properly handle the case
where an appropriate reg-to-reg copy is not available.

llvm-svn: 55375
2008-08-26 18:03:31 +00:00
Owen Anderson bf05ebaccf Add support for fast isel of non-constant fptosi instructions.
llvm-svn: 55373
2008-08-26 17:44:42 +00:00
Chris Lattner 54ef9f5831 typo fix.
llvm-svn: 55355
2008-08-26 06:07:47 +00:00
Dan Gohman 2e834906b9 Actually recycle SDNode allocations. SelectionDAG is using
RecyclingAllocator, but this change is needed for the nodes
to actually be recycled. This cuts SelectionDAG's memory
usage high-water-mark in half in some cases.

llvm-svn: 55351
2008-08-26 01:44:34 +00:00
Owen Anderson 8dd01ccdd8 Add a RetVT parameter to emitted FastISel methods, so that we will be able to pass the desired return
type down.  This is not currently used.

llvm-svn: 55345
2008-08-25 23:58:18 +00:00
Evan Cheng 2c067325d6 Unbreak build.
llvm-svn: 55342
2008-08-25 22:20:39 +00:00
Owen Anderson 126afc5cb9 Expand bitcast support in fast isel to support bitcasts of non-constant values by emitting reg-reg copies.
llvm-svn: 55340
2008-08-25 21:32:34 +00:00
Owen Anderson 32635dbfb2 Add support for fast isel of (integer) immediate materialization pattens, and use them to support
bitcast of constants in fast isel.

llvm-svn: 55325
2008-08-25 20:20:32 +00:00
Chris Lattner f4bd5cf3dd make sure to flush the stream after dumping, to make sure it goes out immediately.
llvm-svn: 55288
2008-08-24 18:28:30 +00:00
Chris Lattner 838aff36dd get MachineConstantPool off std::ostream, onto raw_ostream. It would be
really nice if someone converted MachineFunction::print to raw_ostream.

llvm-svn: 55268
2008-08-23 22:53:13 +00:00
Chris Lattner 0c19df4871 Switch the asmprinter (.ll) and all the stuff it requires over to
use raw_ostream instead of std::ostream.  Among other goodness,
this speeds up llvm-dis of kc++ with a release build from 0.85s
to 0.49s (88% faster).

Other interesting changes:
 1) This makes Value::print be non-virtual.
 2) AP[S]Int and ConstantRange can no longer print to ostream directly, 
    use raw_ostream instead.
 3) This fixes a bug in raw_os_ostream where it didn't flush itself 
    when destroyed.
 4) This adds a new SDNode::print method, instead of only allowing "dump".


A lot of APIs have both std::ostream and raw_ostream versions, it would
be useful to go through and systematically anihilate the std::ostream 
versions.

This passes dejagnu, but there may be minor fallout, plz let me know if
so and I'll fix it.

llvm-svn: 55263
2008-08-23 22:23:09 +00:00
Dan Gohman 48a3623591 Make MBBMap a DenseMap instead of a std::map.
llvm-svn: 55220
2008-08-23 02:44:46 +00:00
Dan Gohman eb0cee91f6 Move the point at which FastISel taps into the SelectionDAGISel
process up to a higher level. This allows FastISel to leverage
more of SelectionDAGISel's infastructure, such as updating Machine
PHI nodes.

Also, implement transitioning from SDISel back to FastISel in
the middle of a block, so it's now possible to go back and
forth. This allows FastISel to hand individual CallInsts and other
complicated things off to SDISel to handle, while handling the rest
of the block itself.

To help support this, reorganize the SelectionDAG class so that it
is allocated once and reused throughout a function, instead of
being completely reallocated for each block.

llvm-svn: 55219
2008-08-23 02:25:05 +00:00
Dan Gohman 95d1056831 Avoid creating shift-by-zero SDNodes in the common case of
i8* getelementptr. DAGCombine eliminates these, but this is
a fairly common case.

llvm-svn: 55214
2008-08-23 01:06:51 +00:00
Dan Gohman ac37f9a9be Move SelectionDAG's constructor out of line.
llvm-svn: 55212
2008-08-23 00:50:30 +00:00
Dan Gohman 2db3f8a095 Reapply r55191 and r55192.
llvm-svn: 55205
2008-08-22 21:28:19 +00:00
Bill Wendling fc4f64eed0 Reverting r55190, r55191, and r55192. They broke the build with this error message:
{standard input}:17:bad register name `%sil'
make[4]: *** [libgcc/./_addvsi3.o] Error 1
make[4]: *** Waiting for unfinished jobs....
{standard input}:23:bad register name `%dil'
{standard input}:28:bad register name `%dil'
make[4]: *** [libgcc/./_addvdi3.o] Error 1
{standard input}:18:bad register name `%sil'
make[4]: *** [libgcc/./_subvsi3.o] Error 1

llvm-svn: 55200
2008-08-22 20:51:05 +00:00
Dan Gohman 04968da460 Fix the InsertBranch call.
llvm-svn: 55192
2008-08-22 19:26:10 +00:00
Dan Gohman 87ff7058e7 Support non-fallthrough unconditional branches in FastISel.
llvm-svn: 55191
2008-08-22 19:21:41 +00:00
Dan Gohman a2292c0d34 Add FastISel support for PHINodes. Machine PHI nodes
are not yet updated properly, but that's a separate
task.

