aboutsummaryrefslogtreecommitdiff
path: root/lib/CodeGen/SelectionDAG
AgeCommit message (Collapse)Author
2008-02-13Avoid setting bits that aren't demanded.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47098 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-13Simplify some logic in ComputeMaskedBits. And change ComputeMaskedBitsDan Gohman
to pass the mask APInt by value, not by reference. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47096 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-13Teach LegalizeTypes how to expand and promote CTLZ,Duncan Sands
CTTZ and CTPOP. The expansion code differs from that in LegalizeDAG in that it chooses to take the CTLZ/CTTZ count from the Hi/Lo part depending on whether the Hi/Lo value is zero, not on whether CTLZ/CTTZ of Hi/Lo returned 32 (or whatever the width of the type is) for it. I made this change because the optimizers may well know that Hi/Lo is zero and exploit it. The promotion code for CTTZ also differs from that in LegalizeDAG: it uses an "or" to get the right result when the original value is zero, rather than using a compare and select. This also means the value doesn't need to be zero extended. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47075 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-13In SDISel, for targets that support FORMAL_ARGUMENTS nodes, lower thisChris Lattner
node as soon as we create it in SDISel. Previously we would lower it in legalize. The problem with this is that it only exposes the argument loads implied by FORMAL_ARGUMENTs after legalize, so that only dag combine 2 can hack on them. This causes us to miss some optimizations because datatype expansion also happens here. Exposing the loads early allows us to do optimizations on them. For example we now compile arg-cast.ll to: _foo: movl $2147483647, %eax andl 8(%esp), %eax ret where we previously produced: _foo: subl $12, %esp movsd 16(%esp), %xmm0 movsd %xmm0, (%esp) movl $2147483647, %eax andl 4(%esp), %eax addl $12, %esp ret It might also make sense to do this for ISD::CALL nodes, which have implicit stores on many targets. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47054 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-13teach dag combiner how to eliminate MERGE_VALUES nodes.Chris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47052 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-13Support legalizing insert_vector_elt on targets where the elementNate Begeman
type is not legal. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47048 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-13Convert SelectionDAG::ComputeMaskedBits to use APInt instead of uint64_t.Dan Gohman
Add an overload that supports the uint64_t interface for use by clients that haven't been updated yet. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47039 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-12Generalize getCopyFromParts and getCopyToParts toDuncan Sands
handle arbitrary precision integers and any number of parts. For example, on a 32 bit machine an i50 corresponds to two i32 parts. getCopyToParts will extend the i50 to an i64 then write half of the i64 to each part; getCopyFromParts will combine the two i32 parts into an i64 then truncate the result to i50. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47024 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-11Generalize the handling of call and return arguments,Duncan Sands
in preparation for apint support. These changes are intended to have no functional effect. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46967 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-11From Chris' review: use isa instead of explicitly using classof.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46964 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-11From Chris' review: minor corrections in comments.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46963 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-11From Chris' review: use cast instead of dyn_cast with an assert.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46962 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-11Added "Profile" method to APFloat for use with FoldingSet.Ted Kremenek
Added member template "Add" to FoldingSetNodeID that allows "adding" arbitrary objects to a profile via dispatch to FoldingSetTrait<T>::Profile(). Removed FoldingSetNodeID::AddAPFloat and FoldingSetNodeID::APInt, as their functionality is now replaced using the above mentioned member template. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46957 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-11Add a isBigEndian method to complement isLittleEndian.Duncan Sands
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46954 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-10Rename MRegisterInfo to TargetRegisterInfo.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46930 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-10Add truncate and AssertZext result expansion.Duncan Sands
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46926 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-10Return "(c1 + c2)" instead of yet another ADD node (which made this aBill Wendling
no-op). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46922 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-08Change ConstantSDNode to store an APInt instead of a uint64_t, andDan Gohman
begin adding some methods to use it this way. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46899 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-08Avoid needlessly casting away const qualifiers.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46876 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-07Follow Chris' suggestion; change the PseudoSourceValue accessorsDan Gohman
to return pointers instead of references, since this is always what is needed. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46857 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-06Re-apply the memory operand changes, with a fix for the staticDan Gohman
initializer problem, a minor tweak to the way the DAGISelEmitter finds load/store nodes, and a renaming of the new PseudoSourceValue objects. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46827 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-05Fix PR1975: dag isel emitter produces patterns that isel wrong flag result.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46776 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-04Typo.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46725 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-04I don't see how NodeUpdated can be called with aDuncan Sands
ReadyToProcess node - add an assertion to check this. Add an assertion to NodeDeleted that checks that processed/ready nodes are indeed not deleted. It is because they are never deleted that none of the maps can have a deleted node as the source of a mapping. It does however seem to be possible in theory to have a deleted value as the target of a mapping, however this has not yet been spotted in the wild. Still mulling on what to do about this. [The theoretical situation is this: a node A is expanded/promoted/whatever to a newly created node B. Thus A->B is added to a map. When the subtree rooted at B is legalized it is conceivable that B is deleted due to RAUW on a node somewhere above it]. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46705 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-03Fix typoChris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46682 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-03handle the case where a node can become ready to processChris Lattner
