aboutsummaryrefslogtreecommitdiff
path: root/lib/CodeGen/SelectionDAG
AgeCommit message (Collapse)Author
2010-03-01add a new OPC_SwitchOpcode which is semantically equivalentChris Lattner
to a scope where every child starts with a CheckOpcode, but executes more efficiently. Enhance DAGISelMatcherOpt to form it. This also fixes a bug in CheckOpcode: apparently the SDNodeInfo objects are not pointer comparable, we have to compare the enum name. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97438 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-28eliminate GetInt1/2Chris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97426 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-28hoist the new isel interpreter out of DAGISelHeader.hChris Lattner
(which gets #included into the middle of each target's DAGISel class) into a .cpp file where it is only compiled once. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97425 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-28enhance the new isel to handle the 'node already exists'Chris Lattner
case of MorphNodeTo directly. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97417 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-28simplify this code, return only ever has zero or one operands.Chris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97408 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-27Re-apply 97040 with fix. This survives a ppc self-host llvm-gcc bootstrap.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97310 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-26Move dbg_value generation to target-independent FastISel,Dale Johannesen
as X86 is currently the only FastISel target. Per review. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97255 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-25Fix ExpandVectorBuildThroughStack for the case where theDan Gohman
operands are themselves vectors. Based on a patch by Micah Villmow for PR6338. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97165 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-25Revert r97064. Duncan pointed out that bitcasts are defined inDan Gohman
terms of store and load, which means bitcasting between scalar integer and vector has endian-specific results, which undermines this whole approach. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97137 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-24clean up various VT manipulations, patch by Micah Villmow! PR6337Chris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97072 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-24Make getTypeSizeInBits work correctly for array types; it should returnDan Gohman
the number of value bits, not the number of bits of allocation for in-memory storage. Make getTypeStoreSize and getTypeAllocSize work consistently for arrays and vectors. Fix several places in CodeGen which compute offsets into in-memory vectors to use TargetData information. This fixes PR1784. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97064 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-24convert cycle checker to smallptrset, add comments and make itChris Lattner
more elegant. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97059 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-24revert david's patch which does not even build.Chris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97057 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-24Use a SmallPtrSet as suggested by Chris.David Greene
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97056 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-24Speculatively revert r97011, "Re-apply 96540 and 96556 with fixes.", again inDaniel Dunbar
the hopes of fixing PPC bootstrap. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97040 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-24When forming SSE min and max nodes for UGE and ULE comparisons, it'sDan Gohman
necessary to swap the operands to handle NaN and negative zero properly. Also, reintroduce logic for checking for NaN conditions when forming SSE min and max instructions, fixed to take into consideration NaNs and negative zeros. This allows forming min and max instructions in more cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97025 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-24Change the scheduler from adding nodes in allnodes orderChris Lattner
to adding them in a determinstic order (bottom up from the root) based on the structure of the graph itself. This updates tests for some random changes, interesting bits: CodeGen/Blackfin/promote-logic.ll no longer crashes. I have no idea why, but that's good right? CodeGen/X86/2009-07-16-LoadFoldingBug.ll also fails, but now compiles to have one fewer constant pool entry, making the expected load that was being folded disappear. Since it is an unreduced mass of gnast, I just removed it. This fixes PR6370 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97023 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-24add node #'s to debug dumps.Chris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97019 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-24Re-apply 96540 and 96556 with fixes.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97011 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-23make selectnodeto set the nodeid to -1. This makes it more akin toChris Lattner
creating a new node then replacing uses. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97000 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-23fix a bug in findNonImmUse (used by IsLegalToFold) where nodes withChris Lattner
no id's would cause early exit allowing IsLegalToFold to return true instead of false, producing a cyclic dag. This was striking the new isel because it isn't using SelectNodeTo yet, which theoretically is just an optimization. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96972 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-23Print node ID's in dumps and views if set.Chris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96971 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-23Speed up cycle checking significantly by caching results.David Greene
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96956 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-19Revert commits 96556 and 96640, because commit 96556 breaks theDuncan Sands
dragonegg self-host build. I reverted 96640 in order to revert 96556 (96640 goes on top of 96556), but it also looks like with both of them applied the breakage happens even earlier. The symptom of the 96556 miscompile is the following crash: llvm[3]: Compiling AlphaISelLowering.cpp for Release build cc1plus: /home/duncan/tmp/tmp/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp:4982: void llvm::SelectionDAG::ReplaceAllUsesWith(llvm::SDNode*, llvm::SDNode*, llvm::SelectionDAG::DAGUpdateListener*): Assertion `(!From->hasAnyUseOfValue(i) || From->getValueType(i) == To->getValueType(i)) && "Cannot use this version of ReplaceAllUsesWith!"' failed. Stack dump: 0. Running pass 'X86 DAG->DAG Instruction Selection' on function '@_ZN4llvm19AlphaTargetLowering14LowerOperationENS_7SDValueERNS_12SelectionDAGE' g++: Internal error: Aborted (program cc1plus) This occurs when building LLVM using LLVM built by LLVM (via dragonegg). Probably LLVM has miscompiled itself, though it may have miscompiled GCC and/or dragonegg itself: at this point of the self-host build, all of GCC, LLVM and dragonegg were built using LLVM. Unfortunately this kind of thing is extremely hard to debug, and while I did rummage around a bit I didn't find any smoking guns, aka obviously miscompiled code. Found by bisection. r96556 | evancheng | 2010-02-18 03:13:50 +0100 (Thu, 18 Feb 2010) | 5 lines Some dag combiner goodness: Transform br (xor (x, y)) -> br (x != y) Transform br (xor (xor (x,y), 1)) -> br (x == y) Also normalize (and (X, 1) == / != 1 -> (and (X, 1)) != / == 0 to match to "test on x86" and "tst on arm" r96640 | evancheng | 2010-02-19 01:34:39 +0100 (Fri, 19 Feb 2010) | 16 lines Transform (xor (setcc), (setcc)) == / != 1 to (xor (setcc), (setcc)) != / == 1. e.g. On x86_64 %0 = icmp eq i32 %x, 0 %1 = icmp eq i32 %y, 0 %2 = xor i1 %1, %0 br i1 %2, label %bb, label %return => testl %edi, %edi sete %al testl %esi, %esi sete %cl cmpb %al, %cl je LBB1_2 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96672 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-19Transform (xor (setcc), (setcc)) == / != 1 toEvan Cheng