llvm-svn: 55187
2008-08-22 17:37:48 +00:00
Dan Gohman 49e19e906f Factor out the predicate check code from DAGISelEmitter.cpp
and use it in FastISelEmitter.cpp, and make FastISel
subtarget aware. Among other things, this lets it work
properly on x86 targets that don't have SSE, where it
successfully selects x87 instructions.

llvm-svn: 55156
2008-08-22 00:20:26 +00:00
Dan Gohman 2af34bd309 Add libcalls for the new rounding opcodes.
llvm-svn: 55133
2008-08-21 18:38:14 +00:00
Dan Gohman c6337ac069 Add libm-oriented ISD opcodes for rounding operations.
llvm-svn: 55130
2008-08-21 17:55:02 +00:00
Dan Gohman 6a7461ad9b Have FastISel skip the multiply by 1 for getelementptr on i8*.
llvm-svn: 55129
2008-08-21 17:37:05 +00:00
Dan Gohman efb7d2d03d MVT::getMVT uses iPTR for pointer types, while we need the actual
intptr_t type in this case. FastISel can now select simple
getelementptr instructions.

llvm-svn: 55125
2008-08-21 17:25:26 +00:00
Dan Gohman 75ea0b83c5 Elements in DeadNodeSet are checked for use_empty() before they
are actually deleted, so it's not necessary to remove re-used
nodes from the set.

llvm-svn: 55123
2008-08-21 16:24:54 +00:00
Dan Gohman fe9056584b Basic fast-isel support for instructions with constant int operands.
llvm-svn: 55099
2008-08-21 01:41:07 +00:00
Evan Cheng 4b5c038cd0 Type of first GEP operand is always the same as the target pointer type.
llvm-svn: 55097
2008-08-21 01:19:11 +00:00
Dan Gohman 6a0780cdd7 Fix unused variable warnings.
llvm-svn: 55089
2008-08-20 23:53:10 +00:00
Evan Cheng 864fcc198d First cut, un-optimized (and untested) fast isel lowering of GetElementPtrInst.
llvm-svn: 55085
2008-08-20 22:45:34 +00:00
Dan Gohman a4305cec93 Simplify the BuildMI calls even more.
llvm-svn: 55077
2008-08-20 21:10:53 +00:00
Dan Gohman 02c84b8910 Simplify FastISel's constructor argument list, make the FastISel
class hold a MachineRegisterInfo member, and make the
MachineBasicBlock be passed in to SelectInstructions rather
than the FastISel constructor.

llvm-svn: 55076
2008-08-20 21:05:57 +00:00
Dan Gohman 43d1c7c607 Dump the instruction that foiled ISel even when -debug is not used.
llvm-svn: 55075
2008-08-20 20:47:32 +00:00
Dan Gohman 07a34a5f69 Make more use of the BuildMI API.
llvm-svn: 55072
2008-08-20 18:16:32 +00:00
Dan Gohman 24e8f0cfe6 Minor code reorganization.
llvm-svn: 55071
2008-08-20 18:10:48 +00:00
Dan Gohman 2471f6ce0f Minor whitespace cleanup.
llvm-svn: 55070
2008-08-20 18:09:38 +00:00
Dan Gohman 39a5ffb03f Fix 80 column violation.
llvm-svn: 55069
2008-08-20 18:09:02 +00:00
Evan Cheng 7b9cd58596 Kill off SimpleBBISel, it's replaced by FastISel.
llvm-svn: 55067
2008-08-20 17:50:32 +00:00
Dan Gohman 837c13a029 Disable DAGCombine's alignment inference in "fast" codegen mode.
llvm-svn: 55059
2008-08-20 16:30:28 +00:00
Dan Gohman 2da2bedc72 Change the FoldingSetNodeID usage for objects which carry
alignment and volatility information, such as loads and
stores, to reduce the number of integer values added to
the FoldingSetNodeID.

llvm-svn: 55058
2008-08-20 15:58:01 +00:00
Dan Gohman f6aa60ff71 Use BitVector instead of std::vector<unsigned char>.
llvm-svn: 55054
2008-08-20 14:58:41 +00:00
Dan Gohman c63a46ef39 Avoid an empty-if-body warning in release builds.
llvm-svn: 55050
2008-08-20 14:00:56 +00:00
Dan Gohman e8f9a00424 Fix FastISel to recognize that the last block in the function does
not have a fall-through successor.