multiple times due to a RAUW. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46680 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-03Use the new infrastructure for listening to node updates to Chris Lattner
keep the LegalizeTypes node flags up to date when doing a RAUW. This fixes a nasty bug that Duncan ran into and makes the previous (nonbuggy case) more efficent. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46679 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-03the world doesn't need my debugging code.Chris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46678 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-03Change the 'global modification' APIs in SelectionDAG to take a newChris Lattner
DAGUpdateListener object pointer instead of just returning a vector of deleted nodes. This makes the interfaces more efficient (no more allocating a vector [at least a malloc], filling it in, then walking it) and more clean. This also allows the client to be notified of nodes that are *changed* but not deleted. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46677 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-03Generalize the SDOperand->SDOperand form of Chris Lattner
SelectionDAG::ReplaceAllUsesWith to handle replacement of an SDOperand with *any* sdoperand, not just one for a node with a single result. Note that this has a horrible FIXME'd hack in it to work around PR1975. This should be removed when PR1975 is fixed. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46674 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-03add a -view-legalize-types-dags option, for viewing the dags going into ↵Chris Lattner
legalize types. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46672 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-02SDIsel processes llvm.dbg.declare by recording the variable debug ↵Evan Cheng
information descriptor and its corresponding stack frame index in MachineModuleInfo. This only works if the local variable is "homed" in the stack frame. It does not work for byval parameter, etc. Added ISD::DECLARE node type to represent llvm.dbg.declare intrinsic. Now the intrinsic calls are lowered into a SDNode and lives on through out the codegen passes. For now, since all the debugging information recording is done at isel time, when a ISD::DECLARE node is selected, it has the side effect of also recording the variable. This is a short term solution that should be fixed in time. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46659 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-01Remove the nasty LABEL hack with a much less evil one. Now ↵Evan Cheng
llvm.dbg.func.start implies a stoppoint is set. SelectionDAGISel records a new source line but does not create a ISD::LABEL node for this special stoppoint. Asm printer will magically print this label. This ensures nothing is emitted before. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46635 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-01Rename RecordLabel to RecordSourceLine because that's what it is doing.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46628 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-31Revert 46556 and 46585. Dan please fix the PseudoSourceValue problem and ↵Evan Cheng
re-commit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46623 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-31Add a comment for a nasty short term hack.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46610 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-31Add an extra operand to LABEL nodes which distinguishes between debug, EH, ↵Evan Cheng
or misc labels. This fixes the EH breakage. However I am not convinced this is *the* solution. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46609 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-31Add more thorough error checking for NULL register classes.Christopher Lamb
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46605 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-31Rename ISD::FLT_ROUNDS to ISD::FLT_ROUNDS_ to avoid conflictingDan Gohman
with the real FLT_ROUNDS (defined in <float.h>). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46587 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-31Create a new class, MemOperand, for describing memory referencesDan Gohman
in the backend. Introduce a new SDNode type, MemOperandSDNode, for holding a MemOperand in the SelectionDAG IR, and add a MemOperand list to MachineInstr, and code to manage them. Remove the offset field from SrcValueSDNode; uses of SrcValueSDNode that were using it are all all using MemOperandSDNode now. Also, begin updating some getLoad and getStore calls to use the PseudoSourceValue objects. Most of this was written by Florian Brander, some reorganization and updating to TOT by me. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46585 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-30Treat the label for the first @llvm.dbg.stoppoint the same way as the ↵Evan Cheng
dbg_func_start label. Make sure nothing else is inserted before them. Note this solution might be somewhat fragile since ISD::LABEL may be used for other purposes. If that ends up to be an issue, we may need to introduce a different node for debug labels. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46571 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-30A semi-gross fix for a debug info issue. When inserting the "function start" ↵Evan Cheng
label (i.e. first label in the entry block) take care to insert it at the beginning of the block. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46568 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-30Even though InsertAtEndOfBasicBlock is an ugly hack it still deserves a ↵Evan Cheng
proper name. Rename it to EmitInstrWithCustomInserter since it does not necessarily insert instruction at the end. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46562 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-30Factor the addressing mode and the load/store VT out of LoadSDNodeDan Gohman
and StoreSDNode into their common base class LSBaseSDNode. Member functions getLoadedVT and getStoredVT are replaced with the common getMemoryVT to simplify code that will handle both loads and stores. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46538 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-29When expanding an operand, it's not the result valueDuncan Sands
type that matters but the operand type. This fixes 2008-01-08-IllegalCMP.ll which crashed with the new legalize infrastructure because SETCC with result type i8 and operand type i64 was being custom expanded by the X86 backend. With this fix, the gcc build gets as far as the first libcall. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46525 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-29Use empty() instead of comparing size() with zero.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46514 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-29Fix a typo in a comment.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46513 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-29Fix a typo in a comment.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46508 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-29Properly expand extract-element for non-power-of-2 codegenNate Begeman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46486 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-29Handle 'X' constraint in asm's better.Dale Johannesen
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46485 91177308-0d34-0410-b5e6-96231b3b80d8