(xor (setcc), (setcc)) != / == 1. e.g. On x86_64 %0 = icmp eq i32 %x, 0 %1 = icmp eq i32 %y, 0 %2 = xor i1 %1, %0 br i1 %2, label %bb, label %return => testl %edi, %edi sete %al testl %esi, %esi sete %cl cmpb %al, %cl je LBB1_2 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96640 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-18Some dag combiner goodness:Evan Cheng
Transform br (xor (x, y)) -> br (x != y) Transform br (xor (xor (x,y), 1)) -> br (x == y) Also normalize (and (X, 1) == / != 1 -> (and (X, 1)) != / == 0 to match to "test on x86" and "tst on arm" git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96556 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-17Make the non-temporal bit "significant" in MemSDNodes so they aren'tDavid Greene
CSE'd or otherwise combined with temporal MemSDNodes. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96505 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-17sink special case "cannotyetselect" for intrinsics out of the Chris Lattner
tblgen splatted code into the implementation. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96460 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-16There are two ways of checking for a given type, for example isa<PointerType>(T)Duncan Sands
and T->isPointerTy(). Convert most instances of the first form to the second form. Requested by Chris. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96344 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-15Fix a memory leak. Patch by Nicolas Geoffray.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96295 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-15Split SelectionDAGISel::IsLegalAndProfitableToFold toEvan Cheng
IsLegalToFold and IsProfitableToFold. The generic version of the later simply checks whether the folding candidate has a single use. This allows the target isel routines more flexibility in deciding whether folding makes sense. The specific case we are interested in is folding constant pool loads with multiple uses. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96255 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-15Add non-temporal flags and remove an assumption of default arguments.David Greene
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96240 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-15Uniformize the names of type predicates: rather than having isFloatTy andDuncan Sands
isInteger, we now have isFloatTy and isIntegerTy. Requested by Chris! git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96223 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-11Use array_pod_sort instead of std::sort for improved code size.Jakob Stoklund Olesen
Use SmallVector instead of std::vector for better speed when indirectbr has few successors. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95879 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-11Remove duplicate successors from indirectbr instructions before building the ↵Jakob Stoklund Olesen
machine CFG. This makes early tail duplication run 60 times faster when compiling the Firefox JavaScript interpreter, see PR6186. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95831 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-10The previous fix of widening divides that trap was too fragile as it depends ↵Mon P Wang
on custom lowering and requires that certain types exist in ValueTypes.h. Modified widening to check if an op can trap and if so, the widening algorithm will apply only the op on the defined elements. It is safer to do this in widening because the optimizer can't guarantee removing unused ops in some cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95823 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-10Fix "the the" and similar typos.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95781 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-10Now that ShrinkDemandedOps() is separated out from DAG combine. It sometimes ↵Evan Cheng
leave some obvious nops which dag combine used to clean up afterwards e.g. (trunk (ext n)) -> n. Look for them and squash them. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95757 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-10Emit an error for illegal inline asm constraint (which uses illegal type) ↵Evan Cheng
rather than asserting. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95746 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-10Fix comments to reflect renaming elsewhere.Dale Johannesen
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95730 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-09Only dump output in debug mode.David Greene
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95711 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-09move target-independent opcodes out of TargetInstrInfoChris Lattner
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. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95687 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-08Apply the 95471 fix to SelectionDAGBuilder as well;Dale Johannesen
we can get in here if FastISel gives up in a block. (Actually the two copies of this need to be unified. Later.) git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95579 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-08In guaranteed tailcall mode, don't decline the tailcall optimizationDan Gohman
for blocks ending in "unreachable". git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95565 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-06After Victor's latest commits I am seeing nullDale Johannesen
addresses in dbg.declare; ignore this for the moment to prevent things from breaking. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95471 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-05When the scheduler unfold a load folding instruction it move some of the ↵Evan Cheng
predecessors to the unfolded load. It decides what gets moved to the load by checking whether the new load is using the predecessor as an operand. The check neglects the cases whether the predecessor is a flagged scheduling unit. rdar://7604000 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95339 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-04Fix typo Duncan noticed.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95322 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-04It's too risky to eliminate sext / zext of call results for tail call ↵Evan Cheng
optimization even if the caller / callee attributes completely match. The callee may have been bitcast'ed (or otherwise lied about what it's doing). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95282 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-03Revert 94937 and move the noreturn check to codegen.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95198 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-03Allow all types of callee's to be tail called. But avoid automatic tailcall ↵Evan Cheng
if the callee is a result of bitcast to avoid losing necessary zext / sext etc. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95195 91177308-0d34-0410-b5e6-96231b3b80d8
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2026-03-04 15:33:14 +0000