llvm-svn: 55033
2008-08-20 01:17:01 +00:00
Dan Gohman 98265cae87 Fix a leak in the FastISel code that Chris pointed out.
llvm-svn: 55031
2008-08-20 00:56:17 +00:00
Dan Gohman 847ebb90b8 Add support for running SelectionDAG if FastISel fails. This is under
a command-line option, so that the default behavior is an abort, which
is useful for exposing code that isn't supported yet.

llvm-svn: 55028
2008-08-20 00:47:54 +00:00
Dan Gohman f6884373c2 Fix FastISel to recognize unhandled operands, such as constants
that aren't available as virtual registers (for now).

llvm-svn: 55026
2008-08-20 00:35:17 +00:00
Dan Gohman b16a7783c5 Add FastISel support for floating-point operations.
llvm-svn: 55021
2008-08-20 00:23:20 +00:00
Dan Gohman a3e4d5a5e1 Add FastISel support for several more binary operators.
llvm-svn: 55020
2008-08-20 00:11:48 +00:00
Dan Gohman 697284fe0a Add code to call FastISel, and a command-line option to enable it.
llvm-svn: 55015
2008-08-19 22:33:34 +00:00
Dan Gohman 214343fbbe Support unconditional fall-through branches in FastISel.
llvm-svn: 55014
2008-08-19 22:31:46 +00:00
Dan Gohman 547ce65467 Use the BuildMI overload that sets up a destination register
instead of the one that doesn't and then adding it manually.

llvm-svn: 55006
2008-08-19 20:46:54 +00:00
Dan Gohman c55fdcc935 Handle the case where target-specific fastisel code doesn't have
a desired opcode.

llvm-svn: 55005
2008-08-19 20:43:22 +00:00
Chris Lattner 5d2a9a4ae6 don't use the result of WriteTypeSymbolic or WriteAsOperand.
llvm-svn: 54978
2008-08-19 04:44:30 +00:00
Gordon Henriksen d930f913e6 Rename some GC classes so that their roll will hopefully be clearer.
In particular, Collector was confusing to implementors. Several
thought that this compile-time class was the place to implement
their runtime GC heap. Of course, it doesn't even exist at runtime.
Specifically, the renames are:

  Collector               -> GCStrategy
  CollectorMetadata       -> GCFunctionInfo
  CollectorModuleMetadata -> GCModuleInfo
  CollectorRegistry       -> GCRegistry
  Function::getCollector  -> getGC (setGC, hasGC, clearGC)

Several accessors and nested types have also been renamed to be
consistent. These changes should be obvious.

llvm-svn: 54899
2008-08-17 18:44:35 +00:00
Gordon Henriksen bcef14d2e4 Factor GC metadata table assembly generation out of Collector in preparation for splitting AsmPrinter into its own library.
llvm-svn: 54881
2008-08-17 12:56:54 +00:00
Chris Lattner 17f7165f84 Rework the routines that convert AP[S]Int into a string. Now, instead of
returning an std::string by value, it fills in a SmallString/SmallVector
passed in.  This significantly reduces string thrashing in some cases.

More specifically, this:
 - Adds an operator<< and a print method for APInt that allows you to 
   directly send them to an ostream.
 - Reimplements APInt::toString to be much simpler and more efficient
   algorithmically in addition to not thrashing strings quite as much.

This speeds up llvm-dis on kc++ by 7%, and may also slightly speed up the
asmprinter.  This also fixes a bug I introduced into the asmwriter in a
previous patch w.r.t. alias printing.

llvm-svn: 54873
2008-08-17 07:19:36 +00:00
Dan Gohman c44423853a Make FastISel's constructor protected, and give it a destructor.
llvm-svn: 54793
2008-08-14 21:51:29 +00:00
Dan Gohman 550c9af91f Improve support for vector casts in LLVM IR and CodeGen.
llvm-svn: 54784
2008-08-14 20:04:46 +00:00
Dan Gohman 6134fbccef Fix a bogus srem rule - a negative value srem'd by a power-of-2
can have a non-negative result; for example, -16%16 is 0. Also,
clarify the related comments. This fixes PR2670.

llvm-svn: 54767
2008-08-13 23:12:35 +00:00
Dan Gohman 7e3c392248 Allow SelectionDAG to create EXTRACT_VECTOR_ELT nodes with
non-constant indices. Only a few of the peephole checks require
a constant index.

llvm-svn: 54764
2008-08-13 21:51:37 +00:00
Dan Gohman b2226e21c3 Initial checkin of the new "fast" instruction selection support. See
the comments in FastISelEmitter.cpp for details on what this is.
This is currently experimental and unusable.

llvm-svn: 54751
2008-08-13 20:19:35 +00:00
Dan Gohman a7b8aed469 Rename SelectionDAGISel's FastISel to Fast, to begin to make
room for the new FastISel instruction selection code.

llvm-svn: 54749
2008-08-13 19:47:40 +00:00
Dan Gohman 23785a1679 Correct the filename in the top-of-file comment.
llvm-svn: 54688
2008-08-12 17:42:33 +00:00
Dan Gohman 127bb03b8c Take the FrameOffset into account when computing the alignment
of stack objects. This fixes PR2656.

llvm-svn: 54646
2008-08-11 18:27:03 +00:00
Evan Cheng 38aa7de6e9 Add skeleton of simple basic block instruction selector.
llvm-svn: 54522
2008-08-08 07:27:28 +00:00
Bruno Cardoso Lopes de5161fdf2 Add the remaining fp_round libcalls:
FPROUND_F80_F32, FPROUND_PPCF128_F32, 
FPROUND_F80_F64, FPROUND_PPCF128_F64

Support for soften float fp_round operands is added, Mips 
needs this to round f64->f32.

Also added support to soften float FABS result, Mips doesn't 
support double fabs results while in 'single float only' mode.

llvm-svn: 54484
2008-08-07 19:01:24 +00:00
Evan Cheng 0638115a6e Factor code that finalize PHI nodes, jump tables, etc. out of SelectBasicBlock. No functionality changes.
llvm-svn: 54438
2008-08-07 00:43:25 +00:00
Owen Anderson 7c42ac4133 Remove the -disable-correct-folding option, which was ugly and is no longer needed.
llvm-svn: 54361
2008-08-05 18:27:54 +00:00
Dan Gohman e955c481fd Fix several const-correctness issues, resolving some -Wcast-qual warnings.
llvm-svn: 54349
2008-08-05 14:45:15 +00:00
Owen Anderson bbeb8f0807 This option doesn't need to be a target option. It can be in SDISel instead.
llvm-svn: 54336
2008-08-05 00:27:28 +00:00
Owen Anderson a102290bdc - Fix SelectionDAG to generate correct CFGs.
- Add a basic machine-level dead block eliminator.

These two have to go together, since many other parts of the code generator are unable to handle the unreachable blocks otherwise created.

llvm-svn: 54333
2008-08-04 23:54:43 +00:00
Dan Gohman 90c724cadc Fix SDISel lowering of PHI nodes to use ComputeValueVTs.
This allows it to work correctly on aggregate values.
This fixes PR2623.

llvm-svn: 54331
2008-08-04 23:42:46 +00:00
Dan Gohman 6e023e63cd Fix SDISel lowering of zeroinitializer and undef to use ComputeValueVTs.
This allows it to work correctly on nested aggregate values.
This fixes PR2625.

llvm-svn: 54330
2008-08-04 23:30:41 +00:00
Dale Johannesen c31eb205c1 Add a flag to disable jump table generation (all
switches use the binary search algorithm) for
environments that don't support it.  PPC64 JIT
is such an environment; turn the flag on for that.

llvm-svn: 54248
2008-07-31 18:13:12 +00:00
Dan Gohman 345d63ccf2 Improve dagcombining for sext-loads and sext-in-reg nodes.
llvm-svn: 54239
2008-07-31 00:50:31 +00:00
Dan Gohman 88e0df0c91 Move SelectionDAG::viewGraph() out of line; as an inline function
it isn't always visible to gdb.

llvm-svn: 54228
2008-07-30 18:48:53 +00:00
Dan Gohman 2fe4352691 Don't look for leaf values to store when lowering stores of
empty structs. This fixes PR2612.

llvm-svn: 54226
2008-07-30 18:36:51 +00:00
Nate Begeman 82f1925708 Fix broken CellSPU lowering, re-instate braces in Legalize
llvm-svn: 54168
2008-07-29 19:07:27 +00:00
Nate Begeman d63495ff25 Disable a fix in the previous patch, since it breaks CellSPU.
The CellSPU codegen is broken, but needs to be fixed before we can
put this back in.

llvm-svn: 54164
2008-07-29 18:28:31 +00:00
Nate Begeman fecbc8cff1 Add vector shifts to the IR, patch by Eli Friedman.
CodeGen & Clang work coming next.

llvm-svn: 54161
2008-07-29 15:49:41 +00:00
Dan Gohman 804c95df52 Fold the useful features of alist and alist_node into ilist, and
a new ilist_node class, and remove them. Unlike alist_node,
ilist_node doesn't attempt to manage storage itself, so it avoids
the associated problems, including being opaque in gdb.

Adjust the Recycler class so that it doesn't depend on alist_node.
Also, change it to use explicit Size and Align parameters, allowing
it to work when the largest-sized node doesn't have the greatest
alignment requirement.

Change MachineInstr's MachineMemOperand list from a pool-backed
alist to a std::list for now.

llvm-svn: 54146
2008-07-28 21:51:04 +00:00
Dan Gohman 68e45a361b Make the ScheduleDAG's GraphRoot edge be blue and dashed too, like
the SelectionDAG's.

llvm-svn: 54129
2008-07-27 22:46:49 +00:00
Dan Gohman 2ce6f2ad5e Rename SDOperand to SDValue.
llvm-svn: 54128
2008-07-27 21:46:04 +00:00
Dan Gohman 91e5dcb680 Tidy SDNode::use_iterator, and complete the transition to have it
parallel its analogue, Value::value_use_iterator. The operator* method
now returns the user, rather than the use.

llvm-svn: 54127
2008-07-27 20:43:25 +00:00
Dan Gohman bb5f43ed4d Rename isOnlyUseOf to isOnlyUserOf.
llvm-svn: 54124
2008-07-27 18:06:42 +00:00
Duncan Sands d9374421ea Some binary operations were being treated as
unary operations!  Add support for softening
some additional unary operations like fp_to_sint.

llvm-svn: 54122
2008-07-27 12:28:43 +00:00
Mon P Wang 7334350d31 When splitting a vector shuffle, fixed which type we used for the hi part
llvm-svn: 54007
2008-07-25 01:30:26 +00:00
Dan Gohman 9268601d8a Use AliasAnalysis::pointsToConstantMemory in SDISel to avoid unnecessary
dependencies with constant load nodes. This allows them to be scheduled
freely.

llvm-svn: 54001
2008-07-25 00:04:14 +00:00
Dan Gohman fa1211f69b Enable first-class aggregates support.
Remove the GetResultInst instruction. It is still accepted in LLVM assembly
and bitcode, where it is now auto-upgraded to ExtractValueInst. Also, remove
support for return instructions with multiple values. These are auto-upgraded
to use InsertValueInst instructions.

The IRBuilder still accepts multiple-value returns, and auto-upgrades them
to InsertValueInst instructions.

llvm-svn: 53941
2008-07-23 00:34:11 +00:00
Duncan Sands 775e509525 LegalizeTypes support for VSETCC. Fixes PR2575.
llvm-svn: 53938
2008-07-22 23:54:03 +00:00
Evan Cheng b8ff223f26 Fix pr2566: incorrect assumption about bit_convert. It doesn't not have to output a vector value. Patch by Nicolas Capens!
llvm-svn: 53932
2008-07-22 20:42:56 +00:00
Dan Gohman 57c749294c Make the GraphRoot edge look like a chain edge, which is more accurate,
and use the right result number, in the off chance that the graph root
has multiple result values.

llvm-svn: 53923
2008-07-22 17:52:59 +00:00
Dan Gohman ebeccb44cf Fix grammaros in comments.
llvm-svn: 53884
2008-07-21 22:38:59 +00:00
Dan Gohman f1dc362547 Enhance the GraphWriter support for edge destinations, and teach the
SelectionDAG graph writer to make use of them. Now, nodes with multiple
values are displayed as such, with incoming edges pointing to the
specific value they use.

llvm-svn: 53875
2008-07-21 21:06:55 +00:00
Dan Gohman a6191cde79 After early-lowering the FORMAL_ARGUMENTS node, delete it.
llvm-svn: 53874
2008-07-21 21:04:07 +00:00
Dan Gohman 581cc87f57 Add titles to the various SelectionDAG viewGraph calls
that include useful information like the name of the
block being viewed and the current phase of compilation.

llvm-svn: 53872
2008-07-21 20:00:07 +00:00
Duncan Sands b0e3938651 Add VerifyNode, a place to put sanity checks on
generic SDNode's (nodes with their own constructors
should do sanity checking in the constructor).  Add
sanity checks for BUILD_VECTOR and fix all the places
that were producing bogus BUILD_VECTORs, as found by
"make check".  My favorite is the BUILD_VECTOR with
only two operands that was being used to build a
vector with four elements!

llvm-svn: 53850
2008-07-21 10:20:31 +00:00
Duncan Sands 6b418e750d Softfloat support for FDIV. Patch by
Richard Pennington.

llvm-svn: 53773
2008-07-18 21:18:48 +00:00
Duncan Sands 694228b47d Eliminate unused variable.
llvm-svn: 53772
2008-07-18 21:07:41 +00:00
Duncan Sands 32e387c461 Revert 53729, after waking up in the middle of
the night realising that it was wrong :)  I
think the reason the same type was being used
for the shufflevec of indices as for the actual
indices is so that if one of them needs splitting
then so does the other.  After my patch it might
be that the indices need splitting but not the
rest, yet there is no good way of handling that.
I think the right solution is to not have the
shufflevec be an operand at all: just have it
be the list of numbers it actually is, stored
as extra info in the node.

llvm-svn: 53768
2008-07-18 20:12:05 +00:00
Dan Gohman 7168de7872 When printing MemOperand nodes, only use print() for
PseudoSourceValue values, which never have names. Use getName()
for all other values, because we want to print just a short summary
of the value, not the entire instruction.

llvm-svn: 53738
2008-07-17 21:12:16 +00:00
Duncan Sands 656b256a1a Use a legal type for elements of the vector_shuffle
mask.  These are just indices into the shuffled vector
so their type is unrelated to the type of the
shuffled elements (which is what was being used before).
This fixes vec_shuffle-11.ll when using LegalizeTypes.
What seems to have happened is that Dan's recent change
r53687, which corrected the result type of the shuffle,
somehow caused LegalizeTypes to notice that the mask
operand was a BUILD_VECTOR with a legal type but elements
of an illegal type (i64).  LegalizeTypes legalized this
by introducing a new BUILD_VECTOR of i32 and bitcasting
it to the old type.  But the mask operand is not supposed
to be a bitcast but a straight BUILD_VECTOR of constants,
causing a crash.

llvm-svn: 53729
2008-07-17 19:28:41 +00:00
Dan Gohman 1705968102 Add a new function, ReplaceAllUsesOfValuesWith, which handles bulk
replacement of multiple values. This is slightly more efficient
than doing multiple ReplaceAllUsesOfValueWith calls, and theoretically
could be optimized even further. However, an important property of this
new function is that it handles the case where the source value set and
destination value set overlap. This makes it feasible for isel to use
SelectNodeTo in many very common cases, which is advantageous because
SelectNodeTo avoids a temporary node and it doesn't require CSEMap
updates for users of values that don't change position.

Revamp MorphNodeTo, which is what does all the work of SelectNodeTo, to
handle operand lists more efficiently, and to correctly handle a number
of corner cases to which its new wider use exposes it.

This commit also includes a change to the encoding of post-isel opcodes
in SDNodes; now instead of being sandwiched between the target-independent
pre-isel opcodes and the target-dependent pre-isel opcodes, post-isel
opcodes are now represented as negative values. This makes it possible
to test if an opcode is pre-isel or post-isel without having to know
the size of the current target's post-isel instruction set.

These changes speed up llc overall by 3% and reduce memory usage by 10%
on the InstructionCombining.cpp testcase with -fast and -regalloc=local.

llvm-svn: 53728
2008-07-17 19:10:17 +00:00
Duncan Sands 7e5edf1a1f LegalizeTypes support for what seems to be the
only missing ppc long double operations: FNEG
and FP_EXTEND.

llvm-svn: 53723
2008-07-17 17:35:14 +00:00
Duncan Sands d9256a7ceb Turn LegalizeTypes back off again for the moment:
it is breaking Darwin bootstrap due to missing
functionality.

llvm-svn: 53721
2008-07-17 17:06:03 +00:00
Duncan Sands 77a3d05f1e Factorize some code for determining which libcall to use.
llvm-svn: 53713
2008-07-17 02:36:29 +00:00
Dan Gohman 2714059079 Fix the result type of a VECTOR_SHUFFLE+BIT_CONVERT dagcombine. This
was turned up by some new SelectionDAG assertion checks that I'm
working on.

llvm-svn: 53687
2008-07-16 16:13:58 +00:00
Duncan Sands 2d28e281e9 Add support for promoting and expanding AssertZext
and AssertSext.  Needed when passing huge integer
parameters with the zeroext or signext attributes.

llvm-svn: 53684
2008-07-16 16:03:07 +00:00
Duncan Sands e766b4230e Reorder methods alphabetically. No functionality change.
While this is not a wonderful organizing principle, it
does make it easy to find routines, and clear where to
insert new ones.

llvm-svn: 53672
2008-07-16 11:41:33 +00:00
Duncan Sands c359055fa9 Turn on LegalizeTypes by default.
llvm-svn: 53671
2008-07-16 11:36:51 +00:00
Dan Gohman 1e5aa12b7d SelectionDAG::AssignNodeIds is unused.
llvm-svn: 53636
2008-07-15 18:29:32 +00:00
Dan Gohman 1d093846b5 Don't sort SDNodes by their addresses in SelectionDAG::dump. Instead,
just use the AllNodes order, which is at least relatively stable
across runs.

llvm-svn: 53632
2008-07-15 18:18:54 +00:00
Duncan Sands 0f1a1cdcf8 LegalizeTypes support for fabs on ppc long double.
llvm-svn: 53613
2008-07-15 15:02:44 +00:00
Duncan Sands 6162e0377c LegalizeTypes support for promotion of bswap.
In LegalizeDAG the value is zero-extended to
the new type before byte swapping.  It doesn't
matter how the extension is done since the new
bits are shifted off anyway after the swap, so
extend by any old rubbish bits.  This results
in the final assembler for the testcase being
one line shorter.

llvm-svn: 53604
2008-07-15 10:18:22 +00:00
Duncan Sands 202225cdf8 LegalizeTypes support for promotion of SIGN_EXTEND_INREG.
llvm-svn: 53603
2008-07-15 10:14:24 +00:00
Duncan Sands b9b5a671d3 Reorder the integer promotion methods alphabetically.
No change in functionality.

llvm-svn: 53602
2008-07-15 10:12:34 +00:00
Mon P Wang 97432f4f1b Fixed potential bug if the source and target of a bit convert have different alignment
llvm-svn: 53590
2008-07-15 05:28:34 +00:00
Dan Gohman adec96f438 Reapply 53476 and 53480, with a fix so that it properly updates
the BB member to the current basic block after emitting
instructions.

llvm-svn: 53567
2008-07-14 18:19:29 +00:00
Dan Gohman e7c8387616 Improve debug output for MemOperandSDNode. PseudoSourceValue nodes
don't have value names, so use print instead of getName() to get a
useful string.

llvm-svn: 53563
2008-07-14 17:51:24 +00:00
Duncan Sands 673cf1836b I don't think BUILD_PAIR can have a vector result.
Remove support for this.

llvm-svn: 53559
2008-07-14 17:34:19 +00:00
Duncan Sands 0ca9a38f68 Tighten up some checks. Fix FPOWI splitting for
non-power-of-two vectors.

llvm-svn: 53558
2008-07-14 17:33:37 +00:00
Duncan Sands a30cbd9797 An INSERT_VECTOR_ELT can insert a larger value
than the vector element type.  Don't forget to
handle this when the insertion index is not a
constant.

llvm-svn: 53556
2008-07-14 17:32:02 +00:00
Duncan Sands 693185bcee According to the docs, it is possible to have an
extending load of a vector.  Handle this case when
splitting vector loads.  I'm not completely sure
what is supposed to happen, but I think it means
hi should be set to undef.  LegalizeDAG does not
consider this case.

llvm-svn: 53555
2008-07-14 17:27:46 +00:00
Duncan Sands b766084cb0 There should be no extending loads or truncating
stores of one-element vectors.  Also, neaten the
handling of INSERT_VECTOR_ELT when the inserted
type is larger than the vector element type.

llvm-svn: 53554
2008-07-14 17:22:31 +00:00
Duncan Sands d47d2d6b12 Ignore TargetConstant with an illegal type. These
are used for passing huge immediates in inline ASM
from the front-end straight down to the ASM writer.
Of course this is a hack, but it is simple, limited
in scope, works in practice, and is what LegalizeDAG
does.

llvm-svn: 53553
2008-07-14 17:15:45 +00:00
Evan Cheng ef8412c822 Back out 53476 and 53480 for now. Somehow they cause llc to miscompile 179.art.
llvm-svn: 53502
2008-07-12 01:38:51 +00:00
Dan Gohman 02c7c6cb33 Include a frame index in the "fixed stack" pseudo source value
instead of using the frame index for the SVOffset, which was
inconsistent.

llvm-svn: 53486
2008-07-11 22:44:52 +00:00
Dan Gohman ed087a62dc Fix an obsolete top-level comment.
llvm-svn: 53481
2008-07-11 22:39:58 +00:00
Dan Gohman f4cd404e6f Factor out debugging code into the common base class.
llvm-svn: 53480
2008-07-11 22:36:22 +00:00
Dan Gohman 36a69373dc Add support for putting NamedRegionTimers in TimerGroups, and
use a timer group for the timers in SelectionDAGISel. Also,
Split scheduling out from emitting, to give each their own
timer.

llvm-svn: 53476
2008-07-11 21:54:34 +00:00
Dan Gohman 0597e5b697 Trim unnecessary #includes.
llvm-svn: 53471
2008-07-11 20:38:31 +00:00
Duncan Sands 121641d601 Remove an apparently useless routine: there should
be no need to split the result of a vector RET node,
since they are always already legal.

llvm-svn: 53462
2008-07-11 17:02:09 +00:00
Duncan Sands 3e7d0fa3ca It is pointless to turn a UINT_TO_FP into an
SINT_TO_FP libcall plus additional operations:
it might as well be a direct UINT_TO_FP libcall.
So only turn it into an SINT_TO_FP if the target
has special handling for SINT_TO_FP.

llvm-svn: 53461
2008-07-11 17:00:14 +00:00
Duncan Sands 37b7322b35 Add two missing SINT_TO_FP libcalls.
llvm-svn: 53460
2008-07-11 16:57:02 +00:00
Duncan Sands d9948110a6 Port a shift-by-1 optimization from LegalizeDAG: it
was presumably added after the rest of the code was
copied to LegalizeTypes.

llvm-svn: 53459
2008-07-11 16:54:57 +00:00
Duncan Sands 927a3648d5 Add support for 128 bit shifts and 32 bit shifts
on 16 bit machines.

llvm-svn: 53458
2008-07-11 16:52:29 +00:00
Chris Lattner 87909d0629 Fix a bug in the soft-float handling of FCOPYSIGN that Duncan noticed
when working on legalizetypes.  Both legalizetypes and legalizeops now
produce hte same code for CodeGen/ARM/fcopysign.ll.

llvm-svn: 53435
2008-07-10 23:46:13 +00:00
Chris Lattner 17b234cf9b make legalize types be a command line option: -enable-legalize-types.
llvm-svn: 53434
2008-07-10 23:37:50 +00:00
Duncan Sands abdcac66dc Add support for 128 bit multiplicative operations.
Lack of these caused a bootstrap failure with Fortran
on x86-64 with LegalizeTypes turned on.  While there,
be nice to 16 bit machines and support expansion of
i32 too.

llvm-svn: 53408
2008-07-10 15:35:05 +00:00
Duncan Sands 5e6d1402c2 Add a mysteriously missing libcall, FPTOSINT_F80_I32.
Be nice to 16 bit machines by supporting FP_TO_XINT
expansion for these.

llvm-svn: 53407
2008-07-10 15:33:02 +00:00
Duncan Sands 303524be58 Fix a FIXME: use an apint in CTTZ legalization.
llvm-svn: 53406
2008-07-10 15:30:54 +00:00
Duncan Sands e78352a125 Remove PromoteIntRes_FP_ROUND - not sure what it
was doing there: FP_ROUND returns a float, not an
integer.

llvm-svn: 53405
2008-07-10 15:29:55 +00:00
Duncan Sands 4ac3984fc5 Make sure the alignment of the temporary created
in CreateStackStoreLoad is good enough for both
the source and destination types.

llvm-svn: 53404
2008-07-10 15:26:17 +00:00
Duncan Sands d4c09df689 Make the LegalizeType method naming scheme more regular.
llvm-svn: 53403
2008-07-10 15:25:04 +00:00
Duncan Sands 74f23ff45c Don't barf when dumping a constant that contains
a ginormous value (eg: i128 -1).

llvm-svn: 53402
2008-07-10 11:23:14 +00:00
Dan Gohman 7d94c49db9 Simplify hasNUsesOfValue and hasAnyUsesOfValue even more. This
makes their special-case checks of use_size() less beneficial,
so remove them. This eliminates all but one use of use_size(),
which is in AssignTopologicalOrder, which uses it only once for
each node, and so can reasonably afford to recompute it, as
this allows the UsesSize field of SDNode to be removed
altogether.

llvm-svn: 53377
2008-07-09 23:03:14 +00:00
Dan Gohman 7a510c2990 hasAnyUseOfValue can check SDUse nodes of its users directly instead
of examining every operand of every user.

llvm-svn: 53374
2008-07-09 22:39:01 +00:00
Dan Gohman db4504fa57 Move MemoryVT out of LSBaseNode into MemSDNode, allowing the
getMemOperand function to be moved into the base class as well
and made non-virtual.

llvm-svn: 53372
2008-07-09 22:08:04 +00:00
Dan Gohman 89e71d48b8 Move the IsVolatile and SVOffset fields into the MemSDNode base
class, and store IsVolatile and Alignment in a more compact form.
This makes AtomicSDNode slightly larger, but it shrinks LoadSDNode
and StoreSDNode, which are much more common and are the largest of
the SDNode subclasses. Also, this lets the isVolatile() and
getAlignment() accessors be non-virtual.

llvm-svn: 53361
2008-07-09 21:23:02 +00:00
Duncan Sands 37ab611e8e Remove some unneeded includes.
llvm-svn: 53289
2008-07-09 12:08:25 +00:00
Duncan Sands 5e266c914a Redo LegalizeTypes soft float support for
SINT_TO_FP and UINT_TO_FP.  This now produces
the same code as LegalizeDAG (the previous
code was based on a mistaken idea of what
LegalizeDAG did in this case).

llvm-svn: 53288
2008-07-09 12:07:22 +00:00
Duncan Sands b9e63db718 Forgot to update the chain result when softening
loads.

llvm-svn: 53287
2008-07-09 11:15:31 +00:00
Duncan Sands ed811f0ec1 LegalizeTypes soft float support for FP_TO_SINT and
FP_TO_UINT.

llvm-svn: 53286
2008-07-09 11:13:46 +00:00
Duncan Sands 8090f8576f LegalizeTypes support for powi soft float.
llvm-svn: 53285
2008-07-09 11:11:47 +00:00
Duncan Sands c52d3bf646 Make the role of MVT::i32 clearer here, and add a
note since it is not clear whether it is correct.

llvm-svn: 53284
2008-07-09 08:07:41 +00:00
Evan Cheng 7898e98026 Missed alignment argument on stores lowered from memcpy.
llvm-svn: 53281
2008-07-09 06:38:06 +00:00
Dan Gohman 919936815e const-ify SelectionDAG::getNodeValueTypes.
llvm-svn: 53264
2008-07-09 00:00:42 +00:00
Dan Gohman e8d8d2ea42 Factor out the code for computing an alignment value, and make it
available to getAtomic in addition to just getLoad and getStore,
to prevent MachineMemOperands with 0 alignment.

llvm-svn: 53261
2008-07-08 23:46:32 +00:00
Evan Cheng 34ef1db87c Do not CSE DEBUG_LOC, DBG_LABEL, DBG_STOPPOINT, DECLARE, and EH_LABEL SDNode's. This improves compile time slightly at -O0 -g.
llvm-svn: 53246
2008-07-08 20:06:39 +00:00